Use of Crystal MD Simulations to Speed Up Evaluation of Binding Free Energies of Dimannose Deoxy Analogs With M4-P51g-Cyanovirin-N

Abstract

Molecular Dynamics (MD) presents an advanced tool for scoring of the binding free energies (ΔG) between a target protein and a set of candidate substrates, narrowed by extensive virtual screening process. Molecular mechanics(MM)/continuum model approach for evaluation of ΔG includes calculation of (i) a critically important solvation energy electrostatic contribution by means of solving the Poisson-Boltzmann (PB) or generalized Born (GB) equation and (ii) nonpolar component estimated from the solvent accessible area (SA) of solutes. Both, MM/PBSA and MM/GBSA, methods imply averaging of ΔG over a set of snapshots generated through, preferably, explicit solvent MD simulation which is the most time demanding step. In this work we demonstrated a possibility to reduce time of conformational sampling using crystal environment simulation. Amber10 program and FF99SB/GLYCAM06 force fields combination were employed for MD simulations and free energy calculations. Six 3, 4 and 3,4-deoxy dimannose analogs were studied as ligands of the m4-P51G-CVN mutant of the potent anti-HIV carbohydrate binding agent Cyanovirin-N (1). The use of crystal over solution simulations results in at least 8 times faster generation of the equivalent length trajectories. Binding free energy ΔA estimated from crystal NVT simulation trajectories shows 0.93 and 0.94 correlation with ΔG from solution NPT simulations for MM/PBSA and MM/GBSA approaches, respectively. We also evaluated performance of the relatively new GLYCAM06 carbohydrate force field and found reasonable agreement between calculated ΔG and experimental value. Results of this study further support our earlier hypothesis about importance (for CVN specificity) of the eight-component H-bond interactions of dimannose and protein main chain atoms that also was recently observed in experiment. [1] Vorontsov and Miyashita (2009) Biophys. J., 97.