Membrane Cholesterol Content Modulates ClC-2 Gating and Sensitivity to Oxidative Stress*

ClC-2 is a broadly expressed member of the voltage-gated ClC chloride channel family. In this study, we aimed to evaluate the role of the membrane lipid environment in ClC-2 function, and in particular the effect of cholesterol and ClC-2 distribution in membrane microdomains. Detergent-resistant and detergent-soluble microdomains (DSM) were isolated from stably transfected HEK293 cells by a discontinuous OptiPrep gradient. ClC-2 was found concentrated in detergent-insoluble membranes in basal conditions and relocalized to DSM upon cholesterol depletion by methyl-β-cyclodextrin. As assessed by patch clamp recordings, relocalization was accompanied by acceleration of the activation kinetics of the channel. A similar distribution and activation pattern were obtained when cells were treated with the oxidant tert-butyl hydroperoxide and after ATP depletion. In both cases activation was prevented by cholesterol enrichment of cells. We conclude that the cholesterol environment regulates ClC-2 activity, and we provide evidence that the increase in ClC-2 activity in response to acute oxidative or metabolic stress involves relocalization of this channel to DSM.

ClC-2 is a broadly expressed member of the voltage-gated ClC chloride channel family. In this study, we aimed to evaluate the role of the membrane lipid environment in ClC-2 function, and in particular the effect of cholesterol and ClC-2 distribution in membrane microdomains. Detergent-resistant and detergentsoluble microdomains (DSM) were isolated from stably transfected HEK293 cells by a discontinuous OptiPrep gradient. ClC-2 was found concentrated in detergent-insoluble membranes in basal conditions and relocalized to DSM upon cholesterol depletion by methyl-␤-cyclodextrin. As assessed by patch clamp recordings, relocalization was accompanied by acceleration of the activation kinetics of the channel. A similar distribution and activation pattern were obtained when cells were treated with the oxidant tert-butyl hydroperoxide and after ATP depletion. In both cases activation was prevented by cholesterol enrichment of cells. We conclude that the cholesterol environment regulates ClC-2 activity, and we provide evidence that the increase in ClC-2 activity in response to acute oxidative or metabolic stress involves relocalization of this channel to DSM.
ClC-2 is a broadly expressed member of the ClC family of voltage-gated chloride channels (1). Although a number of roles have been suggested for this channel to date, its biological significance still remains a fundamental question. It has been suggested to participate in epithelial transport processes (2,3) and in the control of intracellular Cl Ϫ concentration in neurons expressing inhibitory ␥-aminobutyric acid receptors (4). In this respect, ClC-2 mutations have been found in patients with idiopathic generalized epilepsy (5). In addition, ClC-2 has been proposed to be the molecular candidate for the native cardiac inward-rectifying chloride channels (6). It has also been speculated that the physiological role of these channels in the heart may become more prominent under pathological conditions, such as ischemia and hypoxia (6).
ClC-2 is activated by hyperpolarization, extracellular acidification, and hypotonic cell swelling (7). Several studies have demonstrated additional regulation by protein kinases (8 -10) and modulation of its activity by the redox status of the cell (11). Control of channel trafficking and activity by interaction with various protein partners has been documented (9,12,13). However, even though the synthetic lipid lubiprostone has been shown to activate ClC-2 in vitro (14) and chloride secretion in the intestine (15), the influence of membrane lipids on ClC-2 activity has not yet been investigated.
Among membrane lipid components, cholesterol plays an active role in regulating the lipid bilayer dynamics and structure. Experimental evidence indicates that it is involved in the assembly and maintenance of sphingolipid and cholesterol-rich microdomains (16,17). These are known as "detergent-insoluble membranes" (DIM), 3 or most commonly "lipid rafts." These structures are implicated in signal transduction, intracellular trafficking of lipids and proteins, and translocation of solutes across the membrane (18,19). DIM have been proposed to concentrate various membrane proteins such as signal transduction enzymes, membrane receptors, ion transporters, and channels (20,21). Since the first demonstration of Kv2.1 channel localization to lipid microdomains (18), additional reports suggested that other types of cationic channels are associated with lipid rafts, such as Na ϩ (22) and Ca 2ϩ channels (23), Kir2.1 (24), the pacemaker channel HCN4 (25), or TRP (26), and that the cholesterol-rich environment is important for their activity. Yet little is known about the possible association of anionic channels with lipid rafts, except for the cystic fibrosis transmembrane conductance regulator, which has been proposed to relocalize to DIM upon stimulation of the inflammatory response (27). Cholesterol has also been shown to regulate volume-activated anion current in endothelial cells (28,29), but in the absence of the molecular identity of these volume-dependent channels, their precise localization is unknown.
