A Region Directly Following the Second Transmembrane Domain in γENaC Is Required for Normal Channel Gating*

We used a yeast one-hybrid complementation screen to identify regions within the cytosolic tails of the mouse α, β, and γ epithelial Na+ channel (ENaC) important to protein-protein and/or protein-lipid interactions at the plasma membrane. The cytosolic COOH terminus of αENaC contained a strongly interactive domain just distal to the second transmembrane region (TM2) between Met610 and Val632. Likewise, γENaC contained such a domain just distal to TM2 spanning Gln573–Pro600. Interactive domains were also localized within Met1–Gln54 and the last 17 residues of α- and βENaC, respectively. Confocal images of Chinese hamster ovary cells transfected with enhanced green fluorescent fusion proteins of the cytosolic tails of mENaC subunits were consistent with results in yeast. Fusion proteins of the NH2 terminus of αENaC and the COOH termini of all three subunits co-localized with a plasma membrane marker. The functional importance of the membrane interactive domain in the COOH terminus of γENaC was established with whole-cell patch clamp experiments of wild type (α, β, and γ) and mutant (α, β, and γΔQ573-P600) mENaC reconstituted in Chinese hamster ovary cells. Mutant channels had about 13% of the activity of wild type channels with 0.33 ± 0.14 versus 2.5 ± 0.80 nA of amiloridesensitive inward current at –80 mV. Single channel analysis of recombinant channels demonstrated that mutant channels had a decrease in Po with 0.16 ± 0.03 versus 0.67 ± 0.07 for wild type. Mutant γENaC associated normally with the other two subunits in co-immunoprecipitation studies and localized to the plasma membrane in membrane labeling experiments and when visualized with evanescent-field fluorescence microscopy. Similar to deletion of Gln573–Pro600, deletion of Gln573–Arg583 but not Thr584–Pro600 decreased ENaC activity. The current results demonstrate that residues within Gln573–Arg583 of γENaC are necessary for normal channel gating.

Activity of integral membrane proteins is regulated, in general, by the following two means: post-translational modifica-tion, and discretionary interaction with accessory, regulatory proteins and/or lipids. These two modalities of regulation are not necessarily mutually exclusive and impact function by influencing several parameters, including protein localization and kinetics. Ion channels are integral membrane proteins that play fundamental roles in many diverse cellular processes. Similar to other membrane proteins, ion channel activity is, in part, a manifestation of channel kinetics and cellular locale.
The amiloride-sensitive epithelial Na ϩ channel (ENaC) 1 is an ion channel localized to the luminal plasma membrane of epithelial cells (1)(2)(3). Activity of this channel is the rate-limiting step in Na ϩ transport across electrically tight epithelium. Thus, ENaC plays a pivotal role in Na ϩ and concomitant water (re)absorption across many epithelial tissues. This channel, consequently, is centrally positioned as an effector for systemic hormones and other factors that modulate blood pressure. Gain and loss of function mutations in ENaC and its regulatory pathways, indeed, cause blood pressure disorders in humans associated with aberrant Na ϩ and water metabolism (4). Although it is accepted that ENaC activity is dynamically modulated by regulation of channel localization to the luminal membrane, little is actually known about the cellular control points and queues impinging upon this modulation. In addition, the specific residues and domains within the channel itself important to localization and control of channel activity remain obscure.
ENaC is a member of the Deg/ENaC superfamily of ion channels (3,5). This superfamily contains a functionally diverse array of channels that all share a common tertiary structure with members having two-transmembrane spanning regions, a large extracellular ectodomain and two short cytosolic tails. Channels within this superfamily play important roles in sensory perception, including taste, touch, hearing, nociception, and neurotransmission, as well as vectorial Na ϩ transport across epithelia. In native epithelia, ENaC is composed of three homologous but distinct subunits: ␣, ␤, and ␥. Canessa et al. (6,7) and Lingueglia et al. (8) were the first to identify the molecular correlates of ENaC. Most results suggest that the functional channel has a stoichiometry of two ␣ and one ␤ and ␥ subunit (9,10); however, the alternative that the functional channel is composed of three copies of each of the three subunits has also been proposed (11). Heterologously expressed ␣ENaC alone and together with either ␤or ␥ENaC also forms homomeric and heterodimeric channels, although with much de-creased activity and slightly different biophysical characteristics from the endogenous channel in native epithelia (6,7,(12)(13)(14).
The cytosolic tails of ENaC are believed to be regulatory domains and/or effector sites that impinge on channel gating and locale. Recent findings from our laboratory showing that the NH 2 terminus of ␣ENaC and the COOH termini of all three subunits contain domains involved in protein-protein and/or protein-lipid interactions localized to the plasma membrane are consistent with such a possibility (15). Moreover, deletion of the entire NH 2 terminus from any subunit inactivates ENaC (16). Conversely, deletion of the complete COOH tail of ␤and ␥but not ␣ENaC activates the channel (17,18). Deletion of the latter half of the COOH tail of ␣ENaC, however, does increase activity (19). These results, as well as others, suggest that the cytosolic tails of ENaC are involved in both positive and negative regulation of channel activity. Interestingly, the COOH terminus of ENaC subunits are the least well conserved portions of the channel. It has been hypothesized that these regions may impart the well described tissue-and species-specific regulation of ENaC by allowing differential interaction with site-specific intermediary effector/accessory proteins (5).
