NCC27, a 27-kDa homolog of the intracellular chloride channel p64, was recently described as a chloride channel in nuclear membrane. We probed human Northern blots for NCC27 and found an ∼1.7-kb message in all tissues examined, including kidney, the transcript being most abundant in heart and skeletal muscle. NCC27-specific antisera was raised to a COOH-terminal peptide derived from the NCC27 coding region. Using this antisera, we find NCC27 is expressed in an intracellular vesicular compartment in HeLa cells, PancI cells, and macrophages. In human and mouse kidney, NCC27 is expressed at low levels in most cells of the kidney. NCC27 is highly expressed in glomeruli, in periarterial smooth muscle, and in the apical membrane of a subset of cortical tubule cells. Double staining with nephron segment-specific lectins indicates that the NCC27-expressing cells are proximal tubule cells.
- vascular smooth muscle
chloride channels carry out a number of essential functions in the kidney. Like cells elsewhere in the body, kidney cells use chloride channels for defense against hypotonic swelling, in acidification of intracellular compartments, and in exocytosis (1, 11). In renal and other epithelial cells, chloride channels also play a more specialized role in the process of transepithelial chloride transport (7).
A host of distinct chloride channel activities has been reported in cells derived from kidney (3, 4, 7, 13, 19, 24, 25, 30). A number of chloride channel proteins are known to be expressed in the kidney, most notably, cystic fibrosis transmembrane conductance regulator (CFTR) (16), several members of the ClC family of chloride channels (12, 27), and p64 (14). The precise roles of each of these proteins in chloride permeabilities throughout the kidney are still largely unknown. It seems likely that as-yet-unknown proteins account for at least some of the observed channel activity.
The chloride channel protein p64 was originally identified by biochemical purification from bovine kidney microsomal membranes (13). The protein is selectively expressed in the limiting membranes of what appear to be regulated secretory vesicles in kidney tubule cells and in T84 cells (18). When exogenous p64 is expressed inXenopus oocytes or PancI cells, which do not normally express p64, the protein is excluded from the plasma membrane, further supporting the identification of p64 as a channel of intracellular membranes (14, 18).
The sequence of p64 has been reported from cDNA (14). The protein consists of 428 amino acids with a predicted molecular weight of 49 kDa. As demonstrated by in vitro translation, the apparent molecular weight of 64 kDa is independent of posttranslational modification. Thus p64 demonstrates an aberrantly high apparent molecular size by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Based on hydropathy analysis, p64 has been proposed to have two transmembrane segments, both of which are in the COOH-terminal half of the molecule.
It has become apparent that p64 is a member of a closely related multigene family. A protein of apparent molecular weight of 62 kDa, which is antigenically related to p64, has been identified as the chloride channel in osteoclast ruffled border (20). A second distinct p64 homolog named NCC27 (27-kDa nuclear chloride channel), proposed to function as a channel in nuclear envelope, was recently cloned from human monocytic cells (28). A partial cDNA from rat brain encoding this protein has also been reported (9). NCC27 is quite similar to the COOH-terminal half of p64 but is missing any sequence corresponding to the amino-terminal half of the molecule. Expression of this protein in CHO cells resulted in increased chloride channel activity both in plasma membrane and in nuclear membrane (28).
In this study, we report independent cloning and characterization of NCC27 from the human pancreatic cancer cell line, PancI. We find that NCC27 is widely expressed in human tissue, including kidney. Immunofluorescent studies using novel NCC27-specific antisera demonstrate that, in the kidney, the protein is highly expressed in the apical membrane of proximal tubule, in glomeruli, and in periarterial smooth muscle.
MATERIALS AND METHODS
Cell culture. PancI cells (15) were obtained from the American Type Culture Collection (Bethesda, MD). HeLa cells and recombinant vaccinia stock were obtained from Dr. Andrey Shaw (Washington University, St. Louis, MO). Both cell types were grown in DMEM with 10% fetal calf serum, 100 μg/ml penicillin, and 10 μg/ml streptomycin. Primary cultures of human macrophages (29) were a gift from Dr. Eric Brown (Washington University). CHO-1 cells were obtained from Dr. Keith Hruska (Washington University) and were grown in DMEM-Ham’s F12 (1:1) with 5% fetal calf serum, 100 μg/ml penicillin, and 10 μg/ml streptomycin.
cDNA cloning. Randomly primed cDNA was prepared from poly(A)+-selected RNA from PancI cells using the Zap-cDNA synthesis kit and inserted into the λZap vector (Stratagene, La Jolla, CA). The library was screened by hybridization at low stringency with a probe containing the entire coding region of bovine p64. An isolate named PG11 was found to contain a complete coding region essentially identical to that reported for NCC27 (28). PG11 was used as starting material for the subsequent subcloning and expression experiments.
