Molecular and Functional Characterization of a Calcium-sensitive Chloride Channel from Mouse Lung*

A protein (mCLCA1) has been cloned from a mouse lung cDNA library that bears strong sequence homology with the recently described bovine tracheal, Ca2+-sensitive chloride channel protein (bCLCA1), bovine lung endothelial cell adhesion molecule-1 (Lu-ECAM-1), and the human intestinal Ca2+-sensitive chloride channel protein (hCLCA1). In vitro, its 3.1-kilobase message translates into a 100-kDa protein that can be glycosylated to an approximately 125-kDa product. SDS-polyacrylamide gel electrophoresis from lysates of mCLCA1 cDNA-transfected transformed human embryonic kidney cells (HEK293) reveals proteins of 130, 125, and 90 kDa as well as a protein triplet in the 32–38 kDa size range. Western analyses with antisera raised against Lu-ECAM-1 peptides show that the N-terminal region of the predicted open reading frame is present only in the larger size proteins (i.e. 130, 125, and 90 kDa), whereas the C-terminal region of the open reading frame is observed in the 32–38 kDa size proteins, suggesting a posttranslational, proteolytic processing of a precursor protein (125/130 kDa) into 90 kDa and 32–38 kDa components similar to that reported for Lu-ECAM-1. Hydrophobicity analyses predict four transmembrane domains for the 90-kDa protein. The mCLCA1 mRNA is readily detected by Northern analysis and byin situ hybridization in the respiratory epithelia of trachea and bronchi. Transient expression of mCLCA1 in HEK293 cells was associated with an increase in whole cell Cl−current that could be activated by Ca2+ and ionomycin and inhibited by 4,4′-diisothiocyanatostilbene-2,2′-disulfonic acid, dithiothreitol, and niflumic acid. The discovery of mCLCA1 opens the door for further investigating the possible contribution of a Ca2+-sensitive chloride conductance to the pathogenesis of cystic fibrosis.

order that is caused by mutations in the CF gene encoding the cystic fibrosis transmembrane conductance regulator CFTR (1,2). Such mutations generally lead to loss of CFTR protein from the plasma membrane of affected cells (3), causing abrogation not only of the primary CFTR function of a cyclic AMP-regulated, linear, small conductance chloride channel (4), but also of the CFTR regulatory control over sodium and other chloride channels (5)(6)(7)(8)(9)(10). Although the extent of the CFTR protein defect, which underlies a particular CFTR mutation, is likely to be uniform throughout all the affected tissues of a CF patient, the severity of the CF pathology commonly varies from one tissue to another (3). This heterogeneity in the degree of disease is mirrored in CFTR (Ϫ/Ϫ) mice (11)(12)(13)(14), strongly suggesting that an alternative, non-CFTR-regulated chloride channel activity might account for attenuating CF disease in some tissues. Indeed, when tissues of CFTR (Ϫ/Ϫ) mice were subjected to detailed measurements of chloride conductance, it was observed that epithelia with mild CF pathology (e.g. airway epithelia, pancreas), but not epithelia with severe CF pathology (e.g. small and large intestines), expressed such an alternative, non-CFTR-mediated chloride conductance (14,15).
Three structurally distinct classes of chloride channel proteins have been identified as potential alternates to CFTR (reviewed in Refs. 16 and 17). The first class includes the superfamily of ligand-gated chloride channels (e.g. the glycine and GABA A receptors) (18,19). These chloride channels typically have four transmembrane domains and require assembly to pentamers for proper function. The second class of chloride channels comprises the ClC gene family (20). ClC channels may have as many as twelve transmembrane domains and predominantly function as voltage-gated channels. Their tissue expression varies greatly from one ClC member to another, with mutations in these genes accounting for diverse chloride conductance disorders. Finally, there is a newly emerging family of Ca 2ϩ -sensitive chloride channels (CLCAs). The first member of this group is the bovine tracheal epithelium CLCA (termed here bCLCA1) (17). Two related proteins have since been cloned in our laboratory and their unique posttranslational protein processing characterized in detail. These proteins include the lung endothelial cell adhesion molecule Lu-ECAM-1 (21), which is primarily expressed in bovine aortic endothelial cells and lung venular endothelium (22), and the human intestinal hCLCA1 (23).
