Fig 1.
Flowchart summarising the immunoinformatic methods in the rational design of the vaccine.
The formation of vaccine construct in silico includes 1. Protein sequences retrieval and processing, 2. Epitopes identification, 3. Epitopes selection, 4. Structure prediction and molecular docking and 5. Codon optimisation. Tools and web servers used in the analyses are in bold. CTL: Cytotoxic T Lymphocytes; HTL: Helper T Lymphocytes; LBC: Linear B-cells; Ig: immunoglobulin; 2D: secondary structure; 3D: tertiary structure.
Fig 2.
Three proposed organisation (A, B and C) of the multi-epitope vaccine constructs, MEVC-A, MEVC-B and MEVC-C. The orientation of the epitopes made based on previous literatures [26, 33, 34] were tested for the efficacy of the vaccine in silico. The defensin adjuvant (pink) was joined to the epitopes by the EAAAK linker (light pink). CTL epitopes (green) were joined together with AAY linker (light green). The HTL epitopes (orange) were separated with GPGPG linker (light orange) while LBC epitope (blue) were joined with KK linkers (light blue). The epitopes were then arranged from the most conserved epitopes to the least conserved extracted from the receptor-binding domain (RBD) S1 subunit and S2 (HR) subunits. hBD3, human β-defensin 3 (adjuvant); CTL, Cytotoxic Lymphocytes; HTL, Helper T Lymphocytes; LBC, Linear B-cells [26, 33, 34].
Table 1.
Final CTL epitope candidate within receptor-binding domain and heptad repeat domains of the SARS-CoV-2 S protein with their antigenicity value and conservancy percentage.
Table 2.
Final HTL epitope candidates within receptor-binding domain and heptad repeat domains of the SARS-CoV-2 S protein with their binding core, antigenicity value and conservancy percentage.
Table 3.
Population coverage analysis using IEDB resource tool calculated based on the set of 19 T lymphocytes epitopes selected for the formulation of the vaccine construct globally and specific to geographical regions where the SARS-CoV-2 strains were retrieved.
Fig 3.
Multi-epitope vaccine construct B (MEVC-B) arrangement.
The signal peptide, Usp45, was added before the hBD3 adjuvant. The EAAAK linker was used to connect the adjuvant and the LBC epitopes which were arranged first in the construct, epitopes from the receptor-binding domain (RBD) region followed by heptad repeat (HRs) region. The LBC epitopes were separated by KK linker. The HTL epitopes were joined with the GPGPG linker while the CTL epitopes were joined with the AAY linker. The epitopes were arranged from RBD region followed by epitopes from HRs region. The 6×His-tag was added to the C-terminal linked with RVRR linker.
Fig 4.
Pro-SA web server Z-score plot of 3D structure models for the MEVC-B.
Before (panel A) and after (panel B) refinement with GalaxyRefine web server showed no significant difference in the Z-scores. The black dot in the blue shaded region represents our MEVC-B. The negative value of the Z-score indicates that the predicted 3D structure for the MEVC-B was reliable and stable as the predicted structure score value lies within the Z-scores of experimentally determined structures. Reprinted from ProSA server (https://prosa.services.came.sbg.ac.at/prosa.php) [46] under a CC BY license, with permission from PLOS ONE, original copyright 2023.
Fig 5.
Ramachandran plot of 3D structure models for MEVC-B.
Plot from panel A showed the Ramachandran plot before refinement and panel B for the refined model structures generated with PROCHECK web server. The most favoured regions were labelled in white letters “A”, “B”, “L”. The black squares and triangles represent the residues that were within favourable regions. Red squares represent residues that were within the disallowed region of the Ramachandran plot. The residues were slightly more concentrated in the most favoured region and also had slightly more residues within the disallowed region after refinement compared to before refinement. A good quality model structure is expected to have more than 90% residues [47]. Reprinted from PROCHECK web server (https://www.ebi.ac.uk/thornton-srv/software/PROCHECK/) [47] under a CC BY license, with permission from PLOS ONE, original copyright 2023.
Table 4.
Re-validation of five models structures of model 2 MEVC-B generated by GalaxyRefine with ProSA, ERRAT and PROCHECK web servers.
Fig 6.
Molecular docked model structure of the MEVCB-TLR4-MD-2 complex.