Northern blots. Human multiple tissue Northern blot was purchased from Clon-Tech (Palo Alto, CA). Poly(A)+ RNA was prepared from PancI cells by standard methods. Three micrograms of poly(A)+ RNA was separated by denaturing electrophoresis, blotted, and probed as described (14). Glyoxal-denatured DNA molecular weight marker 3 (Boehringer-Mannheim, Indianapolis, IN) was separated on the same gel. The molecular weight standards were cut from the rest of the gel prior to transfer; stained in 500 mM NaCl, 100 mM Tris, pH 8, and 1 μg/ml ethidium bromide at 4°C overnight; and photographed. Blots were probed at high stringency with a 300-bp EcoR I fragment from the NCC27 coding region (positions 576–871 of published sequence).
Generation of antisera. Antiserum 823 was generated against a peptide derived from the COOH terminus of NCC27. The peptide NH2-CDEEIELAYEQVAK, corresponding to positions 226–238 of the predicted NCC27 amino acid sequence with an added amino-terminal cysteine, was synthesized and purified by the Protein and Nucleic Acid Chemistry Lab at Washington University. The peptide was coupled to keyhole limpet hemocyanin usingm-maleimidobenzoyl-N-hydroxysuccinimide as described (8). A rabbit was immunized with the coupled peptide by Cocalico Biologicals (Reamstown, PA), resulting in a high-titer antiserum.
An amino-terminal his-tagged version of NCC27 was generated by subcloning the insert from plasmid PG11 into the pQE30 vector (Qiagen, Chatsworth, CA). The his-tagged protein was expressed inEscherichia coli and purified by Ni-affinity chromatography, according to procedure provided by the manufacturer. The his-tagged protein ran with an apparent molecular weight of ∼34 kDa by SDS-PAGE and was recognized by the crude 823 antisera on Western blots. Purified his-tagged NCC27 was immobilized on an Aminolink Column (Pierce, Rockford, IL) and used to affinity purify the antisera as described (8). The affinity-purified antibody, designated AP823, was used for all experiments.
Preparation of proteins from cultured cells. To generate whole cell preparations, cultured cells were rinsed with PBS (135 mM NaCl, 10 mM sodium phosphate, pH 7.0) and then solubilized in 150 mM NaCl, 25 mM Tris, pH 8.0, 1 mM EDTA, 1% Triton X-100, and 0.1 mM phenylmethylsulfonyl fluoride (PMSF) at 4°C for 15 min. The solution was collected, and insoluble material was removed by centrifugation at 15,000 rpm in a microcentrifuge at 4°C for 10 min. The supernatant is taken as the whole cell extract.
Nuclei from CHO-1 cells were prepared by Nonidet P-40 (NP-40) lysis of cells, followed by centrifugation of nuclei through a sucrose cushion, exactly as described by others to generate nuclei for patch-clamp studies of NCC27 (28).
Overexpression of PG11 in HeLa cells.Vaccinia T7-driven expression was carried out essentially as described (6, 22). In brief, HeLa cells were plated in 3.5-cm dishes at 5 × 105 cells/dish (∼50% confluence) in DMEM with 10% newborn calf serum. The next day, cells were rinsed with serum-free medium. Cells were infected with vaccinia in 0.75 ml of DMEM at a multiplicity of infection of ∼10 and incubated at 37°C for 30 min. Five micrograms of plasmid DNA and 15 μl of Lipofectace (GIBCO-BRL; Life Technologies, Bethesda, MD) were combined in 0.75 ml of DMEM and added to the infected cells. The infected/transfected cells were then incubated at 37°C in 5% CO2 for 10 h. Cells were solubilized in 500 μl of SDS-PAGE loading buffer. Twenty-five microliters of each sample were analyzed by Western blotting.
In vitro translation. Plasmid PG11 was linearized with BamH I and transcribed with T7 RNA polymerase as previously described (14). Resulting capped RNA was translated using a reticulocyte lysate (Promega, Madison, WI) in the presence of 35S-labeled methionine. Products were separated by SDS-PAGE on the same gel with unlabeled PancI lysate. After electrophoresis, the gel was cut in half. The PancI lane and molecular weight markers were transferred to nitrocellulose and probed for p34 with affinity-purified 823 antisera. The in vitro translation products and molecular weight markers were stained with Coomassie blue, dried, and subjected to autoradiography.
