Molecular Cloning and Characterization of p64, a Chloride Channel Protein from Kidney Microsomes*

Chloride channels were previously purified from bovine kidney cortex membranes using a drug affinity column. Reconstitution of the purified proteins into artificial liposomes and planar bilayers yielded chlo- ride channels. A 64-kDa protein, p64, identified as a component of this chloride channel was used to gener- ate antibodies which depleted solubilized kidney membranes of all chloride channel activity. This antibody has now been used to identify a clone, H2B, from a kidney cDNA library. Antibodies, affinity-purified against the fusion protein of H2B also depleted solu- bilized kidney cortex from all chloride channel activity. The predicted amino acid sequence of p64 shows that it contains two and possibly four putative trans- membrane domains and potential phosphorylation sites by protein kinase A, protein kinase C, and casein ki- nase 11. There was no significant homology to other protein (or DNA) sequences in the data base. The pro- tein is expressed in all cells tested. Expression of its mRNA in Xenopus laevis oocytes led to the insertion of a protein with the appropriate molecular mass in microsomes but not in the plasma membrane. It is likely that p64 represents the chloride channel of intracellular organelles.

Chloride channels are present in the plasma membranes of neurons, fibroblasts, lymphocytes, muscle, and epithelia (1)(2)(3)(4). In these cells they are important for mediating a variety of functions, including control of the membrane potential and the regulation of transepithelial ion absorption and secretion. Chloride channels are also present in some intracellular organelles, such as Golgi and endocytic vesicles where they are frequently present in parallel to a proton-translocating ATPase and serve to regulate the pH of these organelles (5)(6)(7)(8)(9)(10). Based on single channel behavior, chloride channels exhibit * This work was supported in part by United States Public Health Service Grants DK39532 and DK41146 and the Cystic Fibrosis Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) L16547.
$ Supported by Physician-Scientist Award DK01336 during the performance of this work.  marked diversity in conductance, current-voltage relation, and regulation by modulators. There is probably no cell without one or another type of chloride channel.
The sequences of four chloride channels have recently become available. The neuronal ligand-gated chloride channel (11) is structurally homologous to the nicotinic acetylcholine receptor. A voltage-gated chloride channel from Torpedo electric organ was recently identified by expression cloning (12), and a homologue (ClC-1) was found in mammalian skeletal muscle (13). More recently another homologue (ClC-2) was found to be ubiquitously expressed (14). This family of proteins have at least 1 2 putative transmembrane domains and cause the expression of voltage-gated chloride channels in Xenopus oocytes. A kidney chloride channel was identified from Madin-Darby canine kidney cells which leads to the expression of a nucleotide-regulated channel whose sequence does not display the typical hydrophobic a helices associated with membrane proteins (15). Finally, introduction of the cystic fibrosis gene product, the cystic fibrosis transmembrane conductance regulator (CFTR) (16),l into heterologous cells leads to expression of cyclic AMP-activated chloride channels (17,18). Furthermore, when the positive charges in the first or sixth transmembrane domain were eliminated, the anion selectivity was altered (19). Reconstitution of overexpressed and purified CFTR into planar bilayers resulted in a chloride channel with similar characteristics to the CAMP-regulated channels found in native epithelia, suggesting that CFTR itself is a chloride channel (20).
In contrast to the genetic studies mentioned above, a biochemical approach to the characterization of chloride channels has implicated two additional proteins. Ran and Benos (21) purified and reconstituted a 38-kDa protein from bovine trachea. We tested the indanyloxyacetic acids (IAA) as inhibitory ligands for epithelial chloride channels (22) and identified one of them, IAA-94, as a ligand that had an inhibitory and binding potency in the micromolar range. We purified IAA-binding proteins from bovine kidney cortex using an IAA affinity column and demonstrated that the purified material contained chloride channel activity by two independent criteria (23). First, we showed that purified IAA-binding proteins when incorporated into liposomes supported voltage-sensitive chloride uptake. Furthermore, when these liposomes were fused with planar lipid bilayers, single chloride channels were observed. The material purified by IAA columns consisted of four major proteins, and one, a 64-kDa protein (p64), elicited a monospecific antiserum which immunodepleted all recon-stitutable chloride channel activity from solubilized bovine renal cortex membranes, suggesting that p64 is a component of a kidney chloride channel (24). Using immunoblots, the anti-p64 sera recognized proteins with similar apparent molecular masses in a variety of epithelial and non-epithelial cells from different species. In addition, this antibody stained the apical membrane and intracellular organelles of epithelial cells, suggesting that p64 may be related to some plasma membrane and vacuolar chloride channels. The two most abundant proteins purified by the IAA affinity column (molecular masses, 100 and 27 kDa) were identified by NH2terminal sequence to be cellular proteins that are known to bind to ethacrynic acid, the parent structure of IAA (24). Hence, we concluded that they were drug-binding proteins which were enriched by IAA affinity chromatography but were unrelated to the chloride channel.
