The mechanisms controlling the volume-regulated anion current (VRAC) are incompletely elucidated. Here, we investigate the modulation of VRAC by cellular cholesterol and the potential involvement of F-actin, Rho, Rho kinase, and phosphatidylinositol-(4,5)-bisphosphate [PtdIns(4,5)P2] in this process. In Ehrlich-Lettre ascites (ELA) cells, a current with biophysical and pharmacological properties characteristic of VRAC was activated by hypotonic swelling. A 44% increase in cellular cholesterol content had no detectable effects on F-actin organization or VRAC activity. A 47% reduction in cellular cholesterol content increased cortical and stress fiber-associated F-actin content in swollen cells. Cholesterol depletion increased VRAC activation rate and maximal current after a modest (15%), but not after a severe (36%) reduction in extracellular osmolarity. The cholesterol depletion-induced increase in maximal VRAC current was prevented by F-actin disruption using latrunculin B (LB), while the current activation rate was unaffected by LB, but dependent on Rho kinase. Rho activity was decreased by ∼20% in modestly, and ∼50% in severely swollen cells. In modestly swollen cells, this reduction was prevented by cholesterol depletion, which also increased isotonic Rho activity. Thrombin, which stimulates Rho and causes actin polymerization, potentiated VRAC in modestly swollen cells. VRAC activity was unaffected by inclusion of a water-soluble PtdIns(4,5)P2 analogue or a PtdIns(4,5)P2-blocking antibody in the pipette, or neomycin treatment to sequester PtdIns(4,5)P2. It is suggested that in ELA cells, F-actin and Rho-Rho kinase modulate VRAC magnitude and activation rate, respectively, and that cholesterol depletion potentiates VRAC at least in part by preventing the hypotonicity-induced decrease in Rho activity and eliciting actin polymerization.
- cell swelling
- phospholipid phosphatidylinositol-(4,5)-bisphosphate
in most cell types, volume regulation after cell swelling is dependent on the activation of separate conductive pathways for K+ and Cl− efflux (24). This was first proposed for Ehrlich ascites tumor (EAT) cells (23), in which the swelling activated anion current (VRAC) has been studied extensively (9, 52). The role of VRAC is, however, not limited to cell volume regulation after swelling, but includes regulation of, e.g., cell cycle progression, angiogenesis, and apoptotic cell death (40, 45). VRAC has been extensively characterized with respect to biophysical and pharmacological properties (40, 44), yet its molecular identity is unknown, and despite some progress in recent years (40), the mechanisms by which VRAC is regulated remain incompletely understood.
The early events elicited by cell swelling and resulting in VRAC activation have been proposed to include a reduction in intracellular ionic strength (56, 58, 66), and mechanical events such as membrane unfolding (23, 44, 49). In many cells including EAT cells, osmotic cell swelling is associated with a decrease in cellular F-actin content and/or disruption of cortical F-actin (20, 26, 29, 48, 50). The role of F-actin in regulation of VRAC has been studied in several cell types with varying conclusions (29, 35, 43, 70), possibly confounded by methodological issues regarding the use of cytochalasins to disrupt F-actin (for discussion, see Ref. 34).
Changes in membrane cholesterol content have been found to modulate the function of VRAC (30, 56) and other membrane transporters and channels (16, 19, 32, 56), as well as of numerous plasma membrane receptors (5). Cholesterol is a major component of the plasma membrane ordered microdomains known as lipid rafts and caveolae (4, 60), which play important roles in the control of signaling events, membrane ion transport, and membrane trafficking (59). Cholesterol depletion has been shown to disrupt the structure of rafts (4), including caveolae (67), and consequently, the effects of cholesterol on membrane protein function are frequently taken to indicate the involvement of rafts and/or caveolae. However, 1) cholesterol is known to bind directly to membrane proteins, thereby regulating their function (17), and 2) an increase in cholesterol content generally increases bilayer thickness (8), and stiffness (37), which modulate the “membrane deformation energy,” a parameter dependent on bilayer thickness, mechanical properties, and monolayer equilibrium curvature, which has been used to describe the overall impact of these parameters on ion channel properties (25, 38). The membrane deformation energy contributes to the overall energy cost of channel opening/closing, as shown for, e.g., gramicidin channels and voltage-dependent Na+ channels (31, 32).
In a cell (as opposed to a simple lipid vesicle), the effects of cholesterol on membrane deformability reflect both the phospholipid composition of the membrane and effects of cholesterol on cytoskeletal organization (see Ref. 69). Lipid rafts and caveolae interact with the actin cytoskeleton (22, 28, 36, 57, 61; see also Ref. 59), an interaction that may be disrupted by cholesterol depletion (21), and it appears that the effects of cholesterol content on membrane function are at least in some cases due to changes in the cellular distribution and content of F-actin (3, 6, 27).
Recent studies (30, 36) in bovine aortic endothelial (BAE) cells provided evidence that VRAC activity under modest osmotic gradients is increased by cholesterol depletion, an effect suggested to reflect changes in the cost of bilayer deformation and/or a (negative) regulation of VRAC by segregation to rafts/caveolae (30, 56). This, on the other hand, seems in contrast to the recent proposal that VRAC activation is dependent on the integrity of caveolae (63), and that caveolin-1 may increase VRAC availability in the plasma membrane and/or be involved in the signaling events leading to VRAC activation (64).
Small GTP binding (G) proteins of the Rho family are major regulators of F-actin polymerization (see Ref. 49) and have been implicated in the cytoskeletal reorganization elicited by cell volume perturbations (11, 13) as well as in the regulation or modulation of VRAC by cell swelling (7, 14, 42, 47). In vascular endothelial cells (7, 42) and NIH3T3 cells (47) the activity of the Rho-associated kinase (Rho kinase; ROK) is required for VRAC activation. RhoA and ROK associate with rafts and caveola and translocate to caveolae in an activation-dependent manner (62, 65), raising the possibility that the effect of cholesterol depletion on VRAC might involve changes in Rho kinase activity.
