Interactions between Subunits of the Human Epithelial Sodium Channel*

The human epithelial sodium channel (hENaC) mediates Na+ transport across the apical membrane of epithelia, and mutations in hENaC result in hypertensive and salt-wasting diseases. In heterologous expression systems, maximal hENaC function requires co-expression of three homologous proteins, the α, β, and γhENaC subunits, suggesting that hENaC subunits interact to form a multimeric channel complex. Using a co-immunoprecipitation assay, we found that hENaC subunits associated tightly to form homo- and heteromeric complexes and that the association between subunits occurred early in channel biosynthesis. Deletion analysis of γhENaC revealed that the N terminus was sufficient but not necessary for co-precipitation of αhENaC, and that both the N terminus and the second transmembrane segment (M2) were required for γ subunit function. The biochemical studies were supported by functional studies. Co-expression of γ subunits lacking M2 with full-length hENaC subunits revealed an inhibitory effect on hENaC channel function that appeared to be mediated by the cytoplasmic N terminus of γ, and was consistent with the assembly of nonfunctional subunits into the channel complex. We conclude that the N terminus of γhENaC is involved in channel assembly.

The human epithelial sodium channel (hENaC) mediates Na ؉ transport across the apical membrane of epithelia, and mutations in hENaC result in hypertensive and salt-wasting diseases. In heterologous expression systems, maximal hENaC function requires co-expression of three homologous proteins, the ␣, ␤, and ␥hENaC subunits, suggesting that hENaC subunits interact to form a multimeric channel complex. Using a co-immunoprecipitation assay, we found that hENaC subunits associated tightly to form homo-and heteromeric complexes and that the association between subunits occurred early in channel biosynthesis. Deletion analysis of ␥hENaC revealed that the N terminus was sufficient but not necessary for co-precipitation of ␣hENaC, and that both the N terminus and the second transmembrane segment (M2) were required for ␥ subunit function. The biochemical studies were supported by functional studies. Co-expression of ␥ subunits lacking M2 with full-length hENaC subunits revealed an inhibitory effect on hENaC channel function that appeared to be mediated by the cytoplasmic N terminus of ␥, and was consistent with the assembly of nonfunctional subunits into the channel complex. We conclude that the N terminus of ␥hENaC is involved in channel assembly.
Functional studies indicate that epithelial Na ϩ channels are composed of at least three homologous proteins, the ␣, ␤, and ␥ENaC subunits (20 -24). Amino acid sequence analysis of the subunits and biochemical studies of rat ␣ENaC (25)(26)(27) suggest that the subunits possess cytoplasmic N and C termini, two transmembrane domains (M1 and M2), and a large glycosylated, cysteine-rich extracellular domain. Since all three subunits are required for maximal hENaC function, it has been hypothesized that the subunits interact to form a multimeric channel complex. In addition, genetic studies of the degenerins suggest that mechanosensory function and dominant neurodegeneration require three gene products (14 -16). The goal of this study was to answer some fundamental questions about the structure of hENaC. First, do the hENaC subunits interact with each other and with themselves? Second, which parts of an hENaC subunit are necessary for channel function and which are responsible for interactions with other subunits? EXPERIMENTAL PROCEDURES DNA Constructs-The cDNAs encoding secreted alkaline phosphatase, and ␣, ␤, ␥, ␣ S594X , ␤ R566X , ␥ K576X hENaC (all in pMT3) are described elsewhere (4,22,23,28). Epitopes were introduced into fulllength hENaC subunits, immediately upstream of the stop codon, using the Muta-Gene phagemid in vitro mutagenesis kit (Bio-Rad). The FLAG epitope (DYKDDDDK) was inserted into ␣hENaC (␣ FLAG ), ␤hENaC (␤ FLAG ), and ␥hENaC (␥ FLAG ), and a hemagglutinin (HA) epitope (YPY-DVPDYA) was inserted into ␤hENaC (␤ HA ) and ␥hENaC (␥ HA ). In ␣ FLAG , the FLAG epitope replaced the last 8 C-terminal residues of ␣hENaC. In the other tagged subunits, the epitope was inserted immediately after the most C-terminal residue. Epitope-tagged subunits were cloned into pMT3 for expression. ␥ E518X and ␥ D268X were constructed by single-stranded mutagenesis of ␥hENaC in pcDNA3 (Invitrogen, San Diego, CA). ␥ ⌬3-53 , ␥ L54X , and ␥ I102X were amplified by polymerase chain reaction using ␥hENaC as a template. In ␥ ⌬3-53 , the FLAG epitope was inserted after the most C-terminal residue. ␥ L54X and ␥ I102X were cloned into pcDNA3 and ␥ ⌬3-53 into pMT3 for expression.
