Renal epithelial protein (Apx) is an actin cytoskeleton-regulated Na+ channel.

Apx, the amphibian protein associated with renal amiloride-sensitive Na+ channel activity and with properties consistent with the pore-forming 150-kDa subunit of an epithelial Na+ channel complex initially purified by Benos et al. (Benos, D. J., Saccomani, G., and Sariban-Sohraby, S. (1987) J. Biol. Chem. 262, 10613-10618), has previously failed to generate amiloride-sensitive Na+ currents (Staub, O., Verrey, F., Kleyman, T. R., Benos, D. J., Rossier, B. C., and Kraehenbuhl, J.-P. (1992) J. Cell Biol. 119, 1497-1506). Renal epithelial Na+ channel activity is tonically inhibited by endogenous actin filaments (Cantiello, H. F., Stow, J., Prat, A. G., and Ausiello, D. A. (1991) Am. J. Physiol. 261, C882-C888). Thus, Apx was expressed and its function examined in human melanoma cells with a defective actin-based cytoskeleton. Apx-transfection was associated with a 60-900% increase in amiloride-sensitive (Ki = 3 μM) Na+ currents. Single channel Na+ currents had a similar functional fingerprint to the vasopressin-sensitive, and actin-regulated epithelial Na+ channel of A6 cells, including a 6-7 pS single channel conductance and a perm-selectivity of Na+:K+ of 4:1. Na+ channel activity was either spontaneous, or induced by addition of actin or protein kinase A plus ATP to the bathing solution of excised inside-out patches. Therefore, Apx may be responsible for the ionic conductance involved in the vasopressin-activated Na+ reabsorption in the amphibian kidney.

Sodium reabsorption by polarized renal epithelia is initiated by the activation of apical epithelial Na ϩ channels whose activity is controlled by hormones, including vasopressin. However, molecular information about the protein(s) responsible for the vasopressin-sensitive, and protein kinase A-regulated Na ϩ conductance is ill-defined (6,7). A 130 -160-kDa protein, most abundant in the apical membrane of A6 renal tubular epithelial cells, has been recently cloned (1). The apical protein from Xenopus laevis or Apx 1 may correspond to the 150-kDa subunit of the A6 cell Na ϩ channel complex (2), which is responsible for Na ϩ channel activity (8). Expression of Apx in Xenopus oocytes, however, failed to induce amiloride-sensitive Na ϩ channel activity (1). In this report, the possibility was explored that Apx, although representing the epithelial Na ϩ channel pore, may be under tonic inhibition by proteins associated with the Na ϩ channel complex, including the actin cytoskeleton (9). Actin, for example, activates epithelial Na ϩ channels in A6 cells, which are also inhibited by actin-cross-linking proteins such as actinbinding protein (ABP) and its homolog filamin (3). Therefore, Apx was expressed in human melanoma cells (ABP(Ϫ)), deficient in ABP-280 (homolog of filamin). ABP(Ϫ) cells are devoid of an organized actin cytoskeleton and have defective ion channel regulation (10,11). Expression of Apx alone was sufficient to induce Na ϩ -selective and amiloride-sensitive ion channel activity, which was regulated by the actin cytoskeleton.

Cell Culture and Transfection of cDNA Encoding for Apx into Melanoma Cells
Cell Culture-The ABP-280-deficient human melanoma cell line (M2) was grown as described previously (10).
Plasmid Construction-The Blu-Apx vector, containing the fulllength cDNA encoding Apx and cloned to Bluescript M13-SK (Stratogen, CPI), was digested with the restriction enzymes NotI and XhoI. The 4.9-kilobase pair fragment containing the full-length Apx cDNA was purified and subcloned to the eukaryotic expression vector pcDNA3, containing a neomycin-resistant gene promoter (Invitrogen, CA), to form pcDNA3-Apx.
Melanoma Cell Transfection-Transfection of M2 cells with pcDNA3-Apx was conducted by the calcium phosphate precipitation technique as described previously (10). After recovery, cells were grown in a selection medium containing G418 (1 mg/ml) which efficiently kills these cells (10). Six G418-resistant clones were picked after 16 -18 days and further cultured individually.