Panel B showed the homodimer TLR4, labelled as Chain A (purple) and Chain B (magenta), the MD-2 structures coloured in orange and MEVC-B structure in cyan. Panel A and C showed the binding interactions between the docked complex. The MEVC-B structure interacted with both chains of TLR4 (panel A and C) but only one chain of the MD-2 in the docked complex (not shown). Reprinted from PDBsum Generate web server (https://www.ebi.ac.uk/thornton-srv/databases/pdbsum/Generate.html) [72] under a CC BY license, with permission from PLOS ONE, original copyright 2023.
Fig 7.
Molecular docked model structure of the MEVCB-TLR1-TLR2 complex.
The docked complex showed that there is no interaction predicted between the MEVC-B and TLR2 chain in protein-protein complex (panel A). The binding interactions between TLR1 of the TLR1-TLR2 heterodimer (red) and MEVC-B (cyan) were visualised with Pymol and PDBsum server in Panel B. The interaction within the complex is only limited between the MEVC-B with TLR-1 which is shown in the top-view of the docked complex (panel C). Reprinted from PDBsum Generate web server (https://www.ebi.ac.uk/thornton-srv/databases/pdbsum/Generate.html) [72] under a CC BY license, with permission from PLOS ONE, original copyright 2023.
Fig 8.
Molecular docked model structure of the MEVCB-TLR1-TLR2b complex.
The docked complex showed that the interaction between the MEVC-B with TLR1 (panel C) and TLR2 chain (panel A). The binding interactions between the TLR1-TLR2 heterodimer (red and pink) and MEVC-B (cyan) were visualised with Pymol and PDBsum server in Panel B. The binding residues in the interaction were not included in the reported ligand-binding residues (panel B and C). Reprinted from PDBsum Generate web server (https://www.ebi.ac.uk/thornton-srv/databases/pdbsum/Generate.html) [72] under a CC BY license, with permission from PLOS ONE, original copyright 2023.
Table 5.
Docking results of the top five cluster with the most negative lowest energy scores the on the docked complexes between MEVC-B and TLR molecules generated by ClusPro web server and the binding affinity between the complex.
Table 6.
Docking results of the docked complexes between MEVC-B and TLR molecules for the top five cluster with the Z-score and HADDOCK score generated by HADDOCK web server and the binding affinity between the complex.
Table 7.
Docking results of the individual epitopes and TLR molecules for the top 10 highest binding affinity and the most negative lowest energy scores generated by ClusPro web server.
Fig 9.
Energy minimisation and equilibration of the system for MD simulations.
The potential energy of the system of both docked complexes were minimised prior to the molecular dynamic simulations and the plots showed that the minimum potential energy of both docked complexes were stable over the course of 100-ps simulation (panel A). The system in both protein-protein complex was equilibrated under two phases: pressure (NPT) and temperature (NVT) ensembles. The graph of the NPT and NVT equilibration phases for the protein complexes in panel B and C, respectively, showed that the system in both protein complexes were instantly stabilised and reached the desired pressure and temperature. The graphs were plotted using the XMGRACE plotting tool [71].
Fig 10.
RMSD calculations of the docked protein complexes from MD simulations.
The RMSD calculation of C-alpha atoms of each chain in the MEVCB-TLR4-MD-2 (Panel A) and MEVCB-TLR1-TLR2 (Panel B) complex. The RMSD value assessed the structural stability of the protein in the complexes over the 100ns MD simulations. The graphs were plotted using the XMGRACE plotting tool [71].
Fig 11.
RMSF calculations of the C-alpha atom of each chain in the docked protein complexes from MD simulations.
The RMSF of proteins in the MEVCB-TLR4-MD-2 (Panel A) and MEVCB-TLR1-TLR2 (Panel B) to show the flexibility of the C-alpha atom of each residues in all the chains of the protein-protein complexes. The RMSF of the MEVC-B from both complexes were compared (Panel C) to identify the high peak regions in the plot which were caused by the loop regions in the MEVC-B structure (Panel D). The graphs were plotted using the XMGRACE plotting tool [71].
Table 8.
Analysis on the binding interactions at different time frame during the 100ns MD simulation of MEVCB-TLR4-MD-2 and MEVCB-TLR1-TLR2.
Fig 12.
Radius of gyration (Rg) values of the proteins.
The compactness of the proteins in MEVCB-TLR4-MD-2 (Panel A) and MEVCB-TLR1-TLR2 (Panel B) complexes were calculated using GROMACS and visualised in Rg value. The graphs were plotted using the XMGRACE plotting tool [71].