Fig 1.
PfIleRS model from Saccharomyces cerevisiae and Candida albicans templates.
(A) The colour bar showing the residue ranges defining each domain. (B) Homology model of PfIleRS showing cartoon representation of the functional domains: N-terminal Rossmann fold (NT-RF; blue), Zinc finger 1 hinge (ZnF1; yellow), Connective Peptide I (Editing domain; red), Zinc finger 2 hinge (ZnF2; yellow), Connective Peptide 2 and 3 (olive and orange), C-terminal Rossmann fold (CT-RF), Anticodon Binding Domain (ABD; purple) and Junctional domain (cyan). The HYGH and KMSKR signature motifs are shown in green and magenta spheres, respectively. (C) Multiple sequence alignment of PfIleRS target with 7D5C and 6LDK templates showing secondary structure predictions (number of alpha helices and beta sheets across the PfIleRS reference structure) with HYGH, ZnF1, ZnF2 and KMSKR conserved motifs in black blocks; the asterisks indicate the start and end of the modelling sequence [20,28,29].
Fig 2.
Potential orthosteric and allosteric pockets in PfIleRS and HsIleRS, and their predicted druggability attributes.
(A) PfIleRS orthosteric and allosteric pockets. (B) HsIleRS orthosteric and allosteric pockets. (C) Table of quality assessment parameters for each predicted pocket in PfIleRS: SiteScore, Size, DScore, Volume, Hydrophobicity, and Hydrophilicity. (D) Table of quality assessment parameters for each predicted pocket in HsIleRS.
Fig 3.
Validation and calibration of AutoDock Vina docking parameters.
(A) Redocking validation of AMP to PfIleRS (orange) aligned with modelled AMP (pink). (B) Redocking validation of HsIleRS using AMP (orange) aligned with modelled AMP (pink). (C) ROC curve used for docking parameters calibration, indicating the optimal point where all active inhibitors were identified (red curve peak). (D) Enrichment Curve for docking active inhibitors and decoys to PfIleRS, achieving a BEDROC of 0.76 during AutoDock Vina calibration (red curve).
Fig 4.
SANCDB selection pipeline for identifying potential allosteric modulators of PfIleRS.
(A) Blind docking to PfIleRS holo protein. (B) Docking of extracted potential allosteric compounds to HsIleRS for selectivity evaluation. (C) Redocking of selected compounds to PfIleRS to identify their binding pockets. (D) Selection criteria based on binding scores used to identify the nine potential hit compounds (coloured in red). (E) Two-dimensional representation of SANCDB potential hits against PfIleRS. (F) Bar graph showing the distribution of binding affinities of selected compounds in PfIleRS and HsIleRS. The binding pockets indicated in the bar are for PfIleRS.
Fig 5.
Protein-ligand interactions of PfIleRS allosteric pocket 1 potential SANCDB hit compounds.
All possible interactions formed between the PfIleRS pocket 1 and the SANCDB compounds are shown with their respective binding modes.
Fig 6.
Global trajectory analysis of PfIleRS and HsIleRS last 80 ns equilibrated portions.
(A) PfIleRS backbone-RMSD violin plots of the holo states and PfIleRS-SANC complexes. (B) HsIleRS backbone-RMSD violin plots for the holo and HsIleRS-SANC complexes. (C) Orthosteric pocket (masked residues defining the pocket) backbone-RMSD violin plots of holo and PfIleRS-SANC complexes. (D) Selected SANCDB compounds RMSD violin plots for 100 ns simulation.
Fig 7.
A) Holo average RMSF values mapped to structure, with largest RMSF residues (top 10%: 91 residues) shown as sticks and shown in bar plot. AMP and Ile ligands are coloured magenta and green, respectively. The largest RMSF regions are located primarily in loop regions of the editing and junctional domains. B) Delta RMSF heatmap. The blue colour indicates a decrease from the holo-averaged protein, while the red colour indicates an increase. Residues per system with the largest change in RMSF (top/bottom 3% delta: 54 residues) relative to holo are shown with an “x”. Residues in the heatmap not shown with an “x” are not in the top/bottom 3% in their respective system; these are included for comparison purposes. Junctional domain residues R1036 → T1054 are consistent in the holo replicates but have increased RMSF in all but two SANC systems. The largest active site RMSF change was seen in PfIleRS-SANC456, and then PfIleRS-SANC522.
