https://orcid.org/0000-0002-8068-4144
Figures
Abstract
Liver cancer is the sixth most frequent malignancy and the fourth major cause of deaths worldwide. The current treatments are only effective in early stages of cancer. To overcome the therapeutic challenges and exploration of immunotherapeutic options, broad spectral therapeutic vaccines could have significant impact. Based on immunoinformatic and integrated machine learning tools, we predicted the potential therapeutic vaccine candidates of liver cancer. In this study, machine learning and MD simulation-based approach are effectively used to design T-cell epitopes that aid the immune system against liver cancer. Antigenicity, molecular weight, subcellular localization and expression site predictions were used to shortlist liver cancer associated proteins including AMBP, CFB, CDHR5, VTN, APOBR, AFP, SERPINA1 and APOE. We predicted CD8+ T-cell epitopes of these proteins containing LGEGATEAE, LLYIGKDRK, EDIGTEADV, QVDAAMAGR, HLEARKKSK, HLCIRHEMT, LKLSKAVHK, EQGRVRAAT and CD4+ T-cell epitopes of VLGEGATEA, WVTKQLNEI, VEEDTKVNS, FTRINCQGK, WGILGREEA, LQDGEKIMS, VKFNKPFVF, VRAATVGSL. We observed the substantial physicochemical properties of these epitopes with a significant binding affinity with MHC molecules. A polyvalent construct of these epitopes was designed using suitable linkers and adjuvant indicated significant binding energy (>-10.5 kcal/mol) with MHC class-I and II molecule. Based on in silico cloning, we found the considerable compatibility of this polyvalent construct with the E. coli expression system and the efficiency of its translation in host. The system-level and machine learning based cross validations showed the possible effect of these T-cell epitopes as potential vaccine candidates for the treatment of liver cancer.
Citation: Zafar S, Bai Y, Muhammad SA, Guo J, Khurram H, Zafar S, et al. (2025) Molecular dynamics simulation based prediction of T-cell epitopes for the production of effector molecules for liver cancer immunotherapy. PLoS ONE 20(1): e0309049. https://doi.org/10.1371/journal.pone.0309049
Editor: Sheikh Arslan Sehgal, Cholistan University of Veterinary and Animal Sciences, PAKISTAN
Received: May 22, 2024; Accepted: August 5, 2024; Published: January 3, 2025
Copyright: © 2025 Zafar et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting information files.
Funding: The author(s) received no specific funding for this work.
Competing interests: NO authors have competing interests.
1. Introduction
Liver cancer (LC) is one of the major cause of deaths throughout the world. It is a progressive, multifactorial and prevalent disease accounts for almost 1 million cases around the globe recorded in 2020 [1] indicating nearly 90% occurrence of all liver cancer [2]. Unfortunately, it has been observed that LC is extremely non-effective to standard cytotoxic chemotherapies [3]. Even today, due to less understanding of pathophysiological mechanism, classical diagnostic and therapeutic options, we are facing challenges to cure liver cancer [4]. To overcome this situation, we need to develop more safer, effective and targeted therapeutic strategies. Recent studies found a sub-class of cells with stem cell properties that are involved in the initiation of the tumor makes the tumor persistence, ability to relapse, metastasis, radio resistance and chemoresistance [5]. This causes the tumor cells to withstand with both traditional and innovative therapeutic regimens like Sorafenib [6].
It has been reported that adaptive immunity cannot be effective enough to kill cancer cells, leading the cancer cells to become resistant to the immune system. Initially, the immune system brings about a strong immune response. This response causes some cancer cells to eradicate, while others with low immunogenicity escape detection and proliferate [7]. Macrophages, dendritic cells, natural killer (NK) cells are key innate immune effector molecules in cancer patients [8]. Tumor cells are modified cells always attempting to evade the immune system, which results in a reduction in the expression of antigen-presenting receptors of the Major Histocompatibility Complex class I (MHC-I). Due to the lack of self-recognition, these MHC-I knocked-down cells fell prey to NK cells, which kill them [9]. Dendritic cells serve as a link between the innate and adaptive immune systems. Distinctive to innate immunity, which responds immediately, adaptive immunity takes a few days to respond [10]. T cells only recognize and digest these antigens when they are dispensed by specialized innate immune cells [11]. B cells are more notable because they secrete antibodies (humoral immunity) that exert anticancer functions through activating complement, increasing cellular cytotoxicity, and arbitrating cancer cell phagocytosis [12].
One of the immunotherapeutic approaches to cure cancer is using therapeutic cancer vaccines [13]. The aim behind availing these vaccines for the treatment of LC is to identify tumor associated antigens (TAA) and tumor specific antigens (TSA) as potential vaccine candidates [11]. Computational approaches are utilized to predict and map new vaccine peptides using in-silico analysis [14]. In this regard, epitope-based vaccines have been made possible by improvements in epitope prediction for the preparation of these preparations [15]. Compared to conventional vaccines, epitope-based vaccines provide a number of benefits, such as the capacity to target certain antigens and the potential to elicit a higher immune response. Epitopes are specifically functional in preventing diseases, developing vaccines, detecting and treating diseases that is why clinical and biological researchers are interested in them [16]. To develop a significant liver cancer vaccine, scientists are now employing computational methods to anticipate B and T cell epitopes.
The recent approaches involve the expertise in the field of bioinformatics, genetics, and immunology. Instead of isolation, culturing and studying the cells and tissues, researchers are identifying promising vaccine candidates by computational techniques and data mining. Utilizing only specific antigenic proteins, these techniques not only save time and resources but has the potential to produce safer vaccines. Here are some already approved reverse vaccinology-based vaccines include COVID-19 vaccine against virus [17], Human papillomavirus (HPV) vaccine (Cervarix, Gardasil) [18], Shingles vaccine (Shingrix) for the treatment of secondary varicella infection [19], Meningococcal B vaccine to provide the protection against bacterial meningitis [20].
T-cell epitopes are more significant in developing immunotherapeutic vaccines as they have the potential to provoke other components of immune system. These vaccines are reported more effective against various cancers in pre-clinical and clinical trials [21]. To achieve broad spectral therapeutic effect, preferably T-cell epitopes are designed from conserved regions of tumor associated or tumor specific proteins, as cancer cells find it difficult to escape T-cell mediated immunity [22]. It has been observed that T-lymphocytes provide more robust cell mediated immunity compared to B-lymphocytes that are capable of producing humoral immunity [22, 23].
In this study, we used consolidative and system-level framework to identify the potential liver cancer vaccine candidates involving data mining, antigenicity predictions, genomic conservations, protein-protein interaction studies, molecular interaction and binding affinity analysis, and machine learning and MD simulation based cross-validation techniques. Prediction of T- cell epitopes for the production of effector molecules to cure liver cancer opens the door for the development of novel and efficient therapeutic vaccine that may greatly improve the prognosis and treatment options (Fig 1).
T-cell epitopes provoke and enhance the action of immune response to eradicate the tumor cells.
2. Materials and methods
2.1 Ethical statement
There is no human and animal involvement. The ethical statement is not applicable to this study. Our system level framework involves a variety of immunoinformatic and machine learning based analysis, data mining, software and tools to predict the T-cell epitopes of liver cancer (S1 Table).
2.2 Retrieval and screening of proteins
Human Protein Atlas (HPA) [24] (https://www.proteinatlas.org/), UniProt (https://www.uniprot.org/), NCBI (https://www.ncbi.nlm.nih.gov/), and Liverome databases [25] (http://liverome.kobic.re.kr/) were used to find liver cancer associated proteins. The FASTA format of these protein sequences were retrieved followed by the screening based on antigenicity, molecular weight, and subcellular localization. The molecular weight was determined by the ExPASY server [26] (https://web.expasy.org/compute_pi/) and the antigenicity was predicted using the VaxiJen tool [27] (http://www.ddg-pharmfac.net/vaxijen/). We determined the Sub cellular localization of these proteins was determined by CELLO (http://cello.life.nctu.edu.tw/) and UniProt servers [28]. To predict and screen the proteins based on their optimal expression sites, preferably liver tissues, a GenomicScape server was applied [29] (http://genomicscape.com/).
2.3 Identifying tumor-associated and tumor-specific proteins
Selected proteins were further categorized as tumor-specific antigens (TSA) and tumor-associated antigens (TAA). TAA were located on all normal and cancerous cells of the body but over expressed in tumor cells while TSA are located on cancerous cells only and are specific to these cells. We used TRON Cell Line Portal [30] (http://celllines.tron-mainz.de/) and Tumor-Specific Neo-Antigen Database (TSNAsb) (http://biopharm.zju.edu.cn/tsnadb/browse/) to identify the TAAs and TSAs [31]. These servers comprehensively elaborate the neoantigens indicating specific tumor antigen and a potential target of a particular type of cancer immunotherapy.
