Atomistic Simulations of Ion Channels - Computational Neuroscience: A Comprehensive Approach - page 40

Figure 2.4
Snapshot of an ion channel from an MD simulation of alm K18 (see Section 2.3.3).
The 8 helices are shown as blue ribbons; the lipids are shown in purple, water in
blue. Chloride ions are green spheres, potassium ions red spheres. (See color insert.)
atom types), q i and q j are (partial) atomic charges, and r ij is the distance between
two atoms. Using this potential function the forces (the derivative of the potential
with respect to position) on all atoms in the system of interest are calculated and
used to solve classical equations of motions to generate a trajectory of all atoms in
time. An example of a simulation system is given in Figure 2.4. This system contains
a model ion channel, an explicit lipid bilayer, water, and salt.
This trajectory is the primary result of the simulation, from which specific details
of the system can be analyzed. This is an exciting idea, because atoms can be fol-
lowed as they move in real time on a timescale of up to ca. 100 ns, although longer
simulations have also been reported. In principle, any properties that depend on co-
ordinates, velocities or forces can be calculated, given sufficient simulation time. No
assumptions are required about the nature of the solvent, there is no need to choose
dielectric boundaries because all atoms are explicitly present, and in principle all
interactions (water-ions, water-protein, water-lipid, lipid-protein etc.) are incorpo-
rated. This method automatically includes the dynamics of the ion channel protein
itself as well as any dynamic effects of the lipids on the ion channel.
Molecular dynamics simulations have been applied by many groups, to a large
selection of ion channels as well as to homology models of complex ion channels [91,
111]. Although atomistic simulations have the significant advantage of providing a
detailed view of ion, water and protein dynamics, it remains challenging to link such
simulations with macroscopic observed properties. Molecular dynamics simulations