INTRODUCTION

Whereas ion channels have been found on the membranes of various intracellular organelles, it has only been recently that patch clamping studies have suggested their existence at the nuclear membrane. The nuclear pore complexes have been considered the site of communication and exchange between the nucleus and cytoplasm (1, 2). Studies of traffic across the nuclear envelope have in general conformed to the paradigm that ions and small metabolites with diameters of less than 3-4 nm passively diffuse across the nuclear envelope (3). Thus, the concept of nuclear membrane ion channels seems at variance with the generally accepted views of the nuclear envelope, with their function at this location seeming redundant. However, this does not appear to be the case.

The first demonstration of ionic conductances in the nuclear membrane were in mouse zygote pronuclei (4). Further evidence then followed with the demonstration of ion-selective channels in avian erythrocytes (5), in mouse oocyte germinal vesicles, in nuclei from two-cell embryos and liver (6), in the nuclei of cardiac myocytes (7), and in rat hepatocyte nuclei (8). Mak and Foskett (9) described the presence of inositol 1,4,5-trisphosphate-dependent receptor channels in isolated nuclei which were activated by inositol 1,4,5-trisphosphate, inhibited by heparin and selective to calcium ions. Similarly, Pasyk and Foskett (10), have shown chloride channel activity in isolated nuclei from CHO¹ cells. More recently, following fractionation and reconstitution of inner and outer nuclear membrane fractions into lipid bilayers, Rousseau et al. (11) have shown the presence of two types of chloride channels. The use of calcium ion imaging techniques has also demonstrated variations in calcium ion concentrations between the nucleus and cytoplasm, again suggesting a selective uptake or retention of these ions by the nucleus (12, 13).

With growing electrophysiological data for the existence of nuclear ion channels, it is clear that the cloning and isolation of these proteins will greatly assist in determining their structure and function. This paper describes the molecular cloning and characterization of what we believe to be the first chloride ion channel protein of the nuclear membrane and only the second cloned ion channel found at this location. As part of a project investigating gene expression in "activated" monocytoid cells, a subtracted cDNA library designed to enrich for activation-associated genes was prepared using U937 cells. U937 cells were used to generate the cDNA library, since they are a human histiocytic lymphoma cell line with monoblastic characteristics that can undergo in vitro differentiation into a cell with monocyte-like characteristics (14). Exposure of U937 cells to various factors including all-trans-retinoic acid (RA) or interferon-, can induce this change.These differentiated cells can be activated with phorbol 12-myristate 13-acetate (PMA), much like normal monocytes.

Screening of the subtracted cDNA library for genes associated with activation resulted in the isolation of clone 4, which was found to code for a novel 241-amino acid protein, designated nuclear chloride channel-27 (NCC27). It is homologous to a cDNA clone, BOVCCP64A (GenBank^TM accession number L16547[GenBank]), which codes for a bovine chloride channel protein, p64, believed to localize to internal organelles (15). NCC27 is a novel chloride channel protein that localizes principally to the nucleus of all cells studied. Patch clamp studies of CHO-K1 cells transfected with NCC27 revealed an increased level of chloride ion channel activity at their nuclear membrane. NCC27 was found to be widely expressed in various cells and cell lines, and its gene expression was regulated by activation stimuli.

EXPERIMENTAL PROCEDURES

A subtraction library was constructed using U937-derived mRNA. The driver cDNA library (U937-RA) was synthesized from U937 differentiated with 1 µM RA for 3 days, and the tester cDNA library (U937-RP) was synthesized from U937 cells treated for 3 days with 1 µM RA, followed by 160 nM PMA activation for 3 h. The cDNA was cloned into the EcoRI site of the ZAP II vector (Stratagene) using the ZAP-cDNA Gigapak II Gold Cloning Kit (Stratagene). Each library was then converted to pBluescript SK plasmid (Stratagene) using the autoexcision feature of ZAP II. Single-stranded DNA from the two libraries was used to prepare the subtraction library using the Subtractor kit (Invitrogen) according to the manufacturer's instructions.

Clone 4 was fully sequenced bidirectionally by the dideoxy method (16) using the Sequenase version 2.0 kit (U.S. Biochemical Corp.). Sequence similarity searches of GenBank^TM, EMBL, and SwissProt data bases were performed with the FASTA (17) and BLITZ (18) programs. Predictions of the protein's secondary structure were performed using the algorithm of Kyte and Doolittle (19) (window size was 10 residues).