Our results suggest that another class of anionic channel, ClC-2, is significantly present in cholesterol-rich microdomains (DIM) in both transfected HEK293 cells and native tissues. DIM disruption by cholesterol depletion increases the kinetics of channel activation, whereas membrane enrichment in cholesterol exerts an opposite effect. We also provide evidence that the previously demonstrated increase in ClC-2 activity in response to acute oxidative or metabolic stress involves dissociation of this channel from DIM.

EXPERIMENTAL PROCEDURES
Constructs, Cells, and Animals-The cDNA of rat ClC-2 was kindly provided by T. J. Jentsch (Hamburg, Germany). The cDNA was subcloned either in pIRES-EGFP, which enables the expression of a nontagged ClC-2 protein, or pQBI25-fC1, which generates an N-terminal GFP-ClC-2 fusion protein.
HEK293 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% serum at 37°C, 5% CO 2 atmosphere. Cells were transiently transfected with pQBI25-fC1 plasmid or stably transfected with a linearized pIRES-EGFP plasmid, and selection was done with 500 g/ml G418. In a set of experiments, tissues were obtained from C57BL/6 mice (Centre de Distribution, Typage and Archivage CNRS, Orléans, France).
Treatments-All chemicals were from Sigma except where indicated otherwise. For cholesterol depletion, cells were preincubated for 1 h in serum-free Dulbecco's modified Eagle's medium followed by 30 min in the presence of 10 mM methyl-␤-cyclodextrin (m␤CD) at 37°C. Cholesterol replenishment was performed by removal of m␤CD-containing medium and incubation for 45 min in a solution containing a cholesterol-m␤CD complex (water-soluble cholesterol; Sigma), at a final concentration of 400 g/ml cholesterol and 2.5 mM m␤CD. Cholesterol enrichment prior to any treatment was performed by exposing cells for 2 h to the same cholesterol-m␤CD complex. In control patch clamp analyses, the m␤CD analogue ␣-cyclodextrin was utilized. Oxidation was stimulated by 30 min of incubation in 1.5 mM tert-butyl hydroperoxide (TBH). Inhibition of glycolysis was achieved by incubating cells for 3 h in a glucose-free Ringer's solution containing 140 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM Hepes, pH 7.3, and 2 mM glutamine. ATP depletion was performed by adding 10 mM 2-deoxyglucose, and 1 M antimycin A to the Ringer's solution. Cells were further incubated for 90 min.
Preparation of Lipid Membrane Microdomains-Cells were grown to confluence, washed twice in cold PBS, and transferred to centrifuge tubes. Two additional washes were performed by centrifugation at 800 ϫ g for 2 min. The pellet was resuspended in 2 ml of lysis buffer (1% Triton X-100, 25 mM Tris-HCl, 1 mM EDTA, 150 mM NaCl, 1 mM NaVO 4 , 1 mM NaF, and 1 tablet/10 ml of Complete-Mini protease inhibitor mixture; Roche Diagnostics) and incubated at 4°C for 20 min. Tissue samples from heart, brain, and kidney were homogenized in a hypotonic solution (20 mM Tris-HCl, 25 mM NaCl, 2.5 mM MgCl 2 , 1 mM EDTA) and incubated in lysis buffer with or without 1% saponin. Cell lysates were passed 10 times through a 21-gauge needle, and 200 l of lysate were mixed with 400 l of a 60% Opti-Prep solution (Axis-Shield, Oslo, Norway) (40% final concentration). A three-step discontinuous OptiPrep gradient was prepared by layering 1.2 ml of 30% OptiPrep in detergentfree lysis buffer and 600 l of buffer alone on top. Gradients were centrifuged at 55,000 rpm for 2 h at 4°C in a Beckman XL-70 ultracentrifuge, using a swinging 55-Ti rotor. After spinning, eight fractions of 300 l were recovered from each gradient. DIM reach their isopycnic point at the interphase 0/30% OptiPrep (fractions 2 and 3). In most experiments, fractions were pooled to obtain DIM (fractions 2-4) and DSM (fractions 6 -8) preparations.
