- Email: [email protected]
Exploring the Gating Mechanism in the ClC Chloride Channel via Metadynamics
doi:10.1016/j.jmb.2006.06.034
J. Mol. Biol. (2006) 361, 390–398
Exploring the Gating Mechanism in the ClC Chloride Channel via Metadynamics Francesco Luigi Gervasio 1 ⁎, Michele Parrinello 1 , Matteo Ceccarelli 2 and Michael L. Klein 3 1
Computational Science, Department of Chemistry and Applied Biosciences, ETH Zürich, USI Campus, Via Giuseppe Buffi 13, CH-6900 Lugano, Switzerland 2
Department of Physics and CNR-SLACS University of Cagliari, Campus Monserrato IT-09042 Monserrato, Italy 3
Center for Molecular Modeling and Department of Chemistry, University of Pennsylvania, Philadelphia, PA 19104-6323, USA
Computer simulations have been used to probe the gating mechanism in the Salmonella serovar typhimurium chloride channel (st-ClC). Specifically, the recently developed metadynamics methodology has been exploited to construct free energy surfaces as a function of the positions of either one or two chloride ions inside the pore, the position and protonation state of the key E148 residue, and the number of water molecules coordinating the translocating ions. The present calculations confirm the multi-ion mechanism in which an ion-push-ion effect lowers the main barriers to chloride ion translocation. When a second anion is taken into account, the barrier for chloride passage through the E148 narrow region is computed to be 6 kcal/ mol in the wild-type channel, irrespective of the protonation state of the E148 residue, which is shown to only affect the entrance barrier. In the E148A mutant, this barrier is much lower, amounting to 3 kcal/mol. The metadynamics calculations reported herein also demonstrate that before reaching the periplasmic solution, chloride ions have to overcome an additional barrier arising from two different effects, namely the rearrangement of their solvation shell and a flip in the backbone angles of the residues E148 and G149, which reside at the end of the αF helix. © 2006 Elsevier Ltd. All rights reserved.
*Corresponding author
Keywords: chloride translocation; ClC channel; gating; molecular dynamics; metadynamics
Introduction ClC chloride channels are expressed in a wide range of organisms, and play a role in various biological functions. In eukarya they regulate transepithelial transport of ions, the electrical excitability of the plasma membrane in skeletal muscle cell and acidification of the stomach and intracellular vesicles.1,2 In bacteria, ClC channels are probably involved in acid shock resistance 3 and act as a pump, exchanging two Cl− anions for one proton.4 In the last decade many salient properties of the ClC-type channels have been clarified. Among these is the unique manner in which the conduction is turned on or off, a process defined as “fast gating”. Electrophysiological studies have shown that the Abbreviations used: WT, wild-type; st-ClC, Salmonella serovar typhimurium chloride channel; FES, free energy surface; CV, collective variables. E-mail address of the corresponding author: [email protected]
fast gating is voltage-dependent and the opening of the channel is facilitated by chloride ions in the extracellular solution, 5–8 and possibly by low external pH.7–9 At physiological pH the fast gating is activated by the permeating anion in what is seemingly a coupled conduction and gating mechanism.5,10 Moreover the observation of an anomalous mole fraction in a mixed solution of chloride and nitrate anions suggested a multi-ion conduction process.5 The recent publication of the crystal structure of a bacterial ClC chloride channel in closed11 and open form10 revealed a complex geometry of the channel and suggested possible gating mechanisms. The structure is homodimeric, with each monomer having 18 α-helices arranged in an antiparallel architecture that forms an intracellular and extracellular vestibule separated by a narrow and substantially curved pore approximately 15 Å long. In the closed structure the pore was found blocked by a glutamate (E148). This evidence, taken together with the open nature of the two mutants (E148A and E148Q), provides the link between the
0022-2836/$ - see front matter © 2006 Elsevier Ltd. All rights reserved.
ClC Chloride Channel Gating Mechanism
glutamate residue E148 and the gating mechanism. Additional support is provided by the fact that, at low external pH, both the mutant and the wild-type channel are more conductive8,9 However, a recent crystallographic study has shown that pH has no effect on the structure and ion binding properties of Ec-ClC. 12 In the mutant structures, where the channel is in the open state, three chloride ions are inside the channel: one close to the intracellular entrance to the pore (Sint), one at the filter binding site (Scen), and one in place of the E148 (Sext). Understanding the permeation mechanism of bacterial ClC channels at an atomistic level is of great importance in its own right, since they are one of the few currently available types of ion channel structures, and as such may thus serve as a prototype for understanding the analogous eukaryotic channels, given their extensive homology. Indeed, there is good evidence that both the ClC channels and the Ec-ClC transporters have conserved modes of ion binding.13–15 The availability of the crystal structure made possible the computational study of the conduction mechanism of the transporter.16–21 These studies clarified many aspects of the mechanism. The favorable docking positions of the chloride inside the pore were studied in depth as well as the qualitative electrostatic effects of strongly conserved charged residues.21,18 The multi-ion nature of the conduction process was verified,21 a king-of-the-hill mechanism proposed 16 in which two Cl ions compete for a favorable position at the pore center and free energy profiles evaluated.16,17,20 Despite such studies, questions still remain either unanswered or only partially answered. One relevant question is the role played by the residues R147 and E148 in the gating mechanism. It is known that the anion conductance is affected by the charge state of R147,22,23 since mutations of the homologous residues in ClC-1 and ClC-0 were found to alter the anionic selectivity sequence and to increase the cation permeability or to render the channel dis-functional. In agreement with these experiments the Monte Carlo simulations of Miloshevsky and Jordan found that, when R147 is neutralized by proton transfer to E148, an energy barrier to the ion permeation appears. This result confirms the importance of the strictly conserved positive charge. They also found that a mutation of E148 (E148A and E148Q) as well as its neutralization, creates an electrostatic trap that blocks the anion at a mid-membrane position. Building on this, they propose that the displacement of the trapped anion is made possible by a new ion that comes from the periplasm and pushes the first Cl−. The overall result is similar to the king-of-the-hill mechanism proposed by Cohen and Schulten,16 in which a lone ion bound to the center of the ClC pore is pushed out by a second ion that takes its place. The difference between the two proposals is that in the latter case the protonation (and maybe deprotonation) of E148 plays a fundamental role.
391 The role of E148 protonation is also underlined in the study by Bostick and Berkowitz.20 who calculated by molecular dynamics/umbrella-sampling the potential of mean-force of an anion penetrating the channel from the periplasm. Their potential of mean force is in agreement with the results of Miloshevsky and Jordan and predicts that the protonation of E148 dramatically lowers the barrier to the anion entrance and forms an electrostatic trap. Both of these studies predict a high permeation barrier for the wild-type (WT) channel, while in the study by Cohen and Schulten the barrier is predicted to be only a few kcal/mol. This difference could be due to the fact that in the latter case the channel was prepared in an open state and assumes a king-ofthe-hill scenario by pushing simultaneously two ions. None of the above mentioned studies addressed explicitly the role of the ion hydration/dehydration process, whose importance has been underlined among others by Roux et al.21 and in a different channel by Warshel et al.24 Moreover the role of possible conformational changes of residues in the pore region was not considered. Accordingly, in the present study by using classical molecular dynamics (MD) and taking advantage of the recently developed metadynamics methodology25–27 the movement of the chloride from the proximity of the cytoplasmic site to the external solution and vice versa has been probed in the Salmonella st-ClC channel and in its E148A mutant. Simulations have also been performed with two different protonation states for the E148 residue. The present results confirm the previously postulated multi-ion conduction mechanism in which an ion-push-ion effect lowers the main barriers to chloride ion translocation. In the WT channel there are two barriers to the translocation of the Cl−. The first is met between Scen and Sext and is 6 kcal/mol high in the presence of an ionpush-ion effect and the formation of a salt-bridge between E148 and R147. The latter plays a key role in the opening of the E148 gate that allows the translocation of the Cl−. The second barrier is found between Sext and the periplasmic solution both in the WT and in the E148A mutant and it is due to the need for rearrangement of the ion’s solvation shell and a flip in the backbone of the residues E148 and G149. Since St-ClC is a transporter exchanging two Cl− anions for one proton, the effect of the protonation of E148 was studied in order to understand whether or not there is any coupling between H+ and Cl− . The present simulations indicate that protonation of E148 induces an asymmetry between the entry of the chloride from Sext to Scen and vice versa. Indeed the Cl− enters the protonated pore without barrier, but once inside it is trapped by a strong electrostatic interaction. The mutation E148A has the effect of eliminating this barrier both for the ingoing and outgoing process (Figure 1).
