The epithelial Na+ channel (ENaC), composed of three subunits (αβγ), plays a critical role in salt and fluid homeostasis. Abnormalities in channel opening and numbers have been linked to several genetic disorders, including cystic fibrosis, pseudohypoaldosteronism type I and Liddle syndrome. We have recently identified the ubiquitin‐protein ligase Nedd4 as an interacting protein of ENaC. Here we show that ENaC is a short‐lived protein (t1/2 ∼1 h) that is ubiquitinated in vivo on the α and γ (but not β) subunits. Mutation of a cluster of Lys residues (to Arg) at the N‐terminus of γENaC leads to both inhibition of ubiquitination and increased channel activity, an effect augmented by N‐terminal Lys to Arg mutations in αENaC, but not in βENaC. This elevated channel activity is caused by an increase in the number of channels present at the plasma membrane; it represents increases in both cell‐surface retention or recycling of ENaC and incorporation of new channels at the plasma membrane, as determined by Brefeldin A treatment. In addition, we find that the rapid turnover of the total pool of cellular ENaC is attenuated by inhibitors of both the proteasome and the lysosomal/endosomal degradation systems, and propose that whereas the unassembled subunits are degraded by the proteasome, the assembled αβγENaC complex is targeted for lysosomal degradation. Our results suggest that ENaC function is regulated by ubiquitination, and propose a paradigm for ubiquitination‐mediated regulation of ion channels.
Amiloride‐sensitive epithelial Na+ channels, located at the apical membrane of Na+‐transporting epithelia, play an essential role in the control of salt and fluid homeostasis in the kidney and the colon (Rossier and Palmer, 1992), as well as in fluid clearance from the alveolar space at birth and during pulmonary oedema (Saumon and Basset, 1993; O'Brodovich, 1995). These channels are tightly regulated by hormones such as aldosterone, vasopressin and insulin as well as PKA/cAMP, PKC, Ca2+ and G proteins (Garty and Palmer, 1997).
The amiloride‐sensitive epithelial Na+ channel (ENaC) was cloned recently from rat (Canessa et al., 1993, 1994a; Lingueglia et al., 1993, 1994), human (McDonald et al., 1994, 1995; Voilley et al., 1994) and other species, and shown to be composed of three homologous subunits, α‐, β‐ and γENaC. Each subunit consists of two transmembrane regions, a large extracellular domain and cytosolic N‐ and C‐termini (Canessa et al., 1994b; Renard et al., 1994; Snyder et al., 1994) including proline‐rich regions in each C‐terminus (Rotin et al., 1994; Staub et al., 1996). Proper function of ENaC is crucial, as indicated by the finding that a gene knockout of αENaC is lethal due to the inability of the −/− mice to clear fluid from their lungs shortly after birth (Hummler et al., 1996). Abnormally high ENaC activity has been also demonstrated in cystic fibrosis patients (Stutts et al., 1995). Moreover, several inherited human disorders were recently linked by genetic linkage analysis to mutations in the ENaC subunits. These include the salt‐wasting disease pseudohypoaldosteronism type 1 (PHA1, Chang et al., 1996; Strautnieks et al., 1996) and Liddle syndrome, an inherited autosomal dominant form of human hypertension (Liddle et al., 1963; Botero‐Velez et al., 1994). In all these disorders, alteration in channel gating and/or numbers has been associated with the disease (Chang et al., 1996; Firsov et al., 1996).
We have recently identified the ubiquitin–protein ligase Nedd4 as an interacting protein of ENaC (Staub et al., 1996), and therefore wanted to investigate whether ENaC is ubiquitinated in living cells and whether such putative ubiquitination is involved in regulating channel number and function. Ubiquitination of cellular proteins usually serves to tag them for rapid degradation (reviewed in Ciechanover, 1994; Jentsch and Schlenker, 1995). It involves the covalent attachment of ubiquitin or a polyubiquitin tree onto lysine residues in target proteins. Several enzymes are involved in this process, including a ubiquitin‐activating enzyme (E1), a ubiquitin‐conjugating enzyme (E2) and a ubiquitin–protein ligase (E3). In most studies described to date, ubiquitinated proteins are degraded by the 26S proteasome, a cytosolic threonine protease complex (Jentsch and Schlenker, 1995; Hilt and Wolf, 1996). Extensive work has implicated the involvement of the ubiquitin‐proteasome system in regulating/degrading numerous cytosolic proteins, including several cell cycle proteins (e.g. cyclin A, B, Clb5p, cln 1, 2, 3p, SIC1) (reviewed in Deshaies, 1995), NFκB and IκB (Palombella et al., 1994), and the nuclear proteins c‐Jun (Treier et al., 1994) and p53 (Scheffner et al., 1993). In addition, recent evidence suggests that ER‐associated protein degradation of either misfolded or improperly assembled protein complexes involves the proteasome in either a ubiquitin‐dependent or ‐independent fashion (reviewed in Brodsky and McCracken, 1997). For example, the immature form of CFTR becomes ubiquitinated in the ER and degraded by the proteasome (Jensen et al., 1995, Ward et al., 1995).
