Proton-Dependent Inhibition, Inverted Voltage Activation, and Slow Gating of CLC-0 Chloride Channel

CLC-0, a prototype Cl− channel in the CLC family, employs two gating mechanisms that control its ion-permeation pore: fast gating and slow gating. The negatively-charged sidechain of a pore glutamate residue, E166, is known to be the fast gate, and the swinging of this sidechain opens or closes the pore of CLC-0 on the millisecond time scale. The other gating mechanism, slow gating, operates with much slower kinetics in the range of seconds to tens or even hundreds of seconds, and it is thought to involve still-unknown conformational rearrangements. Here, we find that low intracellular pH (pHi) facilitates the closure of the CLC-0’s slow gate, thus generating current inhibition. The rate of low pHi-induced current inhibition increases with intracellular H+ concentration ([H+]i)—the time constants of current inhibition by low pHi = 4.5, 5.5 and 6 are roughly 0.1, 1 and 10 sec, respectively, at room temperature. In comparison, the time constant of the slow gate closure at pHi = 7.4 at room temperature is hundreds of seconds. The inhibition by low pHi is significantly less prominent in mutants favoring the slow-gate open state (such as C212S and Y512A), further supporting the fact that intracellular H+ enhances the slow-gate closure in CLC-0. A fast inhibition by low pHi causes an apparent inverted voltage-dependent activation in the wild-type CLC-0, a behavior similar to those in some channel mutants such as V490W in which only membrane hyperpolarization can open the channel. Interestingly, when V490W mutation is constructed in the background of C212S or Y512A mutation, the inverted voltage-dependent activation disappears. We propose that the slow kinetics of CLC-0’s slow-gate closure may be due to low [H+]i rather than due to the proposed large conformational change of the channel protein. Our results also suggest that the inverted voltage-dependent opening observed in some mutant channels may result from fast closure of the slow gate by the mutations.

4 68 millisecond time scale, while slow gating operates on the order of ~seconds to hundreds of seconds 69 [26]. Because the slow-gating mechanism appears to control the two pores simultaneously, it is also 70 called "common" gating, and the closure of the slow gate "inactivates" the channel. Based on single-71 channel behaviors of CLC-0, when the slow gate closes, Clconduction through the channel pores is 72 shut, and the functional activities of the fast gate are not observable.  High-resolution CLC structures reveal three Cl --binding sites in the ion-transport 87 pathways of CLC proteins: the external (S ext ), central (S cen ) and internal sites (S int ) [18]. However, S ext 88 can also be occupied by the negatively charged sidechain of a glutamate residue (corresponding to E166 89 in CLC-0), which is thought to be the fast gate (called E gate). Swinging this E-gate away from S ext is 6 113 in Fig. 2 B, where the current of the V490W mutant is activated by membrane hyperpolarization but 114 not by depolarization. Interestingly, we discover that such a hyperpolarization-induced channel opening 115 can also occur in WT CLC-0 in the presence of low pH i (Fig. 2 C). In this paper, we study the relation 116 between the inverted voltage-dependent channel opening and the intracellular H + effect on the slow 117 gating in CLC-0. Voltage protocol (protocol I) for recordings. A full protocol consists of 12 recording sweeps.

121
One sweep includes a prepulse voltage step at +60 mV (50 ms) followed by one of the various 122 test voltage steps (70 ms) from +60 mV to -160 mV in -20 mV voltage steps, and followed by a   Because the slow-gate opening of CLC-0 tends to be minimal at voltages of the resting 175 membrane potential of HEK293 cells or above (Chen 1998), WT CLC-0 was partially inactivated when 176 membrane patches were excised. To activate the current of CLC-0 for experiments, five 50-ms pulses 9 178 The procedure was repeated multiple times until the slow gate was maximally opened (based on the 179 observation that the recorded current was no longer increased by this current-activation procedure). For 180 mutants with a mostly open slow-gate (such as C212S or Y512A) no such current activation procedure 181 was necessary.

182
Three types of experimental protocols were employed for the experiments presented in this 183 paper. Protocol I was used for evaluating the quasi steady-state voltage-dependent current activation. To evaluate the process of the change of Clcurrent upon reducing pH i , we employed a voltage 203 protocol (protocol II) containing a voltage step of +60 mV (50 ms) followed by a tail voltage step at -204 100 mV for 70 ms. Such a voltage protocol was used to mimic the experimental protocol of the previous 205 studies using whole cell recording methods on channels expressed in Xenopus oocytes [38,39]. To 206 present the experimental results, the current at the +60 mV voltage step was measured and plotted 207 against the time of the recording. ISI, which was either 1 or 4 sec, also refers to the time interval 208 between the end of one recording sweep and the beginning of the following sweep. The membrane 209 voltage at the ISI in this protocol was also 0 mV.

210
The third voltage protocol (protocol III), like protocol II, was also used to assess the current 211 inhibition process after applying a high [H + ] i except that the application and removal of the low pH i 212 solution was conducted at a constant membrane voltage. For these experiments, the membrane voltages 213 ranged from +60 mV to -60 mV. However, experiments with a slow inhibition process (in relatively 214 high pH i conditions such as pH i = 5.5 or 6) at some negative voltages were technically difficult due to 215 stability problems of the excised patches. Estimates of the time constant of such slow inhibition 216 processes may thus be less precise.

