ClC-3 Chloride Channels Facilitate Endosomal Acidification and Chloride Accumulation*

We investigated the involvement of ClC-3 chloride channels in endosomal acidification by measurement of endosomal pH and chloride concentration [Cl–] in control versus ClC-3-deficient hepatocytes and in control versus ClC-3-transfected Chinese hamster ovary cells. Endosomes were labeled with pH or [Cl–]-sensing fluorescent transferrin (Tf), which targets to early/recycling endosomes, or α2-macroglobulin (α2M), which targets to late endosomes. In pulse label-chase experiments, [Cl–] was 19 mm just after internalization in α2M-labeled endosomes in primary cultures of hepatocytes from wild-type mice, increasing to 58 mm over 45 min, whereas pH decreased from 7.1 to 5.4. Endosomal acidification and [Cl–] accumulation were significantly impaired in hepatocytes from ClC-3 knock-out mice, with [Cl–] increasing from 16 to 43 mm and pH decreasing from 7.1 to 6.0. Acidification and Cl– accumulation were blocked by bafilomycin. In Tf-labeled endosomes, [Cl–] was 46 mm in wild-type versus 35 mm in ClC-3-deficient hepatocytes at 15 min after internalization, with corresponding pH of 6.1 versus 6.5. Approximately 4-fold increased Cl– conductance was found in α2M-labeled endosomes isolated from hepatocytes of wild-type versus ClC-3 null mice. In contrast, Golgi acidification was not impaired in ClC-3-deficient hepatocytes. In transfected Chinese hamster ovary cells expressing ClC-3A, endosomal acidification and [Cl–] accumulation were enhanced. [Cl–] in α2M-labeled endosomes was 42 mm (control) versus 53 mm (ClC-3A) at 45 min, with corresponding pH 5.8 versus 5.2; [Cl–] in Tf-labeled endosomes at 15 min was 37 mm (control) versus 49 mm (ClC-3A) with pH 6.3 versus 5.9. Our results provide direct evidence for involvement of ClC-3 in endosomal acidification by Cl– shunting of the interior-positive membrane potential created by the vacuolar H+ pump.

Endosomal acidification is involved in a variety of cellular processes, including receptor and ligand sorting, vesicular fusion and budding, and protein degradation (1)(2)(3). Acidification is an active process in which H ϩ is pumped into the endosome lumen by a bafilomycin-sensitive vacuolar H ϩ ATPase. Charge balance requires the accompaniment of inward H ϩ movement by outward movement of cations such as K ϩ and/or inward movement of anions such as Cl Ϫ . Ion substitution, inhibitor, and endosomal pH/Cl Ϫ measurements suggest that Cl Ϫ is the principal ion in most cells responsible for shunting the interiorpositive endosomal potential produced by active H ϩ entry (4 -10). In addition to ion conductances and pump rates, endosomal acidification depends on the magnitude of H ϩ leak, the buffer capacity and density of fixed charges in the endosome lumen, the pH and ionic content of cytoplasm, and the rate and nature of endosome fusion/budding events.
It has been proposed without direct evidence that ClC-type Cl Ϫ channels are responsible for some or all endosomal Cl Ϫ conductance in many cell types, and hence they may be a major determinant of endosomal acidification. There is evidence for expression of at least five members of the ClC Cl Ϫ channel family (ClCs 3-7) in intracellular vesicles, including endosomes (reviewed in Ref. 11). ClC-3 is localized in synaptic vesicles in neurons, where it is thought to provide an electric shunt for the proton pump (12). Studies in transfected cells suggest that a shorter ClC-3 isoform (ClC-3A) is localized in late endosomes and lysosomes, where it may regulate organellar pH (13,14). The function of a Golgi-localized longer ClC-3 isoform (ClC-3B) is not known (14,15). ClC-5 is localized primarily in early endosomes in renal proximal tubules, where it appears to be involved in receptor-mediated endocytosis though its function as an electrical shunt has not yet been shown (16). A recent study reported ClC-4 expression in endosomal membranes and a possible role in endosomal acidification and trafficking in epithelial cells (17). Human mutations and/or knock-out mouse studies have implicated the involvement of ClC-3 in hippocampal degeneration (12,18), ClC-5 in renal stone formation in Dent's disease (19,20), and ClC-7 in osteopetrosis and retinal degeneration (21). A variety of mechanisms have been proposed by which defective endosomal Cl Ϫ conductance and acidification could produce these defects.