2.4 Protein-Protein Interaction (PPI) and network analysis
The purpose of the PPI network is to overview the interactions between the target proteins and the host proteins presenting the overview of the directly and indirectly associated proteins in cancer. Human Annotated Protein-Protein Interactions (HAPPI) database (http://discovery.informatics.uab.edu/HAPPI/) was used to find all the proteins that interact with our target (selected) protein [32]. The network was constructed by using Cytoscape (https://cytoscape.org/) software version 3.10.1. In this network, nodes represented the proteins and edges showed the interaction [33]. Cancer associated and TSAs/TAAs nodes were differentiated and highlighted by different colors.
2.5 Prediction of T cell epitopes, conservation and antigenicity analysis
The software HLAPred (http://crdd.osdd.net/raghava/hlapred/) was considered for designing T-cell epitopes [34]. FASTA sequences were uploaded to predict MHC Class I and II epitopes, population coverage and multiallelic epitopes were analyzed. The UCSC Xena server [35] (https://xena.ucsc.edu/) was employed for epitope conservation to analyze the broad spectral therapeutic effect of these proteins. Epitopes were screened on the basis of antigenicity, immunogenicity [26] by using the VaxiJen server [27] and Immune Epitope Database (IEDB) [36] (https://www.iedb.org/) respectively.
2.6 Development of polyvalent construct
To determine the sequence of epitopes in the final polyvalent construct, all of the screened epitopes were analyzed to find the best compatible combination [37]. Linkers including AAY and GPGPG were used for MHC class-I and II constructs respectively [38, 39]. Initially, two epitopes were linked by the selected likers to build the 3D model using Chimera, followed by the cross-validation using HADDDOCK ‘GURU’ interface, refinement and C-port modules [40]. While developing polyvalent construct, clusters of double epitopes were linked with the third and fourth epitopes followed by the same cross-validation analysis to find the best affinity models.
2.7 Safety profile assessment
Safety profile parameters of epitopes and polyvalent construct were examined including stability, SVM score (toxicity) and allergenicity of the conserved epitopes [41]. Support Vector Machine (SVM) model of amino acid composition of epitopes was used to predict the toxicity through online servers ToxinPred and PredSTP [42] (http://crdd.osdd.net/raghava/toxinpred). Stability of a peptide in test tube or in synthesized drug form is indicated by the value of instability index. Predicting allergenicity helps ensure vaccine safety. Epitope’s allergenicity was predicted by AllerTop v. 2.0 [43, 44] (https://www.ddg-pharmfac.net/AllerTOP/method.html).
2.8 Physicochemical properties analysis
The physicochemical properties of epitopes were analyzed to observe hydropathicity (GRAVY), charge, theoretical pI, half-life, aliphatic index, molecular weight and solubility [45]. A thorough understanding of the hydropathicity of each individual amino acid help to anticipate the overall three-dimensional structure of a protein from its amino acid sequence. ExPASy and Peptide 2.0 were utilized to forecast the hydropathicity of peptides [26]. For a protein to be soluble in water it must have charge on it and for this purpose, we used ProtParam (https://web.expasy.org/protparam/) to determine the charge, theoretical pI and aliphatic index. Half-life of peptide is different in different species, we determined the half-life of peptides in human using PEPlife, ToxinPred and ExPASy. Similarly, the molecular weight was determined by the ExPASY server [26] CamSol (http://www-vendruscolo.ch.cam.ac.uk/camsolmethod.html) was used to analyze the solubility of individual epitopes [46] while solubility of polyvalent construct was determined by the Protein-sol server [47] (https://protein-sol.manchester.ac.uk/). The immunodominant potential vaccine target against Brucella melitensis (C1) was used as positive control to compare the characteristics of designed polyvalent construct [48].
2.9 Proteasomal cleavage analysis
The processing and presentation of antigens via MHC represent a highly complex mechanism of proteasomal cleavage. A critical step in this mechanism is the recognition of specific antigen sequences for presentation to immune cells [49]. TAPs (Transporters associated with antigen processing) were also involved in antigen presentation to MHC after cleavage with proteasomes. We have carried out this analysis by NetChop (IEDB) server aiding in identifying potential sites within protein sequences susceptible to proteasomal cleavage [50].
2.10 In silico cytokine release evaluation
To analyze the potential immunogenicity and efficacy of our epitopes, we predicted the cytokine release (IFN-γ, IL-4, IL-6, and IL-10) from the activated T-lymphocytes using IFNepitope [51], IL-4Pred [52] (http://crdd.osdd.net/raghava/il4pred/), IL-6Pred [53] (https://webs.iiitd.edu.in/raghava/il6pred/), and IL-10Pred [54] (http://crdd.osdd.net/raghava/IL-10pred/) servers respectively. These interleukins play a complex and multifaceted role in liver cancer diagnosis and treatment. Th2 cells are a particular subset of immune cells that can be activated by IL-4. IL-6 has the capacity to trigger apoptosis and can stimulate CD8+ T cells along with other immune cells that target cancer directly [55]. IL-10 can strengthen immune responses against malignancies by encouraging the development of tumor-specific T lymphocytes [56].
2.11 3D Modelling and quality assessment
Robetta (https://robetta.bakerlab.org/) and I-Tesser (https://zhanggroup.org/I-TASSER/) tools were used to construct the 3D illustrations of the whole protein [57]. The sequence of the epitope was highlighted in its native protein model after it was processed in ChimeraX [58] (https://www.cgl.ucsf.edu/chimerax/). The 3D models of epitopes and polyvalent construct were generated using PEP-FOLD server (https://bioserv.rpbs.univ-paris-diderot.fr/services/PEP-FOLD/) and I-Tesser [59, 60] as it is an essential step to predict the binding energy.
The quality of these models were assessed by the ERRAT [61] (https://saves.mbi.ucla.edu/) and MolProbity [62] tools. The PEP-FOLD server presented the QMEAN (Qualitative Model Energy Analysis) scores highlighting the homology and similarity of predicted models with experimental structures [60]. Similarly, Ramachandran plots [63] explained the total number of amino acids available in favorable and unfavorable regions. In silico bonding between the chains was checked by performing disulfide engineering (http://cptweb.cpt.wayne.edu/DbD2/index.php) to maintain and improve the stability of construct. Introduction of disulfide bonds also held the protein in specific 3D conformation to become more immunogenic by exposing the certain immune stimulating regions [48].
2.12 Molecular docking of epitopes and polyvalent construct
The binding energies of the selected epitopes were investigated by molecular docking using Molecular Operating Environment (MOE) software [64]. The lesser the binding energy, greater the affinity of the ligand (epitopes) with MHC class-I (PDB ID: 1W72) and MHC class-II (PDB ID: IBX2) as potential targets [65]. This binding interprets the appropriate orientation and attachment of the ligands to the target in order to produce a stable complex [66]. The epitopes of MHC-I and MHC-II were checked for their affinity with their respective receptors included MHC class-I receptor (HLA-A1) with PDB ID: 1W72 [67] and MHC class-II receptor (HLA-DR2) having PDB ID: 1BX2 [68].
ClusPro (https://cluspro.org/help.php) server was utilized to dock the polyvalent construct with their target molecules of class MHC class-I and II [69]. Additionally, the interaction of ligand-target complex were visualized by LigPlot+ software [70].
2.13 Molecular dynamics simulation of construct
The structure of the polyvalent construct was stabilized by GROMACS, an updated MD modeling tool [71] that provides a comprehensive platform for preparation, analysis, and calculations. In order to maintain the stability, a number of parameters were operated including velocity, pressure, temperature, energy minimization, RMSD, RMSF, and Rg (radius of gyration). Optimized Potential for Liquid Simulation force-field was selected and the protein was controlled in a rhombic dodecahedron cubic box to accommodate solvent molecules. Solvation was performed by SPCE water model followed by electro-neutralization and energy minimization under steepest descend algorithm. Program GMX energy was applied to observe some or all of the energies and other parameters like density, box-volume, box-sizes, pressure tensor, and total pressure. One can select from a list of set of energies, such as potential, kinetic, or total energy, or from a list of individual energies, such as dihedral or Lennard-Jones energies. The center of mass velocity is defined as:
For the real-time monitoring of the system’s overall mass, . However, it is directed to eliminate the center-of-mass velocity with each step.
