Cloning of an Epithelial Chloride Channel from Bovine Trachea*

We have isolated and cloned a novel epithelial Cl (cid:50) channel protein from a bovine tracheal cDNA expres- sion library using an antibody probe. The antibody ( (cid:97) p38) was raised against a 38-kDa component of a ho-mopolymeric protein that behaves as a Ca 2 (cid:49) /calmodulin kinase II-, DIDS-, and dithiothreitol (DTT)-sensitive, an- ion-selective channel when incorporated into planar lipid bilayers. The full-length cDNA is 3001 base pairs long and codes for a 903-amino acid protein. The clone does not show any significant homology to any other previously reported Cl (cid:50) channel sequence. Northern analysis of bovine tracheal mRNA with a cDNA probe corresponding to the cloned sequence revealed a band at 3.1 kilobases, suggesting that close to the full-length sequence has been cloned. The full-length open reading frame (2712 base pairs) has been expressed in Xenopus oocytes and in mammalian COS-7 cells. In oocytes, expression of the clone was associated with the appear- ance of a novel DIDS-, and DTT-sensitive, anion-selec-tive conductance that was outwardly rectified and exhibited a reversal potential close to 0 mV. Whole-cell patch clamp studies in COS-7 cells transfected with the clone identified an ionomycin-, and DTT-sensitive chloride conductance that was not apparent in mock-trans- fected or control cells. In vitro translation studies have shown that the primary transcript codes for a protein of a novel protein following screening of a bovine tracheal cDNA expression library with polyclonal antibodies generated from the reduced 38-kDa tracheal protein. 0.8 mCi/ml Tran 35 S-label (ICN). Ca- nine pancreatic microsomes determine the of co-translational glycosylation of the termi- prior 8% SDS-PAGE gels (23). electrophoresis, gels on autoradiography Autoradiography intensifying (cid:50) lengths of time. X-ray control. Currents were digitally recorded and filed for later analysis using the pCLAMP program. Seal resistances were assessed at the conclusion of each experiment to ensure the stability of each preparation. Tracheal Cl (cid:50) Channel Incorporation into Planar Lipid Bilayers— Tracheal chloride channel protein was immunopurified from solubilized bovine tracheal membrane vesicles as described previously (15), using a polyclonal rabbit antibody generated against the 38-kDa form of the protein. The purified protein was incorporated into liposomes as de- scribed previously (16) and incorporated into planar lipid bilayer membranes (composed of a mixture of diphytanoyl phosphatidylethanol- amine/diphytanoyl phosphatidylserine/oxidized cholesterol (20 mg/ml) in a 2:1:2 (w/w/w) ratio), in the presence of a symmetrical solution of 100 m M KCl, 10 m M EGTA, and 10 m M MOPS (pH 7). Data analysis was performed as described previously (16).

We have isolated and cloned a novel epithelial Cl ؊ channel protein from a bovine tracheal cDNA expression library using an antibody probe. The antibody (␣p38) was raised against a 38-kDa component of a homopolymeric protein that behaves as a Ca 2؉ /calmodulin kinase II-, DIDS-, and dithiothreitol (DTT)-sensitive, anion-selective channel when incorporated into planar lipid bilayers. The full-length cDNA is 3001 base pairs long and codes for a 903-amino acid protein. The clone does not show any significant homology to any other previously reported Cl ؊ channel sequence. Northern analysis of bovine tracheal mRNA with a cDNA probe corresponding to the cloned sequence revealed a band at 3.1 kilobases, suggesting that close to the full-length sequence has been cloned. The full-length open reading frame (2712 base pairs) has been expressed in Xenopus oocytes and in mammalian COS-7 cells. In oocytes, expression of the clone was associated with the appearance of a novel DIDS-, and DTT-sensitive, anion-selective conductance that was outwardly rectified and exhibited a reversal potential close to 0 mV. Whole-cell patch clamp studies in COS-7 cells transfected with the clone identified an ionomycin-, and DTT-sensitive chloride conductance that was not apparent in mock-transfected or control cells. In vitro translation studies have shown that the primary transcript codes for a protein migrating at 140 kDa under reduced conditions, significantly larger than the polypeptide recognized by ␣p38. We therefore suggest that either the 140-kDa translated product is a prepro form of the 38-kDa subunit of the previously identified bovine tracheal anion channel and that the primary transcript is post-translationally cleaved to yield the final product, or that the cloned channel and the previously identified bovine tracheal anion channel protein share an epitope that is recognized by the ␣p38 antibody.
Recent experimental evidence indicates that epithelia such as the mammalian trachea, kidney, and intestine contain different types of Cl Ϫ channel including those sensitive to cAMP, Ca 2ϩ , volume, and voltage (1). Chloride channels may be lo-cated on both basolateral and apical membranes, where they participate in NaCl transport in both absorptive and secretory cells (2). However, studies of epithelial Cl Ϫ channels at the biochemical and molecular level have been severely limited because of a lack of appropriate pharmacological and molecular probes.