Similarly, the phospholipid phosphatidylinositol (4,5) bisphosphate [PtdIns(4,5)P2], a major regulator of F-actin organization, has been shown to associate with lipid rafts (28, 57; see also Ref. 59), and loss or redistribution of membrane PtdIns(4,5)P2 was proposed to underlie the complex F-actin reorganization in cholesterol-depleted human skin fibroblasts (27).
Thus the present study was initiated to determine the effects of cholesterol depletion and enrichment on isotonic and swelling-activated VRAC currents, and to investigate the involvement of the F-actin cytoskeleton, Rho, Rho kinase, and PtdIns(4,5)P2 in the modulation of VRAC in ELA cells by cellular cholesterol content.
MATERIALS AND METHODS
Unless otherwise stated, reagents were from Sigma-Aldrich (St. Louis, MO), Merck (Darmstadt, Germany), or Baker (Deventer, The Netherlands), and of the highest analytical grade. DIDS was dissolved at 50 mM and niflumic acid at 5 mM, both in DMSO. Stock solutions of Y27632 and latrunculin B (LB) (both from Calbiochem, Darmstadt, Germany) were prepared at 10 mM in ddH2O and DMSO, respectively. The PtdIns(4,5)P2 analog, PtdIns(4,5)P2 dioctanoyl (Cayman, Ann Arbor, MI) was prepared as a 1 mM stock in chloroform:methanol:H2O (3:2:1), aliquotted, and stored at −80°C until use. The monoclonal PtdIns(4,5)P2 antibody was purchased from Assay Designs (Ann Arbor, MI). The antibody was diluted in PBS buffer containing 10% newborn calf serum and 0.05% NaN2, and hence the corresponding control experiments were performed in the presence of this vehicle. In a third approach to modulating PtdIns(4,5)P2 levels, PtdIns(4,5)P2 was sequestered using neomycin (Sigma-Aldrich). Neomycin is an antibiotic which has previously been shown to bind PtdIns(4,5)P2 (2) and inhibit the coupling between PtdIns(4,5)P2 and the actin cytoskeleton (27, 28). Neomycin was dissolved at 10 mM in the normal growth medium, the pH of which was readjusted with NaOH, and the cells were incubated for 22–26 h in this medium before experiments.
The model system used was Ehrlich-Lettre ascites (ELA) cells, an adherent subtype of the EAT cells, in which VRAC is well characterized (52). ELA cells were grown in RPMI-1640 medium (Sigma-Aldrich) fortified with 10% fetal calf serum and 1% penicillin/streptomycin (Invitrogen, Carlsbad, CA) at 37°C and 5% CO2. Cells were passaged every fourth day and used from passages 5 to 30.
Whole cell recording.
For patch-clamp experiments, the cells were grown on coverslips in 40 mm petri dishes. Membrane currents were recorded from single ELA cells using the fast whole cell mode of the patch-clamp technique. The isotonic bath solution (300 mosmol/l) contained (in mM) 90 NaCl, 0 KCl, 1 MgCl2, 1 CaCl2, 10 HEPES, and 110 mannitol, pH 7.4 (pH adjusted with Tris in all solutions for electrophysiology). In the hypotonic bath solution, osmolarity was decreased to 255 and 190 mosmol/l by adjusting the amount of mannitol to 65 and 0 mM, respectively. Where indicated, Na+ was replaced by N-methyl-d-glucamine (NMDG+) in equimolar amounts in the extracellular solution. The intracellular pipette solution (295 mosmol/l) contained (in mM) 90 CsCl, 2 MgCl2, 10 EGTA, 10 HEPES, 1.5 Na2-ATP, 0.1 Na2-GTP, and 80 mannitol, pH 7.4. For measurement of TRPM4 Ca2+ currents, two solutions were used. The bath solution was composed of 156 NaCl, 6 CaCl2, 1 MgCl2, 10 glucose, and 10 HEPES, and buffered at pH 7.4 with NaOH. The pipette solution was composed of 20 CsCl, 100 CsAsp, 1 MgCl2, and 10 HEPES; pH 7.4 with NaOH. The osmolarity of solutions was routinely checked with the use of an osmometer (Knaur; Berlin, Germany).
Patch pipettes were pulled from borosilicate glass capillaries of 1.7 mm OD (Hounisens Laboratories, Riskov, Denmark) on a pipette puller (model pp-830; Narishige, Tokyo, Japan). The filled pipettes had resistances of 4.5–6.5 MΩ. A hydraulic micromanipulator (model WR-6, Narishige) was used for the positioning of pipettes and an Ag−/AgCl wire served as the reference electrode. For anion selectivity studies, the reference electrode was 94 mM KCl in 3% agar.
Pipette offsets, series resistances (not >11 MΩ), and capacitive transients were compensated on the patch-clamp amplifier (model EPC7, List Electronic, Darmstadt, Germany). Currents were digitized with an analog-to-digital converter (model 1401+, Cambridge Electronic Design) at 500 Hz and filtered with the built-in four-pole Bessel filter at 3 kHz. Data acquisition and analysis were done with the Cambridge Electronic Design patch- and voltage-clamp software (version 6.41). The holding potential was −40 mV and voltage ramps from −80 to +80 mV and of 2.6-s duration were applied every 15 s, with a 500-ms prepulse at −80 mV preceding each ramp. Time-dependent kinetics were investigated by stepping the potential from −80 to +80 mV in 20 mV steps of 1 s, separated by 2-s intervals at the holding potential. Cells were continuously perfused at a rate of 2 ml/min with a complete exchange of the bath solution every 0.5 min. All anion channel experiments were performed at room temperature (18–21°C). However, because TRPM4 is not active at room temperature in ELA cells, these experiments were performed at 37°C by mounting recording chamber on a heated stage system (Broke).