Recent studies by other investigators support the hypothesis that a protein of approximately 64 kDa may be a component of a vacuolar chloride channel. Brain clathrin-coated vesicles and kidney endosomes respond to cyclic AMP-dependent protein kinase by opening a chloride channel and phosphorylating a protein of similar molecular mass (10,25,26). These results suggest that p64, which we have purified, might be a chloride channel of intracellular organelles. We report here the primary sequence of p64.

EXPERIMENTAL PROCEDURES
Identification of H2B and Sequence Analysis-All RNA samples were prepared from quick-frozen slaughterhouse specimens or cultured cells by the method of Chomczynski and Sacchi (27). Oligo(dT)cellulose enrichment of poly(A+) RNA was carried out by standard methods (28). Oligo(dT) random-primed cDNA was synthesized from bovine kidney cortex poly(A+) RNA using the kit from Bethesda Research Laboratories and the ends rendered blunt with T4 DNA polymerase. The oligo(dT)-primed cDNA was ligated to EcoRI adapters (Promega, Madison WI), whereas the randomly primed cDNA was methylated with EcoRI methylase, ligated to EcoRI linkers, and digested to completion with EcoRI. The cDNA was inserted into the EcoRI site of XZap (Stratagene, La Jolla CA), packaged, and amplified to yield libraries of about lo6 independent recombinants each.
The oligo(dT)-primed library was screened with the anti-p64 antiserum as described (29). One clone, designated H2B, was identified and was used as a hybridization probe to screen both random primed and oligo(dT)-primed libraries. Inserts from individual X clones were transferred to plasmid pBluescriptSK-(Stratagene) using the autoexcision feature of XZap. Sequences were determined from the ends of overlapping restriction fragment subclones or from internal synthetic oligonucleotides using the Sequenase system (United States Biochemical Corp.) or the AB1 automatic DNA sequencer. Maps of the cDNA clones used to determine the sequence are shown in Fig  3A. Bacterial extracts containing p64 fusion protein were prepared by solubilization of bacterial pellets from stationary phase cultures in Laemmli loading buffer.
Anchored polymerase chain reaction was carried out essentially as described by Loh et al. (30). First strand cDNA was derived from bovine kidney cortex poly(A) RNA-primed with antisense oligonucleotide which maps to position 212-181 in the full-length sequence. Excess primer was removed by spinning through a Sephacryl S-400 column, and the reverse transcription product was tailed with terminal transferase and dGTP under conditions designed to yield products with dG tails of 15-30 bases. Polymerase chain reaction was carried p~ d(GGGAATTCAAGCTTGGATCCCGGG), 0.1 p~ d(GGGAAT out with Taq polymerase (United States Biochemical Corp.) and 1 TCAAGCTTGGATCCCGGGC,,), and 1 p~ specific antisense primer mapping to positions 177-158 in the p64 sequence. The annealing step was 55 "C. The products were rendered blunt with T4 DNA polymerase, cut with BamHI, ligated with pBluescript (Stratagene) that had been cut with EcoRV and BamHI and used to transform Escherichia coli strain XL1-Blue (Stratagene). Desired clones were identified by colony hybridization with a third specific p64 antisense oligonucleotide mapping to positions 139-120.
Expression Constructs-Plasmid pNH containing the translational start site and entire coding region was derived from the 5' end of cDNA pNT and the 3' end of cDNA H2B spliced at the ClaI site a t position 780. This recombinant cDNA was constructed in pBluescript KS-such that T7 transcript would contain the sense strand of cDNA. The first stop codon in the p64 reading frame occurs immediately upstream of the first PstI site in pNH. Stop codons in all three reading frames occur shortly downstream of the second PstI site of pNH. Plasmid pNHAP was constructed by deleting the internal PstI site in pNH, resulting in the new occurrence of stop codons in all three reading frames immediately downstream of the endogenous stop codon. Plasmid pNHAPH was derived from pNHAP by deletion of the internal HincII to PstI fragment, resulting in the generation of a stop codon in all frames of the p64 coding region. A map for these constructs is shown in Fig. 3A.