The most well described regulation of ENaC involving the cytosolic tails of the channel is down-regulation of activity upon binding of the ubiquitin ligase Nedd4. The WW domains within Nedd4 target this protein and similar ligases to PY motifs (XPPXY) in the distal portions of the ENaC COOH termini promoting ubiquitination of the NH 2 terminus of ␣and ␥ENaC subunits and subsequent internalization of the channel (20,21). The cytosolic COOH tails of ENaC also contain a tyrosinebased endocytic tag overlapping the PY motif (YXXL) (22) that in some instances is functionally independent of the PY motif at least in ␥ENaC (15). Moreover, COOH tails contain SH3 binding domains (23). Such a domain in ␣ENaC binds the SH3 domain within ␣-spectrin and has been implicated in localizing the channel to the luminal membrane in epithelia. In addition, the COOH termini of ␤and ␥ENaC may impact ENaC open probability by promoting channel closing (24,25). All of these COOH-terminal domains, described previously (15), are more distal than the membrane reactive domain we recently identified in the COOH tail of ␥ENaC.
The NH 2 terminus of ␣ENaC is required for normal channel function (16). Overexpression of a peptide containing this region of ␣ENaC acts as a competitive inhibitor of wild type channels. Channels missing the first 109 residues of ␣ENaC, in addition, have decreased activity; however, they localize to the plasma membrane (16,26). This region also contains another possible endocytic tag (KGDK) (16) and a well conserved 5amino acid tract containing a glycine (Gly 95 ) crucial to normal channel gating (27). Interestingly, ␣-rENaC is differentially spliced to produce ␣ENaC subunits with unique NH 2 termini (28). The functional ramifications of this have yet to be determined.
Similar to the NH 2 terminus and to the COOH terminus of the other subunits, the COOH terminus of ␣ENaC plays a role in modulating channel activity. Binding of actin to the COOHtail of ␣-rENaC increases channel open probability but decreases conductance (24). The COOH terminus of ␣ENaC also contains a region that supports channel activity and is involved in kinase regulation of the channel (19). In particular, residues Pro 595 and Gly 596 in ␣ENaC are critical to normal localization of the channel to the plasma membrane.
Thus, there is convincing evidence that the cytosolic tails of ENaC subunits affect channel activity by impacting both channel locale and gating. However, only a few residues and specific domains within these regions of ENaC have been identified in detail and linked to function. Gupta and Canessa (29) reported previously that heterologous expression of ␣and ␤-rENaC results in yeast becoming salt-and amiloride-sensitive, demonstrating that this recombinant channel is active in this background. In the current study, we built on this earlier work by

EXPERIMENTAL PROCEDURES
Materials-All chemicals were of reagent grade and purchased from either Sigma or Fisher unless noted otherwise. The BCA Protein Assay was from Pierce. All materials used in Western blot analysis were from Bio-Rad. The monoclonal anti-Myc antibody was from Clontech (Palo Alto, CA), and the anti-HA was from Roche Applied Science. Antimouse horseradish peroxidase-conjugated 2 o antibody was from Kirkegaard & Perry Laboratories (Gaithersburg, MD). ECL reagents were from PerkinElmer Life Sciences. All DNA sequencing was performed by the molecular biology core facility at the University of Texas Health Science Center, San Antonio. The Saccharomyces cerevisiae cdc25H yeast strain (cdc25H: MAT␣ ura3-52 his3-200 ade2-101 lys2-801 trp1-90 leu2-3,112 cdc25-2 (ts) Gal ϩ ) and the pSOS plasmid were from Stratagene (La Jolla, CA). The pECFP-M, pEGFP-F, pDsRed2-N, pCMV-Myc, and pCMV-HA plasmids were from Clontech. The plasmids encoding mouse ENaC subunit cDNAs have been described previously (30) and were the gift from Dr. T. R. Kleyman.
Mutagenesis of pSOS-ENaC and pMyc-␥ENaC Constructs-Mutagenesis of pSOS-ENaC and its derivatives was completed using QuikChange (Stratagene) site-directed mutagenesis per the manufacturer's instructions. The primers and templates used to create the deletion and truncation mutations used in the current study are listed in Table I. The pMyc-␥ENaC ⌬Q573-P600 deletion mutant was generated with 5Ј-GCCGCCAGTGGGCCCTGGATACGG and 5Ј-CCGTATC-CAGGGCCCACTGGCGGC upstream and downstream primers, respectively, in conjunction with full-length pCMV-myc-␥ENaC. The pMyc-␥ENaC ⌬T584-P600 and -␥ENaC ⌬Q573-R583 deletion mutants were generated with the 5Ј-CCCGTAGGCGGGCCCTGGATACG and 5Ј-CG-TATCCAGGGCCCGCCTACGGG upstream and downstream primers, respectively, and the 5Ј-GCCGCCAGTGGACACCACCCTCC and 5Ј-GGAGGGTGGTGTCCACTGGCGGC upstream and downstream primers, respectively, in conjunction with full-length pCMV-myc-␥ENaC. All constructs were sequenced to ensure proper mutagenesis and to confirm orientation, reading frame, and sequence identity.
Yeast One-hybrid Complementation Screen-The one-hybrid screen used in the current study has been described previously (15). In brief, S. cerevisiae cdc25H yeast transformed with pSOS-ENaC constructs were initially plated on SD/glucose (ϪLeu) agar at 24°C. Colonies were allowed to form over a 4-day period. Colonies were then patched from the source plate in duplicate onto SD/glucose (ϪLeu) plates to create two identical arrays. Arrays were grown in parallel at permissive (24°C) and restrictive (37°C) temperatures. Colony formation (growth) was quantified after 2 days (refer to Fig. 1). Cdc25 is the yeast homologue of hSOS and is required for Ras-dependent growth. The yeast strain used in the current study has temperature-sensitive Cdc25 that is not functional at restrictive temperatures (37°C). Because pSOS encodes a truncated hSOS incapable of membrane localization, hSOS-ENaC fusion proteins only activate Ras signaling and thus initiate yeast growth when information in the ENaC portion of the hybrid protein enables the full fusion protein to localize at or near the plasma membrane. Digital images of yeast plates were captured with a DC4800 Zoom Digital Camera (Eastman Kodak Co.) interfaced with a personal computer running Kodak IFS Core software.