Fractionation of kidney membranes.Brush-border membranes were prepared from mouse kidney as described (2). In brief, kidneys were prepared from three normal mice and rinsed in ice-cold PBS. Cortex was dissected and minced and then suspended in homogenization buffer (10 mM mannitol, 2 mM HEPES, pH 7.4, and 100 μM PMSF) at 10 ml/g of tissue. The sample was homogenized on ice, first in a loose-fitting glass tissue grinder and then with 10 strokes in a tight-fitting Teflon homogenizer. Debris was removed with a 2-min spin at 2,000 rpm. The supernatant was collected, and a sample was retained as crude lysate. MgCl2 was added to 5 mM final concentration, and the sample was mixed at 4°C for 15 min. The sample was centrifuged at 1,500g (3,500 rpm in Sorvall SS34 rotor) for 12 min at 4°C. The supernatant was collected and centrifuged at 20,000 g (14,000 rpm in Sorvall SS34) for 45 min at 4°C. The supernatant from this spin was saved to prepare soluble and microsomal fractions (see below). The pellet was resuspended in homogenization buffer. MgCl2 was added, and sample was subjected to differential centrifugation as before. The pellet from the final 20,000-g spin was taken as the brush border preparation. The supernatant from the first 20,000-g spin was centrifuged at 100,000 g (40,000 rpm in Beckman Ti70.1 rotor) for 1 h at 4°C. The supernatant was taken as the soluble fraction, and the pellet was resuspended to make the microsomal fraction.
Because of limited material, a brush border preparation was not attempted from human kidney. Instead, human kidney cortex was homogenized in 250 mM sucrose, 10 mM imidazole, 1 mM EDTA, and 0.1 mM PMSF. Nuclei and cellular debris were removed by centrifugation at 2,000 g (4,000 rpm, Sorvall SS34 rotor) for 10 min at 4°C. The supernatant was taken as a crude lysate, which was then centrifuged at 100,000g (40,000 rpm in a Beckman Ti70.1 rotor) for 1 h at 4°C, yielding a membrane pellet and a soluble fraction.
Alkaline extraction of membrane fractions was carried out by bringing a membrane pellet up in 10 mM Na2CO3(pH 10.5) or 100 mM Na2CO3(pH 11.5), as indicated, and incubating on ice for 30 min. The sample was then spun at 100,000 g for 1 h to yield the alkali-washed membrane pellet. To assess detergent solubility, 30 μg (protein) of the pH 10.5 alkali-washed human kidney membrane fraction were suspended in 50 mM HEPES, pH 7.0, 100 mM NaCl, and 1% Triton X-100 and incubated on ice for 15 min. The solution was then centrifuged at 40,000 rpm as before, yielding a Triton-soluble fraction and a Triton-insoluble pellet.
Western blots. Samples were brought to 50 mM Tris, pH 6.8, 2% SDS, 0.1% bromophenol blue, 10% glycerol, and 20 mM dithiothreitol (DTT) and heated to 95°C for 5 min. One-tenth volume 500 mM iodoacetamide was then added to prevent an artifact on the blots related to excess DTT (17). Sample were separated by SDS-PAGE using standard methods (8). Low-range molecular weight standards from Bio-Rad (Richmond, CA) were used. Protein concentrations were determined with the bicinchoninic acid assay (Pierce Biochemical). After electrophoresis, proteins were transferred to nitrocellulose and probed with 1:2,000 dilution of AP823, using peroxidase-conjugated secondary antibodies and the Supersignal chemiluminescent detection system (Pierce). For preabsorption, the diluted antibody was incubated with 10 μg/ml peptide for 15 min before adding antibody to the blot.
Immunofluorescence. Cultured cells were grown on glass coverslips, rinsed with PBS, and fixed in 100% methanol at −20°C for 5 min. Cells were then rinsed with PBS twice and blocked with PBS-FG (PBS containing 5% goat serum and 0.02% fish gelatin) for 1 h at room temperature. Cells were exposed to the primary antibody by inverting the coverslip on a 20-μl drop of AP823 antiserum diluted 1:20 in PBS-FG and incubated at room temperature for 2 h. For preabsorption, the peptide antigen was added to the diluted antibody to a final concentration of 10 μg/ml and incubated for 15 min before exposure to the cells. Coverslips were washed four times with 10-min exchanges of PBS-FG, mounted in Vectashield (Vector, Burlingame, CA), and photographed with a Zeiss Axiophot epifluorescence microscope.
