ClC-2 Contributes to Native Chloride Secretion by a Human Intestinal Cell Line, Caco-2*

It has been previously determined that ClC-2, a member of the ClC chloride channel superfamily, is expressed in certain epithelial tissues. These findings fueled speculation that ClC-2 can compensate for impaired chloride transport in epithelial tissues affected by cystic fibrosis and lacking the cystic fibrosis transmembrane conductance regulator. However, direct evidence linking ClC-2 channel expression to epithelial chloride secretion was lacking. In the present studies, we show that ClC-2 transcripts and protein are present endogenously in the Caco-2 cell line, a cell line that models the human small intestine. Using an antisense strategy we show that ClC-2 contributes to native chloride currents in Caco-2 cells measured by patch clamp electrophysiology. Antisense ClC-2-transfected monolayers of Caco-2 cells exhibited less chloride secretion (monitored as iodide efflux) than did mock transfected monolayers, providing the first direct molecular evidence that ClC-2 can contribute to chloride secretion by the human intestinal epithelium. Further, examination of ClC-2 localization by confocal microscopy revealed that ClC-2 contributes to secretion from a unique location in this epithelium, from the apical aspect of the tight junction complex. Hence, these studies provide the necessary rationale for considering ClC-2 as a possible therapeutic target for diseases affecting intestinal chloride secretion such as cystic fibrosis.

It has been previously determined that ClC-2, a member of the ClC chloride channel superfamily, is expressed in certain epithelial tissues. These findings fueled speculation that ClC-2 can compensate for impaired chloride transport in epithelial tissues affected by cystic fibrosis and lacking the cystic fibrosis transmembrane conductance regulator. However, direct evidence linking ClC-2 channel expression to epithelial chloride secretion was lacking. In the present studies, we show that ClC-2 transcripts and protein are present endogenously in the Caco-2 cell line, a cell line that models the human small intestine. Using an antisense strategy we show that ClC-2 contributes to native chloride currents in Caco-2 cells measured by patch clamp electrophysiology. Antisense ClC-2-transfected monolayers of Caco-2 cells exhibited less chloride secretion (monitored as iodide efflux) than did mock transfected monolayers, providing the first direct molecular evidence that ClC-2 can contribute to chloride secretion by the human intestinal epithelium. Further, examination of ClC-2 localization by confocal microscopy revealed that ClC-2 contributes to secretion from a unique location in this epithelium, from the apical aspect of the tight junction complex. Hence, these studies provide the necessary rationale for considering ClC-2 as a possible therapeutic target for diseases affecting intestinal chloride secretion such as cystic fibrosis.
The physiological significance of ClC-2, a ubitquitously expressed member of the ClC family of chloride channels (1) is not fully understood. Based primarily on studies of native ClC-2 message and protein expression, roles for ClC-2 in neuronal and epithelial tissue have been proposed (2,3). ClC-2 channel activity has been implicated in the regulation of neuronal responses to GABA-A receptor interaction (2). In non-neuronal cells, ClC-2 function has been linked to volume regulation, and in epithelial cells, it has been linked specifically to chloride secretion (3).
Immunolocalization studies of ClC-2 revealed that it is situated on the apical surface of airway epithelial cells in neonatal rat airways (4). Subsequent Ussing chamber studies showed that luminal acidity promoted chloride secretion in neonatal airway cells via a cadmium-sensitive channel (5). These findings prompted us to suggest that ClC-2, a channel that exhibits these properties, may mediate chloride secretion in neonatal rat airways. In addition, Schwiebert et al. (6) reported that ClC-2-like currents are present in airway epithelial cells derived from an adult patient with cystic fibrosis (CF) 1 and suggested chloride transport via ClC-2 may be able to compensate for defective or absent CFTR chloride channels in the CF airway epithelium. Further, because ClC-2 message can be detected in intestinal epithelial tissue obtained from cftr-knockout mice, Joo et al. (7) also suggested that there is the potential for ClC-2 to provide a bypass pathway for chloride transport in CF affected intestines. Despite these intriguing observations, there was no direct molecular evidence to suggest that ClC-2 contributes to native chloride secretion.
In the present work, we assessed the role of ClC-2 in chloride secretion in the Caco-2 cell line, a cell line that models the human small intestinal epithelium (8,9). Using immunofluorescence and confocal microscopy, we confirmed that ClC-2 protein is endogenously expressed in the plasma membrane of Caco-2 cells. Further, we show that it is uniquely situated at the apical aspect of the tight junctions between cells in fully differentiated monolayers of these cells. Using an antisense strategy, we show that endogenously expressed ClC-2 mediates currents across the plasma membrane of single Caco-2 cells and, finally, that ClC-2 can contribute to native anion secretion across Caco-2 cell monolayers.