The following cell lines were obtained from the American Type Tissue Collection (Rockville, MD). Cell culture materials were obtained from Commonwealth Serum Laboratories (Australia), unless otherwise indicated. U937, RAW 264.7, K562, U266B1, and HUT 78 cells were maintained in RPMI 1640 media, supplemented with 10% FCS, 2 mM glutamine, 20 mM HEPES, and 5 µg/ml gentamycin. HL-60, KG-1, Raji, and Daudi cells were maintained in the same media except with a final concentration of 20% FCS, while THP-1 cells had (2 × 10⁵ M) 2-mercaptoethanol (Merck) added. CCD-34LU cells were grown in Dulbecco's modified Eagle's medium containing 10% FCS, 20 mM Hepes, 2 mM glutamine, and 5 µg/ml gentamycin. Chinese hamster ovary cells (CHO-K1) were grown in Dulbecco's modified Eagle's medium/F12 medium (Life Technologies, Inc., Australia) containing 5% FCS.

Human umbilical vein endothelial cells were a gift from Mr. T. Magoulis (CFI, Australia). Monocytes and lymphocytes were obtained from buffy coat leukocyte concentrates from normal human donors by centrifugation over Ficoll-Paque (Pharmacia, Australia). Monocytes were isolated by adherence and cultured in a modified serum-free medium according to the method of Bennett et al. (20, 21). Serum-free medium consisted of Iscove's modification of Eagle's medium (Life Technologies, Inc.) supplemented with 4 mM L-glutamine, 2 mg/ml FCS, 5 ng/ml sodium selenite (Sigma), soy bean lecithin (0.1 mg/ml) (ICN Biomedicals), insulin (5 mg/ml)(Boehringer Mannheim, Germany), and transferrin (5 mg/ml) (Boehringer Mannheim). Lymphocytes were isolated as the nonadherent cell population, removed from the adherent macrophages, and cultured in RPMI 1640 containing 1% (v/v) FCS, 2 mM glutamine, and 20 mM HEPES.

For recombinant interferon- stimulation, 1000 units/ml was added to the culture medium for a specified length of time. Recombinant interleukin-2 (IL-2) (50 units/ml) and IL-4 (100 units/ml) were purchased from Boehringer Mannheim. RA was used at a final concentration of 10⁶ M, PMA at 160 nM final concentration, and lipopolysaccharide at 10 µg/ml for a specified length of time. All were purchased from Sigma.

All RNA samples were prepared by the method of Chomczynski and Sacchi (22). Total RNA (20 µg/lane) was separated on a formaldehyde/agarose denaturing gel, in 1 × MOPS buffer and then transferred to Hybond N⁺ membrane (Amersham) and baked for 2 h at 80 °C. RNA ladder (Boehringer Mannheim) was used as a molecular size marker. Membranes were then probed with a cDNA insert from clone 4, labeled with [-³²P]dCTP using Megaprime Labeling kit (Amersham) or with a 28 S oligonucleotide (TCCGTCCGTCGTCCTCCTC) end-labeled with [-³²P]ATP. Hybridization was overnight at 65 °C (37 °C for the 28 S oligo probe), with the final wash in 0.1% SSPE, 0.1% SDS for 5 min (1% SSPE, 0.1% SDS for 5 min for the 28 S oligo probe). Blots were exposed using XAR-5 x-ray film (Eastman Kodak Co.) with Cronex Lightening Plus intensifying screens (Dupont) at 70 °C.

The recombinant glutathione S-transferase (GST) fusion protein (GST-NCC27) was expressed and purified using pGEX-4T-1 vector system (AMRAD-Pharmacia, Australia) and the Escherichia coli strain HB101, following the manufacturer's protocol. The open reading frame of clone 4 was PCR-amplified using oligonucleotides incorporating restriction enzyme sites to allow for directional cloning: sense primer, TCCCCGACGGATCCATGGCT; antisense primer, CCCAGCGGCCGCTTATTTGA. The DNA was restriction enzyme-digested and then cloned in frame into the HindIII/NotI site of pGEX-4T-1. The isopropyl-1-thio--D-galactopyranoside-induced recombinant protein (rNCC27) was purified from the bacterial homogenate on glutathione S-Sepharose (AMRAD-Pharmacia). The fusion protein was at this stage either eluted from the column with 10 mM reduced glutathione in 50 mM Tris-HCl (pH 8.0) or cleaved from GST while still bound to the column using biotin-labeled thrombin (Novagen), which was then removed by the streptavidin-agarose affinity matrix (Novagen).