Immunoblotting-For whole-cell protein analysis, 20 l of gradient fractions were resuspended in modified 2ϫ Laemmli containing 7 M urea. For surface protein analysis, cells were previously incubated for 30 min at 4°C in PBS containing 0.5 mg/ml sulfo-NHS-biotin (Pierce) and subjected to OptiPrep gradient as described above. Biotinylated (surface) proteins were recovered by incubation of gradient fractions with agarose bead-immobilized monomeric avidin (Pierce). The avidin-containing pellets were resuspended in modified Laemmli. Samples were heated at 37°C for 2 h and subjected to SDS-PAGE in an 8% polyacrylamide gel containing 3.5 M urea. The amount of ClC-2 in each fraction was estimated by Western blot. Briefly, proteins were transferred onto nitrocellulose membranes and blocked for 1 h with 5% nonfat milk diluted in TBS/Tween 0.1%. Membranes were incubated with an anti-ClC-2 antibody (clone H-90; Santa Cruz Biotechnology, Santa Cruz, CA) diluted at 1:500. Raft marker flotillin-1 (Flo-1) was detected by a monoclonal antibody (BD Biosciences) diluted at 1:1000 and a nonraft marker transferrin receptor by a monoclonal antibody (Zymed Laboratories Inc.) diluted at 1:1000. Horseradish peroxidase-coupled secondary antibodies were enabled to visualize protein signal after incubation with ECL solution. In some experiments, the protein concentration of each fraction was determined by the DC method (Bio-Rad), and equal amounts of protein per well were loaded.
Immunocytochemistry-HEK293 cells stably expressing wild-type ClC-2 were grown on polylysine-precoated coverslips, fixed with 4% formaldehyde, and permeabilized with 0.1% Triton in PBS. Cells were blocked with 1% bovine serum albumin in PBS/Triton and incubated at 4°C overnight with the primary ClC-2 antibody (ACL-002; Alomone Labs, Israel; final dilution 1:250 in PBS/Triton). After washing and blocking with 5% normal goat serum, cells were incubated with the secondary antibody (Alexa-fluor 594; Molecular Probes). Glass coverslips were mounted using the Vectashield mounting medium (Vector Laboratories) and examined by confocal laser microscopy (Axiovert 100 M; Zeiss, Jena, Germany). Controls were carried out on cells expressing the pIRES vector alone or by peptide competition using ClC-2 antibodies preincubated for 1 h with a 5-fold excess of the corresponding peptide antigens.
Electrophysiology-Patch clamp experiments were performed at room temperature as described previously (13). Current recordings were performed using either the classical or the nystatin-perforated whole-cell patch clamp configuration. Nystatin stock solution (50 mg/ml) was prepared daily in Me 2 SO. Aliquots were diluted (1:250) in the internal solutions and sonicated for at least 30 s. The bath solution contained the following (in mM): 150 NaCl, 2 CaCl 2 , 1 MgCl 2 , 10 Hepes-Na ϩ , and 35 sucrose, pH 7.4, adjusted with NaOH. The pipette solution contained the following (in mM): 60 NaCl, 2 MgCl 2 , 35 Na 2 SO 4 , 10 Hepes, pH 7.3, adjusted with NaOH. Changes in liquid junction potential were calculated and taken into account. For recordings using the standard whole-cell configuration, the bath solution contained the following (in mM): 140 NaCl, 2 CaCl 2 , 1 MgCl 2 , 10 Hepes-Na ϩ , and 20 sucrose, pH 7.4, adjusted with NaOH. The pipette solution contained the following (in mM): 29 NaCl, 33 CsCl, 71 sodium gluconate, 1 MgCl 2 , 10 Hepes-Na ϩ , 2 EGTA, 2 MgATP, pH 7.3. Currents were recorded by application of regular voltage pulses of desired length and amplitude from a holding potential of 0 mV with an interval of at least 60 s to allow current deactivation. Time courses for activation and deactivation were described by fitting a mono-or double exponential plus a constant term as shown in Equation 1, where I(t) is current as a function of time; A f , A s , and A 0 are fast, slow, and time-independent constants, respectively, and f and s are the fast and slow time constants. The relative open probability as a function of voltage was estimated from measurements of the initial currents at 40 mV after each negative test voltage jump. The conductance values were adjusted by using a Boltzmann distribution as shown in Equation 2, where G, G o , and G max are conductance as a function of voltage, residual conductance, and maximal conductance (extrapolated), respectively. V 0.5 is the voltage at which 50% activation occurs, and k is the slope factor. All measured values are presented as means Ϯ S.E. The significance of differences between means was determined using either paired or unpaired t test when appropriate.