Transcription/Translation-CsCl purified plasmid DNA was digested with XbaI and used as a template for transcription with viral T 7 RNA polymerase (31). Capped transcripts were obtained by carrying out the reaction in the presence of 0.5 mM m7CpppG and 0.05 mM G T P (32). Yield was determined by incorporation of [w3'P]UTP and the presence of intact RNA confirmed by gel electrophoresis of the products. 0.4 pg of in vitro transcription product was translated in a nuclease-treated reticulocyte lysate (Promega) in the presence of 40 pCi of [35S]methionine, and the products were separated on 10% SDS-PAGE and detected by autoradiography.
Northern Blot Analysis-Two micrograms of poly(A) RNA from each bovine tissue or cultured cell and 5 fig of RNA from shark rectal gland were denatured in 50% dimethyl sulfoxide, 6% deionized glyoxal, 10 mM sodium phosphate, pH 6.5, a t 50 "C for 1 h, then separated on a 0.7% agarose gel run in 10 mM sodium phosphate, pH 6.5, with constant recirculation of buffer. Hind111 restriction fragments of X DNA were used as approximate molecular size markers. The RNA was transferred to Biotrans Membrane (ICN, Costa Mesa, CA) and probed with the cDNA insert from plasmid H2B, labeled with [a-32P]dCTP using the random priming kit from Boehringer Mannheim. The bovine samples were probed a t high stringency with the final wash in 0.1 X SSC, 0.1% SDS at 60 "C for 30 min. The with a final wash in 0.1 X SSC, 0.1% SDS at 55 "C for 15 min. human and shark samples were probed at somewhat lower stringency Preparation of Anti-H2B Antibodies-Guinea pig polyclonal antisera were prepared against gel slices of 64-kDa proteins purified on an IAA-23 column. These antisera, termed anti-p64, were found to immunodeplete chloride channel activity (24).
The anti-p64 serum was affinity-purified to generate an anti-H2B antibody and used for Western blot and immunodepletion experiments. E. coli lysates expressing the H2B insert were subjected to preparative SDS-PAGE and transferred to nitrocellulose. After blocking with 5% milk, the blots were incubated for 18 h with anti-p64 antisera preabsorbed against E. coli lysates lacking the insert. The bound antibody was located using an alkaline phosphatase-linked goat anti-guinea pig antisera to probe a vertical strip of the nitrocellulose. The bound antibody was eluted from the indicated region of the remaining nitrocellulose using 0.2 M glycine, pH 2.3. Control sera (non-H2B) were obtained by identical elution of anti-p64 antisera from preparative SDS-PAGE gels of proteins from E. coli transfected with vectors lacking the H2B insert at the same molecular mass region as H2B. For Western blot, kidney cortex vesicles were solubilized with n-octyl glucoside and separated on a Sephadex HR200 gel filtration column (2.5 X 90 cm). The fraction containing reconstitutable 36CI uptake activity was subjected to SDS-PAGE and immunoblot assay with anti-H2B H2B was cloned into an expression vector pEX2, and a lacZ-p64 fusion protein was generated in the bacterial host N4830-a (33). The fusion protein was constructed by inserting the large PstI fragment of H2B into the unique PstI site of the parent plasmid. Antibodies were raised in rabbits by immunizing with the partially purified fusion protein.
Reconstitution of Chloride Channels and Immunodepletion of Actiuity-Solubilized kidney membranes (3 mg/ml) were incubated for 18 h a t 4 "C with anti-H2B (ie. anti-p64 antibodies affinity-purified against the lacZ-H2B fusion protein), control antisera (anti-p64 affinity-purified against the same region of E. coli proteins), or no IgG. Antibody-protein complexes were removed with 50 pl/ml of protein A-Sepharose preincubated with 10% guinea pig albumin. The immunodepleted protein was then reconstituted into 10 mg/ml asolectin by detergent dialysis in the presence of 700 mM sucrose, 10 mM KCl, 10 mM HEPES, pH 7.0, for 24 h and 200 mM KC1, 10 mM HEPES, p H 7.0, for a subsequent 12 h. Proteoliposomes were frozen a t -70 "C and when required rapidly thawed and sonicated for 25 s in a bath sonicator.