Modulation and estimation of cellular cholesterol level.
ELA cells were depleted of cholesterol by incubation with 5 mM methyl-β-cyclodextrin (MβCD) without cholesterol in RPMI-1640 medium for 1 h. Cholesterol-enriched ELA cells were prepared by incubation with 5 mM MβCD saturated with cholesterol (625 μM) for 3 h. The MβCD-cholesterol solution was prepared as previously described (10). To evaluate the cholesterol content, lipids were extracted from cell monolayers by incubating these for 30 min in isopropanol at room temperature, while lipid free cell content was subsequently solubilized in 1 M NaOH. This procedure quantitatively extracts all lipids (33). Extracted lipids were concentrated by partially evaporating isopropanol in purpose made apparatus, and cholesterol content was subsequently measured using the “cholesterol CHOD-PAP method” (Roche, Basel, Switzerland). This is a standard enzymatic reaction for spectrophotometric analysis, where the cholesterol content is linear to the absorbance at 500 nm. Cell protein was estimated from the lipid free NaOH solution using a BCA protein kit (Bio-Rad, Hercules, CA) with bovine serum albumin as a standard.
Neomycin pretreatment to sequester cellular PtdIns(4,5)P2.
PtdIns(4,5)P2 was sequestered using neomycin (Sigma-Aldrich). Neomycin is an antibiotic which has previously been shown to bind PtdIns(4,5)P2 (2) and inhibit the coupling between PtdIns(4,5)P2 and the actin cytoskeleton (27, 28). Neomycin was dissolved at 10 mM in the normal growth medium, the pH of which was readjusted with NaOH, and the cells were incubated for 22–26 h in this medium before experiments.
Immunocytochemistry and confocal laser-scanning microscopy.
For immunocytochemistry experiments, cells were grown to a confluence of 60–70% on no. 1 glass coverslips. Where indicated, cells were enriched or depleted of cholesterol as described above. For disruption of F-actin using LB, cells were incubated with LB (10 μM) for 1 h during the cholesterol depletion treatment. Before fixation, the cells were incubated for 2 min in either standard isotonic salt solution composed of (in mM) 143 NaCl, 5 KCl, 1 MgSO4, 1 Na2HPO4, 1 CaCl2, 3.3 MOPS, 3.3 TES, and 5 HEPES, or, where indicated, in the mannitol-containing isotonic solution used for electrophysiological experiments. The cells were fixed in 4% paraformaldehyde for 15 min on ice, permeabilized for 10 min in 0.1% Triton X-100, and blocked for 30 min in blocking buffer (PBS) composed of (in mM) 136.89 NaCl, 2.68 KCl, 8.1 Na2HPO4, and 1.47 KH2PO4 with 2% BSA. F-actin was stained with either FITC- or rhodamine-conjugated phalloidin (2 U/ml, Molecular Probes, Leiden, The Netherlands) in blocking buffer for 45 min at room temperature. The nuclei were stained with 4,6-diamidino-2-phenylindole (DAPI; 1:300, Molecular Probes) for 5 min in blocking buffer. After being stained, the cells were washed extensively in blocking buffer and mounted using N-propyl galleate mounting medium (Sigma-Aldrich). Confocal microcopy was performed using a Leica DM IRB/E microscope coupled to a confocal laser scanning unit (model TSC NT; Leica Lasertechnik, Heidelberg, Germany). Excitation of DAPI, FITC, and rhodamine was carried out using the 364 nm UV, and the 488 or 567 nm argon/krypton laser lines, respectively. Images were taken using a ×40/1.25 numerical aperture planapochromat objective, a 1.0 Airy disk pinhole, and an optical slice thickness of 1 μm. When multiple dyes were used, images were acquired by sequential scanning. Images (512 by 512 pixels) were frame averaged and presented in RGB pseudocolor.
Purification of Rho-GTP.
Rho-GTP content was estimated using a commercial pull-down assay (Cytoskeleton, Denver, CO). Briefly, the cholesterol content of cells grown to 80% confluence in 10 cm petri dishes was modulated as described above, and cells were washed once with PBS and stimulated for 3 min with isotonic or hypotonic solution at room temperature. Cells were quickly washed in PBS, 500 μl of lysis binding buffer containing a protease inhibitor cocktail were added, cells were scraped off, and homogenates were transferred to Eppendorf tubes. Homogenates were cleared by centrifugation at 20,000 g for 5 min at 4°C. The cleared lysates (500 μl) were used for analysis of Rho-GTP content, and the remaining lysate for protein measurement and estimation of total Rho, as follows: 500 μl of cleared lysate were transferred to 20 μl of RBD purification bead solution, and 5 μl of protein inhibitor cocktail were added to the solution. Lysate and RBD beads were incubated at constant rotation for 1 h at 4°C. RBD beads were washed once with lysis binding buffer and once with washing buffer (5,000 g, 3 min, 4°C). RBD beads with bound Rho-GTP were spun down as above, the supernantant carefully removed, and the RBD bead pellet dissolved in 10 μl of NuPage LDS sample buffer (Invitrogen, Carlsbad, CA) containing 29% DTT and frozen at −80°C. Total Rho and purified Rho-GTP content were analyzed by SDS-PAGE and Western blot analysis as described below. For positive controls, the cells were transfected with Myc-tagged, constitutively active Rho (RhoQ63L, a kind gift from Dr. Andras Kapus, St. Michael's Hospital, Toronto, ONT, Canada), using Lipofectamine 2000 (Invitrogen), and 4 μg of DNA per 10 cm dish. Forty-eight hours after transfection, the cells were lysed and Rho-GTP isolated as described above.
SDS-PAGE and Western blot analaysis.