The temperature is given by the total kinetic energy of the N-particle system:
Protein temperature was stabilized using NVT isothermal-isochoric ensemble equilibration at 100 ps, up to a specified value. The quantities we want to compute were obtained from a constant temperature (NVT) ensemble, also known as the canonical ensemble, although the direct use of molecular dynamics also includes the NVE (constant number, constant volume, constant energy ensemble). To simulate constant temperature, GROMACS can employ the Berendsen weak-coupling method [72], the extended ensemble Nosé-Hoover scheme [73, 74], the Andersen thermostat [75], or a velocity-rescaling scheme [76]. The Berendsen algorithm simulates a weak coupling with first-order kinetics to an external heat bath with a fixed temperature T0. This algorithm has the effect on the maintenance of system temperature from T0 in accordance with:
This indicates that the temperature deviation decays exponentially with a given time constant τ. The benefit of this coupling technique is that the coupling strength can be adjusted to the required specifications.
Difference between (Ekin) kinetic energy and virial (Ξ) provide the pressure tensor P.
At every nPC steps, the Berendsen method rescales the coordinates and box vectors with a matrix μ. This results in a first-order kinetic relaxation of the pressure towards a specified reference pressure P0, as:
Pressure and temperature were maintained at 1 bar and 310 k, respectively.
Energy minimization (EM) was done at the 5000 steps in order to get the final energy minimized structure. Conjugate gradients, steepest descent and l-bfgs (limited-memory Broyden-Fletcher-Goldfarb-Shanno quasi-Newtonian minimizer) are three methods for minimizing energy in GROMACS. EM is merely one of the mdrun program’s options. Steepest descent algorithm was used as it is simple and robust method. The vector of all 3N coordinate is referred to as the vector r. It is necessary to first specify a maximum displacement h0 (e.g., 0.01 nm). Potential energy and force F were computed first. The calculation of new positions was done by:
Where ‘Fn’ was the force, or the negative gradient of the potential V, and ℎn is the maximum displacement. The largest scalar force on any atom is indicated by the notation ‘max(|Fn|)’.
Radius of gyration (Rg) was used to measure the compactness of structure was measured, Rg is calculated by:
where ri is position of atom i with respect to the molecule’s center of mass and mi is the mass of atom i. This is very helpful in characterizing the polymers of solutions and proteins. The software will also yield the radius of gyration around the coordinate axis (or, alternatively, primary axes) by simply adding the radii components orthogonal to each axis, for example,
The program GMX rms can be used to compute the root mean square deviation (RMSD) of specific atoms in a molecule with respect to a reference structure by least-square fitting the structure to the reference structure (t2 = 0) and then computing the RMSD.
Where position of atom ‘i’ at time ‘t’ is indicated by ri(t). It should be noted that fitting does not always require the use of the same atoms for the RMSD calculation. For example, a protein is often fitted on its backbone atoms (N, Cα, and C), but its RMSD can be calculated of the protein as a whole or of the backbone.
Alternatively, one can calculate RMSD (t1, t2) using a structure at time t2 = t1− τ, instead of comparing the structures to the initial structure at time t = 0 (such as a crystal structure). This provides some understanding of the mobility in relation to τ. Another way to create a good graphical representation of a trajectory is to create a matrix using the RMSD as a function of t1 and t2. Transitions in a trajectory will be evident in such a matrix if any exist. Fit-free method program was also run to calculate RMSD as:
where the distance ‘rij’ between atoms at time ‘t’ is compared with the distance between the same atoms at time 0.
The stabilized vaccine construct was subjected to temperature, pressure, and density stabilization using an NPT ensemble with 50,000 steps total for the entire procedure. For the equilibrated and stabilized construct, an MD simulation with 50,000 steps was performed at 50 ns.
2.14 Immune simulation analysis of construct
An immune simulation of a polyvalent construct’s sequence in the form of its amino acid sequence is performed using the C-IMMSIM (https://kraken.iac.rm.cnr.it/C-IMMSIM/index.php) server. The humoral and cellular responses of a mammalian immune system in response to the vaccine design are described by this web server.
In conclusion, C-IMMSIM presents images that include representations of the major cell classes of the myeloid lineage (macrophages (M) and dendritic cells) as well as the lymphoid lineage (T helper lymphocytes (Th), cytotoxic T lymphocytes (CTL), B lymphocytes, and antibody-producer plasma cells, PLB) [77]. The initial population of lymphocytes is determined by parameters in C-IMMSIM, including simulation volume, host haplotype selection, injection schedule (both time and dose), and random seed. In order to generate an effective and long-lasting immune response, the study’s simulated parameters included a vaccine devoid of lipopolysaccharides (LPS) and number of Antigen (Ag) to inject were set at 1000, three vaccine doses were set at time intervals of 1,63 and 126-time steps (1 time step is equal to 8 hours) then simulation volumes and steps were set to 10 and 1050, respectively. The other perimeters were kept as default.
2.15 In silico codon optimization and cloning of construct
In E. coli (strain K12), the expression level of the multi-epitope vaccine was measured using the Java Codon Adaptation Tool (J.Cat server: (http://www.prodoric.de/JCat). The reverse translational analysis was performed using this server. The output is cDNA sequence, which was further examined for codon optimization to get a compatible sequence with prokaryotic expression vector. To get the maximum expression outcomes, J.Cat provide the GC content and codon adaptation index (CAI) value [78]. Finally, polyvalent construct was then cloned using SnapGene software (version 5.2.3) into the pET-28a (+) plasmid [79]. For restriction and cloning, Hind III and Bam HI restriction sites were incorporated at the polyvalent construct’s N and C terminals, respectively.
3. Results
3.1 Protein retrieval and screening
We accessed 546 liver cancer associated proteins from the databases and from this total, 125 proteins are overexpressed in liver cancer patients. So, these cancerous proteins were selected for further screening and their potential ability of antigenicity. Off these 125 proteins, we predicted 21 are extracellular, 33 are membrane-bound, and 71 are intracellular proteins (Fig 2).
We found 15 antigenic proteins from the total sum of extracellular and membrane bound proteins based on the significant cut off parameters of VaxiGen database (>0.5). As larger molecular weight is associated with more immunogenicity [36], therefore, we analyzed the molecular weight of these 15 proteins and based on this parameter (>30kDa), ten proteins were retained (Table 1). The differential expression sites of these 10 proteins in different body tissues were investigated and finally 8 proteins were selected based on their significant expression in liver. These proteins includes CDHR5 (Cadherin-related family member 5), AMBP (Alpha-1-microglobulin/bikunin precursor), CFB (Complement factor B), VTN (Vitronectin), APOBR (Apolipoprotein B receptor), FETA (Alpha-fetoprotein), A1AT (Alpha 1-antitrypsin) and APOE (Apolipoprotein E) (Fig 3).
3.2 Prediction of tumor-specific and tumor-associated proteins
We predicted the tumor specific antigens (TSAs) and tumor associated antigens (TAAs) from the list of 8 selected proteins. Off eight, 5 proteins including VTN, CDHR5, APOBR, AFP and APOE are TSAs, while the remaining CFB, AMBP and SERPINA proteins are TAAs (Table 1). During data mining of liver cancer patients, we observed mutations in selected TSAs. For example, CDHR5 protein possessed the mutation at position 107 and indicated the substitution of glutamine with histidine. The other TSA protein VTN was mutated at the 253 positions, presenting isoleucine is replaced with asparagine. Similarly, in case of APOBR and AFP antigens, we analyzed mutation at 435 and 57 position respectively where aspartic acid is changed into glycine and glutamic acid is replaced with pyrrolysine.
3.3 Prediction, screening of T-cell epitopes and conservation analysis
We predicted the multiallelic T-cell epitopes (9-mer residues) of selected proteins. Epitopes were ranked and selected on the basis of significant molecular weight, antigenicity and immunogenicity. These epitopes were antigenic having the antigenicity threshold value of >0.5 [27] and immunogenic with the significant cutoff parameter of >0.1. MHC class-I specific epitopes including LGEGATEAE, LLYIGKDRK, EDIGTEADV, QVDAAMAGR, HLEARKKSK, HLCIRHEMT, LKLSKAVHK, EQGRVRAAT (Table 2) and MHC class-II including VLGEGATEA, WVTKQLNEI, VEEDTKVNS, FTRINCQGK, WGILGREEA, LQDGEKIMS, VKFNKPFVF, VRAATVGSL were selected (Table 3).
Population coverage analysis of these multiallelic epitopes showed the effective immune response among different individuals. In conservation analysis, we found more than 90% conserved domains of these epitopes indicating the broad spectral effect (S2 and S3 Tables).