Purified recombinant fusion protein GST-NCC27 was used to immunize rabbits for the production of polyclonal antibodies (Institute of Medical & Veterinary Science, South Australia). Preimmune and test bleed sera, were screened by immunofluorescent staining of nontransfected and clone 4-transfected CHO-K1.

Eukaryotic expression of NCC27 was performed by directionally cloning the coding region of the clone 4 construct described above into the expression vector pRc/CMV (Invitrogen) at the HindIII/NotI sites. A second construct (clone4FLAG) was also cloned into pRc/CMV vector, which incorporated the sequence for the 8-amino acid FLAG peptide (DYKDDDDK) at the carboxyl terminus of the recombinant protein NCC27. The insert-containing pRc/CMV constructs were then transfected into monolayers of CHO-K1 cells. CHO-K1 cells (80% confluent) were transfected for 24 h in 35-mm² dishes using 9 µl of Lipofectamine reagent (Life Technologies, Inc.) and 1 µg of DNA, as recommended by the manufacturer. Stable transfectants were selected with 1000 µg/ml G418 (Boehringer Mannheim). Subclones were isolated and then screened by immunofluorescent staining using anti-m2 (anti-FLAG) monoclonal antibody (IBI) and anti-NCC27 polyclonal antisera.

Cells were seeded onto 8-well glass chamber slides (NUNC) one day prior to staining, washed twice with PBS, and either fixed and permeabilized for 15 min at room temperature in PBS containing 3.5% (v/v) formaldehyde and 0.1% (v/v) Triton X-100 for internal cell staining or fixed with 4% paraformaldehye for 30 min for surface staining. The cells were stained with primary antibody for 30 min at 4 °C. Antibodies used were anti-FLAG antibody (1:1000 dilution), control class-matched monoclonal antibody, anti-CD4 (OKT4 (550 mg/ml), a gift from Dr. Margaret Cooley (CFI, Australia) (1:50)), anti-NCC27 rabbit polyclonal antisera (1:1000), and control preimmune rabbit sera (1:1000). A further control of anti-NCC27 polyclonal antibody (1:2000), which had been preincubated for 30 min at room temperature with 50 µg of purified rNCC27 protein, was also included. Following the reaction with primary antibody, the cells were washed in PBS and then incubated with secondary sheep anti-mouse or sheep anti-rabbit fluorescein isothiocyanate-conjugated polyclonal antibody (Silenus Laboratories, Australia) for 45 min at room temperature. Cells were then mounted and viewed using UV fluorescence microscopy, with excitation at 488 nm and by confocal microscopy using a Sarastro 2000 CLSM (Molecular Dynamics), with a plan apochromat × 60/1.40 NA oil immersion lens and an argon-ion class II laser.

SDS-polyacrylamide gel electrophoresis and Western blotting were performed by standard techniques (23). Cells were solubilized on ice for 15 min, in 1 × lysis buffer (50 mM Tris-HCl, pH 7.0, 0.5% (v/v) Nonidet P-40, 3.7% (w/v) EDTA) including leupeptin and phenylmethylsulfonyl fluoride (Boehringer Mannheim). Samples were clarified by centrifugation at 1000 × g for 5 min. 6 × SDS sample buffer (0.35 M Tris-HCl, pH 6.8, 10.28% (w/v) SDS, 36% (v/v) glycerol, and 0.012% (w/v) bromphenol blue) was then added to protein from 1 × 10⁵ cells and loaded onto a 15% SDS-polyacrylamide gel (Bio-Rad, Australia). This was followed by tank electrotransfer of the samples to nitrocellulose membrane (Bio-Rad). Western blots were blocked for 1 h at 37 °C (or overnight at 4 °C) with 3% (w/v) bovine serum albumin, 0.05% (v/v) Tween-20 in PBS and then probed for 30 min at room temperature using either anti-NCC27 (1:2000) or anti-FLAG antibody (IBI) (1:1500). Control antibodies were anti-GST polyclonal (1:5000) (AMRAD-Pharmacia) and preimmune rabbit sera (1:2000). After washing, secondary biotin-labeled anti-mouse Ig or anti-rabbit Ig polyclonal antibodies were applied (1:2000) (Amersham) for 30 min at room temperature, after which time blots were again washed. This was followed by a 30-min incubation at room temperature with streptavidin-biotinylated horseradish peroxidase complex (1:2000) (Amersham). After final washing, the staining was visualized using Renaissance chemiluminescence reagent (DuPont) according to the manufacturer's instructions and exposure of the blots to XAR-5 x-ray film (Kodak).

