Molecular Dynamics Computations for Proteins: A Case Study in Membrane Ion Permeation

Abstract

Computer simulation can provide an atomic‐level view of protein structure and function that is unattainable with experiments alone. In this chapter, we demonstrate how molecular dynamics (MD) simulation can reveal the microscopic mechanisms of biomolecular activity. In particular, we illustrate the quantitative value of MD simulation by exploring ion channel proteins that allow selective permeation of charged molecules across cell membranes to control our nervous systems and chemical activity in the body. We begin by introducing the basic statistical mechanical formulation and methodological aspects of modern day MD simulations. We then discuss how to analyze a simulation to obtain mechanistic insight through calculation of various molecular properties. Special attention is given to quantitative thermodynamic calculations, which are the driving forces of protein function. We explain how to calculate free energies and equilibrium constants using potential of mean force (PMF) and free energy perturbation (FEP) techniques as well as the use of more advanced sampling approaches. These methods are then used to study ion permeation through the gramicidin A (gA) channel as well as the energetics of arginine side chain translocation across a lipid bilayer to assess models of voltage‐gated ion channel activation.

Atomistic Simulations of Membrane Ion Channel Conduction, Gating, and Modulation

“…In the simple case of water, correct gas-phase properties have been sacrificed to get accurate condensed-phase properties in most additive water models. 91 A water dipole changes dramatically between gas and water phases, and this cannot be included in a fixed charge model. A polarizable model, however, can simultaneously achieve accurate gas-phase (dipole moment, water structure, and energetics) and condensed-phase (density, heat of vaporization, and dielectric constant) properties.…”

“…One may also group interactions into contributions that help explain the process, such as direct interactions and indirect strain energies, involving subset of terms. 91 Such analyses help us understand how ion−ion, ion− ligand, and ligand−ligand energy terms contribute to complex formation within an ion channel. A fine balance of these large opposing contributions, determined by system geometry and force field parameters, may control ion conduction, as explained below, as well as some channel activation and inactivation mechanisms.…”

“…The first is to follow the time evolution of structure and to visualize the interactions of molecules over time, the second is to sample equilibrium configurations from phase space for the calculation of ensemble averages related to experimental observables, and the third is to force the system to follow a process of interest that may not otherwise occur in an unbiased simulation time frame. We first discuss the basic MD method, as it applies to membrane proteins, although several detailed texts on MD simulation can be found elsewhere. ,− The sole purpose of this section is to establish MD as an engine for generating accurate representations of the equilibrium configurations of the all-atom system to enable statistical averages and free energy estimation pertaining to studies of ion channel function.…”

“…In the simple case of water, correct gas-phase properties have been sacrificed to get accurate condensed-phase properties in most additive water models . A water dipole changes dramatically between gas and water phases, and this cannot be included in a fixed charge model.…”

“…One may further decompose energies by computing separate Coulomb and vdW contributions to particular interactions, or analyze related functions, such as packing scores with inverse r 6 weighting (analogous to the attractive part of the LJ potential, representing dispersion interactions, and the rate of magnetization transfer in NMR). One may also group interactions into contributions that help explain the process, such as direct interactions and indirect strain energies, involving subset of terms . Such analyses help us understand how ion–ion, ion–ligand, and ligand–ligand energy terms contribute to complex formation within an ion channel.…”