Total Rho and precipitated Rho-GTP protein samples were resolved by SDS-PAGE under denaturating and reducing conditions using 12% Bis-Tris gels with NuPage Mes SDS running buffer and an XCell system (model E19001; Novex, San Diego, CA). The protein concentration of cleared total cell lysates was estimated using a BCA protein kit (Bio-Rad) with BSA as standard. The amount of protein loaded per well was 20 μg for total Rho, and an amount corresponding to 0.74 mg of unpurified whole cell protein for purified Rho-GTP. Separated proteins were electrotransferred to Protran nitrocellulose membranes (Schleicher and Schuell, Keene, NH), membranes were stained with 1% Ponceau S Red solution (Sigma-Aldrich), blocked in 5% nonfat dry milk in 1× TBST (composed of 0.01 M Tris·HCl, pH 7.4, 0.15 M NaCl, and 0.1% Tween 20) for 75 min, incubated for 2 h with monoclonal anti-Rho antibody (Cytoskeleton) 1:500 in blocking buffer, washed extensively in TBST, and incubated for 1 h in with alkaline phophatase-conjugated goat-anti mouse antibody (Jackson ImmunoResearch, West Grove, PA), 1:500 in blocking buffer, all at room temperature. Membranes were washed extensively in TBST and developed using BCIP/NBT solution (Kirkegaard and Perry Laboratories, Gaithersburg, MD). Band intensity was evaluated by densitometric scanning using UN-SCAN-IT software (version 5.1, Silk Scientific, Orem, UT).
Data treatment and statistical analysis.
Data are presented as means ± SE, with n denoting the number of cells tested. For comparison of data sets, Student's t-tests for paired and unpaired data were employed as appropriate. In patch-clamp experiments, the percent inhibition by DIDS and niflumic acid were calculated as (1) where Icontrol refers to the control current, and Idrug refers to the current measured in the presence of the drug. Permeability sequences were calculated from the experimentally determined Vrev shifts using the modified Goldman-Hodgkin-Katz equation as follows: (2) where Px/PCl gives the relative permeability of the given anion (X) to Cl−; [Cl−]o is the Cl− concentration in the standard bath solution, [Cl−]s and [X−]o are the concentrations of Cl− and the substituting anion in the substituted medium, respectively, ΔVrev is the measured shift in reversal potential, and F, R, and T have their usual meanings.
Characterization of swelling-activated Cl− current in ELA cells.
VRAC has not previously been characterized in ELA cells, and it was therefore important to first establish the biophysical and pharmacological characteristics of the current. Perfusion with hypotonic solution resulted in cell swelling (not shown) and development of a current which was readily reversible upon return to isotonic conditions, and which exhibited time-dependent inactivation and flat tail currents at depolarized potentials (Fig. 1A). The current was outwardly rectifying and the current magnitude was greater after exposure to a 36% hypotonic solution (190 mosmol/l) than to a 15% hypotonic solution (255 mosmol/l) (Fig. 1, B–D). The current Vrev was −12 ± 0.5 mV (n = 46) at 190 mosmol/l, compared with the calculated Vrev for Cl− at 0 mV under these conditions. To investigate the possible contamination by cation currents, several experiments were performed, in which all extracellular Na+ was substituted by NMDG+. This resulted in a small shift in Vrev (−13.61 ± 0.63 mV to −15.01 ± 0.47 mV, n = 7, P < 0.01). However, returning to NaCl solution did not cause the reciprocal change in Vrev (−14.46 ± 1.04 mV, n = 7; P > 0.5), indicating that the shift in Vrev was not due to cation current modulation. Instead, as discussed in Ref. 53, this deviation from the theoretical value is likely to reflect a dilution of the intracellular solution, arising because the osmotic water inflow into Ehrlich cells is faster than the solution exchange via the patch pipette. From the Nernst equation, it may be calculated that the discrepancy between the calculated and experimental Vrev corresponds to a 38% dilution of the intracellular solution in the 190 mosmol/l bath solution. In the 15% hypotonic solution, Vrev was −5.63 ± 0.95 mV (n = 4) in the presence, and −9.78 ± 1.07 (n = 4, P < 0.05) in the absence of Na+. Because this shift was reversible when the cells were reexposed to the Na+-containing bath solution, this indicates that a cation current is responsible for a minor fraction of the current measured in the standard hypotonic solution. The relative permeabilities of the current to various anions were estimated from the shifts in Vrev upon anion substitution, using the modified Goldman-Hodgkin-Katz equation (Fig. 1E). The anion permeability sequence obtained was the following: PSCN: PI: PBr: PCl: PAsp = 1.3 ± 0.03 (n = 8): 1.2 ± 0.02 (n = 8): 1.1 ± 0.01 (n = 8): 1: 0.06 ± 0.004 (n = 4), i.e., an Eisenman I permeation sequence. A linear correlation was noted between the relative permeability of the ion and the Jones-Dole viscosity B coefficient, a measure of ion-water interactions (Fig. 1F; see also discussion). Finally, application of DIDS (50 μM) inhibited the current in a voltage-dependent manner, with 75% and 11% inhibition at +40 mV and −55 mV, respectively, whereas inhibition by another Cl− channel inhibitor, niflumic acid (5 μM), was small and voltage-independent, calculated at ∼20% and 14%, respectively, at +40 and −55 mV (Fig. 1, G and H). Thus the swelling-activated Cl− current in ELA cells exhibits characteristics similar to those described for the swelling-activated anion current, VRAC, in a wide range of cell types. Specifically, it may be noted that the only notable difference between the VRAC currents in ELA cells and the suspension cell line EAT cells, of which ELA cells are an adherent subtype, is that the degree of voltage-dependent inactivation in ELA cells observed in the present study was substantially lower than that in EAT cells (52).
Effects of cholesterol enrichment and depletion on activation of VRAC.
The next series of experiments investigated the relationship between cellular cholesterol content, F-actin organization, and VRAC activity. Cellular cholesterol content was modified using MβCD in the absence or presence of added cholesterol (see materials and methods). This resulted in a 47% increase, or a 44% decrease, respectively, in total cholesterol content in ELA cells (Fig. 2A).
