Molecular cloning and characterization of an intracellular chloride channel in the proximal tubule cell line, LLC-PK1.

CLC5 is an intracellular chloride channel of unknown function, expressed in the renal proximal tubule. The subcellular localization and function of CLC5 were investigated in the LLC-PK1 porcine proximal tubule cell line. We cloned a cDNA for the porcine CLC5 ortholog (pCLC5) that is predicted to encode an 83-kDa protein with 97% amino acid sequence identity to rat and human CLC5. By immunofluorescence, pCLC5 was localized to early endosomes of the apical membrane fluid-phase endocytotic pathway and to the Golgi complex. Xenopus oocytes injected with pCLC5 cRNA exhibited outwardly rectifying whole cell currents with a relative conductance profile (nitrate Cl(-) approximately Br(-) > I(-) > acetate > gluconate) different from that of control oocytes. Acidification of the extracellular medium reversibly inhibited this outward current with a pK(a) of 6.0 and a Hill coefficient of 1. Overexpression of CLC5 in LLC-PK1 cells resulted in morphological changes, including loss of cell-cell contacts and the appearance of multiple prominent vesicles. These findings are consistent with a potential role for CLC5 in the acidification of membrane compartments of both the endocytic and the exocytic pathway and suggest that its function may be important for normal intercellular adhesion and vesicular trafficking.

Functional evidence for chloride channels has been demonstrated in many intracellular organelles including the endoplasmic reticulum, Golgi, endosomes, lysosomes, and synaptic vesicles (1). Although their molecular identity in most cases is unknown, members of several structurally distinct chloride channel gene families such as p64 (2,3), cystic fibrosis transmembrane regulator (4), and CLC6 (5) have recently been found to be expressed intracellularly. Dent's disease is an Xlinked inherited disorder characterized by hypercalciuria, nephrocalcinosis, and renal failure, associated with a Fanconilike syndrome, low molecular weight proteinuria, and rickets (6 -8), which is caused by inactivating mutations in the renal chloride channel, CLC5 (9). CLC5 has been shown to be expressed intracellularly in endosomes of the renal proximal tubule, where it colocalizes with the vacuolar H ϩ -ATPase (10 -13), but its physiological role is unknown. It has been postulated that CLC5 may provide an electrical shunt to dissipate the potential gradient across the endosomal membrane generated by electrogenic proton transport into the endosomal lumen. Thus, loss of CLC5 function would be predicted to impair endosomal acidification, which is normally required for endocytosis (14), trafficking to lysosomes (15), and recycling back to the surface (16). In this model, low molecular weight proteinuria in Dent's disease occurs because of failure of endocytosis from the proximal tubule lumen, which is the normal pathway for reabsorption of low molecular weight proteins (17). The occurrence of a Fanconi-like syndrome may be explained by internalization of apical membrane sodium-coupled solute transporters into endosomal vesicles but failure to recycle these transporters back to the surface. The hypercalciuria in this disease appears to be absorptive in origin (18) and may be due to abnormal regulation of proximal tubule 25-hydroxyvitamin D 1-hydroxylase activity. How this might be possible, given that the 1-hydroxylase is located on the inner membrane of mitochondria, remains unknown.
To understand the cellular role of CLC5 and test these hypotheses, it will be critical to have a cell culture model that normally expresses CLC5 and physiologically resembles native proximal tubule epithelium. LLC-PK1 is a well characterized cell line that is an excellent model of the proximal tubule epithelium (19). It shares many of the key properties of native proximal tubules that CLC5 might be predicted to regulate, including expression of apical transporters such as the sodiumhydrogen exchanger, NHE3 (20), and the high affinity sodiumglucose cotransporter, SGLT1 (21), endosomal expression of megalin (22), and the vacuolar H ϩ -ATPase (23), receptor-mediated endocytosis of proteins from the luminal surface (24), and 25-hydroxyvitamin D 1-hyroxylase activity (25). We report here the cloning of a porcine CLC5 ortholog (pCLC5), its localization to Golgi and endosomal vesicles of LLC-PK1 cells, and its functional characterization by heterologous expression in the Xenopus oocyte system.

MATERIALS AND METHODS
Tissue Culture and Northern Blot Analysis-LLC-PK1 cells (American Type Culture Collection, catalog number CL-101) were cultured to confluence on plastic plates in Dulbecco's modified Eagle's medium with 5% fetal bovine serum at 37°C in 5% CO 2 . Transient transfections were performed on 70 -80% confluent cells by lipofection using Lipo-fectAMINE Plus Reagent (Life Technologies, Inc.), according to the manufacturer's instructions. Expression of enhanced green fluorescent protein by cotransfection of pEGFP-C2 (CLONTECH) was routinely used to assess transfection efficiency (typically 10 -20% of cells) and identify positive cells. Northern blots under high stringency conditions (hybridization at 65°C in the presence of 50% formamide, final washes at 65°C in 0.1ϫ SSCP) were performed on poly(A) ϩ RNA isolated from these cells, using digoxigenin-labeled antisense riboprobes of rat CLC5, CLC4, and CLC3 (11,26) and detected by chemiluminescence with CDP-Star and the Dig Genius System (Roche Molecular Biochemicals).