EXPERIMENTAL PROCEDURES
Caco-2 Cell Culture-Caco-2 cells were obtained from the American Type Culture Collection (Manassas, VA). They were grown in Earl's ␣-minimum essential medium (Wisent Inc.) containing 10% fetal calf serum, with 2 mM glutamine, 100 units penicillin G, and 100 g/ml streptomycin sulfate at 37°C in an atmosphere of 5% CO 2 , 95% air. For patch clamp studies, cells were used 1-2 days after plating onto 35-mm coverslips (Fisher). For Ussing chamber studies of chloride currents and assessment of ClC-2 localization by confocal microscopy, Caco-2 cells were seeded at high density (1 ϫ 10 6 cells ml Ϫ1 , 500 l/filter) and grown to confluency on clear Snapwell (Costar) filters (pore size, 4 m; diameter, 12 mm). Filters were cultured at 37°C in an atmosphere of 5% CO 2 , 95% air, for 2 weeks with medium replacement every 2-3 days. The formation of an intact monolayer was assessed by measuring transepithelial resistance in Ussing chambers. Transepithelial resistance was calculated using Ohm's law, from measurements of the change in short circuit current measured (I sc , A) upon passing 1 mV across the epithelium. Monolayers were considered acceptable when the transepithelial resistance exceeded 500 Ohms/cm 2 and the transepithelial potential difference exceeded 2 mV. Alternatively, for iodide efflux studies and assessment of protein expression by Western blotting and confocal microscopy following transfection, Caco-2 cells were grown for 4 -6 days on glass coverslips to achieve a differentiated phenotype, as documented by Sood et al. (9).
Northern Analysis-Total RNA was isolated from Caco-2 cell monolayers using the Trizol method as recommended by the supplier (Life Technologies, Inc.). RNA (2 g) was analyzed on agarose gels (1%) containing 0.6 M formaldehyde and transferred to Hybond-N membranes (Amersham Pharmacia Biotech). Blots were cross-linked with UV radiation and hybridized with mouse-specific ClC-2 cDNA fragments radiolabeled by random priming (10). Final conditions of washing included 0.2ϫ SSC (sodium chloride/sodium citrate) with 0.1% SDS at 60°C. The blots were exposed to X-Omat film (Kodak) for 24 -72 h at Ϫ70°C with one intensifying screen.
Western Analysis-Caco-2 cells, grown on 60-mm plastic dishes, were washed with phosphate-buffered saline containing 10 mM mono-dibasic mix, pH 6.8, and 150 mM NaCl, final pH 7.2, and incubated with 900 l of lysis buffer containing 1% Triton X-100, 120 mM NaCl, 10 mM Tris, 25 mM KCl, 25 mM MgCl 2 , 1.8 mM CaCl 2 , and protease inhibitors leupeptin (10 g/ml) and apotinin (10 g/ml), 1 mM benzamidine, 10 M E64, and 2 mM dithiothreitol. The cells were scraped off, and the suspension was mixed by vortexing. The supernatant was then centrifuged for 10 min at 4°C at 48,000 ϫ g to isolate a crude membrane preparation. Following protein assay of the supernatant, 50 g of this preparation was analyzed by SDS-polyacrylamide gel electrophoresis (8% gel) using anti-ClC-2 antibody at a concentration of 2 g/ml. This polyclonal antibody, as described previously, was generated against a GST fusion peptide containing amino acids 31-74 of rat ClC-2 (rClC-2 cDNA kindly provided by T. Jentsch). The ClC-2-specific antibody was immunopurified from a matrix of GST-N-peptide coupled on an activated agarose column as previously described (11,12). The monoclonal antibody against ␤-actin (Sigma, anti-␤-actin clone AC-74) was used at 1/1000 dilution. Immunoreactive protein was detected using the ECL system (Amersham Pharmacia Biotech).
Immunofluorescence-Immunofluorescence labeling was performed on Caco-2 cells grown on 35-mm circular coverslips or on clear 35-mm, 0.4-m pore Snapwell (Corning Costar) filters. The pattern of ClC-2 labeling was identical regardless of the support employed. Cells were fixed with paraformaldehyde AM (4% in phosphate-buffered saline) and permeabilized with 0.5% Triton X-100 in phosphate-buffered saline. Cells were incubated for 0.5 h at 25°C in 5% normal goat serum, 0.05% Triton X-100 in Tris-buffered saline containing 10 mM Tris-Cl and 150 mM NaCl with final pH 8, and then for 2.5 h with the polyclonal antibody against ClC-2 (30.7 g/ml) or overnight in the refrigerator with the polyclonal antibody against ClC-3 (30 g/ml) (Alomone Labs Ltd., Jerusalem, Israel). Then the cells were washed and incubated with Texas Red-conjugated or fluorescein isothiocyanate-conjugated anti-rabbit secondary antibody (0.02 mg/ml; Molecular Probes) and washed again before mounting. For colocalization studies of ClC-2 and the tight junction protein occludin, the above procedure was followed by additional incubation with the monoclonal anti-occludin antibody (0.002 mg/ml, Zymed Laboratories Inc., Missisauga, Canada) for 1 h and washed. Cells were then incubated with Texas Red-conjugated anti-mouse secondary antibody (0.02 mg/ml; Molecular Probes) and washed before mounting. For the competition studies, the anti-ClC-2 antibody was preincubated with 2-fold excess of the antigenic fusion peptide overnight at 4°C before incubation. Slides were viewed on an Olympus Vanox AHBT3 microscope using epifluorescence, and images were captured using the Image Pro Plus program (Cybernetics, L.P.). For confocal microscopy, sections (each 0.7 m in thickness) were viewed with a 100x objective a Leica TCS 4D microsope, and the images were captured using the SCANware 5.01 program.
