Expression of human pICln and ClC-6 in Xenopus oocytes induces an identical endogenous chloride conductance.

pICln is a protein that induces an outwardly rectifying, nucleotide-sensitive chloride current (ICln) when expressed in Xenopus oocytes, but its precise function (plasma-membrane anion channel versus cytosolic regulator of a channel) remains controversial. We now report that a chloride current identical to ICln is induced when Xenopus oocytes are injected with human ClC-6 RNA. Indeed, both the pICln and the ClC-6 induced current are outwardly rectifying, they inactivate slowly at positive potentials and have an anion permeability sequence NO3− > I− > Br− > Cl− > gluconate. Cyclamate, NPPB, and extracellular cAMP block the induced currents. The success rate of current expression is significantly increased when the injected Xenopus oocytes are incubated at a higher temperature (24 or 37°C) prior to the analysis. In addition, the ICln current was detected in 6.2% of noninjected control Xenopus oocytes. We therefore conclude that the ICln current in Xenopus oocytes corresponds to an endogenous conductance that can be activated by expression of structurally unrelated proteins. Furthermore, functional, biochemical, and morphological observations did not support the notion that pICln resides in the plasma membrane either permanently or transiently after cell swelling. Thus, it is unlikely that pICln forms the channel that is responsible for the ICln current in Xenopus oocytes.

pI Cln is a 26-kDa acidic protein that is ubiquitously expressed not only in various mammalian cell lines and tissues but also in Xenopus oocytes (1)(2)(3). Expression of mammalian pI Cln in Xenopus oocytes induces an outwardly rectifying, nucleotide-sensitive Cl Ϫ current, I Cln , that slowly inactivates at positive potentials (4). Phenotypical similarities between I Cln and the volume-activated Cl Ϫ current, I Cl, swell (for a review of I Cl, swell see Ref. 5), have prompted functional models that describe pI Cln either as the volume-activated Cl Ϫ channel itself or as a regulator thereof (2, 6 -8). However, using Xenopus oocytes, we have recently demonstrated that I Cln and the endogenous I Cl, swell are two different currents that can be discriminated by biological, biophysical, and pharmacological criteria (9).
Irrespective of the relation between pI Cln and I Cl, swell , one major unresolved problem is how expression of mammalian pI Cln in Xenopus oocytes generates a specific Cl Ϫ current. This was originally explained by assuming that pI Cln was a plasma membrane-spanning protein that constituted the anion channel itself (4). However, Krapivinsky et al. (2) showed that the majority of pI Cln resided in the cytosol, and they concluded that pI Cln was a cytosolic regulator of an endogenous, plasma membrane-located anion channel. Recently, it has been suggested that pI Cln , in spite of its cytosolic location, can still function as a plasma membrane-located anion channel if one assumes a bimodal distribution for pI Cln with a subfraction residing in the plasma membrane (6,7). Moreover, the ratio between cytosolic and plasma-membrane pI Cln could vary, depending on specific stimuli such as cell swelling (10).
To further characterize the I Cln chloride current and to address the role of pI Cln , we have expressed human pI Cln in Xenopus oocytes. First of all, we observed that the success rate of I Cln expression in Xenopus oocytes injected with pI Cln RNA markedly depended on the incubation temperature of the oocytes. Furthermore, the I Cln current was also triggered in Xenopus oocytes injected with RNA coding for human ClC-6, a protein that is structurally unrelated to pI Cln . Finally, we also argue against a plasma membrane location for pI Cln .

EXPERIMENTAL PROCEDURES
PCR, 1 reverse transcription-PCR, and Vector Construction-A human cDNA clone for pI Cln (accession number X91788; see Ref. 1) was PCR-mutagenized by replacing the 5Ј-untranslated region with an EcoRI/HindIII/BamHI linker. This allowed subcloning of human pI Cln cDNA as a BamHI fragment in the RNA transcription vector pGEMHE (11), yielding the pGEMHE/EHBhIClnORF vector. The human ClC-6 clone was constructed as follows. First we amplified by reverse transcription-PCR the 5Ј-end of human ClC-6 (nucleotides 202-1229 of the published open reading frame; see Ref. 12). Reverse transcription of 1 g of total RNA of human K562 cells (ATCC CCL 243) was performed as described (13) except that random primers were used instead of oligo(dT) primers. The PCR reaction was carried out with Pfu polymerase (Stratagene, La Jolla, CA) following the manufacturer's instructions. The PCR fragment was then digested with BamHI (in forward primer) and XbaI (internal site at nucleotide 792). The remainder of the ClC-6 open reading frame (nucleotides 793-2610) and part of the 3Јuntranslated region were isolated as a XbaI-HindIII fragment from the HA0519 clone. This is a partial human ClC-6 cDNA clone isolated from a human myeloid cell line by Miyajima and co-workers (EMBL/Gen-Bank TM accession number D28475). The BamHI-XbaI PCR fragment and the XbaI-HindIII fragment were ligated in a pBluescript vector digested with BamHI-HindIII. Nucleotides 1-201 of the open reading frame were then amplified by reverse transcription-PCR from human K562 RNA and inserted upstream of the ClC-6 sequence via a NcoI site. This created a pBluescript vector containing the complete ClC-6 open reading frame. The ClC-6 sequence was then subcloned in pGEMHE as BamHI-HindIII fragment.