cDNA Cloning and Generation of Expression Constructs-To clone the pCLC5 cDNA, degenerate oligonucleotide primers (sense, ATHTCT-GCNCAYTGGATGAC; antisense, TAYTTRTTRAARCARTGRCA) were designed to the conserved amino acid sequences of human (27), rat (28), and Xenopus (29) CLC5. By homology-based reverse transcription-polymerase chain reaction (PCR) 1 amplification of LLC-PK1 cell mRNA, a 427-base pair cDNA was isolated that was highly homologous to CLC5 in other species. The remainder of the cDNA encompassing the coding sequence was cloned by a combination of primer walking and rapid amplification of cDNA ends and then sequenced in both directions by the dideoxynucleotide chain termination method. To exclude PCR misincorporation errors, each nucleotide sequence within the open reading frame was corroborated at least three times using overlapping clones amplified from independent PCRs. As a positive control, a rat CLC5 cDNA (rCLC5) containing the full coding region, identical to the published sequence (28), was amplified by PCR and cloned in a similar fashion. Constructs encoding the rCLC5 coding region with a FLAG epitope tag (DYKDDDDK) at either the N or the C terminus and an optimized Kozak translation initiation sequence were generated by PCR-based oligonucleotide mutagenesis and verified by DNA sequencing.
To generate mammalian expression constructs for transfection into LLC-PK1 cells, pCLC5 and rCLC5 cDNA were cloned into the plasmid vector, pcDNA3 (Invitrogen), under the control of a constitutive cytomegalovirus promoter. To generate vectors with convenient restriction sites suitable for expression in Xenopus oocytes, we adapted a pTLN II vector (30) in which the NcoI site and occult Kozak sequence had previously been deleted by mung bean exonuclease and replaced with an ApaI site (gifts of Dr. Peying Fong and Dr. Joseph Mindell). The polylinker was next removed with ApaI and BstEII, the ends made blunt with Klenow polymerase, and the polylinker of pcDNA3.1-zeo (Invitrogen) excised with PmeI and inserted into this site in both orientations to generate the vectors pOX(ϩ) (oriented so that SP6 transcription proceeds from AflII to ApaI), and pOX(Ϫ). These vectors have the advantage that inserts cloned into the polylinker can be easily swapped as a restriction cassette into any of the pcDNA3.1 and pIND series of vectors for mammalian cell expression, which have the same polylinker, without the need to create new restriction sites. The coding sequences of pCLC5 and rCLC5 were cloned into the KpnI and XbaI sites of pOX(ϩ).
Protein Preparation and Western Blot Analysis-In vitro translation of pCLC5 and rCLC5 cDNA was performed in the presence or absence of canine microsomal membranes using the TNT-coupled reticulocyte lysate system (Promega), and the products were electrophoresed on a 7.5% denaturing SDS-polyacrylamide gel and electrotransferred onto polyvinylidene difluoride membrane. Calibration of molecular weights was achieved by electrophoresing in parallel a lane of unstained protein molecular weight standards. In the negative control sample, cDNA was omitted from the reaction. To detect all translated products, biotinylated lysine tRNA (Transcend tRNA, Promega) was included in the translation reaction, and the membrane was blotted with streptavidinalkaline phosphatase followed by a chemiluminescent substrate, according to the manufacturer's instructions. To detect CLC5 immunoreactivity, Transcend tRNA was omitted, and the membranes were immunoblotted with the affinity-purified polyclonal antibody fractions, C1 and C2, exactly as described previously (11). To detect FLAG epitope-tagged proteins, immunoblots were performed with the M2 monoclonal antibody (Sigma) at a concentration of 10 g/ml.
To detect pCLC5 in LLC-PK1 membranes, confluent cultured cells were washed in phosphate-buffered saline (PBS), scraped off the plate, and suspended in sucrose/histidine buffer containing 0.25 M sucrose, 30 mM histidine, 1 mM EDTA, 2ϫ Complete protease inhibitor mixture (Roche Molecular Biochemicals), adjusted to pH 7.4. The cells were homogenized by 6 passes through a 25-gauge needle, centrifuged at 1000 ϫ g for 5 min to sediment nuclei, and the post-nuclear supernatant centrifuged at 100,000 ϫ g for 20 min to yield a microsomal membrane pellet. This was then resuspended in 2% SDS-containing gel-loading buffer for electrophoresis and immunoblotting as described above. Deglycosylation was performed by boiling the protein samples for 2 min in phosphate-buffered saline (PBS) containing 50 mM ␤-mercaptoethanol, 10 mM EDTA, and 0.1% sodium dodecyl sulfate and then adding 0.5% Nonidet P-40 and 18 units/ml N-glycosidase F (Sigma) and incubating 12-16 h at 37°C. Immunofluorescence Staining-Native or transfected LLC-PK1 cells grown on glass coverslips to 85% confluence were washed in PBS and fixed in methanol at Ϫ80°C for 10 min. To entrap fluorescent markers in early endosomes of the apical endocytotic pathway, LLC-PK1 cells were incubated prior to fixation in Dulbecco's modified Eagle's medium containing fluorescein isothiocyanate (FITC)-labeled conjugates (all from Sigma) of dextran (1 mg/ml, average molecular mass 12 kDa), albumin (1 mg/ml), or Ricinus communis agglutinin (ricin, 0.1 mg/ml) for 15 min at 37°C.