Cholesterol and Phospholipid Analysis-Cholesterol and phospholipid content were monitored by thin layer chromatography. The lipid-containing organic phase was obtained from gradient fractions by liquid-liquid extraction with 6 volumes of chloroform/methanol (2:1, v/v), centrifuged at 800 ϫ g for 3 min, and the resulting lower phase was aspirated. For cholesterol analysis, aliquots of 4 l of the different samples and cholesterol standard were applied to HP-K plates (Whatman, Clifton, NJ), developed in chloroform/acetone (95:5 v/v), stained with the CuSO 4 reagent, and developed by charring at 170°C. For phospholipid analysis, two sequential mobile phases were utilized as follows: chloroform/triethylamine/ ethanol/water (30:30:34:8) and hexane/diethyl ether (100:4.5). The resulting bands were analyzed by the Imagemaster software (Amersham Biosciences).

ClC-2 Is Partially Localized to Cholesterol-rich Microdomains in HEK293 Cells and Mouse
Tissues-To evaluate the microdomain distribution of ClC-2, stably transfected HEK293 cells were subjected to 1% Triton X-100 incubation and a discontinuous gradient. After analysis of an equal volume aliquot of each fraction, ClC-2 was found almost equally distributed in fractions 3 and 6 ( Fig. 1A, 1st panel). However, when an equal protein amount per fraction was loaded, ClC-2 was found mostly associated with fraction 3 (Fig. 1A, 2nd panel). Fractions 3 and 6 correspond to DIM and DSM, respectively, as confirmed by the distribution of DIM markers flotillin-1, cholesterol, and sphingomyelin and of DSM marker transferrin receptor. Under these migration conditions, ClC-2 can be detected either as a single or as a double band when overexpressed in HEK293 cells. As this pattern differs from the one that is usually described, we aimed to confirm the identity of ClC-2. Both bands, which are absent from cells transfected with the empty plasmid (mock), were identified by mass spectrometry as ClC-2 (data not shown). A, stably transfected HEK293 cells were lysed in a buffer containing 1% Triton X-100 and subjected to ultracentrifugation in a discontinuous OptiPrep gradient. Either equal protein (eq. prot.) amounts (2nd panel) or equal volumes (eq. vol.) (rest of panels) from each gradient fraction (1-8) were loaded onto a gel and ClC-2, flotillin-1 (Flo-1), and transferrin receptor (TR) were detected by Western blotting. Cholesterol and sphingomyelin were analyzed by high performance TLC. B, tissue homogenates from mouse heart, brain, and kidney were subjected to microdomain fractionation, and ClC-2 was detected in the same conditions as in A (equal volume loaded). C, densitometric quantification of ClC-2 amount per total protein in DIM and DSM from each model utilized. Results are expressed as means of three experiments Ϯ S.E.
We also analyzed the distribution of endogenous ClC-2 in mouse heart, brain, and kidney. In these tissues, a significant fraction of ClC-2 was detected in DIM (Fig. 1B). Addition of 1% saponin (a cholesterol-solubilizing detergent) to the lysis buffer induced a decrease in the amount of DIM-associated ClC-2, suggesting that DIM localization in these tissues is cholesteroldependent (not shown). Fig. 1C shows the quantification of ClC-2 concentration by total protein in DIM and DSM in cells and tissues, which reveals a tissue-specific variability in the partial association of this channel with cholesterol-containing (DIM) fractions. Thus, ClC-2 presence per protein content is about 2-fold higher in DIM than in DSM in HEK293 cells, almost equally distributed in both domains in brain and heart, and about 3-fold lower in DIM versus DSM in kidney (Fig. 1C).
In subsequent experiments, and based on the distribution of DIM and DSM markers, fractions 2-4 were pooled as DIM, and fractions 6 -8 as DSM. The effect of cholesterol content modulation was first evaluated biochemically. As shown in Fig. 2A, treatment with the cholesterol-depleting agent m␤CD (10 mM, 30 min) led to a relocalization of ClC-2 to DSM. This effect was reversed by cholesterol replenishment, as found by addition of the cholesterol-m␤CD complex, which significantly reduced the presence of ClC-2 in DSM, and returned to control values. Fig. 2B shows the effect of cholesterol enrichment by direct incubation of cells with the cholesterol-m␤CD complex, which redistributed ClC-2 to DIM as compared with the control. The densitometric analysis indicates that ClC-2 relocalization to DSM following m␤CD treatment is significant both at the cell surface and in the total cell. The effect of cholesterol-m␤CD complex alone on ClC-2 distribution is also significant in both pools (Fig. 2C). Fig. 2D shows the changes in cholesterol content of DIM and DSM induced by the different conditions tested. m␤CD induced a significant decrease in cholesterol in both domains. This effect was reversed by subsequent cholesterol replenishment, whereas cholesterol enrichment significantly increased its content in DSM. These results clearly indicate a link between cholesterol content and ClC-2 solubility.