ENaC Expression in CHO Cells-CHO cells were maintained in culture with Dulbecco's modified Eagle's medium supplemented with 10% FBS and antibiotics (penicillin and streptomycin) by using stand-ard methods (31). For patch clamp analysis and confocal imaging, cells were plated on coverglass chips treated with 0.01% polylysine. Plated cells were transfected with four plasmids encoding ␣-, ␤-, and ␥ENaC and GFP using the PolyFect reagent (Qiagen, Valencia, CA) per the manufacturer's recommendations. In brief, cells ϳ60% confluent in a 35-mm dish were treated with 2.5 g of total plasmid cDNA for 24 -48 h. Cells were used for patch clamp analysis up to 96 h after transfection and were maintained in culture in the presence of 10 M amiloride replenished daily. Cells used for protein analysis were grown in 100-mm dishes, transfected with 4 g of total plasmid cDNA, and extracted 24 -48 h after transfection.
Confocal Imaging-Transfected CHO cells were grown on number 0 coverglass, fixed in 4% paraformaldehyde, and mounted using Vectashield (Vector Laboratories, Burlingame, CA). Confocal images were collected using a ϫ60 (1.3 NA) oil-immersion lens on a Nikon Eclipse TE2000 (Nikon Instruments, Melville, NY) inverted microscope fitted with a Cascade Photometric CCD camera (Roper Scientific, Tucson, AZ), the CARV confocal fluorescence imaging unit (Kinetic Imagine, Weston, Ontario, Canada), and a Lambda 10-2 filter wheel (Sutter Instruments, Novato, CA). This unit is driven by the Metamorph program suite (Universal Imaging Corp., Downingtown, PA) and interfaced with a piezosystem (Piezosystem Jena, Hopedale, MA) for Z-series imaging. For DsRed2 a triple pass polychroic emission filter (D/F/TXRD 62002; Chroma Technology Corp., Brattleboro, VT) was used in conjunction with a single pass excitation filter. For ECFP, images were collected using the yellow fluorescent protein/cyan fluorescent protein dual pass polychroic emission filter (cyan fluorescent protein/yellow fluorescent protein 51017v2; Chroma Technology Corp.) in conjunction with a single pass excitation filter. EGFP was visualized with a single pass dichroic emission filter (endow GFP 41017; Chroma Technology Corp.) in conjunction with a single pass excitation filter. With these filter sets, fluorophores were easily discriminated with no bleed through (see Fig. 2).
Evanescent-field (EF) Fluorescence Microscopy-To selectively illuminate the plasma membrane and its associated channel subunits, we used EF microscopy. Cells used for EF microscopy were plated on glass coverslips and fixed as above for confocal imaging. Methods followed closely those described previously by Almers and colleagues (32,33). In brief, EF microscopy was performed using an inverted TE2000 microscope with through-the-lens fluorescence imaging. EF illumination was generated by total internal reflection fluorescence (TIRF) after the light beam struck the interface between the glass coverslip and cellular plasma membrane at a glancing angle (34). Samples were viewed through a Plan Apo TIRF ϫ60 oil-immersion, high resolution (1.45 NA) objective (Nikon). TIRF generates an EF that declines exponentially with increasing distance from the interface between the cover glass and plasma membrane illuminating only a small optical slice of the cell (ϳ200 nm) including the plasma membrane. Thus, with TIRF only fluorophores in the plasma membrane and its immediate vicinity contribute to emission, whereas those deeper in the cell do not (see Fig. 8A). DsRed2 and EGFP-F were excited with green HeNe and argon lasers, respectively, with emissions subsequently passing through 543-and 488-nm single pass filters, respectively. This system was also interfaced with a mercury lamp with appropriate dichroic excitation and emissions filter sets enabling wide field epifluorescence imaging of DsRed2 and EGFP-F. Images were collected and processed as above with a CCD camera interfaced to a PC running Metamorph software.

Patch Clamp Recording and Single Channel Analysis-Whole-cell macroscopic current recordings of ENaC reconstituted in CHO cells
were made under voltage clamp conditions using standard methods (31). Prior to patch clamp analysis, cells were rinsed of culture media and amiloride and patched at room temperature under constant perfusion in a bath solution of (in mM) 160 NaCl, 1 CaCl 2 , 2 MgCl 2 , and 10 HEPES (pH 7.4, 320 mOsm). Pipette solution was (in mM) 145 KCl, 5 NaCl, 2 MgCl 2 , 0.5 CaCl 2 , 10 EGTA, 10 HEPES (pH 7.4), 3.0 ATP, and 0.1 GTP (330 mOsm). Current recordings were acquired with a PC-505B patch clamp amplifier (Warner Instruments; Hamden, CT) interfaced via a Digidata 1320A (Axon Instruments, Union City, CA) with a PC running the pClamp 8.1 suite of software. All currents were filtered at 1 kHz. Both a family of test pulses stepping by 20-mV increments (500 ms each separated by 400 ms) form Ϫ120 to ϩ100 mV, and voltage ramps (100 ms) over the same range were used to generate currentvoltage (I-V) relations. The whole-cell capacitance was routinely compensated and was approximately ϳ12 picofarads for CHO cells. Series resistances, on average 2-5 megohms, were also compensated. Currents, however, were not leak-corrected. For all experiments, holding potential was 30 -50 mV. For voltage steps, steady state whole-cell currents were routinely measured 100 -200 ms from the start of each pulse.