Patch Clamp Studies of Caco-2 Cells-Caco-2 cell membrane currents were measured using conventional whole cell patch clamp technique (13). Patch clamp electrodes were prepared from borosilicate glass capillaries (outer diameter, 1.5 mm; inner diameter, 1.18 mm) with an inner filament (World Precision Instruments, Inc., Sarasota, FL) on a Narishige PP-83 patch electrode puller using the standard two-pull technique. The tip resistance was 3-5 M ⍀ when filled with pipette solution (see below for composition). Whole cell currents were measured using an Axopatch 200A patch clamp amplifier (Axon Instruments, Foster City, CA) and were filtered at 100 Hz with a 6-pore Bessel Filter. Sampling rate was 4 kHz for most data, and junction potentials were corrected. Voltage clamp protocols were generated using pCLAMP software (version 7, Axon Instruments) via a Pentium II computer interfaced with a 1200 series Digidata (Axon Instruments) The same software package was used both for data acquisition and analysis. Current-voltage relationships were determined in a stepwise clamp protocol. From a holding potential of Ϫ30 mV, voltage pulses of 3.0 s were applied from Ϫ160 to ϩ40 mV in 20-mV increments. The bath solution contained 140 mM N-methyl-D-glutamine (NMDG) chloride, 2 mM MgCl 2 , 2 mM CaCl 2 , 5 mM HEPES, whereas the pipette solution contained 140 mM NMDG chloride, 2 mM MgCl 2 , 2 mM EGTA, and 5 mM HEPES. Both pipette and bath solutions were adjusted to pH 7.4 and 260 mOsm. In experiments in which the response to hypotonic shock was studied, the bath solution contained 110 mM NMDG chloride, 2 mM CaCl 2 , 2 mM MgCl 2 , 5 mM HEPES, pH 7.4, and the osmolarity was adjusted to 303 mOsm with sucrose. The hypotonic bath solution was made as above, maintaining equal ionic strength and pH, except that the osmolarity was adjusted to 228 mOsm with sucrose as assessed using a 5500 Vapor pressure osmometer (Wescor, Johns Scientific Inc.). Caco-2 cells were subjected to hypotonic shock using a gravity-fed superfusion system.
ClC Constructs-The ClC-2 sense construct was made by directional cloning of the rat (r)ClC-2 open reading frame (kindly provided by T. Jentsch, Hamburg, Germany) with BamHI (5Ј) and EcoRI (3Ј) linkers into the BamHI and EcoRI restriction sites of the eukaryotic vector pCDNA 3.1 (ϩ) (Promega, Madison, WI). The rat ClC-2 sequence shares 77% identity with the human sequence at the nucleotide level. The antisense ClC-2 construct was made by cloning the ClC-2 open reading frame into pCDNA 3.1(Ϫ) vector such that the reversed restriction sites on this vector would reverse the orientation of the open reading frame to create the antisense plasmid. The antisense murine ClC-4 construct (Clcn4, a gift from E. Rugarli, Milano, Italy) was made by cloning the ClC-4 open reading frame with BamHI (5Ј) and EcoRI (3Ј) into pCDNA 3.1(Ϫ). The murine ClC-4 construct shares 73% sequence identity with the human sequence.
Intranuclear Injection of Plasmid-Caco-2 cells were microinjected with plasmids at day 1 after plating on glass coverslips for patch clamp experiments. In this procedure, the Eppendorf microinjector 5246 system, the micromanipulator 5171 system, and a Nikon Diaphot inverted microscope were used. Nuclear microinjection was performed with the Z (depth) limit option, using 0.3-s injection duration and 40 -60 hPa injection pressure. Injection femtotips were pulled from borosilicate glass capillaries with an internal diameter of 0.5 Ϯ 0.2 m. Plasmids were diluted to a final concentration of 50 g/ml for sense ClC-2 plasmids, 50 and 300 g/ml for antisense ClC-2, and 300 g/ml antisense ClC-4. The injection buffer contained in 50 mM HEPES, 50 mM NaOH, 40 mM NaCl, pH 7.4. Fluorescein isothiocyanate-labeled dextran (0.5%, Sigma) was also added to the injection medium to identify successfully microinjected cells.
Transient Transfection of Caco-2 Cell Monolayers-For transfection of Caco-2 cells with antisense ClC-2 DNA, the Lipofectin transfection protocol was followed (Life Technologies, Inc.). Briefly, ϳ2 ϫ 10 5 cells were seeded on 35-mm tissue culture plates in culture medium supplemented with serum. Cells were then incubated at 37°C in a 5% CO 2 incubator overnight to allow them to reach 80% confluency (ϳ10 6 cells/ plate). Two solutions, one containing 2 g of antisense ClC-2 cDNA (dissolved in serum-free medium) in 100 l of OPTI-MEM I reduced serum medium and the second containing 20 l of Lipofectin reagent in 100 l of OPTI-MEM I reduced serum medium were mixed and incubated at room temperature for 45 min. Following addition of 800 l of OPTI-MEM I reduced serum medium, this transfection mixture was applied to the cells (after washing them with serum-free medium). The cells were then incubated at 37°C in a 5% CO 2 incubator for 48 h, after which the transfection medium was replaced with normal medium containing 10% serum and antibiotics (as described above). 24 h later, cells were harvested for immunoblot analysis or studied by iodide efflux assay.