Antibodies, Immunoprecipitations, and Western Blots-The anti-HA monoclonal antibody, 12CA5, was obtained from Boehringer Mannheim, and the anti-FLAG monoclonal antibody, M2, was from Eastman Kodak Co. The anti-N␣ and anti-N␥ polyclonal antisera were raised against peptides corresponding to the N-terminal 20 amino acids of ␣ and ␥hENaC, respectively. COS-7 cells were transfected by electroporation and grown in medium containing 10 M amiloride. Two to three days post-transfection, cells were lysed at 4°C in lysis buffer (150 mM NaCl, 50 mM Tris, pH 7.4, 0.4 mM phenylmethylsulfonyl fluoride, 20 g/ml aprotinin, 20 g/ml leupeptin, 10 g/ml pepstatin A) with detergent. Metabolic labeling with [ 35 S]methionine (Amersham Corp.) was performed prior to cell lysis, as described previously (29). To determine the most appropriate method for solubilization of hENaC subunits, we tested lysis buffer with several different detergents. 1% digitonin, 1% Nonidet P-40, and radioimmunoprecipitation buffer (1% Triton X-100, 0.1% SDS, and 1% sodium deoxycholate) each effectively solubilized ␣ FLAG , whereas 1% CHAPS and 2.5% octyl glucoside were both relatively ineffective (Fig.  1A). 1% Triton X-100, without SDS or deoxycholate, also effectively solubilized ␣ FLAG . Lysates were cleared by centrifugation at 70,000 rpm at 4°C for 30 min, then incubated with primary antibody overnight at 4°C. For immunoprecipitations, we used 5 g/ml anti-FLAG antibody, 2 g/ml anti-HA antibody, or a 1:200 dilution of anti-N␥ rabbit antisera. Antigen-antibody complexes were precipitated with immobilized protein A (Pierce), and precipitates were washed twice in lysis buffer with detergent, then once in lysis buffer alone. For high salt washes, precipitates were first washed three times in lysis buffer containing 500 mM NaCl and detergent. Following washes, precipitates were boiled for 5 min in sample buffer (4% SDS, 65 mM Tris, pH 6.8, 100 mM dithiothreitol, 20% glycerol, and 0.005% bromphenol blue). Proteins were separated on 8% polyacrylamide gels using SDS-polyacrylamide gel electrophoresis. Radioactive proteins were detected by autoradiography. Western blots were blocked with 5% bovine serum albumin, and incubated first with primary antibody (2 g/ml anti-FLAG or a 1:1000 dilution of anti-N␣), then with a horseradish peroxidase-coupled secondary antibody (Amersham Corp.) at a 1:10,000 dilution. Proteins were detected by enhanced chemiluminescence (Pierce).
Expression and Electrophysiological Analysis in Xenopus Oocytes-hENaC subunits were expressed in albino Xenopus laevis oocytes (Nasco, Fort Atkinson, WI) by nuclear injection of plasmid DNA. DNA was injected at concentrations ranging from 40 to 70 ng/l. Following injection, oocytes were incubated at 18°C in modified Barth's solution or low Na ϩ modified Barth's solution (23), then studied 1 day later. Whole-cell currents were measured using the two-electrode voltageclamp technique, as we have previously described (4). During recording, oocytes were bathed in frog Ringer's solution (116 mM NaCl, 2 mM KCl, 0.4 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, pH 7.4). A maximal dose of amiloride (100 M) was added directly to the bath, and amiloridesensitive current was determined by subtracting current following amiloride application from the baseline current. Oocyte membrane proteins were isolated according to the protocol of Tucker et al. (30), then detected by Western blot as described above.