Modulation of Cholesterol Content Affects Channel Function-The effect of cholesterol depletion or enrichment on channel function was analyzed in transfected HEK293 cells using the whole-cell configuration of the patch clamp technique. Because preincubation of cells with m␤CD did not allow us to regularly obtain tight seals, in most experiments tests were performed after cell break in using nystatin. ClC-2 currents were recorded in response to negative voltage jumps followed by a ϩ40-mV step. Fig. 3A shows traces obtained by hyperpolarization to Ϫ120 mV in the same cell, before and after exposure to 10 mM m␤CD for 30 min. Current activation was clearly faster after cell treatment by the drug. By contrast, cholesterol enrichment by preincubation of cells for 2 h with the cholesterol-m␤CD complex resulted in slower kinetics of current activation. Mean current densities at Ϫ120 mV were not significantly different before and after treatment of 18 cells with m␤CD (Ϫ1.11 Ϯ 0.11 nA and Ϫ1.12 Ϯ 0.08 nA, respectively), whereas current amplitude was reduced when cells were cholesterol-enriched (Ϫ0.65 Ϯ 0.08 nA, n ϭ 7, p Ͻ 0.005). No changes were observed either in cells incubated with 10 mM of the inactive stereoisomer ␣-cyclodextrin or in cells transfected with enhanced GFP alone (not shown). As described previously, ClC-2 current activation at negative potentials and deactivation at ϩ40 mV displayed fast and slow components (30). Quantification of changes in the activation or deactivation kinetics was done by fitting Equation 1 (see "Experimental Procedures") to the current relaxations at Ϫ120 and ϩ40 mV. Fig. 3B shows the time constants of current activation. Both constants were statistically decreased by m␤CD treatment, whereas they were increased after cholesterol enrichment. When cholesterol-depleted cells by m␤CD were further incubated for 45 min with the cholesterol-m␤CD complex, activation constants were similar to the values measured in control cells, indicating that the effect of m␤CD could be reversed by cholesterol replenishment. The fractional amplitudes of the fast (12.7 Ϯ 2.1), slow (73.2 Ϯ 2.5), and instantaneous (16.9 Ϯ 2.2) components were not statistically affected by treatments (not shown). The kinetics of channel closure at a positive potential were not modified by cholesterol depletion or enrichment treatments, as shown in Table 1.
The voltage dependence of currents was estimated in some cells by recording families of currents evoked at negative and positive potentials. Representative current traces elicited from a holding potential of 0 mV in response to negative test pulses followed by pulses to ϩ40 mV are illustrated in Fig. 3C. Steadystate activation curves (Equation 2) are illustrated in Fig. 3D. m␤CD treatment and cholesterol enrichment resulted in a rightward and leftward shift of the P o /V curve, respectively. The V 0.5 and slope factors were Ϫ109 Ϯ 2 and Ϫ25 Ϯ 2 mV (n ϭ 15) for control cells, Ϫ94 Ϯ 5 and Ϫ20 Ϯ 2 mV for m␤CD-treated cells (n ϭ 6), and Ϫ137 Ϯ 4 and Ϫ27 Ϯ 3 mV for cholesterolenriched cells (n ϭ 5). From these curves it can be observed that the fraction of channels opened in the range of physiological potentials (around Ϫ60 to Ϫ80 mV) was substantially higher in cells treated by m␤CD as compared with cells enriched in cholesterol.