Until very recently, the primary amino acid sequence for many of these anion channels was unknown. The first epithelial Cl Ϫ channel to be cloned was the (CFTR) 1 protein (3). CFTR acts as a small conductance, linear, cAMP/protein kinase Asensitive anion channel that is insensitive to inhibition by DIDS (1,4), and exhibits an anion selectivity of Br Ϫ Ͼ Cl Ϫ Ͼ I Ϫ . Subsequently, Jentsch and co-workers (5) cloned and expressed a voltage-dependent Cl Ϫ channel (ClC-2) from rat heart and brain. This clone encodes a protein with a predicted M r of 99,000 and is 50% homologous to both the Torpedo electroplax Cl Ϫ channel (ClC-0) and rat muscle Cl Ϫ channel (ClC-1; Refs. 6 and 7). The message for this ClC-2 channel was found by Northern analysis in a number of epithelial cell types, including the pancreas, kidney, intestine, and lung, and in several cell lines, such as T 84 cells, Chinese hamster ovary cells, and COS cells, and in the CFPAC-1 cell line. Uchida et al. (8) cloned a rat kidney Cl Ϫ channel (ClC-K1), using degenerate oligonucleotide primers constructed from conserved regions of ClC-0, -1, and -2 in a PCR-based cloning strategy. This channel was primarily expressed in renal thick ascending limb. The currents expressed in oocytes were DIDS-sensitive, had a selectivity of Br Ϫ Ͼ Cl Ϫ Ͼ I Ϫ , and were postulated to be regulated by osmotic stress in vivo. A protein kinase A-sensitive gastric chloride channel, highly homologous to ClC-2, has also been isolated from the rabbit stomach (9), the selectivity of which was reported to be I Ϫ Ͼ Cl Ϫ Ͼ NO 3 Ϫ . Paulmichl et al. (10), using the renal Madin-Darby canine kidney cell line, cloned and expressed a protein termed pICl n , the expression of which was associated with the appearance of an outwardly rectifying, Ca 2ϩ -insensitive, DIDS-and NPPB-blockable Cl Ϫ channel with a predicted M r of 26,000. Unlike CFTR, nucleotides can also block this channel. Landry et al. (11) cloned a 64-kDa protein that functions as a Cl Ϫ channel in intracellular organelles.
Recent experiments using CFTR "knockout" mice have indicated that a Ca 2ϩ -mediated Cl Ϫ secretory pathway is up-regulated in the nasal mucosa of these animals and that Cl Ϫ secretion is therefore a process molecularly distinct from activation of the CFTR pathway in these cftr/cftr mice (12,13). It has been shown that this Ca 2ϩ -dependent Cl Ϫ channel can compensate for the lack of CFTR in the CFTR knockout mouse and thus accounts for the absence of significant airway pathology in this cystic fibrosis mouse model. Transduction of normal CFTR in monolayers of human cystic fibrosis nasal epithelia using adenoviral vectors suppressed Ca 2ϩ -dependent Cl Ϫ secretion, indicating that functional expression of the calcium-mediated pathway is inversely dependent on the expression of functional CFTR (14). Thus, the Ca 2ϩ -dependent Cl Ϫ secretory pathway is of functional importance in human cystic fibrosis and may prove amenable to pharmacological manipulation and amelioration of the airway disease.
Our laboratory has purified and functionally reconstituted a Cl Ϫ channel protein from bovine tracheal epithelium using anion and cation exchange chromatography followed by immunopurification (15,16). We have shown that this protein is phosphorylated by Ca 2ϩ /calmodulin-dependent protein kinase II, and acts as a Ca 2ϩ /calmodulin-dependent protein kinase II-activated anion channel when reconstituted into planar lipid bilayers (17). However, when incorporated into planar lipid bilayers, this channel cannot be activated through phosphorylation by protein kinase A. The native form of this protein is 140 kDa and it forms anion-selective, 25-30 picosiemens channels in 150 mM NaCl (16) with an anion selectivity of I Ϫ Ͼ NO 3 Ϫ Ͼ Br Ϫ Ͼ Cl Ϫ . The channel is inactivated by disulfide reduction procedures, and the reduced form of the channel, which appears as a 36 -38-kDa protein, is not associated with channel activity in planar lipid bilayers. Of the other Cl Ϫ channels purified or cloned to date, none are Ca 2ϩ -dependent, and all exhibit different anion discrimination profiles. We here report the cloning of a novel protein following screening of a bovine tracheal cDNA expression library with polyclonal antibodies generated from the reduced 38-kDa tracheal protein.

Methods
Isolation of cDNA Clones-Bovine tracheal epithelium was scraped from fresh trachea obtained from a local slaughterhouse and immediately added to the lysis buffer of the Fast Track mRNA isolation kit (Invitrogen). Poly(A) ϩ mRNA isolation was then performed according to the manufacturer's instructions. A cDNA library was constructed in the Unizap vector (Stratagene) using 5 g of tracheal poly(A) ϩ mRNA, according to the manufacturer's instructions. A polyclonal antibody (␣p38), raised against a tracheal Cl Ϫ channel (15), was used to screen ϳ 10 6 clones. In the initial screening, nine positive clones were identified and purified by three further rounds of immunoscreening. All positives were excised into pBluescript using R408 helper phage. Doublestranded DNA was subjected to alkaline denaturation and sequenced by a modified Sanger dideoxynucleotide protocol using Sequenase 2.0 (U.S. Biochemical Corp.) and [ 35 S]dATP (1000 Ci/mmol, DuPont NEN). All computer analyses were performed using the Genetics Computer Group sequence analysis software package (Center for AIDS Research, University of Alabama at Birmingham).