Epitope-tagged Subunits-To determine if the epitope-tagged subunits retain structural and functional integrity, we analyzed them in COS-7 cells and Xenopus oocytes. Like wild-type subunits, epitope-tagged subunits expressed in COS-7 cells were glycosylated. Following treatment of immunoprecipitated ␣ FLAG , ␤ FLAG , or ␥ FLAG with protein N-glycanase, the high molecular weight glycosylated form of each subunit was reduced to the low molecular weight unglycosylated form (Fig.  1B). Fig. 1C shows that expression of epitope-tagged subunits in Xenopus oocytes resulted in amiloride-sensitive Na ϩ currents that were indistinguishable from wild-type hENaC currents, suggesting that insertion of C-terminal epitope tags did not grossly disrupt hENaC structure or cell surface targeting. The epitope-tagged subunits were also functional in mammalian (Fischer rat thyroid cell) epithelia (not shown).

Co-immunoprecipitation of Full-length hENaC Subunits-
To test the hypothesis that hENaC subunits form heteromeric complexes, we examined the interaction of hENaC subunits in transfected COS-7 cells. In these experiments, we immunoprecipitated one subunit and detected the other subunit by Western blot; this strategy allowed us to distinguish subunits despite their similar molecular weights. With the co-precipitation assay, we detected all potential heteromeric combinations of two hENaC subunits: ␣-␤, ␣-␥, and ␤-␥ (Fig. 2). The presence of a third hENaC subunit was not required for the association of any two hENaC subunits, indicating that each subunit was able to interact with either of the other subunits. In each case, both unglycosylated and glycosylated forms of one hENaC subunit immunoprecipitated with the other subunit, suggesting that the interaction between subunits may have occurred prior to glycosylation, in the endoplasmic reticulum. When lysate from cells expressing one subunit was mixed with lysate from cells expressing another subunit, interactions were not detected. Under identical immmunoprecipitation conditions, we could not detect an interaction between ␣hENaC and the cystic fibrosis transmembrane conduct-FIG. 1. Epitope-tagged hENaC subunits. A, solubilization of hENaC by several detergents. COS-7 cells were transfected with ␣ FLAG , ␤, and ␥. Cells were lysed with various detergents, and ␣ FLAG was immunoprecipitated and detected by Western blot with anti-FLAG antibody. B, glycosylation of epitope-tagged subunits. COS-7 cells were transfected with ␣ FLAG , ␤ FLAG , or ␥ FLAG . Cells were lysed with 1% digitonin, and ␣ FLAG , ␤ FLAG , or ␥ FLAG was immunoprecipitated with anti-FLAG antibody. Immunoprecipitated protein was incubated at 37°C overnight with or without protein N-glycanase (PNGase), and FLAG-tagged subunits were detected by Western blot with anti-FLAG antibody. C, function of epitope-tagged subunits. Xenopus oocytes were injected with the indicated DNA constructs, and amiloride (100 M)sensitive current (I amiloride ) was measured with a two-electrode voltageclamp. In this experiment, epitope-tagged subunits were expressed in combinations that were used for co-immunoprecipitation studies. Individual epitope-tagged subunits were also fully functional when expressed with two wild-type subunits in Xenopus oocytes (not shown). Each group represents an average current (ϮS.E.) from at least four oocytes.
FIG. 2. hENaC subunits form heteromeric complexes. COS-7 cells were transfected with the indicated constructs, and lysed with 1% digitonin. HA-tagged subunits were immunoprecipitated with anti-HA, immunoprecipitates were washed with lysis buffer containing isotonic salt (150 mM NaCl) and 1% digitonin. FLAG-tagged subunits were detected with anti-FLAG antibody. For mixing experiments, we mixed lysate from cells expressing one subunit with an equal volume of lysate from cells expressing a different subunit. Mixing was done prior to immunoprecipitation.
ance regulator (not shown), suggesting that hENaC subunits did not interact nonspecifically with other transmembrane proteins. To test the strength of the heteromeric interactions, we varied the immunoprecipitation conditions; each interaction withstood all the conditions tested, including a nonionic detergent (Triton X-100) with 500 mM salt (Fig. 3, A-C).