Here, we perform homology-based screening of a mouse lung cDNA library using a bovine Lu-ECAM-1 probe (21). A 3.1-kb cDNA is identified and found to be highly homologous to bovine Lu-ECAM-1, bovine tracheal bCLCA1, and human intestinal hCLCA1 (17,21,23). The open reading frame encodes a 100-kDa protein product that upon expression in HEK293 cells produces glycosylated protein products of 130 and 125 kDa. These protein products are posttranslationally processed into 90-and 38/32-kDa glycosylated components in a manner identical to Lu-ECAM-1 (21). Transfection of mCLCA1 into HEK293 cells combined with whole cell patch-clamp recording confirmed expression of a novel Ca 2ϩ -sensitive Cl Ϫ conductance. The resemblance between the tissue expression patterns of mCLCA1 and CFTR supports previous electrophysiological data for the existence of two (and possibly more) distinct independently regulated chloride channels in the same cell type and underscores the potential importance of CLCAs in CF (24).

MATERIALS AND METHODS
Screening of a Mouse cDNA Library and Sequencing-A mouse lung cDNA library (Stratagene) was screened with the EcoRI-BglII fragment of the Lu-ECAM-1 cDNA (2.4 kb) (21). Hybridization was performed at 65°C in 5ϫ SSC, 5ϫ Denhardt's solution, and 0.2% SDS solution overnight with agitation. Blots were washed in 2ϫ SSC, followed by several washes in 0.2ϫ SSC, 0.2% SDS at room temperature for a total time of 30 min. Subclones of each positive clone were isolated by the in vivo excision protocol provided by the manufacturer of the cDNA library (Stratagene). Automated sequencing was performed by the Cornell University DNA sequencing facility using dRhodamine Terminator cycle sequencing on an ABI Prism 377 DNA sequencer (PE Applied Biosystems). The BLAST program was used for homology searches in existing data bases (25), and the Megalign of the DNAStar package (Lasergene) for multiple sequence alignment.
Construction of Full-length Mouse mCLCA1 cDNA-5Ј rapid amplification of cDNA ends (5Ј RACE kit; Life Technologies, Inc.) was used to clone the 5Ј end of the mouse cDNA from a pool of mouse lung poly(A) ϩ RNA (CLONTECH). A gene-specific primer 5Ј-GAA CCT TGC CAG GGG CCG-3Ј (nucleotides 2367 to 2350) was employed to reversely transcribe the cDNA from mouse lung mRNA. A second nested genespecific primer 5Ј-CCA CGT GCT TCT GCG ATT GCA C-3Ј (nucleotides 836 to 875) and a primer recognizing the 5Ј-terminal tag served to polymerase chain reaction amplify the 5Ј end of the cDNA. Polymerase chain reaction products were cloned into the pGEM-T vector (Promega). A full-length chimera was generated by fusing the RACE product clone with the longest cDNA clone in pBluescript (Stratagene), using the unique PmlI restriction site in the shared overlapping region. To confirm the existence of the resulting contiguous ORF in lung, primers corresponding to its ends were used to amplify the entire ORF for direct sequencing.
Northern Analysis and in Situ Hybridization-A mouse multiple tissue Northern blot (CLONTECH) was probed with a radioactively labeled HindIII fragment from the mCLCA1 ORF (2.2 kb). Hybridization was done at 65°C in 5ϫ SSC, 5ϫ Denhardt's solution, and 0.2% SDS solution overnight with agitation. Blots were washed in 2ϫ SSC followed by several washes in 0.2ϫ SSC, and 0.2% SDS solution at room temperature for a total time of 30 min.