Modulation of Cholesterol Content Affects Plasma Membrane Channel Distribution-Subcellular ClC-2 localization
was assessed by two approaches as follows: (i) by confocal immunocytochemistry, using a ClC-2 antibody previously validated by complete absence of labeling in tissues from ClC-2 KO mice (31), and (ii) by examining the GFP fluorescent signal of a GFP-ClC-2 construct transiently expressed in HEK293 cells. Confocal images of middle sections of cells under different cholesterol conditions are shown in Fig. 4A. Similar patterns of ClC-2 labeling were obtained by the two detection methods. Prior to any treatment, ClC-2 exhibited a markedly punctate staining pattern in the cytoplasm and at the plasma membrane, suggesting the concentration of this protein in certain membrane microdomains (Fig. 4A, panels a and b). Cholesterol depletion (Fig. 4A, panels c and d) induced a discrete modification in the membrane distribution of channels, which became more regularly scattered over the membrane surface. By contrast, incubation of cells with cholesterol-m␤CD complex for 2 h clearly diminished plasma membrane staining (Fig.  4A, panels e and f ). The specificity of ClC-2 immunostaining was assessed either by the absence of labeling in cells transfected with an empty plasmid (Fig. 4A, panel g) or by the absence of labeling when ClC-2 antibodies were preincubated with an excess of the corresponding peptide antigen (Fig. 4A,  panel h). The decrease in ClC-2 expression at the plasma mem- A, upper panel, currents elicited at Ϫ120 mV from a holding potential of 0 mV in the same cell, before and after cholesterol depletion by 30 min treatment with m␤CD. The negative pulse was followed by a pulse to ϩ40mV; lower panel, current elicited at Ϫ120 mV in a cell that was enriched in cholesterol by exposure for 2 h to a cholesterol (400 g/ml)-m␤CD (2.5 mM) complex. B, means Ϯ S.E. of time constants of the fast (⌻ f ) and slow components (⌻ s ) of current activation (control, n ϭ 18; m␤CD, n ϭ 18; m␤CD-cholesterol complex, n ϭ 7; m␤CD followed by 45 min of incubation with the m␤CDcholesterol complex: n ϭ 6. *, p Ͻ 0.05, paired t test. C, representative families of current traces recorded from a holding potential of 0 mV to test pulses delivered every 60 s from ϩ40 to Ϫ180 mV in 20-mV steps. These pulses were followed by a pulse to ϩ40mV. The duration of the pulses was increased at less negative voltages in order to approximate full activation of the conductance, and their lengths were adapted at the negative potentials to avoid changes in [Cl Ϫ ] i and consequently in the reversal potential of the current. For illustration purposes, the beginning of the tail currents at ϩ 40 mV was set at the same time. D, open probability P o /V curves in control conditions (n ϭ 15, circle), m␤CD-treated cells (n ϭ 6, square), or cholesterol-enriched cells (n ϭ 5, triangle). brane when cells were enriched in cholesterol was confirmed by Western blotting of surface-biotinylated ClC-2 (Fig. 4B).

TABLE 1 Current deactivation parameters in ClC-2-transfected HEK293 cells subjected to different treatments
Oxidative and Metabolic Stress Affect Channel Function-In an attempt to find a physiopathological significance for the cholesterol-dependent regulation of ClC-2 activity, we tested oxidative and metabolic stress as exogenous stimuli that have been proven to activate the channel (11,32). Exposure of ClC-2-transfected cells to 1.5 mM TBH, an H 2 O 2 analogue, resulted in a progressive increase in current amplitude from Ϫ1.1 Ϯ 0.06 to Ϫ1.93 Ϯ 0.19 nA (n ϭ 8), reaching a new steady state in about 20 min. An example of current traces recorded at Ϫ120 mV in a transfected cell before and after TBH treatment is illustrated in Fig. 5A. Because oxidation is known to affect volume-and Ca 2ϩ -regulated Cl Ϫ currents, we investigated whether it was able to activate basal currents in cells transfected with an empty plasmid. In this case, no current was induced by oxidation (not shown), indicating that the increase in current amplitude in ClC-2-transfected cells was indeed because of ClC-2 activation. Moreover, the maximal current could be inhibited by 72.2 Ϯ 4.6% (n ϭ 4) using 300 M Zn 2ϩ . The current amplitude increase induced by the oxidant was clearly associated with faster kinetics of current activation. Fig. 5B shows that both cur-rent activation constants were decreased. The amplitude of the fast component was not statistically different before and after TBH application, whereas TBH induced an increase in the fraction of the instantaneous component at the expense of the slow component fraction (Fig. 5C). The quantitative parameters of the deactivation process are given in Table 1 and indicate that channel closure was less affected than activation. Oxidation also resulted in a significant rightward shift of the P o /V curve. Half-maximal activation was shifted from Ϫ109 Ϯ 2 to Ϫ69 Ϯ 6 mV (n ϭ 5, p Ͻ 0.005). Remarkably, preincubation of cells for 2 h with the cholesterol-m␤CD complex completely prevented the increase in current amplitude and the kinetic changes of current activation induced by TBH (Fig. 5D).
As already reported (32), when ATP was replaced by 2 mM AMP in the pipette solution, current amplitudes were higher (Ϫ2.32 Ϯ 0.31 nA (n ϭ 9) versus Ϫ1.27 Ϯ 0.10 nA (n ϭ 28) in the presence of ATP) and the rate of channel opening increased (Fig. 6A). Both slow and fast time constants of current activation were significantly decreased (Fig. 6B), and the weight of the slow component was reduced, whereas the fraction of the fast and instantaneous components were slightly increased (Fig. 6C). By contrast, current amplitude and time constants were not affected by AMP in cholesterol-enriched cells (Fig. 6, B-D).