Expression of Human pI Cln and ClC-6 in Xenopus Oocytes and Electrophysiology-The pGEMHE/ClC-6 vector was cut with SphI, and the pGEMHE/EHBhIClnORF vector was cut with HindIII. The linearized DNAs were purified (QIAEX desalting and concentration protocol; Qiagen) and in vitro transcribed with T7 RNA polymerase (RiboMAX system, Promega). RNAs were extracted with phenol/chloroform and ethanol-precipitated. Stage V-VI Xenopus oocytes were isolated by partial ovariectomy and defolliculated by collagenase treatment. Between 2 and 4 h after defolliculation, oocytes were injected with 50 nl of 1-100 ng/l human pI Cln or ClC-6 RNA. The oocytes were then incubated at 18°C for 2-4 days in ND-96 solution supplemented with gentamycin sulfate (50 mg/ml). ND-96 contains 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM Hepes, pH 7.5.
Whole-cell currents from oocytes were recorded using the two-microelectrode voltage clamp technique. Resistances of voltage and current electrodes filled with 3 M KCl were 0.5-2 M⍀. Current was sampled at 500-or 2000-s intervals and filtered at 1 or 0.1 kHz, respectively, using a quadrupole low pass Bessel filter. To eliminate the effect of voltage drop across the bath-grounding electrode, the bath potential was actively controlled. Linear components of capacity and leak currents were not subtracted. The bath solution was ND-96 except for the study of the anion permeability sequence, where NaCl (96 mM) was replaced with various sodium anions (96 mM). These experiments were performed using an agar bridge. Three voltage protocols were used during the analysis: (i) in step protocol 1, from a holding potential of Ϫ20 mV, oocytes were clamped for 800 ms from Ϫ100 mV to ϩ100 mV spaced 20 mV; (ii) in step protocol 2, oocytes were held at Ϫ70 mV, and 800-ms pulses to Ϫ40, 0, or ϩ40 mV were applied; and (iii) in the linear voltage ramp protocol, oocytes were held at Ϫ20 mV, and a linear voltage ramp from Ϫ100 mV to ϩ100 mV (0.4 V/s) was applied. All electrophysiological experiments were performed at room temperature on Xenopus oocytes that been permanently incubated at 18°C or that had been subjected to a temperature elevation (24°C for 3 h or 37°C for 30 min) prior to the electrophysiological analysis.
Permeability ratios (P X /P Cl ) for various anions were calculated using the formula, P X /P Cl ϭ (Cl e ϫ exp(⌬E ϫ F/RT) Ϫ Cl rest )/X e , with ⌬E being the shift in reversal potential, Cl e the extracellular Cl Ϫ concentration in ND-96, X e the extracellular anion concentration in anion-substituted ND-96, and Cl rest the remaining Cl Ϫ concentration in the anion-substituted media. Numerical data are represented as mean Ϯ S.E.
Expression of human pI Cln in Xenopus oocytes was verified by Western blot analysis using a polyclonal anti-pI Cln antiserum (1). Noninjected and pI Cln RNA-injected oocytes were lysed in a hypotonic buffer containing 25 mM Tris-HCl, pH 7.5, 20 mM NaCl, 2.5 mM EGTA. The lysate was centrifuged, and the supernatant was stored at Ϫ20°C. 50 g of Xenopus protein extract was analyzed on Western blot as described previously (1).
Preparation of Cytosolic and Microsomal Protein Fractions and Immunoblotting-Subcellular fractionation of a human endothelial cell line (EA.hy926 cell line; see Ref. 14) was performed by step centrifugation. Cultured cells were washed with phosphate-buffered saline (PBS) containing 1 mM EDTA, trypsinized, pelleted, and resuspended in 1 ml of isotonic PBS or 1 ml of hypotonic PBS (0.6 ml of PBS, 0.4 ml of H 2 O). After a 5-min incubation at room temperature or at 37°C, cells were lysed by sonication. Nuclei plus mitochondria were sedimented by centrifugation for 10 min at 10000 ϫ g. The supernatant was then fractionated into cytosolic and microsomal fractions by ultracentrifugation at 100000 ϫ g for 30 min. The microsomal fraction was resuspended in 20 mM Tris-HCl, pH 7.4, supplemented with 300 mM sucrose.
