Computational protocol: Dynamic water patterns change the stability of the collapsed filter conformation of the KcsA K+ channel

Similar protocols

Protocol publication

[…] The simulations were conducted by the NAMD molecular simulation package [] and the results were analyzed by VMD [] and displayed by Chimera []. We used PDB files 3F7V and 1K4D for the collapsed filter structures and 1K4C for the conductive filter structure for the KcsA K+ channel. The CHARMM force field parameters were used, c22 for proteins [] with CMAP corrections included [], and c27 for lipids [, ]. Standard parameters for ions were used, and the modified parameters for carbonyl oxygen and cation interactions were included [–]. Each channel structure was embedded in the POPC lipid bilayer and then put in a box filled with the TIP3P water molecules. The KCl concentration was set to 0.15 M in each simulation box. In each structure, the residue E71 of each chain was protonated, which was reported as an important factor in the KcsA channel simulations []. The all-atom molecular dynamics simulations were conducted under the constant temperature of 310 K and constant pressure of 1 atm. The temperature was controlled by the Langevin thermostat with the damping coefficient of 1 ps-1, and the pressure was controlled by the Nose-Hoover Langevin piston with the period of 200 ps and the damping time constant of 50 ps. Particle Mesh Ewald method was used for the electrostatic interactions. The nonbonded interactions were truncated at 12 Å and the switching functions were applied starting at 10 Å. Bonds involving the hydrogen atoms were constrained using the SHAKE algorithm. Simulations were conducted with the time step of 2 fs. The r-RESPA multiple time step scheme was used, where the short-range nonbonded forces were updated every 2 fs and the long-range electrostatic forces were updated every 4 fs.The filter structure in the 3F7V file has the collapsed conformation. After initial equilibrations, we pulled the K+ ion at the S4 site in the outward direction with the constant force of 3.5 nN applied every 10 fs until it moved to the S2 site that switched the collapsed filter to the conductive conformation. This structure was equilibrated for 3 ns by restraining the structure containing the conductive filter conformation (residues T74 to L81) with the harmonic restrains where the force constant was set to 1 kcal·mol-1·Å-2. Then we placed three K+ ions at the S0, S2, and S4 sites and two water molecules at the S1 and S3 sites in this conductive filter. After 2 ns equilibration, we pulled K+ upward out of the selectivity filter of the 3F7V channel structure, and observed that in some case the water molecule percolated deeply behind the collapsed filter following the filter conformational switch process. Using these results, we placed water molecules in the different patterns behind the collapsed filter chains. All the patterns were randomly designed by arbitrarily placing one to three water molecules over the three percolating sites behind each filter chain. After equilibrations, we selected eight patterns for the study and denoted them as models M1-M8 in the paper, where K+ ions were placed in the S1 and S4 sites in each collapsed filter. Models M9 and M10 directly used the PDB files 1K4D and 1K4C as the initial structures, where the K+ ions and the water molecules were placed in the filters as shown in , respectively. The ten models were equilibrated further for 5–10 ns. And in some cases, we applied the extra force to restrain the K+ ions in the S1 and S4 sites inside the filter. From the simulations examined, we observed that the patterns composed of a small number of water molecules changed, typically more water molecules moved behind the collapsed filter bearing few percolated waters. This can happen over several nanoseconds. In addition, the ions and the water occupancies inside the selectivity filter may also change. Thus to compare the model structures as designed, we used only short equilibration runs in the analyses (5–10 ns).To compare the filter stabilities of the ten models, the K+ ions inside the filter were pulled upward through the filter under the constant force (ranged from 2 to 4.5 nN) applied every 10 fs. The force was turned off when the ion moved out of the filter, defined as the location of the ion above the center of Cα of D80. The procedure conducted under each force value was repeated six times for each model system. And the average value of ln(Iapp/Iapp,max) versus the force curves were fitted using the natural logarithm of the equation: IappIapp,max=11+e−β⋅(F−F1/2)⋅d(1) Where F1/2 describes the force required when the current I reaches one half of the maximum value, and d is the apparent distance that the ion moves along under the given force. β equals 1/kBT, where kB is Boltzmann’s constant and T is the Kelvin temperature. Note that the curves fitted by cannot be compared to the experimental data due to two reasons. First, the forces used in the simulations are large and are applied to the filter ions every 10 fs (in order to relax the local structure), which is different from the real experiment. Second, we only study the single ion permeation event at the initial stage of the filter recovering from the collapsed conformation whereas in experiment multiple ions can continuously go through the filter after the filter begins to conduct ions. Thus the current I studied here has the different meaning than that measured in the experiment, so we use the symbol Iapp. Therefore, the curves fitted by are used only to compare the stabilities of the different collapsed filter conformations that undergo the procedures in the same condition. […]

Pipeline specifications

Software tools NAMD, VMD
Application Protein structure analysis