The purpose of this study was to define the role of ClC-3 Cl Ϫ channels in endosomal acidification and Cl Ϫ accumulation. The kinetics of endosomal pH and [Cl Ϫ ] were measured in intact cells using two models: primary cultures of hepatocytes from wild-type and ClC-3 knock-out mice, and Chinese hamster ovary (CHO) 1 cells after transfection with control or ClC-3A-encoding cDNA. In addition, measurements were made in Golgi to examine organelle-specific effects in ClC function. Hepatocytes were chosen as a readily cultured cell type that natively expresses ClC-3 (22,23). CHO cells were chosen as an easily transfected cell type that expresses little or no ClC-3 and in which endosomal acidification and Cl Ϫ accumulation have been studied extensively (9, 24 -27). The methodology for measurement of endosomal [Cl Ϫ ] by ratio imaging fluorescence microscopy was established and validated recently by the Verkman laboratory. Cells are imaged after pulse label-chase with fluorescent ligands (transferrin, Tf, or ␣ 2 -macroglobulin, ␣ 2 M) containing Cl Ϫ -sensitive and -insensitive chromophores (9). We report here that ClC-3 deletion in mice remarkably impairs endosomal acidification and [Cl Ϫ ] accumulation in hepatocytes and that transfection of ClC-3A into CHO cells enhances endosomal acidification and [Cl Ϫ ] accumulation.

EXPERIMENTAL PROCEDURES
Hepatocyte Cell Culture from Wild-type and ClC-3-deficient Mice-Wild-type and ClC-3 knock-out mice on a 129SV and C57BL/6 mixed genetic background were generated by targeted gene disruption as described previously (18). The ClC-3 knock-out mice were smaller than wild-type litter mates and had a higher mortality at 3-4 weeks. Hepatocytes were isolated by a modification of the two-step liver perfusion method (28). After anesthesia, livers were perfused with liver perfusion medium (35 ml over 10 min) and then liver digest medium (35 ml over 10 min) (Invitrogen). Isolated hepatocytes were filtered, washed, and suspended in Williams' medium E (Invitrogen) containing 5% fetal bovine serum and 2 mM glutamine. Cells were plated on 18-mm-diameter round coverslips that were coated with rat tail collagen (20 g/cm 2 ; BD Biosciences) and placed in a 5% CO 2 -95% air incubator at 37°C. The medium was replaced with HepatoZYME-SFM (Invitrogen) after 3 h of culture. Measurements of pH and [Cl Ϫ ] were done after 2-3 days of culture, at which time cells were nearly confluent.
Construction of cDNA Encoding Mouse CLC-3-mRNA was isolated from mouse kidney (Oligo Direct mRNA kit; Qiagen) and cDNA was reverse transcribed using oligo(dT) (SuperScript II preamplification kit; Invitrogen). PCR amplification was performed with the cDNA template and primers designed from the mouse ClC-3A sequence (GenBank TM accession number NM_007711). Primers were engineered with BglII and XbaI restriction sites: sense, 5Ј-GAAGATCTCAATGACAAATGG-AGGCAGCATTAATAGC-3Ј; antisense, 5Ј-GCTCTAGAGTTGAACATT-ATTGAAGCGGGGTC-3Ј. The PCR product was subcloned into plasmid pcDNA6/Myc-His (Invitrogen), which introduces a c-Myc tag at the NH 2 terminus. The full-length insert was confirmed by sequence analysis.
CHO Cell Culture and Transfection-CHO-K1 cells (ATCC number CCL-61) were grown in Ham's F12K medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were cultured on 18-mm-diameter round coverslips at 37°C in 95% air-5% CO 2 and used just prior to confluence.
