The Cl– channel ClC-2 is expressed in transporting epithelia and has been proposed as an alternative route for Cl– efflux that might compensate for the malfunction of CFTR in cystic fibrosis. There is controversy concerning the cellular and membrane location of ClC-2, particularly in intestinal tissue. The aim of this paper is to resolve this controversy by immunolocalization studies using tissues from ClC-2 knockout animals as control, ascertaining the sorting of ClC-2 in model epithelial cells and exploring the possible molecular signals involved in ClC-2 targeting. ClC-2 was exclusively localized at the basolateral membranes of surface colonic cells or villus duodenal enterocytes. ClC-2 was sorted to the basolateral membranes in MDCK, Caco-2 and LLC-PK1-μ1B, but not in LLC-PK1-μ1A cells. Mutating a di-leucine motif (L812L813) to a di-alanine changed the basolateral targeting of ClC-2 to an apical location. The basolateral membrane localization of ClC-2 in absorptive cells of the duodenum and the colon is compatible with an absorptive function for this Cl– channel. Basolateral targeting information is contained in a di-leucine motif (L812L813) within CBS-2 domain at the C-terminus of ClC-2. It is speculated that ClC-2 also contains an apical sorting signal masked by L812L813. The proposal that CBS domains in ClC channels might behave as regulatory sites sensing intracellular signals opens an opportunity for pharmacological modulation of ClC-2 targeting.
Cystic fibrosis (CF) is an autosomal recessive disease that affects the epithelium of the gastrointestinal tract, sweat ducts, respiratory tract and the male reproductive system. The disease is caused by more than a thousand different mutations in the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR), a plasma membrane protein. Mutations affect the trafficking or various aspects of the function of CFTR, which has been found to be a Cl– selective channel (Riordan et al., 1989; Sheppard and Welsh, 1999) (see www.genet.sickkids.on.ca/cftr for current mutation list). CFTR is localized to the apical membranes of secretory cells where it underlies a highly regulated permeability to Cl–. In the small and large intestine, Cl– secretion mediated by CFTR is thought to be the driving force for fluid secretion. The secretory process is triggered by increases in cAMP (or cGMP), which enhance Cl– efflux from the cells mediated by the CFTR Cl– channel and is accompanied by Na+ and water efflux. CFTR is, therefore, central in a process of fluid secretion that will clear mucus and bacteria from external epithelial surfaces and its lack of function is thought to be the cause of mucus obstruction in CF-affected organs.
It has long been reasoned that if alternative Cl– efflux pathways existed that could be pharmacologically activated in CF epithelia, this might help alleviate the secretion-related aspects of the disease. Mouse airway and pancreas possess Ca2+-activated secretion independent of CFTR (Anderson and Welsh, 1991; Clarke et al., 1994), but the molecular identity of the Cl– channels involved has remained elusive. The participation of Ca2+-activated Cl– channels in intestinal secretion is controversial (Strabel and Diener, 1995).
Work with CF airway cells led to the proposal that ClC-2, one of four plasma membrane ClC Cl– channels that occur in mammals (Thiemann et al., 1992), could substitute for CFTR (Schwiebert et al., 1998). More recently, it has been proposed that ClC-2 can mediate Cl– secretion in the intestine (Gyömörey et al., 2000). The authors presented evidence for apical and tight-junction localization of ClC-2 in the mouse small intestine. An apical localization and function in fluid secretion is also postulated for ClC-2 in lung epithelium (Blaisdell et al., 2000). In contrast with these results, we and others have shown ClC-2 transcript in surface epithelium of guinea pig distal colon, identified a basolateral membrane localization of ClC-2 in surface epithelium and demonstrated a functional Cl– conductance with rectification and selectivity properties consistent with this subcellular distribution of ClC-2 (Catalán et al., 2002; Lipecka et al., 2002; Catalán et al., 2004). ClC-2 knockout mouse models present blindness and male infertility and, it has been argued, transepithelial transport defects could be responsible for these alterations (Bösl et al., 2001; Nehrke et al., 2002). Dual ClC-2 and CFTR knockout mice do not exhibit a worsening of the CF phenotype, suggesting indirectly that ClC-2 does not contribute to secretion and might, therefore, not be located apically (Zdebik et al., 2004).