In Xenopus oocytes, expression of ␣hENaC alone generates current, suggesting that ␣ may form a homomultimer. Because expression of ␤ and ␥ alone fails to generate current, it is possible that those subunits do not form homomultimers. To test these possibilities we employed two different strategies. First, we tested the ability of full-length subunits to co-precipitate smaller subunits that lacked the cytoplasmic C terminus. These truncated subunits (␣ S594X and ␤ R566X ) are functional and are similar to mutant ␤ and ␥ subunits found in patients with Liddle's syndrome (1)(2)(3)(4). Deletion of the C terminus speeds migration on an SDS-polyacrylamide gel electrophoresis gel, allowing us to distinguish metabolically labeled ␣ S594X and ␤ R566X from full-length protein. Fig. 4A shows that ␣ FLAG co-precipitated ␣ S594X and ␤ FLAG co-precipitated ␤ R566X , indicating that both ␣ and ␤ subunits could form homomultimers. Fig. 4 also shows that ␣ FLAG co-precipitated ␤ R566X , and ␤ FLAG co-precipitated ␣ S594X , consistent with the data shown in Figs. 2 and 3. Our second strategy was to immunoprecipitate one full-length subunit with the anti-HA antibody, and then detect co-precipitating subunits with the anti-FLAG antibody. As shown in Fig. 4B, ␤ FLAG co-precipitated with ␤ HA and ␥ FLAG co-precipitated with ␥ HA . These results indicated that each of the subunits can associate to form homomultimers. Therefore, the inability of ␤ and ␥ to generate current when expressed alone is not likely the result of an inability to associate.
Biochemical Interactions between ␣hENaC and Truncated ␥ Subunits-To define the site(s) of interaction between hENaC subunits, we deleted portions of one subunit, ␥hENaC, then tested those truncated ␥ subunits for their ability to associate with full-length ␣hENaC. First, we made progressive truncations from the C terminus of ␥ (Fig. 5A). Deletion of the C terminus, M2, the large extracellular domain, and M1 of ␥ did not abolish its ability to interact with ␣ (Fig. 5, B and C). A construct that consisted of only the cytoplasmic N terminus (␥ L54X ) was able to precipitate ␣ (Fig. 5B). Thus, the N terminus of ␥ was sufficient for an interaction with ␣. Although constructs that contained M1 (␥ K576X , ␥ E518X , ␥ D268X , and ␥ I102X ) interacted with glycosylated ␣, we detected an interaction only between ␥ L54X and unglycosylated ␣. To determine if the N terminus was necessary for an association with ␣, we tested ␥ ⌬3-53 , a ␥ construct that lacked only the N terminus (Fig. 5A). As shown in Fig. 5D, ␥ ⌬3-53 immunoprecipitated ␣hENaC. This result indicated that the N terminus of ␥ was not required for association with ␣ and that other sites of intersubunit association likely exist.
