A New Subunit of the Epithelial Na (cid:1) Channel Identifies Regions Involved in Na (cid:1) Self-inhibition*

The epithelial Na (cid:1) channel (ENaC) is the apical entry pathway for Na (cid:1) in many Na (cid:1) -reabsorbing epithelia. ENaC is a heterotetrameric protein composed of homologous (cid:2) , (cid:3) , and (cid:4) subunits. Mutations in ENaC cause severe hypertension or salt wasting in humans; and con-sequently, ENaC activity is tightly controlled. According to the concept of Na (cid:1) self-inhibition, the extracellular Na (cid:1) ion itself can reduce ENaC activity. The molecular basis for Na (cid:1) self-inhibition is unknown. Here, we de scribe cloning of a new ENaC subunit from Xenopus laevis ( (cid:5) xENaC). (cid:5) xENaC can replace (cid:2) xENaC and formed functional, highly selective, amiloride-sensitive Na (cid:1) channels when coexpressed with (cid:3) xENaC and (cid:4) xENaC. Channels containing (cid:5) xENaC showed strong inhibition by extracellular Na (cid:1) . This Na (cid:1) self-inhibition was significantly slower than for (cid:2) xENaC-containing channels. Using site-directed mutagenesis, we show that the proximal part of the large extracellular domain controls the speed of self-inhibition. This suggests that this region is involved in conformational

ubiquitin-mediated endocytosis. This feedback mechanism depends on interaction of a "PY" motif at the carboxyl terminus of ENaC subunits with the ubiquitin ligase Nedd4-2 (6,7). However, the way the intracellular Na ϩ concentration is sensed is unknown.
In contrast, the model of Na ϩ self-inhibition proposes that the extracellular Na ϩ ion itself can reduce the ENaC activity (5, 8 -10). Na ϩ self-inhibition has been best described in amphibian tissues such as Rana skin (8 -11) and Necturus urinary bladder (12). In these tissues, an increase in the extracellular Na ϩ concentration leads to a decline of the Na ϩ current within Ͻ5 s. The speed of self-inhibition with no measurable increase in intracellular Na ϩ concentration has led to the proposal that the extracellular Na ϩ ion itself mediates self-inhibition. Selfinhibition leads to efficient reduction of Na ϩ transport whenever the extracellular Na ϩ concentration rises. This would protect the cell from Na ϩ overload and distribute Na ϩ transport more evenly along an epithelium such as the collecting duct. However, the molecular basis for Na ϩ self-inhibition is unknown.
ENaC has been best characterized in Xenopus laevis among amphibian species. In Xenopus, five ENaC subunits are known. Besides the ␣, ␤, and ␥ subunits (13), which assemble into the classical heteromeric ENaC with an ␣ 2 ␤␥ stoichiometry (1), two isoforms of ␤xENaC and ␥xENaC have been identified (␤ 2 xE-NaC and ␥ 2 xENaC, respectively) (14). These additional isoforms are probably due to polyploidy of Xenopus. Channels containing the ␥ 2 subunit are characterized by a relatively high affinity for Na ϩ , leading to current saturation at low (20 mM) Na ϩ concentration (14). However, Na ϩ self-inhibition is not easily observed with any of the possible combinations of subunits. Very recently, using fast solution exchange, Chraibi and Horisberger (15) observed self-inhibition with recombinant ␣␤␥xENaC channels. However, self-inhibition was too fast to be reliably determined.
Here, we report the cloning of a new ENaC subunit from X. laevis (⑀xENaC). ⑀xENaC-containing channels displayed strong Na ϩ self-inhibition with a reduction of the current amplitude of ϳ70%. Na ϩ self-inhibition of ⑀xENaC-containing channels was significantly slower than that of ␣xENaCcontaining channels, facilitating the investigation of this process. In this study, we identify a region in the extracellular domain of the channel protein that controls the speed of inactivation.

EXPERIMENTAL PROCEDURES
Cloning of ⑀xENaC-We cloned a 220-bp PCR product showing homology to other members of the DEG/ENaC gene family from stage V-VI oocytes of X. laevis as described (16). This PCR product was used to design primers for rapid amplification of 5Ј-and 3Ј-cDNA ends (RACE) (⑀xENaC-5ЈRACE, 5Ј-CCCAAACTGCCAGTACATCATAGC-3Ј; and ⑀xENaC-3ЈRACE, 5Ј-GGCACTGTCCGTCTCAACTGCTC-3Ј) and for the nested PCRs (⑀xENaC-5ЈNRACE, 5Ј-GAGCAGTTGAGACGGA-* This work was supported by Grant FG 1-0-0 from the Attempto Research Group Program of the University of Tü bingen (to S. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AJ440222.