PCR Analysis-Bovine tissue RNAs from trachea, lung, brain, renal papilla, and liver were isolated as described (19). RNA from control Xenopus oocytes was isolated by homogenization in buffer containing 1% SDS as described (20). Reverse transcription of 1 g of total isolated RNA was performed using avian myeloblastosis virus-reverse transcriptase (Promega) for 60 min at 42°C using oligo(dT) 15 (Promega) to prime the reaction. Approximately 200 ng of the reverse transcribed cDNA was amplified by PCR for 30 cycles using 2 units of Taq polymerase (Promega)/reaction, under the following conditions: denaturation, 94°C, 45 s; annealing, 58°C, 45 s; extension, 72°C, 2 min. Primers were designed corresponding to a 24-bp region bracketing the initial ATG codon (5Ј-AAAATGGTGCCTCGTCTGACTGTC-3Ј) and to a 21nucleotide region initiating at 704-bp downstream (5Ј-ACTCCCTTG-CAGTCTGGGATT-3Ј). In addition, primers designed to amplify a 150-bp product from bovine ␤-actin were used in parallel reactions as controls. The sense and antisense bovine actin primers were 5Ј-CAAT-GTGCCCATCTATGAGG-3Ј and 5Ј-GCTTCTCTTTGATGTCGCGC-3Ј, respectively. In the case of PCR reactions using transcribed Xenopus RNA, human ␤-actin primers (Stratagene) that amplified a 650-bp product were used for control PCR reactions. PCR products were electrophoresed through 2% agarose (Nu-Sieve, FMC)/Tris-borate-EDTA gels and stained with ethidium bromide.
Subcloning into Xenopus and pMT3 Expression Vectors-A modified version of pGEM11zf Ϫ containing the 3Ј-and 5Ј-untranslated regions of the Xenopus ␤-globin gene with BglII restriction sites 3Ј and 5Ј to the insert (21) was a kind gift of Dr. Doug Melton (Harvard University, Cambridge MA). The putative Cl Ϫ channel insert subcloned into the vector was obtained by PCR of the 2712-bp open reading frame using Vent thermostable DNA polymerase (New England Biolabs), under the following conditions: denaturation, 94°C, 1 min; annealing, 52°C, 1 min; extension, 72°C, 3 min; 30 cycles; and purification by gel electrophoresis. Two adaptors, containing a BglII restriction site were synthesized as follows: 5Ј, 5Ј-GATCTCGTTGCTGTCG-3Ј; 3Ј, 5Ј-CGACAG-CAACGA-3Ј. The sequence of the insert was verified by restriction mapping of the PCR-generated open reading frame and by limited dideoxy sequencing of the double-stranded cDNA at 3Ј and 5Ј ends.
A PCR-based strategy was adopted to subclone the Ca-CC open reading frame into the COS cell expression vector pMT3 (a kind gift of Dr. Chris Miller, Brandeis University, Waltham, MA). Two primers, encompassing either an EcoRI site and a Kozak start translation sequence (5Ј-GG2AATTCCGCCGGCAAAAATGGTGCCTCGTCTGACT-3Ј) or a NotI site (5Ј-ATAGTTTAGC2GGCCGCATTCTTAAACTATAG-ATAAAATCAT-3Ј), were used to amplify a PCR reaction run under identical conditions to those described above. The sequence of the insert cloned into pMT3 was verified by restriction mapping and limited dideoxy sequencing at the 3Ј and 5Ј ends. Ligation with T 4 ligase and transformation into XLI-Blue were all as described previously (22).
In Vitro Transcription and Translation-cRNA was transcribed from the oocyte expression vector using the Ribomax kit from Promega. Approximately 25 g of the pGEM11zf Ϫ vector containing the cDNA insert was linearized with NotI and transcribed using T7 polymerase. The transcription reaction was performed in the presence of a methylguanosine cap structure (m 7 5Ј(G)ppp(G)5Ј, 2.4 mM final concentration), to enhance transcript stability. Transcription was performed for 4 h at 37°C. At the end of this period, 25 units of RQ1-DNase (Promega), was added to the reaction to digest the template DNA, and the reaction continued for a further 30 min at 37°C. The transcribed RNA was checked by electrophoresis through a 1% agarose/formaldehyde denaturing gel. Antisense cRNA was in vitro transcribed from a eukaryotic expression vector, pcDNA I, into which the insert had been subcloned using conventional techniques (20). This vector lacked the 3Ј-and 5Ј-flanking regions of the Xenopus ␤-globin gene. The vector was linearized with FspI and in vitro transcribed with SP6. The antisense transcript was also synthesized with a methylguanosine cap structure. Transcribed sense cRNA (2 g) was in vitro translated at 30°C for 60 min using rabbit reticulocyte lysate pretreated with micrococcal nucle-ase (Promega) in the presence of 0.8 mCi/ml Tran 35 S-label (ICN). Canine pancreatic microsomes were added to determine the extent of co-translational glycosylation of the protein. The reaction was terminated by the addition of SDS sample buffer Ϯ 50 mM DTT. Samples were heated to 95°C for 5 min prior to loading onto 8% SDS-PAGE gels (23). Following electrophoresis, the gels were fixed, fluorographed using EN 3 HANCE (DuPont NEN) according to the manufacturer's instructions, dried, and placed on autoradiography film. Autoradiography was performed (with intensifying screens) at Ϫ70°C using Kodak XAR-5 film for appropriate lengths of time. X-ray film was developed using an automatic processor.