This was performed essentially as outlined by Ausubel et al. (23). Briefly, 1-10 µg of each protein sample was dissolved in 100 µl of 50 mg/ml CNBr (Pierce) in 70% formic acid. A second sample of each protein was dissolved in 100 µl of 70% formic acid without CNBr. Samples were incubated for 48 h at room temperature, after which time they were lyophilized. 1 × SDS sample buffer was then added, and Western blot analysis was undertaken as previously indicated.

A modification of the method described by Ausubel et al. (23) was used to isolate nuclei. Confluent CHO-K1 cells were washed three times with PBS. The cells were then detached with 0.02% EDTA and washed with PBS once more. All solutions used from this point contained leupeptin, pepstatin, and aprotinin (all at 2 µg/ml), and Pefabloc (0.5 mM) (all from Boehringer Mannheim). The cell pellet was resuspended in 3 ml of ice cold hypotonic buffer (10 mM MgCl₂, 1 mM KCl, 5 mM HEPES, pH7.4) and centrifuged at 800 × g for 5 min. The pellet was then resuspended in 3 times its own volume of hypotonic buffer and incubated on ice for 10 min to allow the cells to swell. The swollen cells were disrupted with 30 strokes of a Dounce homogenizer, and a crude nuclear pellet was obtained by centrifugation of the homogenate for 2 min at 800 × g. The nuclei were purified using Optiprep iso-osmotic density medium (Nycomed, Norway) according to the manufacturer's instructions. Briefly, nuclei were resuspended in 1 ml of homogenization solution (0.25 M sucrose, 25 mM KCl, 10 mM MgCl₂, 20 mM Tricine, pH 7.8), combined with 1 ml of solution C (50% (w/v) Optiprep, 25 mM KCl, 5 mM MgCl₂, 20 mM Tricine-NaOH, pH 7.8). For centrifugation, 2 ml of the nuclear suspension were underlaid with 2 ml of 30% Optiprep, followed by 1 ml of 35% Optiprep. The sample was centrifuged for 40 min at 10,000 × g at 4 °C, and the nuclei were collected at the 25-30% interface. The nuclear suspension was diluted 4-fold in hypotonic buffer and centrifuged at 800 × g for 2 min to remove the remaining Optiprep. The pellet was resuspended in 200 µl of hypotonic buffer and freeze-thawed three times, and the insoluble material was washed three times. The insoluble nuclear fraction was resuspended in 100 µl of hypotonic buffer containing 1 unit of DNase I and incubated for 30 min at 37 °C. The insoluble nuclear fraction was solubilized in nonreducing SDS electrophoresis sample buffer and separated on a 15% SDS-polyacrylamide gel, transferred to nitrocellulose, and probed with polyclonal anti-NCC27 antibody as described above.

Nuclei were isolated based on the method of Wotton et al. (24). Briefly, confluent CHO-K1 were detached with 0.02% EDTA and then washed twice in ice-cold PBS. Cells were resuspended in ice-cold lysis buffer (3 mM MgCl₂, 1 mM KCl, 10 mM Tris-HCl, pH 7.4) and lysed on ice with 2.0% Nonidet P-40. Following lysis, the nuclei were centrifuged through a 30% sucrose cushion at 800 × g for 5 min at 4 °C. The nuclear pellet was then gently resuspended in intracellular solution (140 mM KCl, 1.2 mM MgCl₂, 5 mM EGTA, and 10 mM HEPES, pH 7.2, with KOH). Integrity of the nuclei was checked by staining with toluidine blue and trypan blue and viewed by phase contrast microscopy. The nuclei were then placed into 35-mm² dishes for electrophysiological study.

