2.1. Global receptor Features do not significantly deviate in simulations of different holo states of InhA compared to the apo form. Comparison of the average backbone RMSD values of the apo (3.3 Å) and ligand-bound (1 – 2 Å) forms of InhA obtained from simulations of the docked structures points to the structural stability acquired after ligand binding (Figure 2B and Table S1), where all eight pyrazinone derivatives (I – VIII, Figure 1) provide greater stability to the receptor (average RMSD of 1.5 Å) compared to isoniazid (average RMSD of 2.0 Å, Table S1). More importantly, in all complexes, RMSD becomes uniform towards the end of simulation, (Figures S2 – S3), which also correlates with the RAC values that approach zero towards the end of each simulation (Figures 2A and S4 – S5), indicating structural convergence. Further, similar RoG values of the apo form (17.9 Å) and the ligand-bound forms (17.6 Å – 18.3 Å) suggest that despite dynamic fluctuations in RoG values over the simulation time, ligand binding does not significantly perturb the overall InhA structure (Table S2 and Figures 2D and S6 – S7). Furthermore, no significant differences are observed in the solvent accessible surface area of the apo and ligand-bound forms of the protein (Table S2 and Figures 2E and S8 – S9). Overall, this suggests that ligand-binding mostly affects the active site of InhA.
2.2. Change in flexibility within the InhA structure is significant in all the protein:ligand complexes compared to the apo form. Previous studies suggest that smaller scattering along the first two components obtained from principal component analysis indicates stable nature of the complex due to correlated motions [55, 56]. On similar lines, PCA analysis of our simulated complexes suggests that the complexed protein is more compact as compared to the apo protein (Figures S10 – S12), thus revealing reduced flexibility within the ligand binding site. Further, the movement on the PCA2 axis shows the negative loading [i.e., the scattering of dots (each dot showing the single conformation) is more towards negative side of vertical axis], compared to the apo protein (Figures S10 – S12). This suggests that all the compounds (I – VIII) would cause a reduced flexibility, resulting in inhibition of the target enzyme (Figures S10 – S12) [55]. Although the dynamics of PCs of all ligands is very similar, close comparison reveals greater movement of complexes containing pyrazinone derivatives towards negative axis compared to isoniazid, indicating their higher inhibition potential (Figure S12). Further, among the pyrazinone derivatives, the movement along the PCA2 axis in N4-ethyl substituted ligands (Figure S10) is more towards positive side in comparison to N4-methoxybenzyl substituted ligands (Figures 7 and S11). This indicates the better inhibition efficiency of the N4-methoxybenzyl substituted ligands towards InhA.
RMSF analysis reveals fluctuations at residues 100 – 120, 150 – 160 and 205 – 220 in the ligand bound form as well as the apo protein, although these fluctuations are comparatively less prominent in the ligand-bound protein (Figures 2C, S13 and S14). Further, these fluctuations are reduced in complexes of pyrazinone derivatives compared to isoniazid, which further substantiates the greater inhibitory potential of pyrazinone derivated compared to isoniazid (Figures S7 and S14). Furthermore, DCCM plots shows reduced anti-correlation between residues in complexes containing pyrazinone derivatives in comparison to the isoniazid:InhA complex, whereas the apo protein exhibit the highest degree of anti-correlated motion (Figures S15 – S19), which points to the attained stabilization upon ligand binding. Overall, these analyses provide clues to the inhibitory potential of pyrazinone derivations towards InhA.
2.3. Arg194 is the common anchor point for all ligands within the InhA binding pocket. Pre-simulation molecular docking calculations reveal that all pyrazinone derivatives (I – VIII, Figure 1) interact with InhA at a congruous binding site (Figure 3), that involves Arg194 as the common anchoring point (Figures 4 and 5). Further, the hydrogen-bonding interactions of Arg194 with each of these ligands is retained during simulations (45-99% occupancies, Figure 6) and are distinctly observed in the last simulation frame for most complexes (Figures 8 – 9). Furthermore, except complex I, the interaction of Arg194 with the ligand is retained even in the optimized isolated quantum chemical models of all complexes, which points to the intrinsically stable nature of these interactions (Figures 10 and 11). Although the docked structures reveal hydrogen bonding of the side chain of Arg194 with the carbonyl oxygen (complexes I, II, III and VI) or N1 (complexes IV, V, VII and VIII), MD simulations and QM calculations reveal the interaction of Arg194 with carboxylate and/or carbonyl oxygen atoms of the pyrazinone skeleton of each ligand (Figures 8 – 11). Altogether, this suggests that Arg194 plays an important role in binding of pyrazinone derivatives.
In addition to Arg194, the interaction with Ile193 is important for ligand binding. Specifically, although docking reveals that Ile193 interacts with only two ligands (IV and VI), simulations suggest significant (43.6% - 96.1%) occupancy of this interaction in complexes containing five (i.e., II, III, V, VII and VIII) of the eight ligands (Figure 6). Further, except the complex containing ligand VI, the interaction of Ile193 is retained in the last frame of the simulations of all complexes (Figures 8 – 9), suggesting that this residue plays an important role in active-site organization and ligand binding. This is further substantiated by QM calculations, which reveal intrinsically stable interaction of this residue with four (i.e., III, V, VII and VIII) ligands (Figures 10 – 11). More importantly, analysis of the structures from the last simulation frame of each complex reveals that although ligand:protein interactions slightly differ between the docked and simulated structures, all pyrazinone derivatives remain intact within the binding pocket till the end of simulation (Figures 4, 5, 8 and 9).