Immunoblot Analysis-A fusion protein was generated by the Center for AIDS Research Core Facility at the University of Alabama Birmingham from a 570-bp BamHI fragment of the Ca-CC cDNA clone (nucleotides 1170 -1740), using the pET 21a/T7 expression system (Novagen) with a C-terminal histidine tag. The resultant 24-kDa polypeptide was denatured with 8 M urea, purified by Ni 2ϩ affinity chromatography, dialyzed, and used to generate a rabbit polyclonal antibody as described previously (24) without further modification of the antigen. Immunoblot analysis of cation exchange-purified bovine tracheal membrane vesicles was performed as described previously (15).
Expression in Xenopus Oocytes-Xenopus oocytes were removed and defolliculated by collagenase digestion as described previously (22). Twenty-four hours after defolliculation, stage V/VI oocytes were injected with either 50 nl of nuclease-free water, or 50 nl of water containing either 25 ng sense or 25 ng sense ϩ 25 ng antisense cRNA. Recording was performed 2 days postinjection in ND-96 Ringer's solution (96 mM NaCl, 2.4 mM KCl, 2 mM CaCl 2 , 1.8 mM MgCl 2 , 5 mM HEPES, pH 7.4), as described previously (22). In some experiments, membrane vesicles were made from oocytes injected with either water or Ca-CC sense cRNA as described previously (25). Vesicles were subsequently fused to the lipid bilayer for physiological recording.
Transfection of COS-7 Cells and Whole-cell Patch Clamp Recording-COS-7 cells were grown to approximately 80% confluency in Dulbecco's modified Eagle's medium containing 2 mM L-glutamine, 1% penicillin/ streptomycin, and 10% fetal bovine serum at 37°C. Cells were transfected using either chloroquine/dextran (26) or Lipofectamine (Life Technologies, Inc.). We had initially subcloned the Ca-CC open reading frame into two eukaryotic expression vectors, pcDNA1 (Invitrogen) and pMT3. However, we were not able to achieve successful transfections with pcDNA1-Ca-CC and thus used the pMT3-Ca-CC construct in all of our subsequent experiments. In the case of chloroquine/dextran transfection, cells were transfected using 20 g of pMT3-Ca-CC/100-mm dish, using 100 M chloroquine and 250 g/ml dextran in medium in the absence of serum and antibiotics. Cells were incubated for 6 h following transfection and then shocked for 3 min with 10% Me 2 SO. Cells were incubated with 100 M chloroquine alone for a further 2 h before being allowed to recover for 2 days in normal medium ϩ 10% fetal bovine serum, 1% penicillin/streptomycin. In the case of transfection with Lipofectamine, 5 g of plasmid ϩ 25 g of lipid were added to 500 l of Dulbecco's modified Eagle's medium (serum and antibiotic-free) and allowed to incubate for 30 min at room temperature. After this time, the lipid/DNA mix was diluted 5-fold with Dulbecco's modified Eagle's medium and added to a 100-mm dish of cells. Cells were incubated at 37°C overnight and then fed with Dulbecco's modified Eagle's medium (containing 10% fetal bovine serum, 1% antibiotics) for 2 days prior to electrophysiological recording. Mock-transfected cells were transfected with pMT3 plasmid without insert, and control cells were carried through the transfection protocol in the absence of exogenous DNA.
Whole-cell patch clamp recording was carried out using standard techniques. Briefly, the cells were gently scraped, washed in serum-free medium, and added to a chamber placed on the stage of an inverted microscope. The pipette solution contained 112 mM N-methyl-D-glucamine-Cl, 1 mM EGTA, 0.366 mM CaCl 2 , 2 mM MgCl 2 , 5 mM HEPES, pH 7.2. The bath solution contained 112 mM N-methyl-D-glucamine-Cl, 30 mM sucrose, 1 mM CaCl 2 , 2 mM MgCl 2 , 5 mM HEPES, pH 7.2. Thus, chloride was the main conductive ion. All experiments were performed at room temperature (24 Ϯ 2°C). Filled pipettes were fitted to a suction line and an EPC-7 (List Electronics, Darmstadt, Germany) patch clamp amplifier headstage. Liquid junction potentials were compensated and checked for stability on immersion of the pipette into the bath. The patch pipette was placed onto the cell, and suction was applied until a seal resistance of Ͼ 5 G⍀ was achieved. After the seal was formed, a sharp suction pulse was applied to form the whole-cell configuration. The membrane potential was clamped at 0 mV and pulsed between Ϯ 100 mV in 20-mV increments at 500-ms intervals under computer control. Currents were digitally recorded and filed for later analysis using the pCLAMP program. Seal resistances were assessed at the conclusion of each experiment to ensure the stability of each preparation.
