Translational vaccinomics and structural filtration algorithm to device multiepitope vaccine for catastrophic monkeypox virus

Abstract

Recent outbreak of monkeypox disease commenced in April 2022, and on May 7, the first confirmed case was reported. The world health organization then designated monkeypox disease as a public health emergency of international outrage on July 23, after it spread to 70 non-endemic nations in less than 15 days. This catastrophic viral infection encourages the development of antiviral therapeutics due to the lack of specific treatments with negligible adverse effects. This analysis developed a highly immunogenic multiepitope subunit vaccine against the monkeypox virus using an in silico translational vaccinomics technique. Highly antigenic B cell and T cell (HTL and CTL) epitopes were predicted and conjugated with the help of unique linkers. An adjuvant (β-defensin) and a pan-HLA DR sequence were attached at the vaccine construct's N-terminal to invoke a robust immunological response. Additionally, physiochemical, allergic, toxic, and antigenic properties were anticipated. Interactions between the vaccine candidate and the TLR3 demonstrated that the vaccine candidate triggers a robust immunological response. Finally, the stability is confirmed by the molecular dynamics study. In contrast, the modified vaccine candidate's ability to produce a protective immune response were verified by an immune dynamics simulation study conducted via C-ImmSim server. This study validates the generation of B cell, Th cell, and Tc cell populations as well as the production of IFN‐γ.

Graphical abstract

1. Introduction

The outbreak of the monkeypox virus (MPXV) in 2022 has terrified people, alerted scientists, and caused concern due to its stealthy and quick-spread nature [1]. Monkeypox is an unusual viral zoonotic disease caused by MPXV, a member of the Poxviridae family's Orthopoxvirus genus and a close relative of the variola virus well known to cause smallpox disease [2]. The virus was initially identified in 1958 in a Danish laboratory in monkeys, earning the nomenclature MPXV. The first human MPXV case was identified in 1970 in Zaire (DRC). Prior to the 1970s, this disease was thought to be camouflaged by smallpox [3]. In accordance with a genomic comparison between the human MPXV and the variola virus conducted by Shchelkunov et al., the central region of the MPXV genome is 96.3% identical to the smallpox virus [4]. According to various experts, the 2022 outbreak would be driven by the termination of smallpox vaccination (40 years ago), which left almost 70% of the population unprotected and allowed other closely related orthopoxviruses, such as monkeypox, to spread more quickly [5,6]. The disease is mostly brought on by Clade I and Clade II of the MPXV, particularly Clade IIb, accounting for the 2022 outbreak. Clade II, formally the West African strain, has a milder illness and ∼1% mortality rate, whereas the Congo basin or Central African strain of the MPXV causes a more severe infection with a ∼10% fatality rate [7]. As of September 1, 2022, 100 countries have positively reported more than 52,500 confirmed cases of MPXV, seven of that previously reported monkeypox, according to the Centers for Disease Control and Prevention (CDC). 26 countries have reported more than 100 cases, while nine worst-affected have reported more over 1000 cases [8]. Various non-human primates and animals like tree squirrels, rabbits, rodent and may act as a reservoir as MPXV can naturally infect these species [9]. There are currently no specific pharmacological therapies available to treat this disease. However, several antivirals (such as tecovirimat, brincidofovir, and cidofovir) have been repurposed to manage monkeypox symptoms.

For sudden outbreaks or emergencies, tecovirimat, brincidofovir, and cidofovir are the best drugs approved by US FDA. These drugs have been used to treat other viral infections. Tecovirimat and brincidofovir were initially used to treat smallpox in children and adults. There is no big analysis of the effectiveness of these two antiviral drugs, but these drugs work against orthopoxviruses in in vitro and animal studies. Similarly, cidofovir, has shown effectiveness in the treatment of cytomegalovirus retinitis in patients with acquired immunodeficiency syndrome [10].

Nevertheless, all these medications have serious adverse effects on the individual suffering from an infection. These drugs are licensed to treat smallpox or other orthopoxviruses-related diseases [11]. Regarding vaccinations, JYNNEOS (for individuals at risk for serious adverse events) and ACAM2000 (for people with HIV or other immunocompromised conditions) are the two vaccines that are currently in use for MPXV. Both are live vaccines and FDA-approved for smallpox. JYNNEOS cannot replicate effectively within human cells and generates an immune response two weeks after obtaining the second dose.

In comparison, ACAM2000 is replication proficient and emits a protective immunity four weeks after acquiring the dose. Besides, adverse reactions, including enlargement of the lymph nodes, fever, rashes, and redness, have been linked to these immunizations. It is advised against administering these vaccines to anyone allergic to any vaccine component. However, JYNNEOS is safe for HIV-positive or immunocompromised patients [12]. One more antibody treatment is available, FDA-approved VIGIV (Vaccinia Immune Globulin Intravenous). VIGIV contains Immunoglobulin G prepared from the plasma of people who have received smallpox vaccinations and used to treat complications introduced by the vaccinia virus. Monkeypox and other orthopoxviruses can be treated with VIGIV [13]. Till September 2, 2022, a total of 11 studies have been registered for monkeypox on [14], supported by the US National Library of Medicine. No study is reported for early phase 1, two studies in Phase 1 clinical trial, one in Phase 2, one in Phase 3, no study has been found to register in Phase 4, and one is not applicable [14].