Immunostaining of endogenous pCLC5 with the C1 antibody, using amplification by the direct tyramide signal amplification kit (PerkinElmer Life Sciences), was performed exactly as described previously (11). For double-staining studies with the Na ϩ -K ϩ -ATPase, a monoclonal antibody (gift of Dr. Kevin Bush) was used without dilution. For immunodetection of the FLAG epitope, the M2 anti-FLAG monoclonal antibody (Sigma) was used, diluted 1:100 into PBS containing 1% bovine serum albumin, 0.3% Triton X-100, and 0.2% skimmed milk. ARF1 rabbit polyclonal antibody (gift of Dr. Vladimir Marshansky), used as a marker of the Golgi complex (31,32), was applied at a dilution of 1:250, also in the presence of 0.3% Triton X-100.
Slides were visualized with a Bio-Rad MRC-1024 confocal kryptonargon laser scanning microscope. For double-labeled slides, images were acquired sequentially for each fluorophore in single label mode to minimize "bleed-through" between channels. Each pair of images was then imported into Adobe Photoshop 3.0, false color added, and merged to generate dual-color images.
Heterologous Expression of CLC5 in the Xenopus Oocyte System-pOX(ϩ) plasmid constructs encoding pCLC5 and rCLC5 cDNA were linearized with MluI, and capped cRNA was transcribed with SP6 polymerase. Stage V/VI Xenopus laevis oocytes were digested with collagenase, defolliculated, and injected with 50 nl of cRNA (1 g/l) or sterile water for negative controls and then incubated at 18°C in ND96 (containing in mM: NaCl 96, KCl 2, CaCl 2 1.8, MgCl 2 1, HEPES 5, adjusted to pH 7.4) with 50 g/ml gentamicin for 4 -6 days.
To detect CLC5 expression in the oocytes, total microsomal membranes were prepared. Oocytes were suspended in sucrose/histidine buffer, lysed by triturating 30 times through a 200-l volume pipette tip, and then centrifuged at 1000 ϫ g for 10 min to sediment cellular debris. The floating layer of yellow lipid was removed with a cottontipped applicator, and the supernatant was recovered and centrifuged at 100,000 ϫ g for 20 min to yield a microsomal membrane pellet.
To identify the subset of proteins expressed at the cell surface, a modification of a published surface biotinylation method (33) was used. Between 40 and 80 oocytes were washed in OR2 (containing in mM: NaCl 82.5, KCl 2, MgCl 2 1, Na phosphate 10, adjusted to pH 7.4) and then incubated in OR2 containing 4 mg/ml EZ-Link sulfo-NHS-biotin (Pierce) for 10 min at room temperature (22°C). The reaction was then stopped by adding 0.5 M glycine, pH 7.4, and the oocytes were washed twice in glycine and twice in OR2. The oocytes were then suspended in lysis buffer containing 2% Nonidet P-40, 150 mM NaCl, 2 mM CaCl 2 , 20 mM Tris-HCl, 2ϫ Complete protease inhibitor, adjusted to pH 7.4. The oocytes were lysed by trituration, centrifuged at 1000 ϫ g for 10 min, the lipid layer removed, and the supernatant dialyzed overnight against SAV buffer containing 0.3% Nonidet P-40, 0.5 M NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM Tris-HCl, adjusted to pH 8. The lysate was then centrifuged at 16,000 ϫ g for 30 min to remove insoluble material. The supernatant was incubated with streptavidin-agarose beads overnight at 4°C. The next day, the agarose was sedimented by brief centrifugation, and the supernatant, which contains non-biotinylated proteins, was recovered. The pellet was washed 4 times in SAV buffer, and the biotinylated proteins were recovered by solubilization in 1% SDS. The protein recovery of each fraction was determined by a detergent-compatible colorimetric assay (DC Protein Assay, Bio-Rad). Equal amounts of each protein sample (25 g) were loaded in each lane of a denaturing 7.5% SDS-polyacrylamide gel for electrophoresis and immunoblotted with C1 antiserum.