In situ hybridization was performed on formalin-fixed murine lung and trachea sections with single-stranded digoxigenin-labeled sense or antisense RNA probes (24). Bound probe was detected by alkaline phosphatase-conjugated antidigoxigenin antibodies and nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate as substrates (Boehringer Mannheim). Sections were counterstained with methyl green. Controls were as described previously by our laboratory (26).
In Vitro Translation-The TNT TM coupled transcription and translation system (Promega) was employed for the in vitro expression of full-length mCLCA1 cDNA. Canine microsomes were used to glycosylate the 35 S-labeled translation product according to Stratagene's recommended protocol. 5-l aliquots of each of the 50-l translation reactions were run on a 10% SDS-PAGE gel.
Transfection of HEK293 Cells-mCLCA1 cDNA was cut from the pBluescript vector (Stratagene) using the restriction enzymes SacI and PvuI. The 3Ј exonuclease activity of the Klenow polymerase was employed to generate blunt ends, and the resulting fragment was inserted into the tetracycline-sensitive mammalian expression vector, pTetsplice (Life Technologies, Inc.) at the EcoRV site. HEK293 cells were cotransfected with mCLCA1 cDNA cloned into pTet-splice, along with the vector expressing the tetracycline activator pTet-tTak, using the LipofectAMINE TM method (Life Technologies, Inc.). For electrophysiological studies, HEK293 cells were also cotransfected with pEGFP1 encoding green fluorescent protein (GFP) (CLONTECH). Cells were analyzed 48 h after initiation of transfection.
Protein Analysis and Western Blotting-Proteins prepared by in vitro translation or from lysates of mCLCA1-transfected HEK293 cells were analyzed by standard SDS-PAGE. Western blots were probed with rabbit polyclonal antibodies CU8 and CU21, generated against anti-Lu-ECAM-1 peptides (21), and used at a dilution of 1:1000 in phosphatebuffered saline. They were developed with enhanced chemiluminescence reagent for detecting horseradish peroxidase activity (Amersham Pharmacia Biotech).
Whole Cell Patch Recording-Whole cell channel activity was recorded on mCLCA1-transfected HEK293 cells, superfused at 1-2 ml/ min with bath solution (112 mM N-methyl-D-glucamine-Cl, 30 mM sucrose, 2 mM CaCl 2 , 2 mM MgCl 2 , 5 mM HEPES, pH 7.4) in the presence or absence of DIDS (300 M), NFA (100 M), or DTT (2 mM). Borosilicate glass electrodes (tip resistance: 4 -9 megaohms) were filled with bath solution containing either 5 mM ATP alone or 5 mM ATP and 1 mM EGTA in the presence of low intracellular Ca 2ϩ concentrations (336 M: free [Ca 2ϩ ] calculated at ϳ25 nM) in experiments designed to examine the effects of ionomycin (2 M) on channel activity. After seal formation at Ͼ1 gigaohm and establishment of the whole cell recording configuration, cells were clamped at ϩ20 mV. Whole cell currents were recorded at room temperature using an Axopatch 200A (Axon Instruments) connected via a TL1 interface (Axon) with a 12-bit resolution to a PC. Records were sampled at 5-10 kHz and filtered at 1-2 kHz with a four-pole Bessel filter. The I-V relationship of mCLCA1 was determined using 300-ms voltage steps from a holding potential of ϩ20 mV to potentials from Ϫ100 to ϩ100 mV at 10 mV intervals. To normalize the measured membrane currents to the membrane capacitance, the capacitative current transiently recorded in response to a 10 mV hyperpolarizing pulse was integrated and divided by the given voltage to yield the total membrane capacitance (C m ) for each cell.