In recent years, it has become apparent that some transmembrane proteins also become ubiquitinated, and that ubiquitination is involved in their subsequent degradation by the endosomal/lysosomal pathway (reviewed in Hochstrasser, 1996). Examples of mammalian transmembrane proteins which become ubiquitinated include several tyrosine kinase receptors (e.g. EGFR, PDGFR, c‐kit), cytokine receptors (e.g. T cell and IgE receptors) (Cenciarelli et al., 1992; Mori et al., 1992; Paolini and Kinet, 1993; Miyazawa et al., 1994; Galcheva‐Garbova et al., 1995) and the growth hormone receptor, in which ligand‐induced, ubiquitination‐dependent endosomal/lysosomal degradation of the receptor was recently demonstrated (Strous et al., 1996). In yeast, the membrane receptor Ste2 (Hicke and Riezman, 1996) and membrane transporters Ste6 and Pdr5 (Kölling and Hollenberg, 1994; Egner and Kuchler, 1996), as well as several yeast amino acid permeases including Gap1 and Fur4 (Hein et al., 1995; Galan et al., 1996), have been shown to become ubiquitinated, a necessary step for their subsequent degradation in the vacuoles. The exact degradation pathway for ubiquitinated transmembrane proteins, and the possible involvement of the proteasome complex in this process, is not known (Hochstrasser, 1996).
In this report we show that ENaC is a short‐lived protein which becomes ubiquitinated in living cells on the α and γ subunits. The rapid turnover of ENaC is sensitive to inhibitors of both the lysosomal and proteasomal degradation systems; whereas excess of unassembled (or misfolded) subunits are degraded by the proteasome, our data suggest that the assembled αβγENaC complex is targeted to the lysosome. Moreover, mutating a cluster of lysines (to arginines) at the N‐terminus of γENaC, or α‐ plus γENaC, causes reduced ubiquitination, slower turnover and increased Na+ channel activity due to increased number of channels at the plasma membrane. Using Brefeldin A treatment, we demonstrate that this increase is caused by an elevated number of mutant channels retained at the cell surface, as well as by increased incorporation of mutant channels into the plasma membrane. These results suggest that ubiquitination of ENaC controls the number of channels at the plasma membrane, and provides a powerful means to regulate channel function.
ENaC subunits are short‐lived proteins with a rapid turnover rate
We have shown previously (Staub et al., 1996) that the ubiquitin protein ligase Nedd4 binds to ENaC. Since ubiquitination of proteins is usually associated with their rapid turnover, we initially investigated the turnover rate of αβγENaC using pulse–chase experiments. For these and all subsequent experiments, we used rat αβγENaC (rENaC), which share ∼85% overall amino acid sequence identity with the corresponding human ENaC (hENaC) subunits (Canessa et al., 1993, 1994a; McDonald et al., 1994, 1995). Pulse–chase experiments were performed with either kidney epithelial MDCK cells or with NIH‐3T3 fibroblasts previously triple‐transfected and stably expressing αβγENaC (Stutts et al., 1995). Cells were labelled for 2 h with [35S]methionine/cysteine and then chased for various periods of time. The αβγENaC proteins were subsequently immunoprecipitated with antibodies directed against each of the ENaC subunits, and separated on SDS–PAGE (Figure 1). Following SDS–PAGE and autoradiography, bands were excised and radioactivity quantified by scintillation counting to determine half‐life (t1/2) of the proteins. As is evident from Figure 1A, α‐ and γENaC expressed in MDCK cells have a rapid turnover rate with a half‐life of ∼1 h; βENaC, on the other hand, has a longer half‐life (t1/2 >3 h). The half‐life of the ENaC subunits measured following 1 h pulse was the same as that following 2 h pulse (not shown). In addition, the turnover rates of αβγENaC expressed in NIH‐3T3 cells were very similar to those expressed in MDCK cells (data not shown).
α‐ and γENaC, but not βENaC, are ubiquitinated in vivo
A rapid turnover of proteins is often associated with ubiquitination‐mediated degradation. We thus wanted to test whether the αβγENaC subunits are ubiquitinated in living cells. For these experiments, we used epitope‐tagged α‐ or γENaC because the antibodies available against these native subunits, unlike the antibodies against native βENaC, are not sensitive enough for immunoblotting. We subcloned HA‐tagged αENaC, FLAG‐tagged γENaC or untagged βENaC into CMV‐based expression vectors. The HA or FLAG tags were added just upstream of the stop codons of α‐ or γENaC respectively, since tagging at these sites had no effect on channel activity (not shown). We then transiently co‐transfected HA‐tagged αENaC, FLAG‐tagged γENaC and βENaC together with a plasmid encoding His‐tagged multiubiquitin (His‐Ub) (Treier et al., 1994) (kindly provided by D.Bohmann) into Hek‐293 cells. These cells were chosen because MDCK cells are not easily amenable to transient transfections. The αβγENaC subunits were expressed together (in Hek‐293 cells) to allow for the formation of a functional channel complex that is able to reach the plasma membrane (Canessa et al., 1994a; Firsov et al., 1996). Twenty‐four hours after transfection, cells were lysed, denatured with 2% SDS, diluted with lysis buffer and the diluted lysates incubated with Ni2+‐NTA beads to allow for binding of the histidinated (and therefore ubiquitinated) proteins. After thorough washes, bound proteins were eluted and separated on SDS–PAGE, followed by immunoblotting with either anti‐HA (i.e. anti‐αENaC), anti‐FLAG (i.e. anti‐γENaC), or anti‐βENaC antibodies directed against the C‐terminus of βENaC (kindly provided by B.C. Rossier). As can be seen in Figure 2A, B and C (bottom panels, lysates), α‐, β‐ and γENaC were all expressed in the transfected cells (in the same experiment). The precipitation of His‐tagged (ubiquitinated) proteins from lysates of these transfected cells (with Ni2+‐NTA beads) revealed high molecular weight ubiquitinated species of αENaC (Figure 2A, top panel, bracket) and γENaC (Figure 2C, top panel, bracket), as detected by anti‐HA or anti‐FLAG antibodies respectively; these ubiquitinated ENaC proteins were only detected in cells co‐transfected with the His‐Ub construct, and not in cells transfected with the ENaC subunits alone or with the His‐Ub construct alone (Figure 2A, B and C, top panels). In contrast to α‐ or γENaC, the βENaC subunit was not ubiquitinated (Figure 2B, upper panel), even though βENaC was strongly expressed in these cells (Figure 2B, lower panel). The ubiquitination of α‐ and γENaC did not seem to be a peculiar property of one cell line, because it was also observed in transfected COS‐7 cells (not shown). These results, therefore, demonstrate that in αβγENaC‐transfected cells, α and γENaC, but not βENaC, are ubiquitinated in vivo.