3.4 Safety profile analysis
We analyzed the toxicity of these epitopes based on the significant threshold value of <5 using the Support Vector Machine (SVM) platform. We observed that entire MHC class-I and II specific epitopes are non-toxic. Based on the stability score of <40, we observed that selected epitopes are stable and possessed long-term effectiveness. We analyzed that these epitopes are non-allergic and do not cause hypersensitivity and anaphylactic reactions (Tables 2 and 3).
3.5 Physicochemical properties of epitopes
We analyzed the hydropathicity, theoretical pI, thermostability, half-life, and solubility of epitopes. The negative value of hydropathicity (GRAVY), ionic nature and solubility score indicated the hydrophilicity and water solubility of epitopes. It was observed that variations in isoelectric point substantially effects the solubility of these epitopes. Peptides with a lower or higher pI value are either positively or negatively charged at neutral pH, this charge can either increase or decrease solubility in aqueous or organic solutions. The half-life of our epitopes was ≤10 hours highlighting their significant bioavailability. We observed that these epitopes are thermostable based on the significant cutoff parameter of aliphatic index (>50) (Table 4).
3.6 Assessment of cytokine release
We analyzed the stimulation of effector molecules in response to check the potential effect of epitopes. Using IFNEpitope and InterleukinPred tools, we observed that these epitopes are good inducers of interferon (IFN-γ). Similarly, we predicted that these epitopes are significantly inducing interleukin-10, interleukin-6 and interleukin-4 [80] (Tables 2 and 3).
3.7 Proteasomal cleavage
Proteasomal cleavage pattern of antigenic proteins was investigated. Each protein possessed several cleavage sites within their sequence. Cleavage was more likely in amino acid sequences with a cleavage score higher than 0.5. The more positive peaks in graph (Fig 4) above the threshold line indicates more cleavage sites compared to the negative peaks below the line. We observed that these epitopes in their respective proteins were cleaved at their C-terminal site for their presentation to major histocompatibility complex molecules.
3.8 Development of polyvalent construct
We designed and constructed the polyvalent sequence of selected epitopes with a molecular weight of 21.43kDa containing a total number of 209 amino acid residues. The entire sequence comprises start codons, epitopes, AAY and GPGPG linkers: MAAYLGEGATEAEAAYLLYIGKDRKAAYEDIGTEADVAAYQVDAAMAGRAAYHLEARKKSKAAYHLCIRHEMTAAYLKLSKAVHKAAYEQGRVRAATGPGPGVLGEGATEAGPGPGWVTKQLNEIGPGPGVEEDTKVNSGPGPGFTRINCQGKGPGPGWGILGREEAGPGPGLQDGEKIMSGPGPGVKFNKPFVFGPGPGVRAATVGSL. 3D model of this polyvalent construct was generated by I-Tesser with a confidence score of -3.52, RMSD value of 13.8±3.9Å and Estimated TM-score of 0.33±0.11. We observed that this polyvalent construct is thermostable and considerably antigenic (score: 0.534) and immunogenic (score: 2.0) with a molecular weight of 21.43 kDa and pI of 7.74 similar to that of (C1) having 13.96kDa and 8.89 respectively. The instability index of the polyvalent construct based on the significant score 21.7 indicated a stable protein complex. With an aliphatic index of 118.13, the final anticipated vaccine construct was a highly thermostable protein. The GRAVY score of our polyvalent construct 0.438. We analyzed the half-life of this construct was 30 hours in mammalian reticulocytes, >20 hours in yeast and >10 hours in Escherichia coli compared to (C1) having the same half-life of 10 hours. Based on substantial cutoff parameter (0.637 score), the solubility analysis of polyvalent has been shown in Fig 5 indicating its water solubility at pH 7.
3.9 Designing of 3D models and quality analysis
3D models of antigenic proteins designed from the Robetta server were used to examine the position of epitopes of MHC class-I (Fig 6) and II (Fig 7) highlighted as green color. We generated 3D models of each epitope by the PEP-FOLD server in PDB format to perform molecular docking. Heatmap graph shows the structural analysis of each amino acid of epitopes. In this graph, the probabilities of structural alphabet on vertical and horizontal axis shows helical (red color), coil (blue) and extended configuration (green color). The QMEAN score of each epitope greater than threshold value 0.5 indicating the predicted structures were similar to the experimentally defined structures from the database (Table 5). Ramachandran plots showed the good quality of 3D models of epitopes indicating 80% residues in favorable and other permissible regions (Fig 8). Similarly, 3D model of polyvalent construct was generated (Fig 9) and we observed that the Ramachandran plot of this model showed the low-energy conformations for psi and phi angles, as well as (59%) of total amino acid residues in favorable region (61.4% in the favored region and 38.6% in the permissible region). Disulfide engineering introduce the strong covalent linkages between cysteine amino acid within the structure of protein. It helps in making the structure more rigid, stable and resistant to degradation within biological system. Disulfide engineered pairs of amino acids are shown in (Table 6) and mutated structure with yellow colored disulfide stick was given in (Fig 10).
(a) Schematic representation of polyvalent construct having 209 residues (b) 3D model of construct obtained from I-Tesser (c) Ramchandran plot indicates the quality of construct.
3.10 Protein-Protein Interaction (PPI) network analysis
The PPI network was created in order to assess how antigenic liver cancer proteins are functionally interacting with host (human) proteins. The eight proteins AMBP, CFB, CDHR5, VTN, APOBR, FETA, A1AT and APOE have direct interaction with each other and indirect interaction with other proteins associated with liver cancer. All of these proteins interacted to form a complicated network in which nodes indicate the proteins and edges shows the interaction. In Fig 11, we segregated this network into three categories presenting antigenic liver proteins as red color nodes, yellow and purple color nodes indicate the tumor specific and tumor associated antigens, while other physiological proteins has been shown as aqua color nodes. We observed that TSAs including ALBU-HUMAN (Albumin), CTNB1-HUMAN (catenin beta-1), HEP2-HUMAN (Heparin cofactor-2), ANT3-HUMAN (antithrombin-III), A2MG-HUMAN (alpha-2 macroglobulin), USH1C-HUMAN (harmonic) and TAAs comprising FETUA-HUMAN (alpha-2-HS glycoprotein), FIBB-HUMAN (fibrinogen beta chain), FITM2-HUMAN (acyl-coenzyme A diphosphatase), PRDX4-HUMAN (peroxiredoxin-4) have direct functional relationship with target proteins.
3.11 Molecular docking of epitopes and polyvalent construct
The in-silico binding affinity (E-score) of the epitopes with MHC molecules (class I and II) was analyzed using MOE software. The binding affinities of the MHC class-I epitopes including LGEGATEAE, LLYIGKDRK, EDIGTEADV, QVDAAMAGR, HLEARKKSK, HLCIRHEMT, LKLSKAVHK, EQGRVRAAT were observed as: -9.57, -8.86, -8.10, -7.99, -8.73, -9.01, -8.51, -8.56 kcal/mol respectively. We observed these epitopes showed interaction with amino acid residues: arginine-A55, glycine-B55, glutamine-A154, asparagine-B53, glutamic acid-A63, and tyrosine-A159 of the MHC-I receptor (Fig 12). Similarly, binding energies of MHC class-II epitopes including VLGEGATEA, WVTKQLNEI, VEEDTKVNS, FTRINCQGK, WGILGREEA, LQDGEKIMS, VKFNKPFVF and VRAATVGSL are -9.20, -11.92, -10.97, -11.16, -10.20, -10.24, -10.06 and -9.82 kcal/mol respectively. In interaction, we found glutamine-A28, glutamic acid-B35, asparagine-C83, leucine-C144, threonine-C80, valine-B50, lysine-A111, arginine-A140, tyrosine-B32, proline-C85, and serine-B54 amino acid residues (Fig 13).
We performed molecular docking of polyvalent construct using ClusPRo software binding energy (E-score) of polyvalent with MHC class-I and II was -1072.3 and -1107.3 respectively. Ligand interactions of construct with MHC class-I and II developed by LigPlot tool. Aspatic acid-122, glutamine-A96-32, arginine-A32-3-234, tyrosine-A27, histidine-A192, threonine-A190, arginine-A202, glutamine-A232-242, tryptophan-A232, proline-A235, alanine-A236, aspartic acid-A238 are the residues found in the interaction. Green and red dashed lines represent hydrogen bonds and salt bridges respectively. The residues involved in hydrophobic interactions are shown with a half circle in red color (Fig 14).