CAGTGCC-3Ј; and ⑀xENaC-3ЈNRACE, 5Ј-GCTATGATGTACTGGCAG-TTTGGG-3Ј). Using the Smart RACE cDNA amplification kit (Clontech, Palo Alto, CA), RACE was performed with poly(A) ϩ RNA from the kidney of an adult X. laevis female. PCR products were subcloned with the TOPO-TA cloning kit (Invitrogen, Groningen, The Netherlands) and sequenced. Full-length ⑀xENaC was assembled from the longest 5Ј-and 3Ј-RACE products. The 5Ј-untranslated region contained an in-frame stop codon upstream of the methionine.
For expression studies, the entire coding sequence of ⑀xENaC was amplified from kidney cDNA by PCR with ExpandHighFidelity (Roche Applied Science, Mannheim, Germany). The PCR primers were ⑀xE-NaC-5Ј (5Ј-GCCGGATCCTTTATTATGGAGTCCACAG-3Ј) and ⑀xE-NaC-3Ј (5Ј-GCCAGATCTTAGGAAAAGGTATTTATTTCCCC-3Ј). Using terminal restriction endonuclease recognition sequences (BamHI and BglII), the PCR product was ligated in the oocyte expression vector pRSSP containing the 5Ј-untranslated region from Xenopus ␤-globin and a poly(A) tail. Several clones from two independent PCRs were sequenced to exclude PCR errors. Mutant cDNAs were constructed by recombinant PCR using standard techniques.
Electrophysiology and Data Analysis-0.1-10 ng of cRNA was injected into stage V-VI oocytes of X. laevis. Oocytes were kept in a low Na ϩ oocyte ringer-2 medium (10 mM NaCl, 72.5 mM N-methyl-D-glucamine, 1 mM Na 2 HPO 4 , 2.5 mM KCl, 5 mM HEPES, 0.5 g/liter polyvinylpyrrolidone, 1 mM MgCl 2 , and 1 mM CaCl 2 ) at 19°C for 1-5 days. Oocytes were superfused with a normal frog ringer solution containing 115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl 2 , and 10 mM HEPES (pH 7.3); and amiloride (50 M)-sensitive current was measured using two-electrode voltage clamp. Ionic selectivity was determined by replacing NaCl with LiCl or KCl. The holding potential was Ϫ60 mV. For determination of apparent amiloride affinity, oocytes were superfused with a solution containing 50 M amiloride. This solution was then replaced with solutions containing amiloride at 0 -10 M. Current values were subtracted from the current measured in the absence of amiloride to obtain the value of the inhibited current. Data were fitted for each experiment, and current values were normalized to the maximal value obtained from the respective fit. All experiments were performed at room temperature (22-26°C).
For determination of the time constant of self-inhibition, the oocytes were superfused with the bath solution at a rate of ϳ20 ml/min. Using a single exponential function, the time constant of self-inhibition of ␣xENaCand ⑀xENaC-containing channels was determined. This time constant was used to calculate the time at which self-inhibition was complete by Ͼ97%. This time was considered as the quasi-steady state.
Na ϩ affinity was determined for the initial current after amiloride washout (I max ) and for the current in the "steady state." For these long measurements, we had to use a slower solution flow rate, limiting the time resolution. Therefore, in these measurements, we considered 2 min as the steady state. Solutions containing different Na ϩ concentrations (140, 90, 35, 10, 3, and 0 mM), in which Na ϩ was replaced by equimolar amounts of N-methyl-D-glucamine, were used. To account for the channel rundown in these long measurements, current values were normalized to currents that were obtained under identical experimental conditions, but with an identical Na ϩ concentration (90 mM). Current values were then normalized to the current measured with 140 mM NaCl. Data were fitted for each experiment. The holding potential was Ϫ100 mV.
Apparent affinity for amiloride and Na ϩ was fitted to the following equation: is the amiloride or Na ϩ concentration, I max is the maximal current, a is the residual current, C 50 is the concentration at which half-maximal response occurs, and n is the Hill coefficient. C 50 values are expressed as mean Ϯ S.D. Error bars on Figs. 3 and 4 represent S.E. Statistical analysis was done with Student's t test.