Mutation of conserved N‐terminal lysines hyperactivates ENaC channel activity
Ubiquitination of proteins is mediated by the addition of ubiquitin or multiubiquitin groups onto Lys residues. Since we identified ubiquitination of α‐ and γENaC in vivo, we wanted to investigate whether mutation of Lys residues, which may serve as the ubiquitin attachment sites, could lead to increased Na+ channel activity. Inspection of the sequences of rat (r) and human (h) αβγENaC (Canessa et al., 1994a; McDonald et al., 1994, 1995) revealed conserved Lys residues at the N‐termini of these subunits (Figure 3). Most of these lysines are also conserved in xENaC, the Xenopus homologue of ENaC (Puoti et al., 1995). The cytosolic C‐termini of these subunits are largely devoid of lysines, and the one or two lysines that are present in β‐ or γrENaC (respectively) are very close to the transmembrane domain and are not conserved. We therefore mutated the conserved lysines at the N‐termini of α‐, β‐ and γrENaC (herein referred to as αβγENaC) to arginines either individually or in groups in those cases where the lysines were clustered close to each other. The mutants generated were as follows (see Figure 3): (i) In αENaC: K47 and K50 (αK47,50R), K108 (αK108R) or αK47,50R plus αK108R (αK47,50,108R). An additional αENaC construct was also generated, in which lysines 23, 26 and 32 (which are not conserved between rat and human) of the αK47,50R construct were mutated to Arg, yielding the mutant αK23‐50R. (ii) In βENaC: K4, K5 and K9 (βK4,5,9R) or K16, K23, K39, K47, K48 and K49 (βK16‐49R). (iii) In γENaC: K6, K8, K10, K12, K13 (γK6‐13R), K26R (γK26R) or K6‐13 plus K26R (γK6‐13+26R).
We then investigated the effect of the above Lys to Arg mutations in αβγENaC on ENaC channel activity. Two‐electrode voltage clamp experiments were performed in which amiloride‐sensitive Na+ currents (INa) were measured in Xenopus oocytes expressing the different Lys to Arg mutants. Our results show that none of the conserved Lys to Arg αENaC mutants, when expressed together with wild type (wt) βγENaC, had any effect on increasing ENaC activity (Figure 4A). This lack of effect on ENaC function was also observed with the extended αK23‐50R mutant (not shown). Similarly, none of the Lys to Arg mutations within the N‐terminus of βENaC, expressed together with wt‐αγENaC, caused any increase in amiloride‐sensitive Na+ currents. In contrast, the expression of the γENaC mutants (γK6‐13R or γK6‐13+26R) together with wt‐αβENaC more than doubled Na+ channel activity (Figure 4A). Moreover, when this γK6‐13R mutant was co‐expressed together with the αK47,50R mutant (along with wt‐βENaC), the hyperactivation of ENaC was potentiated, yielding up to 6‐fold stimulation of Na+ channel activity when compared to the activity of the wt channel (Figure 4A). In contrast, such potentiation was not observed when the βK4,5,9R or the βK16‐49R mutant was co‐expressed with either αK47,50R, γK6‐13R or both mutants (Figure 4A). These results therefore demonstrate that a cluster of conserved Lys residues particularly in γENaC, but also in αENaC, are responsible for the regulation of ENaC activity, and when mutated, lead to an increase in amiloride sensitive Na+ current (INa). Moreover our data show that the Lys to Arg substitutions increase channel activity only when performed on subunits previously shown to be ubiquitinated in vivo (α‐ and γENaC).
Lys to Arg mutations in γENaC, or in α‐ plus γENaC, lead to an increase of Na+ channel numbers at the cell surface
The elevated Na+ channel activity associated with the Lys to Arg mutations in γENaC or α‐ plus γENaC was determined by an increase in amiloride‐sensitive Na+ current. In theory, such an increase can result from either an increase in the number of channels (N) expressed at the cell surface, and/or higher Na+ influx per channel molecule when the open probability (Po) or single channel conductance are increased. Analysis by patch clamp technique of single channel currents of the αK47,50R plus γK6‐13R double mutant (Figure 4B) revealed a high number of active channels per patch without significant changes in single channel conductance for Li+ ions, a highly permeant cation in ENaC. The conductance for Li+ ions of 8–9 pS was similar to that found for wt‐ENaC (Schild et al., 1997), indicating that these Lys to Arg mutations did not affect the conductive properties of the channel. To distinguish between changes in the channel open probability and increased surface expression of ENaC, we used a quantitative binding assay to correlate in individual oocytes the amiloride‐sensitive INa with the number of channel molecules present at the cell surface, as previously described (Firsov et al., 1996). This binding assay uses specific 125I‐labelled antibodies directed against a FLAG epitope introduced into the extracellular domain of each of the αβγENaC subunits (wt or the above Lys to Arg mutants), as described in Materials and methods. Such epitope tagging does not interfere with channel function (Firsov et al., 1996). Figure 6A and B show representative experiments in which amiloride‐sensitive INa and surface expression of wt‐ENaC or Lys to Arg mutants of α‐ and γENaC (αK47,50R; γK6‐13R; αK47,50R + γK6‐13R) were measured in individual oocytes. The Lys to Arg mutants of βENaC were not tested since none of them was able to elevate channel activity (Figure 4A). Comparing the ENaC mutants with wt confirms that the αK47,50R mutations did not increase the amiloride sensitive Na+ current (as shown in Figure 4A) or the number of channel molecules expressed at the cell surface (Figure 5A). For the γK6‐13R or the αK47,50R plus γK6‐13R double mutant, the increase in INa relative to wt‐ENaC was directly proportional to the increase in the number of binding sites recognized by the anti‐FLAG antibodies. The significant linear correlation between the current expressed and the number of ENaC channel molecules indicates that the Na+ current per channel was not significantly altered in the Lys to Arg mutants relative to wt‐ENaC. These results are summarized quantitatively in Figure 5C. Since the INa measured in the Lys to Arg mutants increased linearly with the specific binding of the anti‐FLAG antibodies, we conclude that this increase was caused by an elevated number of channels at the cell surface, and not by changes in the channel open probability. This increase in number of channels at the plasma membrane induced by the Lys to Arg mutations in the α and γ subunits would be consistent with an inhibition of ubiquitination of the channel protein.