Single ion channel activity in isolated nuclei from CHO-K1 were recorded using a "nucleus-attached" configuration at room temperature (23 °C) (4, 25). In other experiments, whole-cell and single-channel currents were recorded from the plasma membrane of intact CHO-K1 (26). Patch pipettes with a tip opening of between 0.9 and 1.5 µm (27) and impedence of 3-7 megaohms were fabricated from thin walled borosilicate glass (Vitrex Microhematocrit Tubes, Denmark). The reference electrode was connected to the superfusing bath with a KCl salt bridge for experiments in which the Cl was either reduced in concentration or substituted by a different anion. The channel currents were amplified and filtered at 1 kHz (3 db point) using an Axopatch 1D amplifier (Axon Instruments) and sampled on line by a microcomputer (IBM 486 compatible) using commercial software and associated A/D hardware (pClamp 6.0/Digidata 1200, Axon Instruments Inc. and Scientific Solutions Inc.). Single-channel open probability was determined from steady-state recordings of 2-min duration, with the nucleus clamped at the resting membrane potential, and quantified as follows,

(Eq. 1)

where P_o represents single-channel open probability, M is the apparent number of channels in the patch, m is each channel in the range 1 to M, and P_m is the open probability associated with the mth channel in the patch. The apparent number of channels in each patch was determined as the number of discrete levels of current observed in the patch clamp recordings. The values for P_m were obtained as the area under the curve for gaussian curves fitted to each of the channel peaks in amplitude histograms constructed for each patch (28). All data are presented as the mean ± S.E., with the number of different cells used to obtain the recordings indicated in parentheses (see "Results"). The solution used for superfusion of the intact CHO-K1 contained 130 mM NaCl, 4.8 mM KCl, 1.2 mM MgCl₂, 1.2 mM NaH₂PO₄, 10 mM HEPES, 12.5 mM glucose, 1 mM CaCl₂, bovine albumin (0.5 mg/ml, fraction V, Sigma), pH 7.4, with NaOH. The solution used for superfusion of the isolated nuclei contained 140 mM KCl, 1.2 mM MgCl₂, 5 mM EGTA, and 10 mM HEPES, pH 7.2, with KOH. The same solution was used to fill the recording pipette for both the "nucleus-attached" and whole-cell recordings. In separate experiments using the whole-cell configuration and clone 4-transfected CHO-K1 cells to detemine the permeability of the channels to different anions, the chloride in the superfusing solution was substituted for nitrate, isothiocyanate, iodide, acetate, bicarbonate, sulfate, or fluoride. The permeability ratio of the anion (X) relative to Cl (P_x/P_Cl) was calculated using the Goldman-Hodgkin-Katz equation from the resultant shift in reversal potential of the I/V curves.

RESULTS

A 1223-base pair cDNA clone (clone 4), was isolated from the U937 subtracted cDNA library, and comparison of its DNA sequence to the GenBank^TM and EMBL data bases revealed significant homology to a bovine chloride channel clone, BOVCCP64A (15), of 1561 base pairs in size (full-length cDNA, 6160 nucleotides). The longest open reading frame of clone 4 extends from nucleotide 219 to 944 (Fig. 1A). The initiation codon is flanked by sequences that concur with the consensus sequence for the initiation of translation (29).

Northern blots of total RNA from various cells and cell lines were probed using the complete clone 4 insert. This resulted in the detection of two mRNA species of approximately 1.2 and 1.0 kb. In the monocytoid cell line U937, clone 4 mRNA is up-regulated by PMA after 24-48 h, with a disproportionate increase in expression of the 1.2-kb band compared with the 1.0-kb band. This response could be further enhanced by pre-exposure of the cells to RA (Fig. 2A). This response appears to be PMA-specific, since lipopolysaccharide had no noticeable effect (data not shown). Similarly, pretreatment of human culture-derived macrophages with interferon- for 9 h followed by IL-2 for 3 h, up-regulates the 1.2-kb band with no detectable 1.0-kb band expression (Fig. 2B).

To determine whether the two mRNA transcripts represented two closely related genes or alternate products of the same gene, Northern blots of total RNA from U937 cells were probed separately with the three EcoRI restriction digest-generated fragments from clone 4 (Fig. 1A). All three fragments were found to hybridize to both mRNA transcripts. The Northern blots were then probed with two shorter fragments generated by PCR, which represent the 5- (primers T3.4 and T7.6) (Fig. 1A) and 3- (primers T3.5 and T7.7) (Fig. 1A) untranslated sequences of clone 4. It was found that only the 3 probe detected both mRNA species, while the 5 probe only detected the 1.2-kb species. It was therefore inferred that the two mRNA transcripts varied in at least the first 120 base pairs of the longer species. This would suggest that the two transcripts are due either to alternate transcription start sites or alternate transcripts.