Single channel current recordings were performed as described previously (35)(36)(37). In brief, all experiments were performed at room temperatures with fire-polished pipettes of borosilicate glass (World Precision Instruments, Sarasota, FL) with tip resistances 4 -7 megohms. All recordings were made in excised, outside-out patches with pipette and FIG. 2. Confocal images of EGFP-ENaC tail hybrids expressed in CHO cells. CHO cells were transfected with both the respective EGFP-ENaC tail hybrid indicated on the left and a ECFPmembrane marker with the exception that EGFP-␥ M1-W56 was expressed with a DsRed2 nuclear marker (noted by asterisk). The 1st column shows localization of fusion proteins collected under EGFP conditions. The middle column shows localization of the co-expressed membrane (or nuclear) marker collected under ECFP (or DsRed2) conditions. The 3rd column contains merged images. For this figure, EGFP was pseudocolored green and ECFP and DsRed2 red.
bath solutions of (in mM) 120 CsCl, 5 NaCl, 2 MgCl 2 , 3 ATP, 0.1 GTP, 5 EGTA, 10 HEPES (pH 7.3), and 160 NaCl, 1 CaCl 2 , 2 MgCl 2 , 10 HEPES (pH 7.4), respectively. Inward currents (cytosol to pipette) are shown as downward deflections. All experiments were acquired using pClamp8.1 software with time and current amplitude data analyzed with this software in conjunction with Igor Pro 4.0 (Wavemetrics Inc., Lake Oswego, OR). Single channel unitary current (i) was determined from the best-fit Gaussian distribution of all-point amplitude histograms. Channel activity (NP o ) was NP o ϭ I/i, where I is the mean total current in the patch and i is unitary current at this voltage (calculated from all-point amplitude histograms). By definition then, current at the closed state is 0. Where appropriate, open probability (P o ) was calculated by normalizing NP o for the total number of estimated channels (N) in the patch as described previously (35).
Expression and Electrophysiology in FRT Epithelia-Function of wild type and mutant ENaC expressed in Fischer rat thyroid (FRT) cells was assayed with standard methods described previously (38,39). In brief, transfected cells (0.07 g of each ␣-, ␤-, and ␥-mENaC) were maintained on permeable filter supports. Na ϩ transport was measured 2-3 days after transfection at 37°C in modified Ussing chambers (Warner Instrument Corp.) with luminal and serosal solutions of (in mM) 135 NaCl, 1.2 CaCl 2 , 1.2 MgCl 2 , 2.4 K 2 HPO 4 , 0.6 KH 2 PO 4 , 10 dextrose, 10 HEPES (pH 7.4; bubbled with O 2 ). Amiloride-sensitive short-circuit current was determined as the difference in current with and without amiloride (10 M) applied to the luminal bath solution.
Western Blot Analyses-Western blot analysis of Myc-and HAtagged ENaC was performed using standard procedures described previously (31,37,40,41). In brief, cells were lysed in gentle lysis buffer, cleared, normalized for total protein concentration, suspended in Lamellia sample buffer and 20 mM DTT, run on 7.5% polyacrylamide gels in the presence of SDS, transferred to 0.45-m nitrocellulose, and probed with antibody in Tris-buffered saline supplemented with 5% dried milk (Nestle, Solon, OH) and 0.1% Tween 20. Primary antibody was applied overnight at 4°C. Blots next were probed with the appropriate horseradish peroxidase-conjugated secondary antibody for 1 h at room temperature. Blots were developed with enhanced chemiluminescent reagents and Kodak BioMax Light-1 film. Digital images of Western blots were generated using a ScanJet4200C (Hewlett-Packard, Houston, TX).
Immunoprecipitation-CHO cells transfected with epitope-tagged ENaC were extracted in gentle lysis buffer. Whole-cell lysates (400 l at 1 g/l total protein) were treated with anti-Myc-agarose or anti-HA antibody plus protein A/G PLUS-agarose overnight at 4°C. Precipitants were then washed 3 times with 400 l of gentle lysis buffer and resuspended in sample buffer. Proteins were analyzed via Western blotting as described above.
Biotinylation-CHO cells were grown and transfected as described above for protein analysis. Biotinylation closely followed that described previously by Heda et al. (42,43). In brief, 48 h after transfection cells were washed 3 times with ice-cold PBS (pH 8.0) and subsequently incubated with 1 mM sulfo-NHS-LC-biotin (in PBS, pH 8.0) for 30 min at 4°C in the dark. Cells were washed 3 times with ice-cold PBS and extracted in gentle lysis buffer. Pre-equilibrated streptavidin-agarose beads were agitated overnight at 4°C with 50 g of total protein. Agarose beads were then washed 6 times with GLB and subsequently resuspended in sample buffer. Proteins were visualized using standard Western blotting described above.
Statistics-Complementation frequency was compared using a z-test on proportions. A p Ͻ 0.05 was considered significant. All patch clamp data are presented as means Ϯ S.E. Paired and unpaired data were compared using appropriate t tests. For multiple comparisons, a oneway analysis of variance with the SNK sub-test was used to establish significances. p Յ 0.05 was considered significant.

RESULTS
A yeast one-hybrid complementation screen was used to identify regions within ENaC involved in protein-protein and/or protein-lipid interactions. Fig. 1 shows results from a typical screen. For this experiment, two hybrids were created: one containing the NH 2 terminus and the other the COOH terminus of ␤ENaC fused to truncated hSOS (catalytic domain) incapable of independently localizing to the plasma membrane. Fig. 1A shows the position of these cytosolic tails relative to the ectodomain and transmembrane domains (TM1 and TM2). Hybrids were overexpressed in Cdc25H yeast. Transfected yeast was maintained on a source plate at the permissive tempera-ture of 24°C (Fig. 1B, top). Colonies were then patched from the source plates to create duplicate arrays with one developed at permissive (Fig. 1B, middle) and the other restrictive (37°C; Fig. 1B, bottom) temperatures. Shown in Fig. 1B are typical results from yeast transfected with the entire NH 2 -(␤ M1-W52 ; left) and COOH-terminal tails (␤ A555-end ; right) of ␤ENaC. These results clearly demonstrate that the COOH-but not NH 2 -terminal tail of ␤ENaC localizes to the plasma membrane in yeast allowing complementation of growth at restrictive temperatures. The COOH-terminal tails of all three subunits and the NH 2 -terminal tail of ␣ENaC complemented growth in this assay (see Fig. 3) (15).