Inhibition of ubiquitination of the Lys to Arg mutants of γENaC, or of α‐plus γENaC
As we observed an increase in channel activity resulting from a greater number of αβγENaC at the plasma membrane in the γK6‐13R mutant, we wanted to test whether such increased cell surface number in ENaC channels is caused by impaired ubiquitination of this mutant. Since direct determination of ENaC ubiquitination in oocytes is not currently feasible due to limits of protein detection (since only a few oocytes can be analysed at a time, not sufficient for biochemical assays), we opted instead to use mammalian cells, which allow the use of a large number of cells per assay and are thus more suitable for biochemical studies. We therefore co‐transfected Hek‐293 cells with the His‐Ub construct together with either wt‐αβγENaC (control), or wt‐αβENaC plus the γK6‐13R mutants of γENaC. As before, the αENaC constructs were HA‐tagged and the γENaC constructs were FLAG‐tagged. Following transfection, cells were lysed, and lysates incubated with Ni2+‐NTA beads to precipitate ubiquitinated proteins. These proteins were then separated on SDS–PAGE, transferred to nitrocellulose and immunoblotted with anti‐FLAG antibodies, to detect the extent of ubiquitination of γENaC. Our results show that ubiquitination of γENaC was indeed inhibited in the γK6‐13R mutant (Figure 6A, upper panel, compare lanes 3 and 4), despite similar expression levels of the wt or mutant γENaC proteins in the lysate of the transfected cells (Figure 6A, lower panel, lanes 3 and 4). The same inhibition of ubiquitination of γENaC was also observed when the γK6‐13R mutant was expressed together with the αK23‐50R mutant of αENaC (Figure 6A, lanes 5 and 6). These results therefore demonstrate that those conserved Lys residues located at the N‐terminus of γENaC which are involved in regulating channel stability at the plasma membrane (γK6‐13) indeed serve as attachment sites for ubiquitin.
We next wanted to determine the extent of ubiquitination of the Lys to Arg mutants of αENaC when expressed either with wt‐βγENaC or with the γK6‐13R mutant of γENaC (plus wt‐βENaC). This was done in order to test whether the potentiation of ENaC hyperactivation and retention at the cell surface observed in the double mutant (i.e. Lys to Arg mutations in both α‐ and γENaC) is associated with reduced ubiquitination of not only γENaC, but also of αENaC. We therefore re‐probed the blot shown in Figure 6A with HA antibodies, to determine the extent of ubiquitination of αENaC in the same experiment. Our results show that the αK23‐50R mutant [similar to the αK47,50R mutant (data not shown)] was still ubiquitinated when co‐expressed with wt‐βγENaC and His‐Ub (Figure 6B, lane 3). In contrast, however, there was a clear and reproducible inhibition of this ubiquitination when the αK23‐50R mutant was co‐transfected with the γK6‐13R mutant (and wt‐βENaC plus His‐Ub) (Figure 6B, lane 4). This reduction in ubiquitination was apparent despite similar expression levels of the wt or mutant αENaC proteins (Figure 6B, lower panel). Thus, the combined removal of lysines 6–13 from γENaC together with N‐terminal lysines from αENaC caused a reduction of ubiquitination of both subunits, in agreement with the observed potentiation of cell surface retention and activity of ENaC under these conditions. None of the above Lys to Arg mutations in α or γENaC, or both, caused any detectible ubiquitination of βENaC (Figure 7C).
Our data show that ENaC turnover is regulated by ubiquitination, and that removal of potential ubiquitination sites on γENaC (γK6‐13R) leads to stabilization of ENaC at the plasma membrane. A prediction of these data is that the half‐life of the γK6‐13R mutant should be prolonged. We therefore generated stable MDCK cell lines which express either FLAG‐tagged wt‐γ (control), or FLAG‐tagged γK6‐13R mutant, together with wt‐αβENaC (the FLAG tag does interfere with channel function, data not shown). We then performed a pulse–chase experiment as described in Figure 1 above. Our results show that γENaC bearing the K6‐13R mutations was indeed more stable (mean t1/2 = 90 min) than the wt‐γENaC (mean t1/2 = 76 min), showing a small but statistically significant different (P = 0.027) increase in half‐life of the protein (Figure 6D).