3.12 Molecular dynamics simulation of polyvalent construct
We used the Groningen Machine for Chemical Simulations (GROMACS) spectro built-in tool to check the stability of the polyvalent vaccine construct complexed with MHC class-I and II. During MD simulation, various parameters such as pressure, temperature, radius of gyration (Rg), root-mean-square fluctuation (RMSF) and root-mean-square density (RMSD) were measured. Optimized Potential for Liquid Simulation (OPLS) force-field was used to control the protein in cubic box for solvation of the protein in water. Genion, a tool of GROMACS applied to neutralize the total charge of the protein using a cutoff scheme (Verlet) and for this purpose 22 Na+ ions and 14 Cl- were added in protein complex of MHC class-I and II respectively. R-programme was used to generate graphs. In case of MHC class-I, the average potential energy of the system was measured -1.58149e+5, at temperature 300.22°K, and the total pressure of 2.042 bar and for MHC class-II, average potential energy was -1.58149e+5, temperature and average pressure was 299.94°K and 5.68 bar respectively. After the equilibration of system, we found the total average density of 1019.05 kg/m3 and 1009.327846 for MHC class-I and II (Fig 15). A radius of gyration for RMSD and RMSF was determined by generating analysis of trajectory after a 50-ns simulation. Rg value of the graph (19.8 to 28.8) indicated that the residues of receptor-ligand complex are closer to each other to make a compact configuration for stable interactions. RMSD graph of simulation reaches a plateau condition indicating the simulation time is sufficient for this protein and confirms the equilibrium of the system. During simulation, we observed that RMSD plot showed the structure stability presenting the sustainable score at 58Å. Additionally, we analyzed no variations in RMSD highlighting the stable complex. We used RMSF to assess the fluctuation in amino acid residues of protein complex and observed that target proteins are responsible for turns and coils in the entire structure. It is demonstrated that the protein complex of our polyvalent construct in its final form has good stability in the solvated dynamic state.
MD simulation of polyvalent vaccine construct with MHC class-I and II (a) Temperature of the system maintained around 300°K (b) Pressure plot of ligand at equilibrium at 100ps (c) Rate of gyration indicating the stability of construct throughout the simulation time (d) RMSD level rises up to 57.87Å indicating the construct’s stability (e) RMSF plot’s peaks indicating the flexibility of complex.
3.13 Immune simulation
During immune simulation, we found significant immunoglobulin titers after primary and secondary injections of the vaccine construct. In Fig 16A, initially, we observed a substantial increase of IgM antibody concentration. Antibody levels (IgM, IgG, IgG1, and IgG2) increases as a result of secondary and tertiary responses during the simulation period and antigen level is reduced. The population of B cells, including memory B cells, increased significantly (Fig 16B) indicated the possibility of memory formation and isotype switching while we observed the increase in B cell proliferation and antigen presentation after immunization (Fig 16C). In response to memory development, TH (helper) and TC (cytotoxic) cell populations was significantly high as shown in Fig 16D–16F. During the secondary vaccine injection, we found higher titer rate of macrophage activity and antigen presentation cells (Fig 16G). After the third injection of vaccine, we analyzed a significant increase of interferon gamma (IFN-γ) and a moderate increase in interleukin 2 (IL-2) (Fig 16H). These findings showed the potential impact of polyvalent construct.
Immune simulation by C-ImmSimm (a) Different types of immunoglobulins are represented by different colors as response to vaccine injection (b) Post injection population of B cell population (c) Per state of B-cell population (d) Helper T-cell population (rise in memory cells) (e) Per state population of T-helper cell (f) Post injection increase in cytotoxic T-cells (g) Per state population of macrophages (h)Significant rise in titer of IFN- and IL-2Production (i) Th1 production.
3.14 Codon optimization and in silico cloning
In silico cloning was used to validate the expression of polyvalent construct in E. coli strain. This construct was optimized using JCAT server containing 639 nucleotides with a GC content of 55.5% (30%-70%), and a codon adaptation index (CAI) value of 0.965 (0.8–1.0). Restrictions sites BamHI and HindIII were inserted at the C and N terminals of the optimized polyvalent construct. This modified construct was then cloned at MCS (multiple cloning sites) between these specific restriction sites of the pET30a (+) vector. The resulting clone has a total length of 5977bp (Fig 17).
4. Discussion
Liver cancer is the fourth most lethal disease responsible for 8.2% of deaths globally. It is a major health and economic burden. Traditional treatment methods like surgery, drugs, and combined therapies have some limitations [81]. This research proposes more economical and resourceful integrated immunoinformatic and reverse vaccinology approaches to find potential liver cancer vaccine candidates. In this regard, T-cell epitopes predictions are more effective towards vaccine development against liver cancer [82].
In our study, we used a comprehensive immunoinformatic based framework to design a therapeutic vaccine against the liver cancer. We started screening from 125 liver cancer proteins and finally found 8 proteins which are antigenic, extracellular and highly expressed in liver tissues, making them easily accessible to the immune system [83].
We focused on the prediction of 8 MHC class-I and II T-cell epitopes. MHC class-I epitopes were 100% conserved across various populations of the selected proteins, with the exception of HLEARKKSK, which had a probability of SNP occurring at location H135Y, making them perfect targets for vaccine development [84]. Three epitopes of MHC class-II (WGILGREEA, VLGEGATEA, VEEDTKVNS) have some SNP variations with probability of more than 20%. The study explained by Sukumar et al [85] claimed, these variations are still relatively infrequent and may not significantly impact the effectiveness of epitopes. Overall, the high conservation of these epitopes suggests they are promising candidates for a broadly effective liver cancer vaccine.
Pedro Romero et al [86] and I Farhani et al [87] implemented some bioinformatic tools to predict the antigenicity and immunogenicity of peptides against human melanoma and found three peptides give the best results on the basis of cutoff values of antigenicity and immunogenicity, similarly our predicted T-cell epitopes show antigenicity scores exceeding 0.5 and immunogenicity >0.1 indicating their potential to generate a strong immune response [88]. VaxiJen is a first protein antigenicity prediction tool for bacteria, viruses and tumor. In this server, a significant cutoff threshold of 0.5 is set for all predicted epitopes to get the best possible outcomes amongst different antigens [27]. In a study conducted by Maolin et al [89], authors found 70 unique peptides based on toxicity profile and allergenicity as an important screening parameter. According to AllerTop v. 2.0’s allergenicity prediction [90], we selected non-allergenic epitopes to reduce the possible risk of allergic reactions. Additionally, physicochemical characteristics and safety profile studied by Chiangjong et al [91] and Rathore et al [92] reported stable, safer and effective cancer peptides and highlighted the significance of the physicochemical properties analysis and toxicity prediction. We observed the parallel outcomes indicating the substantial physicochemical properties including non-toxicity, half-life, water solubility, and stability [93]. Similarly, these peptides are capable to activate effector molecules including cytokines and antibodies [91, 94].
For simulation, molecular modeling is important and for this purpose, we used ERRAT server to examine the quality of 3D models of epitopes [95]. Each epitope demonstrated a significant proportion of residues (40% to 100%) in favored and allowed regions indicating overall good geometry [96]. Remarkably, none of the residues were observed in disallowed regions, implying that the predicted structures are unlikely to contain major structural errors. Likewise, all epitopes had positive Q-mean scores, indication a strong level of similarity between the predicted and experimentally determined protein structures [97]. These findings provided further evidence to support the validity and potential effectiveness of these epitopes as vaccine candidates.
According to Vilar et al [98], molecular docking is effective to screen the vaccine candidates and we observed that the interaction of epitopes and HLA (MHCs) molecules based on their binding energies showed significant binding affinities. The computed binding energies of the MHC-I specific epitopes range from -7.93 to -10.09 kcal/mol and MHC-II specific epitopes showed -9.26 to -11.15 kcal/mol presenting considerable interactions between these epitopes and target molecules.
We analyzed that potential polyvalent vaccine candidates could be effective against cancer. In previous studies, it has been reported that such polyvalent candidates showed substantial activity against Wilms tumor-1 (WT1) and human papillomavirus (HPV) cancer [17] against Hepatitis C [99]. Similarly, polyvalent vaccine was designed to analyze the broad spectral effect against respiratory syncytial virus (RSV) using simulation based framework [100]. In our analysis, our polyvalent construct containing 209 amino acid residues including AAY and GPGPG linkers with a molecular weight of 21.43kDa is significant immunogenic 17, [38, 101]. We observed the considerable binding energies (-1072.3 and -1107.3) during molecular interaction. MD simulations were performed to evaluate the stability of the docked complexes with MHC class-Iand II molecules using different statistical algorithms as previously described [102, 103]. In silico cloning is important to develop cost effective vaccines [104]. The codon adaptation index (CAI) value indicated [105] that the polyvalent mRNA codons were compatible with the E. coli system suggesting that the predicted polyvalent production was possible by in silico cloning [106]. The selected epitopes and potential polyvalent construct are efficiently recognized and presented by the immune system triggering an immune response against liver cancer.