Function of Truncated ␥ Subunits in Xenopus Oocytes-Since some pseudohypoaldosteronism type 1-associated mutations are predicted to cause truncation of hENaC subunits (6, 7), we tested the ability of truncated ␥ subunits to contribute to a functional Na ϩ channel complex. In Xenopus oocytes, expression of ␣hENaC alone produces a small amiloride-sensitive Na ϩ current (22). We previously reported that co-expression of ␥ with ␣ produced larger currents, and maximal currents were obtained when all three subunits were co-expressed (23). Moreover, in the absence of ␣hENaC, ␤ and ␥ generated no amiloride-sensitive current. Fig. 6 shows that coexpression of ␥hENaC with the ␣ subunit increased current. However, of the truncated ␥ subunits, only ␥ K576X augmented ␣ subunit-dependent amiloride-sensitive currents. ␥ K576X , which lacked most of the C terminus and was similar to mutant subunits found in patients with Liddle's syndrome, increased the current compared with wild-type ␥. This increase in current was FIG. 3. Heteromeric interactions withstand high salt and nonionic detergents. Transfected COS-7 cells were lysed in solubilization buffer containing 1% digitonin (DG) or 1% Triton X-100 (TX). Immunoprecipitates were washed with lysis buffer containing isotonic salt (150 mM NaCl) or high salt (500 mM NaCl), as indicated. A, cells were transfected with ␣ FLAG and ␥. ␥ was immunoprecipitated with anti-N␥ antibody, and ␣ FLAG was detected by Western blot using anti-FLAG antibody. B, cells were transfected with ␣ FLAG and ␤ HA . ␤ HA was immunoprecipitated with anti-HA antibody, and ␣ FLAG was detected by Western blot using anti-FLAG antibody. C, cells were transfected with ␤ FLAG and ␥. ␥ was immunoprecipitated with anti-N␥ antibody, and ␤ FLAG was detected by Western blot using anti-FLAG antibody.

FIG. 4. hENaC subunits form homomultimers. A, COS-7 cells
were transfected with the indicated constructs. Proteins were metabolically labeled with [ 35 S]methionine, then lysed in 1% Triton X-100. ␣ FLAG or ␤ FLAG was immunoprecipitated with anti-FLAG antibody, and immunoprecipitates were washed under high salt conditions. Proteins were visualized with autoradiography. Thin arrows indicate glycosylated and unglycosylated ␣ FLAG and ␤ FLAG . Bold arrows indicate ␣ S594X and ␤ R566X . Only the more prominent, glycosylated form of ␤ R566X was detected. B, COS-7 cells were transfected with the indicated constructs, then lysed in 1% Triton X-100. ␤ HA or ␥ HA was immunoprecipitated with anti-HA antibody, immunoprecipitates were washed under high salt conditions, and ␤ FLAG and ␥ FLAG were detected by Western blot using anti-FLAG antibody. In this experiment, unglycosylated ␥ FLAG was not observed, and ␤ HA immunoprecipitated a protein that migrated between glycosylated and unglycosylated ␤ FLAG . This protein likely represents an intermediate glycosylated form of ␤ FLAG (L. Prince, P. Snyder and M. J. Welsh, manuscript in preparation).
probably due to increased cell-surface expression of ␣␥ K576X channels, since ␥ K576X lacked a C-terminal sequence thought to mediate endocytosis of hENaC (4,31). Truncated ␥ subunits that lacked the N terminus (␥ ⌬3-53 ) or M2 (␥ E518X , ␥ I102X , and ␥ L54X ) were unable to substitute for wild-type ␥, indicating a functional requirement for the N terminus and M2. Moreover, co-expression of ␥ ⌬3-53 with ␥ L54X did not reconstitute a functional ␥ subunit. The functional requirement for the N terminus may reflect its involvement in intersubunit associations, as we observed in the co-immunoprecipitation assay. The requirement for M2 might be explained by its contribution to the channel pore (32) and suggests that pseudohypoaldosteronism type 1-associated mutants that lack M2 are nonfunctional.
Functional Interactions between Truncated ␥ Subunits and hENaC-We also used a functional assay in Xenopus oocytes to investigate interactions between ␣ and ␥hENaC. Since hENaC subunits interact to form heteromeric complexes, we hypothesized that the interaction of a nonfunctional, truncated ␥ subunit with the hENaC channel complex might have an inhibitory effect on channel function. To promote assembly of truncated subunits into the channel complex, we expressed wild-type ␣, ␤, and ␥ subunits with a relative overabundance of truncated ␥ subunits. To control for nonspecific effects on protein synthesis, some oocytes were injected with a plasmid encoding secreted alkaline phosphatase (28) in place of plasmids encoding truncated ␥ subunits. Expression secreted alkaline phosphatase does not alter the amount of current generated by coexpression of ␣, ␤, and ␥. As shown in Fig. 7A, ␥ E518X , ␥ I102X , and ␥ L54X inhibited amiloride-sensitive current in oocytes coexpressing all three wild-type subunits. Since these constructs also co-precipitated ␣, their inhibitory effect was consistent with a direct interaction with the ␣␤␥ complex, at least in part through the ␣ subunit. To test the possibility that the inhibitory effect required the presence of ␤ or ␥hENaC, we expressed only wild-type ␣hENaC with ␥ I102X or ␥ L54X . In this assay as well, both ␥ I102X and ␥ L54X inhibited amiloride-sensitive current (Fig. 7B). The functional effect of these constructs was consistent with the results of the co-precipitation assay, which showed that the N terminus of ␥hENaC was a site of interaction with ␣.