Clone 4 mRNA was found to be expressed in all of the following cells and cell lines, including the monocytoid cell lines U937, HL-60, THP-1, and KG-1, the mouse monocyte cell line RAW264.7, peripheral blood monocytes and lymphocytes, the B cell lines Daudi, Raji, and U266B1, and the T cell line HUT 78 as well as human umbilical vein endothelial cells, the fibroblast cell line CCD-34LU, and K562 cells (results not shown).

The longest open reading frame of clone 4 codes for a polypeptide of 241 amino acids with a predicted molecular mass of 26,900 Da and pI of 4.85. Comparison of the predicted amino acid sequence of NCC27 with the available protein data bases NBRF and SwissProt again confirmed its close homology to the bovine chloride channel protein p64, with scores of 57% identity and 72% similarity. The 241 amino acids of NCC27 align with p64 at positions 194-437, which represent the carboxyl-terminal half of the protein p64 (Fig. 1B). Extensive searches and multiple alignment comparisons of these two protein sequences to other ion channel proteins revealed no similarities.

Hydrophobicity analysis (19) shows two strongly hydrophobic regions that are long enough to represent transmembrane domains, from amino acids 21-39 and 170-189 (Fig. 1A). Comparison of the hydrophobicity plots for NCC27 with p64 reveals a similar pattern of hydrophobic domains, with the two longest regions found at the same relative position in both proteins (Fig. 1C). Furthermore, analysis of NCC27 for possible motifs revealed two putative nuclear localization sequences, KRR and KKYR, at positions 49-51 and 192-195, respectively (Fig. 1A). There is one putative N-glycosylation site at asparagine 42, one cAMP phosphorylation site at threonine residue 49, five possible casein kinase II phosphorylation sites, four putative protein kinase C phosphorylation sites at serines 27 and 163 and threonines 48 and 77, and five possible N-myristoylation sites.

To localize NCC27, CHO-K1 cells transfected with clone 4, the clone4FLAG construct, or the vector alone were stained using anti-FLAG. The antibody showed bright staining of the clone4FLAG-transfected cells, while there was no visible staining in the other two transfectants (Fig. 3). The antibody staining was brightest throughout the nucleus, including the nuclear membrane, with a small amount of staining also apparent in the cytoplasm and on the plasma membrane (Fig. 3). Staining of fixed but not permeabilized cells with the anti-FLAG antibody revealed a pattern of staining of the clone4FLAG-transfected cells, suggestive of surface membrane distribution (Fig. 3). There was no staining seen in the control-transfected cells (Fig. 3).

The three CHO-K1 transfectants discussed above were also visualized using the polyclonal antibody raised to the GST-NCC27 fusion protein. This showed binding to the nucleus and nuclear membrane in all three of the transfectants (Fig. 3) as well as nontransfected CHO-K1 (data not shown). This suggested constitutive expression of the native protein in transfected and nontransfected cells. A control anti-GST antibody (data not shown) and rabbit preimmune sera showed no comparable staining of any of the cells (Fig. 3). Blocking studies undertaken by preincubation of the anti-NCC27 antibody with purified rNCC27 protein confirmed the specificity of the antibody with no binding being observed in cells stained with blocked anti-NCC27 (Fig. 3).

To provide further evidence supporting the presence of NCC27 on the nuclear envelope, a crude nuclear membrane fraction was analyzed. Purified nuclei were freeze-thawed to disrupt the membranes of the nuclear envelope and washed to remove any remaining nucleoplasmic contents. This fraction is expected to contain the inner and the outer nuclear envelopes, along with the nuclear pore complexes, and the underlying nuclear lamina. NCC27 was found to be present in this insoluble nuclear material (Fig. 4A). This is consistent with the occurrence of either a transmembrane form of the protein or its association with a multiprotein channel complex.

To further examine NCC27 protein expression in CHO-K1 cells, Western blots of whole cell lysates were probed with anti-NCC27 antibody. This showed the presence of a 27-kDa protein in all samples including U937 (Fig. 4B) and nontransfected CHO-K1 (data not shown). Identical blots stained with the anti-FLAG antibody showed binding to a similar sized protein band in only those CHO-K1 cells transfected with the clone4FLAG construct (Fig. 4B). Identical blots probed with control preimmune sera and anti-GST polyclonal antibody showed no specific binding (data not shown).