Immunocytochemistry
Immunocytochemistry was performed as described previously for cultured cells (12). Briefly, the various cell lines were grown on glass coverslips for 2-4 days (80% confluent), as for the patch-clamp experiments. Cells were fixed with either paraformaldehyde-lysine-periodate or 4% paraformaldehyde in phosphate-buffered saline (PBS) for 2 h at room temperature, followed by cell permeabilization with 0.1% Triton X-100 for 4 min. After incubation with PBS containing 1% bovine serum albumin to block nonspecific binding (10 min), coverslips were incubated with the primary antibody diluted in PBS. After extensive washing, goat anti-rabbit IgG coupled to Cy3 (indocarbocyanine, Jackson ImmunoResearch Laboratories, West Grove, PA) was applied (1:400). After further washing in PBS, the coverslips were mounted in Airvol 205 (polyvinyl alcohol; AIR Products, Allentown, PA), sealed, and examined with a Nikon FXA fluorescence microscope. Representative islands of cells were photographed using Kodak TMax 400 film, pushprocessed to 1600 ASA. To quantify the amount of Apx labeled in Apx-7, M2, and A6 cells, pictures from immunocytochemistry studies were printed using identical exposure conditions, scanned, and gray-scale analyzed with NIH image software (version 4.1).

Antibodies
The primary antibody was an immunopurified rabbit polyclonal antibody raised against an Apx fusion protein containing Apx COOH terminus (amino acids 1194 to 1395), and diluted 1:10 (25 g/ml, Fig.  1A). For Fig. 1B, the antibody-containing whole serum was used at a 1:100 dilution.

Whole-cell Current Studies
Actual currents and command voltages were obtained as described previously (11). Currents and command voltages were obtained and driven with a PC-501 patch-clamp amplifier using a 1 gigaohm headstage (Warner Instruments, Hamden, CT) or with a Dagan 3900 (Dagan Corporation, Minneapolis, MN). Signals were filtered at 1 kHz with an eight-pole Bessel filter (Frequency Devices, Haverhill, MA). Data were stored in a hard disk of a personal computer and analyzed with PClamp 5.5.1 (Axon Instruments, Burlingame, CA). Holding potentials refer to the patch-pipette. The patch-pipette and bathing solution was, in mM: 140 NaCl, 5.0 KCl, 0.8 MgSO 4 , 1.2 CaCl 2 , and 10 Hepes, pH 7.4. When indicated, Na ϩ was replaced by an equimolar concentration of K ϩ , all other solutes remaining the same. Likewise in some cases Cl Ϫ was replaced by isethionate.

Single Channel Studies
The cell-attached and excised, inside-out patch-clamp configurations were carried out as described previously (3). Currents and command voltages were obtained and driven with a PC-501 patch-clamp amplifier using a 10 gigaohm headstage (Warner Instruments, Hamden, CT) or with a Dagan 3900 (Dagan Corp., Minneapolis, MN) and further processed as indicated for the whole-cell experiments. Data were further filtered at 25-50 Hz for display purposes. Data from cell-attached or excised, inside-out patches were obtained between Ϯ120 mV, with upward and downward deflections indicating the channel open state at positive and negative holding potentials, respectively. Patch-pipette and bathing solutions were as described for the whole-cell experiments.

Membrane-enriched Preparations
Cultures of confluent M2 and Apx-7 cells were scraped, centrifuged at 3,000 rpm (4°C) for 10 min, and washed twice with PBS. The cell membrane pellets were resuspended in fresh buffer (1 ml) containing, in mM, sucrose 250, Tris-base (pH 7.6) 10, NaCl 50, phenylmethylsulfonyl fluoride 1.0, leupeptin 10 g/ml, and aprotinin 20 g/ml. The suspensions were sonicated at 45% duty cycle (Ultrasonic Processor, model W-375, Ultrasonics Inc.) for two 10-s runs. After addition of 9 ml of protease-inhibitor buffer, sonication was repeated, and the suspensions were centrifuged again at 3,000 rpm for 10 min. The supernatants were further ultracentrifuged at 25,000 rpm for 1 h at 4°C in an L8-80 M Ultracentrifuge (Beckman, Palo Alto, CA), using a swing rotor SW 41 Ti. Pellets were resuspended in 200 l of protease-inhibitor buffer and stored at Ϫ80°C until further use. Protein content was determined using the Bradford method (Bio-Rad).