RESULTS
Cloning of ⑀xENaC-We isolated a cDNA for a new ENaC subunit by homology cloning from X. laevis (see "Experimental Procedures"). It encodes a 604-amino acid protein that displays 42% identity to ␣xENaC and 29 -31% identity to ␤xENaC and ␥xENaC, respectively (Fig. 1A). Identity to ␦ENaC, which has been identified only in humans thus far and has properties similar to those of ␣ENaC (17), is 36%. ␤ 2 xENaC and ␥ 2 xENaC are Ͼ90% identical to ␤xENaC and ␥xENaC, respectively (14); ENaC orthologs from rat and Xenopus are 57-61% identical. Thus, the new subunit does not represent the isoform or species ortholog of an already known ENaC subunit (Fig. 1B); and therefore, we named it ⑀xENaC. ⑀xENaC has two predicted transmembrane domains similar to other ENaC subunits and a large loop between these domains, containing 16 conserved cysteines and seven potential glycosylation sites (Fig. 1A). A proline-rich motif (PPPXY) that is conserved in the carboxyl termini of ␣xENaC, ␤xENaC, and ␥xENaC is missing in ⑀xE-NaC (Fig. 1A). The tissue expression pattern of ⑀xENaC was investigated by reverse transcription-PCR, revealing predominant expression in kidney and bladder and faint expression in brain and skeletal muscle (Fig. 2).
Functional Characterization of ⑀xENaC-We investigated the electrophysiological characteristics of ⑀xENaC by functional expression in Xenopus oocytes. When injected alone in oocytes, ⑀xENaC elicited only small (Ͻ10 nA) amiloride-sensitive Na ϩ currents. Coexpression with either ␤xENaC or ␥xE-NaC increased the current amplitude, but only coexpression with both ␤xENaC and ␥xENaC resulted in full expression with amplitudes on the order of several A (Fig. 3A). This feature resembles ␣ENaC and ␦ENaC (2, 17) and identified ⑀xENaC as an ␣-like subunit, which efficiently co-assembles with ␤ENaC and ␥ENaC to form surface-expressed heteromeric channels. Substitution of extracellular Na ϩ with Li ϩ or K ϩ revealed that channels containing ⑀xENaC are highly selective for Na ϩ over K ϩ , with a Li ϩ Ն Na ϩ Ͼ Ͼ K ϩ permeability sequence, similar to ␣xENaC-containing channels (Fig. 3B). ⑀xENaC-containing channels were blocked by amiloride with an apparent IC 50 of 2.50 Ϯ 0.82 M (n ϭ 12), which is 10 times higher than the IC 50 for ␣xENaC-containing channels (0.22 Ϯ 0.01 M; p Ͻ Ͻ 0.05) (Fig. 3C). Together, ⑀xENaC-containing channels show the electrophysiological hallmarks of ENaCs: high selectivity for Na ϩ and block by amiloride at a low M concentration.
As illustrated in Fig. 4A, the time dependence of the current after washout of a saturating amiloride concentration (50 M) was very different for ␣␤␥xENaCand ⑀␤␥xENaC-expressing oocytes. For both channels, the current amplitude rose within seconds due to unblocking of the ion pore. However, whereas ␣␤␥xENaC-expressing oocytes showed a constant inward current with no strong time dependence over a time period of minutes, the inward current of ⑀␤␥xENaC-expressing oocytes strongly declined within a few seconds to a quasi-steady state (Fig. 4A). At this quasi-steady state (25 s after amiloride washout; see "Experimental Procedures"), the current amplitude was only 28 Ϯ 7% (n ϭ 10) of the peak amplitude (Table I). In addition to this fast current decline of ⑀␤␥xENaC-expressing oocytes, both ⑀␤␥ENaCand ␣␤␥ENaC-expressing oocytes displayed a linear "rundown" of channel activity on a time scale of minutes (Fig. 4A). The time course of this rundown was much slower than the initial fast decline of ⑀␤␥xENaC-expressing oocytes, and both processes can be clearly distinguished. The initial fast decline was well fitted to an exponential function with a time constant of 8.2 Ϯ 1.3 s (n ϭ 10).