Tracheal Cl Ϫ Channel Incorporation into Planar Lipid Bilayers-Tracheal chloride channel protein was immunopurified from solubilized bovine tracheal membrane vesicles as described previously (15), using a polyclonal rabbit antibody generated against the 38-kDa form of the protein. The purified protein was incorporated into liposomes as described previously (16) and incorporated into planar lipid bilayer membranes (composed of a mixture of diphytanoyl phosphatidylethanolamine/diphytanoyl phosphatidylserine/oxidized cholesterol (20 mg/ml) in a 2:1:2 (w/w/w) ratio), in the presence of a symmetrical solution of 100 mM KCl, 10 mM EGTA, and 10 mM MOPS (pH 7). Data analysis was performed as described previously (16). RESULTS A polyclonal antibody (␣p38), raised against a putative Cl Ϫ channel protein isolated from bovine trachea (15), was used to screen approximately 10 6 clones in a bovine cDNA expression library. Nine positives were identified, purified by three additional rounds of screening, and sequenced. The inserts were excised from pBluescript with XhoI and NotI. Dideoxy sequencing of the double-stranded DNA revealed that all of the clones were derived from a single mRNA species but differed in size. The full-length sequence (Fig. 1) consists of 3001 bases of which 255 bases comprise a 3Ј-untranslated sequence followed by a poly(A) ϩ tail. A typical polyadenylation signal sequence (AATAAA), found at bases 13-18 upstream of the poly(A) ϩ tail, suggests that this is not an internally A-rich region. Although we did not locate an upstream stop codon, the ATG at position 19 conforms to a conserved Kozak sequence, with an A at the Ϫ3 position and a G at the ϩ4 position. One long reading frame extends from base 19 of the full sequence (ATG) that encodes methionine and extends to a TAA stop codon at base 2730, coding for a 903-amino acid protein and predicting a primary translation product of 100 kDa. Motif analysis of the predicted amino acid sequence predicts consensus phosphorylation sites for protein kinase A (2 sites), protein kinase C (15 sites), Ca 2ϩ / calmodulin-dependent protein kinase (10 sites), and tyrosine kinase (3 sites). In addition, 12 potential sites for N-linked glycosylation are present. This clone does not show any significant homology to any other sequence in the GenBank TM data base, including the ClC family of chloride channels and p64 (5)(6)(7)(8)(9)11). Hydropathy analysis of the full-length Ca-CC open reading frame using an analysis window of 19 residues, revealed four major potential transmembrane-spanning domains, as well as several minor regions, consistent with a membrane protein (Fig. 2).
In order to determine if the full-length message from which the clones were derived was isolated, Northern blot analysis of bovine tracheal mRNA was carried out under high stringency conditions. As shown in Fig. 3A, a positively hybridized mRNA signal was detected at 3.1 kilobases, suggesting that close to a full-length sequence was cloned. In addition, RT-PCR analysis of mRNA from five different bovine tissues (trachea, lung, liver, brain, and renal papilla) and from Xenopus oocytes was performed to determine if the cloned cDNA was exclusively expressed in bovine trachea. As shown in Fig. 3B, following 30 cycles of PCR, signal of the predicted size (704 bp) was detected only in RNA extracted from the trachea. The appropriate genomic controls were also negative. All other bovine tissues examined, as well as RNA from Xenopus oocytes, were negative using these primers, although the integrity of the RNA from these samples was maintained, as evidenced by the amplification of a 150-bp fragment of bovine ␤-actin from the bovine samples and a 650-bp product using human ␤-actin primers from Xenopus RNA. Following an additional 30 cycles of PCR, RNA samples that had been reverse-transcribed with avian myeloblastosis virus-reverse transcriptase remained negative (results not shown). Although only detected in the trachea among the bovine tissues tested, the Ca-CC clone did crosshybridize with DNA from mouse, rat, dog, rabbit, monkey, and human, as shown in Fig. 3C, suggesting that the clone is conserved between species.
In Vitro Transcription and Translation-In order to determine whether the tracheal cDNA could encode a functional chloride channel protein, the cDNA was subcloned into an oocyte expression vector. This vector was a modified form of pGEM11zf Ϫ and included the 5Ј-and 3Ј-untranslated region of the Xenopus ␤-globin gene flanking the coding region of the tracheal cDNA insert. The Ca-CC insert was modified to include BglII restriction sites 3Ј and 5Ј to the coding region, to facilitate subcloning into the vector. Sense cRNA for injection into Xenopus oocytes was in vitro transcribed from the vector using the T7 promoter. In vitro translation of the cRNA yielded a major polypeptide product that migrated at 100 kDa on SDS-PAGE (Fig. 4). In the presence of canine pancreatic microsomes, the M r of the polypeptide shifted to 140,000. Several smaller protein bands, presumably equivalent to partial transcripts were also detected by autoradiography. Neither the 100nor 140-kDa polypeptide products were reduced to smaller forms by treatment with 50 mM DTT. Antisense cRNA, without the Xenopus ␤-globin-flanking regions, was in vitro transcribed from a separate vector (pcDNA I), using the SP6 promoter.
Immunoblot Analysis-In order to further characterize the relationship between the cloned Ca-CC and the native bovine tracheal protein, we performed an immunoblot analysis using a rabbit polyclonal antibody raised against a fusion protein generated from the Ca-CC cDNA sequence to probe tracheal proteins partially purified by cation exchange of solubilized bovine tracheal apical membrane vesicles separated by SDS-PAGE. As shown in Fig. 5, two separate polyclonal antibodies raised against the 38-kDa component of the native tracheal channel recognized a band migrating at approximately 36 -38,000 under reduced conditions (50 mM DTT), in the semipurified tracheal material, consistent with our earlier observations (15). Similarly, the polyclonal antibody generated against the Ca-CC fusion protein also recognized a polypeptide migrating with an M r of 38,000 in the cation exchange-purified tracheal material, suggesting that both the native tracheal anion channel and the Ca-CC protein share a high degree of immunological identity.