Endosome Labeling and Kinetics of Endosomal [Cl Ϫ ] and pH-Cells were incubated in serum-free medium for 15 min at 37°C before experiments. Labeling of cell surface receptors was done by incubation with Cl Ϫ or pH sensors (300 nM for Tf, 100 nM for ␣ 2 M) for 20 min in PBS (containing 1 mM CaCl 2 and MgCl 2 ) at 4°C. Coverslips were then washed twice with ice-cold PBS containing 1% bovine serum albumin and transferred to a precooled perfusion chamber containing ice-cold perfusate. Sets of BAC and TMR images (for Cl Ϫ ) or FITC and TMR images (for pH) were acquired at specified times after rapid warming by perfusion at 37°C. In some experiments, the perfusate contained bafilomycin A 1 (200 nM).
Golgi Compartment pH-Cells were incubated in serum-free medium for 30 min at 37°C, washed three times with ice-cold PBS, and incubated with the Golgi pH indicator (200 nM) for 30 min in PBS at 4°C. Coverslips were washed three times with ice-cold PBS and maintained at 37°C for 45 min for Golgi targeting. CF and TMR images were acquired for pH determination.
Fluorescence Microscopy and Data Analysis-Cells were imaged using a Leitz upright epifluorescence microscope equipped with a Nipkow-wheel confocal attachment and 14-bit cooled charge-coupled device camera (24). Fluorescence was collected using a ϫ100 oil immersion objective (Nikon plan-apo, numerical aperture 1.4) and appropriate filter sets for BAC (excitation 470 Ϯ 5 nm, dichroic 505 nm, emission 535 Ϯ 20 nm), TMR, and FITC. Image analysis for computation of endosomal/Golgi [Cl Ϫ ] and pH was done using custom software to compute area-integrated background-subtracted pixel intensities (9,24).
Functional Analysis of Cl Ϫ Transport in Endosomes-Hepatocytes were freshly isolated as described above and washed three times in PBS. Cells were incubated with the ␣ 2 M Cl Ϫ indicator (100 nM) for 45 min in PBS at 37°C, washed three times, and then homogenized in ice-cold low Cl Ϫ buffer (120 mM KNO 3 , 20 mM NaNO 3 , 1 mM CaCl 2 , 1 mM MgCl 2 , and 10 mM HEPES, pH 7.4). Homogenates were centrifuged at 3,500 ϫ g for 10 min at 4°C to remove nuclei, mitochondria, and debris. The supernatant containing fluorescently labeled endosomes was preincubated in low Cl Ϫ buffer in the absence or presence of valinomycin (10 M) for at least 10 min at 37°C. To drive Cl Ϫ influx, an isosmolar high Cl Ϫ buffer (120 mM KCl, 20 mM NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 10 mM HEPES, pH 7.4) was added to the stirred vesicle suspension to give 60 mM final [Cl Ϫ ]. BAC fluorescence in the vesicle suspension was measured continuously using a Fluoromax-3 fluorimeter (excitation 470 Ϯ 10 nm, emission 535 Ϯ 10 nm).