Cholesterol enrichment had no effect on VRAC activation after either a 190 mosmol/l or a 255 mosmol/l hypotonic challenge (Fig. 2, B and C). In contrast, cholesterol depletion significantly potentiated VRAC after exposure to 255 mosmol/l of bath solution, but not after the more severe swelling in 190 mosmol/l solution (Fig. 2, D and E). The initial current activation rate, as calculated by linear regression to currents measured between 15 and 90 s after 255 mosmol/l hypotonic challenge was significantly increased in depleted cells. Thus an activation rate of 10.32 ± 1.86 pA pF−1 min−1 was found in depleted cells, whereas the rate was only 4.78 ± 0.78 pA pF−1 min−1 in control cells (P < 0.05).
As pointed out above, there was a small cation component in the current activated by 255 mosmol/l Na+-containing bath solution. However, the shift in Vrev upon switching to 255 mosmol/l NMDG+ solution was similar in depleted cells and control cells (−4.15 ± 1.30 mV and −3.17 ± 0.47, respectively, n = 4, P = 0.51), indicating that the cation permeability of cholesterol depleted cells is similar to that of control cells under these conditions, and hence that the augmenting effect of cholesterol depletion on the current measured in osmotically swollen cells is predominantly or exclusively an effect on VRAC.
We next investigated whether cholesterol depletion similarly activates VRAC in the absence of osmotic swelling. Under isotonic conditions, Vrev in both control (n = 9) and cholesterol depleted (n = 7) cells shifted significantly toward more negative values when Na+ was replaced with NMDG+, indicating a significant contribution from a cation current. Hence, characterization of the cholesterol sensitivity of the isotonic anion current was performed in both Na+ and NMDG+ bath solutions (Fig. 3). As shown, an outwardly rectifying current with time dependent inactivation at depolarized potentials, i.e., highly similar to VRAC, was active in both control and cholesterol depleted cells in NMDG Ringer (Fig. 3, A–D). Similar to the swelling-activated VRAC current, this current was significantly increased in depleted cells compared with control cells, indicating that cholesterol depletion partially activates VRAC under isotonic conditions. The cholesterol depletion-induced increase in isotonic current magnitude was, if anything (P = 0.059), greater in NMDG+ than in Na+ bath solution, substantiating the notion that the cholesterol depletion-sensitive current is carried by anions (Fig. 3, E and F).
Effects of cholesterol enrichment and depletion on F-actin organization.
F-actin organization has been shown to be altered by changes in cellular cholesterol content, and has been implicated in swelling-induced VRAC activation (see Introduction). The effect of cholesterol enrichment and depletion on F-actin organization in ELA cells was assessed by confocal laser scanning microscopy of phalloidin-labeled cells (Fig. 4). As shown in Fig. 4A, neither cholesterol enrichment nor depletion had any detectable effect on cellular morphology. Images from the bottom, middle, and top of control cells (images a–c) illustrate the typical appearance of stress fibers and cortical F-actin in ELA cells. Cholesterol enrichment had no detectable effect on cellular F-actin organization and content (images g–i). In contrast, cholesterol depletion was associated with a marked increase in F-actin labeling intensity, apparently both due to increased cortical F-actin polymerization and increased stress fiber formation (images d–f). Figure 4B shows the quantification of pixel intensities in the phalloidin-labeled cells, under isotonic and hypotonic conditions, in control cells and after cholesterol depletion and enrichment, respectively. As shown, in control cells both cytoplasmic (i.e., predominantly stress-fiber associated) and cortical F-actin levels were reduced by ∼50% after 2 min of hypotonic swelling, consistent with previous findings in EAT cells (53). Under isotonic conditions, cholesterol depletion tended to increase cytoplasmic F-actin, although not quite significantly (P = 0.11), whereas cholesterol enrichment significantly decreased the cortical, but not the cytoplasmic, F-actin level. Cholesterol depletion abolished the decrease in F-actin seen in control cells under hypotonic conditions, and in fact resulted in a significant increase in F-actin levels compared with isotonic control conditions. In contrast, cholesterol enrichment had no effect on the swelling-induced decrease in F-actin levels. The images shown in Fig. 4 are obtained in cells exposed to mannitol-containing, low ionic strength isotonic saline as in the electrophysiology experiments, however, similar results were obtained in the standard isotonic saline (n = 3 and n = 6 for cholesterol enrichment and cholesterol depletion, respectively, data not shown).
Effects of LB on F-actin organization and VRAC activity.
Pretreatment of ELA cells with LB, which reduces the cellular content of polymerized actin by sequestering actin monomers, lead to the essential disappearance of both F-actin stress fibers and cortical F-actin in the ELA cells (Fig. 5A). In control cells, LB treatment had no significant effect on VRAC under hypotonic (Fig. 5B) or isotonic (Fig. 5F) conditions. In cholesterol depleted cells, LB treatment reduced the hypotonic current magnitude to a value similar to that in undepleted control cells (Fig. 5C), whereas in contrast, the isotonic current in depleted cells was further potentiated by LB (Fig. 5G). Notably, the difference in activation rates between control and depleted cells was preserved despite the LB treatment. Thus activation rates of 1.86 ± 0.56 and 6.50 ± 1.18 pA pF−1 min−1 where observed in control and depleted cells, respectively (P < 0.01), yet the cholesterol depletion-induced increase in maximal VRAC current was abolished (Fig. 5D). The maximal swelling-induced VRAC currents are summarized in Fig. 5E.
The above data indicate that the cholesterol depletion-induced increase VRAC might be dependent on the increase in stress fiber content seen in the cholesterol depleted cells. Consistent with this notion, the serine protease thrombin, which stimulates stress fiber formation by activating the small G protein Rho (e.g., Ref. 18), significantly increased the VRAC current magnitude when present (1 U/ml) simultaneously with a modest hypotonic challenge (255 mosmol/l). Similar to cholesterol depletion, thrombin had no effect on VRAC after a more severe hypotonic challenge (190 mosmol/l) (Fig. 6).