Electrophysiological Analysis-Two-microelectrode voltage clamp studies of Xenopus oocytes were performed at room temperature using a Clampator-1B (Dagan Instruments, Minneapolis, MN) controlled by pCLAMP 8.0 software (Axon Instruments, Foster City, CA). Microelectrodes were filled with 3 M KCl and had tip resistances of 0.5-2 m⍀. Oocytes were clamped at a holding potential of Ϫ50 mV, and voltage steps from Ϫ120 to ϩ100 mV in 20 mV increment were applied. Data were acquired at 10 kHz and low pass-filtered at 2 kHz. For control studies, oocytes were superfused with ND96 modified to contain 98 mM NaCl and 0.3 mM CaCl 2 so as to minimize activation of the endogenous calcium-activated chloride conductance in oocytes. In anion substitution experiments, 80 mM chloride was substituted with an equimolar concentration of another anion (except where otherwise specified), and the bath Ag/AgCl electrode was protected by a KCl-agar bridge. In studies of the effect of lowering bath pH, MES was substituted for HEPES as the buffer and titrated to the desired pH with NaOH. Each experimental result reported was confirmed in at least 3 distinct batches of oocytes isolated from different frogs on separate days.

RESULTS
cDNA Cloning and Sequence Analysis of pCLC5-By high stringency Northern blot analysis with a rat CLC5 probe, LLC-PK1 cells appear to express mRNA encoding a CLC5 homolog. A strong band of 9.5 kb was observed, similar to the size of human and rat CLC5 mRNA (28,34), as well as a weaker band of approximately 3 kb, similar in size to the X. laevis CLC5 mRNA (29) (Fig. 1). The pCLC5 cDNA was isolated and sequenced from multiple clones by homology-based reverse transcription-PCR of mRNA from LLC-PK1 cells. It contains a long open reading frame of over 2 kb. The first ATG codon after an in-frame upstream stop codon is not in the context of a suitable Kozak consensus sequence (35), but the second ATG is, and the latter has therefore been assigned as the initiator codon (Fig.  2). The resultant open reading frame of 2238 base pairs predicts a protein of 746 amino acids with a molecular mass of 83 kDa (Fig. 3A). The protein sequence of pCLC5 is the same length as and highly homologous with rat CLC5 (rCLC5, 97% amino acid identity (28)) and is also very similar to human CLC5 (98% amino acid identity in overlapping region (27)) except for the absence of a 20-residue stretch at the N-terminal of human CLC5. We conclude that pCLC5 is the porcine ortholog of CLC5.
The hydropathy plot of pCLC5 is very similar to that of other CLC proteins and predicts 13 hydrophobic domains (Fig. 3B). Based on glycosylation scanning and protease protection studies in CLC1 (36), 10 of these domains are believed to span the membrane, with both N and C termini located on the cytosolic side of the membrane. pCLC5 has two potential N-glycosylation sites (Fig. 3A). One at residue 408 is predicted to be in the extracellular loop between the 7th and 8th transmembrane segments, is conserved in all CLC family members, has been shown to be used in vitro in CLC-K1, CLC-K2, and CLC1 (36,37), and is likely a bona fide glycosylation site. The other site at residue 38 is predicted to be intracellular. There are two overlapping cAMP-dependent phosphorylation consensus sequences in the cytosolic loop between the 6th and 7th transmembrane domain and four potential protein kinase C phosphorylation sites that are predicted to be located on the cytosolic side (Fig. 3A).
pCLC5 Is Immunoreactive with the C1 Antibody-We previously generated polyclonal antiserum against a fusion protein containing the C terminus of rCLC5 (11), and we affinitypurified from it two fractions, C1 (which reacts primarily with rCLC5 but also cross-reacts with CLC3 and CLC4), and C2 (which is specific for rCLC5). To determine if these sera could be used for immunodetection of pCLC5, we first in vitro translated pCLC5. As shown by the streptavidin blot in Fig. 4A, pCLC5 was successfully translated to yield a protein with an apparent molecular mass of 65 kDa, identical to the apparent size of in vitro translated rCLC5, and absent from the negative control lane. By Western blotting, the C1 antibody reacts strongly with the translated pCLC5 protein (Fig. 4A), but C2 serum did not recognize pCLC5 at all (data not shown). Furthermore, C1 recognizes a band of the same apparent size in LLC-PK1 100,000 ϫ g membranes (Fig. 4B).