To further confirm that the purified rNCC27 protein was the same as the 27-kDa protein detected by Western blot analysis of U937 whole cell lysates, CNBr cleavage was undertaken. Since the predicted NCC27 protein sequence encodes only two methionine residues, with the second at amino acid position 32, CNBr cleavage would result in a molecular weight reduction to 23.5 kDa. Samples of both purified rNCC27 and soluble protein from U937 whole cell lysates were CNBr-treated. The resultant products analyzed by Western blotting clearly indicated that both rNCC27 and native NCC27 from U937 cells were cleaved by CNBr resulting in a reduction of the 27-kDa band to the predicted size of approximately 23.5 kDa (Fig. 4C). Additional immunoreactive bands can be seen in the CNBr-cleaved rNCC27 sample (Fig. 4C, lane 3). These probably represent the effect of contaminating uncleaved rNCC27-GST fusion protein. GST itself is known to contain numerous internal methionine residues and is susceptible to CNBr cleavage, which would result in the extra bands noted on this heavily loaded blot.

To confirm that NCC27 was constitutively expressed in CHO-K1 cells, Northern blots of total RNA from both clone 4-transfected CHO-K1 and nontransfected control CHO-K1 were probed with the entire clone 4 insert. Both samples were found to express the clone 4 mRNA 1.2-kb transcript (Fig. 2C). This was in keeping with the studies using the antibody to rNCC27, which suggested that native protein as well as recombinant protein were detected in the clone 4-transfected cells.

Ionic currents in both clone 4-transfected and vector only control-transfected CHO-K1 were recorded using patch clamp electrophysiology. Using the whole-cell configuration, the macroscopic conductance of the cell membrane of intact cells was 4.6-fold greater in clone 4-transfected cells (3.70 ± 0.08 nanosiemens, n = 14) compared with the vector only control-transfected cells (0.81 ± 0.06 nanosiemens, n = 3). Single-channel recordings of the clone 4-transfected cells revealed single channels with a conductance of 22 ± 5 picosiemens (n = 6, 95% confidence interval, 9-35 picosiemens; Fig. 6). Thus, transfection of CHO-K1 with clone 4 introduced ion channel activity that was not normally present at the plasma membrane of these cells. This was found to be due to the activity of chloride ion channels, since simultaneous dilution of the extracellular cations and anions shifted the reversal potential of the current-voltage relation toward the Nernst potential for Cl (Fig. 5). Furthermore, the chloride channels were permeable to other anions to varying degrees. The permeability sequence, with the relative permeability to Cl in parentheses, was as follows: SCN (1.45) > F (1.28) > Cl (1.0) > NO₃ (0.82)  I (0.81) = HCO₃ (0.81) > acetate (0.67).

Patch clamp recordings were also undertaken of intact isolated nuclei from both clone 4- and vector only control-transfected cells. The isolated nuclei, were seen as intact spheres when viewed under high power phase contrast microscopy. Some nuclei showed varying degrees of cytoplasmic debris associated with them. Only those nuclei that showed a sharp outline were chosen for the patch clamp experiments. Single-channel recordings of the nuclear membrane obtained from clone 4-transfected CHO-K1 revealed single channels with a conductance of 33 ± 4 picosiemens (n = 6, 95% confidence interval, 23-43 picosiemens). The conductance of these channels was not significantly different from those recorded from the plasma membrane (p = 0.12, not significant, Student's t test). Although electrophysiologically similar channels were also present in the nuclear membrane of control CHO-K1 transfected with the expression vector only, their activity was 9.5-fold less than in the clone 4-transfected cells (Fig. 6). Specifically, the single-channel open probability for clone 4-transfected cells was 0.19 ± 0.08 (n = 7) compared with 0.02 ± 0.02 (n = 5) in cells transfected with the vector only (p = 0.02, Mann-Whitney test). Furthermore, in none of the vector-only transfections was there more than one level of single-channel current transitions, whereas in the nuclei from clone 4-transfected cells, integral multiples of the transition level were commonly observed, indicating the presence of several active ion channels in the patch of membrane sealed by the recording micropipette.

DISCUSSION

Several lines of evidence indicate that NCC27 acts as an ion channel or forms part of an ion channel complex. These include its protein structure, which is consistent with that of other ion channel proteins, its strong homology to p64, a bovine chloride channel protein of internal organelles (15), and electrophysiological and immunohistochemical studies of transfected CHO-K1 showing plasma membrane and nuclear membrane chloride ion channels.