%CI uptake assays were performed as described previously using the method of Garty (22,34). In brief, external chloride was removed by applying the vesicles to a gluconate-exchange column and eluting with 250 mM sucrose (22). 5 X lo6 cpm/ml of "CI was added. Immediately half the sample was removed and valinomycin added to a final concentration of 5 p~. At the indicated time points 0.5-ml samples were taken and extravesicular "CI removed by passing the vesicles through an 8-cm anion exchange column. uptake was then measured by liquid scintillation counting.
Electrophysiology of Xenopus laevis Oocytes-Oocytes were always completely denuded of follicular epithelium by treating them for 1-2.5 h with collagenase followed by incubation in hypertonic K aspartate for 45 min. p64 mRNA was injected and the oocytes were assayed 3, 6, or 7 days after injection. Changes in membrane conductance were assayed by measuring the current changes in response to voltage pulses using a standard 2-microelectrode voltage clamp set at a holding potential of -100 mV. The maximum change in current was recorded in response to 50 p~ cytosolic CAMP for determination of GC~,~AMP) or 10 p~ bath A23187 for CclCc.,. Incubations were at 18 "C, and recordings were made a t room temperature (21-25 "C). The  HEPES (titrated to pH 7.4 with NaOH). Also included were penicillin 100 pg/ml), streptomycin (100 pg/ml), and gentamicin (50 pg/ml).
Preparation of Membrane Vesicles from Xenopus Oocytes-Intracellular vesicles were isolated by minor modifications of the method of Evans and Kay (35). 40-50 oocytes were washed with ice-cold homogenization buffer (50 mM NaCI, 10 mM MgC12,20 mM Tris-HCI, pH 7.6, 1 mM phenylmethanesulfonyl fluoride, and 1 mM leupeptin) and homogenized in a loose-fitting glass homogenizer with 10 strokes using 10-20 pl of buffer/oocyte. The homogenate was layered on a 400-pl cushion of 20% sucrose in homogenizing buffer and spun for 30 min in a table top Microfuge at the maximum speed. The supernatant, composed largely of the cytosol, was discarded and the pellet was rehomogenized in 200 pl of phosphate-buffered saline (PBS) containing 1% Nonidet P-40 and 1 mM phenylmethanesulfonyl fluoride using 10 strokes and spun for 10 min in the Microfuge at top speed. The supernatant, now comprising the vesicle fraction was then analyzed by SDS-PAGE and immunoblotting methods.
Biotinylation (Ref. 36)"Cells were washed twice with PBS containing calcium and magnesium and incubated in PBS containing 0.5 mg/ml NHS-SS-biotin (Pierce Chemical Co.) at 4 "C for 30 min while slowly shaking. The reaction was stopped by washing twice with 0.5 M Tris, pH 8.0. The cells were then solubilized in RIPA buffer (1 ml/ T-75 plate or 20 oocytes) and the nonsolubilized material separated by centrifugation in the Microfuge at top speed for 30 min. The supernatant was added to 50 pl of streptavidin-conjugated Sepharose beads (Sigma). After a 30-min incubation a t 4 "C with gentle shaking, the beads were washed four times with RIPA buffer containing 0.5 M NaCl followed by two washes with RIPA buffer. The beads were then mixed with 50 pl of SDS-PAGE sample buffer and boiled for 3 min to release the disulfide-linked proteins from the beads which were then removed by centrifugation. The eluted proteins were then subjected to SDS-PAGE and electroblotting followed by immunoblotting using anti-p64 antibodies.