pCLC5 and rCLC5 Exhibit Anomalous Electrophoretic Mobility on Denaturing Gels-Because the apparent molecular mass of pCLC5 (and of rCLC5) on our denaturing gels (65 kDa) is lower than the 83-kDa size predicted from its amino acid sequence, and anomalous migration on protein gels has not been reported for any other CLC protein, the potential contribution of post-translational modification of pCLC5 and rCLC5 proteins by glycosylation and proteolysis was assessed. In vitro translated pCLC5 and rCLC5 polypeptides could both be glycosylated by canine microsomal membranes, yielding broad bands that migrate with an apparent molecular mass of ϳ80 -85 kDa (Fig. 4B), identical in size to the C1-immunoreactive band in rat kidney cortex (11). This suggests that mature glycosylated CLC5 protein (as found in rat kidney cortex) migrates at 80 -85 kDa, whereas the unglycosylated polypeptide (as found in LLC-PK1 cells) migrates at 65 kDa. Consistent with this, deglycosylation by N-glycosidase F collapsed the C1-immunoreactive band in rat kidney cortex from 80 to 85 kDa down to 65 kDa, but had no effect on the 65-kDa band in LLC-PK1 cells (Fig. 4B).
The finding of bands that migrate with a molecular weight that is consistently lower than predicted could also be explained by initiation of translation from an initiator codon downstream of the predicted start site, giving rise to a protein truncated at the N terminus or by proteolytic degradation. To address these possibilities, we generated epitope-tagged rCLC5 constructs that encode the FLAG octapeptide epitope fused directly to the N or C terminus. Both constructs could be translated in vitro and yielded bands that run only very slightly higher on the gel than those of the untagged proteins (Fig. 4A). Furthermore, both tagged polypeptides were immunoreactive with a monoclonal antibody directed against the FLAG epitope, demonstrating that neither the N nor the C terminus had been truncated. We conclude that the position of pCLC5 and rCLC5 bands on denaturing protein gels is the consequence of anomalous electrophoretic mobility of the native, full-length polypeptide and is not related to glycosylation or proteolysis.
Localization of CLC5 in LLC-PK1 Cells-Punctate immunofluorescent staining with C1 was detected in intracellular vesicles of native LLC-PK1 cells, particularly in the perinuclear region (Fig. 5). In some cells, staining was asymmetrically concentrated in a dense juxtanuclear patch suggestive of local- ization to the Golgi apparatus. By double staining for the Na ϩ -K ϩ -ATPase, which is located on the lateral and, to a lesser extent, the basal membrane and then confocal microscopy with vertical section reconstruction, we observed that C1 staining was absent from apical or basolateral plasma membranes. As C1 might potentially cross-react with another isoform such as CLC4, which is also expressed in LLC-PK1 cells (Fig. 1), we confirmed the subcellular localization of CLC5 by heterologous expression of our FLAG-tagged CLC5 construct. In LLC-PK1 cells transiently transfected with this construct, the epitopetagged CLC5 exhibited two distinct patterns of distribution, either in scattered intracellular vesicles (Fig. 6, A and B) or in an asymmetric juxtanuclear density (Fig. 6, C and D), very similar to that of native pCLC5. Interestingly, some transfected cells exhibited a grossly abnormal rounded appearance with loss of the normal borders with neighboring cells and were packed intracellularly with vesicles that were prominently visible by light microscopy and positively stained for CLC5 by immunofluorescence (Fig. 6E).
Colocalization of CLC5 in LLC-PK1 with Organelle-specific Markers-We hypothesized that the distribution of CLC5 in LLC-PK1 cells might reflect its localization in apical membrane endosomal vesicles and the Golgi complex. To investigate whether CLC5 is localized in endosomes, we used fluoresceinlabeled albumin and dextran as putative markers of the recep-tor-mediated and fluid-phase endocytotic pathway, respectively (14,38), and fluorescein-labeled ricin as a putative panendosomal marker (39). LLC-PK1 cells transfected with FLAGtagged rCLC5 were exposed to each of these fluorescent markers at their apical surface over a sufficient length of time for them to be entrapped in apical membrane early endosomal vesicles and were then fixed, stained with the FLAG primary antibody and a Texas-red-conjugated secondary antibody, and visualized by confocal microscopy. As shown in Fig. 7, the red fluorescence due to FLAG antibody binding showed spatial overlap with the green fluorescence of endocytosed dextran but not with that of albumin or ricin, suggesting that CLC5 colocalizes with entrapped dextran in early endosomes of the fluidphase endocytotic pathway. As a negative control, cells that were not exposed to a green fluorescent marker were also stained with the FLAG antibody and Texas Red secondary antibody (Fig. 7, bottom row). Confocal images acquired in an identical manner from this slide showed no overlap of green and red spectra, thus excluding an artifact due to "bleedthrough" of the red fluorescence into the green channel.
CLC5 was also expressed in the Golgi complex (Fig. 7), since we observed colocalization of FLAG antibody binding in the juxtanuclear region with staining by an antibody to the Golgi complex marker, ARF1 (31,32).
