Mechanosensitive channel MscS in the open state: modeling of the transition, explicit simulations, and experimental measurements of conductance

Abstract

Mechanosensitive channels of small conductance (MscS) are ubiquitous turgor pressure regulators found in many walled cells and some intracellular organelles. Escherichia coli MscS acting as a tension-activated osmolyte release valve shows a nonsaturable conductance (1.2 nS in a 39 mS/cm electrolyte) and weak preference for anions. Pursuing the transition pathways in this channel, we applied the extrapolated motion protocol (cycles of displacements, minimizations, and short simulations) to the previously generated compact resting conformation of MscS. We observed tilting and straightening of the kinked pore-forming TM3 helices during the barrel expansion. Extended all-atom simulations confirmed the stability of the open conformation in the bilayer. A 53 degrees spontaneous axial rotation of TM3s observed after equilibration increased the width and polarity of the pore allowing for stable voltage-independent hydration and presence of both cations and anions throughout the pore. The resultant open state, characterized by a pore 1.6 nm wide, satisfied the experimental conductance and in-plane expansion. Applied transmembrane electric field (+/-100 to +/-200 mV) in simulations produced a flow of both K(+) and Cl(-), with Cl(-) current dominating at higher voltages. Electroosmotic water flux strongly correlated with the chloride current (approximately 8 waters per Cl(-)). The selectivity and rectification were in agreement with the experimental measurements performed in the same range of voltages. Among the charged residues surrounding the pore, only K169 was found to contribute noticeably in the rectification. We conclude that (a) the barrel expansion involving tilting, straightening, and rotation of TM3s provides the geometry and electrostatics that accounts for the conductive properties of the open pore; (b) the observed regimen of ion passage through the pore is similar to electrodiffusion, thus macroscopic estimations closely approximate the experimental and molecular dynamics-simulated conductances; (c) increased interaction of the opposing ionic fluxes at higher voltages may result in selectivities stronger than measured near the reversal potential.

Figure 1.

The resting (closed) and open-like conformations derived from the crystal structure of MscS. The entire crystal structure of MscS (pdb 1mxm) shown as a diminished side view (A) and a view from the top (D) presented in the same scale as other models. The transmembrane barrel of MscS in the closed state (B and E), and expanded open-like state (C and F) obtained through extrapolations. Domain colors: TM1 (gold), TM2 (green), TM3a (teal), TM3b (blue and dark blue), reconstructed N-terminal segments (red). The bottom row shows the models in the same scale to illustrate the degree of compaction of the resting state relative to the splayed crystal structure and the expansion associated with opening. In the extrapolated open state (C and F) the gate-keeping leucines (yellow WdV spheres) protrude directly into the pore.

Figure 2.

Extrapolated expansion of the transmembrane barrel. Wire representations of the TM3 barrel in the initial (A) and expanded (B) conformations, with the side and top views. The gate position is represented by α carbons of residues L105 and L109 (balls). In B, 10 similar conformations obtained in independent extrapolated trajectories are shown as a bundle (thin blue lines) around the consensus conformation (thicker tube) closely retracing the average backbone position. Expansion trajectories presented in the coordinates of TM3a helical tilt versus pore diameter (C). Helical tilt was defined as the angle between the pore axis and TM3a helices, measured for the backbone of all residues between V96 to N112. The average surface-to-surface distance at the level of the gate was taken as the constriction diameter. The initial position is denoted with square (around the 7-Å 25° point); the groups of trajectories obtained with the amplification coefficients of 1 and 1.05 are show in blue and red, respectively. The rectangle around the 15-Å, 30° point designates the group of open-like states that satisfy the experimental restraints on the channel geometry. An in-plane expansion area versus conductance plot for nine consecutive conformations from a typical expansion trajectory is shown in D. The green rectangle shows the range for experimentally observed parameters within 10% deviation.

Figure 3.