Confocal Immunofluorescence Study of pI Cln Localization in EA.hy926 Cells-Anti-pI Cln antibodies were first affinity-purified from a polyclonal anti-pI Cln antiserum using a hexahistidine-tagged human pI Cln fusion protein (1), which was bound to a polyvinylidene difluoride microporous membrane strip (Immobilon; Millipore). EA.hy926 endothelial cells were grown on gelatin-coated chamber slides (Lab Tek; Nunc Inc.). Prior to fixation in PBS containing 3% paraformaldehyde, cells were incubated for 5 min with isotonic or hypotonic (40% reduction in tonicity) PBS. Cells were permeabilized with 0.5% Triton X-100 and blocked with 10% goat serum in PBS for 1 h. They were then incubated with the affinity-purified polyclonal anti-pI Cln antibodies for 36 h at 4°C and thereafter with fluorescein isothiocyanate-conjugated antirabbit IgG (Sigma) at room temperature for 1 h. As a control, cells incubated only with the secondary antibodies were analyzed in parallel. Cells were imaged using a Bio-Rad MRC 1000 confocal microscope with a ϫ 40 oil immersion objective and a Kalman filter averaging between 5 and 10 frames. Both confocal (iris confocal aperture, 2 mm), and nonconfocal images (768 ϫ 512 lines) were stored.
Cell Volume Measurements-EA.hy926 cells were grown on glass coverslips, and cell height was monitored as described by Van Driessche et al. (15).

RESULTS
Expression of the pI Cln -associated Chloride Current in Xenopus Oocytes Is Temperature-dependent-It has previously been reported that expression of mammalian pI Cln in Xenopus oocytes resulted in an outwardly rectifying chloride current (I Cln ) that slowly inactivated at positive potentials and that could be blocked by extracellular nucleotides (4,16,17). However, our initial attempts to functionally express human pI Cln in Xenopus oocytes were rather unsuccessful; less than 10% of Xenopus oocytes injected with human pI Cln RNA displayed currents with an I Cln phenotype. The failure to functionally express I Cln could not be ascribed to the defolliculation method, since both collagenase-defolliculated and manually defolliculated Xenopus oocytes gave similar negative results (data not shown). We then accidentally discovered that functional expression in some oocyte batches was strongly promoted by incubating the oocytes at a higher temperature (24°C for 3 h or 37°C for 30 min) prior to the electrophysiological analysis at room temperature ( Fig. 1A). Fig. 1B compares the current amplitude at ϩ100 mV of injected and noninjected Xenopus oocytes when they were kept at 18°C and when the incubation temperature was raised to 24°C for at least 3 h prior to the electrophysiological analysis at room temperature (n ϭ 6 for each experimental condition; all oocytes derived from a single batch). In this experiment, only Xenopus oocytes that had been injected with human pI Cln and that had been subjected to a temperature shift displayed I Cln -type currents (see Fig. 2, A, C, and E, for a characterization of the current). From then on, we routinely included a minimum incubation of 3 h at 24°C in our experimental protocol.
The absence of I Cln -type currents in injected Xenopus oocytes permanently incubated at 18°C could not be explained by deficient translation of human pI Cln RNA. Western blot analysis of oocyte extracts confirmed the presence of human pI Cln protein in Xenopus oocytes that had been injected with human pI Cln RNA and that had been kept at 18°C (see Fig. 1C). The overall quality of the Xenopus oocytes within one batch and among different batches was always checked in parallel experiments in which we injected Xenopus oocytes with RNA coding for a Kv1.1 voltage-gated potassium channel (RCK1) (18). Typical K ϩ currents were observed in all of the injected oocytes independent of the incubation temperature (data not shown).