Theoretically, every viral protein is equally crucial for formulating a vaccine. However, acquiring antigenic proteins will always help in boosting the immune response. The proteins selected for this study are as follows: (1) Chemokine binding protein- Chemotactic cytokines, sometimes referred to as chemokines, are a family of secreted proteins that promote cell migration and are essential for immunity. The poxviruses release a chemokine-binding protein that has a higher affinity for the host chemokines and inhibits their action [15]. (2) Ankyrin (ANK) repeat motif- 33mer residues that make up an ANK motif promote association with several other specific proteins. Signal transduction, cytoskeletal control, and transcriptional activation are just a few of the diverse roles performed by ANK proteins. The nuclear factor-κB (NF-κB) transcriptional pathway has a role in viral suppression, which is manipulated by the interaction of poxviral ANK protein with NF-κB [16,17]. (3) EEV maturation protein- In the absence of these EEV maturation proteins, infectious (viral) particles are produced but persist within cells and do not migrate to the cell membrane. These EEV maturation proteins assist with the transportation of intracellular viral particles to the cell membrane [18]. (4) According to a study on the vaccinia virus by Byrd et al., the I7L gene's product virion core cysteine protease facilitates the cleavage of the three main viral core proteins' precursors: P4a, P4b, and P25K. Any alteration to the I7L triad or even its enzymatic sites would prevent cysteine protease from cleaving the core protein precursor will continue to exist [19]. (5) Entry/fusion complex component- In the case of the vaccine virus, gene G9R codes for a comprising 340 residual protein. The N-terminal myristoylation, the C-terminal transmembrane region, and the 14 cysteine residues are three structural characteristics that are conserved within the protein in all poxviruses. The protein is involved in the viral replication mechanism and is crucial for establishing the entry-fusion complex [20]. (6) DNA binding protein- The 25 kDa virion protein VP8 is encoded by the gene product of the L4R. VP8 has a strong affinity for binding to single- and double-stranded DNA. It exhibits a slightly higher affinity for ssDNA and is similar to RNA [21]. The L4R protein of the vaccine virus plays various roles, including packing and sustaining transcriptional activation by adhering to the ssDNA and collaborating with the I8R protein to unwind the initial promoters [22]. (7) DNA helicase- A18R's gene product, DNA helicase, is essential for the transcription of viral DNA and DNA repair mechanisms. According to Bayliss et al., this A18R gene product exhibits attraction for the DNA-dependent ATPase activity triggered by both ssDNA and dsDNA or DNA-RNA hybrids but not by any of the DNA [23]. (8) Surface glycoprotein- The primary target of the neutralizing antibodies is a surface glycoprotein, as it enables access to the host organism. Most vaccination tactics and pharmacological therapies focus on the Spike glycoprotein as their central focus due to its significant involvement in viral infection and adaptive immunity [24,25].

By predicting the highly antigenic epitopes from the sequences of the MPXV proteins, we have developed a multiepitope subunit vaccine in this study. Additionally, epitopes were predicted using openly accessible theoretical servers. The vaccine candidate's N-terminal was linked with beta-defensin as an adjuvant to produce a potent immunological response. However, several other factors were examined, including antigenicity, allergenicity, toxicity, and physiochemical factors. The proposed vaccine candidate was tested for interaction, stability, and immunological response via docking and immune-simulation experiments (Fig. 1 ).

Fig. 1.

Schematic representation of workflow used for designing efficient and potential multiepitope subunit vaccine using immunoinformatic tools.

2. Methodology

2.1. Leveraging the entire monkeypox virus proteome for the formulation of a vaccine: protein selection

In this investigation, the monkeypox virus proteins were selected from the reference genome accession number MT903344.1. The protein sequences were computed from the NCBI (https://www.ncbi.nlm.nih.gov/) database in the FASTA format. Following sequence retrieval, each protein's sequence similarity to viral proteins and the human genome was examined via subjecting protein sequence to blastP server (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Proteins were sorted based on the scores of human percentage sequence similarity.

2.2. Assessment of the antigenicity, allergenicity, and toxicity of selected proteins

The proteins' antigenicity, allergenicity, and toxicity were evaluated after retrieving protein sequences from the NCBI database. Antigenic proteins trigger a more robust immune response that offers protection and has a higher affinity for the paratopes of the reactive antibody that elicit a protective immune response against the foreign particle [26].

2.2.1. Antigenicity prediction

VaxiJen is the first and most reliable tool for predicting protective antigens without regard to alignment (alignment-free strategy), i.e., prediction is based on the protein's physiochemical characteristics. It enables the classification of antigens without sequence alignment. An application VaxiJenv2.0, a standalone tool (http://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) for antigen prediction based on auto cross-covariance (ACC) that transforms residues into uniform vector according to the principal characteristic of amino acid to get around the shortcomings of previously known alignment-based methods of antigen prediction. Due to its larger dataset, 100 antigens and 100 non-antigens for each dataset of bacterial, viral, and tumor proteins, the VaxiJen server has a prediction efficiency of 70–89% [27,28].

2.2.2. Allergenicity prediction

Protein allergenicity was predicted to determine whether or not chosen proteins can trigger allergic reactions. Allergenic proteins or peptides can trigger a type I hypersensitivity reaction by releasing Immunoglobulin E (IgE) [29]. The AlgPred (https://webs.iiitd.edu.in/raghava/algpred/submission.html) server was implemented to predict the proteins' allergenicity. The dataset for training the model comprises 578 allergens and 700 non-allergens, and AlgPred uses various techniques to evaluate allergic proteins accurately. The servers use a massive collection of IgE epitopes, a support vector machine based on the residual composition of protein, and motif-based techniques, MEME/MAST, together with a BLAST search against the allergic protein, to predict the allergenicity of the proteins. The performance of all these methods has been assessed by utilizing additional reference datasets of 323 allergens and 101,725 non-allergens from the Swiss-Prot database [30]. The FAO/WHO (Food and Agriculture Organization and World Health Organization) guidelines released in 2003 stipulate that the prediction of allergenic proteins should be based on multiple datasets rather than a single set, supported by the AlgPred server [31].

2.2.3. Toxicity prediction

Humans may be affected in diverse ways by toxic proteins from numerous sources, including plants, animals, or microorganisms. This could result in diseases such as paralysis, cholera, tetanus, cellular toxicity, tissue damage, or toxic shock syndrome. The use of toxic proteins during the design of vaccines can even have adverse effects on vaccine recipients, eventually resulting in the vaccine's failure as intended [32]. In this study, the ToxinPred2 (https://webs.iiitd.edu.in/raghava/toxinpred2/) server was used, tool that helps in the prediction of toxic and non-toxic proteins. ToxinPred2 was formulated to overcome the inadequacies of ToxinPred, which could only anticipate the toxicity of peptides or smaller proteins (sequences of length ≤50 residues). ToxinPred2 can predict the toxicity of proteins of any size, in contradiction to ToxinPred. It makes use of machine learning prediction techniques as well as BLAST similarity search, MERCI (Motif-EmeRging and with Classes-Identification-based motif search), and a wide range of information based on large datasets of 8233 toxic and 8233 non-toxic proteins to train the model for proteins' toxicity prediction [32,33].