392
ClC Chloride Channel Gating Mechanism
Figure 1. (a) Top view of the ClC channel showing the chloride ions. Water molecules and the POPC membrane have been removed for clarity. (b) An actual snapshot of the simulation showing the selectivity filter of the St-ClC, the two αhelices pointing toward the narrow region and two chloride ions occupying Sint and Scen, respectively.
ClC Chloride Channel Gating Mechanism
Results and Discussion In a preliminary metadynamics run the hydration of the pore was found to vary as the Cl− moved towards the exit. This observation pointed to the fact that the initial hydration of the pore was likely unsatisfactory. However, on repeating several times the calculation of moving through the pore new pairs of Cl−s, eventually a situation is achieved where the degree of hydration and the free energy surface (FES) did not vary significantly. Subsequent metadynamics trajectories described below were run starting from this structure. From Scen to Sext Since previous studies have discussed multiple occupancy of the selectivity filter,5,28,12 and a kingof-the-hill mechanism had already been proposed, it is instructive to use metadynamics as a vehicle to compare the permeation of a single Cl− with the correlated permeation of two Cl ions. The choice of collective variables (CVs) in the former case is the Z of one chloride and the Z of the center of mass of the key residue E148 and in the latter case the Z of the two chlorides that in the following will be denoted as Z1 and Z2, for upper and lower ion, respectively. The calculations underpinning the FESs reported in Figure 2 have been repeated in a number of independent runs. These gave the same results within the expected accuracy of the method,27 which in the present case, due to the slow diffusion of the chosen CVs, is 2 kcal/mol. These runs were started from a pre-equilibrated and consistently hydrated system where one Cl− is in Scen and the
393 second, when present, is in Sint. In the first set of metadynamics runs, only one ion occupies the channel. As the energy wells are filled the upper ion is pushed toward Sext but finds a barrier around Z1 = −1.5 Å that is approximately the experimental position of E148. Eventually the energy added is sufficient to push open the E148 gate and the ion translocates to Sext (Figure 2(a)). The barrier of a lone ion passing through the gate is 10 kcal/mol. Once the hydrogen bonds from the residue E148 to the backbone N-H are broken, the metadynamics trajectories show that E148 rapidly forms a stable salt-bridge with R147. This observation could explain how the charge state of R147 affects the conductance of the channels.22,23 In the second set of metadynamic runs two ions are in the pore. As the energy wells are being filled the lower ion (Z2) is pushed toward Scen, the upper (Z1) toward Sext. Again a barrier is found around Z1 = −1 Å due to the presence of E148. Eventually, the first ion is pushed over the barrier from Scen to Sext. The minimum free energy path (in yellow in Figure 2(b)) shows that when one ion pushes the other, the barrier is 6 kcal/mol. If the lower ion is far from the upper the barrier to the translocation is 8 kcal/mol, slightly lower that that obtained with a single chloride ion. Given the high error bar on the barriers derived from metadynamics runs and to quantify with more precision the barrier to E148 opening, we also performed a 2D umbrella sampling with two ions in the pore using as CVs the projection along the pore axis of the center of mass two chloride ions and the projection along the pore axis of the position of the carboxy group of E148. The result is shown in Figure 3. In this case the lower ion
Figure 2. Free energy surface as a function of (a) the projection of the distance from the centroid of the channel along the vertical axis of one chloride ion Z1 and of the E148 residue ZE148 and (b) the positions of two chloride ions Z1 and Z2 occupying the pore simultaneously. The isosurfaces are one every two kcal/mol and the white regions are the most energetically favorable. The total metadynamics trajectories spanned 4 ns. The approximate position of Sint, Scen and Sext are reported. In (a) only one ion is in the pore occupying preferentially Scen. In (b) the minimum free energy path is shown as a yellow arrow. This path corresponds to a coordinated movement of the two ions with the lower ion going from Sint to Scen pushing the upper from Scen to Sext. The red arrow shows the path corresponding to the upper ion translocating without the ion-push-ion effect.
394
Figure 3. Free energy surface obtained by umbrella sampling as a function of the projection along the pore axis of the center of mass of two chloride ions and the projection along the pore axis of the position of the carboxy group of E148. The isosurfaces are one every kcal/mol and the deep-blue regions are the most energetically favorable. The total sampling trajectories spanned 15.2 ns. The deeper minimum corresponds to the ions in Sint, Scen and the E148 gate closed.
remains in Sint and the resulting barrier of 9 kcal/ mol is consistent with that obtained with metadynamics when the ion-push-ion effect is absent (Figure 2(b) red arrow). Moreover we can quantify with greater accuracy the barrier to E148 opening as being 3–4 kcal/mol. From the comparison of the two FESs we can estimate that the ion-push-ion mechanism accounts for a lowering of the barrier 4 kcal/mol. The opening of the E148 gate itself and the slight rearrangement of water molecules and residues in the pore account for the remaining 6 kcal/mol.
ClC Chloride Channel Gating Mechanism
By inspecting Figure 2(b) we notice that in the FES a region around (−4,−4) is populated. This does not mean that the two Cl ions are overlapping, since their separation always remains greater than 3 Å. Instead the lower Cl− was pushed into a lateral pocket. A similar phenomenon was observed by Cohen and Schulten and discarded as an artifact. However, we consistently found this Cl− location in many metadynamics runs. To better understand the origin of the barrier to the ion translocation, we reconstructed the FES as a function of Z1 and the distance of the E148 carboxylic carbon from the Cζ of residue R147. The new FES confirms that the barrier to the ion translocation from Scen to Sext, in the presence of a second ion in the channel, is due mainly to the breaking of the hydrogen bonds with the backbone followed by the formation of a salt-bridge with R147 (the gate opening) that contributes 6 kcal/mol. The effect of E148A mutation and E148 protonation As in laboratory experiments, much can be learned from mutation and protonation experiments. For this reason, the next focus of attention was the E148A mutant and the WT where E148 was neutralized, which is expected to play a major role as evidenced by the previous discussion as well as both experimental and theoretical evidence. The result obtained by running metadynamics as a function of Z1 and Z2 distances in the WT channel with a neutralized E148 is shown in Figure 4. The protonation of E148 has a clear effect on the FES of chloride translocation as can be appreciated from a comparison of Figures 2(b) and 4(a). In the neutralized channel the minimum at 1,−5 is shallower and a new minumum appears at −2.5,−7. This is due to an ion-trap effect and the concurrent
Figure 4. (a) The free energy surface as a function of Z1 and Z2 coordinates of chloride ions in a WT channel in which the residue E148 is protonated. The isosurfaces are one every two kcal/mol. The metadynamics trajectory was run for 3.7 ns. (b) A scheme illustrating the free energy profile along the reactive trajectory translocating chloride ions from Sint, Scen to Scen,Sext in the WT and E148H pores.