Absorption and secretion in colon and small intestine epithelium are believed to occur in different cellular compartments. Surface epithelium of the colon, or the villi in the small intestine, are mainly absorptive while the crypts are the site of secretion (Welsh et al., 1982; Kunzelmann and Mall, 2002). This compartmentalization is compatible with the distribution of the membrane transport systems involved in secretion. In particular, CFTR, is solely expressed in the crypt epithelium (Trezise et al., 1992). Therefore, if ClC-2 is expressed in villi of small intestine and surface epithelium in the colon (Catalán et al., 2002; Lipecka et al., 2002; Catalán et al., 2004), and if involved in transepithelial transport, it would appear that it should participate in an absorptive rather than a secretory process.
The dynamic segregation of proteins to morphologically and functionally separate basolateral and apical domains in epithelia is dependent upon sorting mechanisms operating in the biosynthetic and recycling pathways. These mechanisms recognize specific sorting signals contained in the structure of the membrane proteins (Yeaman et al., 1999). If ClC-2 is a basolateral membrane protein, it should carry signals in its sequence to permit correct targeting in epithelial cells grown in vitro.
Consideration of whether ClC-2 can substitute for CFTR in CF intestine critically depends on the cellular and subcellular location of the channel. In the present work we aim at resolving the controversy about the location of ClC-2 in the intestine, by improving the quality of immunolocalization and taking advantage of the availability of ClC-2 null mice to provide a strict parallel negative control. Our data show conclusively that ClC-2 has a basolateral location in absorptive cells in large and small intestine. In addition, we demonstrate basolateral sorting in vitro for ClC-2 in epithelial cell lines, which is dependent upon the AP-1B clathrin-adaptor complex and is encoded in a C-terminus CBS-2 di-leucine motif.
Materials and Methods
Tissue samples taken from both wild type and knockout mice were fixed by immersion in Bouin's fluid (75 ml of saturated picric acid, 25 ml of 40% formaldehyde and 5 ml of glacial acetic acid) for 48 hours at room temperature. Fixed samples were dehydrated in ethanol and embedded in Histosec (Merck, Darmstadt, Germany).
Sections of 5 μm thickness (3 μm for consecutive sections), were dewaxed and hydrated before immunostaining. Pseudoperoxidase activity was eliminated by treatment with absolute methanol and 10% hydrogen peroxide. Before immunostaining, all tissues were microwaved with 10 mM sodium citrate pH 6.0 or 50 mM Tris-HCl pH 10.0 for 5 minutes at 90°C at a maximum power of 1150 W. Twenty minutes after treatment, sections were washed with three changes of 50 mM Tris-HCl pH 7.8 and then incubated overnight with various dilutions of the following antibodies: (1) an anti-peptide antibody directed to residues 888-906 of rat ClC-2 (1:100 to 1:500). This antibody recognizes the C-terminus epitope RSRHGLPREGTPSDSDDKC of rat ClC-2 that is identical in the mouse molecule (ACL-002, Alomone Labs, Jerusalem, Israel); (2) an antibody to NKCC1 (Kurihara et al., 1999) (1:1000 to 1:5000), kindly provided by R. James Turner; or (3) an antibody to α2 subunit of H+/K+ ATPase (Codina et al., 1998) (1:1000 to 1:2500), kindly given by Juan Codina. Anti-ClC-2 antibodies initially tested also included sc-16429 and sc-16430 (Santa Cruz Biotechnology, Santa Cruz CA, USA, 1:100 to 1:2000 dilutions). Dilutions were performed in 50 mM Tris-HCl pH 7.8 containing 1% immunoglobulin-free bovine serum albumin (Sigma Aldrich, St Louis MO, USA). Bound immunoglobulins were detected with the LSAB+ biotin/streptavidin-peroxidase kit (Dako, Carpinteria, CA, USA) and peroxidase was visualized by incubation with 3-amino-9-ethylcarbazol (AEC, Dako) for 20 minutes or 3,3′-diaminobenzidine (DAB, Dako) for 5 minutes. Controls were carried out using tissues prepared under identical conditions from knockout mice. Other controls included the omission of primary antibody or its replacement by non-immune serum of same origin and at the same dilution (Catalán et al., 2002).
For double immunolabeling, sections were incubated first with anti-NKCC1 antibody following the procedure described above and peroxidase was developed using 3,3′-diaminobenzidine. Following completion of the first labeling reaction, the sections were rigorously washed in tap water, then distilled water, and the brownish color darkened using the silver methenamine technique (Rodríguez et al., 1984). Next, sections were washed with distilled water and 50 mM Tris-HCl pH 7.8 before incubation with anti-ClC-2 antibody. Peroxidase was developed with 3-amino-9-ethylcarbazol that produced an intense red color. Finally, sections were contrasted with hematoxylin and mounted with the aqueous mounting medium Mowiol™ (Polysciences Inc., Warrington, PA, USA).