To test the effect of oxidation on ClC-2 localization in membrane microdomains, cells were treated with 1.5 mM TBH. This induced relocalization of ClC-2 to DSM that could be reversed by cholesterol replenishment (Fig. 7A). Relocalization was prevented when cells where incubated with the cholesterol-m␤CD complex prior to oxidative stimulation (not shown). Incubation with TBH for 30 min produced a significant decrease in ClC-2 localization in DIM at the surface, without affecting total distribution in the cell (Fig. 7B). However, TBH incubation did not change membrane cholesterol levels, suggesting that ClC-2 relocalization is not because of cholesterol depletion (not shown). The effect of ATP depletion on ClC-2 distribution in DIM and DSM is shown in Fig.  7C. ATP depletion resulted in relocalization of ClC-2 to DSM.
The biochemical data indicate that cholesterol depletion, oxidation, and ATP depletion are able to displace ClC-2 from DIM, suggesting that oxidative and metabolic stress might activate ClC-2 by changing the lipid environment of the protein.
Although oxidation does not change the membrane cholesterol content (Fig. 2D), prevention of its effect by cholesterol enrichment would support this interpretation.

DISCUSSION
We report for the first time that a member of the ClC family, namely ClC-2, partially associates with cholesterol-rich mem-brane microdomains and that its channel activity is increased upon displacement from the former by either cholesterol depletion or oxidative/metabolic stress. This would be in agreement with previous studies showing that both in transfected HEK293 cells (13,33) and in native tissue (33), ClC-2 channels are not randomly distributed throughout the plasma membrane, as immunoreactivity appears localized in clusters. Our results indicate that in HEK293 cells this channel is concentrated in DIM (Fig. 1A), which could explain this patchy membrane distribution.
The presence of ClC-2 in DIM was also observed in heart, brain, and kidney, although at a lesser and variable extent, and it was diminished when the cholesterol-solubilizing detergent saponin was added to the lysis buffer. The fact that a partial cholesterol-dependent association with DIM exists in tissues endogenously expressing this channel, and that this association varies among tissues, reinforces the idea that its differential localization may have functional implications. Given that only a fraction of ClC-2 protein is present in the low density DIM, this suggests that the association of ClC-2 with DIM is either tightly regulated or unstable. This led us

Cholesterol Modulates ClC-2 Gating
to hypothesize that subtle changes in ClC-2 distribution may determine important variations in channel activity. This hypothesis was confirmed by the results of patch clamp experiments.
An obvious question concerns the mechanism involved in the association of ClC-2 with DIM domains. Several possibilities have been proposed for other integral membrane proteins, such as their direct association with lipids by the specific properties of their transmembrane domains (34,35) or by direct association with cholesterol, as shown for caveolins (36). Alternatively, protein-DIM anchoring might occur through protein-protein interactions, for instance by specific domains that mediate association with the DIM-resident flotillins (37). In addition, a cysteine-rich region has been found to be responsible for direct targeting of epidermal growth factor receptor to caveolae/rafts (38), whereas palmitoylated PSD-95 may function as a DIM scaffolding protein for Kv1.4 (39). Additional mutation experiments would be necessary to elucidate the molecular mechanism involved in ClC-2 association with DIM.
DIM disruption increased channel activity, as evidenced by faster current activation and a shift of the activation curve toward less negative potentials. Opposite effects were observed following cholesterol enrichment of the membrane. Nevertheless, the precise mechanism of ClC-2 modulation by cholesterol depletion or enrichment is presently unknown. It has been postulated that DIM function as a sorting device in the cell membrane to bring ion channels into contact with signaling proteins that regulate their function (40). In this context, according to our results, DIM may act as a reservoir of ClC-2 and a platform to allow interaction with inactivating proteins. Potential DIM-associated inactivating partners could be protein kinase C, tyrosine kinases, and epidermal growth factor receptor (8,41). Another putative regulator of ClC-2 is actin. The assembly of actin filaments at DIM and their association with the cortical cytoskeleton are required to stabilize these signaling domains (42)(43)(44). DIM disruption and dispersion of ClC-2associated proteins by m␤CD may abolish an electrostatic interaction of ClC-2 with actin, which has been shown previously to regulate the activity of this channel (45).