We investigated the dependence of both the initial and steady-state currents of ⑀␤␥xENaC-expressing oocytes on ex-tracellular Na ϩ by equimolar replacement of increasing amounts of Na ϩ with N-methyl-D-glucamine. The initial current after amiloride washout saturated at ϳ100 mM Na ϩ with an apparent EC 50 of 15.3 Ϯ 5.3 mM (n ϭ 14), similar to ␣xE-NaC-containing channels (20.5 Ϯ 7.2 mM (n ϭ 8); p ϭ 0.03) (Fig.  4B). However, after the time-dependent decline in the current Transmembrane domains were predicted for ⑀xENaC with the TMpred program (available at www.ch.embnet.org/). GenBank TM /EBI accession numbers are AJ440222 for ⑀xENaC, U23535 for ␣xENaC, U25285 for ␤xENaC, and U25342 for ␥xENaC. B, evolutionary relationship between ENaC subunits and other family members (FaNaCh, the FMRFamide receptor from the snail Helix aspersa; ripped pocket (RPK) and pickpocket (PPK) from Drosophila; acid-sensing ion channels (ASIC) from rat (r); and DEG-1, MEC-4, and MEC-10, degenerins from Caenorhabditis). Highly divergent sequences at the N and C termini as well as in the proximal part of the extracellular loop were deleted, and the alignment and the tree for the phylogram were established by neighbor joining with ClustalX. The tree was then imported in TreeView and rooted with DEG-1, MEC-4, and MEC-10 as an outgroup. h, human. amplitude, the steady-state current saturated with an apparent EC 50 of 4.1 Ϯ 3.0 mM (n ϭ 14; p Ͻ Ͻ 0.05) (Fig. 4B). Thus, the time-dependent decline was seen only at high (Ͼ10 mM) Na ϩ concentration (Fig. 4B).
The rapid inhibition with high concentrations of extracellular Na ϩ was also observed after switching from an extracellular solution containing a low (1 mM) Na ϩ concentration to one containing a high concentration (115 mM). Initially, the current amplitude rapidly rose due to the strongly increased driving force for the inward Na ϩ current. But then, the current of ⑀␤␥xENaC-expressing oocytes relaxed by 70% with a time constant of 5.9 Ϯ 1.5 s to a quasi-steady-state value (n ϭ 10) (Fig.  4C), comparable to the behavior after amiloride washout. However, rapidly changing the extracellular Na ϩ concentration uncovered this fast inhibition also for ␣␤␥xENaC-expressing oocytes. Currents relaxed by ϳ40% with a time constant of 1.6 Ϯ 0.2 s (n ϭ 10) (Fig. 4C). Thus, inhibition of channels containing ␣ENaC is significantly faster than that of channels containing ⑀ENaC (p Ͻ Ͻ 0.05). The speed of inhibition of channels formed by ␣␤␥xENaC probably leads to an underestimation of the peak current. Therefore, the apparent ratio of steady-state to peak current may be overestimated for these channels.
Such a fast inhibition with high concentrations of extracellular Na ϩ is known as Na ϩ self-inhibition (8 -10). The characteristic feature of self-inhibition is that it is mediated by the extracellular Na ϩ ion itself and not by a rise in the intracellular Na ϩ concentration. However, due to the large current amplitude on the order of several A in our experiments, a rapid increase in the intracellular Na ϩ concentration was difficult to prevent. To show that a rise in the intracellular Na ϩ concentration is not responsible for the rapid current decline, we therefore used an experimental setup that allowed us to use different concentrations of extracellular Na ϩ leading to an inward current of comparable size. This was achieved by using a low extracellular Na ϩ concentration together with a large negative holding potential and a high extracellular Na ϩ concentration together with a small positive holding potential. These experiments were performed successively on the same ⑀␤␥xENaC-expressing oocyte. As shown in Fig. 4D, amiloride washout gave rise to an inward current in both cases. The initial amplitude of this current was 1.48 Ϯ 0.20-fold larger in the 2 mM Na ϩ solution (n ϭ 5; p ϭ 0.05), indicating a larger Na ϩ influx under this condition. However, despite the smaller Na ϩ influx, a rapid current decline was observed only in the 115 mM Na ϩ solution (the current amplitude 30 s after the peak was 43 Ϯ 3% of the peak amplitude (n ϭ 5); p Ͻ 0.05). In the 2 mM Na ϩ solution, no significant current decline was registered (the current amplitude 30 s after the peak was 101 Ϯ 2% of the peak amplitude (n ϭ 5); p ϭ 0.5) (Fig. 4D). These results strongly suggest that the rise in the extracellular (and not intracellular) Na ϩ concentration was the basis of the rapid current decline.