Planar Lipid Bilayer Studies
Membrane-enriched vesicles were either fused to planar lipid bilayers painted onto a 0.1-mm hole in a 13-mm reconstitution polystyrene cuvette (Warner Instrument Corp., Hamden, CT), as described by Alvarez (13), or painted directly onto the lipid bilayer. The phospholipid composition of the lipid bilayers was 1-palmitoyl-2-oleyl-sn-glycero-3phosphatydylethanolamine:1-palmitoyl-2-oleyl-sn-glycero-3-phosphatydylcholine (7:3, v/v, Avanti Polar Lipids, Alabaster, AL) in n-decane (Aldrich) to a final concentration of 14 and 6 mM, respectively. Two g of membrane-enriched preparations from Apx-7 or M2 cells (diluted 2:3 with 2 M sucrose) were either added into the cis chamber or directly painted onto the lipid bilayer with a glass rod. The cis and trans solutions contained NaCl 600 mM, MOPS 10 mM, CaCl 2 20 M, pH 7.0, and NaCl 50 mM, MOPS 10 mM, CaCl 2 20 M, pH 7.0, respectively. Input signals were acquired with a PC-501A patch/whole-cell clamp amplifier via a 10 Gohm head-stage for lipid bilayers (Warner Instrument Corp., Hamden, CT). The output currents were low pass filtered at 1 kHz with an eight-pole Bessel filter (Frequency Devices), digitized at 37 kHz, 14 bits with a VR-10B digital data recorder (Instrutech Corp., Great Neck, NY), and stored on a video cassette recorder (JVC Corp., Fairfield, NJ) until further analysis with PClamp 5.5.1 (Axon Instruments, Burlingame, CA) as for the patch-clamp studies.

Actin and Associated Proteins
G-actin (Sigma) was used without further purification and was stored until the time of the experiment at Ϫ80°C in 200-l aliquots at approximately 5-10 mg/ml in a solution (depolymerizing buffer) containing, in mM: Tris-HCl, 2; ATP, 0.5; CaCl 2 , 0.2; and ␤-mercaptoethanol, 0.5; pH 8.0.

Data Analysis
Data were expressed as the mean Ϯ S.E., where n equals the number of patches analyzed. The mean conductance was obtained by regression analysis. Data were compared by Student's t test. The Goldman-Hodgkin-Katz equation was used to calculate the perm-selectivity ratio (P Naϩ /P Kϩ ) from the whole-cell and excised inside-out patch-clamp experiments where Na ϩ was equimolarly replaced by K ϩ : where E r is the measured reversal potential; P Naϩ and P Kϩ , are the Na ϩ and K ϩ permeabilities, respectively; and K p , Na p and K b , Na b , are the Na ϩ and K ϩ concentrations in pipette (p) and bathing solution (b), respectively. For the amiloride inhibition studies, the percent change of the whole-cell conductance as a function of the various concentrations of amiloride (A) was bestfitted to one binding site using the equation: 100 ϫ ␥/␥ ctrl ϭ 100 Ϫ {100(A/(A ϩ K i ))} from which the inhibition constant K i was obtained.