Five-millimeter blocks of fresh human and mouse kidney were quick-frozen in Tissue-Tek embedding medium, and ∼4-μm-thick frozen sections were obtained. Sections were fixed on the slide while still frozen by immersion in methanol at −20°C for 5 min. Section were rinsed with PBS and then blocked with PBS-FG for 1 h. Affinity-purified antibody was diluted 1:20 in PBS-FG and incubated with the sections for 2 h at room temperature. For preabsorption, the diluted antibody was incubated with 10 μg/mol peptide for 15 min prior to addition to the section. Samples were washed four times with PBS-FG and then incubated for 1 h with tetramethylrhodamine isothiocyanate (TRITC)-conjugated, affinity-purified goat anti-rabbit antibody (Boehringer-Mannheim). For samples double stained with lectin, the secondary antibody was diluted into PBS-FG with 0.1 mM MgCl2 and 0.1 mM MnCl2 along with 10 μg/ml of the appropriate FITC-conjugated lectin (Sigma, St. Louis, MO). Sections were washed four times, with the last incubation going overnight at 4°C. Sections stained only with antibody were mounted in Vectashield (Vector). Sections costained with lectin were mounted in PBS supplemented with Mg, Mn, and Ca as above.
A cDNA library from PancI cells was constructed and screened by hybridization at low stringency with a probe to bovine p64, yielding a number of overlapping isolates. The protein encoded by these clones was found to be identical to that reported for NCC27 (28) with one discrepancy: we find the nucleotide at position 405 of the published NCC27 sequence to be C instead of G and, hence, the amino acid encoded at position 63 to be glutamine instead of glutamic acid. One isolate, named PG11, was found to contain the complete NCC27 coding region, oriented so that the transcript from the T7 promoter is sense RNA. This plasmid was used for subsequent expression and cloning experiments.
A 300-bp EcoR I fragment encoding the COOH-terminal half of the molecule was used to probe RNA from PancI cells, as well as a series of human tissues including heart, brain, placenta, lung, liver, skeletal muscle, kidney, and pancreas. As shown in Fig.1, we find a transcript of ∼1.7 kb in each of the human tissues and in the PancI cells. The transcript is particularly abundant in heart and skeletal muscle.
NCC27 shows aberrant mobility by SDS-PAGE. We raised an antisera, named 823, against a peptide derived from the COOH-terminal domain of NCC27. The antisera was affinity purified over amino-terminal his-tagged NCC27 immobilized on an agarose support. AP823 was used to probe a Western blot of PancI and human macrophage total cellular protein as shown in Fig.2 A. The antibody recognizes a protein of apparent molecular weight 34 kDa in both cell types, and this signal is preabsorbed by incubation of the antibody with the peptide antigen.
To be certain that the 34-kDa signal recognized by our antibodies is in fact due to the protein encoded by our cDNA, we expressed the PG11 cDNA in HeLa cells using a vaccinia/T7-driven system and probed for NCC27 as shown in Fig. 2 C. Equal fractions of vaccinia-infected cells transfected with a control plasmid (Fig.2 B, lane 1) or with PG11 cDNA (Fig.2 B, lane 2) were solubilized, separated by SDS-PAGE, blotted, and probed with AP823. In control HeLa cells, a 34-kDa protein is present that is indistinguishable from that seen in PancI cells. With expression of exogenous PG11, the 34-kDa protein becomes more abundant, and no new protein band is detected (Fig.2 B, lane 2). The 34-kDa signal in both control and PG11-expressing cells is blocked by preabsorption with the peptide antigen (Fig. 2 B,lanes 3 and4). Thus NCC27 protein expressed from cDNA is identical in mobility to the endogenous protein in HeLa cells recognized by our NCC27 antibodies.
NCC27 has been reported to reside in nuclei of CHO-1 cells, and chloride channel activity found by patch-clamp studies of nuclear envelope has been attributed to NCC27. Whole cell lysates and nuclei were prepared from CHO-1 cells. The method for isolating nuclei was identical to that used by others to study p34 in nuclei using patch-clamp technique (28). Ten micrograms of total cell protein and of isolated nuclei from CHO-1 cells were separated alongside macrophage total cell protein and probed with AP823 (Fig.2 C). The antibody recognizes a 34-kDa protein in CHO-1 cells, which is identical in mobility to the previously noted macrophage protein. No NCC27 is detectable in the nuclear preparation.