We tested next whether domains identified as membranereactive in the yeast screen yielded comparable results in mammalian cells. Fig. 2 shows representative confocal images of CHO cells transfected with a membrane marker (ECFP-M; Fig.  2, middle column) and one of the six different EGFP-ENaC hybrid proteins containing the cytosolic tails of each subunit (1st column). Fig. 2, 3rd column, shows merged images. For this figure, the NH 2 tail of ␥ENaC was overexpressed with a nuclear marker (DsRed2-N; noted with asterisk). These results clearly show that all three of the COOH termini of ENaC subunits co-localize with the membrane marker. Moreover, the NH 2 terminus of ␣ENaC, but not ␤ and ␥, also overlapped with the membrane marker. These findings in a mammalian expression system are entirely consistent with findings in yeast (see Figs. 1 and 3) (15). Together they strongly argue that domains within the cytosolic tails of ENaC contain residues involved in protein-protein and/or protein-lipid interactions localized to or near the plasma membrane. Most important, the results in Fig.  2 validated the yeast screen. Because the yeast screen measures only membrane-localized interactions, localization is easier to grade compared with confocal images, which can be influenced by background fluorescence resulting, in part, from immature and cytosolic pools of the hybrid. Thus, our yeast screen may be more sensitive than traditional confocal micros- copy with respect to identifying domains involved in proteinprotein interactions localized to the plasma membrane.
We further defined the domains within ENaC tails interacting at the plasma membrane using the one-hybrid screen. Shown in Fig. 3 is a summary graph of complementation frequency (right) for various truncation and deletion mutants of pSOS-ENaC tail hybrids. The portion of ENaC (Fig. 3, gray rectangles) tested in each construct is noted relative to TM1 or TM2 (black boxes). A membrane interactive domain was localized to the first half (Met 1 -Gln 54 ) of the NH 2 terminus of ␣ENaC. Upon dividing this peptide, both halves complemented suggesting that domains within both regions localized to the membrane. Residues included within Met 610 -Val 632 in ␣ENaC, which directly follow TM2, contained a domain strongly interactive at the plasma membrane. Similarly, the region from Ala 568 -Arg 583 in ␥ENaC, which also directly follows TM2, contained a domain strongly interactive at the plasma membrane. Residues Cys 595 -End in ␤ENaC and Thr 584 -Pro 600 in ␥ENaC also contained membrane interactive domains; however, these constructs resulted in punctate growth suggesting a weak interaction that is approaching the sensitivity threshold of the screen.
The functional relevance of the membrane interactive domains in the COOH tail of ␥ENaC was determined by using a reconstituted mammalian expression system. Fig. 4A shows a representative family of currents from CHO cells transfected with (Myc-tagged) ␣-, ␤-, and ␥-mENaC elicited by test pulses (20-mV increments) from Ϫ120 to 100 mV in the absence (left) and presence (right) of 10 M amiloride. Fig. 4B shows a family of currents elicited by the same voltage steps in untransfected CHO cells in the absence and presence of amiloride. Fig. 4C is a typical whole-cell I-V relation of a CHO cell transfected with ␣-, ␤-, ␥-mENaC in response to voltage ramps. Short and long dashed lines note currents in the absence and presence of amiloride, respectively, with their difference being the solid line. The graph of Fig. 4D summarizes the effects of amiloride on inward currents at Ϫ80 mV in CHO cells transfected with ␣-, ␤-, and ␥-mENaC (n ϭ 10).
Typical families of currents elicited by voltages steps in CHO cells transfected with (Myc-tagged) ␣-, ␤-, and ␥ ⌬Q573-P600 -mENaC in the absence (left) and presence (right) of amiloride are shown in Fig. 5A. (Note the scale difference with Fig. 4A.) Fig. 5B compares representative I-V relations for wt (solid line) and mt (dashed line) channels in response to voltage ramps. Fig. 5C summarizes a population study of the amiloride-sensitive, macroscopic, inward Na ϩ current at Ϫ80 mV for wt and mt channels with mt channels having significantly less current than wt with 0.33 Ϯ 0.14 compared with 2.5 Ϯ 0.80 nA (n Ն 7). This was a single-blinded population study of four distinct parallel transfections where the end user did not know which was the experimental group. The inset in Fig. 5C shows that mt ␥ENaC expressed equally as well as wt in CHO cells. Moreover, no overt difference was observed in the frequency of seals that had amiloride-sensitive currents in the wt versus mt groups (f ϳ 0.6).