Impaired Acidification and Cl
Ϫ Accumulation in ClC-3-deficient Hepatocytes-Measurements were made in primary cultures of hepatocytes generated from wild-type and ClC-3 null mice. Fig. 1A shows a similar appearance of control and ClC-3-deficient hepatocytes by light microscopy. Fig. 1B shows colocalization of structures labeled with FITC-conjugated Tf (top) and ␣ 2 M (bottom) with markers of early (EEA) and late (Rab7) endosomes, respectively. Fig. 1C shows fluorescence micrographs of hepatocytes after labeling with BAC-dextran-␣ 2 M-TMR and BAC-dextran-Tf-TMR at low temperature, followed by chase for 15 min (Tf) or 45 min (␣ 2 M) at 37°C. A typical endosomal pattern was seen in the control and ClC-3-deficient hepatocytes, as found for internalization of Tf and ␣ 2 M in many cell types (2,9,27,30,31). A membrane pattern was observed just after labeling, with progressive maturation over time (not shown). Cellular fluorescence was reduced to near zero when up to 100-fold excess unlabeled Tf or ␣ 2 M was included in the labeling solution, confirming a receptor-mediated internalization mechanism. Fig. 2 summarizes the kinetics of endosomal acidification and Cl Ϫ accumulation in control and ClC-3-deficient hepatocytes. Initial pH in Tf-labeled endosomes was ϳ7.2, decreasing over 15 min to 6.1 in control and 6.5 in ClC-3-deficient hepatocytes ( Fig. 2A, left). Endosomal [Cl Ϫ ] increased in parallel from an initial low value of 16 to 46 mM in control and 35 mM in ClC-3-deficient hepatocytes ( Fig. 2A, right). As reported for CHO cells (9), the low [Cl Ϫ ] early after endocytosis is probably the consequence of a Donnan potential produced by surface negative charges that excludes Cl Ϫ during endosome formation. Endosomal acidification and Cl Ϫ accumulation were blocked by inclusion of the H ϩ pump inhibitor bafilomycin in the perfusate. Fig. 2B shows that pH in ␣ 2 Mlabeled endosomes at 45 min was reduced to 5.4 in control and 6.0 in ClC-3-deficient hepatocytes. Corresponding [Cl Ϫ ] was increased to 58 mM in control and 43 mM in ClC-3deficient hepatocytes. As found for early/recycling Tf-labeled endosomes, acidification and Cl Ϫ accumulation were bafilomycin-sensitive and significantly impaired in ClC-3 deficiency.
Golgi Acidification Is Not Altered by ClC-3A Expression-Measurements of Golgi pH were done in hepatocytes from wild-type and ClC-3 null mice. As reported in other cell types (9, 32), the Golgi lumen was labeled with fluorescent CTb, which is targeted to the Golgi by a retrograde transport mechanism. Fig. 3A shows overlapping localization of FITC-labeled CTb with three Golgi protein markers: giantin for the cytoplasmic face of Golgi (33), GM130 for cis-Golgi (34), and boron dipyrromethene difluoride TR-ceramide, a red fluorescent Golgi-specific probe (35). The CTb fluorescently labeled compartment probably represents mainly trans-Golgi, though there may be some cis-and medial-Golgi labeling. Fig. 3B (top) shows micrographs of hepatocytes after Golgi compartment staining, with CF (green) and TMR (red) fluorescence at 45 min after perfusion. Golgi pH, as determined by ratio imaging of the pH-sensitive CF chromophore and the pH-insensitive TMR chromophore, was not different in the ClC-3-deficient hepatocytes (Fig. 3B, bottom).
Enhanced Acidification and Cl Ϫ Accumulation in ClC-3Atransfected CHO Cells-RT-PCR and immunostaining analysis showed the absence of ClC-3 transcript and protein in CHO cells. To determine the influence of expressed ClC-3A on endosomal acidification and Cl Ϫ accumulation, CHO cells were transfected with cDNA encoding c-Myc epitope-tagged mouse ClC-3A. Control cells were transfected with cDNA encoding c-Myc alone. Immunofluorescence with c-Myc and ClC-3 antibodies showed ClC-3A expression in the transfected CHO cells in an endosomal pattern (Fig. 4A). A diffuse cytoplasmic/nuclear pattern was seen for c-Myc expression in control cells. Immunoblot analysis showed the expected molecular size of 84 kDa for c-Myc-tagged ClC-3A in the transfected cells (Fig. 4A,  bottom). dextran-Tf-TMR and BAC-dextran-␣ 2 M-TMR were characteristic of the ligands in early and late endosomes, respectively, as also seen for FITC and TMR labeled with Tf and ␣ 2 M (not shown). Fig. 5 summarizes the kinetics of endosomal pH and [Cl Ϫ ] in control and ClC-3A-expressing CHO cells. Transfection with ClC-3A produced significant enhancements of endosomal acidification and Cl Ϫ accumulation in Tf-labeled early/recycling endosomes compared with control cells that were transfected with plasmid containing only the c-Myc tag (Fig. 5A). Endosomal H ϩ and Cl Ϫ entry were coupled in an ϳ1:1 molar ratio as deduced using the measured endosomal buffer capacity of 28 mM H ϩ /pH unit (9); for example, the pH drop of 0.72 units at 15 min in control endosomes corresponds to 20 mM H ϩ entry, similar to the increase in Cl Ϫ of 19 mM at 15 min. In ␣ 2 M-labeled endosomes, ClC-3 expression produced significantly enhanced endosomal acidification (pH 5.8 Ϯ 0.1 versus 5.2 Ϯ 0.2) and Cl Ϫ accumulation (42 Ϯ 3 versus 53 Ϯ 3 mM) (Fig. 5B).