Ussing Chamber Analysis-Short circuit current measurements were performed on Caco-2 cells grown to confluency on Snapwell clear filters, with a surface area of 1.13 cm 2 (Corning Costar). The average transepithelial resistance of the cells used was 2421 Ϯ 357.2 ⍀ cm 2 . Filters were inserted into Ussing chambers and bathed in a buffer composed of 110.4 mM NaCl, 27.5 mM mannitol, 2.4 mM K 2 HPO 4 , 0.8 mM KH 2 PO 4 , 10 mM glucose, 10 mM HEPES, and 1 mM CaCl 2 , gassed with 95% O 2 and heated to 37°C. After 5-8 min of measuring basal current the apical solution was changed to one lacking (27.5 mM) mannitol for 20% hypotonic shock (HTS; 80% isotonicity; determined using a 5500 Vapor pressure osmometer, Wescor, Johns Scientific Inc.).
Iodide Efflux-Caco-2 cells grown on coverslips (80%) were transfected either with antisense ClC-2 plasmid in the pCDNA vector or with vector alone as described under "Experimental Procedures." Only monolayers possessing 1 ϫ 10 6 cells after the entire transfection protocol were used for subsequent assays. The transfected cells were iodide loaded according to established methods (14, 15) using a 1-ml Ringers Nitrate loading buffer at pH 7.4 containing 136 mM NaI, 4 mM KNO 3 , 2 mM CaNO 3⅐4H2 O, 2 mM MgNO 3⅐6H2 O, 11 mM glucose, and 20 mM HEPES. The cells were incubated for 1 h at 37°C in the presence of 5% CO 2 in the above buffer. After this incubation period, the coverslips covered with iodide-loaded cells were washed for a total of 15 s in three separate baths containing Ringers nitrate efflux buffer (110 mM NaNO 3 , 4 mM KNO 3 , 2 mM CaNO 3 ⅐4H 2 O, 2 mM MgNO 3 ⅐6H 2 O, 11 mM glucose, and 20 mM HEPES, pH 7.4) with the osmolarity adjusted to 300 mOsm with added sucrose. After this washing period, efflux of cellular iodide was assessed continuously after placing the coverslip into an isotonic solution (the above Ringers Nitrate efflux buffer) or hypotonic solutions (osmolarity adjusted 228 mOsm). Iodide efflux (measured as a change in mV) was assessed over a 5-min period, using an iodide sensing electrode (Fisher). Changes in voltage were acquired using the FETCHEX data acquisition program (pCLAMP 6.04, Axon Inst.) and data analyzed using FETCHAN software.
Statistics-Patch clamp measurements are presented as the means Ϯ S.E. Most statistical analyses were performed using the Student's unpaired test. Results obtained in Ussing chamber studies and in Patch clamp studies with hypotonic shock were analyzed using the Student paired test. Differences between two groups were considered significant with p values Ͻ0.05.

ClC-2 Message and Protein Are Expressed in Caco-2 Cells-
ClC-2 mRNA in Caco-2 cells was detected by Northern blot analysis as a 4.6-kb transcript (Fig. 1A). This size transcript plus a smaller transcript of ϳ3.3 kb in size has been detected in several other tissues and cell lines, as well, including the colonic epithelial cell line T 84 (6). Immunoblots (Fig. 1B) using a polyclonal antibody directed against ClC-2 (12) showed that ClC-2, migrating as a 90 -97-kDa protein, is expressed in Caco-2 cells. This signal was competed using the GST-ClC-2 fusion peptide against which the antibody was raised, not GST alone, indicating its specificity for ClC-2.
Immunofluorescence labeling using the above antibody suggests that ClC-2 protein localizes to plasma membrane and/or submembranous vesicles in Caco-2 cells (Fig. 2). Further, this signal was specific because it was competed using the antigenic peptide described above.
ClC-2 Is Functionally Expressed on the Plasma Membrane of Caco-2 Cells-Whole cell patch clamp studies were performed to determine whether ClC-2 is functional in the membrane of Caco-2 cells. Because previous studies in heterologous systems showed that ClC-2 expression conferred chloride currents were activated by hyperpolarization and by hypotonic shock (1, 11, 16), we assessed whether these manipulations could activate ClC-2 endogenously expressed in Caco-2 cell membranes by patch clamp electrophysiology. We functionally isolated aniondependent currents by using intracellular (pipette) and extracellular (bath) solutions which contained NMDG chloride as the predominant salt. We chose a voltage step protocol that has been used in previously published studies of ClC-2 (11,16). Briefly, from a holding potential of Ϫ30 mV, the membrane potential was stepped by 20-mV increments from Ϫ160 to ϩ 40 mV. Because ClC-2 currents have not been reported to be ATP-dependent and to minimize the contribution by the ATPdependent, swelling-activated outwardly rectifying chloride channel, volume-sensitive organic osmolyte anion channel (17), MgATP was not included the patch pipette solutions. As shown in Fig. 3 (A and C), currents typical of those previously associated with ClC-2 expression, i.e. showing activation with hyperpolarizing voltage steps and an inwardly rectifying currentvoltage relationship, were detected in Caco-2 cells. At the hyperpolarized membrane potential of Ϫ160 mV, these currents had a magnitude of Ϫ37 pA/pF Ϯ 0.96 (n ϭ 8). These hyperpolarization-activated currents reversed close to the estimated equilibrium potential of chloride (E Cl ϭ 0 in symmetrical NMDG chloride solutions (Fig. 3C, circles).