Collectively, these results show that ENaC stability and activity is regulated by ubiquitination. The primary target for this ubiquitination‐mediated regulation is a cluster of lysine residues (K6–13) at the N‐terminus of γENaC, since mutation of these lysine residues was sufficient to both increase channel number at the plasma membrane and to reduce its ubiquitination. αENaC, but not βENaC, is also playing a role in this regulation of channel stability by ubiquitination.
Accumulation of Brefeldin A‐resistant pool of the Lys to Arg ENaC mutant at the cell surface
To test whether the impaired ubiquitination and increased channel number and activity of the Lys to Arg ENaC mutants is associated with increased arrival of newly synthesized channels at the plasma membrane, decreased degradation of cell surface‐associated channels, or both, we tested the effect of Brefeldin A (BFA) on ENaC activity in Xenopus oocytes. Brefeldin A inhibits the ER to Golgi transport of newly synthesized membrane proteins, and has been previously shown to be effective in Xenopus oocytes (Geering et al., 1996). Thus, cRNA of wt‐αβγENaC (wt) or αK47,50R+γK6‐13R (with wt‐β) ENaC was injected into Xenopus oocytes and following overnight incubation, oocytes were treated with 10 μg/ml BFA at time 0 (arrow) for up to 8 h, and amiloride‐sensitive INa, representing active channels at the cell surface, measured by the two‐electrode voltage clamp technique. Our results show that whereas the majority (∼80%) of wt ENaC activity disappeared by 8 h exposure to BFA, 43% (±7%) of the αK47,50R+γK6‐13R mutant channel activity was BFA‐resistant (Figure 7). The accumulation of this BFA‐resistant pool was apparent after 4 h incubation, and suggests that a significant fraction of the αK47,50R+γK6‐13R mutant, unlike the wt ENaC, is retained at the cell surface or is recycled between endocytic vesicles and the plasma membrane. However, the amount of BFA‐sensitive pool at 8 h incubation was also greater in the αK47,50R+γK6‐13R mutant than in the wt channel (average of 9.9 μA versus 5 μA respectively), suggesting that the rate of arrival of newly synthesized channels to the plasma membrane is also elevated in the Lys to Arg mutant. Thus, loss of ubiquitination sites in ENaC likely leads to both an increase in incorporation and increased retention/recycling of the channel at the plasma membrane.
ENaC degradation is sensitive to endosomal/lysosomal and to proteasome inhibitors
Degradation of ubiquitinated proteins is carried out either by the proteasomes (usually of cytosolic or misfolded/misassembled proteins in the ER) or by the endosomal/lysosomal system in the case of several transmembrane proteins. In order to determine the mechanism(s) responsible for the rapid degradation (of total cellular pool) of ENaC seen in Figure 1, we tested the effect of inhibitors of either the proteasome or the endosomal/lysosomal system on the half‐life of αβγENaC expressed in MDCK cells using pulse–chase experiments. Our results show that when the potent proteasome inhibitor lactacystin (10 μM; Fenteany et al., 1995) was added during the 2 h pulse and the subsequent chase period, there was an attenuation of channel degradation (2.1‐ and 1.7‐fold increase in half‐life of α and γENaC respectively; Figure 8A). This suggests an involvement of the proteasome in ENaC degradation. However, chloroquine (0.4 mM), a weak base that dissipates the late endosome/lysosome acidic pH thereby inhibiting proteolysis in these compartments, also led to a significant inhibition (2.7‐ and 1.3‐fold increase in half‐life of the α and γ subunits respectively) of ENaC degradation when added during the chase period (Figure 8B). Our results, therefore, suggest that both the proteasomes and the endosomes/lysosomes are likely to play a role in the degradation of ENaC. One possible explanation for this dual mode of degradation may be that the individual ENaC subunits which are unassembled are targeted for proteasome degradation, whereas the properly assembled αβγENaC subunits are subject to degradation by late endosomes/lysosomes. To test whether the unassembled subunits are indeed degraded by the proteasomes and not by the lysosomes, we used a MDCK cell line which expresses high levels of HA‐tagged αENaC alone (Staub et al., 1996). Despite high levels of expression, these cells display very little amiloride‐sensitive Na+ currents when compared with MDCK cells expressing αβγENaC together, as determined by whole cell patch clamp analysis (T.Ishikawa, Y.Marunaka and D.Rotin, unpublished). Moreover, expression of αENaC alone in Xenopus oocytes leads to a severely reduced ENaC activity due to a reduced number of channels at the cell surface (Canessa et al., 1994a; Firsov et al., 1996). We therefore tested whether αENaC expressed alone is ubiquitinated and whether it is degraded by proteasomes. Figure 9B shows that transfection of αENaC into Hek‐293 cells together with the His‐Ub construct led to a massive ubiquitination of this subunit. Moreover, pulse–chase experiments performed with αENaC stably expressed in MDCK cells reveal a strong inhibition of αENaC degradation by 10 μM lactacystin (Figure 9A, upper panel), whereas chloroquine (0.4 mM) was largely ineffective (Figure 9A, lower panel). These results, therefore, demonstrate that expression of the αENaC subunit alone, which can not properly assemble into a channel in the absence of βγENaC, is targeted for proteasomal degradation.