Human pI Cln and ClC-6 Induce an Identical Cl Ϫ Current when Expressed in Xenopus Oocytes-The kinetics, the anion permeability, and the pharmacology of I Cln currents induced by heterologous expression of mammalian pI Cln in Xenopus oocytes have been previously characterized (4,9). I Cln features are identical in manually defolliculated (4) and collagenasedefolliculated (9) Xenopus oocytes, and they are summarized in Fig. 2, A, C, and E. Briefly, I Cln is an outwardly rectifying current that slowly inactivates at positive membrane potentials (at least ϩ60 mV). It is an anion-selective current with a permeability sequence NO 3 Ϫ Ͼ I Ϫ Ͼ Br Ϫ Ͼ Cl Ϫ Ͼ gluconate. The permeability ratios P X /P Cl calculated from shifts in reversal potential in anion-substituted media are as follows (n ϭ 5): ClC-6 is a recently described member of the ClC chloride channel family (12). The functional characteristics and the physiological role of ClC-6 are still unknown, since functional expression of ClC-6 in Xenopus oocytes was reported to be negative (12). When we injected Xenopus oocytes with human ClC-6 RNA, we initially also obtained negative results. However, when we preincubated the Xenopus oocytes at a higher temperature prior to the analysis, we observed an outwardly rectifying current that inactivated slowly at positive potentials FIG. 1. Expression of the I Cln chloride current is temperaturedependent. A, collagenase-defolliculated Xenopus oocytes derived from a single batch were either injected or not injected with human pI Cln RNA and then incubated at 18°C for 3 days. Oocytes were tested at room temperature using the voltage step protocol shown at the top (800-ms pulses from Ϫ100 to ϩ100 mV in 20-mV steps; holding potential, Ϫ20 mV). Noninjected oocytes with or without temperature elevation and injected oocytes that were continuously kept at 18°C displayed a negative phenotype (background current in left trace). pI Cln -injected oocytes acquired a fast activating, outwardly rectifying current (right trace; see Fig. 2 for analysis of the current) only when they had been incubated at 24°C for 3 h prior to the analysis. Traces correspond to typical responses. B, plot showing the averaged current amplitude Ϯ S.E. during an 800-ms pulse at ϩ100 mV of noninjected and pI Clninjected oocytes with or without temperature elevation (n ϭ 6 for each experimental condition; all oocytes derived from one single batch). C, immunoblot analysis of the expression of human pI Cln in collagenasedefolliculated Xenopus oocytes incubated at 18°C. 50 g of soluble protein extract of Xenopus oocytes injected with pI Cln RNA (I) or of noninjected Xenopus oocytes (NI) were separated by SDS-polyacrylamide gel electrophoresis and electroblotted. The polyclonal anti-pI Cln antiserum identifies the heterologously expressed human pI Cln as a 39-kDa band in the injected oocytes. The 41-kDa band present in both lanes corresponds to the endogenous Xenopus pI Cln .
FIG. 2. Human pI Cln and ClC-6 induce an identical chloride current when expressed in Xenopus oocytes. Human pI Cln (A, C, and E) and human ClC-6 (B, D, and F) were expressed in collagenasedefolliculated Xenopus oocytes, and currents were recorded using the two-electrode voltage clamp method after oocytes had been transferred to 24°C for 3 h or more. The currents described in A-F are typical examples of the induced currents (n Ն 5). A and B, currents were recorded in ND-96 using a voltage step protocol (800-ms pulses from Ϫ100 to ϩ100 mV in 20-mV steps; holding potential, Ϫ20 mV). The I/V plot shows the current amplitude at the beginning (10 ms; filled circles) and at the end (open circles) of the pulse. C and D, a linear voltage ramp protocol (Ϫ100 mV to ϩ100 mV; 0.4 V/s) was used to determine the effect of extracellular anions. Recordings were performed in modified ND-96 in which NaCl (96 mM) was replaced with 96 mM of NaNO 3 (curve 1), NaI (curve 2), NaBr (curve 3), sodium gluconate (curve 5), or sodium cyclamate (curve 6). Curve 4 corresponds to unmodified ND-96. Shifts in reversal potential were used to calculate the permeability ratios (see "Results"). E and F, the induced currents are sensitive to cAMP and NPPB. Currents were recorded in ND-96 (curve 1) supplemented with 5 mM cAMP (curve 2) or 100 M NPPB (curve 3) using an identical voltage ramp protocol as in C and D.
pI Cln and ClC-6 Induce Identical Chloride Currents (at least ϩ60 mV; see Fig. 2, B, D, and F, for a description of the current). For example, in one batch of Xenopus oocytes, all oocytes (either injected with ClC-6 RNA or H 2 O-injected; n Ն 10 for each condition) remained negative when they were continuously kept at 18°C. In contrast, 10 of 14 oocytes injected with ClC-6 RNA acquired an I Cln phenotype after a temperature shock (either Ͼ3 h at 24°C or 30 min at 37°C; data not shown). In this experiment, 1 of 14 H 2 O-injected control Xenopus oocytes also became positive after temperature elevation.