2.3. Potential helper T lymphocytes (HTL) epitopes prediction and selection for vaccine designing

Helper T lymphocytes (HTL) are critical players in activating cellular and humoral immunity and identifying infections by attaching to specific HTL epitopes on MHC class II molecules [34]. This study predicted the possible helper T lymphocyte epitope via the Immune Epitope Database and Analysis Resource (IEDB) server. The National Institute of Allergy and Infectious Diseases (NIAID) provides financial assistance for the IEDB server, freely accessible at https://www.iedb.org/. It has a vast dataset of T-cell epitopes, and the epitope prediction relies on experimental data gathered from studies on humans, primates, and other species associated with infectious diseases, transplantation, allergies, and autoimmunity [35]. FASTA-formatted antigenic protein sequences were submitted to the IEDB database, and alleles were chosen based on the 7-allele HLA reference set common to any geographical location.

2.4. Potential CTL epitopes prediction and selection for vaccine engineering

Host cytotoxic T lymphocytes (CTL) are crucial in stimulating the immune system because they can identify particular viral peptides. Because MHC class I molecules bind and present CTL epitopes on T cells' surfaces, CTL epitope prediction is crucial [36]. In this case, the CTL epitopes are predicted via the NetCTL-1.2 webserver (http://www.cbs.dtu.dk/services/NetCTL). The forecast depends on three different factors: the proteasomal cleavage of the proteins, the effectiveness of the transport of the Transporter associated with Antigen Presentation (TAP) molecules, and the binding affinity of the peptides to the MHC class I molecule. The artificial neural network predicts MHC class I binding efficiency and proteasomal cleavage of the proteins, while the weight matrix estimates TAP transport efficiency. The server can predict up to 12 supertypes of MHC class I CTL epitopes, and ANN prediction results in a log-transformed value corresponding to IC50 in nM units [37].

2.5. Potential B-Lymphocytes epitopes prediction and selection for vaccine designing

The B cell receptors (BCR) identify B cell epitopes that could provoke a cellular or humoral immune response [38]. This study predicted the linear B cell epitopes from the antigenic proteins using the web-basedABCpred server. The server employs the ANN to determine the epitope regions that are advantageous when selecting a synthetic vaccine candidate or used to diagnose a disease. The input protein sequence is furnished in the FASTA format with a protein name for identification without any heading or file name. The server's accuracy is 65.93%, and it will ignore any additional characters, including spaces and numbers. The threshold can be set between +0.1 and + 1.0; we have chosen the +0.5 threshold because it exhibits the highest degree of accuracy. However, raising the threshold will improve specificity while lowering sensitivity. The server output can be either graphical or tabular, depending on the user's preference [38,39].

2.6. Epitopes joining via linkers for the development of potential vaccine candidate

With the aid of unique spacer sequences called linkers, all the anticipated potentially antigenic epitopes were integrated to construct a complete vaccine candidate. HTL epitopes were joined using the GPGPG spacer sequence, whereas CTL and B-cell epitopes were merged using the AAY and KK spacer sequences. The study reported that the MHC class II molecule engages the KK linker to carry out proteasomal cleavage for the antigen processing [40] AAY, and the universal linker GPGPG prohibits the establishment of junctional epitopes and improves protein's flexibility [41]. Another linker, EAAAK, a helix-forming linker, was also employed to engage the adjuvant at the N-terminal of the Vaccine construct to increase its immunogenicity. In addition, a pan HLA DR-binding epitope (PADRE) was incorporated just after the adjuvant to facilitate the formulation of immunotherapeutic vaccines. PADRE, an adjuvant agonist, is a universal peptide that aids in activating antigen-specific CD4⁺ T cells [42,43].

2.7. Determination of antigenicity, allergenicity, toxicity, and physiochemical properties of the designed vaccine candidate

The Swiss Institute of Bioinformatics (SIB) and Expasy jointly manage the ProtParam server (protein analysis tool) to compute different physiochemical parameters of the proposed vaccine candidate. The computed physiochemical parameters include the Mol. Wt., half-life, aliphatic index, instability index, theoretical pI, and grand average of hydropathicity index (GRAVY) [44]. This server is freely available at https://www.expasy.org/resources/protparam. Along with this, antigenicity, allergenicity, and toxicity were also predicted via VaxiJen v2.0 (http://www.jenner.ac.uk/VaxiJen), AllerTOP v2.0 (https://www.ddg-pharmfac.net/AllerTOP/), and ToxinPred2 (https://webs.iiitd.edu.in/raghava/toxinpred2/), respectively [41,45].

2.8. Tertiary (3D) structure prediction of the vaccine candidate

The protein 3D structure prediction tool, Robetta (https://robetta.bakerlab.org/), developed by the Baker lab at the University of Washington, was assessed to determine the tertiary structure of the designed vaccine construct after merging all epitopes and predicting their physiochemical properties [46]. The CAMEO (Continuous Automated Model EvaluatiOn) continuously assesses the accuracy and dependability of the anticipated structure [47]. RoseTTAFold, a deep learning-based approach that constantly ranks in the top spot by CAMEO, accurately predicts the construction of numerous protein chains. The input sequence was submitted in the FASTA format [47].

2.9. 3D structure refinement, model quality assessment, and structure validation

The best-predicted model was identified, and the GalaxyRefine web server (http://galaxy.seoklab.org/refine) was used to refine the predicted structure based on the results acquired from the Robetta web server. The server's refining technique has been successfully examined and approved in CASP10. In order to relax the model using a molecular dynamics simulation, the server first rebuilds the side chains and performs side-chain packing. The server's refinement method, as demonstrated by CASP10, performs best at enhancing the local and global quality of the refining model [48,49].