Evolution of the open pore in unrestrained MD simulations. Characteristic frames from all-atom simulations of the relaxed resting (A) and open (B) states in a fully hydrated POPC bilayer in the presence of 0.2 M KCl. The coordinates for the resting state are taken from Anishkin et al. (2008). The protein is represented as solvent-accessible surface with areas designated as apolar (white), aliphatic lining of the gate (yellow), polar (green), negatively charged (red), and positively charged (blue). The Cl⁻ and K⁺ ions are shown as pink and cyan spheres, respectively. The hydrocarbon chains of lipids are shown as yellow wires. Panel C shows computed hydration energy profiles for the open pore lining before (red) and after (blue) 4-ns MD simulation of the open state. D and E show the reorientation of TM3s that led to the surface polarity increase illustrated in C: conserved leucines (L105, L109, L111, and L115) interact more tightly after a 4-ns simulation and brief symmetry annealing (E) as compared with the initial extrapolated model (D). The helices rotate clockwise, looking from the periplasm. Water-accessible surface inside the equilibrated open pore with two completely solvated ions (F).

Figure 4.

Electrostatics maps for the open conformation of the MscS barrel. The model obtained first through extrapolated motion was relaxed in a 4-ns unrestrained simulation in a POPC bilayer at 10 dyne/cm followed by 1-ns symmetric annealing (see Materials and methods). The channel with the softly restrained backbone was then additionally preequilibrated with 1.5 M KCl at zero voltage. The maps were averaged over 4 ns. (A) Distribution of electric potential due to the structural charges on the protein, totally unscreened by the medium. (B) The equilibrium densities of ions calculated inside the column (4 × 4 Å) positioned axially through the pore. The densities in the column do not reflect ion concentrations near the walls in wider vestibules. (C) Electrostatic potential in the system in the presence of lipids, water, and ions. Note that the color scale in C reflects a 40 times smaller variation of voltage compared with A. The scale is centered such that white color on C corresponds to the averaged potential in the aqueous bulk.

Figure 5.

Explicit simulations of ionic conductance through the transmembrane domain of MscS. The pore occupancy with ions in the course of two 16-ns simulations at +100 mV (A) and −100 mV (B). The occupancies were measured for the region of the pore (−12 > z > −32 Å) encompassing the gate. Panel C shows the cumulative charge transferred by ions passing through the gate region during these two simulations. The direction of ion movement is denoted by the traces. Integrals of ion and water passage events for two 8-ns simulations at +200 and −200 mV (D). The scale for the number of permeated water molecules is on the right and reflects ∼8 waters per ion. The bulk concentration of KCl is 1.5 M.

Figure 6.

A summary of conductance simulations and comparison with experiments. (A) Simulated I-V curve obtained without the cage domain (open circles), and curve position adjusted to represent the entire channel including the estimated resistance of the cytoplasmic cage (filled circles). The simulated curves can be compared with experimental I-V curves obtained on single channels (filled triangles) and multichannel patches (open triangles). The horizontal axis shows pipette voltages with positive branch representing hyperpolarizations. Note that measurements on multichannel patches allowed to explore a wider range of voltages (open triangles, see Fig. 7 and legend for technical details). (B) The ratio of Cl⁻ to K⁺ ions occupying the gate region of the pore (triangles) and the fraction of current carried by Cl⁻ simulated at different transmembrane voltages (diamonds). (C) Electroosmotic water flux associated with Cl⁻ current computed between −200 and +200 mV.

Figure 7.

Experimental measurements of conductive properties of MscS in native spheroplast membranes in a wide range of membrane potentials. (A) A typical response of channel population to a ramp of pressure at +40 mV pipette voltage. Single-channel steps are seen in the magnified trace. (B) Single-channel I-V curves obtained in symmetrical 0.2 and 1.5 M KCl in excised inside-out patches. The currents were normalized to the specific bulk conductivites of the recording buffer. Current–voltage relationships measured in an extended range of voltages using the “pulse” regimen. (C) Ensemble current measurements. A response of a patch to a train of 50-ms pulses. The pipette pressure was gradually ramped up to a saturating level, and then a train of voltage pulses was applied. The magnified top part of one current response was fit with an exponent, whose extrapolation shows the magnitude of the current at the onset of the pulse. (D) I-V plots for WT MscS obtained by the single-channel (filled symbols) and ensemble (open symbols) methods in 0.2 and 1.5 M KCl. (E) I-V plots for WT and K169Q MscS measured with the single-channel method. The inset shows the entire curves, whereas the main panel presents the magnified view of the lower left quadrant where the WT and K169Q curves deviate the most. The small error bars represent standard deviation based on three independent experiments.