Docking further reveals that all N4-ethyl ligands containing an aromatic moiety at C6 (i.e. I, II and III) hydrogen bond with Arg194, Thr195 and Gln215, although the C6-chloro-substituted ligand (IV) additionally interacts with Ile193 (Figure 4). However, the four N4-methoxybenzyl substituted analogues (V – VIII) interact with Arg194, Thr195 and Met198, although ligands containing 2-thiophene (V), phenyl (VII) or chloro (VIII) substitution at C6 additionally hydrogen bond with Gln215 (Figure 5). Further, a salt bridge interaction is observed between the negative carboxylate group of each ligand and the positively charged guanidinium side chain of Arg194 (Table 1). Furthermore, MD simulations suggest that the hydrogen bonding occupancies of the complexes containing N4-methoxybenzyl substituted ligands (maximum occupancy of 97.2% for V, 76.9% for VI, 98.5% for VII and 72.2% for VIII) are comparable to that of complexes containing N4-ethyl substituted ligands (80.7% for I, 91.9% for II, 93.6% for III and 99.6% for IV, Figure 6). In comparison, isoniazid interacts at different binding site of InhA, and hydrogen bonds with Pro150, Met154, Pro155 and Thr161 (Figure S22), although the last MD frame and QM-optimized structure shows the interaction of isoniazid with Asp149, Tyr157 (Figure S23) and Asp149, Arg152, Pro155 (Figure S24), respectively.
The neutral (i.e., COOH) form of each ligand was also examined for comparison. Docking reveals that like the complexes containing the carboxylate-ionized form, Arg194 is the common anchoring residue for each ligand (Figure S22). Further, except IV, Arg194 interacts with all neutral ligands in MD simulations (80 – 90% occupancy, Figure S21) and QM-optimized structures (Figure S24). Similarly, the interaction with Ile193 was also observed in simulations (Figure S21) and QM optimized structures (Figure S24), and distinctly in the last simulation frame of most complexes (Figure S23). However, due to significant similarities in the interactions of neutral and ionized ligand forms and greater importance of the ionized ligand form at physiological pH, only the complexes containing ionized ligands are discussed in the subsequent sections.
2.4 Ligands show variable hydrophobic interactions with InhA. In addition to hydrogen bonding, the ligands form hydrophobic contacts with the receptor through their N4 and C6 moieties (Table 1). Molecular docking reveals that Met198 most commonly forms hydrophobic contacts with ligands (i.e., with II, III, IV, V, VII and VIII, Table 1), although this interaction is retained in last frame of simulation only by complex III (Table 1). Further, Ile 201 forms hydrophobic contacts with all the four docked N4-methoxybenzyl substituted ligands (V – VIII), where Tyr157 and Arg152 form additional hydrophobic contacts with three (V, VII and VIII) and two (VI and VII) ligands, respectively (Table 1). In contrast, except ligand IV, Thr195 interacts with N4-ethyl substituted ligands, whereas Ala197 additionally interacts with ligands I and III (Table 1). However, interaction of ligand with Ile201 in last frame of simulation is retained in complex V and VII only and introduced in complex II, whereas interaction with Thr195 is retained in complex II and III and interaction with Tyr157 is retained in complex VIII. Further, new hydrophobic interactions occur in simulations (interaction with Val237 in complex I; Pro155, Ile 201, Val 237 in complex II; Phe46, Met146 in complex V; Arg190, Ile193 in complex VII and Phe148 in complex VIII), while the interactions observed in docked structures are lost (Table 1). However, in case of isoniazid bound InhA, the interaction of Tyr157 with isoniazid is observed in both docked as well as simulated complex, along with interaction of Pro150 (Table S2). Overall, this analysis suggests that due to their weak nature, hydrophobic interactions show variable behaviour in simulations.
2.5. N4-methoxybenzyl Substituted Compounds show More Effective Binding with InhA. Docking reveals that N4-methoxybenzyl substituted ligands (V – VIII) show better receptor binding (docking scores of –7.4 to –8.2 kcal mol-1) than N4-ethyl substituted ligands (I – IV, docking scores of –5.4 to –6.5 kcal mol-1, Table 2), where all ligands show stronger binding than isoniazid (–5.4 kcal mol-1, Tables 2 and S4). Stronger binding of N4-methoxybenzyl substituted ligands is further substantiated by post-simulation MM-GBSA binding energy values which are significantly higher (–36.0 to –44.3 kcal mol-1) compared to N4-ethyl substituted compounds (–27.5 to –36.7 kcal mol-1, Table 2). On similar lines, QM calculations on reduced models reveal that the average binding strength of N4-methoxybenzyl substituted ligands (–241.5 kcal mol-1) is higher than four N4-ethyl substituted compounds (–175.6 kcal mol-1). Overall energy calculations (Docking score + MMGBSA binding energies + QM binding energies) summarise the enhanced stability of complexes containing N4-methoxybenzyl substituted ligands. Further, out of the four N4-methoxybenzyl substituted ligands, ligand VII with C6-chloro substitution show greatest stability (Table 2).