NCC27 localizes principally to the nucleus, including the nuclear membrane, of both transfected and control-transfected CHO-K1, although small amounts were also noted in the cytoplasm and on the cell membrane. Patch clamp studies of isolated nuclei from transfected and control-transfected cells revealed chloride ion channels at the nuclear membrane with significantly more activity noted in the nuclei of clone 4-transfected cells. The detection of chloride channel activity in the nuclei of control-transfected cells coincides with the immunofluorescent staining and Western blot studies, which indicate that NCC27 is constitutively expressed in CHO-K1. Analysis of the NCC27 putative protein sequence revealed two nuclear localization motifs that may be acting as the target sequences for the migration and retention of NCC27 in the nucleus.

The ion channel activity detected in clone 4-transfected cells at their plasma membrane is likely to be largely due to spillover resulting from overexpression of the recombinant protein, since control-transfected and nontransfected CHO-K1 showed no ion channel activity at this location. This, nonetheless, proved useful for electrophysiological studies, since the channel-forming properties of NCC27 appeared to be the same at both locations, with the conductance of single channels at the cell membrane found to not be significantly different from that of the single channels recorded from the nucleus. This allowed initial characterization of the channel to be carried out at the plasma membrane, which is more easily accessible.

The immunohistochemical staining pattern observed in the CHO-K1 also revealed some cytoplasmic staining, suggesting that NCC27 may also be localizing to other intracellular organelles, as was reported for p64, where staining was noted in association with the Golgi and other compartments toward the periphery of the cell (30).

NCC27 and p64 are not homologous to any other ion channel proteins in the protein data bases, which would suggest that together they constitute a new class of ion channel proteins associated with internal organelles. NCC27 is not, however, the human homologue of p64, since Northern blot analysis of mRNA from human tissue reveals that the apparent human homologue of p64 has an mRNA of 6.5 kb (15) whereas NCC27 mRNA is only 1.2 kb. Furthermore, bovine p64 is a much larger protein (64 kDa) than NCC27 (27 kDa), and the amino-terminal half of p64 bears no homology to NCC27.

Our electrophysiological data also support the hypothesis that NCC27 and p64 constitute a new class of ion channel proteins. For example, the anion permeability sequence of the NCC27 chloride channel differs from that of the cystic fibrosis transmembrane regulator chloride channel (31) and other chloride channels that are associated with secretion in exocrine glands (32, 33, 34, 35). Furthermore, a chloride channel with a single channel conductance of 30 picosiemens from the inner nuclear envelope of sheep cardiac ventricular myocytes has recently been described (11).

The presence of free NCC27 protein in the nucleoplasm is a puzzling observation and is unusual for an ion channel protein. Furthermore, it is unclear how the protein is regulated between these two states, in a free form in the nucleoplasm and then translocating to the nuclear membrane. One possibility is that insertion of NCC27 into the nuclear membrane is a calcium-dependent process, as occurs in the annexin family (36). Alternatively, NCC27 may be acting as a regulatory subunit of a multiprotein chloride ion channel complex and therefore itself might not represent the ion-conducting unit of the chloride channel complex.

The two mRNA transcripts detected by clone 4 are up-regulated by various activation stimuli, including PMA and IL-2, which also results in the preferential expression of the 1.2-kb transcript. This regulation was also found to be associated with the maturational state of the cell. For example, the response of U937 cells to PMA activation was enhanced in those cells that had prior exposure to RA (a known differentiating agent of monocytoid cells) compared with those that had no prior RA treatment. While the two mRNA transcripts of NCC27 are believed to be alternate transcripts or alternate start sites, they are unlikely to involve the coding region of the protein and thus have no effect on the structure of the ion channel protein itself. The sum total of this data suggests that an increased number of NCC27 channel protein molecules may in some way be required for cell activation.

Northern blot analysis of clone 4 also revealed that the gene is expressed in a wide variety of cells and cell lines of human origin as well as in the murine cell line RAW 264.7 and in CHO-K1. Furthermore, the fact that the anti-NCC27 antibody cross-reacts with native protein in CHO-K1 indicates that the protein is not only widely expressed throughout various tissues and cells but is also highly conserved across species.

The wide distribution of clone 4 mRNA and its regulation by activation stimuli, as well as the fact that the encoding protein localizes to the nucleus and nuclear membrane, suggest that NCC27 is involved in some basic biological function in all cells. Such a role could be involvement in the regulation of either cell replication or gene transcription, where alteration of the nuclear chloride ion concentration may either serve to limit or facilitate this process. Its localization to the nuclear envelope and function as a nuclear ion channel suggest that it would act to control the ionic concentrations in the perinuclear cisternae as well as controlling transmembrane potentials.