Post-model refinement, the SWISS-MODEL's Structure Assessment tool (https://swissmodel.expasy.org/assess) was used to verify the model's quality and structure. Numerous metrics were examined, including MolProbity [50], and the QMEAN score (https://swissmodel.expasy.org/qmean/), to assure the model quality. The protein model quality was assessed using MolProbity v4.4 at the local (residue level) and global (for the structural framework) levels. A Ramachandran plot was evaluated to analyze the energetically efficient region for the dihedral angles of the backbone for the amino acid residues involved in the model structure. The newly added term, QMEANDisCo, demonstrated the per-residue quality evaluation utilizing distance constraints ensembles and is an extension of QMEAN [51]. Ramachandran plot server (https://zlab.umassmed.edu/bu/rama/) was utilized for structural validation, before and after refinement [52].

2.10. Interaction study between the vaccine and immune cell TLR-3 receptor

The ClusPro docking web server (https://cluspro.bu.edu/) was employed to execute a molecular interaction analysis between immune receptor TLR3 and the designed vaccine construct. Boston University and Stony Brook University collaborate to maintain the server, which is freely accessible for academic research. ClusPro requires two proteins, one as a receptor and the other as a ligand, to provide the optimum docking pose. The ligand is kept movable, the receptor is fixed, and the interaction energy is represented by a correlation function [53]. The server uses the correlation-based Fast Fourier Transform (FFT) PIPER docking software for rigid docking [54]. The server implements the following procedures [1]: FFT-based docking via rigorous conformation sampling [2]; RMSD-based clustering to detect low energy conformations; and [3] model refining utilizing the CHARMM force field to eliminate steric conflicts. Finally, the server algorithm extracts the top 10 models based on the final scores [55].

2.11. Molecular dynamics simulation of designed vaccine and immune cell TLR-3 receptor

WebGRO for macromolecular simulations (https://simlab.uams.edu/) was used to explore the stability of the vaccine candidate and the TLR3 immune receptor at the microscopic level in the simulated physiological setting. WebGRO uses the GROMOS96 43a1 forcefield, the GROMACS software package that calculates both non-bonding and bonding interactions [56]. The SPC water model with 3-site rigid and charges was employed, along with a triclinic solvent box. The solvent box was regulated and neutralized using Na⁺ and Cl⁻ ions employing periodic boundary conditions [57]. Five thousand frames per simulation using the steepest descent method was carried out to minimize energy and to determine whether there were steric clashes between the vaccine design and the TLR3 receptor. Additionally, the macromolecular system was maintained at 300 K at 1 ATM pressure using NPT and NVT ensembles [[58], [59], [60], [61]]. Finally, we computed the structural deviations, Root Mean Square Deviation (RMSD), Root Mean Square Fluctuation (RMSF), Radius of gyration (Rg), and hydrogen bonding via executing 100 ns macromolecular simulation [62].

2.12. Immune simulation of the designed monkeypox virus vaccine for generation of immune response

Immunological dynamics simulation was used to virtually assess the immune response developed by the designed multiepitope vaccine. Utilizing the online C-ImmSim webserver developed by Rapin et al. [63], it was possible to execute this in-silico immune response experiment (https://www.iac.cnr.it/∼filippo/projects/c-immsim-online.html). The more sophisticated version of C-ImmSim is consistent with the earlier iteration and uses amino acid strings instead of bit strings. The server uses the vaccine construct's sequence as an input parameter. Although some vaccines require a 2, 3, or 6-month gap between the first and second immunizations or booster doses, research suggests that a one-month dosing interval is best for most vaccines [59]. In order to explore the dose-dependent immunological response, we have decided on the one-month time frame based on the published literature for our vaccine candidate. The repeated exposure of X following the vaccination dose was employed to examine the immune response elicited by the vaccine candidate in response to the encroaching pathogen. Thus, a single dose of an X vaccine molecule was administered every time, and a simulation was carried out for approximately X days. The produced immunological response was then evaluated using the Simpson index (D).

3. Results and discussion

3.1. Screening and sorting of immunogenic protein and their sequence retrieval from the NCBI database

Monkeypox virus 2022 outbreak isolate MPXV-UK_P2 (accession MT903344.1) was selected for protein sequence retrieval. The NCBI database obtained the whole genome sequence and coding sequence [64]. Total ten MPXV proteomes were retrieved from GenBank: Chemokine binding protein (Cop-C23L) [AccessionURF91554.1], TNF receptor (Cop-C22L) [Accession QNP13601.1], Ankyrin (Cop-C19L) [Accession QNP13602.1], Secreted EGF-like protein (Cop-C11R) [Accession QNP13604.1], EEV maturation protein (Cop-F12L) [Accession QNP13637.1], Virion core cysteine protease (Cop-I7L) [Accession QNP13663.1], Entry-fusion complex component (Cop-G9R) [Accession QNP13674.1], DNA binding protein (Cop-L4R) [Accession QNP13678.1], DNA helicase (Cop-A18R) [Accession QNP13725.1], and Surface glycoprotein (B21R) [Accession QNP13774.1]. Before shortening MPXV proteins, NCBI blastP (protein-protein BLAST) was used to compare their sequence similarity to human proteomes (taxid:9606). TNF receptor and Secreted EGF-like protein were determined to have sequence similarities of 45.70% and 48.98%, respectively, and were thus removed from further analysis since these proteins have more than 45% sequence similarity. Except for ankyrin, which shares 24.65% of its sequence with the human proteome, the remaining proteins do not significantly resemble human proteins. The blastP program is again employed to determine the similarity between these proteins and other poxviruses. Interestingly, all MPXV proteins share at least 90% of their sequence with the cowpox virus (Supplementary Table 1).

3.2. Antigenicity, allergenicity, and toxicity prediction

After shorting and selecting potential antigenic protein from the MPXV, antigenicity, allergenicity, and toxicity were predicted to cross-check the vaccine candidate's properties. Here, ANTIGENpro and VaxiJen v2.0 were employed as two separate servers for antigenicity prediction. The rationale behind employing two distinct servers is that one protein (Surface glycoprotein) has 1880 amino acid residues, making it challenging to predict antigenicity via ANTIGENpro due to the server's sequence submission limitation. As a result, VaxiJen, another highly trustworthy server, was utilized to forecast antigenicity. All proteins were predicted antigenic via ANTIGENpro servers; the same antigenicity was predicted for all proteins by VaxiJen v2.0, all proteins were antegenic except for DNA helicase, which shows non-antigenic behavior (Supplementary Table 2).