As shown in Fig. 3 (B and C), whole cell chloride currents in Caco-2 cell were stimulated by 25% hypotonic shock. We detected an increase in membrane currents from Ϫ38 pA/pF to Ϫ84 pA/pF within 3-5 min after bath dilution at Ϫ160 mV (Fig.  3). The mean current-voltage (I/V) curves of ClC-2 currents before and after application of hypotonic shock reversed close to the chloride equilibrium potential (ϩ5.6 Ϯ 0.47 mV) as shown in Fig. 3C. The average current at Ϫ160 mV following hypotonic shock (Ϫ65 Ϯ 8 pA/pF) was elevated when compared with currents measured in isotonic condition (Ϫ32 Ϯ 2 pA/pF) (p ϭ 0.0208). The I/V relationship of the HTS-stimulated chloride currents was less inwardly rectifying than that observed in isotonic solutions (Fig. 3C). A similar change in the I/V relationship was observed with HTS in chloride currents specifically conferred by ClC-2 expression in Xenopus oocytes and Sf9 cells and has been attributed to an alteration in the inactivation gate of ClC-2 (11,16). These results indicate that native ClC-2 expression at the cell surface of Caco-2 cells is associated with appearance of chloride currents with activation and conductance properties similar to those conferred by ClC-2 expression in heterologous expression systems (1,11,16,18).
We used an antisense strategy to confirm that the above currents were mediated by ClC-2 because the pharmacological approach lacks specificity. First, we confirmed that transient transfection of ClC-2 antisense cDNA (see "Experimental Procedures") successfully reduced ClC-2 protein expression by Western analysis of cell lysates from ClC-2 antisense-transfected Caco-2 cells. Using the NIH Imaging Program, we found that there is a 70% decrease in ClC-2 protein quantity in antisense ClC-2 transfected Caco-2 cells relative to control (vector-alone transfected cells). We verified by assessing ␤-ac-tin expression that differences in protein loading could not account for the decrease in ClC-2 expression in the antisense transfected cells (Fig. 4A). Furthermore, we examined the effects of antisense ClC-2 transfection on immunolabeled ClC-2 detected by fluorescence confocal microscopy. DNA coding for green fluorescence protein (GFP) was cotransfected with antisense ClC-2 (or empty vector as a control) into Caco-2 cells to identify transfected cells (Fig. 4B). We used an imaging program (Scion Corp.) to compare the ClC-2 immunofluorescence intensity in antisense ClC-2 and in vector transfected Caco-2 cells. We found that the fluorescence intensity of the signal (red) corresponding to membrane expression of ClC-2 was reduced by ϳ75% in antisense ClC-2 transfected Caco-2 cells (24.9 units Ϯ 4.3, n ϭ 13, p Ͻ 0.0001) relative to the intensity of the ClC-2 signal in mock transfected cells (105.8 units Ϯ 3, n ϭ 10). Immunofluorescence corresponding to expression of ClC-3, a related family member, was not affected by antisense ClC-2 transfection (Fig. 4C). The signal detected using this ClC-3 antibody in immunofluorescence studies can be competed using the antigenic peptide used to raise the antibody, confirming its specificity (19). Fig. 4C shows that the ClC-3 immunofluorescence (red) in antisense ClC-2 and GFP cotransfected Caco-2 cells (106.1 units Ϯ 3.2, n ϭ 20) was similar to that in vector and GFP cotransfected cells (105.9 units Ϯ 2.2, n ϭ 17, p ϭ 0.97). Interestingly, our studies show that unlike ClC-2, immunoreactive ClC-3 appears to be primarily expressed in intracellular membranes, although there is signal detected at the cell surface in a subpopulation of cells.
For patch clamp studies, we manipulated ClC-2 expression using intranuclear plasmid injection technique (20,21), because this method permits control of plasmid copy number and hence has greater precision in manipulating the level of antisense expression. Fluorescein isothiocyanate-dextran was coinjected with the plasmid to permit identification of manipulated cells. We found that microinjection of antisense ClC-2 cDNA into Caco-2 cells decreased the ClC-2-like currents in a dosedependent manner (Fig. 5, A and B). The negative whole cell current measured at Ϫ160 mV decreased from Ϫ37 pA/pF Ϯ 1 (n ϭ 8) in uninjected cells to Ϫ26 Ϯ 2 pA/pF (n ϭ 5, p ϭ 0.0003) and Ϫ12 Ϯ 1 pA/pF (n ϭ 10, p Ͻ 0.0001) in 50 and 300 g/ml antisense ClC-2 cDNA injected cells, respectively (Fig. 5, A and  B). To allow direct comparison of current amplitude in ClC-2 antisense injected and uninjected Caco-2 cells, we normalized these currents to the currents at Ϫ160 mV in uninjected cells (Fig. 5B). As shown in Fig. 5B, the normalized ClC-2 currents in cells injected with 50 or 300 g/ml antisense plasmid were decreased by 29 Ϯ 6 and 68 Ϯ 2%, respectively.