5. Conclusion
Immunotherapy against cancer stimulates the patient’s immune system to recognize and eliminate cancerous cells. It is crucial to determine the antigenic proteins present on malignant cells. The application of this computational approach led to the identification of potent MHC class-I and II T-cell epitopes that potentially be used in the development of a vaccine against liver cancer. This research study suggests that these potential epitopes have significant binding affinity and immunogenicity to produce a therapeutic immune response against liver cancer.
Supporting information
S1 Table. List of bioinformatics tools/ software used in this research study.
https://doi.org/10.1371/journal.pone.0309049.s001
(DOCX)
S2 Table. Prediction, conservational analysis and population coverage of multiallelic MHC class-I T-cell epitopes.
https://doi.org/10.1371/journal.pone.0309049.s002
(DOCX)
S3 Table. Prediction, conservational analysis and population coverage of multiallelic MHC class-II T-cell epitopes.
https://doi.org/10.1371/journal.pone.0309049.s003
(DOCX)
References
- 1. Rumgay H, Arnold M, Ferlay J, Lesi O, Cabasag CJ, Vignat J, et al. Global burden of primary liver cancer in 2020 and predictions to 2040. J Hepatol. 2022;77(6):1598–606. pmid:36208844
- 2. Feng GS, Hanley KL, Liang Y, Lin X. Improving the efficacy of liver cancer immunotherapy: the power of combined preclinical and clinical studies. J Hepatol. 2021;73:104–14. pmid:32715491
- 3. Singal AG, Lampertico P, Nahon P. Epidemiology and surveillance for hepatocellular carcinoma: New trends. J Hepatol. 2020;72(2):250–61. pmid:31954490
- 4. Shi JF, Cao M, Wang Y, Bai FZ, Lei L, Peng J, et al. Is it possible to halve the incidence of liver cancer in China by 2050? Int J Cancer. 2021;148(5):1051–65. pmid:32997794
- 5. Dimri M, Satyanarayana A. Molecular Signaling Pathways and Therapeutic Targets in Hepatocellular Carcinoma. Cancers. 2020;12(2):491. pmid:32093152
- 6. Liu S, Xu Z, Fang Z, Zhang D, Qin Z, Fan L, et al. The Combination of Age, International Standardized Ratio, Albumin and γ-Glutamyl Transpeptidase (AIAG), Tumor Size and Alpha Fetoprotein (AFP) Stage as the Prognostic Model for Hepatitis B-Related Hepatocellular Carcinoma. Int J Gen Med. 2021;14:4291–301.
- 7. Nishida N, Kudo MJO. Oncogenic signal and tumor microenvironment in hepatocellular carcinoma. Oncology. 2017;93(Suppl. 1):160–4. pmid:29258072
- 8. Zheng Y, Li Y, Feng J, Li J, Ji J, Wu L, et al. Cellular based immunotherapy for primary liver cancer. J Exp Clin Cancer Res. 2021;40(1):1–16.
- 9. Kubes P, Jenne C. Immune responses in the liver. Annu Rev Immunol. 2018;36:247–77. pmid:29328785
- 10. Shuai Z, Leung MW, He X, Zhang W, Yang G, Leung PS, et al. Adaptive immunity in the liver. Cell Mol Immunol. 2016;13(3):354–68. pmid:26996069
- 11. Wang SZ, Lee SD, Sarkar D, Lee HM, Khan A, Bhati C, et al. Immunological characterization of hepatocellular carcinoma. Hepatoma Res. 2021.
- 12. Sharonov GV, Serebrovskaya EO, Yuzhakova DV, Britanova OV, Chudakov DM. B cells, plasma cells and antibody repertoires in the tumour microenvironment. Nat Rev Immunol. 2020;20(5):294–307. pmid:31988391
- 13. Sim H-W, Knox J. Hepatocellular carcinoma in the era of immunotherapy. Curr Probl Cancer. 2018;42(1):40–8. pmid:29150141
- 14. Soria-Guerra RE, Nieto-Gomez R, Govea-Alonso DO, Rosales-Mendoza S. An overview of bioinformatics tools for epitope prediction: implications on vaccine development. J Biomed Inform. 2015;53:405–14. pmid:25464113
- 15. Paul S, Dillon MB, Arlehamn CSL, Huang H, Davis MM, McKinney DM, et al. A population response analysis approach to assign class II HLA-epitope restrictions. J Immunol. 2015;194(12):6164–76. pmid:25948811
- 16. Kazi A, Chuah C, Majeed ABA, Leow CH, Lim BH, Leow CY. Current progress of immunoinformatics approach harnessed for cellular-and antibody-dependent vaccine design. Pathog Glob Health. 2018;112(3):123–31. pmid:29528265
- 17. Zaheer T, Waseem M, Waqar W, Dar HA, Shehroz M, Naz K, et al. Anti-COVID-19 multi-epitope vaccine designs employing global viral genome sequences. PeerJ. 2020;8:e9541. pmid:32832263
- 18. Szarewski A. HPV vaccine: Cervarix. Expert Opin Biol Ther. 2010;10(3):477–87. pmid:20132062
- 19. Maltz F, Fidler B. Shingrix: a new herpes zoster vaccine. Pharmacy and Therapeutics. 2019;44(7):406. pmid:31258310
- 20. Masignani V, Pizza M, Moxon ER. The development of a vaccine against meningococcus B using reverse vaccinology. Front Immunol. 2019;10:441439. pmid:31040844
- 21. Rosenberg SA, Restifo NP. Adoptive cell transfer as personalized immunotherapy for human cancer. Science. 2015;348(6230):62–8. pmid:25838374
- 22. Abbas AK, Lichtman AH, Pillai S. Basic Immunology E-Book: Basic Immunology E-Book: Elsevier Health Sciences; 2023.
- 23. Pardoll DM. The blockade of immune checkpoints in cancer immunotherapy. Nat Rev Cancer. 2012;12(4):252–64. pmid:22437870
- 24. Uhlen M, Zhang C, Lee S, Sjöstedt E, Fagerberg L, Bidkhori G, et al. A pathology atlas of the human cancer transcriptome. Science. 2017(6352). pmid:28818916
- 25. Lee L, Wang K, Li G, Xie Z, Wang Y, Xu J, et al., editors. Liverome: a curated database of liver cancer-related gene signatures with self-contained context information. BMC Genomics; 2011: Springer.
- 26. Gasteiger E, Hoogland C, Gattiker A, Wilkins MR, Appel RD, Bairoch A. Protein identification and analysis tools on the ExPASy server. The Proteomics Protocols Handbook. 2005:571–607.
- 27. Doytchinova IA, Flower DR. VaxiJen: a server for prediction of protective antigens, tumour antigens and subunit vaccines. BMC Bioinform. 2007;8(1):1–7. pmid:17207271
- 28. Yu CS, Chen YC, Lu CH, Hwang JK. Prediction of protein subcellular localization. Proteins: Struct, Funct, Bioinf. 2006;64(3):643–51. pmid:16752418
- 29. Kassambara A, Rème T, Jourdan M, Fest T, Hose D, Tarte K, et al. GenomicScape: an easy-to-use web tool for gene expression data analysis. Application to investigate the molecular events in the differentiation of B cells into plasma cells. PLoS Comput Biol. 2015;11(1):e1004077. pmid:25633866
- 30. Scholtalbers J, Boegel S, Bukur T, Byl M, Goerges S, Sorn P, et al. TCLP: an online cancer cell line catalogue integrating HLA type, predicted neo-epitopes, virus and gene expression. Genome Med. 2015;7(1):1–7.
- 31. Wu J, Zhao W, Zhou B, Su Z, Gu X, Zhou Z, et al. TSNAdb: a database for tumor-specific neoantigens from immunogenomics data analysis. Genom Proteom Bioinform. 2018;16(4):276–82. pmid:30223042
- 32. Chen JY, Mamidipalli S, Huan T. HAPPI: an online database of comprehensive human annotated and predicted protein interactions. BMC Genomics. 2009;10(S1):S16.
- 33. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13(11):2498–504. pmid:14597658
- 34. Brusic V, Rudy G, Harrison LC. Prediction of MHC binding peptides using artificial neural networks. Complex Syst: Mech Adapt. 1994:253–60.