The protein's allergenicity was predicted using AlgPred servers using an amino acid-based SVM module prediction technique with a cut-off of −0.4. DNA helicase, EEV maturation protein, and Entry/fusion complex components are non-allergens. However, it is anticipated by the server that surface glycoprotein, ss/dsDNA binding protein, ankyrin, virion core cysteine protease, chemokine binding protein, and ankyrin are allergens. Since the threshold was set very high, most proteins are expected to be predicted as an allergen. AlgPred predicts allergenicity values and positive and negative prediction scores using input from the FASTA sequence. (Supplementary Table 2). In addition to their allergenic behavior, these proteins are included in our study. According to the immunoinformatics approach, only antigenic peptides known as epitopes are crucial for developing vaccine constructs; the entire protein is no longer necessary. The chosen epitopes should not trigger allergic reactions.

Using the Hybrid (RF + BLAST + MERCI) machine learning algorithm and the threshold value of 0.6, ToxinPred2 effectively predicts toxicity. Except for ankyrin, determined to be toxic, the ToxinPred2 server demonstrated that all MPXV proteins were non-toxic. Supplementary Table 2 lists all the predicted values, including the ML, MERCI, BLAST, and hybrid scores.

These analyses revealed the high antigenicity of each selected protein, for whom the sequence may be exploited to make a vaccine for the MPXV.

3.3. Selection of potential epitope to ameliorate the immune response

To strengthen the protective immune response, plausible epitope prediction is necessary. The most effective and trustworthy web servers were trained to predict all the epitopes based on the data gathered from the experiments and their validation. While predicting HTL, CTL, and B cell epitopes, those epitopes should be selected that can stimulate CD4⁺ T cells, CD8⁺ T cells, and B cells, respectively. Protective immunological responses depend heavily on activating CD4⁺ T cells, CD8⁺ T cells, and B cells [65,66].

3.3.1. HTL epitopes prediction

The IEDB tool predicts 12mer antigenic HTL epitopes for each relevant MPXV protein. The two most significant factors, percentile rank and IC₅₀ value, were used to select potential antigenic peptides. The IEDB server's IC₅₀ score evaluates the peptides' affinity for MHC class II molecules. The MHC class II molecule seems to have the maximum binding affinity when the IC₅₀ value is less than 50 nM, the midpoint interaction when the IC₅₀ value is less than 500 nM, and the worst binding affinity whenever the IC₅₀ value is less than 5000 nM. The percentile and IC₅₀ scores were inversely related [67]. All of the selected HTL epitopes have SMM-aligned IC₅₀ values of less than 95 nM and percentile scores of less than 0.7, whereas the chemokine binding protein has an IC₅₀ value of 148 nM and a percentile score of 1.1 (Table 1 ).

Table 1.

Potential antigenic HTL epitope prediction via IEDB web server and their characteristics.

S. No.	Proteins	AA Seq	Peptide	Position	Percentile Rank	SMM align IC50 (nM)	Allele	Region
1	Chemokine binding protein (Cop-C23L)	246	DVIIKVTKQDQT	40–51	1.1	148	DRB4*01:01	North America
2	Ankyrin (Cop-C19L)	588	LEAIHSDKRISF	505–516	0.25	70	DRB1*03:01	Europe, Western Asia
3	EEV maturation protein (Cop-F12L)	635	IKIIYDLNAVTT	551–556	0.51	72	DRB1*03:01	Europe, Western Asia
4	Virion core cysteine protease (Cop-I7L)	423	TEFHHYNNFYFY	265–276	0.69	85	DRB1*15:01	Europe, Oceania, South-East Asia
5	Entry/fusion complex component,	340	EYVHIGPLTKDK	41–52	0.09	23	DRB5*01:01	North America
6	ss/dsDNA binding protein (VP8) (Cop-L4R)	251	STSMRLNAIYGF	188–199	0.38	94	DRB3*01:01	North America
7	DNA helicase, transcript release factor (Cop-A18R)	492	SPDVLIVVSRHL	176–187	0.47	21	DRB1*07:01	Europe, South-East Asia, North Africa
8	Surface glycoprotein; B21R	1880	LLVIILILAIYI	1829–1840	0.08	8	DRB1*15:01	Europe, Oceania, South-East Asia

3.3.2. Prediction of CTL epitope

The CTL epitopes in the above protein sequences were predicted using the NetCTL-1.2 web server using the default values of predefined threshold (0.75), TAP transport efficiency (0.05), and C terminal cleavage (0.15). The most widely dispersed supertypes, A2, A3, and B7, were considered, and epitopes were predicted for each supertype separately. The selection of these three supertypes was supported because they represent about 88% of the global population [59,68]. Finally, based on the server's scores, 24 CTL epitopes were selected from the predicted epitopes: 8 for the A2 supertype, 8 for the A3 supertype, and 8 for the B7 supertype (Table 2 ).

Table 2.

CTL epitope prediction via NETCTL-1.2

S. No.	Proteins	A2	Score	A3	Score	B7	Score
1	Chemokine binding protein (Cop-C23L)	IVLACMCLV	1.0105	ISHKKVSYK	1.5494	SPAITREEA	1.0464
2	Ankyrin (Cop-C19L)	ISDTDLYTV	0.7693	MLYGKNHYK	1.7479	NTRFNPSVL	0.9279
3	EEV maturation protein (Cop-F12L)	MMNITILEV	1.3326	GLACYRFVK	1.5494	RPEIDVLPF	1.5619
4	Virion core cysteine protease (Cop-I7L)	FLADKKMTL	1.3830	SLFMILCTR	1.1266	LPTSIPLAY	1.0963
5	Entry/fusion complex component (Cop-G9R)	VIWILIVAI	0.9540	KLHLISLLS	0.8405	VPVNRAKVV	1.5207
6	ss/dsDNA binding protein (Cop-L4R)	SMLSIFNIV	1.3355	NIVPRTMSK	1.2994	SVSTKYTPI	0.7240
7	DNA helicase (Cop-A18R)	SLLKMEYNL	1.1020	REHMVFFYK	1.2415	VEVEPGSSF	0.4656
8	Surface glycoprotein; B21R	KMADIQTRI	1.3293	RTYSAMTIK	1.5986	SPLTRKGAT	1.3516

3.3.3. B cell epitope prediction

Using the ABCpred web application, B cell epitopes were predicted just at the standard threshold (0.51) at 10 mer window range for epitope prediction using the overlapping filter for each of the antigenic proteins from MPXV. Based on the server's anticipated scores and ranking, eight B cell epitopes (one from every protein) were selected (Table 3 ). The highest score represents the epitopes' highest binding affinities [69].