RESULTS
Cloning and Sequencing of p64"We had previously generated an antibody against a 64-Da protein that was purified using an IAA-23 column. This antibody, termed anti-p64 immunodepleted chloride channel activity from solubilized bovine renal cortex membranes (24). We constructed a cDNA expression library from bovine renal cortex and screened it with anti-p64 antisera to obtain a single clone, H2B. H2B contained a 2.4-kb insert and expressed a fusion protein with @-galactosidase with an apparent molecular mass of -70 kDa (Fig. 1A). LacZ sequences contributed about 7 kDa to fusion proteins expressed in pBluescript; approximately 63 kDa is predicted to be contributed by the cDNA insert. To assess the identity of this clone, anti-p64 antiserum was affinity-purified against the overexpressed H2B fusion protein, and the resulting fraction is termed anti-H2B. As a control antiserum, we "affinity-purified" the same anti-p64 serum against the 70-kDa region of E. coli proteins. These two affinity-purified Anti-p64 antiserum was affinity-purified against the H2B fusion protein (circles) or against the same region of an E. coli lysate which was transformed with a vector without an insert (triangles). Bovine kidney cortex microsomes were solubilized and incubated with the two batches of sera; the antigen-antibody complexes were then precipitated with protein A beads and the supernatant proteins reconstituted as described under "Experimental Procedures." 36Cl uptake was then measured in the presence (open symbols) or absence (closed symbols) of valinomycin to collapse the membrane potential. Voltagesensitive transport is the difference between the two curves. The results shown are the average of three independent experiments. antibodies were used in two assays. First, on a blot of proteins from bovine renal cortex membranes, anti-H2B antiserum stained a 64-kDa protein just as did the parent antibody prepared against he purified kidney protein (Fig. 1B). Antisera, immunopurified against proteins from E. coli which has been transfected with a vector lacking the H2B insert, failed to stain the bovine renal proteins. Second, anti-H2B antiserum immunodepleted reconstitutable chloride channel activity from solubilized bovine renal cortex membranes (Fig.  2).
Bovine kidney cortex membrane proteins were solubilized with octyl glucoside and incubated with the affinity-purified anti-H2B antibodies. The antigen-antibody complexes were removed by protein A beads, and the remaining proteins were reconstituted into phospholipid vesicles. The vesicles were formed with a chloride gradient such that chlorideinsid, >> chloride,.,i,.
H2B was used to screen the original oligo(dT)-primed library, and a subclone of the 5' end was used to screen a randomly primed library. Overlapping clones were obtained and sequenced (Fig. 3). The 5' end of the message was obtained with anchored polymerase chain reaction using nested oligonucleotide primers. A partial restriction map of the cDNA clones is shown in Fig. 3A. The sequence of the entire coding region was obtained from at least two independently derived cDNA clones.
The full-length cDNA is 6160 nucleotides (Fig. 2 B ) . The first methionine codon is a t nucleotide 157 and is preceded by a stop codon at position 60 in the same reading frame. The sequence around the first ATG conforms with the Kozak sequences for translation initiation sites (38). The first stop codon occurs at nucleotide 1468, resulting in an open reading frame coding for a protein of 437 amino acids. There is an extremely long 3"untranslated region which ends in a poly(A) tail preceded by a typical polyadenylation sequence a t nucleotide 6116. There is no extended open reading frame in the 3'untranslated region.
The predicted translation product has a calculated molecular mass of 49,008 daltons. T o confirm the identity of the deduced translational initiation and termination sites, a construct named pNH containing the entire coding region downstream from a T7 viral promoter was assembled from individual cDNAs and modified in one of two manners (Fig. 3A). Translational stop codons were introduced a t nucleotide 1516 (numbered according to the full-length sequence), 48 nucleotides downstream of the endogenous termination codon, in plasmid pNHAP and at position 1294 within the open reading frame in plasmid pNHAPH. The parent plasmid and each construct with introduced termination codons were used as templates in an in vitro transcription/translation assay using T 7 RNA polymerase and a reticulocyte lysate (Fig. 4). Both pNH and pNHAP with predicted translation products of 49,008 Da yielded products with an identical apparent molecular mass of 64 kDa on SDS-PAGE. pNHAPH with a predicted translation product of 42,306 daltons yielded a product with an apparent molecular mass of 58 kDa. Thus we conclude that the predicted initiation and termination codons are correctly identified and that p64 has aberrant mobility on SDS-PAGE.
Neither the nucleotide nor the predicted amino acid sequence of p64 bear significant homology to any known gene or gene product. The deduced amino acid sequence is markedly rich in acidic residues and has a predicted PI of 4.14. A hydrophobicity analysis of the predicted amino acid sequence by the method of Kyte and Doolittle (39) is shown in Fig. 3C. There are at least two potential transmembrane domains at amino acids 201-236 and 367-385. No putative signal sequence was discerned, suggesting that the NH, terminus of the protein would be predicted to be in the cytoplasm. This large amino terminus section of the protein contains consen- sus sequences for phosphorylation by protein kinase C, tyrosine kinase, and casein kinase 11. An octapeptide sequence with a high density of negatively charged amino acids (in single letter code, Q(E, A or G)SD(P or S)EEP, is repeated four times, and partially for a fifth, in this first intracellular domain. There is no significant homology to this specific sequence motif in the data base except for heavily negatively charged proteins. A more focused search regarding calciumbinding proteins did not result in any similarity to this motif, and in particular the motif is not homologous to an E-F hand sequence. One possibility is that it represents a site of proteinprotein interaction. The sequence between the two putative transmembrane domains is predicted to be extracellular, and it contains one potential N-glycosylation site. The COOH terminus of the protein is predicted to contain the second intracellular domain and contains a single consensus sequence for protein kinase A phosphorylation at Ser4:".