RESULTS
Six neomycin-resistant clones, named Apx-2, and Apx-4 through Apx-8, were produced under standard selection conditions (G418, 1 mg/ml) by stable transfection of M2 (ABP(Ϫ)) melanoma cells with the full-length cDNA for Apx. The presence of Apx in the G418-resistant clones was first determined by immunolocalization of Apx with immunopurified anti-Apx antibodies raised against a fusion protein containing the amino acid sequence 1194 to 1395 from Apx. Immunocytochemical analysis indicated that all six clones expressed various amounts of intracellular Apx. A punctate plasma membrane distribution of Apx, however, was also observed in Apx-7 cells (Fig. 1A). No labeling of Apx was detected in the parent, nontransfected human melanoma cells (M2, Fig. 1A). Immunocytochemical analysis, however, showed no specific staining for the recently cloned epithelial Na ϩ channel, rENaC (14), in the M2 cells (data not shown). Quantification of Apx labeling was carried out after staining with serum containing antibodies against Apx. The immunofluorescence labeling indicated that the relative fluorescence observed in Apx-7 cells was statistically higher than in M2 cells (background fluorescence, Fig.  1B). As a positive control, immunocytochemical analysis of Apx was also assessed in A6 epithelial cells, which expresses high levels of the protein, and which displayed labeling comparable to that of Apx-7 (Fig. 1B).
after replacement of Na ϩ by K ϩ , thus consistent with a permselectivity ratio (P Naϩ /P Kϩ ) of 4.3. The dose-response decrease of the whole-cell currents by amiloride indicated an affinity of 3.3 M (Fig. 3A), and a maximal inhibition of 62% (3.3 Ϯ 0.1 nS/cell, n ϭ 6, p Ͻ 0.001). Apx-7 whole-cell currents were also inhibited by the amiloride analog benzamil (Ͻ1 M) but not by N-(ethyl-N-isopropyl)-amiloride (100 M; data not shown). A strong positive correlation was found between Apx expression by the various Apx clones and the whole-cell conductance (as indicated below), thus indicating a direct link between the two parameters (Fig. 2D). Although most clones (4/6) had a higher cationic conductance compared to M2 cells (Fig. 2D, Table I), no correlation was found between the whole-cell and the amiloride-sensitive conductance, thus most clones were insensitive to amiloride (Fig. 3B). Cyclic AMP stimulation of Apx-7 cells induced a 264 Ϯ 136% (n ϭ 8) increase of the whole-cell currents in 8 out of 14 experiments (57%, p Ͻ 0.05), although no effect was observed on M2 cells (Ϫ47 Ϯ 11%, n ϭ 6, NS). Addition of the actin cytoskeleton disrupter cytochalasin D (10 g/ml), however, had no effect on the whole-cell currents of either M2 or Apx-7 cells (72 Ϯ 48%, n ϭ 6, and 44 Ϯ 34%, n ϭ 4, respectively).
The molecular nature of the Apx-induced Na ϩ currents was further studied at the single channel level (Fig. 4). Spontaneous ion channel activity was observed in 50% (n ϭ 28) of cell-attached patches. The current-voltage relationship (Fig.  4C) was highly similar to that of Na ϩ channels previously reported in A6 epithelial cells (4,15). After excision, Na ϩ channel currents had a single channel conductance of 6.2 Ϯ 0.6 pS (n ϭ 5) in symmetrical Na ϩ (140 mM NaCl or sodium isethionate, Fig. 4, A and B). In 3/8 quiescent patches (38%), addition of protein kinase A plus ATP (1 mM) induced Na ϩ -selective single channel currents of 7.2 Ϯ 0.2 pS (n ϭ 3, Fig. 4D). Addition of monomeric actin (1 mg/ml), in the presence or in the absence of ATP (1 mM), to excised, inside-out patches of Apx-7 cells, also induced and/or increased Na ϩ channel activity in 78% of the experiments tested (n ϭ 9, Fig. 4, F and G). Actininduced Na ϩ channels had a linear single channel conductance of 6.2 Ϯ 0.3 pS (n ϭ 6, Fig. 4H) and a perm-selectivity sequence of Na ϩ Ͼ Li ϩ Ͼ K ϩ (Fig. 4, H and I) with a 4:1 perm-selectivity ratio between Na ϩ and K ϩ (Fig. 4H).
Membrane-enriched fractions from Apx-7 cells were also reconstituted into lipid bilayers to assess for cation-selective ion channel activity (Fig. 5). Either spontaneous or actin-induced ion channel activity was observed in 4/4 experiments in symmetrical 150 mM NaCl (Fig. 5A) and 3/3 experiments in 300 mM NaCl (Fig. 5B). Reconstituted channels had a single channel conductance of 11-15 pS (n ϭ 4) and 18 pS (n ϭ 3) under either symmetrical 150 mM or 300 mM NaCl, respectively. Addition of 10 M amiloride abolished ion channel activity under either condition. In asymmetrical NaCl (600:50 mM NaCl, cis:trans, respectively) spontaneous Na ϩ channel activity was observed only in 1/14 experiments but was readily activated in 9/14 experiments by addition of actin Ϯ ATP to the cis side of the chamber (Fig. 5C). Reconstituted channels had a single channel conductance of approximately 11 pS (n ϭ 3, Fig. 5C), rectified in a cation-selective manner, and were inhibited by addition of amiloride (0.1 M) added to the trans but not to the cis-side of the chamber (Fig. 5C). The reversal potential (E r ), 60.9 Ϯ 7.7 mV (n ϭ 3), was consistent with the expected value of 63.6 mV for Na ϩ -selective currents. A perm-selectivity value P Naϩ /P Kϩ of 4 -6 was obtained in asymmetrical Na ϩ /K ϩ conditions (150 mM NaCl in cis, 150 mM KCl in trans). Membranepreparations from M2 cells showed no channel activity (n ϭ 4).