In view of the similarities between the ClC-6-induced current and I Cln , we then went on to compare in greater detail these two currents in Xenopus oocytes (Fig. 2). The ClC-6induced currents were outwardly-rectifying, and they inactivated slowly at positive potentials. We quantified the rectification by calculating the ratio of the current amplitude at ϩ55 mV to that at Ϫ95 mV. These potentials were chosen, since they are approximately equidistant from the Cl Ϫ equilibrium potential in Xenopus oocytes. In ND-96 the rectification score (I ϩ55mV /I Ϫ95mV ) of the ClC-6-induced current was 6.9 Ϯ 1.5 (n ϭ 20), which is comparable with the 6.2 Ϯ 0.8 score of the pI Cln -induced current. The ClC-6-induced currents reversed between Ϫ25 and Ϫ35 mV in ND-96, and they depended on the presence of extracellular anions, since substitution of extracellular NaCl with various sodium anion solutions changed the current amplitude and shifted the reversal potential of the induced current. Based on shifts in the reversal potential we obtained the following permeability sequence: NO 3 Ϫ Ͼ I Ϫ Ͼ Br Ϫ Ͼ Cl Ϫ Ͼ gluconate. P X /P Cl ratios calculated from shifts in reversal potential were as follows (n ϭ 5): 1.37 Ϯ 0.04 (NO 3 Ϫ ), 1.24 Ϯ 0.03 (I Ϫ ), 1.12 Ϯ 0.02 (Br Ϫ ), and 0.46 Ϯ 0.03 (gluconate). The ClC-6-induced current was blocked by cyclamate. Furthermore, NPPB blocked the ClC-6-induced current to a similar degree as the pI Cln -induced current (81 Ϯ 5% block with 100 M at ϩ80 mV; n ϭ 5). Extracellular cAMP only inhibited the outward current (45 Ϯ 8% block with 5 mM cAMP at ϩ80 mV versus 3.4 Ϯ 1.8% block at Ϫ80 mV; n ϭ 5). Thus, expression of two structurally nonrelated proteins in Xenopus oocytes induced a chloride current with identical biophysical and pharmacological characteristics.

I Cln Is Also Present in Noninjected or H 2 O-injected Xenopus
Oocytes-A possible explanation for the observation that pI Cln and ClC-6 induce an identical Cl Ϫ current in Xenopus oocytes is that these proteins activate, directly or indirectly, a Cl Ϫ conductance that is endogenously present in Xenopus oocytes (see "Discussion"). One observation in favor of this interpretation is that an I Cln -type current is occasionally observed in noninjected or H 2 O-injected Xenopus oocytes. In a survey of 81 noninjected or H 2 O-injected Xenopus oocytes, we observed in 5 oocytes (6.2%) an I Cln phenotype (outward rectification; slow inactivation at potentials of at least ϩ60 mV; reversal potential about Ϫ30 mV). Similarly, Paulmichl et al. (4) reported that 36 of 943 H 2 O-injected oocytes (3.8%) displayed an I Cln phenotype.

Control Xenopus Oocytes Contain a Conductance That Is Phenotypically Identical to the Nucleotide-resistant Mutant I Cln -Expression of mutated pI Cln proteins in
Xenopus oocytes suggested a close link between pI Cln structure and I Cln phenotype, since mutations in a putatively extracellular, glycine-rich region led to mutant currents that could no longer be blocked by extracellular nucleotides such as cAMP (4). However, these mutations also changed the kinetics and the Ca 2ϩ dependence: in contrast to the wild type I Cln , the mutant current activated slowly at positive potentials, and reducing extracellular Ca 2ϩ decreased its amplitude. It was therefore concluded that pI Cln was a plasma membrane-spanning protein with an extracellular nucleotide binding site and extracellular Ca 2ϩ binding sites. However, this interpretation should be treated cau-tiously, since the mutant phenotype (slow activation, dependence on extracellular Ca 2ϩ , tail currents) resembles a conductance that is endogenously present in Xenopus oocytes (e.g. the steady state current in Ref. 19). We demonstrated this by eliciting the "mutant phenotype" in noninjected Xenopus oocytes using the same test protocols as Paulmichl et al. (4). All Xenopus oocytes possessed a current that slowly activated during an 800-ms voltage step to ϩ40 mV from a holding potential of Ϫ70 mV (Fig. 3). The time constant of activation at ϩ40 mV was 105 Ϯ 17 ms (n ϭ 15), which is identical to the time constant reported by Paulmichl et al. (4) (100 Ϯ 25 ms). The amplitude of this current at the end of an 800-ms pulse at ϩ40 mV varied between 0.1 and 2.1 A (0.62 Ϯ 0.14 A; n ϭ 15). This current depended on extracellular Ca 2ϩ , since perfusion of the oocytes with ND-96 containing 1 mM EGTA reduced its amplitude to 60 Ϯ 3% (n ϭ 4). Furthermore, it was not sensitive to 5 mM extracellular cAMP (n ϭ 5). This observation demonstrates that a current with a phenotype identical to the mutant I Cln is also present in noninjected control Xenopus oocytes.
Evidence against a Plasma Membrane Location of pI Cln -Subcellular fractionation of mammalian cells has revealed that the majority of pI Cln resides in the cytosol and that a small part of pI Cln associates with the microsomal fraction (2, 10). In addition, it has been reported for NIH3T3 fibroblasts and LL-FIG. 3. Noninjected Xenopus oocytes contain an endogenous conductance that resembles the mutant I Cln phenotype. A, noninjected, collagenase-defolliculated oocytes were clamped at Ϫ70 mV, and a depolarizing pulse to Ϫ40, 0, or ϩ40 mV for 800 ms was given. Note the appearance of a slowly activating current at ϩ40 mV as well as the inward tail currents. This current is phenotypically identical to the mutant I Cln current reported by Paulmichl et al. (4). B, a step to ϩ40 mV from a holding potential of Ϫ70 mV was applied either in ND-96 or in ND-96 containing 1 mM EGTA. Note the reduction of the current amplitude when extracellular Ca 2ϩ is lowered. This current is phenotypically identical to the mutant I Cln current reported by Paulmichl et al. (4). C, a step to ϩ40 mV from a holding potential of Ϫ70 mV was applied either in ND-96 or in ND-96 containing 5 mM cAMP.
