Role of CSF1R 550th-tryptophan in kusunokinin and CSF1R inhibitor binding and ligand-induced structural effect

Abstract

Binding affinity is an important factor in drug design to improve drug-target selectivity and specificity. In this study, in silico techniques based on molecular docking followed by molecular dynamics (MD) simulations were utilized to identify the key residue(s) for CSF1R binding affinity among 14 pan-tyrosine kinase inhibitors and 15 CSF1R-specific inhibitors. We found tryptophan at position 550 (W550) on the CSF1R binding site interacted with the inhibitors' aromatic ring in a π-π way that made the ligands better at binding. Upon W550-Alanine substitution (W550A), the binding affinity of trans-(-)-kusunokinin and imatinib to CSF1R was significantly decreased. However, in terms of structural features, W550 did not significantly affect overall CSF1R structure, but provided destabilizing effect upon mutation. The W550A also did not either cause ligand to change its binding site or conformational changes due to ligand binding. As a result of our findings, the π-π interaction with W550's aromatic ring could be still the choice for increasing binding affinity to CSF1R. Nevertheless, our study showed that the increasing binding to W550 of the design ligand may not ensure CSF1R specificity and inhibition since W550-ligand bound state did not induce significantly conformational change into inactive state.

Figure 1

π–π stacking from the bound ligand with W550 in CSF1R^WT. (A) Trans-(−)-kusunokinin (black), (B) Trans-(+)-kusunokinin (red), (C) Pexidartinib (blue), and (D) Imatinib (green) were depicted and the dash line indicated π–π interaction between the amino acid and the ligand. The protein and ligand structures were generated using Visual Molecular Dynamics (VMD) software version 1.9.3

Figure 2

Root-mean square distance (RMSD) plot of the protein backbone. (A) Trans-(−)-kusunokinin in complex with CSF1R^W550A (red) or CSF1R^WT (black). (B) Trans-(+)-kusunokinin in complex with CSF1R^W550A (red) or CSF1R^WT (black). (C) Pexidartinib in complex with CSF1R^W550A (red) or CSF1R^WT (black). (D) Imatinib in complex with CSF1R^W550A (red) or CSF1R^WT (black). (E) Ligand-free CSF1R^W550A (red) or CSF1R^WT (black). The values were calculated from three independent simulation systems and represented as mean values.

Figure 3

Root-mean square fluctuation (RMSF) plot of the α-carbon position in the protein backbone. (A) Trans-(−)-kusunokinin in complex with CSF1R^W550A (red) or CSF1R^WT (black). (B) Trans-(+)-kusunokinin in complex with CSF1R^W550A (red) or CSF1R^WT (black). (C) Pexidartinib in complex with CSF1R^W550A (red) or CSF1R^WT (black). (D) Imatinib in complex with CSF1R^W550A (red) or CSF1R^WT (black). (E) Ligand-free CSF1R^W550A (red) or CSF1R^WT (black). The values were calculated from three independent simulation systems and represented as mean values. Residue numbers were re-arranged according to AMBER procedure.

Figure 4

Alignment of ligand-free CSF1R kinase structure. Yellow structure represented CSF1R^WT and grey structure represented CSF1R^W550A. Blue sticks represented the native residue W550 of CSF1R^WT and red sticks represented the mutated residue A550 of CSF1R^W550A. Circle highlighted the substituted site. The protein structures were generated using Visual Molecular Dynamics (VMD) software version 1.9.3

Figure 5

Average distance from CSF1R kinase center to each residue backbone, calculated from the last 800 MD snapshots from the triplicated MD systems. Similar distance patterns suggested no conformational changes between the two models.

Figure 6

Distances between center of ligand and center of protein during simulations. (A) Ligand-CSF1R^WT models and (B) Ligand-CSF1R^W550A models. Distances between trans-(−)-kusunokinin to protein models were represented in black, trans-(+)-kusunokinin in red, imatinib in green and pexidartinib in blue. The values were calculated from three independent simulation systems and represented as mean values.

Figure 7

Difference in distance patterns upon ligand-binding between ligand-bound CSF1R^WT and ligand-bound CSF1R^W550A, calculated from the ligand-free models. (A) Trans-(−)-kusunokinin in complex with CSF1R^W550A (red) or CSF1R^WT (black). (B) Trans-(+)-kusunokinin in complex with CSF1R^W550A (red) or CSF1R^WT (black). (C) Pexidartinib in complex with CSF1R^W550A (red) or CSF1R^WT (black). (D) Imatinib in complex with CSF1R^W550A (red) or CSF1R^WT (black). The light blue highlighted the activation loop (AL) region. The zero line represented the position of each residue from ligand-free CSF1R models. Minus values indicated that residues moved closer to the protein center while plus values indicated that residues moved further from the protein center. Similar distance patterns suggested no conformational changes between the two models. Differences of distance pattern more than 2.0 Å were counted as significant conformation changes due to mutation W550A.

Figure 8

Concept for Distance Analysis. In the wild-type simulation, the distances from the center of structure to amino residues 1 and 2 are denoted by d1_n1 and d1_n2, whereas in the W550A simulation, the distances are denoted by d2_n1 and d2_n2. The varying distances indicated the position of the same amino acid in both wild-type and mutant structures. If d1_n1 is longer than d2_n1, it indicates that amino residue 1 is further from the center, vice versa. In this concept, a significant difference in distance may indicate that the amino residue is in a different position as well as conformational similarity. The protein structures were generated using Visual Molecular Dynamics (VMD) software version 1.9.3.