RESULTS
Cloning and Sequence Analysis of mCLCA1-Low stringency hybridization conditions were employed in the screening of a mouse lung cDNA library with the ORF of the bovine Lu-ECAM-1 cDNA as probe. Positive phages were purified and analyzed by Southern blot hybridizations to verify specificity. The largest of the isolated cDNA fragments was 2.2 kb in length. It lacked the 5Ј end, as determined from sequence comparison with the bovine homolog. The missing 0.9 kb of the mCLCA1 cDNA was obtained using 5Ј-RACE. The full-length mouse mCLCA1 cDNA was assembled by fusing the RACE product to the 2.2-kb cDNA insert. It is 3.1-kb long and encodes a polypeptide of 902 amino acids (Fig. 1). The mCLCA1 amino acid sequence is 71% identical to both of the previously described bCLCA1 (17) and Lu-ECAM-1 (21) protein sequences and 53% identical to human intestinal hCLCA1 protein (23). Moreover, the mCLCA1 sequence between amino acids 308 and 365 is 57% identical to an expressed sequence tag from porcine small intestine (27). Hydropathy analysis of the mCLCA1 ORF predicts it to be a transmembrane protein with at least two and possibly four membrane-spanning domains (data not shown). A four-transmembrane structure is consistent with the extracellular location of five putative glycosylation sites of mCLCA1 and has been suggested for bCLCA1, Lu-ECAM-1, and hCLCA1 (17,21,23).
The sequence alignment of the four members of the CLCA gene family isolated so far indicates conservation throughout the entire length of the sequence, without the compartmentalization of more conserved domains (Fig. 1). However, the Nterminal extracellular domain contains a highly conserved spacing of five cysteine residues, i.e. CX 12-13 CX 4 CX 4 CX 12 C (amino acids 187-223). No significant homologies to any other chloride channel proteins were detected using the BLAST program with either the entire mCLCA1 ORF or parts thereof.
Biochemical Analysis of mCLCA1 Protein-mCLCA1 cDNA was translated in vitro, and proteins analyzed by SDS-PAGE ( Fig. 2A). The unglycosylated translation product resolved as a 100-kDa protein, which corresponded to the size predicted by the mCLCA1 ORF. Using canine microsomes, glycosylation of this protein yielded a product of approximately 125 kDa. The SDS-PAGE pattern of bands was consistent with that observed for the in vitro transcribed/translated Lu-ECAM-1 cDNA (21).
The translation products of mCLCA1 cDNA-transfected HEK293 cells were further analyzed in Western blots using crossreacting antisera generated against various Lu-ECAM-1 peptides. Protein bands of 130, 125, 90 kDa and a triplet of bands of a size range of 32-38 kDa were specific to the mCLCA1 transfected cells (Fig. 2B). Antisera raised against the N-terminal region of Lu-ECAM-1 (CU8) reacted exclusively with the large sized bands of 90, 125, and 130 kDa (Fig. 4A), whereas antisera raised against the C-terminal region of Lu-ECAM-1 (CU21) recognized only the triplet of smaller protein bands (Fig. 2B). This recognition pattern is similar to that observed for Lu-ECAM-1 (21) and suggests that the ORF of mCLCA1 cDNA encodes a precursor protein, represented by alternate glycoforms of 125 and 130 kDa, that is posttranslationally processed into 90-kDa and 38/32-kDa components.
Tissue Expression of mCLCA1 cDNA-The tissue expression pattern was analyzed by probing a multiple tissue Northern blot (CLONTECH) with mCLCA1 cDNA (Fig. 3). Unlike bCLCA1, mCLCA1 seemed to be expressed in a wide variety of tissues. There was strong expression in spleen, kidney, lung, and liver and weak expression in the brain. The signal in liver probably represented gall bladder, which was strongly positive by in situ hybridization (24). Two transcripts of 3.1 kb and 5 kb were detected in most tissues, whereas only the 3.1-kb transcript was present in brain and spleen. The isolated full-length ORF corresponded in size to the smaller 3.1-kb transcript (Fig.  3). The relationship of the two transcripts detected is unclear. The larger transcripts may represent a precursor mRNA species or an unknown highly related homolog. In situ hybridization with digoxigenin-labeled, antisense RNA probes yielded a strong staining reaction with respiratory epithelia of the bronchi and trachea as well as with their submucosal glands (Fig. 4). Sense probe hybridization revealed no staining reaction, confirming specificity of the assay. A comprehensive in situ hybridization study of murine tissues other than those of the respiratory system has been published elsewhere (24). This study reveals high correlation between the tissue expression patterns of mCLCA1 and CFTR.