Cell culture and transfection
MDCK type II cells were grown in Dulbecco's modified Eagles medium (DMEM) containing 10% fetal bovine serum (FBS), 1% penicillin-streptomycin and 1% fungizone. Caco-2 cells were grown in α-modified Eagles medium α-MEM with 10% FBS, 1% penicillin-streptomycin and 1% fungizone. LLCPK1-μ1A and LLCPK1-μ1B were generously supplied by Ira Mellman and were grown in α-MEM containing 10% FBS and 1.8 mg ml–1 geneticin (G418). All cells were cultured in 100 mm plastic dishes, and were subcultured at 90% confluence. Cells were maintained at 37°C in 5% CO2 in air.
For transfection cells were grown upon glass coverslips (four in a 35 mm plate) or 12 mm diameter permeable polyester membranes with 0.4 μm-pores (Costar, Cambridge MA, USA). For all cell lines, transfection was done 2-3 days after confluence had been reached, except for Caco-2 cells, which were grown for 10 days prior to transfection. Transient transfection in cells grown on permeable support, was carried out using lipofectamine 2000 (Invitrogen, Carlsbad CA, USA) and 2 μg plasmid DNA added to the apical side of the monolayer during a 5 hours incubation. Cultures were analyzed 24 hours post-transfection. A similar transfection protocol was followed for cells grown on glass coverslips.
The plasmid ClC-2-GFP used was a human ClC-2 fusion protein with the green fluorescent protein (pEGFP-N1), which has been characterized to show normal functional expression (Niemeyer et al., 2004). Alternatively, a human ClC-2-HA plasmid in the pCR3.1 vector, was modified to carry the hemagglutinin epitope in an extracellular loop between α-helices L and M (see Fig. 6). This construct also presents normal functional expression (I.C., unpublished data). Mutants (Table 1) were obtained by PCR using the overlap extension method (Ho et al., 1989) and verified by sequencing. Plasmid for c-myc-tagged hSlo (Bravo-Zehnder et al., 2000) was a gift from Patricio Orio.
|Residues .||Consensus .||Mutant .||Location .|
|Residues .||Consensus .||Mutant .||Location .|
Notice that this residue 17 is given as H in the GenBank sequence for human ClC-2 (NP004357), which is a polymorphism, being Y in the clone used here (Niemeyer et al., 2004) and in the highly conserved mouse, rat and guinea-pig orthologs
Mutant used was done on the guinea-pig ortholog where its numbering corresponds to 16-61 (Varela et al., 2002). Φ, bulky or hydrophobic residue. BL, basolateral membrane location. Ap, apical membrane location. IR, intracellular retention
For microscope observation, cells grown on glass coverslips were fixed in 4% paraformaldehyde for 30 minutes. Coverslips were mounted on glass slides using a fluorescent mounting medium (Dako). A similar fixation approach was used for filter-grown cells, except that the filters were cut-off to be mounted on glass slides. Cells were permeabilized with 0.2% Triton X-100 in PBS buffer for 10 minutes. After washing and gelatin blocking, cells were incubated with primary antibody for 1 hour. Antibodies used were anti-ZO-1 (Zymed, San Francisco CA, USA) 12 μg ml–1, anti-hemagglutinin epitope (clone 3F10, Roche, Penzberg, Germany) at 0.2 μg ml–1 and anti α-myc epitope 12 μg ml–1, kindly given by Marcela Bravo-Zehnder. After washing, Alexa Fluor 488- or 568-labeled secondary antibodies (Molecular Probes, Eugene OR, USA, both 1:500) were used. Negative controls without primary antibodies were carried out in parallel.
Observation of fluorescent cells was done using a Zeiss LSM 5 Pascal confocal microscope. Excitation used the 488 nm laser and emission was collected between 505 and 550 nm, to detect the GFP or Alexa Fluor 488 fluorescence. For Alexa Fluor 568 fluorescence, the excitation used the 543 nm laser and emission was between 560 and 615 nm. The objectives used were either C-Apochromat 63× 1.4 oil or C-Apochromat 63× 1.2 water. The X-Z sections were obtained from 0.6-0.9 μm-thick z-stack images taken over a thickness of 7-14 μm depending on cell size. All immunolocalization experiments were repeated at least three times with similar results. Representative images are shown throughout.