To determine the contribution of ClC-2 to HTS-stimulated chloride currents in Caco-2 cells, we studied the effects of antisense ClC-2 transfection on the development of hypotonicity-activated chloride currents in Caco-2 cells. As shown in Fig.  5E, the amplitude of the HTS-stimulated chloride currents FIG. 3. ClC-2-mediated chloride currents are activated by hypotonic shock in Caco-2 cells. ClC-2 mediated chloride currents recorded in Caco-2 cells before (A) and after (B) exposure to 25% hypotonic shock. The same voltage step protocol was applied as described previously. C, mean I/V curves of ClC-2 current under isotonic (circle, n ϭ 4) and hypotonic (square, n ϭ 4) conditions. measured in noninjected Caco-2 cells at -160 mV (-65 Ϯ 8 pA/pF, n ϭ 4) was almost four times larger than the currents measured in antisense ClC-2 injected cells (Ϫ15.7 Ϯ 2 pA/pF, n ϭ 7, p Ͻ 0.0001). The current density at ϩ40 mV after HTS was 12.7 Ϯ 2.4 pA/pF in control cells and almost two to three times less in antisense-transfected cells (4.9 Ϯ 0.8 pA/pF, p ϭ 0.0053), suggesting that ClC-2 mediates most of the hypotonicity-activated current at both potentials. Fig. 5F shows the mean current-voltage curves of chloride currents obtained in antisense ClC-2 injected Caco-2 cells before and after application of hypotonic shock. The average current density at Ϫ160 mV following hypotonic shock (Ϫ15.7 Ϯ 2 pA/pF, n ϭ 7) was not significantly different when compared with current densities measured in isotonic conditions (Ϫ13.7 Ϯ 0.9 pA/pF, n ϭ 7, p ϭ 0.26). Hence, antisense ClC-2 transfection abolished hypotonicity activated chloride currents at Ϫ160 mV. At ϩ40 mV, current density determined in antisense-transfected cells was less than half that measured in control cells, confirming that ClC-2 is a major determinant of hypotonicity-activated chloride currents evoked in these particular experimental conditions, i.e. without MgATP added to the pipette solutions. However, antisense ClC-2 did not abolish the hypotonicity-evoked chloride currents at ϩ40 mV, suggesting that channels other than ClC-2 may also contribute to this response at depolarized membrane potentials.

In a Polarized Caco-2 Cell Monolayer, ClC-2 Localizes
Close to the Tight Junction Complex-To determine whether ClC-2 protein exhibits a polarized distribution in fully differentiated intestinal cells, we examined its subcellular localization in confluent Caco-2 cells grown on semipermeable filters. It has been well documented that Caco-2 cells grown on either filters or on glass exhibit a polarized phenotype 4 -6 days after plating (8,9). In such a confluent monolayer, we found that ClC-2 protein exhibited a novel localization pattern. Fig. 6 shows optical sections obtained using a confocal microscopy of confluent Caco-2 cells co-labeled with the polyclonal
anti-ClC-2 antibody and a monoclonal antibody against the transmembrane tight junction protein, occludin (22). The first row of images shows the most apical sections, and the subsequent rows show consecutive images obtained toward the basolateral pole. The ClC-2 signal (red) clearly delineates the plasma membrane in the first row (Fig. 6), whereas the occludin signal (green) is quite faint, suggesting that ClC-2 protein expression is apical relative to this tight junction protein. The yellow signal, indicative of regions of overlap, is weak in this apical section. In the next row of images, the membrane staining corresponding to both proteins and the yellow signal is intense, suggesting that expression of these proteins may overlap in this optical section. The last two rows of sections show the disappearance of the ClC-2 signal from the plasma membrane while the occludin signal remains strong (Fig. 6). Thus, it appears that ClC-2 localization overlaps with the apical aspect of the tight junctions. Identical localization of ClC-2 was observed for confluent monolayers of Caco-2 cells grown on glass coverslips (data not shown).
ClC-2 Contributes to Hypotonicity-stimulated Chloride Secretion in Monolayers of Caco-2 Cells-There may be multiple functional consequences of this unique localization of ClC-2 because the tight junction is known to be critical for regulating solute (including ions) and water flux through the paracellular pathway and as well maintaining the polarization of membrane proteins and lipids (22,23). We reasoned that ClC-2 may contribute to chloride flux in a secretory direction across the apical membrane because its localization not only overlaps with the junction transmembrane protein, occludin, but it also extends into the membrane at the apical aspect of the tight junction. Typically, chloride channels implicated in secretion have been localized to the apical membrane (24). Hence, we investigated the role of ClC-2 in chloride secretion across monolayers of Caco-2 cells.