The results presented here demonstrate collectively that ENaC stability and function are modulated by ubiquitination. Our current results show that ENaC is a short lived protein, with a half‐life of ∼1 h for at least 2 of its subunits. This short t1/2 was detected in ENaC heterologously expressed in Xenopus oocytes, in ENaC transfected into either epithelial (MDCK) cells or fibroblast (NIH‐3T3), or in xENaC endogenously expressed in A6 cells (B.C.Rossier, personal communication). The accumulation of ENaC following treatment of cells with the weak base chloroquine (or with NH4Cl; our unpublished data) which dissipate the acidic pH in late endosomes/lysosomes, suggests the involvement of the endosomal/lysosomal pathway in ENaC degradation. However, our findings that the half‐life of the ENaC subunits is also prolonged when the cells are treated with lactacystin (or with MG132; data not shown), indicate that the proteasome is involved in ENaC breakdown as well. One possible explanation for these dichotomous data may be that a portion of the ENaC subunits which do not become properly assembled in the ER may be degraded in a proteasome‐dependent manner, whereas the properly assembled αβγENaC is targeted to the cell surface and is subsequently degraded by late endosomes/lysosomes. Although we cannot currently preclude the possibility of a sequential course of events whereby lysosomal degradation is followed by proteasomal degradation, we believe different pools (assembled versus unassembled) of ENaC are targeted to the different pathways. This is based on our results demonstrating lactacystin‐sensitive and chloroquine‐insensitive degradation of αENaC when expressed alone, but chloroquine‐sensitivity when expressed together with βγENaC, as well as our preliminary results demonstrating a small additive stabilization of αENaC by lactacystin plus chloroquine relative to each inhibitor alone (O.Staub and D.Rotin, unpublished data). We believe our data represent the first documented case where the same ubiquitinated protein (e.g. αENaC) can be degraded by either the proteasome or the lysosome, depending on its assembly/disassembly status.
Although we propose a lysosomal‐mediated degradation of αβγENaC when properly assembled, we currently do not know how much of that pool of protein is actually located at the plasma membrane at any one time. We believe, however, that the ubiquitination of α‐ and γENaC we observed represents, at least in part, ubiquitinated mature channel at the plasma membrane and not just of excess unassembled ENaC chains. This conclusion is based on our finding of a tight correlation between ubiquitination of ENaC and its number (and function) at the cell surface: mutations of a cluster of Lys residues in γENaC (or α‐ plus γENaC) lead to both a reduction of ubiquitination and increased number of channels at the plasma membrane, with a concomitant elevated Na+ channel activity. Moreover, our Brefeldin A experiments suggest that a significant fraction of this increase in channel numbers in the ubiquitination‐defective ENaC mutant results from increased retention (or recycling) of ENaC at the cell surface. Thus, although we do not currently know at what point(s) during processing and plasma membrane incorporation of ENaC it becomes ubiquitinated, the consequences of such ubiquitination are clearly affecting channel stability at the cell surface.
The fact that the status of ENaC ubiquitination determines its stability at the plasma membrane is intriguing, because so far it is not known how ubiquitination signals for, or is involved in, degradation of transmembrane proteins. In this regard, the involvement of ubiquitination in regulating ENaC degradation at the cell surface is similar to that of other recently described transmembrane proteins. A notable example is the ubiquitination‐dependent vacuolar degradation of the Saccharomyces cerevisae permeases Fur4 and Gap1, which require the presence of intact Rsp5 (the yeast homologue of Nedd4) for this degradation (Hein et al., 1995; Galan et al., 1996). We have previously demonstrated direct binding of ENaC to the ubiquitin‐protein ligase Nedd4, by association of the WW domains of Nedd4 with the PY motifs of the ENaC subunits (Staub et al., 1996). We do not know yet whether Nedd4 participates in the ubiquitination of the ENaC subunits described in this report. We can speculate, however, that if it does, this may have implications for Liddle syndrome, an hereditary form of hypertension caused by deletions/mutations within the PY motifs of β‐ or γENaC (Shimkets et al., 1994; Hansson et al., 1995a,b; Tamura et al., 1996). Such mutations lead to increased Na+ channel activity due to increased channel number and opening at the plasma membrane (Schild et al., 1995, 1996; Snyder et al., 1995; Firsov et al., 1996). The same mutations also lead to an inhibition of binding to Nedd4‐WW domains (Staub et al., 1996). Thus, we can speculate that the increase in retention of ENaC at the plasma membrane associated with Liddle syndrome may involve reduced ubiquitination, possibly by reduced binding to Nedd4. Future experiments will address this possibility.
Regardless of the link to Liddle syndrome, the observation that ENaC is ubiquitinated in living cells, and that this ubiquitination is critical for the regulation of channel degradation or stability at the plasma membrane and hence to channel activity, is by itself important, as it provides the first demonstration of regulation of ion channel function by ubiquitination. Clearly, a tight regulation of ENaC is necessary for maintenance of salt and water homeostasis; impairment of such regulation has been linked to several diseases, including not only Liddle syndrome, but also PHA1 (Chang et al., 1996; Strautnieks et al., 1996), pulmonary oedema (Hummler et al., 1996) and cystic fibrosis (Stutts et al., 1995). It remains to be identified whether any subsets of genetically determined hypertensive patients (Liddle syndrome or otherwise) indeed have mutations in the conserved lysine residues at the N‐termini of α‐ or γENaC which serve as attachment sites for ubiquitin.