HEK-293 cells were grown and transiently transfected with expression plasmids for the ClC-2 construct and πH3-CD8 to identify effectively transfected cells as described previously (Cid et al., 2000). Experiments were performed on cells in 35 mm cell-culture plastic Petri dishes mounted directly on the microscope stage. The bath solution contained (in mM) 140 NaCl, 2 CaCl2, 1 MgCl2, 22 sucrose and 10 HEPES pH 7.4 adjusted with Tris. The pipette solution (35 mM Cl–) contained (in mM) 100 Na gluconate, 33 CsCl, 1 MgCl2, 2 EGTA and 10 HEPES pH 7.4 adjusted with Tris. Liquid junction potentials were calculated (Barry, 1994) and appropriate corrections applied.
Standard whole cell patch-clamp recordings were performed as described previously (Díaz and Sepúlveda, 1995) using an EPC-7 (List, Germany) amplifier. The bath was grounded via an agar-150 mM KCl bridge. Patch-pipettes had resistances of 2-3 MΩ. The voltage pulse generator and analysis programs were from Axon Instruments. The currents generated by transfection were neither observed in untransfected cells nor in cells transfected with the πH3CD-8 plasmid alone.
Specificity of anti-ClC-2 antibody staining in the gastrointestinal tract
C-terminus-directed antibodies have been previously used in localization studies in mouse, rat, human and guinea pig intestine with controversial results. More generally, ClC-2 immunocytochemistry studies have been fraught with uncertainties (Jentsch et al., 2002). To avoid these problems we have examined the localization of ClC-2 by comparing immunocytochemical localization in mouse intestinal epithelium from normal (WT) and ClC-2-null (KO) mice (Nehrke et al., 2002). We used three different anti-ClC-2 antibodies. ACL-002 and sc-16430 were raised against C-terminal peptides derived from the rat sequence (GenBank access N° P35525), which is identical to that of the mouse (GenBank access N° AAD26466); the third antibody, sc-16429, was raised against an N-terminus peptide. In our hands, sc-16430 gave no signal in tissues from WT animals; sc-16429 non-specifically labeled both WT and KO tissues at intracellular crypt and apical surface locations. These antibodies were, therefore, not used further. A third antibody, ACL-002, strongly stained tissues from WT animals without any apparent signal in the tissues of KO animals and was the reagent used here.
The study of the distribution of immunoreactive ClC-2 protein was carried out using an immunoperoxidase technique performed on consecutive tissue sections or by double immunolabeling. Prior to immunostaining, two antigen retrieval procedures were performed, one using sodium citrate at pH 6.0 and the other with Tris-HCl at pH 10.0 in a microwave at 90°C. Both procedures significantly improved the intensity of immunostaining and revealed an identical basolateral distribution of ClC-2 in colon and duodenum. Nevertheless, antigen retrieval using Tris-HCl at pH 10.0 was superior to that obtained after treatment with sodium citrate at pH 6.0 (not shown).
Fig. 1a,d show low magnifications of colon sections from WT and KO animals, respectively. It can be seen that the surface of the WT epithelium expresses a strong label that is absent from non-epithelial tissues, on the same section, and in the section from a KO animal. A higher magnification in Fig. 1b shows that only the surface epithelial cells are labeled while the crypts are devoid of signal. In Fig. 1c, higher magnification shows that immunoreactivity clearly delineated the limit between surface epithelial cells of the colon (arrows). The high magnification examination of colon epithelia from KO animals did not detect immunolabel. A picture similar to that with colonic tissues emerged from the examination of duodenum sections from WT and KO animals (Fig. 1g,h). ClC-2 immunolabel was present at the borders between duodenal villus epithelial cells (g) but was absent from tissue from the KO animals (h). In all cases, ClC-2 immunoreactivity was restricted to the surface epithelium and undetectable in cells forming the crypts (Fig. 1b and Fig. 3a,d).