Expression of pCLC5 in the Xenopus Oocyte System-To investigate its functional properties, pCLC5 was transiently expressed in the Xenopus oocyte system. Negative control oocytes injected with sterile water express an endogenous CLC5 ortholog, xCLC5 (29), that cross-reacts with the C1 antibody and appears as two bands with apparent molecular masses of 65 and ϳ85 kDa on Western blots (Fig. 8A), neither of which can be deglycosylated by N-glycosidase F (data not shown). Oocytes injected with pCLC5 cRNA demonstrate a marked increase in the intensity of these two bands, indicating the synthesis of new CLC5 protein (Fig. 8A). To confirm that pCLC5 is actually expressed on the surface of cRNA-injected oocytes, we labeled all proteins on the oocyte plasma membrane that had primary amines exposed at the extracellular surface by biotinylation, and we then separated them from non-biotinylated proteins by affinity purification with streptavidin-agarose beads and immunoblotted the fractions with C1 antibody (Fig. 8B). In control oocytes, the C1-immunoreactive bands, presumably representing endogenous oocyte xCLC5, were located exclusively intracellularly. In pCLC5 cRNA-injected oocytes, moderately strong bands of the same size became apparent in the biotinylated lane, indicating that CLC5 channels were indeed expressed at the cell surface.
Electrophysiological Properties of pCLC5 Expressed in Xenopus Oocytes-The functional properties of pCLC5 were investigated by two-electrode voltage clamp analysis of Xenopus oocytes. Oocytes injected with pCLC5 cRNA consistently exhibited outwardly rectifying currents that were 2 to 3 times greater than those of water-injected control oocytes (Fig. 9). These outward currents were almost completely abolished when most of the bath chloride was replaced with the large anion, methane sulfonate, indicating that the current is likely carried by chloride. The distilbene anion channel blocker, diisothiocyanostilbene-2,2Ј-disulfonic acid, had no significant inhibitory effect. When iodide was substituted for chloride as the predominant bath anion, outward currents in control oocytes markedly increased while those in pCLC5 cRNA-injected oocytes decreased, strongly suggesting that the currents observed in pCLC5-injected oocytes are indeed due to heterologously expressed pCLC5 channels and not simply due to up-regulation of an endogenous oocyte conductance. To characterize further the relative anion conductance of this channel, a panel of different halide and non-halide anions was tested (Fig. 10B). The relative conductance profile for pCLC5 was as follows: nitrate Ͼ Ͼ Cl Ϫ Ϸ Br Ϫ Ͼ I Ϫ Ͼ acetate Ͼ gluconate; for control oocytes the order was quite different: I Ϫ Ͼ Ͼ Br Ϫ Ϸ nitrate Ͼ gluconate Ϸ acetate. Furthermore, the relative anionic conductance of pCLC5 was strikingly similar to that of rat CLC5 (Fig. 10A).
Effect of Extracellular pH on pCLC5 Currents-As CLC5 channels are predominantly located on membranes of intracellular organelles, they are exposed at their non-cytosolic face to a relatively low pH, which may affect its functional properties. To test this, the whole cell currents in oocytes expressing pCLC5 on the plasma membrane were measured as extracellular (topologically equivalent to the non-cytosolic side of intracellular membranes) pH was decreased (Fig. 11). Acidification of the bath reversibly inhibited the outward current of pCLC5 with an apparent pK a of 6.0, similar to that of control oocytes (pK a 6.3), while having minimal effect on the inward conductance. The outward slope conductance-pH relationship was well fit by a Hill equation with a Hill coefficient of 1.1 Ϯ 0.003, suggesting that a single proton-binding site on the extracellular surface may regulate anion entry. DISCUSSION LLC-PK1 cells are a classic model for renal proximal tubule epithelial cells, and therefore an ideal tissue culture system in which to investigate the physiological role of CLC5. Our results show that LLC-PK1 cells express a CLC5 ortholog that resembles CLC5 in other species in terms of primary structure, predicted topology, and electrophysiological properties (40,41). By Northern and Western blot analysis, we find that pCLC5 is expressed quite abundantly in this cell line.
The finding of multiple bands by Northern blotting suggests the possibility of alternative splice variants, alternative poly- FIG. 7. Subcellular localization of CLC5 by double fluorophore labeling. LLC-PK1 cells were transiently transfected with a FLAG epitope-tagged rCLC5 construct. The red fluorescence in all samples represents immunostaining by the M2 anti-FLAG antibody, detected with a Texas Red-conjugated secondary antibody. The green fluorophore was either an entrapped FITC-conjugated marker of endosomal vesicles (DEX, dextran; ALB, albumin; RIC, ricin), immunostaining with antibody to the Golgi marker, ARF1, followed by detection with a FITC-conjugated secondary antibody (ARF), or was omitted (None). Each row represents a series of images from a single confocal field of one slide, acquired from the red (FLAG) and green (Marker) channels, and merged at the same (Merge) or higher (Zoom) magnification. FIG. 8. Heterologous expression of pCLC5 in Xenopus oocytes, detected by immunoblot with C1 antibody. A, immunoblot of total microsomal membrane proteins isolated from oocytes injected with water or pCLC5 cRNA. B, immunoblot of biotinylated surface proteins (Biot) or non-biotinylated intracellular proteins (Ϫ) isolated from oocytes injected with water or pCLC5 cRNA. The yield of protein from the biotinylation and streptavidin purification is shown below each lane (in g per oocyte); from this total, 25 g was loaded in each lane. adenylation sites, or cleavage of the mature mRNA. However, we found no evidence by cloning for alternative splice variants.