Functional Expression of mCLCA1 in HEK293 Cells-Expression of mCLCA1 in HEK293 cells was associated with the appearance of a novel Ca 2ϩ -sensitive Cl Ϫ conductance as determined from whole cell recordings in the presence and absence of the Ca 2ϩ ionophore ionomycin (2 M) (Fig. 5A). At low intracellular free Ca 2ϩ concentrations, the basal current at ϩ100 mV in mCLCA1-transfected cells was 2.05 Ϯ 1.09 pA/pF (mean Ϯ S.D., n ϭ 5). Upon perfusion of the bath with a solution containing 2 M ionomycin, the current increased to 10.23 Ϯ 3.46 pA/pF (n ϭ 5) (Fig. 6). This experimental manipulation had no significant effect on membrane currents recorded from either nontransfected cells, or cells transfected with the pEGFP1 reporter vector alone (Fig. 6). When subjected to whole cell recordings in the presence of 2 mM Ca 2ϩ in the pipette, basal currents in mCLCA1-transfected HEK293 cells averaged 12.01 Ϯ 6.31 pA/pF (n ϭ 5). Perfusion of 300 M DIDS through the bath reduced the current to 1.84 Ϯ 0.96 pA/pF (n ϭ 5). Similarly, NFA (100 M) and DTT (2 mM) reduced the whole cell current to 2.58 Ϯ 1.19 pA/pF and to 2.59 Ϯ 1.02 pA/pF (n ϭ 3), respectively (Figs. 5B and 6). DISCUSSION Electrophysiological studies and, more recently, molecular cloning have identified a bewildering variety of chloride channels (reviewed in Refs. 16 and 28). These channels differ in their structure (e.g. transmembrane topology), biophysical properties (e.g. ion selectivity, voltage dependence), mode of regulation (e.g. by ligands, calcium, G-proteins, CFTR, etc.), tissue expression patterns, and associated diseases. The significance of Ca 2ϩ -sensitive chloride channels in CF and mouse models of the disease is well established, but their molecular nature has been unknown until quite recently (17,21,23,24). Here we disclose molecular, biochemical, and functional characteristics of such a protein derived from mouse lung (mCLCA1). This protein is homologous to the recently cloned bovine tracheal epithelial bCLCA1 (17), the bovine endothelial Lu-ECAM-1 (21), and the human intestinal hCLCA1 (23). Sequence alignment and comparison between these four proteins show a high level of homology throughout the open reading frame, but especially in the pattern of cysteines (CX 12-13CX 4 CX 4 CX 12 C) in the N-terminal domain of the protein. A similar pattern of conserved cysteine residues is observed in members of the ligand-gated chloride channel family, albeit the functional significance of this motif has not yet been established (29).