395
ClC Chloride Channel Gating Mechanism
lowering of the gate opening barrier. Overall the profile of the reaction is very different (Figure 4(b)). The barrier to the chloride entrance in the pore from the external solution is lower, in agreement with the findings of Bostick and Berkowitz,20 who found that the barrier to chloride translocation through this narrow region of the pore is strongly dependent on the protonation state of E148. To strengthen this conclusion we reran the metadynamics simulations starting with the upper ion in Sext with both E148 protonated and deprotonated. These calculations confirmed that the barrier to the ion translocation from Scen to Sext depends on the protonation state of E148, being 6 kcal/mol in the deprotonated E148 WT channel and less than 4 kcal/mol in the protonated E148 WT channel. The effect of the E148A mutation was studied by several metadynamics runs using different CVs. In each case the barrier to the translocation of the chloride from Scen to Sext was less than 3 kcal/mol. Moreover, the electrostatic trap here is exactly at the position formerly occupied by E148 and stabilizes the chloride ion in Sext more than in Scen (see Figure 6). This is in agreement with the results of Lobet and Dutzler, obtained by measuring the anomalous difference electron density of the E148Q Ec-ClC mutant at high Br− concentration.12
the chloride ion from Sext to the external water solution. An example of this effect is shown in Figure 6 where the FES of the E148A mutant was calculated as a function of the Z position and the number of water molecules coordinating the chloride ion. This barrier is 9 kcal/mol. When repeated, with three chloride ions in the pore starting from Sint, Scen and Sext, respectively the barrier is ≃5 kcal/ mol. Also here the multiple occupancy of the pore has a clear effect. The ion-push-ion mechanism contributes to a lowering of the barrier of 4 kcal/ mol. By analyzing the FESs and the trajectories it was found that this barrier is due to the rearrangement of the water in the solvation shell, clearly appreciable in Figure 6, and to a flip of the φ and ψ angles of the residues 148 and G149. The backbone of these residues initially has the typical dihedral angles of an α-helix and is oriented in such a way to expose the N-H toward Sext. After the dihedral transition the backbone angles assume values typical of a β-sheet and point the C = O towards Sext (see Figure 5). Also, in agreement with the calculations of Bostick and Berkowitz,20 a favorable position for the chloride ion is observed around Z = 11 (Sout), which corresponds to a situation where the chloride is coordinated to R147.
From Sext to the water solution
Conclusions
During the various metadynamics runs that were performed both on the WT and the E148A mutant, invariably a barrier was found to the translocation of
The recently developed metadynamics methodology has been employed to explore chloride ion translocation from the proximity of the cytoplasmic
Figure 6. Shown is the free energy surface as a function of the Z1 coordinate and the number of waters solvating the chloride ion inside the pore of the E148A mutant. In (a) and (b) the pore is occupied by one and three ions, respectively. The isosurfaces are two per kcal/mol. The total metadynamics trajectory was run for 4 ns. When the upper chloride ion is in Sext, Z1 has a value around 0.5. In (a) it is apparent that the barrier from Scen to Sext (Z1 = −4 and 0.5, respectively) is 1– 2 kcal/mol. A larger barrier is found from Sext to the external solution. In (b) where the pore is occupied by multiple ions the FES is flatter and the barrier to the chloride exit is lower.
396
ClC Chloride Channel Gating Mechanism
Figure 5. Two snapshots of the simulation showing the selectivity filter of the St-ClC E148A mutant with a chloride ion (a) in Sext and (b) in the periplasmic water solution. As the chloride ion translocates to the periplasmic solution, the backbone of the residues A148 and G149 (yellow tube) undergoes a conformational change. An important effect of this change is that the C = O dipole of the residue 148 changes direction.
site to the external solution and vice versa for two different protonation states of the residue E148 and for the mutant E148A. The barrier to chloride ion exit through the E148 narrow region is computed to be 6 kcal/mol in the wild-type channel when there is a second chloride ion in the pore pushing from below. In the absence of a second ion, the barrier is 10 kcal/mol, confirming the importance of the multiple occupancy of ions in the filter, a finding that is in agreement with recent experiments.12 The protonation of E148 has a clear effect on the free energy profile and lowers the barrier to the chloride entrance in the pore from the external solution in agreement with previous calculations.20,18 In the E148A mutant the barrier from Scen to Sext was found to be considerably lower than in the wildtype. It was also found that, before reaching the periplasmic solution, the chloride ion has to overcome an additional barrier coming from the rearrangement of the solvation shell and of the residues 148 and G149 located at the end of the αF helix. Also, this barrier is lowered when the pore is occupied by multiple ions. Overall the present calculations are strongly suggestive of a multi-ion conduction mechanism in which an ion-push-ion effect lowers the main barriers to chloride translocation.
Methods All of the present MD simulations were performed using the MD program ORAC, 29 along with the CHARMM27 all atoms force field30 and the TIP3P water
model.31 The initial geometry was obtained from the crystal structure of Salmonella serovar typhimurium ClC (st-ClC) refined at 3.0 Å resolution (PDB code: 1KPL).11 The N-terminal segment of one of the two monomers (chain A) is missing from the crystal structure. This was reconstructed by duplicating the residues 12–32 from chain B and splicing them into chain A. The water molecules inside the narrow pore were manually added after visual inspection. The protein was then placed with the pore oriented along the Z axis of the simulation cell that contained a pre-equilibrated water solvated POPC bilayer. To do so, all the water molecules and POPC chains closer than 1.0 Å from the protein were eliminated. To minimize the number of atoms the simulation cell was chosen to be hexagonal, since the dimeric channel has approximately a rhomboidal shape. The final equilibrated system contains 12,170 TIP3P water molecules and 136 POPC lipids. In addition to the two crystallographic chloride ions, 30 Cl− and 14 K+ ions were added at random positions in the solution to neutralize the charge of the protein and add a total ion concentration of 100 mM. The resulting 69,000 atoms system was then equilibrated in the NPT ensemble for 6 ns in the hexagonal cell with periodic boundary conditions and particle mesh Ewald electrostatics. The NPT ensemble was simulated with the Parrinello-Rahman barostat32,33 and the Nosé-Hoover thermostat34 keeping pressure and temperature at 1 atm and 300 K, respectively. A further thermalization run was performed in the canonical ensemble for 3 ns at 300 K. In the thermalized structure, the Cα atoms showed an RMSD of 1.9 Å from the crystallographic structure. The metadynamics methodology25 was used throughout to accelerate rare events and to reconstruct the FES. In brief, the method is a dynamics in the space of collective coordinates that are evolved with a standard restrained MD supplemented by a history-dependent potential. A Lagrangian metadynamics run consists of a standard MD
397
ClC Chloride Channel Gating Mechanism
run in which we impose harmonic restraints on a set of CVs that can be any function of the atomic positions. The harmonic restraints connect the real CVs to additional variables that feel the effect of the previous history of the run by means of a Gaussian repulsive potential. The Gaussian repulsive potential is added in the current position of the CVs every T time steps. For a more detailed description of metadynamics, the interested reader should refer to the literature.25–27 The principal advantages of using metadynamics are that it is able to escape local minima by overcoming large free energy barriers and makes it possible to reconstruct FES in the space of the chosen CVs. This is possible since as shown25,27 in the limit of a long metadynamics, the sum of the repulsive Gaussian potentials and the FES tends to become flat as a function of the CVs. When this happens the sum of the repulsive Gaussian potentials gives the negative image of the FES within a controlled error that is inversely proportional to the square root of the diffusion coefficient of the CVs and directly proportional to the speed of Gaussian deposition.27 This method has already been applied with success to study events similar to the translocation of Cl−, namely the translocation of antibiotics through a porin35 and the entrance of the tetramethylammonium ion in the acetylcholinesterase gorge.36 Herein metadynamics has been employed to study the cytoplasmic gating mechanism of the St-ClC channel. To this aim, different sets of CVs were employed. Since the orientation of the channel is such that the pore is approximately along the Z-axis, the main CV used was the projection along the Z-axis of the distance of the chloride ions from the centroid of the channel dimer and POPC bilayer (Z). With this convention the residue E148 has a value of Z that is 0 Å in the crystal structure and fluctuates between 0 and 1 Å in the MD when closed; using the nomenclature of Dutzler et al.10 this value corresponds approximately to Sext. An estimate for the expected values of Scen and Sint can be obtained by measuring in the MD run the average position of the residues to which Cl− is experimentally known to be coordinated when occupying those positions.10 This yields a value of Z = −4 for Scen and Z = −10 for Sint. In the case of Scen a value of Z = −4 is also obtained if we superimpose the experimental and relaxed structure by minimizing the RMSD between the Cαs. Metadynamics runs were performed using as CVs the Z of two chloride ions occupying simultaneously one of the pores in the protonated and unprotonated E148 WTchannel as well as in the E148A mutant. Separate metadynamics runs were performed using as CVs the Z of one chloride ion and the coordination number of the E148 with respect to the H-bonded residues, the Z of one chloride ion and the distance of E148 carboxylic carbon from the Cζ of R147, the Z of one chloride ion and the number of coordinating water molecules and the Z of one chloride ion and Z of the E148 residue. The height of the Gaussians was chosen to be 1.0 kj/mol, the deposition time T, 1 ps. A multiple walker version of the metadynamics algorithm was used throughout.37 To obtain an independent measure of the FES and to confirm the results obtained by metadynamics, we performed a 2D umbrella sampling38 on the WT St-ClC using as CVs the projection along the pore axis of the center of mass two chloride ions and the projection along the pore axis of the position of the carboxy group of E148. In each of the sampling simulations each of the CVs was tethered by virtual springs acting along the Z axis with spring constants of 25 kj/mol per Å2. The tethering points for the umbrella potential were never distributed >1 Å apart requiring 32 simulations of 300 ps each and 14
simulations of 400 ps each to sample the region of interest. Starting configurations were chosen from the metadynamics runs and employed after a re-equilibration at 300 K. The resulting FES is in agreement with that previously obtained by metadynamics.