ClC-2 expression in the basolateral membranes of surface cells, but not secretory intestinal cells
To check for the exclusive basolateral localization of ClC-2 immunolabel a comparison was made with the distribution of immunoreactivity for H+-K+-ATPase, a known marker of apical membranes in colon surface epithelium (Gallardo et al., 2001). Fig. 2a,c show consecutive sections stained with anti-ClC-2 and anti-H+-K+-ATPase, respectively. In Fig. 2b,d, higher magnifications can be seen of the portions identified by rectangles in the lower magnification images (Fig. 2a,c). As expected, H+-K+-ATPase was present exclusively at the apical border of cells, a region negative for ClC-2 immunoreactivity. This experiment demonstrated, in addition, that the luminal cell membrane was intact after tissue processing. Again, only the basal and lateral aspects (arrows in b) of surface epithelial cells had marked ClC-2 immunostaining.
To ascertain further whether ClC-2 might be involved in fluid secretion, its distribution in colon and duodenum epithelium was compared with that of the basolateral cotransporter NKCC1, which is known to accumulate Cl– above equilibrium for secretion via CFTR (Matthews et al., 1998; Flagella et al., 1999). Fig. 3a shows immunostaining for ClC-2 and NKCC1 in the colon revealed with AEC (red) and DAB (brown), respectively. There was no overlap in the distribution of ClC-2 and NKCC1 immunoreactivities, as the latter was expressed strictly in the crypt region. Fig. 3b,c show higher magnification images of the surface and crypt regions, respectively, confirming the separation in the expression of ClC-2 and NKCC1. Notice that expression of the cotransporter, as expected, was confined to the basolateral aspect of the cells (Fig. 3c). Similar results were obtained when duodenum sections were studied in the same way (Fig. 3d-f). ClC-2 immunoreactivity was present in the villus region but absent from the crypt, while the reverse situation occurred for NKCC1. As in the colon, NKCC1 staining in the duodenal crypts was basolateral (Fig. 3f).
ClC-2 immunolocalization experiments were also carried out on jejunum, ileum and stomach. There was no obvious staining in these later tissues. In the stomach the antibody dilution was decreased to 1:10 without producing specific staining.
Basolateral targeting of ClC-2 in polarized epithelial cell lines
Specific targeting of membrane proteins has been extensively studied in kidney-derived MDCK and other epithelial cell lines, which form continuous polarized layers in vitro. If ClC-2 is a bona fide basolateral protein, as suggested by its intestinal localization, it should contain defined motifs that will also generate similar targeting in model epithelia. We have used ClC-2 fused with green fluorescent protein at its C-terminus, ClC-2-GFP (Niemeyer et al., 2004), to evaluate the possible polarization of ClC-2 in MDCK cells. Confocal X-Y sections revealed basolateral accumulation of ClC-2-GFP at the marginal borders in two cells acutely transfected within a confluent monolayer of MDCK cells grown on glass support (upper panel of Fig. 4a). This impression was confirmed by examination of X-Z optical sections, where the cells exhibited basolateral, but no apical, expression (bottom panel). A very similar picture was obtained using MDCK cells grown upon permeable support (Fig. 4b). Cell confluence was confirmed by staining with an anti-ZO-1 antibody (Fig. 4c), which revealed a continuous belt around the apical portion of the cells. The targeting of ClC-2 contrasted to that of hSlo, a K+ channel of known apical expression in MDCK cells (Bravo-Zehnder et al., 2000). Fig. 4d shows restricted expression of the c-myc-tagged hSlo α-subunit of the maxi-K+ channel to the apical membrane of a cell within a monolayer grown on glass support, which contrasts with that seen in MDCK cells similarly transfected with ClC-2. To test for a possible cell-specificity of the observed targeting of ClC-2, similar transfection experiments were carried out using polarized Caco-2 cells, derived from human colon. Fig. 4e,f show Caco-2 cells on glass and filter supports, respectively, expressing the ClC-2-GFP fluorescence at their basolateral membranes, with no evidence of apical expression.