We hypothesized that if the truncated ␥ subunits interacted directly with full-length subunits to form nonfunctional channel complexes, these complexes might be degraded by cellular quality control mechanisms, thereby reducing the amiloridesensitive current. By a similar mechanism, Xenopus oocytes degrade inward rectifier K ϩ channels if nonfunctional channel subunits are co-expressed (30). Fig. 7C shows that oocytes co-expressing ␣ FLAG with ␥ I102X or ␥ L54X contained much less ␣ FLAG protein than control oocytes. This result was consistent with our hypotheses that truncated ␥ subunits assemble into an unstable complex with full-length subunits, and that the ␥ N terminus is a site of channel assembly.

DISCUSSION
These studies provide both biochemical and functional data showing protein-protein interactions between hENaC subunits. Previous studies indicated that maximal hENaC function required three homologous subunits (␣, ␤, and ␥), suggesting that the subunits interact to form a heteromeric channel complex. This hypothesis was confirmed by our demonstration that hENaC subunits interact with each other and with themselves. The interactions between hENaC subunits appeared to occur early in biosynthesis, prior to glycosylation. Furthermore, the interactions were relatively strong and were not disrupted by nonionic detergents or high concentrations of salt. In the assembly of some heteromeric channels, such as the nicotinic acetylcholine receptor, interaction between specific subunits requires the presence of a third subunit, suggesting a definite order of assembly (33,34). In contrast, we found that each hENaC subunit was capable of interacting with either of the other subunits. In addition, we detected subunit associations (␤-␥, ␤-␤, and ␥-␥) that do not produce functional Na ϩ channels. It is possible that variation in subunit composition might be a mechanism to generate diversity in channel function and regulation.
To begin to define the subunit domain(s) that participate in intersubunit associations, we performed a deletion analysis of the ␥ subunit and found that the ␥ N terminus alone could co-precipitate ␣. Interestingly, the ␥ N terminus interacted only with unglycosylated ␣ and did not co-immunoprecipitate glycosylated ␣. There are a number of possible explanations. First, the precipitating antibody may not recognize the ␥ N terminus when it is associated with glycosylated ␣. Second, the ␥ N terminus may interact with full-length ␣ early, then dissociate when ␣ is glycosylated. Third, the N terminus may associate with unglycosylated ␣ and prevent its glycosylation. While our experiments here do not address these issues, the third possibility is consistent with our experiments in Xenopus oocytes, which showed an effect of the ␥ N terminus on ␣ biosynthesis.
Our studies of truncated ␥ subunits also indicated that there is not a single site that is essential for intersubunit association. This result was consistent with studies of other channels that have shown multiple sites of subunit interactions (35). One interesting structural feature of hENaC subunits is the large number of highly conserved extracellular cysteines, which could potentially participate in intersubunit disulfide bonds. Although these cysteines were not required for intersubunit In B and C, ␥ subunits were immunoprecipitated with anti-N␥ antisera, and ␣ FLAG was detected by Western blot using anti-FLAG antibody. In D, ␥ ⌬3-53 was immunoprecipitated with anti-FLAG antibody, and ␣ was detected by Western blot using anti-N␣ antisera. In B and D, immunoprecipitates were washed under high salt conditions. In C, only glycosylated ␣ FLAG was observed, probably due to a lower level of protein expression in this experiment. association, it remains possible that extracellular cysteines also contribute to subunit interactions.