To determine whether the mutant ␥ENaC subunit lacking Gln 573 -Pro 600 was capable of appropriately interacting with ␣and ␤ENaC subunits, we overexpressed HA-tagged ␣and ␤ENaC with Myc-tagged ␥ENaC in CHO cells and performed co-immunoprecipitation studies. Both wt and mt ␥ENaC were tested in these experiments. Shown in Fig. 6 are typical Western blots containing the anti-HA precipitant (top) and wholecell lysate (bottom) from CHO cells transfected with wt HA-␣, ␤ENaC in addition to Myc-tagged wt (middle lane) and mt ⌬Gln 573 -Pro 600 (1st lane) ␥ENaC. The last lane contains precipitant and lysate from cells transfected with Myc-tagged wt ␥ENaC alone. Both blots probed with anti-Myc antibody. These studies clearly showed that mt ␥ENaC is as effective as wt in interacting with ␣and ␤ENaC subunits We asked next whether mt ␥ENaC was capable of getting to the plasma membrane. Shown in Fig. 7A is a typical Western blot (cut in half) probed with anti-Myc (top) and anti-Fra-2 (bottom) antibodies. These blots contain the pellet and supernatant from streptavidin precipitations of CHO cells overexpressing EGFP or all three Myc-tagged ENaC subunits (noted by gray boxes). After transfection but before lysing, one sample from the ENaC transfection group was briefly treated with sulfo-NHS-LC-biotin at 4°C to label membrane proteins. These results, consistent with the functional studies in Figs. 4, 5, and 9 -11, clearly show that exogenous ENaC is in the plasma membrane. The cytosolic protein Fra-2 localized primarily to the supernatant in streptavidin-precipitated biotinylated preparations demonstrating good separation of membrane-labeled protein from cytosolic protein. Fig. 7B tested whether the Gln 573 -Pro 600 deletion mutant like that of wild type ␥ENaC localized to the plasma membrane. For this representative experiment, Myc-tagged wt and mt subunits were overexpressed with HA-tagged ␣and ␤ENaC. Cells were washed (right) or treated with sulfo-NHS-LC-biotin to label membrane proteins (left), followed by lysing and streptavidin precipitation. Precipitated proteins are shown in the top blot. The lower blot contains whole-cell lysate from each group. Blots were probed with anti-Myc antibody. These results show that channels containing the mutant ⌬Gln 573 -Pro 600 subunit similar to the wild type ␥ENaC subunit localize to the plasma membrane. These data in conjunction with those in Figs. 5 and 6 showing that the ⌬Gln 573 -Pro 600 mt interacts normally with ␣and ␤ENaC and expresses at levels comparable with wt ␥ENaC but that channels containing the mt subunit have less activity suggest, then, that the ⌬Gln 573 -Pro 600 mutation most likely affects ENaC gating kinetics and not localization to the plasma membrane.
Results in Fig. 8, in complement to those in Fig. 7, show that wild type channels and those containing the ⌬Gln 573 -Pro 600 mutant both reside in the plasma membrane. Fig. 8A shows EF fluorescence (bottom row) and wide-field epifluorescence (top FIG. 5. Channels containing the ⌬Gln 573 -Pro 600 ␥ENaC deletion mutant have less activity. A, typical macroscopic currents in the absence (left) and presence of 10 M amiloride (right) elicited by test pulses from Ϫ120 to 100 mV from a whole-cell patch made on a CHO cell expressing ␣-, ␤-, and ␥ ⌬Q573-P600 ENaC. B, representative amiloride-sensitive I-V relations in response to voltage ramps from 100 to Ϫ100 mV from CHO cells expressing wt (solid line) and mt (dashed line) ENaC. C, summary comparing the inward, amiloride-sensitive macroscopic Na ϩ current at Ϫ80 mV in CHO cells transfected with wt and mt ENaC. This was a single-blinded population study. *, p Ͻ 0.05 versus wt. Inset shows a Western blot that contains whole-cell lysate from CHO cells transfected with Myc-tagged wt and mt ␥ENaC probed with anti-Myc antibody. row) images of CHO cells overexpressing cytosolic DsRed2 (middle column; pseudocolored red) and membrane-localized EGFP-F (1st column; pseudocolored green). Merged images are shown in the last column. These results clearly show, as expected with this technique, that only fluorophores, such as EGFP-F but not DsRed2, localized to or within the vicinity of the plasma membrane (ϳ100 nm) were excited with EF microscopy. For these experiments all parameters, including gain and exposure times, were held constant in both the wide-field and the EF microscopy groups. Upon greatly increasing the gain and exposure time, we could observe a very modest DsRed2 signal with EF microscopy (note shown) most likely resulting from slight penetration into the perimembrane cytosolic region that contained a small percentage of the DsRed2 signal. Fig. 8B shows EF fluorescence (2nd column; pseudocolored green) and wide-field epifluorescence (1st column pseudocolored red) images of CHO cells transfected with HA-␣,␤-mENaC in addition to either Myc-tagged wild type ␥ENaC (top) or the ⌬Gln 573 -Pro 600 mutant (bottom). The last column shows merged images. These cells were permeabilized and exposed to anti-Myc antibody and a rhodamine-conjugated 2 o antibody. These results clearly show that channels containing wt and mt ␥ENaC subunits are both similarly localized to within ϳ100 nm of the cover glass-cell interface most likely in the plasma membrane.
Results in Fig. 9 characterize at the single channel level wt ENaC reconstituted in CHO cells. The current traces in Fig. 9A were from an outside-out patch with CsCl in the pipette and NaCl in the bath. The applied command potential is noted for each trace. Inward current is denoted by downward deflections, and the arrows note the closed state. This patch contained at least six channels. Fig. 9B shows the I-V relation and corresponding channel openings (for voltages ranging from Ϫ40 to Ϫ120 mV) for this type of channel, which was not observed in untransfected cells. This channel had a single channel cord conductance (Ϫ40 to Ϫ120 mV) of 5.8 pS, was Na ϩ -selective, and not voltage-gated. Shown in Fig. 9C are all-point histograms for the current trace in Fig. 9A held at Ϫ60 mV. This Na ϩ -selective channel had all the hallmarks of ENaC described in other expression systems and in native epithelia (44 -47). Fig. 10 tested whether this Na ϩ -selective, 5.8-pS channel was sensitive to amiloride, which is a well described specific inhibitor of ENaC (44 -47). Shown in Fig. 10A are single channel currents from an outside-out patch (no applied potential) made on ␣-, ␤-, and ␥-mENaC transfected cells containing several of these 5.8-pS Na ϩ channels before and after 2 M amiloride treatment. The currents under the three gray lines are shown at a faster time scale in the insets with I being before amiloride treatment and II and III after treatment. Shown in   10B is a macro-patch current from an outside-out patch made on an ␣-, ␤-, and ␥-mENaC transfected cell that contained many channels. This patch was held at 0 mV, and pipette and bath solutions were CsCl and NaCl. Inward current is down with arrows denoting the closed state. Because this macro-patch contained many channels, it was impossible to isolate single transitions; however, upon addition of 2 M amiloride the channels quickly closed entering a flickery blocked state with macro-patch current tending toward 0 pA. Considering the I-V relation in Fig. 8B (see also Fig. 11A), the unitary current at 0 mV for channels within this patch is ϳ0.11 pA. Thus, this patch is estimated to contain at least 40 channels. Indeed, variance analysis suggested that the macro-patch current trace in Fig. 10B was from a patch containing 57 channels with a P o of 0.41. It was not unusual to find this many ENaC in a single patch made from cells expressing wt channels. These results, as well as those in Fig. 5A, show that CHO cells transfected with ␣-, ␤-, and ␥-mENaC contain macroscopic and single channel currents sensitive to amiloride, a finding consistent with them expressing functional heterotrimeric ENaC.