Moreover, we observed self-inhibition of a comparable degree for current amplitudes ranging from 1 to 50 A. Finally, it is well documented that feedback inhibition, which is due to a rise in the intracellular Na ϩ concentration, has a slower time course on the order of minutes (6). Therefore, feedback inhibition probably underlies the slow linear rundown described above, which follows Na ϩ self-inhibition. At least one mechanism of feedback inhibition is internalization of membraneexpressed channels. However, the fast current decline was partially but rapidly reversible after application of an amiloride analog (see below), most likely too fast for a substantial extrusion of the Na ϩ load. Together, our results suggest that the extracellular Na ϩ ion itself was responsible for the fast current decline seen in ⑀xENaC-expressing oocytes.
⑀xENaC and Na ϩ Self-inhibition taining 115 mM NaCl. After the current had relaxed to its steady state, we applied 1 mM BIG. Following an initial decrease in the current amplitude, the current amplitude increased rapidly (Fig. 4E). This behavior revealed that BIG had, in addition to its inhibitory effect, a dominant stimulatory effect on ⑀␤␥ENaC-containing channels (I steady-state /I peak ϭ 0.29 Ϯ 0.04 without BIG and 0.48 Ϯ 0.03 with BIG (n ϭ 8); p Ͻ Ͻ 0.05). After BIG washout, the current amplitude initially increased due to relief of blockade by BIG, but then decreased again to the value before BIG application (Fig. 4E). This behavior can be explained by reversible relief of self-inhibition by BIG. ␣␤␥ENaC-expressing oocytes showed a qualitatively similar reaction to BIG application (Fig. 4E). However, here the effect of BIG was mainly inhibitory (I steady-state /I peak ϭ 0.67 Ϯ 0.03 without BIG and 0.62 Ϯ 0.04 with BIG (n ϭ 6); p Ͻ 0.05). The relief of the rapid inhibition by BIG lends additional support to the interpretation that it reflects Na ϩ self-inhibition.
Identification of Molecular Determinants of Na ϩ Self-inhibition-Self-inhibition of ⑀xENaC-containing channels differed in two ways from that of ␣xENaC-containing channels. It was slower; however, a larger fraction of the channels apparently passed into the inhibited state. Assuming a simple two-state model (Scheme 1), with an active state (A) and an inhibited state (I) linked by an inactivation rate constant (k i ) and an activation rate constant (k a ), the rate constant k i would determine the speed of self-inhibition. The steady-state level that we measured as I steady-state /I peak depends on both k i and k a . Because, in ⑀xE-NaC-containing channels, the kinetics of self-inhibition changed inversely to the steady-state level, this would imply that the ⑀xENaC subunit changes k a as well as k i . Compared with the ␣xENaC subunit, the k i would be reduced, but the k a would be even more reduced, leading to the increased but slower inhibition.
To identify regions in the channel protein responsible for this differential behavior of ␣xENaC and ⑀xENaC, we constructed a set of chimeric channels with parts of ⑀xENaC exchanged with the corresponding parts of ␣xENaC. Self-inhibition of these channels was then examined by rapidly switching from solutions of low (1 mM) to high (115 mM) Na ϩ concentrations (Fig. 5 and Table I). As is shown in Table I, the kinetics of selfinhibition was significantly different (p Ͻ 0.05) between either the wild-type ⑀xENaC or ␣xENaC subunit and all of the chimeric channels. Channels in which only the N terminus was exchanged showed slow self-inhibition, as seen with wild-type ⑀xENaC subunits (C1, ϭ 4.07 Ϯ 1.42 s). However, all of the chimeras in which also the first transmembrane domain was exchanged showed self-inhibition that was significantly faster than the self-inhibition of even wild-type ␣xENaC-containing traces of inward currents after switching from a solution of low (1 mM) to high (115 mM) Na ϩ concentration. D, self-inhibition is not due to an increase in the intracellular Na ϩ concentration. Channels were activated by washout with 50 M amiloride. The extracellular Na ϩ concentration was either 2 mM with a holding potential of Ϫ120 mV (gray trace) or 115 mM with a holding potential of 10 mV (black trace). Experiments were performed successively on the same oocyte. For better comparison, traces have been overlaid. E, BIG relieves inhibition by extracellular Na ϩ . Oocytes were superfused with a solution containing 1 mM NaCl, and channels were then "activated" by fast wash-in of a solution containing 115 mM NaCl. 1 mM BIG was applied after the current had relaxed to the steady state.