A comparison of the in vitro and in vivo translation products of mCLCA1 reveals a complex pattern of protein processing. While the in vitro translation of the full-length mCLCA1 cDNA yields a 100-kDa protein, which can be glycosylated to a 125-kDa size product, in vivo overexpression of the mCLCA1 cDNA in HEK293 cells shows SDS-PAGE bands of 125 and 130 kDa. These two bands most likely represent two alternatively glycosylated forms of the full-length protein (21). The transfected cells also produce a 90-kDa protein band as well as a triplet of bands in the 32-38-kDa size range. Immunoblots show that the N terminus of the ORF is contained in the 90-, 125-, and 130-kDa protein products, whereas the C-terminal region of the ORF is detectable only in the smaller proteins of the 32-38-kDa size range. The absence of expected immunostaining of the 125-and 130-kDa components by the C-terminal-specific antibody presumably reflects the low relative abundance of these components and the poor conservation of the C-terminal region between bovine and mouse. The appearance of three smaller protein components at 32-38 kDa is similar to those observed for Lu-ECAM-1 and reflects differences in glycosylation of the same peptide backbone (21). Taken together, these data show an identical posttranslational processing pattern of mCLCA1 to that reported for Lu-ECAM-1 and hCLCA1, generating peptide products of similar sizes and glycosylation (21,23).
Our present results show that expression of mCLCA1 in HEK293 cells is associated with the appearance of a Ca 2ϩsensitive chloride conductance. Under whole cell conditions, the current was outwardly rectified and inhibited by the anion channel blockers DIDS and NFA as well as the reducing agent DTT. The NFA sensitivity was identical to that of the human intestinal homolog hCLCA1 (23) but in contrast to the pharmacological profile previously reported for bCLCA1 studied in a similar eukaryotic expression system (COS-7 cells) (17). These differences between members of the CLCA family of proteins could therefore provide important insights into the regions of the protein required for drug binding and/or channel gating.
The tissue expression patterns of bCLCA1, Lu-ECAM-1, hCLCA1, and mCLCA1 are quite different. bCLCA1 is expressed exclusively in the respiratory epithelia of trachea and bronchi (17,21), Lu-ECAM-1 predominantly in bovine aortic endothelial cells and endothelia of pulmonary venules (22,30), hCLCA1 in intestinal epithelia (23), and mCLCA1 in many tissues (24). Strong expression of mCLCA1 is recorded in tissues with secretory or ion regulatory functions including epithelia of the mammary gland, the respiratory system, gall bladder, pancreas, kidney, uterus, and epididymis (24). Expression of mCLCA1 is also observed in germinal centers of lymphatic tissues, spermatids, and keratinocytes of the skin, esophagus, and cornea. A precedent for differences in the tissue expression patterns of members of the same chloride channel family exists for the ClC gene family. For example, ClC-1 has been reported to be expressed predominantly in skeletal muscles where its disruption is linked to muscle myotonies (31). In contrast, ClC-Ka and ClC-Kb are expressed only in the kidney, ClC-3, ClC-4, and ClC-5 mostly in kidney and brain, and ClC-2, ClC-6, and ClC-7 are ubiquitously expressed (reviewed in Ref. 16). This heterogeneity in tissue expression suggests differences in function, regulation, and associated disease among the various members of a structurally distinct family of chloride channels.
The reported tissue expression pattern of mCLCA1 overlaps with that of CFTR, suggesting that both might contribute to the pathogenesis of CF (24). In accordance, Boucher and associates (11,13) recently reported that a calcium-mediated chloride secretory pathway was up-regulated in upper respiratory epithelia isolated from CFTR (Ϫ/Ϫ) mice and that this alternate chloride secretion pathway effectively compensated for the lack of CFTR. Correspondingly, transduction of wild-type CFTR into nasal epithelial cells from CF patients suppressed Ca 2ϩ -mediated chloride secretion in these cells, suggesting that the Ca 2ϩ -mediated pathway of chloride secretion is switched off once functional CFTR becomes available (32). These experiments clearly indicate that the Ca 2ϩ -dependent chloride secretory pathway is of importance in human cystic fibrosis and may prove amenable to pharmacological manipulation and amelioration of the degree of CF disease in specific tissues (17), e.g. the respiratory system. Cloning and molecular characterization of a murine Ca 2ϩ -sensitive chloride conductance protein provides a tool to further clarify the complementary roles of Ca 2ϩ -sensitive chloride channels with CFTR in CF.