The experiments in Figs. 11 and 12 tested whether channels containing the mt ⌬Gln 573 -Pro 600 ␥ENaC subunit had a decrease in P o due to changes in gating. Shown in Fig. 11A are typical current traces from CHO cells expressing wt and mt channels. The top trace is from an outside-out patch made from a CHO cell expressing wt channels held 0 mV. The lower three traces are from three different CHO cells expressing mt channels. These patches were held at 0, Ϫ20, and Ϫ60 mV. The unitary current calculated from the all-point histograms in Fig.  11B for each of these channels is reported to the right. In addition, the P o calculated from the area under the histograms in Fig. 11B is also reported with wt having a P o of 0.57 and mt channels of 0.20, 0.07 and 0.17.
The summary graph in Fig. 12A compares P o for each channel with wt and mt channels having 0.67 Ϯ 0.07 and 0.16 Ϯ 0.03, respectively (n Ն 8). As mentioned above, patches made from CHO cells expressing wt channels most often contained several channels. Thus, P o for wt often had to be estimated from patches containing several channels when it was possible to reliably estimate n from all-point histograms. (Only 3 of the 11 patches containing wt channels contained one channel; however, only patches containing 5 channels or fewer were used to calculate P o .) This may have led us to overestimate P o in the wt group. Nevertheless, these results in conjunction with those in Figs The functional domain following TM2 in ␥ENaC was further defined by assessing the activity of channels containing ⌬Gln 573 -Arg 583 and ⌬Thr 584 -Pro 600 mutants. Fig. 13 shows representative currents elicited by voltage ramps (40 to Ϫ100 mV) from CHO cells expressing wild type (A) and ⌬Gln 573 -Arg 583 mutant (B) channels in the absence and presence of amiloride. As summarized in Fig. 13C, the amiloride-sensitive inward current at Ϫ80 mV of 0.16 Ϯ 0.04 nA (n ϭ 19) for channels containing the ⌬Gln 573 -Arg 583 mutant was significantly less than the 2.3 Ϯ 0.55 nA (n ϭ 9) for wild type channels. In contrast to the large effect of deleting Gln 573 -Arg 583 , deleting Thr 584 -Pro 600 had a very modest effect on Na ϩ transport as measured in FRT cells. The relative transport of 1.0 Ϯ 0.06 versus 0.86 Ϯ 0.04 (p ϭ 0.06, n ϭ 18; not shown) for wt and mt channels, respectively, was not significantly different. These results are consistent with the idea that the 11 amino acids between Gln 573 -Arg 583 contain the most significant residues with respect to support of ENaC activity. DISCUSSION The current study used a novel yeast one-hybrid screen to identify regions in the cytosolic tails of ENaC that localized to the plasma membrane complementing an inherent, temperature-sensitive growth defect in the Cdc25H yeast strain. Confocal images of CHO cells transfected with EGFP hybrids made from these cytosolic portions of ENaC and a membrane marker (ECFP-M) validated findings in yeast showing that the NH 2terminal tail of ␣ENaC, but not ␤and ␥ENaC, and the COOHterminal tails of all three subunits co-localized with the membrane marker. Together, these findings support the idea that these regions of ENaC are involved in protein-protein and/or protein-lipid interactions localized to or near the plasma membrane. The yeast screen was used to further localize membrane interacting domains within ENaC subunits. The functional significance of one membrane interacting region, residing within Gln 573 -Pro 600 of ␥ENaC, was accessed in the current study using a reconstituted mammalian expression system where ENaC was overexpressed in CHO cells and studied using patch clamp electrophysiology. This region has not been identified previously as playing a role in ENaC activity. We found that deletion of Gln 573 -Pro 600 within the cytoplasmic COOH tail of ␥ENaC decreased activity by affecting channel gating but not channel localization or oligomerization. The current results support the idea that the Gln 573 -Pro 600 region of ␥ENaC is not involved in channel oligomerization or targeting channels to the membrane but interacts with an as yet undefined membrane resident factor to support normal channel gating and activity. An alternative interpretation that cannot be excluded with the current data set is that, although the ⌬Gln 573 -Pro 600 mt ␥ENaC subunit interacted normally with ␣ and ␤ subunits and was localized to the plasma membrane, the mutant channel is inactive either because it is non-functional or in a prolonged quiescent state. The resulting macroscopic currents and single channel currents measured in mt transfectants then would have resulted from ␣␤ heterodimeric or ␣ homomeric channels. Nonetheless, this alternative also is consistent with the idea that the Gln 573 -Pro 600 region in ␥ENaC interacts with a membrane factor to facilitate formation of normal functional heterotrimeric channels.
Further deletion mutagenesis localized the active region in the COOH terminus of ␥ENaC to reside within residues Gln 573 -Arg 583 . The 11 amino acids between Gln 573 -Arg 583 contain several conserved regions: two tracts, KAK and RRR, containing positively charged residues and an absolutely conserved tryptophan repeat, WW. In consideration of the recent findings (48,49) showing that ENaC activity is modulated by anionic phospholipids, which are well known to interact with positively charged residues, it is provocative to speculate that the conserved charged residues within this region of ENaC are somehow involved in this regulation.