Expression in Xenopus Oocytes-Expression studies were performed in stage V/VI oocytes isolated from Xenopus laevis and defolliculated by collagenase digestion. Oocytes were injected with water, 50 ng of sense, or 25 ng of sense ϩ 25 ng of antisense cRNA and then incubated at 18°C for 48 h prior to recording. Whereas the whole-cell current records of oocytes injected with sense ϩ antisense cRNA were indistinguishable from records made from water-injected oocytes, oocytes injected with sense cRNA alone displayed an increased current at every potential in the absence of calcium ionophore (Fig. 6). At a holding potential of Ϫ100 mV, the mean current (in pA Ϯ S.E.) was Ϫ96 Ϯ 25 (n ϭ 14), in water-injected oocytes, Ϫ157 Ϯ 46 (n ϭ 7) in sense ϩ antisense-injected oocytes, and Ϫ2052 Ϯ 528 (n ϭ 8) in oocytes injected with Ca-CC sense cRNA. At a holding potential of ϩ80 mV, the mean current (in pA Ϯ S.E.) for water-injected oocytes was 457 Ϯ 209 (n ϭ 14); for sense ϩ antisense-injected oocytes, 847 Ϯ 278 (n ϭ 7); and for senseinjected oocytes, 8084 Ϯ 1700 (n ϭ 8). The I/V curve (Fig. 7) was outwardly rectified at higher positive voltages and reversed close to 0 mV, consistent with an anion-selective channel. Fur- In the presence of microsomes, the M r of the major polypeptide shifted to 140,000 (lane 3). Both major protein products were insensitive to reduction by 50 mM DTT (lanes 2 and 4).

FIG. 5. Immunoblot of semipurified bovine tracheal vesicles.
Apical membrane vesicles from bovine trachea were partially purified by cation exchange prior to separation by reducing (ϩ50 mM DTT) 8% SDS-PAGE and transfer to nylon membranes. The blot was probed with either of two different rabbit polyclonal antibodies raised against the 36 -38-kDa component of the bovine anion channel (lanes 1 and 2) or with an antibody generated against a 24-kDa fusion protein derived from the Ca-CC cDNA (lane 3). All three antibodies identified a polypeptide from the tracheal material that migrated with an M r of approximately 36,000 -38,000. ther elevation of [Ca 2ϩ ] i by 1 M ionomycin resulted in activation of the endogenous Ca 2ϩ -activated Cl Ϫ current that could be blocked by niflumic acid (results not shown).
The expressed current was also sensitive to 100 M DIDS (Fig. 8A) and to 1 mM DTT (Fig. 8B), consistent with our previous observations of an anion-selective channel isolated from bovine trachea (16). DIDS, at a holding potential of Ϫ100 mV, decreased the whole-cell current to 61 Ϯ 8% (S.E.) of the pre-DIDS control value, and to 55 Ϯ 7% at a holding potential of ϩ80 mV (n ϭ 4). The reducing agent DTT decreased wholecell current at Ϫ100 mV and ϩ 80 mV to 73 Ϯ 6% and 78 Ϯ 6% of control, respectively (n ϭ 5). In other experiments where we have expressed cRNAs of either epithelial Na ϩ channels or CFTR in oocytes, neither DTT nor DIDS had any effect on the magnitude of the observed currents (results not shown). In contrast, in three separate experiments, the expressed current was not sensitive to 100 M niflumic acid (Fig. 9A), a compound previously identified to inhibit the endogenous Ca 2ϩ -activated Cl Ϫ channel of Xenopus oocytes (18). In the oocyte, the apparent K i for niflumic acid inhibition of the endogenous Ca 2ϩactivated Cl Ϫ channel is 17 M (18). Similarly, the tracheal anion channel, when reconstituted into planar lipid bilayers, was also insensitive to this compound (Fig. 9B). At a holding potential of ϩ40 mV under control conditions, the P o of the bovine tracheal Ca-CC incorporated into the lipid bilayer was 0.38 Ϯ 0.04 (n ϭ 9). Following the addition of 100 M niflumic acid to both the cis and trans sides of the bilayer, the P o remained at 0.36 Ϯ 0.07 (n ϭ 4). Furthermore, and consistent with our previous observations (17), this channel was apparently insensitive to a cAMP elevating mixture of 10 M forskolin and 1 mM isobutylmethylxanthine when expressed in Xenopus oocytes (data not shown).