Acknowledgements We acknowledge the Swiss National Supercomputing Center for computer time. M.L.K. thanks the National Institutes for Health for generous support under GM 40712. M.C. thanks the Caspur Computer Center (Rome, Italy) for using its facilities.
References 1. Estèvez, R. & Jentsch, T. J. (2002). Clc chloride channels: correlating structure with function. Curr. Opin. Struct. Biol. 12, 531–539. 2. Maduke, M., Miller, C. & Mindell, J. A. (2000). A decade of clc chloride channels: structure, mechanism, and many unsettled questions. Annu. Rev. Biophys. Biomol. Struct. 29, 411–438. 3. Iyer, R., Iverson, T. M., Accardi, A. & Miller, C. (2002). A biological role for prokaryotic clc chloride channels. Nature, 419, 715–718. 4. Accardi, A. & Miller, C. (2004). Secondary active transport mediated by a prokaryotic homologue of clc cl- channels. Nature, 427, 803–807. 5. Pusch, M., Ludewig, U., Rehfeldt, A. & Jentsch, T. J. (1995). Gating of the voltage-dependent chloride channel clc-0 by the permeant anions. Nature, 373, 527–531. 6. Chen, Y. T. & Miller, C. (1996). Nonequilibrium gating and voltage dependence of the clc-0 cl- channel. J. Gen. Physiol. 108, 237–250. 7. Rychkov, G. Y., Pusch, M., Roberts, M. L. & Bretag, A. H. (2001). Interaction of hydrophobic anions with the rat skeletal muscle chloride channel clc-1: effects on permeation and gating. J. Physiol. 530, 379–393. 8. Chen, M. F. & Chen, T. Y. (2001). Different fast-gate regulation by external Cl− and H+ of the muscle-type clc chloride channels. J. Gen. Physiol. 118, 23–32. 9. Hanke, W. & Miller, C. (1983). Single chloride channels from torpedo electroplax. Activation by protons. J. Gen. Physiol. 82, 25–45. 10. Dutzler, R., Campbell, E. B. & MacKinnon, R. (2003). Gating the selectivity filter in clc chloride channels. Science, 300, 108–112. 11. Dutzler, R., Campbell, E. B., Cadene, M., Chait, B. T. & MacKinnon, R. (2002). X-ray structure of a clc chloride channel at 3.0 å reveals the molecular basis of anion selectivity. Nature, 415, 287–294. 12. Lobet, S. & Dutzler, R. (2006). Ion-binding properties of the clc chloride selectivity filter. EMBO J. 25, 24–33. 13. Chen, M. F. & Chen, T. Y. (2003). Side-chain charge effects and conductance determinants in the pore of clc-0 chloride channels. J. Gen. Physiol. 122, 133–145. 14. Estèvez, R., Schroeder, B. C., Accardi, A., Jentsch, T. J. & Pusch, M. (2003). Conservation of chloride channel structure revealed by an inhibitor binding site in clc-1. Neuron, 38, 47–59.
398
ClC Chloride Channel Gating Mechanism
15. Engh, A. M. & Maduke, M. (2005). Cysteine accessibility in clc-0 supports conservation of the clc intracellular vestibule. J. Gen. Physiol. 125, 601–617. 16. Cohen, J. & Schulten, K. (2004). Mechanism of anionic conduction across clc. Biophys. J. 86, 836–845. 17. Corry, B., O'Mara, M. & Chung, S.-H. (2004). Conduction mechanisms of chloride ions in clc-type channels. Biophys. J. 86, 846–860. 18. Miloshevsky, G. V. & Jordan, P. C. (2004). Anion pathway and potential energy profiles along curvilinear bacterial clc Cl− pores: electrostatic effects of charged residues. Biophys. J. 86, 825–835. 19. Yin, J., Kuang, Z., Mahankali, U. & Beck, T. L. (2004). Ion transit pathways and gating in clc chloride channels. Proteins: Struct. Funct. Genet. 57, 414–421. 20. Bostick, D. L. & Berkowitz, M. L. (2004). Exterior site occupancy infers chloride-induced proton gating in a prokaryotic homolog of the clc chloride channel. Biophys. J. 86, 1686–1696. 21. Faraldo-Gomez, J. D. & Roux, B. (2004). Electrostatics of ion stabilization in a clc chloride channel homologue from Escherichia coli. J. Mol. Biol. 339, 981–1000. 22. Fahlke, C., Yu, H. T., Beck, C. L., Rhodes, T. H. & George, J. A. L. (1997). Pore-forming segments in voltage-gated chloride channels. Nature, 390, 825–835. 23. Lin, C. W. & Chen, T. Y. (2000). Cysteine modification of a putative pore residue in clc-0. implication for the pore stoichiometry of clc chloride channels. J. Gen. Physiol. 116, 535–546. 24. Braun-Sand, S., Burykin, A., Chu, Z. T. & Warshel, A. (2005). Realistic simulations of proton transport along the gramicidin channel: demonstrating the importance of solvation effects. J. Phys. Chem. B, 109, 583–592. 25. Laio, A. & Parrinello, M. (2002). Escaping free energy minima. Proc. Natl Acad. Sci. USA, 20, 12562–12566. 26. Gervasio, F. L., Laio, A. & Parrinello, M. (2005). Flexible docking in solution using metadynamics. J. Am. Chem. Soc. 127, 2600–2607. 27. Laio, A., Rodriguez-Fortea, A., Gervasio, F. L., Ceccarelli, M. & Parrinello, M. (2005). Assessing the accuracy of metadynamics. J. Phys. Chem. B, 109, 6714–6721.
28. Fahlke, C., Rosenbohm, A., Mitrovic, N., George, A. L., Jr & Rudel, R. (1996). Mechanism of voltagedependent gating in skeletal muscle chloride channels. Biophys. J. 71, 695–706. 29. Procacci, P., Darden, T., Paci, E. & Marchi, M. (1998). Orac: a molecular dynamics program to simulate complex molecular systems with realistic electrostatic interactions. J. Comput. Chem. 18, 1848-1862. 30. MacKerell, A. D., Jr, Bashford, D., Bellot, M., Dunbrack, R. L., Jr, Evanseck, J. D., Field, M. J. et al. (1998). All-atom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B, 102, 3586–3616. 31. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L. (1983). Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79, 926–935. 32. Parrinello, M. & Rahman, A. (1981). Polymorphic transitions in single crystals: a new molecular dynamics method. J. Appl. Phys. 52, 7182. 33. Parrinello, M. & Rahman, A. (1980). Crystal structure and pair potentials: a molecular-dynamics study. Phys. Rev. Letters, 45, 1196. 34. Nose, S. (1984). A molecular dynamics method for simulations in the canonical ensemble. Mol. Phys. 52, 255–268. 35. Ceccarelli, M., Danelon, C., Laio, A. & Parrinello, M. (2004). Microscopic mechanism of antibiotics translocation through a porin. Biophys. J. 87, 58–64. 36. Branduardi, D., Gervasio, F. L., Cavalli, A., Recanatini, M. & Parrinello, M. (2005). The role of the peripheral anionic site and cation-π interactions in the ligand penetration of the human ache gorge. J. Am. Chem. Soc. 127, 9147–9155. 37. Raiteri, P., Laio, A., Gervasio, F. L., Micheletti, C. & Parrinello, M. (2006). Efficient reconstruction of complex free energy landscapes by multiple walkers metadynamics. J. Phys. Chem. B, 110, 3533–3539. 38. Roux, B. (1995). The calculation of the potential of mean force using computer simulations. Comp. Phys. Commun. 91, 275–282.