217
Protocol II & III were used for studying the kinetics of the current relaxation process followed by 218 the pH i perturbation. Both protocols started from a steady-state current level at pH i = 7.4. A lower pH i 219 solution was then applied, and the current was inhibited to different degrees at different speeds 220 depending on pH i (or [H + ] i ). After the current reached a steady state, pH i was changed back to 7.4 and 221 the current may or may not recover. The current relaxation process (current inhibition or current 11 223 inhibition (τ inh ) and the time constant of current recovery (τ rec ). However, for recordings of the WT 224 CLC-0, current recovery was observed only at negative membrane voltages but not at the positive 225 membrane voltages.  (Fig. 2 B, left panel). The current deactivation at hyperpolarization voltages reflects a 242 reduction of P o f , with a kinetics in the millisecond time range.
In some mutant channels of CLC-0, such a normal voltage dependence of channel opening is 244 inverted. For example, using the same voltage protocol I (Fig. 2 A), WT CLC-0 (Fig. 2 B, left panel) 245 and the V490W mutant (Fig. 2 B, right panel) 258 pre-pulse current in the recording with 1-sec ISI retains some outward current. The difference in the 259 outward current between these recordings is also reflected by the instantaneous inward current when 260 the membrane voltage is changed from +60 mV pre-pulse voltage step to the various hyperpolarizing 261 voltage steps. It should be noted that the current at the +60 mV pre-pulse voltage depends on the channel 262 conductance at the end of the ISI following the previous recording sweep. We suspected that 263 intracellular H + may inhibit the current of WT CLC-0 at the 0-mV holding voltage during ISI. Therefore, 264 in a recording with 4-sec ISI, the channel conductance (after being activated by membrane 265 hyperpolarization) was nearly completely inhibited by H + . On the other hand, with 1-sec ISI, the 267 following recording sweep.

268
To confirm this speculation, we employed a continuous recording protocol, in which a sweep 269 of recording contains only a +60 mV voltage step for 50 ms followed by a negative tail voltage step of 270 -100 mV for 70 ms (protocol II, Fig. 3 A). The holding voltage at ISI was 0 mV. Typical experiments 271 are shown in Fig. 3 (Fig. 3 B) is inhibited almost completely while 276 the recording with ISI = 1 sec (Fig. 3 C) still retains significant outward current.    The experiments in Figs. 3 and 4 were conducted using protocol II with a negative tail voltage 305 step (at -100 mV) which activates the current of WT CLC-0 at low pH i . Assessing the effects of 306 membrane voltages on the kinetics of current inhibition and recovery was therefore not accurate. We 307 thus employed a different experimental protocol, namely, altering pH i at constant voltages (protocol 308 III). In such experiments (Fig. 5 A), the current inhibition can still be reasonably fit to a single-309 exponential function. Nonetheless, no current recovery was observed after switching back to the 311 the results obtained with protocol II (Fig. 4), the experiments using protocol III generate a faster current 312 inhibition and a slower current recovery (namely, τ inh , is smaller while τ rec , is larger). In addition, like 313 those experiments in Fig. 4, τ inh from using protocol III also strongly depends on [H + ] i (Fig. 5 B) but 314 not on membrane voltages (Fig. 5 C). On the other hand, membrane voltages affect the current recovery 315 significantly in that the more hyperpolarized the membrane voltage, the smaller the value of τ rec (namely, 316 the faster the current recovery rate) (Fig. 5 D).   It is interesting to note that extrapolating the value of τ inh to a neutral pH i gives a τ inh value of 330 hundreds of sec (Fig. 5 B), which is similar to the relaxation time constant of the CLC-0 slow-gate   (Fig. 7 B), pH i = 4.5 was used because this mutant channel is much less sensitive to H + inhibition.
352 Visual inspection of the three recording traces indicates that the temperature dependence of the low 353 pH i -induced inhibition in C212S is weak. The averaged results shown in Fig. 7 C reveal a large 354 difference of the temperature dependence of τ inh between WT CLC-0 and the C212S mutant. In Fig. 8 355 A, the process of the current inhibition by pH i = 4.5 (pink area) and the process of current recovery 356 upon removing high [H + ] i (light blue area) for WT CLC-0 and the C212S mutant are illustrated. The 357 voltage dependence of the averaged values of τ inh between WT CLC-0 and the C212S mutant are 358 compared in Fig. 8 B, while the comparison of those of τ rec are illustrated in Fig. 8 C. In CLC-0, it is 359 τ rec but not τ inh that is voltage dependent-the more negative the membrane voltage, the smaller the 360 value of τ rec (namely, the faster the recovery from the inhibition). In C212S, a [H + ] i < 1 µM (namely 361 pH i > 6) generates very little inhibition (Fig. 6), so it is technically necessary to employ very low pH i 362 (4-5) to generate inhibition for the experiments. The value of τ inh is small (fast current inhibition) likely 363 because of the high [H + ] i at pH i = 4.5. Interestingly, τ rec is also small in C212S, reflecting a faster current 364 recovery process than that in WT CLC-0. Fig. 8    If the lower sensitivity to intracellular H + inhibition in C212S is due to a reduced slow-gate 385 closure in this mutant, other mutations that also prevent the channel from closing the slow gate may 386 exhibit similar low sensitivity to intracellular H + inhibition. In Fig. 9 A, we compare the inhibition by 387 low pH i solutions (pH i = 5.5 and 4.5 in the upper and lower panel, respectively) between WT CLC-0 388 and another mutant, Y512A, which has also been shown to largely prevent the slow gate from closing 389 [45]. The steady-state [H + ] i -dependent current inhibition in Fig. 9 B indeed shows that the mutant 390 Y512A, like C212S, is also more resistant to the intracellular H + inhibition than WT CLC-0. On the other hand, if the inverted voltage dependent opening of the V490W mutant is due to 402 an excessive slow-gate closure in neutral pH i , the C212S mutation or the Y512A mutation may reduce 403 the excessive slow-gate closure caused by the V490W mutation. In Fig. 9  The two types of CLC family members, Clchannels (such as CLC-0, CLC-1, CLC-2 and CLC-