⑀xENaC and Na ϩ Self-inhibition channels (C2-C5, ϭ 0.22 Ϯ 0.14 to 1.16 Ϯ 0.17 s) (Fig. 5 and Table I), suggesting unspecific effects in these chimeric constructs. Therefore, we constructed variants that retained both intracellular termini and both transmembrane domains from ⑀xENaC and in which only parts of the extracellular loop were exchanged with the corresponding parts of ␣xENaC (C3ex-C5ex) (Fig. 5). A variant in which approximately the first third of the extracellular loop was exchanged showed self-inhibition with an intermediate time course (C3ex, ϭ 2.69 Ϯ 1.05 s), whereas exchange of bigger parts led to self-inhibition that was TABLE I Time constants of self-inhibition and ratio of steady-state to peak current for wild-type and chimeric subunits Channels were activated either by switching from a solution containing 1 mM NaCl to one containing 115 mM NaCl or by washout of 50 M amiloride. The rise time of the current is also indicated. Note the slow rise time of ␣ENaC-containing channels after amiloride washout. All constructs were coexpressed with ␤xENaC and ␥xENaC. ND, not determined.

FIG. 5. Identification of molecular determinants of Na ؉ self-inhibition.
Shown is a scheme of wild-type and chimeric subunits and representative current traces for each construct. All constructs were coexpressed with ␤xENaC and ␥xENaC. Oocytes were superfused with a solution containing 1 mM NaCl, and channels were then activated by fast wash-in of a solution containing 115 mM NaCl. Vertical and horizontal bars correspond to 1 A and 30 s, respectively. In construct C1, amino acids up to (not including) Thr 49 of ⑀xENaC were replaced with the corresponding amino acids of ␣xENaC; in C2, up to Gln 68 ; in the C3 constructs, up to Pro 258 ; in the C4 constructs, up to Met 383 ; and in the C5 constructs, up to Leu 510 . similar to that of ␣xENaC-containing channels (C4ex, ϭ 1.79 Ϯ 0.67 s; and C5ex, ϭ 1.27 Ϯ 0.49 s). Altogether, these results suggest that crucial regions that determine the kinetic differences between ⑀xENaC and ␣xENaC reside in the extracellular loop.
In addition to the time course of self-inhibition, we analyzed the ratio of steady-state to peak current (I steady-state /I peak ) for the chimeric constructs (Table I). Interestingly, for the chimeric channel C2, which contained the cytoplasmic N terminus together with the first transmembrane domain of ␣xENaC, more of the peak current was inhibited at steady state (90%) than for ⑀xENaC. However, inhibition of this channel was faster than for ␣xENaC. This shows that a low I steady-state /I peak ratio can also be measured for rapidly inhibited channels, suggesting that the difference in I steady-state /I peak between ␣xE-NaC-and ⑀xENaC-containing channels is real. For all of the chimeras that contained part of the extracellular loop of ␣xE-NaC, I steady-state /I peak was high (ranging from 0.65 Ϯ 0.05 to 0.85 Ϯ 0.08) (Fig. 5 and Table I). This result suggests that the extracellular loop, especially the proximal third, determines not only the inactivation rate constant k i , but also the activation rate constant k a (see Scheme 1 above).
Finally, we addressed the molecular basis for the lower apparent amiloride affinity of ⑀xENaC-containing channels.
Amiloride is an open channel blocker, and amino acids at the beginning of the second transmembrane domain are involved in amiloride binding (20). Sequence alignment of all known ␣-like subunits (Fig. 6, upper) shows that there is only one amino acid at the beginning of the second transmembrane domain that is different in ⑀xENaC compared with all other ␣-like subunits: Trp 513 . Replacement of Trp 513 with the amino acid present in ␣xENaC confirmed that Trp 513 is responsible for the low apparent amiloride affinity of ⑀xENaC-containing channels. Channels containing the single amino acid exchange W513L showed a significantly increased apparent affinity for amiloride (IC 50 ϭ 0.38 Ϯ 0.33 M (n ϭ 11); p Ͻ Ͻ 0.05). In addition, these channels showed a slow current rising phase and no apparent self-inhibition after amiloride washout (Fig. 6, lower, left trace).