Alternatively, the observed kinetic changes could be due to direct effects of the altered lipid composition on channel activity and, subsequently, to significant variations in the membrane phase conformation. It has been reported that swelling-activated chloride conductance is enhanced by caveolin-1 overexpression and recruitment of channels from an intracellular pool in a cholesterol-dependent manner (46,47). Moreover, volume-regulated anion current has been shown to be increased by m␤CD treatment and inhibited by cholesterol enrichment by a mechanism involving a shift in the equilibrium between the channel open and closed states (29). ClC-2 opening most likely involves two gates as follows: a fast gate controlling the individual protopores and a slow gate controlling both protopores simultaneously (30,48). Opening of both gates is highly dependent on temperature, suggesting important conformational changes in the channel (30). The two gates are tightly coupled so that conforma-tional changes leading to the opening of the slow gate may facilitate the fast opening mechanism (30,48). Accordingly, we have shown that both components are affected by changes in membrane cholesterol content. The exact mechanism by which cholesterol may influence the gating properties of a channel is not known, but as discussed by other authors, the increased rigidity conferred to the membrane by cholesterol may impede the conformational changes necessary for channel opening (24,28). If this were the case, ClC-2 conformation might be sensitive to specific changes in the phase or rigidity of the surrounding membrane environment. In this respect, an attractive hypothesis could link cholesterol-dependent membrane fluidity with the pH-dependent movement of the R-helix described for bacterial ClC (49), possibly affecting the affinity of the central Cl Ϫbinding site (50).
Another intriguing aspect of these results is the potential effect of cholesterol on ClC-2 trafficking. It has been established that alterations in membrane cholesterol content modulate the rate of endocytosis and recycling of endosomes. For example, cholesterol depletion inhibits clathrin-coated pit budding (51), increasing the retention of several receptors or transporters at the plasma membrane (52,53), and promotes the recruitment of volume-regulated anion current channels at the membrane (54). Conversely, cholesterol increase stimulates endocytosis (55), causes Golgi vesiculation in many cell types, and impairs export of the vesicular stomatitis virus glycoprotein protein (56). Our results show a decrease in the presence of ClC-2 channels at the plasma membrane following cholesterol enrichment (Fig. 4A, panels e and f), which can account in part for the decrease in current density. The greater proportion of ClC-2 associated with DIM after cholesterol enrichment (Fig.  2B) suggests that the lower presence of ClC-2 at the surface under the same conditions (Fig. 4, A and B) occurs at the expense of DSM-associated ClC-2. As shown in Fig. 2D, cholesterol enrichment leads to a significant increase in cholesterol content only in DSM. This specific change in DSM lipid composition (not changing its detergent-solubility properties) could be involved in the decrease in overall surface ClC-2. Whether this effect implies an increased endocytosis rate or an impaired export to the plasma membrane remains to be determined, but it demonstrates that cholesterol is involved in ClC-2 trafficking.
Oxidation and reduction are known to modulate the function of a variety of proteins, including ion channels (57-60), often implying modification of cysteine residues. Two members of the ClC family have been found sensitive to oxidation, namely ClC-0 (61) and ClC-2 (11), when expressed in Xenopus oocytes. However, in both reports, the targets for oxidation-reduction have not been identified. Here we show that acute oxidation of cells results in the release of ClC-2 from DIM and activation of ClC-2 current. Likewise, we show that ATP depletion, which has been previously reported to activate ClC-2 (8,30), leads to its relocalization to DSM. This metabolic stress-dependent release of ClC-2 from DIM is in agreement with the reported partial redistribution of Glut1 out of DIM domains following azide-dependent ATP depletion (62). Moreover, we show that cholesterol enrichment completely prevented the effect of the oxidant, as well as channel activation by AMP. Therefore, these results suggest that relocalization to DSM may be a necessary step for ClC-2 activation in response to oxidative and metabolic stress. Nevertheless, a more complex mechanism may be involved, given that m␤CD did not increase current amplitude as much as oxidation and ATP depletion.
In this study we tried to decipher the relationship between channel activity and membrane composition. In this respect, activation of ClC-2 by oxidative and metabolic stress is greatly dependent on the membrane cholesterol content and probably on its rigidity and phase conformation. ClC-2 regulation by the membrane microdomain environment governed by cholesterol content presents a clear physiopathological relevance. DIM localization may provide a way to keep the protein in a low activation state. Variation of the membrane lipid composition, either as a consequence of certain pathologies, by the use of pharmacological compounds, or by altering the balance between oxidants, antioxidant defenses, and oxidation substrates, may increase or decrease channel activity and modulate the response of ClC-2 to metabolic stress. Conversely, appropriate modulation of membrane cholesterol content, as well as specific phospholipids and fatty acids, may provide a powerful tool to activate or inactivate Cl Ϫ currents in those pathologies involving an alteration of chloride secretion, such as cystic fibrosis.