- 35. Goldman MJ, Craft B, Hastie M, Repečka K, McDade F, Kamath A, et al. Visualizing and interpreting cancer genomics data via the Xena platform. Nat Biotechnol. 2020;38(6):675–8. pmid:32444850
- 36. Calis JJ, Maybeno M, Greenbaum JA, Weiskopf D, De Silva AD, Sette A, et al. Properties of MHC class I presented peptides that enhance immunogenicity. PLoS Comput Biol. 2013;9(10):e1003266. pmid:24204222
- 37. Dominguez C, Boelens R, Bonvin AM. HADDOCK: a protein− protein docking approach based on biochemical or biophysical information. J Am Chem Soc. 2003;125(7):1731–7. pmid:12580598
- 38. Ali M, Pandey RK, Khatoon N, Narula A, Mishra A, Prajapati VK. Exploring dengue genome to construct a multi-epitope based subunit vaccine by utilizing immunoinformatics approach to battle against dengue infection. Sci Rep. 2017;7(1):9232. pmid:28835708
- 39. Pandey RK, Bhatt TK, Prajapati VK. Novel immunoinformatics approaches to design multi-epitope subunit vaccine for malaria by investigating anopheles salivary protein. Sci Rep. 2018;8(1):1125. pmid:29348555
- 40. Van Zundert G, Rodrigues J, Trellet M, Schmitz C, Kastritis P, Karaca E, et al. The HADDOCK2. 2 web server: user-friendly integrative modeling of biomolecular complexes. J Mol Biol. 2016;428(4):720–5. pmid:26410586
- 41. Muhammad SA, Zafar S, Rizvi SZ, Imran I, Munir F, Jamshed MB, et al. Experimental analysis of T cell epitopes for designing liver cancer vaccine predicted by system-level immunoinformatics approach. Am J Physiol Gastrointest Liver Physiol. 2020;318(6):G1055–G69. pmid:32363891
- 42. Gupta S, Kapoor P, Chaudhary K, Gautam A, Kumar R, Raghava GP, et al. In silico approach for predicting toxicity of peptides and proteins. PloS One. 2013;8(9):e73957. pmid:24058508
- 43. Dimitrov I, Bangov I, Flower DR, Doytchinova I. AllerTOP v. 2—a server for in silico prediction of allergens. J Mol Model. 2014;20(6):1–6. pmid:24878803
- 44. Safavi A, Kefayat A, Abiri A, Mahdevar E, Behnia AH, Ghahremani F. In silico analysis of transmembrane protein 31 (TMEM31) antigen to design novel multiepitope peptide and DNA cancer vaccines against melanoma. Mol Immunol. 2019;112:93–102. pmid:31079006
- 45. Zafar S, Bai B, Guo J, Muhammad SA, Naqvi STQ, Shabbir MN, et al. Experimental Study of Potential CD8+ Trivalent Synthetic Peptides for Liver Cancer Vaccine Development Using Sprague Dawley Rat Models. Biomed Res Int. 2022;2022. pmid:35686237
- 46. Sormanni P, Amery L, Ekizoglou S, Vendruscolo M, Popovic B. Rapid and accurate in silico solubility screening of a monoclonal antibody library. Sci Rep. 2017;7(1):1–9.
- 47. Hebditch M, Carballo-Amador MA, Charonis S, Curtis R, Warwicker J. Protein–Sol: a web tool for predicting protein solubility from sequence. Bioinformatics. 2017;33(19):3098–100. pmid:28575391
- 48. Malik M, Khan S, Ullah A, Hassan M, Haq Mu, Ahmad S, et al. Proteome-Wide Screening of Potential Vaccine Targets against Brucella melitensis. Vaccines. 2023;11(2):263. pmid:36851141
- 49. Pishesha N, Harmand TJ, Ploegh HL. A guide to antigen processing and presentation. Nature Reviews Immunology. 2022;22(12):751–64. pmid:35418563
- 50. Nielsen M, Lundegaard C, Lund O, Keşmir C. The role of the proteasome in generating cytotoxic T-cell epitopes: insights obtained from improved predictions of proteasomal cleavage. Immunogenetics. 2005;57(1):33–41. pmid:15744535
- 51. Dhanda SK, Vir P, Raghava GP. Designing of interferon-gamma inducing MHC class-II binders. Biol Direct. 2013;8(1):1–15. pmid:24304645
- 52. Dhanda SK, Gupta S, Vir P, Raghava G. Prediction of IL4 inducing peptides. Clin Dev Immunol. 2013. pmid:24489573
- 53. Dhall A, Patiyal S, Sharma N, Usmani SS, Raghava GP. Computer-aided prediction and design of IL-6 inducing peptides: IL-6 plays a crucial role in COVID-19. J Bioinform. 2021;22(2):936–45.
- 54. Nagpal G, Usmani SS, Dhanda SK, Kaur H, Singh S, Sharma M, et al. Computer-aided designing of immunosuppressive peptides based on IL-10 inducing potential. Sci Rep. 2017;7(1):1–10.
- 55. Masjedi A, Hashemi V, Hojjat-Farsangi M, Ghalamfarsa G, Azizi G, Yousefi M, et al. The significant role of interleukin-6 and its signaling pathway in the immunopathogenesis and treatment of breast cancer. Biomed Pharmacother. 2018;108:1415–24. pmid:30372844
- 56. Hamidullah , Changkija B, Konwar R. Role of interleukin-10 in breast cancer. Breast Cancer Res Treat. 2012;133:11–21. pmid:22057973
- 57. Kim DE, Chivian D, Baker D. Protein structure prediction and analysis using the Robetta server. Nucleic Acids Res Spec Publ. 2004;32(suppl_2):W526-W31. pmid:15215442
- 58. Pettersen EF, Goddard TD, Huang CC, Meng EC, Couch GS, Croll TI, et al. UCSF ChimeraX: Structure visualization for researchers, educators, and developers. Protein Sci. 2021;30(1):70–82. pmid:32881101
- 59. Shen Y, Maupetit J, Derreumaux P, Tufféry P. Improved PEP-FOLD approach for peptide and miniprotein structure prediction. J Chem Theory Comput. 2014;10(10):4745–58. pmid:26588162
- 60. Thévenet P, Shen Y, Maupetit J, Guyon F, Derreumaux P, Tufféry P. PEP-FOLD: an updated de novo structure prediction server for both linear and disulfide bonded cyclic peptides. Nucleic Acids Res Spec Publ. 2012;40(W1):W288–W93. pmid:22581768
- 61. Colovos C, Yeates TO. Verification of protein structures: patterns of nonbonded atomic interactions. Protein Sci. 1993;2(9):1511–9. pmid:8401235
- 62. Prisant MG, Williams CJ, Chen VB, Richardson JS, Richardson DC. New tools in MolProbity validation: CaBLAM for CryoEM backbone, UnDowser to rethink “waters,” and NGL Viewer to recapture online 3D graphics. Protein Sci. 2020;29(1):315–29. pmid:31724275
- 63. Tam B, Sinha S, Wang SM. Combining Ramachandran plot and molecular dynamics simulation for structural-based variant classification: Using TP53 variants as model. Comput Struct Biotechnol J. 2020;18:4033–9. pmid:33363700
- 64. Jakhar R, Dangi M, Khichi A, Chhillar AK. Relevance of molecular docking studies in drug designing. Current Bioinformatics. 2020;15(4):270–8.
- 65. Tomar N, Singh V, Marla S, Chandra R, Kumar R, Kumar A. Molecular docking studies with rabies virus glycoprotein to design viral therapeutics. Indian J Pharm Sci. 2010;72(4):486. pmid:21218060
- 66. Zhang Y. I-TASSER server for protein 3D structure prediction. BMC bioinform. 2008;9:1–8. pmid:18215316
- 67. Hülsmeyer M, Chames P, Hillig RC, Stanfield RL, Held G, Coulie PG, et al. A major histocompatibility complex· peptide-restricted antibody and T cell receptor molecules recognize their target by distinct binding modes: crystal structure of human leukocyte antigen (HLA)-A1· MAGE-A1 in complex with FAB-HYB3. J Biol Chem. 2005;280(4):2972–80. pmid:15537658
- 68. Smith KJ, Pyrdol J, Gauthier L, Wiley DC, Wucherpfennig KW. Crystal structure of HLA-DR2 (DRA* 0101, DRB1* 1501) complexed with a peptide from human myelin basic protein. J Exp Med. 1998;188(8):1511–20. pmid:9782128
- 69. Alekseenko A, Ignatov M, Jones G, Sabitova M, Kozakov D. Protein–protein and protein–peptide docking with ClusPro server. Protein Structure Prediction. 2020:157–74. pmid:32621224
- 70. Agrawal R, Punarva H, Heda GO, Vishesh Y, Karunakar P. VinaLigGen: a method to generate LigPlots and retrieval of hydrogen and hydrophobic interactions from protein-ligand complexes. J Biomol Struct Dyn. 2023:1–4. pmid:37822199
- 71. Páll S, Zhmurov A, Bauer P, Abraham M, Lundborg M, Gray A, et al. Heterogeneous parallelization and acceleration of molecular dynamics simulations in GROMACS. J Chem Phys. 2020;153(13). pmid:33032406
- 72. Berendsen HJ, Postma Jv, Van Gunsteren WF, DiNola A, Haak JR. Molecular dynamics with coupling to an external bath. J Chem Phys 1984;81(8):3684–90.