Interestingly, the general region just distal to TM2 in ␣ENaC, including residues Met 610 -Val 632 , identified as local- FIG. 9. Single channel characterization of mENaC reconstituted in CHO cells. A, single channel current traces from outside-out patches containing several small conductance, Na ϩ -selective ion channels made from CHO cells transfected with ␣-, ␤-, and ␥ENaC. Bath and pipette were NaCl and CsCl, respectively. Patches were held at pipette potentials of 10, Ϫ40, and Ϫ60 mV. Inward current is downward, and arrows denote closed states. B, single channel I-V relation for ENaC overexpressed in CHO cells. Bath and pipette solutions are as above. Sample channel openings for Ϫ40 through Ϫ120 mV are shown to the left of the I-V. The chord conductance (from Ϫ40 to Ϫ120 mV) for ENaC was 5.8 pS. C, all-point histograms for the current traces in 8A at Ϫ60 mV. This patch had at least 6 active channels. The unitary current i at this voltage was ϳ0.36 pA.
izing to the membrane in the current study has been identified previously by Volk et al. (19,50) as playing an essential role in kinase regulation of the channel. This group showed that ki-nase regulation associated with this region was important for proper localization of ENaC to the plasma membrane. Specifically, these authors report that Pro 595 of ␣-hENaC (correspond- FIG. 10. Reconstituted ENaC is sensitive to amiloride. A, single channel currents from an outside-out patch made from a CHO cell expressing ␣-, ␤-, and ␥-mENaC. Bath and pipette solutions are as above. Holding potential was 0 mV. Inward current is down, and arrows denote closed state. Vertical arrow notes application of 2 M amiloride. Gray lines I, II, and III define regions before and after amiloride shown below at an expanded time scale. B, current in an outside out macro-patch. Solutions are as above, holding potential is 0 mV, and arrow denotes closed state. The vertical arrow notes application of 2 M amiloride. The ϳ5-s time to inactivation is consistent with the dead-space and necessary time for bath exchange in our perfusion system. Note that after amiloride, flickery single channel events can be observed. This macro-patch likely contained at least 55 channels (refer to "Results").
FIG. 11. Single channel characterization of channels containing the ⌬Gln 573 -Pro 600 deletion mutant. A, single channel current traces from outside-out patches created on CHO cells expressing wt (top) and mt (lower three) channels. Conditions are as stated above. The traces for mt channels are from three distinct experiments with patches held at 0, Ϫ20, and Ϫ60 mV. B, all-point histograms for these current traces. Open probability, calculated as the area under the best-fit line to these Gaussian distributions, is indicated above each graph.
ing to Pro 622 in ␣-mENaC) was necessary for normal channel activity and suppression of activity by the broad-spectrum kinase inhibitor staurosporine. Interpretations of these earlier data as presented by the authors included that Pro 595 is involved in an interaction between ENaC and a staurosporinesensitive kinase, a phosphoprotein, and alternatively, a linker protein that then binds a kinase or phosphoprotein. It is our contention that these earlier findings lend further support to the functional relevance of domains identified with the yeast one-hybrid screen in the present study. Notably, this region of ␣ENaC, in addition, is well conserved across species and also within the ␦ENaC subunit (51).
There are currently no reports that we are aware of specifically suggesting that the first 54 residues in the NH 2 terminus of ␣ENaC identified in the current study as membrane interactive are important for positive regulation of ENaC activity. However, Chalfant et al. (16) have identified a possible endocytic tag within this region corresponding to KGDK 50 that may play a role in down-regulation of ENaC activity. Moreover, we find the recent reports (28) provocative, showing that the NH 2 terminus of ␣-hENaC is alternatively spliced and include alternative first exons to form unique ␣ENaC subunits. Whereas in oocytes the alternative ␣ENaC subunits do not show functional differences, it remains to be determined whether there is a tissue-specific activity difference between these apparently interchangeable subunits or whether these subunits respond in unique manners to different regulatory pathways. It is not clear at this time if there are similar NH 2 -terminal variant ␣ENaC subunits in other species arising from differential splicing and inclusion of alternative exons; however, the early but not latter portions of the NH 2 -terminal tails of ␣ENaC in general are heterogeneous across species. It is exciting to speculate that this apparently hyper-flexible region may possibly contribute to the noted species-and tissue-specific differences in channel activity.
Awayda et al. (12) demonstrated previously that injection of a 30-amino acid peptide with identity to the distal portion of the COOH terminus of the ␤ subunit into Xenopus laevis oocytes expressing ENaC quickly decreased activity. From this and other results, it was concluded in this earlier study that this region of ENaC exerts tonic inhibition of channel activity with the implication being that this region of ENaC interacts with other regions of the channel to negatively regulate P o . An alternative is that the exogenous peptide blocked activity by competing for a positive regulatory protein. Nonetheless, these previous findings provide support for the idea that the most distal portion of ␤ENaC plays a functional role in control of channel activity. We demonstrate in the current study that the last 17 residues in ␤ENaC are membrane-interactive, which is consistent with the possibility that this region plays some role in development of modulatory protein-protein and/or proteinlipid interactions. Consistent with this idea, Dinudom et al. (52) demonstrated that a peptide identical to the distal most 10 residues in ␤ENaC inhibited ENaC activity in intralobular duct cells most likely by competing for a required positive regulatory protein.
We interpret the current results collectively as showing that specific residues/microdomains, of which Gln 573 -Arg 583 in ␥ENaC had not been identified previously, play important roles in control of ENaC activity. The current results demonstrate that residues within Gln 573 -Arg 583 interact with a membrane resident factor to support ENaC openings. We expect through continued use of the one-hybrid screen and reconstituted expression system described in the current study to gain further insight about the molecular mechanisms and channel microdomains involved in ENaC regulation.