Molecular dynamics simulations and drug discovery

Abstract

This review discusses the many roles atomistic computer simulations of macromolecular (for example, protein) receptors and their associated small-molecule ligands can play in drug discovery, including the identification of cryptic or allosteric binding sites, the enhancement of traditional virtual-screening methodologies, and the direct prediction of small-molecule binding energies. The limitations of current simulation methodologies, including the high computational costs and approximations of molecular forces required, are also discussed. With constant improvements in both computer power and algorithm design, the future of computer-aided drug design is promising; molecular dynamics simulations are likely to play an increasingly important role.

Figure 1

The different conformations of the acetylcholine binding protein from Lymnaea stagnalis. Portions of the protein have been removed to facilitate visualization. (a) The protein in a closed conformation with nicotine bound (PDB ID: 1UW6), shown in blue. (b) The protein in an open conformation (PDB ID: 1YI5) with the same nicotine conformation superimposed on the structure, shown in pink. (c) Ribbon representations of the two conformations.

Figure 2

A schematic showing how a molecular dynamics simulation is performed. First, a computer model of the receptor-ligand system is prepared. An equation like that shown in Figure 3 is used to estimate the forces acting on each of the system atoms. The positions of the atoms are moved according to Newton's laws of motion. The simulation time is advanced, and the process is repeated many times. This figure was adapted from a version originally created by Kai Nordlund.

Figure 3

An example of an equation used to approximate the atomic forces that govern molecular movement. The atomic forces that govern molecular movement can be divided into those caused by interactions between atoms that are chemically bonded to one another and those caused by interactions between atoms that are not bonded. Chemical bonds and atomic angles are modeled using simple springs, and dihedral angles (that is, rotations about a bond) are modeled using a sinusoidal function that approximates the energy differences between eclipsed and staggered conformations. Non-bonded forces arise due to van der Waals interactions, modeled using the Lennard-Jones potential, and charged (electrostatic) interactions, modeled using Coulomb's law.

Figure 4

The thermodynamic cycle used for selecting alchemical transformations. Typically, one wishes to calculate the free energy of binding, ΔG_bind, shown across the top. However, it is generally impractical to run a molecular dynamics simulation long enough to capture an entire binding event. Instead, a series of alchemical transformations are performed using molecular dynamics simulations. ΔG_protein is the change in free energy that occurs when a bound ligand is 'annihilated'. ΔG is the change in free energy that occurs when an unbound 'ghost' ligand binds to the receptor; however, since a ghost ligand is not able to interact with any solvent or receptor atoms, this energy is always zero. Finally, ΔG_water is the change in free energy that occurs when an unbound ligand in solution is 'annihilated'. A system that proceeds from one state around this free-energy cycle only to return to the same initial state should have no change in total free energy; consequently, ΔG_bind = ΔG_water - ΔG_protein.