Edited by M. Levitt (Received 8 March 2006; received in revised form 12 June 2006; accepted 14 June 2006) Available online 30 June 2006
J. Mol. Biol. (2006) 361, 390–398
Exploring the Gating Mechanism in the ClC Chloride Channel via Metadynamics Francesco Luigi Gervasio 1 ⁎, Michele Parrinello 1 , Matteo Ceccarelli 2 and Michael L. Klein 3 1
Computational Science, Department of Chemistry and Applied Biosciences, ETH Zürich, USI Campus, Via Giuseppe Buffi 13, CH-6900 Lugano, Switzerland 2
Department of Physics and CNR-SLACS University of Cagliari, Campus Monserrato IT-09042 Monserrato, Italy 3
Center for Molecular Modeling and Department of Chemistry, University of Pennsylvania, Philadelphia, PA 19104-6323, USA
Computer simulations have been used to probe the gating mechanism in the Salmonella serovar typhimurium chloride channel (st-ClC). Specifically, the recently developed metadynamics methodology has been exploited to construct free energy surfaces as a function of the positions of either one or two chloride ions inside the pore, the position and protonation state of the key E148 residue, and the number of water molecules coordinating the translocating ions. The present calculations confirm the multi-ion mechanism in which an ion-push-ion effect lowers the main barriers to chloride ion translocation. When a second anion is taken into account, the barrier for chloride passage through the E148 narrow region is computed to be 6 kcal/ mol in the wild-type channel, irrespective of the protonation state of the E148 residue, which is shown to only affect the entrance barrier. In the E148A mutant, this barrier is much lower, amounting to 3 kcal/mol. The metadynamics calculations reported herein also demonstrate that before reaching the periplasmic solution, chloride ions have to overcome an additional barrier arising from two different effects, namely the rearrangement of their solvation shell and a flip in the backbone angles of the residues E148 and G149, which reside at the end of the αF helix. © 2006 Elsevier Ltd. All rights reserved.
*Corresponding author
Keywords: chloride translocation; ClC channel; gating; molecular dynamics; metadynamics
Introduction ClC chloride channels are expressed in a wide range of organisms, and play a role in various biological functions. In eukarya they regulate transepithelial transport of ions, the electrical excitability of the plasma membrane in skeletal muscle cell and acidification of the stomach and intracellular vesicles.1,2 In bacteria, ClC channels are probably involved in acid shock resistance 3 and act as a pump, exchanging two Cl− anions for one proton.4 In the last decade many salient properties of the ClC-type channels have been clarified. Among these is the unique manner in which the conduction is turned on or off, a process defined as “fast gating”. Electrophysiological studies have shown that the Abbreviations used: WT, wild-type; st-ClC, Salmonella serovar typhimurium chloride channel; FES, free energy surface; CV, collective variables. E-mail address of the corresponding author: [email protected]
fast gating is voltage-dependent and the opening of the channel is facilitated by chloride ions in the extracellular solution, 5–8 and possibly by low external pH.7–9 At physiological pH the fast gating is activated by the permeating anion in what is seemingly a coupled conduction and gating mechanism.5,10 Moreover the observation of an anomalous mole fraction in a mixed solution of chloride and nitrate anions suggested a multi-ion conduction process.5 The recent publication of the crystal structure of a bacterial ClC chloride channel in closed11 and open form10 revealed a complex geometry of the channel and suggested possible gating mechanisms. The structure is homodimeric, with each monomer having 18 α-helices arranged in an antiparallel architecture that forms an intracellular and extracellular vestibule separated by a narrow and substantially curved pore approximately 15 Å long. In the closed structure the pore was found blocked by a glutamate (E148). This evidence, taken together with the open nature of the two mutants (E148A and E148Q), provides the link between the
0022-2836/$ - see front matter © 2006 Elsevier Ltd. All rights reserved.
ClC Chloride Channel Gating Mechanism
glutamate residue E148 and the gating mechanism. Additional support is provided by the fact that, at low external pH, both the mutant and the wild-type channel are more conductive8,9 However, a recent crystallographic study has shown that pH has no effect on the structure and ion binding properties of Ec-ClC. 12 In the mutant structures, where the channel is in the open state, three chloride ions are inside the channel: one close to the intracellular entrance to the pore (Sint), one at the filter binding site (Scen), and one in place of the E148 (Sext). Understanding the permeation mechanism of bacterial ClC channels at an atomistic level is of great importance in its own right, since they are one of the few currently available types of ion channel structures, and as such may thus serve as a prototype for understanding the analogous eukaryotic channels, given their extensive homology. Indeed, there is good evidence that both the ClC channels and the Ec-ClC transporters have conserved modes of ion binding.13–15 The availability of the crystal structure made possible the computational study of the conduction mechanism of the transporter.16–21 These studies clarified many aspects of the mechanism. The favorable docking positions of the chloride inside the pore were studied in depth as well as the qualitative electrostatic effects of strongly conserved charged residues.21,18 The multi-ion nature of the conduction process was verified,21 a king-of-the-hill mechanism proposed 16 in which two Cl ions compete for a favorable position at the pore center and free energy profiles evaluated.16,17,20 Despite such studies, questions still remain either unanswered or only partially answered. One relevant question is the role played by the residues R147 and E148 in the gating mechanism. It is known that the anion conductance is affected by the charge state of R147,22,23 since mutations of the homologous residues in ClC-1 and ClC-0 were found to alter the anionic selectivity sequence and to increase the cation permeability or to render the channel dis-functional. In agreement with these experiments the Monte Carlo simulations of Miloshevsky and Jordan found that, when R147 is neutralized by proton transfer to E148, an energy barrier to the ion permeation appears. This result confirms the importance of the strictly conserved positive charge. They also found that a mutation of E148 (E148A and E148Q) as well as its neutralization, creates an electrostatic trap that blocks the anion at a mid-membrane position. Building on this, they propose that the displacement of the trapped anion is made possible by a new ion that comes from the periplasm and pushes the first Cl−. The overall result is similar to the king-of-the-hill mechanism proposed by Cohen and Schulten,16 in which a lone ion bound to the center of the ClC pore is pushed out by a second ion that takes its place. The difference between the two proposals is that in the latter case the protonation (and maybe deprotonation) of E148 plays a fundamental role.
391 The role of E148 protonation is also underlined in the study by Bostick and Berkowitz.20 who calculated by molecular dynamics/umbrella-sampling the potential of mean-force of an anion penetrating the channel from the periplasm. Their potential of mean force is in agreement with the results of Miloshevsky and Jordan and predicts that the protonation of E148 dramatically lowers the barrier to the anion entrance and forms an electrostatic trap. Both of these studies predict a high permeation barrier for the wild-type (WT) channel, while in the study by Cohen and Schulten the barrier is predicted to be only a few kcal/mol. This difference could be due to the fact that in the latter case the channel was prepared in an open state and assumes a king-ofthe-hill scenario by pushing simultaneously two ions. None of the above mentioned studies addressed explicitly the role of the ion hydration/dehydration process, whose importance has been underlined among others by Roux et al.21 and in a different channel by Warshel et al.24 Moreover the role of possible conformational changes of residues in the pore region was not considered. Accordingly, in the present study by using classical molecular dynamics (MD) and taking advantage of the recently developed metadynamics methodology25–27 the movement of the chloride from the proximity of the cytoplasmic site to the external solution and vice versa has been probed in the Salmonella st-ClC channel and in its E148A mutant. Simulations have also been performed with two different protonation states for the E148 residue. The present results confirm the previously postulated multi-ion conduction mechanism in which an ion-push-ion effect lowers the main barriers to chloride ion translocation. In the WT channel there are two barriers to the translocation of the Cl−. The first is met between Scen and Sext and is 6 kcal/mol high in the presence of an ionpush-ion effect and the formation of a salt-bridge between E148 and R147. The latter plays a key role in the opening of the E148 gate that allows the translocation of the Cl−. The second barrier is found between Sext and the periplasmic solution both in the WT and in the E148A mutant and it is due to the need for rearrangement of the ion’s solvation shell and a flip in the backbone of the residues E148 and G149. Since St-ClC is a transporter exchanging two Cl− anions for one proton, the effect of the protonation of E148 was studied in order to understand whether or not there is any coupling between H+ and Cl− . The present simulations indicate that protonation of E148 induces an asymmetry between the entry of the chloride from Sext to Scen and vice versa. Indeed the Cl− enters the protonated pore without barrier, but once inside it is trapped by a strong electrostatic interaction. The mutation E148A has the effect of eliminating this barrier both for the ingoing and outgoing process (Figure 1).