Expression of truncated ␥ subunits in Xenopus oocytes identified elements of the ␥ subunit that are essential for function. ␥ subunits lacking either the N terminus or M2 were nonfunctional, even though these regions were not required for subunit association. The functional requirement for M2 may be physi-ologically relevant for two reasons. First, alternatively spliced ␣hENaC transcripts with stop codons upstream of M2 have been identified in kidney, lung, and taste tissues (36). Like ␥ E518X , ␥ I102X and ␥ L54X , the ␣ subunits encoded by alternatively spliced messages are nonfunctional. However, our studies suggest that alternatively spliced transcripts, if translated, could associate with full-length subunits, and thus could inhibit hENaC function. Second, our data indicate that pseudohypoaldosteronism-associated mutations that disrupt M2 in hENaC subunits may cause pseudohypoaldosteronism type 1 by abolishing subunit function.
In studies of other multimeric ion channels, the interaction of nonfunctional channel subunits with functional ones disrupted channel function and/or cell surface expression (30,(37)(38)(39). Therefore, as an additional assay of subunit interactions, we examined the functional effects of truncated ␥ subunits on wild-type hENaC channels. In oocytes expressing wild-type ␣, ␤, and ␥ hENaC, or ␣ hENaC alone, ␥ subunits that contained the cytoplasmic N terminus reduced the magnitude of amiloride-sensitive currents. N termini are important in the assembly of other ion channels, including voltage-gated K ϩ channels and the nicotinic acetylcholine receptor (37, 39 -41). In addition, cytoplasmic N termini have been shown to be important regulators of function in channels after they have been assembled (42,43). Interestingly, proteins closely related to hENaC, the degenerins of C. elegans, also appear to be inhibited by overexpression of their cytoplasmic N termini. 2 Thus, the functional effect of the ␥ N terminus may reflect a general property of this family of ion channels.
To determine if the truncated ␥ subunits affected hENaC assembly, we examined the amount of ␣ subunit protein in oocytes co-expressing mutant ␥ subunits. Like other types of cells, Xenopus oocytes possess a quality control mechanism to degrade misfolded or improperly assembled proteins. Co-expression of truncated ␥ subunits caused a dramatic reduction in the amount of ␣, suggesting that a reduction in the number of functional channel proteins was the underlying mechanism of inhibition. This result was consistent with a model in which the truncated ␥ subunits interact with ␣ early in subunit biogenesis, forming unstable complexes of mutant and wild-type protein that are degraded by the cell. FIG. 6. Function of truncated ␥ subunits in Xenopus oocytes. Oocytes were injected with a 1:1 ratio of the indicated DNA constructs. The membrane potential was clamped to Ϫ80 mV, and the current inhibited by 100 M amiloride was measured. Currents were normalized to the average current of the ␣␥ group. Bars represent S.E.; in some cases they are too small to see. Numbers of oocytes studied in each experiment are shown in parentheses. Asterisk indicates a significant difference from control currents (p Ͻ 0.004).

FIG. 7. Inhibitory effects of truncated ␥ subunits on hENaC.
Oocytes were injected with a 1:1:1:4 DNA ratio of ␣ to ␤ to ␥ to truncated ␥ or secreted alkaline phosphatase (A), or a 1:4 DNA ratio of ␣ to truncated ␥ or secreted alkaline phosphatase (B). Amiloride-sensitive currents were measured at a holding potential of Ϫ60 mV in A and at Ϫ80 mV in B. Average currents were normalized to those of control cells, which expressed secreted alkaline phosphatase instead of a truncated ␥. secreted alkaline phosphatase was used as a control for effects on protein synthesis. Bars represent S.E. Asterisks indicate a significant difference from control currents (p Ͻ 0.001). Numbers of oocytes studied in each group are shown in parentheses. C, oocytes were injected with a 1:4 DNA ratio of ␣ FLAG to truncated ␥ or secreted alkaline phosphatase. Oocyte membrane proteins were isolated, then separated by SDS-polyacrylamide gel electrophoresis, and ␣ FLAG was detected by Western blot using anti-FLAG antibody. Longer exposures revealed a small amount of ␣ FLAG protein from oocytes co-expressing truncated ␥ subunits.