The discrepancy between the predicted molecular weight based on the cDNA sequence and the apparent size of the protein recognized by our antisera led us to question whether the protein could be subject to glycosylation or other posttranslational modification. Capped RNA encoding NCC27 was generated in vitro and used to direct protein synthesis by a reticulocyte lysate. The in vitro translation product was separated on the same gel with crude PancI cell lysate. The products of in vitro translation were detected by autoradiography, and PancI native NCC27 was detected by Western blotting. The results demonstrate that the in vitro translation product (Fig.2 D, lane 2) shows mobility identical to that of the native NCC27 in PancI cells as detected with AP823 (Fig.2 D, lane 3). Translation in the presence of microsomal membranes, which are competent to carry out glycosylation and signal peptide cleavage, has no effect on mobility of the translation product (not shown). Thus we conclude that NCC27 is in fact a 27-kDa protein that has aberrantly low mobility on SDS-PAGE gels.
NCC27 is present in kidney where a fraction of the total behaves as an integral membrane protein. To determine distribution of NCC27 in kidney, human kidney cortex and medulla were separately homogenized and fractionated into crude membranes and soluble protein. The membrane pellets were further characterized by alkaline extraction (10 mM Na2CO3, pH 10.5) to remove peripheral membrane proteins. Thirty micrograms of each fraction were separated by SDS-PAGE, blotted, and probed with AP823 as shown in Fig.3 A. The 24-kDa NCC27 signal is present in crude kidney lysate. NCC27 is more abundant in the soluble fraction than the membrane fraction in both cortex and medulla. However, the membrane-associated NCC27 is resistant to alkaline wash of the membrane (Fig.3 A, lane 4), typical for an integral membrane protein. To ensure that the membrane-associated NCC27 is actually in membranes and not part of an insoluble cytoskeletal complex, 30 μg of alkaline-washed membrane fraction from human kidney cortex were suspended in a solution of 1% Triton X-100. The suspension was centrifuged at 100,000 g, and both supernatant and pellet were probed for NCC27 (Fig.3 A, lanes 5–7). The NCC27 associated with alkaline-washed membranes is solubilized by Triton, consistent with its presence in membrane rather than cytoskeleton.
We also probed mouse kidney fractions for presence of NCC27. Because of easier availability of mouse material as opposed to human, we were able to fractionate the mouse kidney membranes further. We were particularly interested whether NCC27 is present in brush-border membranes. Mouse brush-border membranes were prepared from kidney cortex, using a standard magnesium precipitation method (2). The supernatant following precipitation of brush-border membranes was then centrifuged at 100,000g to yield non-brush-border membranes and a soluble fraction. The brush-border fraction was further characterized by extraction at pH 10.5 as above. Fractions were separated and probed for presence of NCC27, and the results are shown in Fig. 3 B. AP823 recognizes a 34-kDa protein, which is present both in the soluble fraction and in membrane fractions from mouse kidney. The protein is more abundant in brush-border membranes than in non-brush-border membranes, which are presumably mostly microsomes. NCC27 in brush-border membranes is resistant to alkaline extraction, again suggesting that it is present as an integral membrane protein. As with the material from cultured cells, the 34-kDa signal in both human and mouse kidney was blocked by preabsorption of the diluted antisera with peptide (not shown).
To assess membrane association more stringently, a crude preparation of mouse kidney membrane was prepared and washed with 10 mM Na2CO3(pH 10.5) or 100 mM Na2CO3(pH 11.5). Washed membranes were fractionated by SDS-PAGE and probed with AP823 as shown in Fig. 3 C. As noted before, the membrane-associated NCC27 is resistant to wash at pH 10.5. A significant fraction of the membrane-associated NCC27 is also resistant to the more stringent wash at pH 11.5, supporting the interpretation that some of the NCC27 is present as an integral membrane protein.
NCC27 is expressed in a vesicular/reticular pattern in cultured cells. AP823 was used to probe a series of cultured cells as shown in Fig. 4. Figure4, A,C, E,F, andH, shows cells stained with antibody, and Fig. 4, B,D, G, and I, shows the same cells processed identically but probed with peptide-preabsorbed antibody. Figure 4,A andB, shows PancI cells;C andD are HeLa cells;E–G are primary cultures of human macrophages; and H andI are CHO-1 cells. In PancI cells, HeLa cells, and macrophages, the antibody primarily stains small vesicular structures that are spread throughout the cytoplasm. Particularly in PancI cells, the antibody also stains large peripheral structures in addition to the finer reticular/vesicular pattern. None of these cells shows prominent nuclear staining. The CHO-1 cells show the cytoplasmic vesicular staining pattern similar to the other cells but also show prominent staining in the nucleus. The nuclear staining is not limited to the nuclear envelope (in which case the expected pattern would be a bright ring of staining defining the nuclear membrane). Instead, we find a diffuse staining of the nuclear contents. The staining of all cell types is blocked by preabsorption of the antibody with specific peptide (Fig. 4,B, D,G, andI).