392
ClC Chloride Channel Gating Mechanism
Figure 1. (a) Top view of the ClC channel showing the chloride ions. Water molecules and the POPC membrane have been removed for clarity. (b) An actual snapshot of the simulation showing the selectivity filter of the St-ClC, the two αhelices pointing toward the narrow region and two chloride ions occupying Sint and Scen, respectively.
ClC Chloride Channel Gating Mechanism
Results and Discussion In a preliminary metadynamics run the hydration of the pore was found to vary as the Cl− moved towards the exit. This observation pointed to the fact that the initial hydration of the pore was likely unsatisfactory. However, on repeating several times the calculation of moving through the pore new pairs of Cl−s, eventually a situation is achieved where the degree of hydration and the free energy surface (FES) did not vary significantly. Subsequent metadynamics trajectories described below were run starting from this structure. From Scen to Sext Since previous studies have discussed multiple occupancy of the selectivity filter,5,28,12 and a kingof-the-hill mechanism had already been proposed, it is instructive to use metadynamics as a vehicle to compare the permeation of a single Cl− with the correlated permeation of two Cl ions. The choice of collective variables (CVs) in the former case is the Z of one chloride and the Z of the center of mass of the key residue E148 and in the latter case the Z of the two chlorides that in the following will be denoted as Z1 and Z2, for upper and lower ion, respectively. The calculations underpinning the FESs reported in Figure 2 have been repeated in a number of independent runs. These gave the same results within the expected accuracy of the method,27 which in the present case, due to the slow diffusion of the chosen CVs, is 2 kcal/mol. These runs were started from a pre-equilibrated and consistently hydrated system where one Cl− is in Scen and the
393 second, when present, is in Sint. In the first set of metadynamics runs, only one ion occupies the channel. As the energy wells are filled the upper ion is pushed toward Sext but finds a barrier around Z1 = −1.5 Å that is approximately the experimental position of E148. Eventually the energy added is sufficient to push open the E148 gate and the ion translocates to Sext (Figure 2(a)). The barrier of a lone ion passing through the gate is 10 kcal/mol. Once the hydrogen bonds from the residue E148 to the backbone N-H are broken, the metadynamics trajectories show that E148 rapidly forms a stable salt-bridge with R147. This observation could explain how the charge state of R147 affects the conductance of the channels.22,23 In the second set of metadynamic runs two ions are in the pore. As the energy wells are being filled the lower ion (Z2) is pushed toward Scen, the upper (Z1) toward Sext. Again a barrier is found around Z1 = −1 Å due to the presence of E148. Eventually, the first ion is pushed over the barrier from Scen to Sext. The minimum free energy path (in yellow in Figure 2(b)) shows that when one ion pushes the other, the barrier is 6 kcal/mol. If the lower ion is far from the upper the barrier to the translocation is 8 kcal/mol, slightly lower that that obtained with a single chloride ion. Given the high error bar on the barriers derived from metadynamics runs and to quantify with more precision the barrier to E148 opening, we also performed a 2D umbrella sampling with two ions in the pore using as CVs the projection along the pore axis of the center of mass two chloride ions and the projection along the pore axis of the position of the carboxy group of E148. The result is shown in Figure 3. In this case the lower ion
Figure 2. Free energy surface as a function of (a) the projection of the distance from the centroid of the channel along the vertical axis of one chloride ion Z1 and of the E148 residue ZE148 and (b) the positions of two chloride ions Z1 and Z2 occupying the pore simultaneously. The isosurfaces are one every two kcal/mol and the white regions are the most energetically favorable. The total metadynamics trajectories spanned 4 ns. The approximate position of Sint, Scen and Sext are reported. In (a) only one ion is in the pore occupying preferentially Scen. In (b) the minimum free energy path is shown as a yellow arrow. This path corresponds to a coordinated movement of the two ions with the lower ion going from Sint to Scen pushing the upper from Scen to Sext. The red arrow shows the path corresponding to the upper ion translocating without the ion-push-ion effect.
394
Figure 3. Free energy surface obtained by umbrella sampling as a function of the projection along the pore axis of the center of mass of two chloride ions and the projection along the pore axis of the position of the carboxy group of E148. The isosurfaces are one every kcal/mol and the deep-blue regions are the most energetically favorable. The total sampling trajectories spanned 15.2 ns. The deeper minimum corresponds to the ions in Sint, Scen and the E148 gate closed.
remains in Sint and the resulting barrier of 9 kcal/ mol is consistent with that obtained with metadynamics when the ion-push-ion effect is absent (Figure 2(b) red arrow). Moreover we can quantify with greater accuracy the barrier to E148 opening as being 3–4 kcal/mol. From the comparison of the two FESs we can estimate that the ion-push-ion mechanism accounts for a lowering of the barrier 4 kcal/mol. The opening of the E148 gate itself and the slight rearrangement of water molecules and residues in the pore account for the remaining 6 kcal/mol.
ClC Chloride Channel Gating Mechanism
By inspecting Figure 2(b) we notice that in the FES a region around (−4,−4) is populated. This does not mean that the two Cl ions are overlapping, since their separation always remains greater than 3 Å. Instead the lower Cl− was pushed into a lateral pocket. A similar phenomenon was observed by Cohen and Schulten and discarded as an artifact. However, we consistently found this Cl− location in many metadynamics runs. To better understand the origin of the barrier to the ion translocation, we reconstructed the FES as a function of Z1 and the distance of the E148 carboxylic carbon from the Cζ of residue R147. The new FES confirms that the barrier to the ion translocation from Scen to Sext, in the presence of a second ion in the channel, is due mainly to the breaking of the hydrogen bonds with the backbone followed by the formation of a salt-bridge with R147 (the gate opening) that contributes 6 kcal/mol. The effect of E148A mutation and E148 protonation As in laboratory experiments, much can be learned from mutation and protonation experiments. For this reason, the next focus of attention was the E148A mutant and the WT where E148 was neutralized, which is expected to play a major role as evidenced by the previous discussion as well as both experimental and theoretical evidence. The result obtained by running metadynamics as a function of Z1 and Z2 distances in the WT channel with a neutralized E148 is shown in Figure 4. The protonation of E148 has a clear effect on the FES of chloride translocation as can be appreciated from a comparison of Figures 2(b) and 4(a). In the neutralized channel the minimum at 1,−5 is shallower and a new minumum appears at −2.5,−7. This is due to an ion-trap effect and the concurrent
Figure 4. (a) The free energy surface as a function of Z1 and Z2 coordinates of chloride ions in a WT channel in which the residue E148 is protonated. The isosurfaces are one every two kcal/mol. The metadynamics trajectory was run for 3.7 ns. (b) A scheme illustrating the free energy profile along the reactive trajectory translocating chloride ions from Sint, Scen to Scen,Sext in the WT and E148H pores.
395
ClC Chloride Channel Gating Mechanism
lowering of the gate opening barrier. Overall the profile of the reaction is very different (Figure 4(b)). The barrier to the chloride entrance in the pore from the external solution is lower, in agreement with the findings of Bostick and Berkowitz,20 who found that the barrier to chloride translocation through this narrow region of the pore is strongly dependent on the protonation state of E148. To strengthen this conclusion we reran the metadynamics simulations starting with the upper ion in Sext with both E148 protonated and deprotonated. These calculations confirmed that the barrier to the ion translocation from Scen to Sext depends on the protonation state of E148, being 6 kcal/mol in the deprotonated E148 WT channel and less than 4 kcal/mol in the protonated E148 WT channel. The effect of the E148A mutation was studied by several metadynamics runs using different CVs. In each case the barrier to the translocation of the chloride from Scen to Sext was less than 3 kcal/mol. Moreover, the electrostatic trap here is exactly at the position formerly occupied by E148 and stabilizes the chloride ion in Sext more than in Scen (see Figure 6). This is in agreement with the results of Lobet and Dutzler, obtained by measuring the anomalous difference electron density of the E148Q Ec-ClC mutant at high Br− concentration.12
the chloride ion from Sext to the external water solution. An example of this effect is shown in Figure 6 where the FES of the E148A mutant was calculated as a function of the Z position and the number of water molecules coordinating the chloride ion. This barrier is 9 kcal/mol. When repeated, with three chloride ions in the pore starting from Sint, Scen and Sext, respectively the barrier is ≃5 kcal/ mol. Also here the multiple occupancy of the pore has a clear effect. The ion-push-ion mechanism contributes to a lowering of the barrier of 4 kcal/ mol. By analyzing the FESs and the trajectories it was found that this barrier is due to the rearrangement of the water in the solvation shell, clearly appreciable in Figure 6, and to a flip of the φ and ψ angles of the residues 148 and G149. The backbone of these residues initially has the typical dihedral angles of an α-helix and is oriented in such a way to expose the N-H toward Sext. After the dihedral transition the backbone angles assume values typical of a β-sheet and point the C = O towards Sext (see Figure 5). Also, in agreement with the calculations of Bostick and Berkowitz,20 a favorable position for the chloride ion is observed around Z = 11 (Sout), which corresponds to a situation where the chloride is coordinated to R147.