Whole-cell Patch Clamp Recordings in Ca-CC Transfected COS-7 Cells-In addition to examining Ca-CC expression in
Xenopus oocytes, we have also expressed the Ca-CC clone in the simian renal cell line, COS-7. As judged visually by light microscopic inspection of cells transfected with a vector containing the lacz gene behind the CMV promoter (pCMV␤gal, a kind gift of Dr. E. Sorscher, Gregory Fleming James Cystic Fibrosis Research Center, University of Alabama at Birmingham) under identical conditions to those used for the transfection of the Ca-CC clone and developed with 5-bromo-4-chloro-3-indoyl ␤-Dgalactoside (Life Technologies, Inc.), at least 30% of the cells were transfected. Consistent with this observation, out of 11 transfected cells patched, 9 showed ionomycin-sensitive increases in whole-cell chloride current. In contrast, in five nontransfected cells (either control or mock-transfected), we never observed activation of ionomycin-sensitive currents. The 2 value for these observations was 6.32, suggesting that the difference in frequency was highly significant (0.02 Ͼ p Ͼ 0.01). As shown in Fig. 10, A and B, chloride currents in neither the control nor the mock-transfected cells were affected by 2 M ionomycin. In contrast, the transfected cells exhibited increased currents in the presence of the calcium ionophore (Fig.  10C). The increased current could be reversed to near preionophore levels by a 20-min washout. Fig. 11 illustrates the wholecell current/voltage relationships before and after cell exposure to ionomycin together with the appropriate difference currents in transfected cells (

panel B) and in control cells (panel A).
Perfusion of the reducing agent DTT (4 mM) through the bath following the addition of ionophore, completely abolished the increase in current due to ionomycin in transfected cells (Fig.  11C). Under whole-cell conditions, the I/V curve was linear and reversed at 0 mV (Fig. 12). Seal resistances were Ͼ1 G⍀. Therefore, the leak (which also reverses at 0 mV) was small as seen in the preionomycin currents.
Planar Bilayer Recording of Vesicles Prepared from Ca-CCexpressing Oocytes-In order to characterize further the Ca-CC channel, we prepared membrane vesicles from oocytes injected with Ca-CC cRNA and fused these vesicles to the lipid bilayer. As shown in Fig. 13, the channel observed was insensitive to niflumic acid (100 M), as previously determined for the native channel biochemically isolated from the tracheal epithelium (Fig. 8B). Under control conditions (100 mM symmetrical KCl), the P o was 0.41 Ϯ 0.07 (mean Ϯ S.D., n ϭ 24), and the single channel conductance was 21 picosiemens. In the presence of 100 M niflumic acid, the P o was 0.42 Ϯ 0.06 (n ϭ 24). However, the channel could be activated by increasing [Ca 2ϩ ], but only from one side of the bilayer, (i.e. the side opposite from which FIG. 6. Representative whole-cell currents of water, sense, or sense ؉ antisense cRNA-injected Xenopus oocytes. Oocytes were injected with water, 50 ng of sense, or 25 ng of sense ϩ 25 ng of antisense Ca-CC cRNA. The holding potential was stepped in 20-mV increments at 500-ms intervals from Ϫ100 mV to ϩ80 mV.

FIG. 7. Mean current/voltage relationship for control and
Ca-CC cRNA-injected oocytes. The I/V curve in the Ca-CC-injected oocyte was outwardly rectified, reversed close to 0 mV, and displayed an increased current at every potential. Mean current was calculated from the averaged value of the last five sample points at the end of the 500-ms pulse.
DIDS inhibits); in the presence of 10 M Ca 2ϩ , the P o increased to 0.60 Ϯ 0.08 (n ϭ 24). These results agree well with our previous observations of the Ca 2ϩ -activated chloride channel isolated from the bovine trachea, which, in the absence of phosphorylation, exhibited a P o of 0.55 in the presence of 10 M Ca 2ϩ (17). The addition of either 10 M DTT or 100 M DIDS reduced the observed P o of the Ca 2ϩ -activated channel to 0.15 Ϯ 0.06 (n ϭ 6) and 0.06 Ϯ 0.02 (n ϭ 5), respectively. Using membrane vesicles isolated from Ca-CC-expressing oocytes, we were also able to determine the ion selectivity of the channel. The channel was both anion-selective (P A Ϫ:P C ϩ ϭ 8:1, n ϭ 7, rev ϭ 53 Ϯ 2 mV), and selective for iodide over chloride (P I :P Cl ϭ 3:1, n ϭ 4, rev ϭ 27 Ϯ 2 mV), as determined under biionic conditions. These findings are also consistent with our previ-  ously reported observations for a native Ca 2ϩ -activated chloride channel purified from the bovine trachea, which exhibited a single channel conductance of 25-30 picosiemens (in symmetrical 150 mM KCl) and showed an I Ϫ Ͼ Cl Ϫ anion selectivity of 2.1:1 (16,17).