The application of hypotonicity to the mucosal or apical surface of intestinal epithelia has been previously shown to stimulate chloride secretion (25,26). However, the molecular basis for this secretory response remained unclear. To determine whether Caco-2 cell monolayers model this response, we measured short circuit current responses to hypotonic shock by Caco-2 cell monolayers mounted in Ussing chambers. Hypotonicity-activated increases in luminally (or apically) directed negative short circuit current (I sc ) across epithelia have been shown to correlate with chloride secretion in previous studies Ussing chamber studies (25). Similarily, we found that reducing the osmolarity of the bath facing the apical membrane of confluent Caco-2 cells (by 20%) evoked a transient increase in luminally directed, negative transepithelial short circuit current (I sc ) (Fig. 7A). I sc increased significantly from Ϫ1.0 Ϯ 0.1 A/cm 2 to Ϫ3.1 Ϯ 0.4 A/cm 2 (n ϭ 14, p Ͻ 0.0001, Student's test for paired data) with this treatment. These findings are consistent with the transient activation of an apical chloride conductance path by luminal hypotonicity in Caco-2 cells.
To determine directly whether ClC-2 contributes to this response, we assessed the effect of antisense ClC-2 transfection. The above Ussing chamber studies required that Caco-2 cell monolayers were grown on semipermeable filter supports. We found that transfection of Caco-2 cells on such filters was not efficient compared with transfection of monolayers on coverslips (see Western analysis on Fig. 4). Hence, we utilized an iodide efflux method to monitor chloride secretion from cells on glass coverslips. This method has been used extensively in the study of chloride secretion through CFTR in epithelial monolayers (14,15). In Fig. 7B, we show traces of the time-dependent efflux of iodide from iodide loaded Caco-2 cell monolayers. Efflux from monolayers (each comprised of 1 ϫ 10 6 cells) into control, iodide-free isotonic solutions reflects efflux through constitutively open anion channels. The rate of iodide efflux is enhanced if monolayers are transferred to hypotonic (iodidefree) solutions (25% isotonicity) as shown in the traces in Fig.  7B, presumably because of the activation of apically localized chloride channels. The extent of iodide efflux measured during the first minute has been plotted in the bar graph shown in Fig.  7B (right panel). We show that in mock (vector-alone) transfected Caco-2 cell monolayers, the rate of iodide efflux is much greater in monolayers exposed to hypotonic solutions relative to the efflux rate from monolayers exposed to isotonic solutions (p ϭ 0.002). In antisense ClC-2 transfected monolayers, the rate of iodide efflux in hypotonic solutions was significantly reduced in comparison to the hypotonicity-evoked efflux measured in monolayers transfected with vector alone (p ϭ 0.001). This result suggests that ClC-2 contributes to hypotonicityactivated anion secretion. On the other hand, overexpression of ClC-2 does not affect the iodide efflux response in hypotonic FIG. 6. Localization of ClC-2 protein to the apical aspect of the tight junctions formed between Caco-2 cells in a confluent monolayer. Consecutive optical cross-sections of confluent Caco-2 cells that have been co-labeled with the polyclonal anti-ClC-2 antibody (red) and a monoclonal anti-occludin antibody (green) were obtained by confocal microscopy. Panels 1-5 extends from the apical toward the basolateral membrane. Panel 2 is the most apical section in which the ClC-2 signal could be detected, and panel 5 is the last section in which membrane staining of ClC-2 could be detected. Pictures were taken using a confocal microscope with a 100ϫ objective. solutions (p ϭ 0.33) but does appear to increase the basal efflux (p ϭ 0.009). Hence, overexpression of ClC-2 confers an increase in basal anion secretion. Hypotonic solutions do not cause a further increase in iodide efflux in cells overexpressing ClC-2, possibly because the high basal permeation rate quickly dissipates the driving force for further iodide flux. DISCUSSION The major aim of this study was to test the hypothesis that ClC-2 contributes to native chloride secretion by intestinal epithelia. In support of this hypothesis, we provided evidence that ClC-2 is endogenously expressed at the plasma membrane of Caco-2 cells where it contributes to native currents. Immunofluorescence studies were used to document cell surface localization and patch clamp studies established that ClC-2-like currents can be detected. We used an antisense strategy to confirm the identity of these native currents as ClC-2. Finally, reduction of ClC-2 protein expression in Caco-2 cell monolayers by ClC-2 antisense transfection resulted in a decrease in hypotonicity evoked iodide efflux, a measure of chloride secretion, indicating that ClC-2 can contribute to chloride secretion.
Although ClC-2-like currents have been previously described in epithelial cells (27)(28)(29), the current work provides the first direct molecular evidence to support a role for ClC-2 in chloride secretion by differentiated epithelial cells. Caco-2 cells have been found to model the functional properties of the human small intestine; hence, our findings suggest that ClC-2 may contribute to chloride secretion by the epithelium of this organ.
Similarily, ClC-2 protein has been localized to the apical membrane of rat neonatal respiratory epithelium (4,5) and may mediate secretion in this tissue. However, ClC-2 is not likely contribute to chloride secretion in all transport epithelia. Preliminary studies in our laboratory have localized ClC-2 to the basolateral membrane of the mouse colon 2 ; hence, ClC-2 protein is likely to mediate chloride reabsorption by this tissue.