From Sext to the water solution
Conclusions
During the various metadynamics runs that were performed both on the WT and the E148A mutant, invariably a barrier was found to the translocation of
The recently developed metadynamics methodology has been employed to explore chloride ion translocation from the proximity of the cytoplasmic
Figure 6. Shown is the free energy surface as a function of the Z1 coordinate and the number of waters solvating the chloride ion inside the pore of the E148A mutant. In (a) and (b) the pore is occupied by one and three ions, respectively. The isosurfaces are two per kcal/mol. The total metadynamics trajectory was run for 4 ns. When the upper chloride ion is in Sext, Z1 has a value around 0.5. In (a) it is apparent that the barrier from Scen to Sext (Z1 = −4 and 0.5, respectively) is 1– 2 kcal/mol. A larger barrier is found from Sext to the external solution. In (b) where the pore is occupied by multiple ions the FES is flatter and the barrier to the chloride exit is lower.
396
ClC Chloride Channel Gating Mechanism
Figure 5. Two snapshots of the simulation showing the selectivity filter of the St-ClC E148A mutant with a chloride ion (a) in Sext and (b) in the periplasmic water solution. As the chloride ion translocates to the periplasmic solution, the backbone of the residues A148 and G149 (yellow tube) undergoes a conformational change. An important effect of this change is that the C = O dipole of the residue 148 changes direction.
site to the external solution and vice versa for two different protonation states of the residue E148 and for the mutant E148A. The barrier to chloride ion exit through the E148 narrow region is computed to be 6 kcal/mol in the wild-type channel when there is a second chloride ion in the pore pushing from below. In the absence of a second ion, the barrier is 10 kcal/mol, confirming the importance of the multiple occupancy of ions in the filter, a finding that is in agreement with recent experiments.12 The protonation of E148 has a clear effect on the free energy profile and lowers the barrier to the chloride entrance in the pore from the external solution in agreement with previous calculations.20,18 In the E148A mutant the barrier from Scen to Sext was found to be considerably lower than in the wildtype. It was also found that, before reaching the periplasmic solution, the chloride ion has to overcome an additional barrier coming from the rearrangement of the solvation shell and of the residues 148 and G149 located at the end of the αF helix. Also, this barrier is lowered when the pore is occupied by multiple ions. Overall the present calculations are strongly suggestive of a multi-ion conduction mechanism in which an ion-push-ion effect lowers the main barriers to chloride translocation.
Methods All of the present MD simulations were performed using the MD program ORAC, 29 along with the CHARMM27 all atoms force field30 and the TIP3P water
model.31 The initial geometry was obtained from the crystal structure of Salmonella serovar typhimurium ClC (st-ClC) refined at 3.0 Å resolution (PDB code: 1KPL).11 The N-terminal segment of one of the two monomers (chain A) is missing from the crystal structure. This was reconstructed by duplicating the residues 12–32 from chain B and splicing them into chain A. The water molecules inside the narrow pore were manually added after visual inspection. The protein was then placed with the pore oriented along the Z axis of the simulation cell that contained a pre-equilibrated water solvated POPC bilayer. To do so, all the water molecules and POPC chains closer than 1.0 Å from the protein were eliminated. To minimize the number of atoms the simulation cell was chosen to be hexagonal, since the dimeric channel has approximately a rhomboidal shape. The final equilibrated system contains 12,170 TIP3P water molecules and 136 POPC lipids. In addition to the two crystallographic chloride ions, 30 Cl− and 14 K+ ions were added at random positions in the solution to neutralize the charge of the protein and add a total ion concentration of 100 mM. The resulting 69,000 atoms system was then equilibrated in the NPT ensemble for 6 ns in the hexagonal cell with periodic boundary conditions and particle mesh Ewald electrostatics. The NPT ensemble was simulated with the Parrinello-Rahman barostat32,33 and the Nosé-Hoover thermostat34 keeping pressure and temperature at 1 atm and 300 K, respectively. A further thermalization run was performed in the canonical ensemble for 3 ns at 300 K. In the thermalized structure, the Cα atoms showed an RMSD of 1.9 Å from the crystallographic structure. The metadynamics methodology25 was used throughout to accelerate rare events and to reconstruct the FES. In brief, the method is a dynamics in the space of collective coordinates that are evolved with a standard restrained MD supplemented by a history-dependent potential. A Lagrangian metadynamics run consists of a standard MD
397
ClC Chloride Channel Gating Mechanism
run in which we impose harmonic restraints on a set of CVs that can be any function of the atomic positions. The harmonic restraints connect the real CVs to additional variables that feel the effect of the previous history of the run by means of a Gaussian repulsive potential. The Gaussian repulsive potential is added in the current position of the CVs every T time steps. For a more detailed description of metadynamics, the interested reader should refer to the literature.25–27 The principal advantages of using metadynamics are that it is able to escape local minima by overcoming large free energy barriers and makes it possible to reconstruct FES in the space of the chosen CVs. This is possible since as shown25,27 in the limit of a long metadynamics, the sum of the repulsive Gaussian potentials and the FES tends to become flat as a function of the CVs. When this happens the sum of the repulsive Gaussian potentials gives the negative image of the FES within a controlled error that is inversely proportional to the square root of the diffusion coefficient of the CVs and directly proportional to the speed of Gaussian deposition.27 This method has already been applied with success to study events similar to the translocation of Cl−, namely the translocation of antibiotics through a porin35 and the entrance of the tetramethylammonium ion in the acetylcholinesterase gorge.36 Herein metadynamics has been employed to study the cytoplasmic gating mechanism of the St-ClC channel. To this aim, different sets of CVs were employed. Since the orientation of the channel is such that the pore is approximately along the Z-axis, the main CV used was the projection along the Z-axis of the distance of the chloride ions from the centroid of the channel dimer and POPC bilayer (Z). With this convention the residue E148 has a value of Z that is 0 Å in the crystal structure and fluctuates between 0 and 1 Å in the MD when closed; using the nomenclature of Dutzler et al.10 this value corresponds approximately to Sext. An estimate for the expected values of Scen and Sint can be obtained by measuring in the MD run the average position of the residues to which Cl− is experimentally known to be coordinated when occupying those positions.10 This yields a value of Z = −4 for Scen and Z = −10 for Sint. In the case of Scen a value of Z = −4 is also obtained if we superimpose the experimental and relaxed structure by minimizing the RMSD between the Cαs. Metadynamics runs were performed using as CVs the Z of two chloride ions occupying simultaneously one of the pores in the protonated and unprotonated E148 WTchannel as well as in the E148A mutant. Separate metadynamics runs were performed using as CVs the Z of one chloride ion and the coordination number of the E148 with respect to the H-bonded residues, the Z of one chloride ion and the distance of E148 carboxylic carbon from the Cζ of R147, the Z of one chloride ion and the number of coordinating water molecules and the Z of one chloride ion and Z of the E148 residue. The height of the Gaussians was chosen to be 1.0 kj/mol, the deposition time T, 1 ps. A multiple walker version of the metadynamics algorithm was used throughout.37 To obtain an independent measure of the FES and to confirm the results obtained by metadynamics, we performed a 2D umbrella sampling38 on the WT St-ClC using as CVs the projection along the pore axis of the center of mass two chloride ions and the projection along the pore axis of the position of the carboxy group of E148. In each of the sampling simulations each of the CVs was tethered by virtual springs acting along the Z axis with spring constants of 25 kj/mol per Å2. The tethering points for the umbrella potential were never distributed >1 Å apart requiring 32 simulations of 300 ps each and 14
simulations of 400 ps each to sample the region of interest. Starting configurations were chosen from the metadynamics runs and employed after a re-equilibration at 300 K. The resulting FES is in agreement with that previously obtained by metadynamics.