Effect of hypotonicity and cholesterol depletion on Rho activity.
Given the above results, in conjunction with the previous findings that Rho is sensitive to cell volume changes (11), and has been shown by us and others to modulate VRAC (42, 47), we hypothesized that cholesterol depletion might alter VRAC currents by affecting Rho activity, and we therefore evaluated Rho activity in the ELA cells using a Rho-GTP pull-down assay (Fig. 7). Rho activity is shown as the amount of active Rho protein relative to total Rho protein. As expected, Rho activity was increased by ∼75% after a 3-min exposure to thrombin (1 U/ml). In contrast, Rho activity was decreased by ∼50% and ∼20% after a 3-min exposure to 190 or 255 mosmol/l hypotonic conditions, respectively. Importantly, in cholesterol depleted cells, Rho activity was significantly increased under isotonic conditions, and the decrease in Rho activity after 255 mosmol/l, yet not after 190 mosmol/l hypotonic conditions, was abolished.
Effect of Rho kinase inhibitor Y27632 on VRAC.
The Rho-Rho kinase pathway is a major regulator of F-actin organization, and we and others have implicated this pathway in VRAC regulation (42, 47). The maximal current magnitude after hypotonic swelling is similar in control cells and in cells treated with Y27632 to inhibit Rho kinase (Fig. 8, A and B). However, the rate of VRAC activation is reduced by Y27632 treatment. As seen in Fig. 8, A and B, a current significantly greater than the isotonic current value develops after 0.75 min in control cells exposed to a 15% hypotonic challenge, yet only after 2.25 min in Y27632-treated cells. An essentially similar pattern was seen in cholesterol-depleted cells (Fig. 8, C and D). Thus, under hypotonic conditions, Rho kinase activity appears to have a stimulatory effect on the rate of VRAC activation, whereas the maximal VRAC magnitude is unaffected. In congruence with this notion, the initial current activation rate (calculated by linear regression analysis of the current 15 to 90 s after hypotonic challenge) after exposure to a 15% hypotonic challenge was decreased nearly 2.5-fold by Y27632 treatment [4.78 ± 0.79 pA pF−1 min−1 (n = 7) in control cells, vs. 1.85 ± 1.18 pA pF−1 min−1 (n = 6) in Y27632-treated control cells, P = 0.0585]. Similarly, in cholesterol-depleted cells, the initial current activation rate was numerically decreased ∼2-fold by Y27632 treatment, from 10.31 ± 1.86 pA pF−1 min−1 (n = 7) to 5.96 ± 2.08 pA pF−1 min−1 (n = 6) in Y27632-treated depleted cells (P = 0.146). Rho kinase inhibition had no effect on the isotonic current magnitude, neither in control cells nor after cholesterol depletion, and did not affect the cholesterol-induced increase in VRAC (Fig. 8, E–G).
Role of PtdIns(4,5)P2 in modulation of VRAC.
PtdIns(4,5)P2 has been implicated in the effects of cholesterol depletion on F-actin, and we therefore evaluated the possible involvement of PtdIns(4,5)P2 in VRAC activation. Inclusion of either 10 or 20 μM of the water soluble PtdIns(4,5)P2 analog PtdIns(4,5)P2 1,2 dioctanoyl in the patch pipette had no detectable effect on VRAC current density [the steady-state current density after osmotic swelling was 25.95 ± 4.00 pA pF−1 (n = 6) in control cells and 20.78 ± 2.52 pA pF−1 (n = 7) with 10 μM PtdIns(4,5)P2 1,2 dioctanoyl, respectively]. Inclusion of a blocking PtdIns(4,5)P2 antibody in the pipette to conversely reduce the amount of available PtdIns(4,5)P2 was also without effect on VRAC activity [the steady-state current density after osmotic swelling was 23.35 ± 10.86 pA pF−1 (n = 6) under control conditions and 26.13 ± 7.20 pA pF−1 (n = 7) in the antibody-treated cells]. Finally, preincubation of the cells with neomycin to sequester PtdIns(4,5)P2 failed to affect VRAC activity [the steady-state current density after osmotic swelling was 23.44 ± 3.35 pA pF−1 (n = 8) under control conditions and 18.39 ± 3.39 pA pF−1 (n = 8) after neomycin treatment]. To ascertain that the lack of effect of PtdIns(4,5)P2 manipulation was not an experimental problem reflecting, e.g., poor solubility of PtdIns(4,5)P2 dioctanoyl, we tested the effect of this compound on the transient receptor potential channel, TRPM4. Several TRPM channels are stimulated by PtdIns(4,5)P2 (51), and PtdIns(4,5)P2 has specifically been shown to rescue TRPM4 from desensitization (71). Inclusion of the PtdIns(4,5)P2 analog (10 μM) in the pipette abolished TRPM4 desensitization as expected for this current [the current density at time 5.5 min after activation of TRPM4 by 0.5 μM Ca2+ via the pipette was 1.84 ± 0.21 pA pF−1 (n = 3) under control conditions and 17.95 ± 5.51 pA pF−1 (n = 3) with 10 μM PtdIns(4,5)P2 1,2 dioctanoyl in pipette solution]. It may also be noted that the same neomycin treatment dramatically affects phosphorylation of ezrin/radixin/moesin proteins in ELA cells (S. F. Pedersen, unpublished observations), suggesting that the lack of effect on VRAC is not due to, e.g., poor incorporation of the drug into ELA cells. Thus our findings indicate that neither increases nor decreases in the cellular PtdIns(4,5)P2 level detectably affect VRAC in ELA cells, arguing against a role of this phospholipid in regulation of the current.