DISCUSSION
The first step to the transepithelial Na ϩ movement of transporting epithelia entails the selective movement of this ion into the cellular compartment. The molecular nature of this first step, although known to involve apical ion channels, is largely unknown. Previous experimental evidence and current studies by various groups including ours indicate that three "types" of apical Na ϩ channels can be found in A6 and related epithelial cells (6), which have been recently characterized in the small conductance (ϳ3-5 pS), highly selective Na ϩ channel with a Na ϩ :K ϩ perm-selectivity ratio higher than 10 (16, 17), a higher mean conductance (7-15 pS) Na ϩ channel with a lower permselectivity ratio (3 to 5) (3, 15, 18, 19), and yet another nonse-lective cation channel with an even higher (23-28 pS) single channel conductance (20). No molecular information is yet available on the various molecular structures underlying the different functional apical channel fingerprints, except for recent studies providing strong evidence to suggest that the small channel type may represent the recently cloned, heterotrimeric EnaC (6). Functional expression of epithelial sodium channel homologs (14) is largely consistent with the channel(s) described by extensive studies of Palmer and Eaton's laboratories (16,17,21,22). The molecular structure underlying the 9 pS Na ϩ channel, however, is apparently unavailable (for a review, see Ref. 6).
Whether one or all of the Na ϩ -permeable apical epithelial channel types may represent different conductance states of the same structure, and/or can function in different "conductance" modes is, despite current interest, as yet unknown. The possibility exists, for example, that a particular ion channel may have more than one functional mode. Various functional studies of the apical channel complex originally described by Sariban-Sohraby et al. (23) and Benos et al. (24,25) indicate that this channel structure may have more than one single channel conductance. Thus changes in the apical Na ϩ permeability of epithelial cells may be the reflection of one or more, not mutually exclusive, possibilities, including the fact that a particular Na ϩ channel may change its conductance properties depending on its environmental conditions (23,24), and/or that different channels may be selectively expressed under different developmental conditions (6). Our studies on the A6 Na ϩ channel, for example, have been conducted on subconfluent cells grown on glass coverslips. Under these conditions single channel currents consistent with the "highly selective" Na ϩ channel are rarely observed which are a common observation of steroidtreated confluent A6 cells grown on permeable supports (16,17,26). Although further studies will be required to explore these various, yet not mutually exclusive, possibilities, previous independent studies have determined the presence of an apical epithelial multimeric protein complex (27). This channel complex contains at least one subunit (150 kDa) which has been directly implicated in the amiloride-sensitive Na ϩ channel activity (2).
Apx is a 120 -170 kDa protein associated with the apical epithelial Na ϩ channel complex (1). The 150-kDa subunit of the A6 Na ϩ channel complex displays Na ϩ channel activity in lipid bilayers (8). Thus, Apx and the 150-kDa subunit may represent the same transmembrane protein. This is further suggested by the fact that antibodies raised against the 150-kDa subunit immunoprecipitate Apx (1) and conversely, antiidiotypic antibodies to Apx immunoprecipitate the 150 kDa subunit of the Na ϩ channel complex (1). The possibility was raised that Apx may be the pore-bearing component of the renal epithelial Na ϩ channel. However, previous studies failed to detect amiloridesensitive Na ϩ channel activity after expression of Apx in Xenopus oocytes (1). This may be accounted for by two different possibilities. First, Apx might be a channel regulator. Another possibility is that although representing an ion channel, Apx could be functionally inhibited by other regulatory