Acknowledgements We acknowledge the Swiss National Supercomputing Center for computer time. M.L.K. thanks the National Institutes for Health for generous support under GM 40712. M.C. thanks the Caspur Computer Center (Rome, Italy) for using its facilities.
References 1. Estèvez, R. & Jentsch, T. J. (2002). Clc chloride channels: correlating structure with function. Curr. Opin. Struct. Biol. 12, 531–539. 2. Maduke, M., Miller, C. & Mindell, J. A. (2000). A decade of clc chloride channels: structure, mechanism, and many unsettled questions. Annu. Rev. Biophys. Biomol. Struct. 29, 411–438. 3. Iyer, R., Iverson, T. M., Accardi, A. & Miller, C. (2002). A biological role for prokaryotic clc chloride channels. Nature, 419, 715–718. 4. Accardi, A. & Miller, C. (2004). Secondary active transport mediated by a prokaryotic homologue of clc cl- channels. Nature, 427, 803–807. 5. Pusch, M., Ludewig, U., Rehfeldt, A. & Jentsch, T. J. (1995). Gating of the voltage-dependent chloride channel clc-0 by the permeant anions. Nature, 373, 527–531. 6. Chen, Y. T. & Miller, C. (1996). Nonequilibrium gating and voltage dependence of the clc-0 cl- channel. J. Gen. Physiol. 108, 237–250. 7. Rychkov, G. Y., Pusch, M., Roberts, M. L. & Bretag, A. H. (2001). Interaction of hydrophobic anions with the rat skeletal muscle chloride channel clc-1: effects on permeation and gating. J. Physiol. 530, 379–393. 8. Chen, M. F. & Chen, T. Y. (2001). Different fast-gate regulation by external Cl− and H+ of the muscle-type clc chloride channels. J. Gen. Physiol. 118, 23–32. 9. Hanke, W. & Miller, C. (1983). Single chloride channels from torpedo electroplax. Activation by protons. J. Gen. Physiol. 82, 25–45. 10. Dutzler, R., Campbell, E. B. & MacKinnon, R. (2003). Gating the selectivity filter in clc chloride channels. Science, 300, 108–112. 11. Dutzler, R., Campbell, E. B., Cadene, M., Chait, B. T. & MacKinnon, R. (2002). X-ray structure of a clc chloride channel at 3.0 å reveals the molecular basis of anion selectivity. Nature, 415, 287–294. 12. Lobet, S. & Dutzler, R. (2006). Ion-binding properties of the clc chloride selectivity filter. EMBO J. 25, 24–33. 13. Chen, M. F. & Chen, T. Y. (2003). Side-chain charge effects and conductance determinants in the pore of clc-0 chloride channels. J. Gen. Physiol. 122, 133–145. 14. Estèvez, R., Schroeder, B. C., Accardi, A., Jentsch, T. J. & Pusch, M. (2003). Conservation of chloride channel structure revealed by an inhibitor binding site in clc-1. Neuron, 38, 47–59.
398
ClC Chloride Channel Gating Mechanism
15. Engh, A. M. & Maduke, M. (2005). Cysteine accessibility in clc-0 supports conservation of the clc intracellular vestibule. J. Gen. Physiol. 125, 601–617. 16. Cohen, J. & Schulten, K. (2004). Mechanism of anionic conduction across clc. Biophys. J. 86, 836–845. 17. Corry, B., O'Mara, M. & Chung, S.-H. (2004). Conduction mechanisms of chloride ions in clc-type channels. Biophys. J. 86, 846–860. 18. Miloshevsky, G. V. & Jordan, P. C. (2004). Anion pathway and potential energy profiles along curvilinear bacterial clc Cl− pores: electrostatic effects of charged residues. Biophys. J. 86, 825–835. 19. Yin, J., Kuang, Z., Mahankali, U. & Beck, T. L. (2004). Ion transit pathways and gating in clc chloride channels. Proteins: Struct. Funct. Genet. 57, 414–421. 20. Bostick, D. L. & Berkowitz, M. L. (2004). Exterior site occupancy infers chloride-induced proton gating in a prokaryotic homolog of the clc chloride channel. Biophys. J. 86, 1686–1696. 21. Faraldo-Gomez, J. D. & Roux, B. (2004). Electrostatics of ion stabilization in a clc chloride channel homologue from Escherichia coli. J. Mol. Biol. 339, 981–1000. 22. Fahlke, C., Yu, H. T., Beck, C. L., Rhodes, T. H. & George, J. A. L. (1997). Pore-forming segments in voltage-gated chloride channels. Nature, 390, 825–835. 23. Lin, C. W. & Chen, T. Y. (2000). Cysteine modification of a putative pore residue in clc-0. implication for the pore stoichiometry of clc chloride channels. J. Gen. Physiol. 116, 535–546. 24. Braun-Sand, S., Burykin, A., Chu, Z. T. & Warshel, A. (2005). Realistic simulations of proton transport along the gramicidin channel: demonstrating the importance of solvation effects. J. Phys. Chem. B, 109, 583–592. 25. Laio, A. & Parrinello, M. (2002). Escaping free energy minima. Proc. Natl Acad. Sci. USA, 20, 12562–12566. 26. Gervasio, F. L., Laio, A. & Parrinello, M. (2005). Flexible docking in solution using metadynamics. J. Am. Chem. Soc. 127, 2600–2607. 27. Laio, A., Rodriguez-Fortea, A., Gervasio, F. L., Ceccarelli, M. & Parrinello, M. (2005). Assessing the accuracy of metadynamics. J. Phys. Chem. B, 109, 6714–6721.
28. Fahlke, C., Rosenbohm, A., Mitrovic, N., George, A. L., Jr & Rudel, R. (1996). Mechanism of voltagedependent gating in skeletal muscle chloride channels. Biophys. J. 71, 695–706. 29. Procacci, P., Darden, T., Paci, E. & Marchi, M. (1998). Orac: a molecular dynamics program to simulate complex molecular systems with realistic electrostatic interactions. J. Comput. Chem. 18, 1848-1862. 30. MacKerell, A. D., Jr, Bashford, D., Bellot, M., Dunbrack, R. L., Jr, Evanseck, J. D., Field, M. J. et al. (1998). All-atom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B, 102, 3586–3616. 31. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L. (1983). Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79, 926–935. 32. Parrinello, M. & Rahman, A. (1981). Polymorphic transitions in single crystals: a new molecular dynamics method. J. Appl. Phys. 52, 7182. 33. Parrinello, M. & Rahman, A. (1980). Crystal structure and pair potentials: a molecular-dynamics study. Phys. Rev. Letters, 45, 1196. 34. Nose, S. (1984). A molecular dynamics method for simulations in the canonical ensemble. Mol. Phys. 52, 255–268. 35. Ceccarelli, M., Danelon, C., Laio, A. & Parrinello, M. (2004). Microscopic mechanism of antibiotics translocation through a porin. Biophys. J. 87, 58–64. 36. Branduardi, D., Gervasio, F. L., Cavalli, A., Recanatini, M. & Parrinello, M. (2005). The role of the peripheral anionic site and cation-π interactions in the ligand penetration of the human ache gorge. J. Am. Chem. Soc. 127, 9147–9155. 37. Raiteri, P., Laio, A., Gervasio, F. L., Micheletti, C. & Parrinello, M. (2006). Efficient reconstruction of complex free energy landscapes by multiple walkers metadynamics. J. Phys. Chem. B, 110, 3533–3539. 38. Roux, B. (1995). The calculation of the potential of mean force using computer simulations. Comp. Phys. Commun. 91, 275–282.
Edited by M. Levitt (Received 8 March 2006; received in revised form 12 June 2006; accepted 14 June 2006) Available online 30 June 2006