The VRAC current in ELA cells was activated in a dose-dependent manner by a reduction in extracellular osmolarity, was outwardly rectifying, exhibited slight time-dependent inactivation at depolarized potentials, and was inhibited by NPPB and DIDS, the latter in a voltage-dependent manner. The permeation sequence was SCN− > I− > Br− > Cl− > Asp−, consistent with an Eisenman I profile. Px/PCl correlated inversely with the Jones-Dole viscosity B coefficient, indicating a reverse correlation between hydration energy and permeability for smaller anions (12, 54, 55). This is indicative of low-energy interactions between ion and channel as also indicated by the Eisenman I sequence (68), and indicates that although VRAC is seemingly a water-permeable channel (39), it is not a simple pore for hydrated ions. Thus VRAC in ELA cells exhibits biophysical and pharmacological characteristics similar to those reported in other cell types (see Refs. 40, 44, 52).
A 44% reduction in cellular cholesterol content resulted in a significant potentiation of VRAC in ELA cells after mild (255 mosmol/l), but not after severe (190 osmol/l), hypotonic stress, in accordance with findings in BAE cells (30). Levitan and coworkers (30) also pointed out the fact that cholesterol depletion activates VRAC only after mild osmotic stress is consistent with a cholesterol-depletion induced increase in the fraction of open channels, rather than an effect on single channel conductance, a notion supported by observation of single-channel events (56). In contrast, an increase in cellular cholesterol content of ∼47% had no detectable effect on VRAC in ELA cells, in apparent contrast to findings in BAE cells (30). This difference is not due to insufficient cholesterol enrichment in ELA cells because the relative cholesterol enrichment was in fact greater than that in the BAE cells (30). More likely, it reflects that the effect of cholesterol on VRAC is already maximal in ELA cells at the normal cellular cholesterol content and under the osmotic conditions tested. Absolute membrane cholesterol levels under control, depleted, and enriched conditions were similar in ELA and BAE cells; hence, other differences in membrane/cortical structure and/or VRAC sensitivity to cholesterol in the two cell types appear likely. In congruence with this, cell type-specific effects of cholesterol were reported for the Na+/H+ exchanger, NHE1 (16). Notably, NHE1, which is a shrinkage-activated transporter, was activated by cholesterol enrichment (16), i.e., an effect of cholesterol opposite to that on VRAC. In conjunction with the finding that cholesterol depletion activated a VRAC-like current in ELA (the present study) and BAE cells (56) also in the absence of cell swelling, this might suggest that changes in cholesterol content affect the volume signal per se or some component necessary for its transduction, and hereby increases the fraction of open VRAC channels. Clearly, changes in cholesterol content are not the volume signal per se: so, what might this putative volume- and cholesterol-sensitive parameter be? It is unlikely to be membrane thickness, which, as estimated from membrane capacitance, is not generally altered by cell swelling (see Ref. 49), a notion confirmed in ELA cells (A. Bergdahl, T. Kjær Klausen, and E. K. Hoffmann, unpublished observations). Cholesterol depletion/enrichment had no detectable effect on membrane capacitance in ELA cells (control cells 36 ± 2.4 pF, n = 21; depleted 36 ± 3.5 pF, n = 13; enriched 31 ± 1.9 pF; n = 14), similar to findings in BAE cells (30). A putative volume- and cholesterol-sensitive parameter could be membrane deformability or stiffness, which is dependent on cortical F-actin. Cholesterol depletion completely abolished the swelling-induced decrease in cortical and stress fiber-associated F-actin in ELA cells, consistent with the view that cell stiffness is increased. Whereas in isolated lipid bilayers, cholesterol depletion is associated with a reduction in stiffness and deformation energy (e.g., Refs. 32 and 37), it appears that the presence of a cortical cytoskeleton may result in the opposite effect on the total elastic properties of the membrane. Thus, in BAE cells, cholesterol depletion was associated with an increase in membrane stiffness which was abolished upon disruption of F-actin by latrunculin A (6). Similarly, in sperm cells, cholesterol depletion was associated with an increase in F-actin content (3). In human skin fibroblasts, cholesterol depletion elicited a more complex F-actin reorganization, including an apparent decrease in stress fiber content, increase in cortical F-actin content, and increased F-actin turnover (27).
Cell swelling is generally associated with F-actin depolymerization, whereas the converse is true for cell shrinkage (see Ref. 49). In ELA cells, the isotonic VRAC current induced by cholesterol depletion was strongly potentiated by F-actin disruption by LB, whereas after cell swelling, LB prevented the cholesterol depletion-induced increase in maximal current magnitude, yet not the cholesterol depletion-induced increase in VRAC activation rate. Taken together, it appears that in ELA cells, F-actin integrity and/or an increase in F-actin is required on the one hand for limiting isotonic VRAC activity in cholesterol-depleted cells, and on the other hand for the cholesterol-depletion-induced increase in VRAC magnitude under hypotonic conditions. Such a scenario would resemble the regulation of the Na+-K+-2Cl− cotransporter NKCC1 by hypertonic shrinkage and F-actin, which appears to be such that F-actin integrity is required both to maintain NKCC1 silent under isotonic steady-state conditions, and for its shrinkage-induced activation (see Ref. 49). Moreover, the differential effects of F-actin disruption on VRAC maximal current and activation rate, respectively, indicate that there are at least two components in the potentiation of VRAC by cholesterol depletion, one F-actin dependent and the other F-actin independent.