Role of μ1B subunit of the AP-1 clathrin adaptor in the basolateral targeting of ClC-2
The AP-1B clathrin adaptor is an epithelium-specific protein complex involved in basolateral sorting (Fölsch et al., 1999). This heterotetrameric complex is composed of subunits termed γ, β1, μ1B and σ1 and differs from a more widely expressed AP-1A complex, which contains the μ1A, rather than μ1B, subunit. AP-1B mediates targeting to the basolateral membrane (Nakatsu and Ohno, 2003). We have used epithelial LLC-PK1 cells expressing the μ1A or μ1B subunit (Fölsch et al., 1999) to test whether the basolateral sorting of ClC-2 shown above is dependent on the AP-1B complex. As shown in Fig. 5a, LLC-PK1 cells transfected with μ1A expressed the channel without any evidence for polarity. A similar result was obtained for cells grown on permeable support (Fig. 5c). LLC-PK1 cells transfected with μ1B, conversely, exhibited a completely different distribution of ClC-2, targeting it exclusively to the basolateral membrane (Fig. 5b,d for cells grown on glass and filters, respectively). Basolateral polarity was quite evident in the X-Z vertical sections, which provide images very close to those seen after transfecting ClC-2 into MDCK or Caco-2 cells. Both these cell lines are known to have endogenous μ1B expression (Fölsch et al., 1999). To check for monolayer formation in these experiments, we used double staining with a ZO-1 zonula occludens antibody and ClC-2-GFP. These experiments demonstrated that both LLC-PK1-μ1A and LLC-PK1-μ1B cells had formed continuous monolayers with ZO-1 immunoreactivity delineating the borders between cells. ClC-2 fluorescence occurred below and above ZO-1 label in LLC-PK1-μ1A cells. By contrast, it was only basal to ZO-1 label in LLC-PK1-μ1B cells (results not shown).
Sorting signals defining the basolateral localization of ClC-2
Transmembrane proteins often contain sorting signals that are small stretches of amino acid sequence, usually cytoplasmic, that determine their localization within the cell. A number of sorting signals have been identified that are selectively recognized and bound by AP complexes. Well-characterized targets of AP complexes include tyrosine and di-leucine signals (Nakatsu and Ohno, 2003). Fig. 6 shows a possible structure for the ClC-2 channel based on that derived from a crystalographic study of S. thyphimurium ClC (Dutzler et al., 2002). Potential tyrosine and di-leucine sorting signals on the cytosolic side of the channel are indicated. There are 10 such sites, and the possible functional significance in basolateral targeting of ClC-2 of some of these was explored by mutagenesis. Table 1 shows a summary of the results. Tyrosines 17, 23 and 26 are highly conserved in rat, mouse and guinea pig ClC-2. An N-terminus deletion mutant of the guinea pig ortholog, encompassing all of these residues (Varela et al., 2002), was tested and no alteration of the basolateral targeting was found. Three further potential tyrosine-based signals (64, 179 and 553) did not affect the basolateral targeting of ClC-2. Four potential di-leucine signals were also tested, with mutation in one of them (L812AL813A) drastically changing the targeting of ClC-2 from basolateral to apical membrane. Confocal fluorescence images in the middle and right panels of Fig. 7A show cells acutely transfected with L812AL813A-ClC-2-HA within a confluent monolayer of MDCK cells grown on glass support, revealing an apical border accumulation. The appearance of fluorescence in the X-Y sections shown in the upper panels was reminiscent of that seen for the apically-targeted hSlo K+ channel (Fig. 4d). The apical border accumulation of L812AL813A-ClC-2-HA was confirmed by examination of X-Z optical sections (bottom panels in Fig. 7A). Left panel in Fig. 7A shows the distribution of a control, non-mutated ClC-2-HA showing the expected basolateral accumulation. To test whether L812AL813A mutation affected ClC-2 channel function, the activity of the channels was studied after transfection into HEK-293 cells. Fig. 7B shows currents evoked by voltage pulses between –170 and 0 mV in cells transfected with L812AL813A-ClC-2-HA. Currents appear similar to those obtained with WT ClC-2 [not shown, but see (Niemeyer et al., 2004)], with slow rates of opening and closing and strict inward rectification. The steady-state activation curve is shown in Fig. 7C. The V0.5 obtained was –104±5 (mean±s.e.m., n=4) and did not differ significantly from that measured for WT ClC-2 (–108±4 mV, n=10). The slope factor was –28±1 mV and appeared to differ slightly from that of WT ClC-2 (–23±1 mV, n=10, P=0.002).
CFTR is a cAMP-activated Cl– channel and its dysfunction in fluid secretory epithelia appears to be the primary cause of CF (Sheppard and Welsh, 1999). The identification of alternative Cl– conductance pathways in secretory epithelia as a way to compensate the loss of CFTR has been a major goal in CF research. The Cl– channel ClC-2 is expressed in various epithelia including the intestine, and, given its possible activation by extracellular acidification, intracellular Cl– or cell swelling (Gründer et al., 1992; Jordt and Jentsch, 1997; Niemeyer et al., 2003), it represents a good candidate as such surrogate conductance. ClC-2 channel has been linked to epithelial Cl– secretion in airway epithelium (Blaisdell et al., 2000), intestinal Caco-2 cells (Mohammad-Panah et al., 2001) and native intestinal epithelium (Gyömörey et al., 2000). In the last two studies ClC-2 was co-localized with the tight junctions, where it was speculated to mediate Cl– efflux into the lumen. More recent experiments, however, report expression of ClC-2 at the basolateral membranes of surface colonic and small intestinal cells (Lipecka et al., 2002; Catalán et al., 2002). This, plus data showing ClC-2-type currents as the dominant anion conductance of enterocyte basolateral membrane (Catalán et al., 2002; Catalán et al., 2004), suggest ClC-2 as a basolateral pathway for Cl– exit from absorptive cells.