Our data suggest that γENaC is the important, if not the predominant subunit which is regulating ubiquitination‐mediated degradation of the Na+ channel. Ubiquitination of αENaC, however, also plays a role in channel stability, which becomes significant only when the key lysine residues in γENaC (K6–13) are lost. We propose therefore that loss of the main ubiquitination target of ENaC (γK6‐13) allows the ubiquitination of αENaC to partially compensate for this loss; consequently, losing those N‐terminal lysines in both γ‐ and αENaC greatly stabilizes the channel. Our measurements of turnover rate of αβγENaC indicate that the t1/2 of βENaC is somewhat longer than that of α‐ and γENaC. We currently do not know whether such a difference in t1/2 is biologically meaningful. However, it is curious that βENaC (when expressed together with αγENaC) does not become ubiquitinated despite the presence of multiple conserved N‐terminal lysine residues, and that mutation to arginines of these conserved lysines did not lead to channel hyperactivation. This suggests that the ubiquitination‐mediated regulation of channel stability is not targeting the β subunit directly, but rather indirectly, via regulation of γ‐ and αENaC. Since ENaC is only efficiently exported to the plasma membrane in the presence of all three subunits (Canessa et al., 1994a; McDonald et al., 1995; Firsov et al., 1996), targeting any subunit for degradation is sufficient to dismantle the channel and abrogate its activity. The reason for the observed partial inhibition of ENaC activity in some of the Lys to Arg mutants of βENaC is not known, but we speculate such mutations may cause the formation of defective, less well functioning channels, perhaps similar to the recently described G37S mutation in the N‐terminus of βENaC which causes PHA1 due to reduced ENaC activity (Chang et al., 1996).
In summary, the work presented here demonstrates a fundamental role for ubiquitination in the regulation of ENaC function, via regulation of channel stability at the plasma membrane. This could provide a previously undescribed mechanism to regulate function of ion channels at the cell surface.
Materials and methods
Plasmids and constructs
Rat αβγENaC (rENaC, hereafter called ENaC) cDNA was used for all studies. Plasmids used for transient transfection into Hek‐293 cells: αENaC was tagged with a triple haemagglutinin (HA) epitope (YPYDVPDY) at its carboxy terminal end just upstream of the stop codon and subcloned into a pCMV4 plasmid (Andersson et al., 1989). The addition of the tag at that position did not interfere with Na+ channel function as assessed by patch clamp analysis (T.Ishikawa, Y.Marunaka and D.Rotin, unpublished). Similarly, a FLAG epitope (DYKDDDDK) was introduced just before the stop codon at the carboxy terminus of γENaC and subcloned into the pCEP4 plasmid (Invitrogen). Lysine to arginine mutants of α‐, β‐ and γENaC were generated by substituting specific lysine residues with arginines using the PCR based mutagenesis described by Nelson and Long (1989). These point mutants, mostly of conserved lysines and all located at the N‐termini of the ENaC subunits, were designated as follows (see Figure 4): In αENaC: K47 and K50 (αK47,50R), K108 (αK108R), K47, K50 and K108 (αK47,50,108R) and K23, 26, 32, 47 and 50 (αK23‐50R); In βENaC: K4,5,and 9 (βK4,5,9R), and K16, 23, 39, 47, 48 and 49 (βK16‐49R); in γENaC: K6, 8, 10, 12, 13 (γK6‐13R), K26R (γK26R), and K6–13 plus K26R (γK6‐13+26R). With the exception of K23, 27 and 32 in αENaC, all the above mutated lysines are conserved between rat, human and Xenopus (Canessa et al., 1994a; McDonald et al., 1994; Puoti et al., 1995). All the above lysine to arginine mutants were subcloned into pSD5 for expression into Xenopus oocytes (Canessa et al., 1994a). For expression in Hek‐293 cells, αK23‐50R was subcloned into pCMV4 and γK6‐13R into pCEP4.
Transfections and cell lines
MDCK cells stably expressing rat α‐, β‐ and γENaC together, or expressing only HA‐tagged αENaC, were previously generated (Stutts et al., 1995; Staub et al., 1996) by transfecting cells with αENaC in pLKneo (Hirt et al., 1992) which contains a dexamethasone‐inducible promoter, together with β‐ and γENaC cloned into a CMV‐promoter based vector and expressed constitutively. MDCK previously transfected with α‐ and βENaC (kindly provided by Drs Canessa and Rossier) were stably transfected with pCEP4 encoding either FLAG‐tagged wt‐γENaC or γK6‐13R ENaC. Cells were maintained in DMEM + 10% fetal bovine serum (FBS) + G418 (0.3 mg/ml)+ 10 μM amiloride, to prevent cellular Na+ overload (+ 0.1 mg/ml hygromycin for the FLAG‐tagged γENaC‐expressing cells). For transient transfections, Hek‐293 cells were transfected using lipofectamine, according to manufacturer instructions (Life Sciences).
For pulse–chase experiments, MDCK cells stably expressing α‐, β‐ and γENaC, HA‐tagged αENaC alone, or FLAG‐tagged wt‐γ or γK6‐13R ENaC, stably co‐expressed with wt‐αβENaC, were grown to confluency in 6‐well plates (Costar) and induced overnight in induction medium [DMEM/10% FBS/20 μM amiloride containing 1 μM dexamethasone and 2 mM Na+ butyrate (MDCK expressing αβγENaC), or 1 μM dexamethasone (MDCK expressing HA‐tagged αENaC)]. Cells were washed three times in depletion medium (Induction medium without FBS, cysteine and methionine) and labelled at 37°C (in 5% CO2 in air) for 2 h in labelling medium [Depletion medium + 0.1 mCi/ml of 35S‐containing cell labelling mix (Promix, Amersham)]. Cells were then washed three times with ice‐cold Induction medium containing 10 mM unlabelled methionine and cysteine, and then chased in the same medium for various periods of times at 37°C (in 5% CO2 in air). Where indicated, lactacystin (10 μM) was added to the pulse and chase solutions, and chloroquine (0.4 mM) was added to the wash and chase solution only. Cells were then washed three times with ice‐cold phosphate buffer saline (PBS), and lysed in lysis buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 1.5 mM MgCl2, 1 mM EGTA, 1% Triton X‐100, 10% glycerol, 1 mM PMSF, 10 μg/ml leupeptin, 10 μg/ml pepstatin A, 10 μg/ml aprotinin). Insoluble material was removed by centrifugation and the soluble fraction denatured with 2% SDS and boiling for 5 min at 95°C. The samples were then diluted with 11 volumes of lysis buffer and immunoprecipitations performed by adding polyclonal sera directed against either the N‐terminus of αENaC (Rotin et al., 1994), the C‐terminus of βENaC or the C‐terminus of γENaC (Duc et al., 1994). Immunocomplexes were collected with protein A–sepharose, separated on 7% SDS–PAGE and detected by autoradiography.