Table 3.

B cell epitope prediction via ABCpred server.

S.No.	Proteins	B-cell epitope	Start position	Score	VaxiJen Antigenicity	Allergenicity
1	Chemokine binding protein (Cop-C23L)	CTEEENKHHM	28	0.73	0.6667 (ANTIGEN)	Non-Allergen
2	Ankyrin (Cop-C19L)	DEDGLTSLHY	173	0.79	0.6942 (ANTIGEN)	Non-Allergen
3	EEV maturation protein (Cop-F12L)	TSYYPLIDII	96	0.82	0.5312 (ANTIGEN)	Non-Allergen
4	Virion core cysteine protease (Cop-I7L)	DLVISKIPEL	6	0.77	0.5288 (ANTIGEN)	Non-Allergen
5	Entry/fusion complex component (Cop-G9R)	PAKLLEYVHI	36	0.74	0.7974 (ANTIGEN)	Non-Allergen
6	ss/dsDNA binding protein (Cop-L4R)	NNLGLGDDKL	78	0.79	1.4054 (ANTIGEN)	Non-Allergen
7	DNA helicase (Cop-A18R)	QSPDVLIVVS	175	0.77	0.7280 (ANTIGEN)	Non-Allergen
8	Surface glycoprotein; B21R	MDMKLFDHAK	421	0.85	0.6838 ANTIGEN)	Non-Allergen

3.4. Merging of potential immunogenic epitopes, adjuvant, and PADRE via specific linkers

Before combining the epitopes, it is crucial to ensure that perhaps the correct epitopes were identified and selected. So, we predicted the antigenicity, allergenicity, and toxicity for all 40 epitopes, such as the finalized 8 B cell epitopes from the ABCpred server, 24 cytotoxic T lymphocyte epitopes picked from the NETCTL-1.2 server, and 8 helper T lymphocyte epitopes from the IEDB server. It is fascinating to note that every epitope was highly antigenic, non-allergenic, and non-toxic, as determined by the server's predictions of their antigenicity, allergenicity, and toxicity. Consequently, after evaluating all these characteristics, epitopes were ultimately selected to develop the novel vaccine construct (Supplementary Tables 3A, 3B, 3C).

After ensuring that selected epitopes are antigenic, non-allergenic, and non-toxic, we further move with epitope joining. Linkers have already been thoroughly studied to combine all the epitopes into a single vaccine construct. Linkers also assist in limiting the mixing of epitopes, which is critical in multiepitope vaccines, where each epitope must have the ability to elicit potential immune responses [69]. AAY, KK, and GPGPG linkers were employed for HTL, CTL, and B cell epitope joining. Additionally, to increase the immunogenicity of the vaccine construct, β -defensin (GIINTLQKYYCRVRGGRCAVLSCLPKEEQIGKCSTRGRKCCRRKK) TLR-3 agonist; a 45-mer adjuvant, was introduced with the aid of a specific EAAAK linker at the N-terminal of the construct [70,71]. The same EAAAK linker was used to attach a 13-mer PADRE sequence (AKFVAAWTLKAAA) immediately post-adjuvant, which aids in activating an antigen-specific CD4⁺ protective immunological response [69].

Finally, employing combinations of CTL, HTL, and B cell epitopes, six distinct vaccine constructs were designed; however, the adjuvant was positioned before the PADRE at the N-terminal each time. The following combinations were employed to create the intensely antigenic vaccine constructs.

• [1]adjuvant-PADRE-HTL-CTL-B-Cell epitope

• [2]adjuvant-PADRE-CTL-HTL- B-cell epitope

• [3]adjuvant-PADRE-B-cell-HTL-CTL epitope

• [4]adjuvant-PADRE-B-cell-CTL-HTL epitope

• [5]adjuvant-PADRE-CTL-B-cell-HTL epitope

• [6]adjuvant-PADRE-HTL-B-cell-CTL epitope

Six independent vaccine constructs were produced after combining the epitopes, adjuvant, and PADRE to create a single immunogenic vaccine candidate. At last, the homology assessment between the constructed vaccine and the host was performed, and no significant similarity was found except for the β -defensin (sequence from 1 to 45) shows 100% sequence similarity with the Homo sapiens beta-defensin 3, which was used as an adjuvant.

Prior to predicting toxicity, antigenicity, and allergenicity, we predicted the antigenic conformational B-cell epitope using the SEMA tool, which further employs a deep transfer learning approach (Supplementary Text) [72].

3.5. Toxicity, antigenicity, and allergenicity prediction of potential engineered vaccine

Antigenicity, allergenicity, and toxicity were predicted to select the best antigenic and highly immunogenic construct from the vaccine candidate combinations. As the vaccine constructs are non-allergen and non-toxic, antigenicity scores for all the six combinations of vaccine construct were predicted via VaxiJen v2.0. Based on the predicted antigenicity score, vaccine construct number 4 (adjuvant-PADRE-B-cell-CTL-HTL epitope) was selected, which shows the highest antigenicity score compared to the other vaccine constructs (Supplementary Table 4).