proteins. Actin, for example, not only activates Na ϩ channels in A6 epithelial cells (3) but is also responsible for the vasopressininduced and protein kinase A-mediated Na ϩ channel activation (5). Further, actin-binding proteins, including the ABP-280 homolog filamin (3), inhibit spontaneous, protein kinase A-and actin-induced Na ϩ channel activity in A6 cells (3,5). The cell volume-regulated activation of cation channels in human mel-anoma cells was also tonically inhibited by ABP-280 (11). Thus, human melanoma cells devoid of ABP-280 may provide a useful model to assess ion channel-actin cytoskeleton interactions.
The data in this report indicate that expression of Apx in cytoskeletally deranged human melanoma cells is associated with ion channels having functional similarities with both the 9-pS apical Na ϩ channel of A6 cells (3,4,15,18,19), and with the pore-forming 150-kDa subunit of the epithelial Na ϩ channel complex originally purified by Benos et al. (2) and Sariban-Sohraby et al. (23). The functional similarities between Apx and the "slightly" Na ϩ -selective apical channel of A6 cells previously reported by Hamilton and Eaton (15), and our own laboratory (3,4,18,19), include an amiloride-sensitive single channel conductance of 6 -9 pS, and a Na ϩ :K ϩ perm-selectivity ratio of 3 to 5. Further distinction between Apx and the highly selective Na ϩ channel entails the perm-selectivity cation se-  quence Na ϩ Ͼ Li ϩ Ͼ K ϩ observed, instead of Li ϩ Ͼ Na ϩ Ͼ K ϩ as described for the highly selective Na ϩ channel (22). However, further support for functional similarities between the 9-pS Na ϩ channel and Apx are based on the fact that this channel protein is the target for protein kinase A activation, a regulatory process which is mediated by the actin cytoskeleton, which can also "bypass" the effect of protein kinase A phosphorylation on channel activation. This does not preclude that other channel proteins are also regulated by actin, nor that protein kinase A regulation is mediated by the cytoskeleton, as we have recently demonstrated for cystic fibrosis transmembrane conductance regulator (28).
Apx transfection into M2 human melanoma cells devoid of the filamin homolog ABP-280 was associated with amiloridesensitive Na ϩ channel activity, which was further modulated by actin. This raised the possibility that Apx may directly interact with actin. A comparison between the actin-binding domain(s) of various actin-binding proteins and Apx indicated that at least two conserved actin binding consensus sites are present in Apx (Table II). Both domains are located close to the putative main cytoplasmic loop of Apx. The first putative actin binding domain showed strong homology with both spectrins, and dystrophins (29,30). Interestingly, recent studies indicate that ENaC interacts with spectrin (31), thus suggesting that similar regulatory features may be shared by both epithelial ion channels. Another putative actin binding domain in Apx shared homology with the actin binding domains of the actinsevering proteins gelsolin and villin, in a region of Apx containing at least five amino acids essential for the actin-binding ability of these proteins. Thus, direct binding of actin to Apx may be critically involved in the regulation of its ion channel activity. The data in this report indicate that Apx may represent the vasopressin-sensitive, and actin-regulated renal Na ϩ channel of A6 cells (3,5), whose regulation by actin may entail a novel feature of ion channels.

TABLE II Comparison between Apx and actin-binding proteins with actin cross-linking and cleaving motifs
Sequence comparison between the actin binding domains of actin-binding proteins containing actin cross-linking domains (A) and actin severing domains (B). Sequences corresponding to ␤-spectrin and dystrophin were obtained from Matsudaira (32). Sequences corresponding to gelsolin and villin were obtained from Arpin et al. (33). The numbers indicate the first amino acid of the protein sequence.

Protein
Amino acid Sequence a Similarity/identity % A. Human