45
Alternate topological models of p64 are possible using the Kyte-Doolittle analysis. The first hydrophobic domain is long enough to cross the membrane twice. If this model is correct, then the sole N-glycosylation site (at amino acid AsnZ3') would be cytoplasmic and the protein kinase A phosphorylation site (Ser4") would be extracellular. A relatively short hydrophobic domain at amino acids 299-311 could also traverse the membrane perhaps as a p-pleated sheet similar to the H5 segment of the voltage-gated sodium and potassium channels (40,411. These considerations emphasize that this sort of analysis merely assists the construction of models which must be proven biochemically. Expression of p64 in Different Cell Types-Using antibodies raised against an H2B fusion protein, we identified proteins of apparent molecular mass of 64 5 kDa in all cell types tested, namely, human colon carcinoma cell line T84, a human pancreatic adenocarcinoma cell line obtained from a patient with cystic fibrosis, CFPAC, and a cell line that had been rescued by transfection with the wild type CFTR gene, a mouse fibroblast cell line, 3T3, and the Sf9 insect ovary cell line (Fig. 5). In the epithelial cell lines there were additional bands a t a molecular mass of 90 kDa and occasionally 120 kDa (in the case of T84) that reacted with this antibody. The kidney expresses several mRNA species that hybridize with p64 sequences, hence it is possible that cells contain more than one protein with sequence homology to p64.
RNA from a variety of tissues and species were probed by gel hybridization with the p64 cDNA as shown in Fig. 6. RNA from bovine kidney cortex, skeletal muscle, and heart have a relatively abundant transcript at approximately 6.5 kb. A transcript of the same size was also seen in lower amounts in kidney medulla and adrenal. All tissues examined had two distinct smaller transcripts at approximately 5.5 and 4.5 kb. In addition, kidney medulla, adrenal, and brain have a 7-kb transcript, and the heart has a smaller transcript of about 2 kb. Essentially the same pattern of transcripts was observed when bovine kidney cortex or heart RNA was probed with cDNA probes limited to the coding region (data not shown). This diversity of transcript size in various tissues suggests that this gene is subject to alternate splicing or that a family of closely related genes exist. These possibilities are being explored.
RNA from chloride-transporting epithelia were also probed with p64 cDNA as shown in Fig. 6. T84 cells and Panc 1 are continuous human cell lines which have well characterized chloride channels. Both of these cell lines have a single major transcript of approximately 5.5 kb which hybridizes to the p64 probe at high stringency. RNA from the shark rectal gland, a chloride-secreting epithelium rich in chloride channels, contains transcripts of approximately 4 and 6 kb that hybridize with the p64 probe at moderate stringency. Thus, at the protein and RNA level, p64 was present in every cell type that was tested.
Expression of p64 and Chloride Channels-We injected RNA transcribed from the T7 promoter construct of the fulllength clone into Xenopus oocytes. No new plasma membrane chloride current appeared on electrophysiological studies. However, a new protein with the appropriate molecular mass appeared. Cell fractionation experiments showed that the protein was incorporated into microsomes (Fig. 7). However, the protein was unable to be targetted to the plasma membrane. We biotinylated the oocytes and precipitated the solubilized proteins by avidin beads. Western blot analysis showed that p64 was not present in the precipitate (Fig. 7). A similar experiment performed on CFPAC and T84 cells, where we had shown previously that this protein reaches the plasma membrane, demonstrated that this method is capable of documenting the presence of p64 (data not shown). The reason why expressed p64 cannot reach the surface is not apparent at present, but it could be due to the need for another protein or that it could contain a yet unidentified retention signal for intracellular organelles.