3.6. Physiochemical characterization of designed protein vaccine constructs

Any vaccine construct's physiochemical characteristics are made up of a variety of physical and chemical parameters, and in this study, these variables were predicted using the ProtParam server. Our specified vaccine construct has an instability index of 36.76; some sources claim it should be under 40. This suggests that the selected vaccine design has a stable natural state. The anticipated half-life of each vaccine design was 30 h for mammalian cell culture, over 20 h for yeast and in vivo, and even more than 10 h for E. coli (in vivo) systems. This statistic illustrates how long it takes for a protein to degrade after it has formed. For all of the developed vaccine candidate combinations, the molecular weight ranges from 64304 to 64413 Da, approximately 64 kDa. The predicted pI of the selected vaccine candidate was 9.49, although the aliphatic index is higher for all of them, demonstrating the excellent stability of the protein side chains. The GRAVY (grand average of hydropathicity) index of 0.079 is determined by summing the hydrophobicity of each residue and dividing it by the number of amino acids. GRAVY represents the hydrophobicity value of the protein chosen for the vaccine candidate (Table 4 ). According to information from sections 3.5, 3.6 represents highly antigenic, extremely stable, probably non-allergenic, and non-toxic. It is also unlikely to release any toxic components once administered. Finally, the vaccine construct's secondary structure and solvent accessibility were predicted using the PredictProtein web server (https://predictprotein.org/) (Supplementary Fig. 2).

Table 4.

Prediction of physiochemical properties of the vaccine constructs by ProtParam server.

S. No.	Vaccine construct Sequence	Mol. wt.	Theoretical pI	Estimated half-life of the vaccine construct	Instability index	Aliphatic index	GRAVY
1	Adjuvant-PADRE-HTL-CTL-B-Cell	64304	9.52	30 h (mammalian reticulocytes, in vitro).	37.05	95.51	0.074
				>20 h (yeast, in vivo).
				>10 h (Escherichia coli, in vivo).
2	Adjuvant-PADRE-CTL-HTL- B-cell	64364	9.53	30 h (mammalian reticulocytes, in vitro).	36.05	94.84	0.062
				>20 h (yeast, in vivo).
				>10 h (Escherichia coli, in vivo).
3	Adjuvant-PADRE-B-cell-HTL-CTL	64413	9.49	30 h (mammalian reticulocytes, in vitro).	36.76	95.02	0.079
				>20 h (yeast, in vivo).
				>10 h (Escherichia coli, in vivo).
4	Adjuvant-PADRE-B-cell-CTL-HTL	64413	9.49	30 h (mammalian reticulocytes, in vitro).	36.76	95.02	0.079
				>20 h (yeast, in vivo).
				>10 h (Escherichia coli, in vivo).
5	Adjuvant-PADRE-CTL-B-cell-HTL	64364	9.53	30 h (mammalian reticulocytes, in vitro).	36.66	94.84	0.062
				>20 h (yeast, in vivo).
				>10 h (Escherichia coli, in vivo).
6	Adjuvant-PADRE-HTL-B-cell-CTL	64304	9.52	30 h (mammalian reticulocytes, in vitro).	36.90	95.51	0.074
				>20 h (yeast, in vivo).
				>10 h (Escherichia coli, in vivo).

3.7. Tertiary (3D) structure prediction of vaccine construct

The Robetta online tool predicts the 3D structure using the vaccine candidate's protein sequence. The web server generates five models' predictions. Based on the lowest estimated angstroms error for each residue, model 1 was chosen as all predicted models have similar confidence scores of 0.42. The range of the confidence score, which indicates the quality of the model, is from 1.0 to 0.0. Considering that the sequence presented here is a vaccine construct, the study's confidence score of 0.42 appears to be good. The results of template-free modeling are consistently low, as no templates are available for modeling. The protein's three-dimensional structure was ascertained using the RoseTTAFold modeling technique. Without such software, it was impossible to predict the tertiary structure of protein within minutes on a single high-end laptop. This could take a year to identify the structure of a protein using conventional laboratory techniques. The RoseTTAFold method was developed in the Bakers lab by Beak et al., [47].

3.8. Structure refinement and validation of modeled tertiary structure

The protein's tertiary structure is refined by GalaxyWeb (a protein structure refinement server), which creates the refined and compact structure using the 3D-modeled pdb structure acquired via the Robetta protein structure modeling server. Model 3 was adopted from the list of models due to its lowest clashing score and the most extraordinary Ramachandran-favored region. The selected model's structural details are provided in Fig. 2 .

Fig. 2.

(A) 3D structure of modeled vaccine candidate; Red color represents adjuvant, Green color shows the multiepitopes, whereas Blue color denotes the PADRE sequences. (B) and (C) represents the Ramachandran plot of unrefined and refined structures, simultaneously. Highly Preferred Conformations are shown in Black, Dark Grey, Grey, and Light Grey (Delta ≥ −2). Preferential conformations are shown by white with a black grid (−2 > Delta ≥ −4), and White with a grey grid suggests untrustworthy conformations (Delta < −4). Highly preferred observations are represented by GREEN crosses, preferred observations by BROWN triangles, and questionable observations by RED circles.

The refined model was submitted to the structure assessment tool of SWISS-MODEL for structural validation. The Ramachandran favored region before refinement and after refinement was 95.90% and 97.85%, while the overall QMEANDisCo score and MolProbity estimations were 0.35 ± 0.05 and 3.19, respectively. The Z-score was −1.86, which indicates the model is good. The QMEAN Z-score indicates how far the model deviates from the mean model. A score of 0.0 denoted a structure that seems similar to the native structure, while −4.0 denotes a poorer model quality (Supplementary Fig. 2).

3.9. Docking of the TLR3 immune receptor with the modeled vaccine construct

The immune response is triggered when an antigenic molecule interacts with the immune receptor (TLR). The interaction between the molecules of the proposed vaccine candidate and the immunological receptor TLR3 was examined by ClusPro protein-protein docking online tool. The server generates the top 29 docked complexes (Supplementary Table 5) based on the PIPER algorithm's docking scores after designating the TLR3 as a receptor and the vaccine construct as a ligand. PIPER uses the following formula to calculate the energy between two interacting proteins:

E = 0.40E_rep ± 0.40E_att + 600E_elec + 1.00E_DARS

Whereas,

E_rep = repulsive energy contribution.

E_att = attractive energy contribution.

E_elec = electrostatic energy contribution.

E_DARS = energy contribution by paired structure-based potential created Decoys as the Reference State (DARS)

Among all the docked complexes, Model 0 had the lowest energy at its maximum, −1303 kcal mol⁻¹. Hence, this complex was selected for further studies (Fig. 3 ).