pI Cln and ClC-6 Induce Identical Chloride Currents CPK1 cells that a reduction of the extracellular osmolarity induces a translocation of pI Cln from the cytosolic to the microsomal fraction (10). Although we did not think that there is a direct link between pI Cln and I Cl, swell (9), we were still interested in specific conditions in which pI Cln could reside in the plasma membrane. We therefore studied the subcellular distribution of human pI Cln in a human endothelial cell line (EA.hy926) in which we had previously documented the expression of the pI Cln protein (1). Functional studies also indicated that EA.hy926 cells were sensitive to changes in extracellular osmolarity (data not shown). A reduction of the extracellular tonicity with 28% activated a volume-sensitive Cl Ϫ current that was similar to I Cl, swell of other endothelial cell lines such as bovine pulmonary artery endothelial cells (CPAE; see Refs. 20 and 21) or human umbilical vein endothelial cells (22). With 140 mM Cs ϩ in the pipette solution and 5 mM Cs ϩ in the bath to block the inward K ϩ rectifier, I Cl, swell in EA.hy926 cells showed outward rectification, slow inactivation at positive potentials, and virtually no voltage dependence. Current densities at Ϫ80 mV in isotonic and hypotonic (28% reduction) conditions were, respectively, 2.8 Ϯ 0.3 pA/picofarad (mean Ϯ S.E.; n ϭ 19) and 49.3 Ϯ 5.2 pA/picofarad (mean Ϯ S.E.; n ϭ 19; p Ͻ 0.001). Using the method described by Van Driessche et al. (15) we directly tested whether EA.hy926 cells swell when the extracellular tonicity was lowered. A hypotonic stimulus (28 or 50% reduction) for 5 min induced an increase in cell height to, respectively, 138 Ϯ 3% (mean Ϯ S.E.; n ϭ 36) or 163 Ϯ 7% (mean Ϯ S.E.; n ϭ 29).
In view of the expression of pI Cln in EA.hy926 cells and in view of their sensitivity to changes in extracellular tonicity, we analyzed the subcellular distribution of pI Cln in these cells under isotonic and hypotonic conditions. Subcellular fractionation followed by Western blot analysis confirmed that the majority of pI Cln was recovered in the cytosolic fraction under isotonic conditions, indicating that pI Cln is a soluble and cytosolic protein (Fig. 4). A minor proportion of pI Cln was recovered in the microsomal fraction, indicating that the method was sufficiently sensitive to detect pI Cln in the membrane fraction under isotonic conditions. Importantly, the ratio between cytosolic and microsomal pI Cln did not change when the cells were subjected to a hypotonic stimulus (40% reduction in extracellular tonicity for 5 min; see Fig. 4). Identical results were obtained when the hypotonic treatment was performed at room temperature or at 37°C.
The association with the microsomal fraction during subcellular fractionation does not necessarily mean that pI Cln is a plasma membrane protein. Indeed, the microsomal fraction is a mixed population consisting of plasma membrane and intracellular membrane vesicles. Moreover, the association can be merely peripheral, for example mediated by ionic interactions between pI Cln and membrane lipids or proteins. To address this problem, we studied the localization of pI Cln by immunofluorescence confocal microscopy in EA.hy926 cells. In this experiment, cells were incubated at room temperature for 5 min either with isotonic (control cells) or hypotonic (60% tonicity) PBS prior to fixation. Confocal sections showed the presence of pI Cln throughout the cytosol (Fig. 5). Importantly, the periphery of the cell was devoid of pI Cln . Furthermore, the intracellular distribution of pI Cln was unaffected when the cells were subjected to a hypotonic stimulus (Fig. 5). Confocal sections revealed an identical pattern of pI Cln distribution for control cells and cells subjected to a hypotonic challenge. DISCUSSION In this study we describe the induction of an identical Cl Ϫ current by expression of two structurally unrelated proteins in Xenopus oocytes. The current induced by human pI Cln and ClC-6 corresponds phenotypically to the I Cln current observed after heterologous expression of other mammalian pI Cln proteins in Xenopus oocytes (4,16,17). In short, I Cln is an outwardly rectifying anion current with slow inactivation kinetics at positive potentials. Its permeability sequence is NO 3 Ϫ Ͼ I Ϫ Ͼ Br Ϫ Ͼ Cl Ϫ Ͼ gluconate. It is blocked by cyclamate, NPPB, and extracellular cAMP. In Xenopus oocytes, the I Cln current can be clearly discriminated from the volume-activated chloride current (9). The success rate of I Cln expression was greatly enhanced in some Xenopus oocyte batches by incubating the Xenopus oocytes injected with pI Cln or ClC-6 RNA for Ն3 h at 24°C or for 30 min at 37°C prior to the electrophysiological analysis. Brandt et al. (12) mention that they incubated ClC-6 injected Xenopus oocytes at 18°C for 2 or 3 days before current measurements. The absence of a temperature elevation may very well explain why they did not observe a chloride current in ClC-6-injected Xenopus oocytes. The temperature dependence was specific for I Cln , since functional expression of a voltagedependent K ϩ channel (RCK1) did not require a preceding temperature shift. In principle, temperature-dependent processes can affect channel expression by interfering with the biosynthesis (translation, membrane insertion, and folding), the transport from rough endoplasmic reticulum to the plasma membrane, and/or the gating mechanism. As for pI Cln , we can rule out a temperature effect on translation of pI Cln RNAs, since we were able to detect human pI Cln in extracts of injected Xenopus oocytes that had only been incubated at 18°C. A similar effect of incubation temperature on channel expression in Xenopus oocytes has been described for epithelial Na ϩ channels, in which case insertion into the plasma membrane seems to be the temperature-dependent step (23).