DISCUSSION
We have presented the cloning and characterization of a gene encoding a component of bovine kidney cortex chloride channels. The protein predicted to be encoded by this gene has a sequence consistent with an integral membrane protein which crosses the membrane at least twice and has target sequences within the proposed cytoplasmic domains for several signal transduction pathways known to affect chloride channel activity in epithelial cells. This gene is expressed to some extent in all cells tested so far. In addition it is expressed

IKidndyl
Xenopus laevis Oocytes Bovine kidney vesicles (10 pg), intracellular membranes from 20 Xenopus oocytes (100 pg), 20 injected or uninjected oocytes were biotinylated and processed as described under "Experimental Procedures." The bands seen at 50 kDa were also seen in the same blots when no primary antibody was used.
at higher levels in specialized cells and tissues known to express high levels of plasma membrane chloride channel activity, i.e. kidney, heart, skeletal muscle, and T84 and Panc 1 cells. Evidence That p64 Is a Chloride Channel Protein-There are two independent lines of evidence that p64 is a chloride channel protein. First, the purification results indicated that the channel activity in IAA-purified material must be due to one or more of the remaining two proteins in the purified material. Second, anti-p64 antibodies immunodeplete solubilized kidney vesicles of chloride channel activity.
Purification and Reconstitution-Two definitive and complementary assays showed that the IAA-purified material contained chloride channels (23). Incorporation into planar bilayers showed single channels, but since this is a single molecule assay, it cannot give information regarding yield or specific activity. In contrast, reconstitution into liposomes and demonstration of voltage-sensitive chloride flux is a semiquantitative assay that allows an order of magnitude calculation of yield. Using this assay we determined previously that the IAA purification scheme resulted in at least a 1000fold purification of the channel (23). Thus as initially reported, the purification method was consistent with one or more of the major proteins purified being the channel itself. Elimination of two other proteins has left p64 as the leading candidate for the channel protein.
Immurwdepletwn-Antibodies immunopurified against p64 fusion proteins immunodepleted all chloride conductance activity of solubilized kidney cortex membranes ( Fig. 2 and Ref. 24). This result indicates that p64 is part of the channel complex itself but does not rule out the possibility that the channel may consist of p64 associated with other proteins. Since p64 has only two transmembrane domains per mon-omer, it seems likely that the conducting complex itself consists of multimers of p64 with or without other components. Indeed, the apparent molecular mass of the reconstitutable channel solubilized in octyl glucoside (or CHAPS) is -400-600 kDa. The stoichiometry of the components awaits further study. The anti-p64 antibody precipitates only the 64-kDa protein, suggesting that the channel complex is composed of homomultimers (24). However, the conditions needed to elute p64 from the antibody (treatment with 0.2% SDS or low pH) by themselves had deleterious effects on reconstitutable channel activity despite attempts at renaturation using the method of Braiman et al. (42). This has prevented us from examining directly whether p64 alone can reconstitute chloride channels. Hence, at the minimum, p64 is a necessary component of the kidney cortex chloride channel complex. The availability of the cDNA should allow us now to overexpress the protein to obtain sufficient quantities for further biochemical and functional studies similar to what was done with CFTR (20).
Isp64 a Vacuolar Chloride Channel?-p64 was purified from microsomal membranes enriched for Golgi markers. It is present in all cells, albeit at low abundance. In some cells its abundance is high enough to allow immunocytochemical localization. This was achieved in CFPAC cells (a pancreatic duct adenocarcinoma cell obtained from a patient with cystic fibrosis (43)) which showed that perinuclear vesicles were stained (24). There is increasing evidence that the chloride channels of endosomes and clathrin-coated vesicles can be activated by protein kinase A (10,25). Furthermore, such phosphorylation caused the appearance of a phosphorylated 65-70-kDa protein (25, 26). All of this leads one to suspect that p64 could be the chloride channel of intracellular membranes.
When p64 was expressed in Xenopus oocytes, it was incorporated into membranes but was not targetted to the plasma membrane. Whether the protein has signals that retain it in intracellular organelles or lacks signals that carry it to the plasma membrane remains to be determined. We demonstrated previously that p64 was present on the apical membrane of some CFPAC cells. Whether this is due to the presence of another protein that can guide it to that domain or that the CFPAC cell expresses a homologue of p64 that has a targeting signal remains to be determined. As shown in the Northern blots there are several homologous messages of p64 in the kidney, suggesting that such diversity of structure might underlie diverse fates of p64.