However, rapidly changing from a low to high Na ϩ concentration also uncovered self-inhibition for this mutant channel (right trace). Self-inhibition was as slow as in wild-type ⑀xE-NaC-containing channels ( ϭ 5.95 Ϯ 1.66 s (n ϭ 10); p ϭ 0.5), confirming that Trp 513 is at the basis of differential amiloride affinity, but not of differential speed of self-inhibition. DISCUSSION Variety of ENaCs in X. laevis-The electrophysiological characteristics of ENaCs have been characterized in different tissues from various species. These characteristics vary with respect to selectivity, amiloride affinity, and single channel conductance (21). Here, we have reported the cloning of ⑀xE-NaC, a new ENaC subunit from X. laevis. We have demonstrated that ⑀xENaC-containing channels show a similarly high Na ϩ /K ϩ selectivity as ␣xENaC-containing channels. Although we did not determine single channel conductance, most of the highly selective ENaCs have a single channel conductance of ϳ5 picosiemens (21). However, in addition to differences in Na ϩ self-inhibition (see below), ⑀xENaC-containing channels are characterized by a 10-fold lower apparent affinity for amiloride. This value matches best with the apparent K i (4.3 M) of an ENaC from Rana skin (22). Our results demonstrate that the lower apparent amiloride affinity is due to the presence of a unique tryptophan at the outer mouth of the ion pore. This finding confirms the amiloride-binding site identified in previous studies by site-directed mutagenesis (20,23).
⑀xENaC is already the sixth ENaC subunit identified in Xenopus. All subunits are expressed in kidney (13,14) and could, assuming that channels contain an ␣-like, a ␤, and a ␥ subunit, theoretically assemble in eight different combinations. Channels formed by only two different subunits would increase the variety of ENaCs in Xenopus even more. Why this remarkable diversity evolved is unclear. Mammalian ENaCs seem to be less variable. ␤ 2 ENaC and ␥ 2 ENaC subunits do not appear to be present in mammalian genomes. Moreover, the closest ⑀xENaC homolog in the draft version of the human genome is ␣ENaC, suggesting that there is no ⑀ENaC subunit in humans. Na ϩ Self-inhibition-Na ϩ self-inhibition had been best described in frog tissues (5). It has, however, also been indirectly shown for native ENaC in rat cortical collecting tubules (24). Here, an apparent Na ϩ affinity of 9 mM was measured for the whole cell Na ϩ current (I Na ), whereas the current through single channels (i Na ) saturates with an apparent K m of 48 mM (24). Because I Na ϭ (i Na )NP o , this implies that either the channel density (N) or the open probability (P o ) decreases as [Na ϩ ] o increases (24). Thus, Na ϩ self-inhibition seems to be a general characteristic of ENaCs. Strikingly, however, most of the studies characterizing recombinant ENaC in heterologous expression systems did not reveal self-inhibition. It was only very recently that Chraibi and Horisberger (15) convincingly showed that human ENaC, which is formed by ␣␤␥ subunits, shows self-inhibition when expressed in Xenopus oocytes. The self-inhibition was observable only after fast amiloride washout or a change in the extracellular Na ϩ concentration, and the current relaxed to a quasi-steady state with a time constant of 3 s. Thus, self-inhibition is fast and can be observed only with a high time resolution. At room temperature, ϳ40% of the initial current was inhibited at steady-state with human ENaC and ϳ20% with rat ENaC. For ␣␤␥xENaC, self-inhibition was too fast to be reliably determined (15). Due to the speed of self-inhibition, the time constants we determined should be taken only as estimates. Indeed, it seems that for all of the native channels, self-inhibition is faster than for ⑀xENaC-containing channels. The time course of self-inhibition has been reported in detail for ENaC from Rana skin and FIG. 6. Identification of the amino acid responsible for the low apparent amiloride affinity of ⑀xENaC. Upper, sequence alignment of the second transmembrane domains of different ENaC subunits; lower, representative current traces of the ⑀xENaC mutant W513L after amiloride washout (left) or after switching from a solution of low (1 mM) to high (115 mM) Na ϩ concentration (right). The substitutioncontaining ⑀ subunit was coexpressed with ␤xENaC and ␥xENaC. r, rat; h, human.