Also of interest, the apparent molecular weight of the pCLC5 protein on discontinuous reducing SDS-polyacrylamide gels (65 kDa) is lower than predicted from the sequence for the unmodified protein (83 kDa). We believe this 65-kDa band represents bona fide pCLC5 because we observed the same band with in vitro translation of our pCLC5 and rCLC5 cDNA clones, as well as by immunoblotting of native LLC-PK1 membranes, deglycosylated rat kidney cortex membranes, and oocytes injected with pCLC5 cRNA. Oocytes express an additional ϳ85-kDa band (also not glycosylated) that would be more consistent with the expected size of the native polypeptide. Furthermore, the 65-kDa band is not simply due to anomalous initiation of translation from a downstream start site or to proteolytic degradation of the native protein, because we have also generated N-and C-terminal epitope-tagged CLC5 constructs that migrate only slightly higher (consistent with the addition of the octapeptide tag). Thus, this likely represents aberrant mobility of the protein by gel electrophoresis, as has been reported for several other proteins (3), although not for any other members of the CLC family.
Our findings also indicate that both pCLC5 and rCLC5 can be glycosylated in vitro; rCLC5 is also glycosylated in vivo in kidney cortex, but endogenous pCLC5 in LLC-PK1 cells and heterologously expressed pCLC5 in oocytes are not. The presence, nature, and extent of glycosylation are clearly dependent on the species and cell-type of the host cell (42), and this is the most likely explanation for our failure to find glycosylation of CLC5 in Xenopus oocytes and LLC-PK1 cells.
By heterologous expression in Xenopus oocytes, we demonstrate that pCLC5 encodes an outwardly rectifying chloride , nitrate (Nitr), or acetate (Acet). Outward conductance was calculated from the slope of the I-V plot between ϩ80 and ϩ100 mV, normalized to the mean conductance in chloride buffer (modified ND96), and plotted on a logarithmic scale to confer equal weight to a proportionate increase or decrease in conductance. White columns, water-injected oocytes; black columns: pCLC5injected oocytes; gray columns: rCLC5-injected oocytes.
channel. Four lines of evidence indicate that the currents observed were likely due to a pCLC5-encoded channel and not up-regulation of an endogenous channel. First, the relative anion conductance sequence of pCLC5-injected oocytes was completely different from that of control, water-injected oocytes. Second, the relative anion conductance (Cl Ϫ Ͼ I Ϫ ) and kinetics (rapidly activating, non-inactivating over 500 ms) of the outwardly rectifying current in pCLC5-injected oocytes were very different from that reported in oocytes stimulated by injection with a variety of different cRNA such as pI Cln and CLC6 (43). In the latter case, slowly inactivating currents were observed at positive potentials, with an I Ϫ Ͼ Cl Ϫ conductance preference, and attributed to nonspecific stimulation of an endogenous oocyte channel. Third, the electrophysiological properties of the channel observed in pCLC5-injected oocytes closely resemble those of rCLC5-injected oocytes (Fig. 6) (40); in the latter case, Friedrich et al. (40) confirmed by site-directed mutagenesis that the observed currents were due to heterologously expressed rCLC5 channels. Finally, we show by immunoblotting of oocyte microsomal membrane and surface-biotinylated proteins that CLC5 is overexpressed and present at the plasma membrane of pCLC5-injected oocytes. In these studies, we showed that our C1 antibody also detects the known endogenous oocyte xCLC5 (29,41), which appears to be located exclusively intracellularly in water-injected controls. Thus, these data alone cannot completely exclude the possibility that the observed immunoblot findings in pCLC5 cRNA-injected oocytes were due to stimulation of expression of endogenous xCLC5 and of trafficking of xCLC5 (or perhaps of xCLC5-pCLC5 heteromultimers) to the oocyte plasma membrane. It is important to note that the entire yield of biotinylated protein was only 3-4% of the total protein in the cell lysate. Since an equal number of micrograms of biotinylated and non-biotinylated protein were loaded in each lane, the actual abundance of CLC5 expressed at the surface was very low relative to the amount of intracellular CLC5 in these oocytes. The failure to express substantial amounts of CLC5 at the plasma membrane presumably explains the modest magnitude of the whole cell currents observed by two-electrode voltage clamp.