To resolve the problem of localization of ClC-2 in intestinal epithelium, we have now used an immunohistochemical approach, with ClC-2 KO mice (Nehrke et al., 2002) as negative controls. ClC-2 was located exclusively in villus cells of duodenal epithelium and surface cells of the colon, being absent from crypts and non-epithelial tissues. The validity of this localization is strongly supported by the complete absence of label in tissues from ClC-2 KO animals. We did not detect ClC-2 immunoreactivity in stomach. This contrasts with studies in rabbit gastric mucosa (Sherry et al., 2001), but agrees with recent results with rabbit, rat and human gastric tissue (Hori et al., 2004). Within surface and villus cells, ClC-2 is localized to basolateral membranes and excluded from apical membranes that, in the colon, could be stained for H+-K+-ATPase, a known apical marker (Gallardo et al., 2001). ClC-2 distribution within duodenal and colonic epithelium did not overlap with the crypt basolateral cotransporter NKCC1, the Cl– accumulation mechanism for CFTR-mediated secretion (Matthews et al., 1998; Flagella et al., 1999). Our data, therefore, provide strong evidence for a basolateral localization of ClC-2 in absorptive cells of duodenum and colon. Parallel exchangers of Na+/H+ and Cl–/HCO3– provide the entry step for NaCl in electroneutral absorption (Binder et al., 1987). Exit of Na+ from the cells is via the Na+ pump and a basolateral Cl– conductive pathway is assumed to allow for efflux of the anion. ClC-2 could provide the basolateral Cl– conductance in electroneutral salt and water absorption (Catalán et al., 2002; Lipecka et al., 2002; Catalán et al., 2004). Such a function would explain the results in double CFTR/ClC-2 knockout mice, which appear to survive better than CFTR-deficient mice (Zdebik et al., 2004).
We find that human recombinant ClC-2 also has a basolateral distribution in polarized MDCK and Caco-2 cells. The sorting of basolateral membrane proteins depends on cytoplasm-facing motifs, tyrosine-containing motifs and di-leucine or di-hydrophopic residue signals, which interact with intracellular adaptor complexes (Rodriguez-Boulan et al., 2004). The list of determinants of basolateral targeting, however, is growing, as exemplified by a single leucine that directs CD147 to that membrane (Deora et al., 2004). A family of adaptor protein complexes termed AP interacts with tyrosine-based and other motifs to direct proteins to different membrane domains. Two such complexes, the heterotetrameric clathrin adaptors AP-1 and -4 have been implicated in basolateral sorting (Robinson and Bonifacino, 2001; Nakatsu and Ohno, 2003; Simmen et al., 2002). Subunits composing the AP-1 complex are γ, β1, μ1 and σ1. AP-1 can exist in the epithelial-specific form AP-1B, that contains μ1B subunit and directs basolateral sorting, or a more widely expressed μ1A-containing AP-1A (Fölsch et al., 1999). AP-1B is involved in basolateral sorting. The kidney epithelial cell line LLC-PK1 was found to mislocalize basolateral proteins to the apical membrane, due to the absence of μ1B subunit (Fölsch et al., 1999). Interestingly, AP-1B is highly expressed in epithelial and endocrine cells, including those of the intestine (Ohno et al., 1999). AP-4 is also capable of promoting the basolateral sorting of some proteins in MDCK cells and, it has been speculated, it might play a role in non-epithelial cells or in those lacking AP-1B (Simmen et al., 2002). We have used LLC-PK1 cell lines stably transfected with either the μ1A or μ1B subunit (Fölsch et al., 1999), and found that while ClC-2 sorted to the basolateral membrane in LLC-PK 1-μ1B cells, it did not display polarity in LLC-PK1-μ1A cells. Several intracellular-facing tyrosine and di-leucine motifs are present in ClC-2. Mutation of one of these, L812L813, mis-sorted ClC-2 to the apical membrane in MDCK cells. This di-leucine motif is conserved in mouse, rat, rabbit and guinea pig ClC-2. It is also present in ClC-0, -1, -Ka and -Kb, all plasma membrane ClC channels, but not in ClC-3, -4, -5, -6 and -7, channels from intracellular compartments (Jentsch et al., 2002).