Fig 8.
AMP binding free energy calculations and per-residue decomposition of holo and PfIleRS-SANC systems.
(A) AMP energy changes in the overall ΔGbind in the three holo states and ligand-bound systems upon allosteric modulation. (B) The holo-average top 2% residues (18 residues) contributing to the binding of AMP (<0.5Kcal/mol). (C) Delta per-residue decomposition of residues contributing to AMP binding in ligand-bound systems vs holo. Relative to the holo system, residues per system with the largest change in ΔGbind (top/bottom 3 residues) are shown with an “x”. Residues in the heatmap not shown with an “x” are not in the top/bottom 3 residues in their respective system; these are included for comparison purposes. The blue colour indicates a decrease from the holo-averaged protein, while the red colour indicates an increase.
Fig 9.
A) Holo average BC values mapped to structure, with largest BC residues (top 5%: 45 residues) shown as spheres on the structure, sticks in the enlarged window, and shown in a bar plot. AMP and Ile ligands are coloured magenta and green, respectively. A major allosteric pathway is seen, spanning the entire length of the enzyme, connecting the junctional domain to the active site and then the editing domain. B) Delta BC heatmap. The blue colour indicates a decrease from the holo-averaged protein, while the red colour indicates an increase. Residues per system with the largest change in BC (top/bottom 3% delta: 54 residues) relative to holo are shown with an “x”. Residues in the heatmaps not shown with an “x” are not in the top/bottom 3% in their respective system; these are included for comparison purposes. SANC ligand-induced BC changes of active site, zinc finger domain, and editing domain may be linked to enzyme functional change shown by the decreased binding affinity of AMP.
Fig 10.
A) Holo average EC values mapped to structure, with largest EC residues (top 5%: 45 residues) shown as spheres on the structure, sticks in the enlarged window, and shown in a bar plot. AMP and Ile ligands are coloured magenta and green, respectively. Two major groups/communities of residues are seen, indicating extensive residue communication in these groups which may be important for enzyme function. B) Delta EC heatmap. The blue colour indicates a decrease from the holo-averaged protein, while the red colour indicates an increase. Residues per system with the largest change in EC (top/bottom 3% delta: 54 residues) relative to holo are shown with an “x”. Residues in the heatmap not shown with an “x” are not in the top/bottom 3% in their respective system; these are included for comparison purposes. In most PfIleRS-SANC systems, the high EC active site group seen on the left in Fig 10A exhibited decreased EC while the high EC active site group on the right exhibited increased EC.
Fig 11.
The blue colour indicates a decrease from the holo-averaged protein, while the red colour indicates an increase. Residues per system with the largest change in CC (top/bottom 3% delta: 54 residues) relative to holo are shown with an “x”. Residues in the heatmap not shown with an “x” are not in the top/bottom 3% in their respective system; these are included for comparison purposes. Many SANC systems show decreased CC of zinc finger and active site residues, including the HYGH and KMSKR motifs.
Fig 12.
Superimposed structures of holo systems (green) and PfIleRS-SANC456 (delta colours) showing delta CC values. AMP and SANC ligands are coloured magenta and cyan, respectively. Holo and PfIleRS-SANC456 Ile ligands are coloured purple and pink, respectively. A) The structure of the active site, zinc fingers, and start of the editing domain are not aligned with the three holo replicates. The decreased CC (blue) is an indication that these regions are spending most of the trajectory further away from the geometric centre relative to the holo average, which corresponds to the structure clustering. B&C) The stable binding of SANC456 in potential allosteric pocket 1 may cause the structural and positional changes of α-helix 2 and α-helix 3 which may lead to the large structural changes/delta CC seen elsewhere.