Functional Analysis of Cl Ϫ Transport in Endosomes-The involvement of ClC-3 Cl Ϫ channels in endosomal acidification and Cl Ϫ accumulation by an electrical shunt mechanism requires functional expression of ClC-3 in endosomes. We developed a strategy to measure Cl Ϫ transport directly in fluorescently labeled endosomes contained in a cellular homogenate. Endosomes were labeled with a BAC-dextran conjugate of ␣ 2 M as done in intact cell studies. After incubation at 37°C for 45 min, cells were homogenized, centrifuged, and suspended in a low Cl Ϫ , high K ϩ buffer containing the K ϩ ionophore valinomycin to make Cl Ϫ influx (rather than counterion transport) rate-limiting. Cl Ϫ influx was measured selectively in fluorescently labeled endosomes in the heterogeneous membrane suspension from the kinetics of decreasing BAC fluorescence in response to mixing of the membrane suspension with an isosmolar high Cl Ϫ buffer. Fig. 6A shows the kinetics of Cl Ϫ influx measured in endosomes from wild-type and ClC-3-deficient hepatocytes. There was little decrease in fluorescence corresponding to Cl Ϫ influx in the absence of valinomycin, indicating a conductive Cl Ϫ entry mechanism. ClC-3 deletion remarkably reduced Cl Ϫ influx as seen from the slower decrease in fluorescence signal. Fig. 6B summarizes average Cl Ϫ influx rates shown as reciprocal exponential time constants fitted to the kinetics of BAC fluorescence. Cl Ϫ influx was significantly reduced by ϳ4-fold in endosomes of ClC-3-deficient hepatocytes, although significant residual influx remained. This result indicates that ClC-3 is the main Cl Ϫ conductive mechanism in the labeled endosomes. The non-zero Cl Ϫ conductance in endosomes from ClC-3-deficient hepatocytes accounts in part for their ability to acidify, albeit less well than hepatocytes from wild-type mice. DISCUSSION Our results provide direct evidence supporting the involvement of an intracellular ClC-type Cl Ϫ channel in organellar acidification. Endosomal acidification and Cl Ϫ accumulation were significantly reduced in hepatocytes from ClC-3-deficient versus wild-type mice and enhanced in ClC-3A-transfected versus control CHO cells. Functional ClC-3 Cl Ϫ conductance in endosomes was demonstrated in cell homogenates from wildtype versus ClC-3-deficient hepatocytes containing fluorescently labeled endosomes. The data suggest that ClC-3 Cl Ϫ conductance provides a quantitatively significant electrical shunt pathway to permit endosomal acidification by the vacuolar H ϩ pump.