How can we explain the finding that pI Cln and ClC-6 induce an identical Cl Ϫ current when expressed in Xenopus oocytes and that a similar current can be observed in a small minority of control Xenopus oocytes? There are, in principle, two alternative possibilities. (i) Both pI Cln and ClC-6 are anion channels FIG. 4. pI Cln is predominantly present in the cytosolic fraction under both isotonic and hypotonic conditions. Western blot analysis of pI Cln distribution in cytosolic (C) and microsomal (M) fractions of EA.hy926 endothelial cells is shown. Cells were first incubated with isotonic (iso) or hypotonic (hypo) PBS and then subfractionated. 50 g of cytosolic proteins and 200 g of microsomal proteins (to facilitate detection in the microsomal fraction) were separated on SDS-polyacrylamide gel electrophoresis and electroblotted. A polyclonal anti-pI Cln antiserum was used to identify pI Cln . The majority of pI Cln was recovered in the cytosolic fraction. Note that for the cytosolic fraction, 50 g correspond to 6% of the total protein mass, whereas for the microsomal fraction, 200 g correspond to 20% of the total protein mass. No shift from the cytosolic to the microsomal fraction was observed when the cells were subjected to a hypotonic stimulus prior to subcellular fractionation.
pI Cln and ClC-6 Induce Identical Chloride Currents that reside in the plasma membrane and that mediate the I Cln current. However, we then have to explain how two structurally unrelated proteins can form a channel with identical biophysical and pharmacological characteristics. Furthermore, this does also not account for the the occasional appearance of I Cln in control oocytes. Finally, as argued below, there is no firm experimental evidence in favor of a plasma membrane location of pI Cln . (ii) The alternative explanation is that pI Cln and ClC-6 activate, directly or indirectly, an identical anion channel that is endogenously present in Xenopus oocytes. Opening of this channel would then lead to the I Cln current. The sporadic appearance of I Cln in control Xenopus oocytes is consistent with this explanation. Whether proteins other than pI Cln or ClC-6 can induce I Cln when expressed in Xenopus oocytes, remains a possibility that cannot be discarded at the moment. Finally, the induction of an endogenous conductance by exogenous pI Cln or ClC-6 is in line with several other reports that have described the activation of other endogenous channels after overexpression of foreign proteins in Xenopus oocytes (24 -29).
In principle, one could argue that the heterologously expressed proteins, i.e. human pI Cln and ClC-6, interact directly or indirectly with pI Cln that is endogenously present in Xenopus oocytes (1,2) and that the Xenopus pI Cln forms the actual channel. This interpretation presupposes that pI Cln is an intrinsic membrane protein that spans the plasma membrane. However, in our opinion there are no compelling arguments in favor of this assertion. (i) The amino acid sequence of pI Cln contains no hydrophobic regions that are sufficiently long to traverse the membrane as an ␣-helix. A structural model has been proposed in which the transmembrane part of pI Cln consists of an amphipathic ␤-sheet (4). However, no experimental evidence in favor of this model has been presented. (ii) The changes in current phenotype (loss of block by extracellular nucleotides, sensitivity to extracellular Ca 2ϩ ) observed after expression of a mutant pI Cln do not necessarily mean that pI Cln spans the plasma membrane. As demonstrated, an identical phenotype was present in noninjected oocytes. (iii) The association of small amounts of pI Cln with the membrane fraction after cell fractionation does not necessarily imply that it is a transmembrane protein inserted into the plasma membrane. Alternative explanations are that pI Cln is a peripheral membrane protein and/or that it associates with intracellular membrane structures (endoplasmic reticulum, Golgi, endosomes, etc.) rather than with the plasma membrane. (iv) The preferential association with the soluble fraction (our data and Ref. 2) as well as the confocal immunofluorescence data point to a cytosolic location. (v) In contrast to Paulmichl et al. (10) we did not observe a shift of pI Cln to the plasma membrane upon reducing extracellular osmolarity. (vi) Krapivinsky et al. (2) have examined the distribution of endogenous pI Cln in Xenopus oocytes, and they were unable to identify pI Cln in oocyte microsomes. We therefore conclude that pI Cln is not a plasma membrane-located channel protein and, consequently, that the I Cln current is not carried by the pI Cln protein. As to ClC-6, it has formally not yet been proven that it resides in the plasma membrane. However, its structural relationship to well documented plasma membrane Cl Ϫ channels such as ClC-0 and ClC-1 and the presence of several hydrophobic segments in the hydropathy analysis strongly suggest that ClC-6 is a membrane protein (12).