The mechanism(s) of regulation of VRAC by F-actin in cholesterol-depleted cells remains to be elucidated. Cholesterol depletion-induced reinforcement of cortical and stress-fiber associated F-actin might somehow facilitate volume sensing, maybe due to a dependence on the overall cytoskeletal tension in the cell, as reported for mechanosensing (e.g., Ref. 1). In apparent contrast to our findings and those of Levitan and co-workers, potentiation of hypotonic VRAC currents by cytochalasin D has been observed in human epithelial cells, and proposed to reflect modulation of VRAC by the F-actin-dependent membrane spring energy (35) and/or augmentation of VRAC activity by membrane unfolding (44). It would appear that at least two opposing effects of F-actin may be working on VRAC, and that their relative roles may differ between cell types, resulting in different net effects of F-actin disrupting agents: 1) an “unfolding” effect, by which F-actin depolymerization (here induced by swelling or by LB) potentiates VRAC, and 2) a potentiation of volume sensing, by which F-actin polymerization (here induced by cholesterol depletion) conversely potentiates VRAC. While not directly addressed here, it should also be kept in mind that different cellular F-actin pools (cortical, stress fiber associated, etc.) may be involved in the various effects of F-actin on VRAC. Further experiments are required to address this, and to verify whether the apparently contradictory findings reflect cell type specific effects or differences in experimental protocols. With respect to the latter it may be noted that high concentrations of cytochalasin D (>1 μM), compared with 5 μM as used in (35) can in fact increase, rather than decrease, actin polymerization (15).
PtdIns(4,5)P2 is a major regulator of F-actin organization, and cholesterol depletion was found to reduce PtdIns(4,5)P2 levels in the plasma membrane (27). Moreover, cell swelling is generally, and also in EAT cells, associated with decreased PtdIns(4,5)P2 levels while the converse is true for cell shrinkage (see Ref. 49). However, neither increasing [using the water-soluble PtdIns(4,5)P2 analog PtdIns(4,5)P2 1,2-dioctanoyl] nor decreasing [using neomycin sequestering or PtdIns(4,5)P2 blocking antibodies] the cellular PtdIns(4,5)P2 level had any detectable effect on VRAC activation in ELA cells, whereas a positive control experiment showed substantial effect of PtdIns(4,5)P2 dioctanoyl on TRPM4 in the same cells. Thus, while such a role of course cannot be unequivocally excluded based on negative findings, the results obtained in the present study do not support a role for PtdIns(4,5)P2 in regulation of VRAC.
Rho family G proteins and their effectors are major regulators of F-actin organization, and their activity has been shown to be increased by cell shrinkage (11). Consistent with this, Rho activity was dose-dependently attenuated by ELA cell swelling. Notably, cholesterol depletion increased isotonic Rho activity and prevented the swelling-induced attenuation of Rho activity after mild hypotonic exposure (255 mosmol/l). The fact that thrombin, which activates Rho and promotes F-actin polymerization, stimulates VRAC (Fig. 6), is consistent with potentiation of VRAC by Rho activity and F-actin polymerization. If this is so, one would expect that under normal conditions, swelling-induced Rho inhibition and F-actin depolymerization tend to attenuate VRAC activity, and hence, that cholesterol depletion may increase the maximal VRAC current (which was inhibited by LB, i.e., F-actin dependent, but unaffected by Y27632, i.e., not Rho kinase dependent), and the rate of VRAC activation (which was Rho kinase but not F-actin dependent) by increasing Rho activity and causing actin polymerization. These findings are in good agreement with previous reports by us and others showing that while the Rho-ROK pathway is not the volume-sensing mechanism per se, it has a potentiating effect on swelling-induced VRAC activity (7, 41, 42, 47). In contrast to the hypotonic VRAC current, the isotonic current, which was also stimulated by cholesterol depletion, was potentiated by F-actin disruption and unaffected by Rho kinase inhibition. This may indicate that the mechanism of stimulation by cholesterol depletion may be at least partially different between these two conditions.
The mechanism by which cholesterol depletion affects Rho activity remains to be determined. RhoA and Rho kinase exhibit activity-dependent translocation to caveolae (62, 65), suggesting a role for cholesterol depletion-induced changes in caveolae organization. Interestingly, caveolae are also sensitive to cell volume perturbations (46), and have been implicated in VRAC regulation (63, 64). On the other hand, preliminary studies revealed no detectable changes in caveolin-1 distribution in cholesterol-depleted ELA cells (S. F. Pedersen, T. Kjær Klausen, and E. K. Hoffmann, unpublished observations). Thus further studies are required to determine the relative roles in the cholesterol-depletion-induced effects on VRAC of the following: 1) changes in membrane stiffness resulting from cytoskeletal reorganization, 2) effects on Rho signaling, and 3) caveolar reorganization.
In conclusion, in ELA cells, cholesterol depletion increased VRAC activation rate and maximal current after a modest (15%), but not after a severe (36%) reduction in extracellular osmolarity. The cholesterol depletion-induced increase in maximal VRAC current and activation rate were dependent on F-actin and ROK, respectively, whereas changes in cellular PtdIns(4,5)P2 had no effect on VRAC. Rho activity was decreased in osmotically swollen cells, and this reduction was prevented by cholesterol depletion, which also increased Rho activity under isotonic conditions. It is suggested that in ELA cells, F-actin, and the Rho-Rho kinase pathway modulate VRAC activity, and that cholesterol depletion increases VRAC currents at least in part by preventing the hypotonicity-induced decrease in Rho activity and promoting F-actin polymerization.
The study was supported by the Danish Natural Sciences Research Foundation Grants 21-01-0507 and 21-04-0535 (to E. K. Hoffmann and S. F. Pedersen) and the Carlsberg Foundation (to C. Hougaard and E. K. Hoffmann).
We are indebted to Dr. Andreas Bergdahl for performing the experiments demonstrating the effect of PtdIns(4,5)P2 on TRPM4. We also gratefully acknowledge Dr. Claus Helix Nielsen (Quantum Protein Centre, Technical University of Denmark), and Robert Nielsen and Susanne Larsen (Institute of Molecular Biology and Physiology, University of Copenhagen) for stimulating discussions. Finally, we express our gratitude to Dr. Andras Kapus (St. Michael’s Hospital, Toronto, ONT, Canada) for kindly providing the RhoQ63L construct.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society