Detection of ENaC ubiquitination in vivo
Hek‐293 cells were transiently co‐transfected with a mixture of plasmids encoding HA‐tagged αENaC, untagged βENaC and FLAG‐tagged γENaC together with a construct encoding His‐tagged ubiquitin (His‐Ub), which expresses eight His‐tagged ubiquitin molecules from a CMV promoter (Treier et al., 1994). Twenty‐four hours after transfection, cells were lysed in lysis buffer (containing also 50 μM n‐acetyl‐l‐leucinyl‐l‐leucinyl‐l‐norleucinal, LLnL), insoluble material removed by centrifugation and the soluble fraction denatured in 2% SDS and boiling at 95°C for 5 min. Samples were then diluted with 11 volumes of lysis buffer, Ni2+‐NTA agarose beads (Qiagen) added to the lysate, and the mixture incubated on a rotator for 4 h at 4°C. The beads containing the histidinated (and thus ubiquitinated) bound proteins were washed twice with HNTG (20 mM HEPES, pH 7.5, 300 mM NaCl, 0.1% Triton X‐100, 10% glycerol, 40 mM imidazole) and three times with lysis buffer. Bound proteins were separated on 7% SDS–PAGE, transferred to nitrocellulose and immunoblotted with either anti‐HA antibodies (Babco), anti‐FLAG antibodies (Kodak), or anti‐βENaC antibodies directed against amino acids 559–636 at the C‐terminus (Duc et al., 1994) followed by HRP‐conjugated secondary antibody and chemiluminescence (ECL/Amersham) or Supersignal ULTRA/Pierce detection.
Expression and function of ENaC channels in Xenopus oocytes
Complementary RNA (cRNA) of each α‐, β‐ and γENaC subunit, or their indicated mutants, were synthesized in vitro and equal amounts of subunit cRNA (5 ng total) at saturating concentration for maximal expression were injected into stage V oocytes. ENaC activity at the cell surface resulting from ENaC expression was determined with the two‐electrode voltage‐clamp technique by measurement of the amiloride‐sensitive inward Na+ current (INa) at −100 mV holding potential, as previously described (Schild et al., 1995, 1996). Where indicated, Brefeldin A (BFA, 10 μg/ml) was added to the oocyte bathing medium 18 h after cRNA injection, and INa measured as described above at 0, 2, 4, 6 and 8 h post BFA addition. A Patch clamp technique was used to record single channel currents in the cell‐attached configuration with Li+ ions as permeant cation on defolliculated oocytes, as described previously (Canessa et al., 1994a; Schild et al., 1997).
Surface labelling of ENaC in oocytes
For binding experiments, the lysine to arginine mutations at the N‐termini of α‐ or γENaC (αK47‐50R; γK6‐13R) were introduced into α‐ or γENaC cDNA (respectively) which already contained a FLAG epitope at the ectodomains, located ∼30 residues downstream of the first transmembrane domain (Firsov et al., 1996). The addition of the FLAG tag at that position has no effect on channel function (Firsov et al., 1996). Surface expression of ENaC channels was determined by specific binding of [125I]M2IgG1 (M2Ab) to oocytes expressing FLAG‐tagged αβγENaC, as described previously (Firsov et al., 1996). The equilibrium dissociation constant of specific [125I]M2IgG1 (M2Ab) binding was 3 nM. Accordingly, 12 nM M2Ab were added to a Modified Barth's Saline (MBS) solution supplemented with 5% calf serum incubating the oocytes. After 1 h incubation at 4°C, each oocyte was washed eight times with 1 ml MBS solution and transferred individually to a gamma counter. The same oocyte was and then used to measure amiloride‐sensitive Na+ current. M2Ab non‐specific binding was detected with oocytes injected with ENaC cRNAs encoding non‐tagged subunits.
We would like to thank Drs C.M.Canessa and B.C.Rossier for MDCK cells expressing αβγENaC and anti‐β‐ and γENaC antibodies, Dr D.Firsov for preparation of iodinated antibodies, Drs M.J.Stutts and R.C.Boucher for the ENaC‐expressing NIH‐3T3 cells, and Dr D.Bohmann for the His‐tagged ubiquitin construct. This work was supported by the Canadian CF Foundation (to D.R.), by the Medical Research Council (MRC) of Canada and a MRC Group Grant in Lung Development (to D.R.), by a grant from the International Human Frontier Science Program (HFSP) (to L.S. and D.R.), and by a grant from the Swiss National Foundation for Scientific Research to L.S. (No. 31‐39435.93). A.C. is supported by the Israeli Academy of Sciences, the German–Israeli Foundation for Research and Scientific Development (G.I.F.), the Council for Tobacco Research, Inc. (CTR), and the UK–Israel Binational Biotechnology Foundation. K.B. is supported by MINERVA and by HFSP Fellowships, O.S. was supported by a Fellowship from the Canadian CF Foundation, and D.R. was a recipient of a Scholarship from the MRC of Canada.
- Copyright © 1997 European Molecular Biology Organization