ClC-3 is expressed in brain, liver, and other tissues (22) and has been reported to localize in intracellular vesicles (12) and the plasma membrane (36,37). The phenotype of ClC-3 knockout mice includes blindness and hippocampal neurodegeneration (12,18). Indirect measurements of ATP-induced acidification by acridine orange uptake/quenching showed reduced uptake in isolated synaptic vesicles of ClC-3 versus wild-type mice (12), and estimated pH in isolated liver vesicles from ClC-3 null mice in the presence of ATP was greater than that of vesicles from wild-type mice (18). However, other reports of ClC-3A and ClC-3B localization (13,14) and possible function as a swelling-activated Cl Ϫ channel (12, 36 -38) draw somewhat contradictory conclusions. Here, we have shown that ClC-3 functions as an intracellular ClC-type Cl Ϫ channel that facilitates endosomal acidification and Cl Ϫ accumulation in intact cells, providing a possible explanation for the phenotypic abnormalities in ClC-3 null mice. Our conclusion about the role of ClC-3 in endosomal acidification in liver does not necessarily apply to other organs/cell types where different endosomal ion channels may be expressed. For example, ClC-5 may be important in endosomal acidification and Cl Ϫ conductance in kidney proximal tubule cells where impaired endocytosis has been demonstrated in ClC-5 null mice (19) and may provide a mechanism for Dent's disease in ClC-5 deficiency in humans.
The partial impairment of endosomal acidification in ClC-3deficient hepatocytes suggests that some, but not all, ion movement accompanying active H ϩ entry involves ClC-3-mediated Cl Ϫ entry. The results in Fig. 6 suggest that ClC-3 provides the principal route for conductance of Cl Ϫ entry. It is difficult to quantify the fraction of ionic movement through ClC-3 versus other channel pathways because of the complex set of determinants of endosomal acidification as described in the Introduction. Endosomes from hepatocytes may express other types of Cl Ϫ channels, such as ClC-4 and/or K ϩ channels, that may be up-regulated as a compensatory response to ClC-3 gene deletion in mice. Also, endosomes in intact cells may differ from those in suspension in that regulatory factors may be different and little endosome fusion occurs. Notwithstanding the difficulties in ascribing the exact fraction of total ionic movement accompanying H ϩ entry, our results here support the involvement of ClC-3 in endosomal acidification.
The CHO cell transfection studies showed that overexpression of ClC-3A in a cell type that does not normally express ClC-3 produced an enhancement in endosomal acidification and Cl Ϫ accumulation. Therefore, endogenous endosomal ion conductance (rather than the H ϩ pump) must be a rate-limiting determinant of endosomal acidification. Thus, regulation of the expression and/or function of ClC-3 or other endosomal ion channels can influence acidification kinetics and steady-state pH. Because Cl Ϫ entry produces an osmotic gradient leading to endosome swelling (24), whereas K ϩ exit accompanying H ϩ influx would produce endosome shrinkage, regulation of endosomal Cl Ϫ or K ϩ conductances is predicted to alter endosome volume. Based on conclusions from our recent study on the role of endosomal Cl Ϫ accumulation and swelling on transgene delivery (25), the ability of overexpression of a ClC-type Cl Ϫ channel to enhance acidification and Cl Ϫ entry provides a potential new tool to enhance gene delivery by non-viral mechanisms.
In summary, measurements of endosomal pH and [Cl Ϫ ] provide direct evidence for the involvement of a ClC-type Cl Ϫ channel in endosomal acidification. The approach used here should be applicable to investigation of the role of other putative intracellular Cl Ϫ channels, such as ClC-4, ClC-6, ClC-7, and AQP6, in organellar acidification. When available, specific high affinity inhibitors of ClC-type Cl Ϫ channels will be useful to resolve the contributions of individual Cl Ϫ channels because Hepatocytes were incubated with BAC-dextran-␣ 2 M-TMR at 37°C for 45 min and homogenized, and an endosome suspension was prepared in high K ϩ /low Cl Ϫ buffer as described under "Experimental Procedures." A, time course of BAC fluorescence after rapid mixing of the endosome suspension with an isosmolar buffer containing 60 mM Cl Ϫ showing reduced fluorescence corresponding to Cl Ϫ influx. Where indicated, valinomycin (10 M) was present in the endosome suspension. B, averaged initial Cl Ϫ influx rates computed from the kinetics of decreasing BAC fluorescence. Reciprocal exponential time constants ( Ϫ1 ) fitted to fluorescence data for individual measurements (filled circles) shown along with means Ϯ S.E. (open circles). *, p Ͻ0.01. quantitative conclusions cannot be rigorously made using cells from knock-out mice or transfected systems.