In approximately one out of every five vesicle incorporations, we observed a channel of similar conductance to the Ca-CC that was also activated by Ca 2ϩ . However, this channel was in all cases sensitive to niflumic acid (included routinely in all experiments) and was observed in vesicles isolated from waterinjected oocytes. The sensitivity of the channel to niflumic acid suggests that the vesicles contained the Ca 2ϩ -activated Cl Ϫ conductance endogenous to the oocyte. DISCUSSION We have previously described the purification and electrophysiological characteristics of an epithelial anion channel isolated from the bovine trachea. Prominent among the channel properties exhibited by this protein are its activation by calcium-mediated agonists (as opposed to cAMP-mediated agonists) and its sensitivity to the reducing agent dithiothreitol (15,17). Using an antibody raised against this protein to screen a bovine tracheal cDNA expression library, we have isolated a candidate cDNA clone. Many of the features exhibited by the cDNA sequence are consistent with a membrane protein regulated by calcium rather than by protein kinase A, e.g. consensus phosphorylation sequences for protein kinase C and calmodulin-dependent protein kinase II, but few for protein kinase A, and consensus sites for N-linked glycosylation. Furthermore, expression of the full-length cDNA in either Xenopus oocytes or mammalian COS-7 cells is associated with the appearance of a Ca 2ϩ -activated anion channel that shares many of the features previously associated with the anion channel purified from the bovine trachea, e.g. activation by calcium, sensitivity to DTT and DIDS, and insensitivity to niflumic acid, even at concentrations 5 times the apparent K i of this compound in the oocyte (18). In addition, when the cDNA clone was expressed in COS-7 cells, the protein exhibited a linear I/V relationship, a characteristic also previously attributed to the purified bovine channel protein when it was incorporated into planar lipid bilayers. In contrast, when expressed in Xenopus oocytes, the I/V curve for the Ca-CC channel was found to be outwardly rectified. This phenomenon has, however, also been observed when CFTR, which also behaves as a linear anion channel when incorporated into planar lipid bilayers or studied under patch clamp recording conditions (1,27), is expressed in Xenopus oocytes (28).
Although the channel expressed by the Ca-CC cDNA clone and the anion channel protein purified from the bovine trachea have many features in common as assessed functionally, one important divergence resides in the different biochemical behaviors of the purified protein as compared with that translated in vitro from the Ca-CC clone. We have previously demonstrated that the anion channel protein purified from the bovine trachea is in its native state a 140-kDa protein and in this form acts as a functional channel when incorporated into planar lipid bilayers (16). When subjected to chemical reduction by the reducing agent DTT, the mobility of the purified protein on SDS-PAGE shifts to 36 -38 kDa, although an intermediate form migrating at 60 -64 kDa can also be observed (15,16). In the reduced form, the protein no longer forms a func- In either control (A) or mock-transfected (B) COS-7 cells, perfusion of 2 M ionomycin had no effect on whole-cell chloride current. In contrast, in cells transfected with the pMT3-Ca-CC construct (C), ionomycin (2 M) increased whole-cell chloride current. This increase could be partially reversed, following a 20-min washout of the ionophore.
FIG. 11. Whole-cell currents in control (A) and Ca-CC-transfected (B and C) COS-7 Cells. Perfusion of 2 M ionomycin into the bath had no effect on whole-cell current recorded from control cells (A). In contrast, in Ca-CC transfected cells, perfusion of ionomycin evoked an increase in whole-cell current (B). The ionomycin-sensitive currents were abolished by the addition of the reducing agent DTT (4 mM) into the bath (C). tional ion channel. Furthermore, ion channel activity associated with the 140-kDa protein incorporated into planar bilayers is markedly inhibited when DTT is added to the bilayer (16). In contrast, the polypeptide product translated from the Ca-CC clone migrates with an M r of 100,000, which shifts to 140,000 following co-translational core glycosylation in the presence of pancreatic microsomes. The relative mobilities of these polypeptide products on SDS-PAGE is not altered by reduction with DTT, even at concentrations as high as 50 mM.
The underlying reason for this difference in biochemical properties between the purified protein and that translated in vitro from the cDNA clone is not immediately apparent. As the clone was initially identified using an antibody raised against the 38-kDa component of the bovine tracheal anion channel, and given the many similarities between the purified protein and the translated polypeptide, it seems likely that the two are related. Moreover, the fact that an antibody raised against a fusion protein generated from the Ca-CC cDNA only recognizes the 36 -38-kDa protein supports this conclusion. One possibility is that the cloned protein is post-translationally cleaved from a 140-kDa initial transcript to the smaller 38-kDa protein, which is then reassembled to form the functional channel, which coincidentally also migrates at 140 kDa. Similar processing mechanisms have been described for several membrane proteins and receptors (29 -31) but have not yet been identified for ion channels. In this regard, the amino acid sequence predicted by the Ca-CC cDNA also predicts several consensus sites for monobasic amino acid cleavage (32). However, a second possibility is that the clone and the purified bovine protein are unrelated but share some epitope identity that is recognized by both the antibody used to screen the cDNA library and the antibody generated against the Ca-CC fusion protein. Resolution of the exact relationship between the clone and the purified protein will require microsequencing of the purified bovine tracheal anion channel protein.
In summary we have identified a cDNA clone from a bovine tracheal cDNA expression library that encodes a protein that behaves as a Ca 2ϩ -activated anion channel when expressed in either Xenopus oocytes or in mammalian COS-7 cells. Importantly, this newly identified channel shares little homology (less than 40%), with the ClC family of voltage-sensitive anion channels identified by Jentsch et al. (5)(6)(7) and, based on limited RT-PCR analysis of the cDNA and sensitivity of the expressed current to niflumic acid, is separate from the Ca 2ϩ -activated Cl Ϫ conductance endogenous to the Xenopus oocyte. The expressed channel displays identical functional characteristics as a Ca 2ϩ /calmodulin-dependent protein kinase II-sensitive, anion-selective channel previously identified in bovine tracheal membrane vesicles (15,16). This channel may thus provide a valuable alternative pathway for pharmacological manipulation of chloride secretion in cystic fibrosis.