Immunolocalization of NCC27 in human kidney. Staining of human kidney cortex with AP823 is shown in Fig. 5. Most cells in the kidney appeared to stain faintly above background. However, three patterns of staining were particularly prominent. The brightest staining occurred on the luminal surface of a subset of tubules in the very deep cortex as shown in Fig. 5, A–C. Figure5, A andB, shows immunofluorescent and corresponding phase-contrast images from deep in the renal cortex. There is intense staining of the luminal surface of some but not all of the tubules in this field. A higher-power view of a similar field is shown in Fig. 5 C, showing clear staining restricted to the apical surface protruding into the tubular lumen. In addition to tubular staining, we also found diffuse prominent staining of glomeruli throughout the cortex (Fig.5 D). We have not addressed which cell types in the glomerulus are responsible for this staining. Finally, we found that the smooth muscle layer of small arteries also stained prominently. Figure 5,E–F, shows immunofluorescent and corresponding phase-contrast images from the deep cortex, showing intense staining of the vascular smooth muscle layer surrounding a small artery, plus staining of the apical domain of neighboring tubules. Figure 5, G–H, are another pair of corresponding immunofluorescent and phase-contrast images from a section of deep cortex that was probed with AP823 after preabsorption with peptide. The images were photographed and processed identically to panels A andB andE andF of Fig. 5. Notice that, although the autofluorescence from the internal elastic lamina of the artery is still present, staining of tubules, glomerulus, and smooth muscle are all blocked by preabsorption.
Immunolocalization of NCC27 in mouse kidney. Mouse kidney sections were also probed for presence of NCC27 as shown in Fig. 6. Figure 6, A andB, shows fluorescent and phase-contrast images of a section stained with AP823, whereas Fig. 6,C andD, show corresponding images from a section stained with the same antibody preabsorbed with peptide. As in the human kidney, we found faint staining in most cells of the mouse kidney, with superimposed prominent staining of apical domains of cortical tubule cells and glomeruli. In contrast to the human staining pattern, the intensely staining cortical tubule segments are not limited to the deep cortex but are present throughout the cortex and were most prominent in the superficial cortex. Clearly both the tubular and glomerular staining are preabsorbed with peptide.
Identification of nephron segments showing apical NCC27. To identify the nephron segments involved in the apical membrane staining, human and mouse kidney sections were double stained with AP823 and with FITC-conjugated lectins specific for known nephron segments. In human kidney, the lectin fromLotus tetragonobulus (LTA) is specific for the brush border of proximal tubule while lectin from peanut agglutinin (PNA) is specific for thick ascending limb and distal nephron (5). Figure 7 shows human kidney sections stained for NCC27 using TRITC-conjugated secondary antibody and with FITC-conjugated LTA (Fig. 7,A–C) or PNA (Fig. 7,D–F). Sections were then photographed with rhodamine filters showing NCC27 staining (Fig. 7,A andD), with fluorescein filters showing lectin staining (Fig. 7, B andE), and with a double exposure showing both superimposed (Fig. 7, Cand F). The data demonstrate that NCC27 colocalizes with LTA staining and is distinct from PNA staining. Thus the cells expressing apical NCC27 are proximal tubule cells.
Similar experiments were carried out using mouse kidney sections. As in humans, LTA staining in mouse is specific for proximal tubule and is most intense in the apical brush border (21). Mouse sections were double stained with AP823 and FITC-conjugated LTA, shown in Fig.8,A–C. As with human kidney, all the nephron segments staining for NCC27 also stain for LTA, indicating that the NCC27-expressing cells are in the proximal tubule. As mentioned above, we found that the most intense staining for NCC27 in mouse was in the superficial cortex which largely consists of the S1 and S2 segments of the proximal tubule. We wished to be certain whether more distal portions of the proximal tubule also stain with NCC27. In mouse, a lectin from Bandeiraea simplicifolia (BSA-II) is specific for the proximal tubule pars recta (21), approximately corresponding with the S3 segment. Mouse sections were costained with AP823 and BSA-II. As shown in Fig. 8, D–F, all the tubules that stain with BSA-II also stain for NCC27, indicating that the S3 segment of mouse proximal tubule, as well as the more proximal segments, expresses NCC27.
