A docking study using atomistic conformers generated via elastic network model for cyclosporin A/cyclophilin A complex.

Anisotropic network model is used to generate a set of distinct conformations for cylophilin A (CypA). The native structure is deformed to different extents along each of the lowest-frequency modes (first 7 modes) both in negative and positive directions. Each node of the elastic network represents either a single atom in the high-resolution model or a single residue in the low-resolution model. Realistic conformations with energies close to or lower than the crystal structure and with satisfactory internal geometry are recovered by energy minimization using implicit solvation model. These conformations are then used for ensemble docking to the ligand cyclosporin A for both a further test of accuracy of generated conformers and exploration of different binding modes. Higher number of correctly docked ligands are obtained for conformations with low deformation factors as a result of lower root mean square distances with respect to crystal structure. Yet, surprisingly, the lowest binding energy is obtained for one of the highly deformed conformations as a result of its special contact with arginine side chain oriented towards binding site. Considering the fact that the cyclic ligand's backbone and protein's side chains are held rigid during docking, the conformers generated by high- and low-resolution elastic network models are almost equally successful in providing the correct binding mode. The shape of the binding pocket that incorporates crucial interaction sites for hydrogen bond formation is found to be another important determining factor for the success of the dock. Also, the small backbone variations of a few Angstroms in magnitude at the loop regions surrounding the binding pocket can cause amino acids' side chains to be displaced by magnitudes of up to 10 A and therefore have a strong influence on the efficiency of the conformational search during docking.