By immunofluorescence staining of endogenous, or heterologously overexpressed epitope-tagged CLC5 in LLC-PK1 cells, we demonstrate that CLC5 is localized to intracellular vesicles. CLC5 colocalizes with a subpopulation of endosomal vesicles that entrap apical FITC-dextran, but not albumin or ricin, after 15 min at 37°C and are therefore presumed to be early endosomes of the apical membrane fluid-phase endocytosis pathway. This is in agreement with our previous studies demonstrating colocalization of FITC-dextran and CLC5 in endosomal vesicles isolated from rat kidney cortex and analyzed by two-color flow cytometry (11), but contradicts the finding by Devuyst and colleagues (12) of colocalization of CLC5 with FITC-albumin but not with FITC-dextran in the opossum kidney cell line.
Based on the striking phenotype of low molecular weight proteinuria in patients with Dent's disease, one might predict that CLC5 would be expressed in the subset of endosomes that clear low molecular weight proteins, such as ␤ 2 -microglobulin, from the proximal tubule lumen. However, the nature of that endocytotic pathway is poorly understood. Recent evidence indicates that certain low molecular weight proteins of physiological importance, such as vitamin D-binding protein and retinol-binding protein, bind to the apical endocytotic receptor, megalin, with an affinity in the 0.1-1 M range and accumulate in urine in knockout mice lacking megalin (44,45), demonstrating unequivocally that their clearance requires receptor-mediated endocytosis. Conversely, uptake of ␤ 2 -microglobulin by the apical membrane of proximal tubule is saturable only when the protein is present at very high concentrations and exhibits a very low apparent affinity (24, 46 -48). It has been proposed that the nature of this interaction is a fairly nonspecific binding to surface anionic charges on the microvilli (46). Thus, the distinction between receptor-and fluid-phase-mediated endocytosis may be a simplistic one, and different low molecular weight proteins may vary widely in their binding affinity for apical membrane surface proteins or the glycocalyx, and this may determine their ultimate endocytotic fate. The assumption that dextran is taken up by endosomes responsible for clearing substances that have no specific interaction with the plasma membrane (i.e. "fluid-phase" endocytosis) is unproven, and so the most conservative interpretation of our findings is simply that CLC5 is excluded from the subset of vesicles of the receptor-mediated pathway that take up albumin. Interestingly, ricin, a lectin that would be expected to bind fairly promiscuously to surface glycosylated proteins and carbohydrates and has been reported to infiltrate all endosomal compartments (39), did not overlap with the distribution of CLC5 and endocytosed dextran, indicating that ricin is a selective marker of an as yet poorly defined subset of the endosomal pathway. FIG. 11. Effect of extracellular pH on whole cell currents in Xenopus oocytes. A, current traces recorded in a single control oocyte and a pCLC5-expressing oocyte as the bath solution pH was progressively reduced from 7.5 to 4.5 by sequential solution changes and then returned to pH 7.5. B, relative outward slope conductance, normalized to the mean conductance at pH 7.5. Curves were fit by non-linear regression to Michaelis-Menten equations with pK a of 6.3 (water-injected) and 6.0 (pCLC5-injected). White circles, water-injected oocytes; black circles, pCLC5-injected oocytes.
We also demonstrate the novel finding of expression of CLC5 in some cells in a juxtanuclear density that corresponds to the Golgi complex, where it colocalizes with the ADP-ribosylation factor, ARF1. Although CLC5 has not previously been found in the exocytotic pathway, the lumen of the Golgi complex is known to be acidified by an electrogenic vacuolar proton pump (49 -51), and this may be important in the post-translational modification (52) and targeting (53) of proteins in the secretory pathway, as well as their retrieval from individual subcompartments (54). It has been proposed that the electrogenic acidification of the Golgi may be limited by the membrane potential and facilitated by the presence of a chloride conductance that dissipates the transmembrane electrical gradient (50,55). Thus, the function of CLC5 may be to provide the counterion conductance required for adequate acidification of the Golgi lumen, analogous to its postulated role in endosomes. However, the existence of such a possible mechanism is highly controversial; for example, a recent study in Vero monkey kidney cells found that Golgi acidification does not reach thermodynamic equilibrium and is determined primarily by a proton leak pathway (51).
Finally, we found that overexpression of CLC5 in transfected cells could also alter their morphology. We observed several cells that were adherent but rounded up and packed densely with prominent vesicles. This may simply have been due to a nonspecific toxic effect of CLC5 overexpression, but the fact that the cells looked viable and remained adherent to their substratum suggests otherwise. Alternatively, this could represent a specific disruption of intercellular interactions such as adhesion or junction formation and of normal vesicular function and trafficking.
In conclusion, our findings indicate that CLC5 is localized in endosomal and Golgi vesicles, consistent with a possible role in acidification of compartments of the endocytic and exocytic pathways, and suggest that its function may be important for normal intercellular adhesion and vesicular trafficking. The LLC-PK1 cell line should provide a powerful model in which to elucidate the role of CLC5 in these cellular processes.