We have reported tissue expression and subcellular distribution of the recently described chloride channel, NCC27. Our results establish three primary points. 1) NCC27 mRNA is present in all tissues we examined, being particularly abundant in heart and skeletal muscle. 2) The 27-kDa NCC27 protein shows aberrant mobility on SDS-PAGE which is not due to posttranslational modification.3) NCC27 is expressed in a cytoplasmic reticular/vesicular pattern in PancI, HeLa, macrophages, and CHO-1 cells. CHO-1 cells also show prominent nuclear staining. In kidney, NCC27 is particularly prominent in the apical membrane of proximal tubule cells, in glomeruli, and in periarterial smooth muscle.
Size and tissue distribution of NCC27 transcripts. We found a single major NCC27 transcript of ∼1.7 kb in each of the human tissues examined and in PancI cells. In contrast, transcripts of 1.2 and 1.0 kb were reported by others to be present in a macrophage-like cell line (28). Denaturing agarose electrophoresis can be inaccurate in determining transcript size. Because we used a different gel system and different molecular size markers, it is uncertain whether our results indicate that most tissues produce a transcript distinct from that in macrophage-like cells or not. We have carried out 3′ rapid amplification of cDNA ends (RACE-PCR) with RNA from PancI cells, using a kit from Life Technologies, and obtained the same 3′ noncoding region as was previously published (not shown). These data suggest that if longer transcripts are present, the additional sequence is most likely to be at the 5′ end of the molecule.
NCC27 is expressed in each of the diverse human tissues probed, suggesting that the protein carries out a function common to most cell types. In addition, there is markedly enhanced expression in both heart and skeletal muscle. This, plus our observation of intense staining of smooth muscle in human kidney sections, suggests that NCC27 carries out some muscle-specific function. Chloride conductance is known to be important in setting the membrane potential in skeletal muscle, but it is well established that this function is carried out by the ClC-1 chloride channel (11, 12). Chloride has also been implicated in the process of calcium release from the sarcoplasmic reticulum, with one hypothesis being that a chloride conductance may provide a short-circuit pathway that would allow channel-mediated calcium efflux (1). Perhaps NCC27 carries out this function in sarcoplasmic reticulum.
Apparent molecular size of p34. The NCC27 cDNA sequence encodes a protein of 26,901 Da predicted molecular mass, but our antibodies recognize a protein of 34 kDa by SDS-PAGE. We are quite confident that the cDNA sequence and, in particular, the translational initiation and termination sequences are correct. We have sequenced through several independent cDNAs containing these regions that are in perfect agreement and our sequence agrees perfectly with the reported NCC27 sequence in these regions.
Several lines of evidence indicate that the 34-kDa protein recognized by our antibodies is indeed the product encoded by NCC27 cDNA.1) The endogenous 34-kDa NCC27 signal of PancI, HeLa, CHO-1, macrophages, and kidney are blocked by preabsorption with the peptide that was used to raise the antisera.2) The antisera recognizes and was affinity purified over recombinant his-tagged NCC27 expressed in bacteria. 3) Expression of NCC27 in HeLa cells results in a 34-kDa protein recognized by our affinity-purified antisera which comigrates with the endogenous NCC27 antigen. 4) In vitro translation of NCC27 results in a protein of molecular mobility indistinguishable from the native cellular protein. Therefore, we conclude that NCC27 is a 27-kDa protein that shows aberrantly slow migration through SDS-PAGE gels, as has already been demonstrated to be the case for p64 (14). Presumably, some characteristic of the sequences in common between p64 and NCC27 causes this aberrant electrophoretic mobility.
Our NCC27 antibody recognizes a 34-kDa protein, in contrast to the 27-kDa protein reported previously. The evidence outlined above that our antibody is in fact recognizing the product of the NCC27 cDNA is conclusive. The similarity in staining pattern of CHO cells with both our antibody and that of Valenzuaela et al. (28) indicate that the antibodies are recognizing the same protein. It should be noted that we denatured our proteins under reducing conditions prior to electrophoresis, whereas samples were not reduced in the previous report (28). In our experience, electrophoresis under nonreducing conditions does not affect the mobility of NCC27 but does greatly decrease the ability of our antibody to detect the protein. It is possible that our anti-peptide antibody does not recognize a nonreduced form of the protein which has different electrophoretic mobility.
Subcellular localization of NCC27. The subcellular localization of NCC27 was characterized biochemically in kidney. In both human and mouse kidney homogenates, we found that only a fraction of the total NCC27 is present in the membrane fraction, with the remainder partitioning with the soluble proteins in a standard tissue preparation. With human material, NCC27 associated with crude membranes was further characterized by alkaline and detergent extractions of the insoluble fraction. Membrane-associated NCC27 was found to be resistant to alkaline wash (pH 10.5) and solubilized with detergent, behavior consistent with that expected for an integral membrane protein.