Fig. 3.

Representation of molecular docking: (A) protein-protein docking between the Vaccine construct (shown in blue) and immune receptor TLR 3 (green color) whereas, red represents the interacting residues; (B) Stick model representation of bonding residues between vaccine (blue) and TLR 3 (red), dotted lines in yellow shows interaction between them.

3.10. Molecular dynamics (MD) simulation of the docked complex

Molecular dynamics simulation was run for 100 ns to examine the vaccine construct and TLR3 immune receptor (docked complex) behavior under simulated physiological conditions. MD simulation was performed to confirm the interaction and stability of the vaccine construct and immune receptor (TLR3) in the virtual physiological condition. The molecular stability of the involved residues was analyzed at the molecular level using WebGRo for molecular simulation, a GROMACS software that employs GROMOS forcefield. During the graphing of the data generated by this tool, RMSD, RMSF, Rg, and hydrogen bonding between the vaccine construct, and TLR3 were analyzed. It is essential to evaluate structural changes and the equilibrium state of the construct. The structural deviation regarding the initial structure was checked using the RMSD method. The complex's RMSD first increases until 50 ns. Still, after that, it reaches equilibrium, stabilizing the complex, with an average RMSD score of 0.92 nm (Fig. 4 A). Similarly, the RMSF of a biomolecule describes the fluctuation of every residue that contributes to the formation of the complex [73]. Fig. 4B depicts the residual instability of the vaccine construct, which is represented in blue, whereas the fluctuation of residues of the TLR3 receptor is indicated in red. The average residual fluctuation of vaccine construct and immune receptor TLR3 was 0.37 nm and 0.24 nm, respectively, within the acceptable range (values closer to the ideal value, i.e., 3.4 Å). Both molecules exhibit typical residual fluctuation, whether vaccine constructs or TLR3 [74]. The radius of gyration (Rg) measures how atoms and molecules are distributed and are frequently used to gauge how compact a protein structure is. A lower Rg value indicates a tightly packed protein structure. The graph illustrates how the complex compacted during the dynamics period, and the average Rg score attained was 3.40 nm (Fig. 4C) [75]. Finally, the molecular dynamics simulation data was used to examine the breaking and formation of hydrogen bonds, which is the main stabilizing interaction. During the dynamical period, there were an average of 916.77 hydrogen bonds (Fig. 4D).

Fig. 4.

Results of Molecular dynamics simulations: (A) graphical representation of Root-mean square deviation, (B) Shows the residual fluctuations of vaccine candidate (blue color) and the TLR 3 (represented by red).

3.11. Immune simulation of vaccine via C-ImmSim server to check immune response

The C-ImmSim server was utilized to evaluate the potential immunological response that the vaccine candidate might induce [63]. With the immune simulation analysis, we have shown that the designed vaccine candidate would be able to generate the potential immune response against the pathogen. In this investigation, the potential immune response generated by the monkeypox virus vaccine candidate was seen, along with progression in secondary and tertiary immune responses. Generation of Immunoglobulin G (IgG1 + IgG2) and Immunoglobulin M (IgM) helped to determine the progression of secondary and tertiary immune responses. Interestingly, proliferative immune responses were also seen as the antigen count was reduced (Fig. 5 A). IFN‐γ concentration was maintained until the 15th day, after which it started to decrease steadily (Fig. 5B). IFN‐γ has mainly generated after NK and T cell activation and regulates the immune response against viruses and bacteria. Additionally, it improves macrophage activation and antigen presentation [76]. Following immunization, the active B cell population appears to be relatively high due to a significant immunological response (Fig. 5C). The number of cytotoxic T lymphocytes (Tc), as well as helper T lymphocyte (Th), increases in a similar manner, and the number of Tc cells, continue to rise with time (Fig. 5D & E).

Fig. 5.

Immune simulation: (A) Generation of IgG1 + IgG2 and IgM, shows the progression of secondary and tertiary immune responses; (B) represents higher IFN‐γ concentration and maintained till 15th day; (C) active B cell population, (D) cytotoxic T lymphocytes and (E) helper T lymphocyte (Th) population increases in a similar manner represents the higher immune response generated by the designed vaccine.

4. Conclusion

Since this is the first time that MPXV cases have been discovered outside of the endemic countries in significant numbers, the sudden rise in instances of the disease has sparked a global alarm. Before the 2022 epidemic, MPXV cases were primarily found in a small number of Central and African nations, including the DRC, Nigeria, Cameroon, and Central Africa. Patients suffering from monkeypox infection are currently administered JYNNEOS and ACAM2000, two live smallpox vaccinations. The virulence upon mutation of the live vaccination is reversed, which is a significant drawback. With this concern, we have designed the multiepitope subunit vaccine using efficient immunoinformatic tools as a translational vaccinomics. The study starts with the selection of antigenic proteins and homology search against human proteomes. Epitopes were predicted via ABCpred, IEDB, and NetCTL 1.2 web servers and only the high-scoring antigenic, non-allergen, and non-toxic epitopes were selected. Epitopes were linked via AAY, KK, and GPGPG linkers, whereas adjuvant and a pan-HLA DR sequence are attached at the N-terminal via EAAAK linker. Molecular dynamics results showed that the engineered vaccine candidate is stable with an average RMSD of 0.92 nm, RMSF of vaccine construct of 0.37 nm, and TLR3 was 0.24 nm. The designed vaccine candidate ultimately produces IgG (IgG1 + IgG2) and IgM, which trigger secondary and tertiary immune responses. The design of the vaccine candidate can induce neutralizing immunity against the monkeypox virus, as shown by the production of B cells, helper T cells, and cytotoxic T cell populations, as well as an IFN‐γ response.

Author contribution

Satyendra Singh: Conceptualization, Data curation, Software analysis, Validation, Writing an original draft. Abhishek Rao: Validation, Writing an original draft. Ketan Kumar: Validation, Writing an original draft. Amit Mishra: Validation, Writing an original draft. Vijay Kumar Prajapati: Conceptualization, supervision, writing the original draft, and finalizing the draft.

Declaration of competing interest

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

Appendix A. Supplementary data

The following is the Supplementary data to this article.