Computational Prediction of the Mode of Binding of Antitumor Lankacidin C to Tubulin

Abstract

Lankacidin C, which is an antibiotic produced by the organism Streptomyces rochei, shows considerable antitumor activity. The mechanism of its antitumor activity remained elusive for decades until it was recently shown to overstabilize microtubules by binding at the taxol binding site of tubulin, causing mitotic arrest followed by apoptosis. However, the exact binding mode of lankacidin C inside the tubulin binding pocket remains unknown, an issue that impedes proper structure-based design, modification, and optimization of the drug. Here, we have used computational methods to predict the most likely binding mode of lankacidin C to tubulin. We employed ensemble-based docking in different software packages, supplemented with molecular dynamics simulation and subsequent binding-energy prediction. The molecular dynamics simulations performed on lankacidin C were collectively 1.1 μs long. Also, a multiple-trajectory approach was performed to assess the stability of different potential binding modes. The identified binding mode could serve as an ideal starting point for structural modification and optimization of lankacidin C to enhance its affinity to the tubulin binding site and therefore improve its antitumor activity.

Figure 1

Structure of (a) lankacidin C and (b) dictyostatin.

Figure 2

Taxol binding site of tubulin shows a great deal of induced fit, posing a real challenge in docking of new ligands. (a) Binding pocket changes shape to accommodate a variety of structurally diverse ligands, including paclitaxel (white, PDB ID: 3J6G(16)), discodermolide (green, PDB ID: 5LXT(17)), epothilone A (purple, PDB ID: 1TVK(18)), and dictyostatin (red, PDB ID: 5MF4(19)). (b) RMSD matrix of the binding pocket residues of the four different aforementioned tubulin PDB structures in angstrom.

Figure 3

Eight different binding modes selected through docking for (a) lankacidin C and (b) dictyostatin. The native binding mode for dictyostatin is shown together with the GOLD1 binding mode in (a), colored in cyan.

Figure 4

Heavy-atom RMSD matrix quantifying, in angstrom, the differences between the eight different binding poses after fitting the atoms of the binding pocket in the case of (a) lankacidin C and (b) dictyostatin.

Figure 5

RMSD change over the course of the simulation for the different binding modes of (a) lankacidin C and (b) dictyostatin. The black lines represent the RMSD of the C_α atoms of the protein. The red lines (usually higher in the plots) represent the RMSD of the ligand atoms. All RMSD values were obtained after least-square fitting of the C_α atoms of the protein and with respect to the initial structure before the simulations.

Figure 6

Results of the analysis of hydrogen bonds between lankacidin C and the protein along the 50 ns trajectories of the eight complexes. The x axis shows the frame number at a density of 100 frames/ns. The numbers to the right of each plot represent the persistence percentage of single, double, and triple hydrogen bonds (bottom to top) over the entire simulation time.

Figure 7

Interactions between the tubulin taxol binding site and (a) lankacidin C in the FRED1 binding mode or (b) dictyostatin in the native binding mode.