- 73. Nosé S. A molecular dynamics method for simulations in the canonical ensemble. Mol Phys. 1984;52(2):255–68.
- 74. Hoover WG. Canonical dynamics: Equilibrium phase-space distributions. Physical review A. 1985;31(3):1695. pmid:9895674
- 75. Andersen HC. Molecular dynamics simulations at constant pressure and/or temperature. J Chem Phys. 1980;72(4):2384–93.
- 76. Bussi G, Donadio D, Parrinello M. Canonical sampling through velocity rescaling. J Chem Phys 2007;126(1). pmid:17212484
- 77. Rapin N, Lund O, Bernaschi M, Castiglione F. Computational immunology meets bioinformatics: the use of prediction tools for molecular binding in the simulation of the immune system. PloS one. 2010;5(4):e9862. pmid:20419125
- 78. Watts A, Sankaranarayanan S, Watts A, Raipuria RK. Optimizing protein expression in heterologous system: strategies and tools. Meta Gene. 2021;29:100899.
- 79. Zhang L, Jiang C, Chen X, Gu J, Song Q, Zhong H, et al. Large‐scale production, purification, and function of a tumor multi‐epitope vaccine: Peptibody with bFGF/VEGFA. ENG LIFE SCI Journal. 2020;20(9–10):422–36. pmid:32944017
- 80. Zhang C, Wang N, Tan HY, Guo W, Chen F, Zhong Z, et al. Direct inhibition of the TLR4/MyD88 pathway by geniposide suppresses HIF‐1α‐independent VEGF expression and angiogenesis in hepatocellular carcinoma. Br J Pharmacol. 2020;177(14):3240–57.
- 81. Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68(6):394–424. pmid:30207593
- 82. Nayak A, Gadnayak A, Sahoo M, Bhairappanavar SB, Rout B, Mohanty JN, et al. Inferring Toll-Like Receptor induced epitope subunit vaccine candidate against SARS-CoV-2: A Reverse Vaccinology approach. bioRxiv. 2020:2020.12. 24.424322.
- 83. Shen Y, Ding Y, Tang J, Zou Q, Guo F. Critical evaluation of web-based prediction tools for human protein subcellular localization. Brief Bioinform. 2020;21(5):1628–40. pmid:31697319
- 84. Adhikari UK, Tayebi M, Rahman MM. Immunoinformatics approach for epitope-based peptide vaccine design and active site prediction against polyprotein of emerging oropouche virus. J Immunol Res. 2018;2018. pmid:30402510
- 85. Sukumar S, Krishnan A, Banerjee S. An overview of bioinformatics resources for SNP analysis. Bioinform adv. 2021:113–35.
- 86. Romero P, Valmori D, Pittet MJ, Zippelius A, Rimoldi D, Lévy F, et al. Antigenicity and immunogenicity of Melan‐A/MART‐1 derived peptides as targets for tumor reactive CTL in human melanoma. Immunological reviews. 2002;188(1):81–96. pmid:12445283
- 87. Farhani I, Yamchi A, Madanchi H, Khazaei V, Behrouzikhah M, Abbasi H, et al. Designing a multi-epitope vaccine against the SARS-CoV-2 variant based on an Immunoinformatics Approach. Current Computer-Aided Drug Design. 2024;20(3):274–90. pmid:37309762
- 88. Bryson CJ, Jones TD, Baker MP. Prediction of immunogenicity of therapeutic proteins: validity of computational tools. BioDrugs. 2010;24:1–8. pmid:20055528
- 89. Tu M, Liu H, Zhang R, Chen H, Fan F, Shi P, et al. Bioactive hydrolysates from casein: generation, identification, and in silico toxicity and allergenicity prediction of peptides. Journal of the Science of Food and Agriculture. 2018;98(9):3416–26. pmid:29280148
- 90. Dimitrov I, Flower DR, Doytchinova I, editors. AllerTOP-a server for in silico prediction of allergens. BMC Bioinform; 2013: BioMed Central. pmid:23735058
- 91. Chiangjong W, Chutipongtanate S, Hongeng S. Anticancer peptide: Physicochemical property, functional aspect and trend in clinical application. Int J Oncol. 2020;57(3):678–96.
- 92. Rathore AS, Arora A, Choudhury SP, Tijare P, Raghava GPS. ToxinPred 3.0: An improved method for predicting the toxicity of peptides. bioRxiv. 2023:2023.08. 11.552911.
- 93. Garg VK, Avashthi H, Tiwari A, Jain PA, Ramkete PW, Kayastha AM, et al. MFPPI–multi FASTA ProtParam interface. Bioinformation. 2016;12(2):74. pmid:28104964
- 94. Ammirati E, Cannistraci CV, Cristell NA, Vecchio V, Palini AG, Tornvall P, et al. Identification and predictive value of interleukin-6+ interleukin-10+ and interleukin-6− interleukin-10+ cytokine patterns in ST-elevation acute myocardial infarction. Circ Res. 2012;111(10):1336–48. pmid:22931953
- 95. Yao Z, Cao D. 4.3. 9 ProCheck module. PyBioMed Documentation. 2023:53.
- 96. Sobolev OV, Afonine PV, Moriarty NW, Hekkelman ML, Joosten RP, Perrakis A, et al. A global Ramachandran score identifies protein structures with unlikely stereochemistry. Structure. 2020;28(11):1249–58. e2. pmid:32857966
- 97. Binette V, Mousseau N, Tuffery P. A generalized attraction–repulsion potential and revisited fragment library improves PEP-FOLD peptide structure prediction. J Chem Theory Comput. 2022;18(4):2720–36. pmid:35298162
- 98. Vilar S, Cozza G, Moro S. Medicinal chemistry and the molecular operating environment (MOE): application of QSAR and molecular docking to drug discovery. Curr Top Med Chem. 2008;8(18):1555–72. pmid:19075767
- 99. Ahmad S, Demneh FM, Rehman B, Almanaa TN, Akhtar N, Pazoki-Toroudi H, et al. In silico design of a novel multi-epitope vaccine against HCV infection through immunoinformatics approaches. International Journal of Biological Macromolecules. 2024;267:131517. pmid:38621559
- 100. Naqvi STQ, Yasmeen M, Ismail M, Muhammad SA, Nawazish-i-Husain S, Ali A, et al. Designing of potential polyvalent vaccine model for respiratory syncytial virus by system level immunoinformatics approaches. BioMed Research International. 2021;2021:1–18.
- 101. Dar HA, Zaheer T, Shehroz M, Ullah N, Naz K, Muhammad SA, et al. Immunoinformatics-aided design and evaluation of a potential multi-epitope vaccine against Klebsiella pneumoniae. Vaccines. 2019;7(3):88. pmid:31409021
- 102. Panwar A, Kumar A. In-silico analysis and molecular dynamics simulations of lysozyme by GROMACS 2020.2. Ann Rom Soc Cell Biol. 2021;25(6):9679–85.
- 103. Ullah A, Rehman B, Khan S, Almanaa TN, Waheed Y, Hassan M, et al. An In Silico Multi-epitopes Vaccine Ensemble and Characterization Against Nosocomial Proteus penneri. Molecular Biotechnology. 2023:1–16.
- 104. Tahir ul Qamar M, Shokat Z, Muneer I, Ashfaq UA, Javed H, Anwar F, et al. Multiepitope-based subunit vaccine design and evaluation against respiratory syncytial virus using reverse vaccinology approach. Vaccines. 2020;8(2):288. pmid:32521680
- 105. Ahmad S, Nazarian S, Alizadeh A, Pashapour Hajialilou M, Tahmasebian S, Alharbi M, et al. Computational design of a multi-epitope vaccine candidate against Langya henipavirus using surface proteins. Journal of Biomolecular Structure and Dynamics. 2023:1–18. pmid:37713338
- 106. Nain Z, Karim MM, Sen MK, Adhikari UK. Structural basis and designing of peptide vaccine using PE-PGRS family protein of Mycobacterium ulcerans—An integrated vaccinomics approach. Mol Immunol. 2020;120:146–63. pmid:32126449