Dynamic ligand-induced-fit simulation via enhanced conformational samplings and ensemble dockings: a survivin example.

Survivin is an anticancer drug target due to its overexpression in tumor cells in a homodimer form. Abbott Laboratories has identified a small molecule binding site near the dimerization interface in a high-throughput-screening (HTS)-NMR experiment. A benchmarking of the binding mode of the compound Abbott8 aided in the search for the ligand-induced-fit receptor structure by exploring the conformational space of the survivin dimer. We performed ensemble dockings with Abbott8 against a large set of conformations sampled via replica exchange molecular dynamics (REMD). This enhanced sampling allowed the reproduction of the holo-NMR experimental binding mode. Surprisingly, the major structural change in the best-REMD snapshot corresponding to the small molecule induced-fit happens in the so-called "survivin mitosis/apoptosis switch loop", consistent with the X-ray crystal structure of survivin-monomer/borealin/INCENP chromosomal passenger complex (CPC), as the distance between Phe93 and Phe101 increased. To verify this hypothetical pathway for the induced-fit conformational change, we utilized morphed intermediate structures that combined the X-ray data and the best-REMD snapshot, and the potential of mean force (PMF) of the survivin dimer was constructed with umbrella sampling (US) followed by a multiple Bennett acceptance ratio estimator (MBAR). It revealed a 3-4 kcal/mol free energy barrier along the reaction coordinate, and the complex is stabilized by the gain of the binding energies of Abbott8. This free energy barrier might prohibit the reproduction of the experimental binding mode from the regular NTP-MD ensemble docking that we had tried. The combination of REMD generalized ensemble sampling with ensemble docking and free energy pathway analysis may provide a novel research protocol for the simulation of protein-ligand induced-fit recognition.