Our results suggest that μ1B epithelial-specific subunit might mediate the basolateral sorting of ClC-2. Mutation of all potential tyrosine-based motifs in ClC-2, however, failed to alter its basolateral sorting in MDCK cells. Surprisingly, mutation of L812L813 motif changed the destination of ClC-2 to the apical membrane. A simple explanation for this observation is that L812L813 interacts with the AP-1B sorting complex. There are no data in the literature, however, to suggest that such a differential interaction between a di-leucine-based signal with AP-1B exists (Bonifacino and Traub, 2003). In addition, the di-leucine sites identified in ClC-2 do not conform to the canonical [DE]XXXL[LI] identified as a basolateral sorting signal. There are, however, examples of unconventional di-leucine motifs promoting basolateral sorting of certain proteins, including the K+ channel KCNQ1 (Jespersen et al., 2004), the nucleotide-pyrophosphatase NPP1 (Bello et al., 2001) and the anion transporter sat-1 (Regeer and Markovich, 2004). In the last example, the signal has threonine instead of the conventional acidic amino acid at position 4 upstream from LL, which coincides with the site identified here (TIFSLL). It is possible that L812L813 acts as a basolateral targeting signal at the trans-Golgi network and that the dependence of ClC-2 distribution upon AP-1B has to do with recycling back to the basolateral membrane from endosomes (Gan et al., 2002). The apical sorting of the L812AL813A-ClC-2 mutant in MDCK cells could be explained by the unmasking of a cryptic apical sorting signal, as happens with the di-leucine basolateral targeting signal of NPP1 (Bello et al., 2001). Recent work using the low-density lipoprotein receptor (LDLR) and the transferrin receptor (TfR) has suggested that AP-1B complex participates in post-endocytic basolateral sorting (Gan et al., 2002). Such a function could explain, in addition, why L812AL813A-ClC-2 is found at the apical membrane in MDCK cells and without polarity in LLC-PK1-μ1A, as the distribution will depend on the kinetics of retrieval and reinsertion of endocytosed protein (Gan et al., 2002). Further work, beyond the scope of the present communication, will be needed to decide between these possibilities and to ascertain in detail the mechanism of basolateral sorting of ClC-2.
In summary, we demonstrate that ClC-2 is a basolateral membrane Cl– channel present in absorptive epithelial cells in small and large intestine. In addition, we demonstrate that this basolateral sorting is dependent on a novel di-leucine motif that, when mutated mis-targets the channel to the apical membrane of MDCK cells. It is interesting to note that the LL motif is within one of two highly conserved C-terminal cystathionine beta synthase (CBS) domains present in ClC-2. These conserved sequences are believed to form AMP or ATP binding sites and act as regulators of protein function (Scott et al., 2004; Niemeyer et al., 2004). Studies in ScClC and ClC-1 channels have suggested that CBS domains are important in determining their subcellular distribution, trafficking and gating (Schwappach et al., 1998; Hebeisen et al., 2004; Estévez et al., 2004). The LL motif of ClC-2 identified here lies within CBS-2. It is conceivable that pharmacological agents interacting with CBS domains to modulate function might also alter targeting. Such mis-localized ClC-2 could act as an alternative permeability pathway in CF. Human CFTR expressed in villus cells of mouse intestine has been show to correct the CF phenotype CFTR KO mice (Zhou et al., 1994). Similarly, the simpler requirements for activation of ClC-2 could make of the L812AL813A-ClC-2 mutant a potential tool for gene therapy strategies in CF.
Supported by Fondecyt Grant 1030627 (F.V.S.) and NIH grant DE09692 (J.E.M.). CECS is a Millennium Science Institute and is funded in part by grants from Fundación Andes, the Tinker Foundation and Empresas CMPC. I.C. is supported by CONICYT Chile. We are grateful to Marcela Bravo-Zehnder, Juan Codina, Ira Mellman and R. James Turner for gifts of antibodies and cell lines, and to Marcela Bravo-Zehnder and Víctor Faúndez for invaluable advice.