The mechanisms for activation of ClC-2 channel function in situ are currently unclear. In neurons, ClC-2 is basally active at resting membrane potentials (2). The results of the present experiments suggest that in epithelial cells, ClC-2 channels are partially active at resting membrane potentials. Patch clamp studies of Caco-2 cells revealed that chloride currents associated with ClC-2 expression, i.e. those endogenous currents inhibited by ClC-2 antisense and augmented by expression of exogenous ClC-2 expression, are activated by membrane hyperpolarization to potentials more negative than Ϫ60 mV. This membrane potential is close to the resting membrane potential cited for gastrointestinal epithelial cells, i.e. from Ϫ50 to Ϫ60 mV (26). Gastrointestinal epithelial cells can, however, reach more hyperpolarized potentials when stimulated by the hormones that act to increase potassium permeability, i.e. acetylcholine and vasoactive intestinal peptide (26,30). The acetylcholine analogue, carbachol, has been reported to cause transient hyperpolarizations of 10 -25 mV in isolated small intestinal crypts. Hence, during stimulation with the above hormones, ClC-2 channels may become further activated. In the present studies we diluted the external solutions to 70 -80% isotonicity to observe stimulation of ClC-2-mediated currents in single cells and iodide efflux from Caco-2 cell monolayers. It remains to be determined whether this experimental maneuver reflects a physiologically relevant stimulus to intestinal epithelial cells such as the generation of osmotic gradients during the concentrative uptake of nutrients (31).
Several previous studies have suggested that luminal hypotonicity induces transepithelial chloride secretion. A study on Necturus enterocytes has shown that a swelling-activated chloride conductance is present in the apical membrane of these cells (32). In addition, cell swelling may stimulate transepithelial chloride secretion in airway epithelium (33), the T 84 colonic epithelial cell line (34,35), HT-29Cl.19A intestinal epithelial cells (36), and rat ileum (25). The molecular identity of the chloride channels contributing to this hypotonicity-activated secretory chloride conductance in these diverse epithelial tissues has not been determined. Our current studies showing that antisense ClC-2 transfection reduced this function suggests that ClC-2 should be considered as a candidate. However, we cannot rule out the possibility that antisense ClC-2 transfection may have caused primary and/or secondary changes in the expression of other proteins over the 48-h transfection time period. It is conceivable that expression of undefined paralogs of human ClC-2 may have been reduced directly by the antisense rodent ClC-2 construct employed in the current studies. Further, although the expression of another channel implicated that in hypotonicity-activated ion conductance ClC-3 (37, 38) did not change, it remains possible that the expression of other unidentified hypotonicity-activated membrane proteins may have altered as a secondary response to antisense ClC-2 transfection.
Our confocal studies of ClC-2 protein localization in confluent Caco-2 cells indicated that upon cell differentiation, ClC-2 protein is predominantly situated at the apical aspect of the tight junctions. This distinctive localization for ClC-2 has also been detected in murine intestinal tissue, and immunogold 2 K. Gyömörey and C. Bear, unpublished data.

FIG. 7.
Application of hypotonic shock to apical membranes of confluent Caco-2 cell monolayers stimulates chloride secretion through ClC-2 channels. A, left panel, Caco-2 cell monolayers grown on semipermeable support were mounted in Ussing chamber to assess short circuit current response to dilution of mucosal solution. A transient increase in negative I sc was activated by exchange of the isotonic mucosal bath with 20% hypotonic mucosal bath as indicated by the bar. Right panel, the mean negative basal transepithelial I sc was increased by hypotonic shock (open bars, n ϭ 14) (p Ͻ 0.05, Student's paired test). B, left panel, iodide efflux from Caco-2 monolayers transfected with pCDNA 3.1(ϩ) vector alone (Vtr) or antisense ClC-2 construct was monitored continuously using an iodide sensing electrode. Monolayers were exposed either to isotonic (Ctl) or 25% hypotonic solutions. Right panel, this bar graph shows the means Ϯ S.D. iodide efflux measurements (iodide efflux during the first minute) from vector-alone transfected and antisense ClC-2 transfected Caco-2 cell monolayers as well as monolayers transfected with ClC-2 in the sense orientation. The solid bars represent data from isotonic solutions, and the open bars represent data from hypotonic solutions. A minimum of four monolayers were studied for each condition. studies showed that ClC-2 resides primarily in the membrane at this site (12). This distribution pattern is not typical for ion channels that have been implicated in chloride secretion. For example CFTR, the chloride channel thought to mediate chloride transport in many epithelial tissues, resides on the brush border membrane (24). Future studies are required to determine the molecular mechanisms that traffic ClC-2 to the apical aspect of the tight junction and those mechanisms that may act to retain the channel at this site. Finally, we have yet to determine whether there is a particular physiological significance for the concentration of an anion channel at the apical aspect of the tight junction. It is well known that the tight junction functions as a size and charge selective gate restricting the paracellular transit of organic and inorganic solutes (22,23). Although our studies suggest that ClC-2, at the apical aspect of this junction, contributes to chloride secretion, its unique localization may also function to regulate the gate functions of the tight junction.
An understanding of the molecular basis for chloride secretion by epithelial tissues is key to identification of future therapies for intestinal secretory diseases such as diarrheal diseases and cystic fibrosis. Our studies indicate that ClC-2 contributes to the native secretory capacity of intestinal tissue. Hence, modification of its function or location may affect the severity of secretory diseases.