Using mouse material, we further fractionated the kidney to obtain membranes enriched for brush border. NCC27 is more abundant in the brush-border fraction than in non-brush-border membranes, and NCC27 in brush border is resistant to alkaline extraction, indicating that it is present in brush border as an integral membrane protein. We subjected unfractionated mouse kidney membranes to more stringent alkaline wash (pH 11.5), again demonstrating alkali-resistant p34 in kidney membranes.
Thus NCC27 shows unusual behavior for a membrane protein: a fraction of the protein behaves as an integral membrane protein, whereas the remainder is soluble. Although such behavior is unusual, it is not without precedent. Bacterial porins shuttle between soluble and membrane-inserted forms (23). The eukaryotic protein pICln, which has been implicated as a cell volume-sensitive chloride channel, has been proposed to shuttle between soluble and membrane-incorporated forms (23). The distribution between soluble and membrane-associated forms of these proteins is a potential mechanism of regulation for an extended group of chloride channels. Whether NCC27 actively shuttles between the cytoplasm and membrane remains to be determined.
Subcellular distribution of NCC27 was probed morphologically by using the affinity-purified antisera to stain cultured cells and kidney sections. In PancI, HeLa, macrophages, and CHO-1 cells, we find NCC27 in a peripheral reticular/vesicular pattern. We have not yet characterized this vesicular p34 compartment any further. As previously noted using different NCC27 antibodies (28), CHO-1 cells show prominent nuclear staining with our affinity-purified antisera, in addition to the cytoplasmic vesicular pattern. The staining pattern suggests presence of the antigen inside the nucleus rather than restricted to the nuclear membrane. To see whether p34 could account for the nuclear membrane chloride channel activity reported by others (28), we probed isolated CHO-1 cell nuclei for NCC27. We failed to find any NCC27 in the nuclear preparation which others have used for electrophysiology. Thus, although the immunohistochemistry convincingly demonstrates NCC27 in nuclei in intact CHO-1 cells, we cannot detect it in isolated nuclei. This preparation of nuclei begins with cell lysis in NP-40. In other cell types, we found that nonionic detergent essentially completely solubilizes cellular NCC27 (not shown). Perhaps the NCC27 is liberated from CHO nuclei by NP-40 extraction, although the previously reported experience (28) indicates that such extraction leaves enough nuclear membrane for patch-clamp experiments.
In the kidney, NCC27 was found to be highly expressed in glomeruli, in vascular smooth muscle, and, most prominently, in the apical membrane of proximal tubule cells. In both human and mouse kidney, we only found prominent apical membrane staining for NCC27 in proximal tubule. However, we found a significant interspecies difference in the pattern of NCC27 expression within proximal tubule. In human kidney, the very deep juxtamedullary proximal tubule showed much more intense staining than more superficial segments, although apical staining was detectable throughout the proximal tubule. Based on position alone, it is likely that the intensely staining cells in human kidney are in the S3 segment of proximal tubule. In contrast, the mouse cells staining most intensely were present in proximal tubules of the superficial cortex, which consists primarily of S1 and S2 segments. In the mouse, we were able to identify the S3 segment with a segment-specific lectin and found that this portion of the mouse nephron also stains for NCC27.
The brush border of the proximal tubule is known to have significant chloride permeability (3, 30), and a number of distinct chloride channels have been described in this membrane (4, 24, 25). Some chloride channel proteins are known to be expressed in this nephron segment, most notably CFTR (16) and some members of the ClC family of chloride channels (11, 12, 27). In most cases, the correlation of one of the observed channel activities with a specific chloride channel protein is not firmly established. The localization of p34 to the apical domain of proximal tubule cells introduces a new entity into the channels known to be present, which could be responsible for some of the observed channel activity. We are not aware of a brush-border chloride channel activity known to show such dramatically distinct distribution along the proximal tubule between mouse and human kidney as is shown by the NCC27 protein.
We also found NCC27 to be prominently expressed in glomeruli and vascular smooth muscle in the kidney. This finding is of particular interest since an indanyloxyacetic acid-inhibited chloride channel has been implicated in endothelin signal transduction in vascular smooth muscle and mesangial cells (10, 26).
We thank Shefalee Kapadia, Amy Driskell, and Joyce Yee for excellent technical assistance. We thank David Seung-Lae Kim for assistance with DNA sequencing.
Address for reprint requests: J. C. Edwards, Renal Division, Barnes-Jewish Hospital, 216 S. Kingshighway, St. Louis, MO 63110.
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant R29-DK-4612 and by a grant form the Barnes-Jewish Hospital Foundation.
- Copyright © 1998 the American Physiological Society