Our interpretation of the present data implies a molecular pathway that links a cytosolic protein, pI Cln , with a plasma membrane anion channel. Krapivinsky et al. (2) have shown that pI Cln forms oligomeric complexes with several cytosolic proteins, one of which has been identified as actin. In addition, a small amount of pI Cln was found to be associated with the membrane cytoskeleton (2). These observations are consistent with pI Cln forming part of a protein-protein interaction cascade that may be responsible for the activation of a plasma membrane anion channel. Clearly, more experimental work is re- FIG. 5. Immunofluorescence study of the subcellular distribution of pI Cln in EA.hy926 endothelial cells under isotonic and hypotonic conditions. Immunofluorescence detection of pI Cln in EA.hy926 endothelial cells with affinity-purified polyclonal anti-pI Cln antibodies and secondary fluorescein isothiocyanate-conjugated anti-rabbit IgG antibodies is shown. A, visualization of pI Cln under isotonic control conditions. The specificity of the signal obtained with the affinity-purified anti-pI Cln antibodies was verified by omitting the primary antibody and incubating the EA.hy926 cells only with the secondary fluorescein isothiocyanate-conjugated antibodies. This procedure did not yield any significant signal above background (B). C and D show at a higher magnification confocal sections through EA.hy926 cells that had been treated with an isotonic (C) or a hypotonic buffer (40% reduction for 5 min; D). These conditions reveal more clearly the cytosolic localization of pI Cln , and importantly, they do not show evidence for a translocation of pI Cln to the plasma membrane after hypotonic stimulation. Parameters for laser microscopy were identical for panels A-B and C-D, respectively. The color scale (see vertical bar in B) ranges from blue (background signal) to white (saturating signal).
pI Cln and ClC-6 Induce Identical Chloride Currents quired to unravel this pathway. Similarly, it is not known how ClC-6 activates the endogenous channel. Since ClC-6 most likely resides in the membrane (see above), it may interact with the anion channel either directly or indirectly via one or more intervening proteins.
At present we do not know whether the activation of I Cln is related to the proper physiological function of pI Cln and of ClC-6 or whether it represents a mere side effect of expressing exogenous proteins in Xenopus oocytes. Yet, the present data directly impinge on the functional models for pI Cln and its proposed physiological role as a volume-regulated anion channel or as a regulator of I Cl, swell (2, 6 -8). Collectively, the lack of experimental evidence that pI Cln is a plasma membrane anion channel and our previous observation that I Cln and I Cl, swell are two different currents (9) argue against a role for pI Cln as a volume-regulated anion channel or as a regulator thereof. As to the physiological role of ClC-6, there is as yet no formal evidence that it is a Cl Ϫ channel, since expression studies have either yielded negative results (12) or the I Cln phenotype (this study). The structural relationship of ClC-6 with proven Cl Ϫ channels such as ClC-0, ClC-1, ClC-2, and ClC-5 is compatible with a role in Cl Ϫ or anion transport, but further experiments are required to clarify this issue.
Finally, although Xenopus oocytes have proven to be a reliable and powerful tool to analyze ion channels, they may contain specific pitfalls when used to identify or characterize exogenous Cl Ϫ currents. First of all, at least four distinct endogenous Cl Ϫ currents have been described in Xenopus oocytes: a Ca 2ϩ -activated Cl Ϫ current (30), a hyperpolarizationactivated Cl Ϫ current (31), a volume-activated Cl Ϫ current (9,32,33), and the I Cln current. Moreover, two of them can be activated by the expression of exogenous proteins. Expression of small integral membrane proteins such as phospholemman, MAT-8, IsK, SYN-C, and NB activates the hyperpolarizationactivated Cl Ϫ current (24 -28), whereas I Cln is induced by pI Cln and ClC-6.