Born-Oppenheimer Ab Initio QM/MM Molecular Dynamics Simulations of Enzyme Reactions

There are two key requirements for reliably simulating enzyme reactions: one is a reasonably accurate potential energy surface to describe the bond-forming/breaking process as well as to adequately model the heterogeneous enzyme environment; the other is to perform extensive sampling since an enzyme system consists of at least thousands of atoms and its energy landscape is very complex. One attractive approach to meet both daunting tasks is Born-Oppenheimer ab initio QM/MM molecular dynamics (aiQM/MM-MD) simulation with umbrella sampling. In this chapter, we describe our recently developed pseudobond Q-Chem-Amber interface, which employs a combined electrostatic-mechanical embedding scheme with periodic boundary condition and the particle mesh Ewald method for long-range electrostatics interactions. In our implementation, Q-Chem and the sander module of Amber are combined at the source code level without using system calls, and all necessary data communications between QM and MM calculations are achieved via computer memory. We demonstrate the applicability of this pseudobond Q-Chem-Amber interface by presenting two examples, one reaction in aqueous solution and one enzyme reaction. Finally, we describe our established aiQM/MM-MD enzyme simulation protocol, which has been successfully applied to study more than a dozen enzymes.