The Epithelial Na+ Channel Is Inhibited by a Peptide Derived from Proteolytic Processing of Its α Subunit*

Epithelial sodium channels (ENaCs) mediate Na+ entry across the apical membrane of high resistance epithelia that line the distal nephron, airway and alveoli, and distal colon. These channels are composed of three homologous subunits, termed α, β, and γ, which have intracellular amino and carboxyl termini and two membrane-spanning domains connected by large extracellular loops. Maturation of ENaC subunits involves furin-dependent cleavage of the extracellular loops at two sites within the α subunit and at a single site within the γ subunit. The α subunits must be cleaved twice, immediately following Arg-205 and Arg-231, in order for channels to be fully active. Channels lacking α subunit cleavage are inactive with a very low open probability. In contrast, channels lacking both α subunit cleavage and the tract αAsp-206-Arg-231 are active when expressed in oocytes, suggesting that αAsp-206-Arg-231 functions as an inhibitor that stabilizes the channel in the closed conformation. A synthetic 26-mer peptide (α-26), corresponding to αAsp-206-Arg-231, reversibly inhibits wild-type mouse ENaCs expressed in Xenopus oocytes, as well as endogenous Na+ channels expressed in either a mouse collecting duct cell line or primary cultures of human airway epithelial cells. The IC50 for amiloride block of ENaC was not affected by the presence of α-26, indicating that α-26 does not bind to or interact with the amiloride binding site. Substitution of Arg residues within α-26 with Glu, or substitution of Pro residues with Ala, significantly reduced the efficacy of α-26. The peptide inhibits ENaC by reducing channel open probability. Our results suggest that proteolysis of the α subunit activates ENaC by disassociating an inhibitory domain (αAsp-206-Arg-231) from its effector site within the channel complex.

Epithelial sodium channels (ENaCs) mediate Na ؉ entry across the apical membrane of high resistance epithelia that line the distal nephron, airway and alveoli, and distal colon. These channels are composed of three homologous subunits, termed ␣, ␤, and ␥, which have intracellular amino and carboxyl termini and two membranespanning domains connected by large extracellular loops. Maturation of ENaC subunits involves furin-dependent cleavage of the extracellular loops at two sites within the ␣ subunit and at a single site within the ␥ subunit. The ␣ subunits must be cleaved twice, immediately following Arg-205 and Arg-231, in order for channels to be fully active. Channels lacking ␣ subunit cleavage are inactive with a very low open probability. In contrast, channels lacking both ␣ subunit cleavage and the tract ␣Asp-206-Arg-231 are active when expressed in oocytes, suggesting that ␣Asp-206-Arg-231 functions as an inhibitor that stabilizes the channel in the closed conformation. A synthetic 26-mer peptide (␣-26), corresponding to ␣Asp-206-Arg-231, reversibly inhibits wild-type mouse ENaCs expressed in Xenopus oocytes, as well as endogenous Na ؉ channels expressed in either a mouse collecting duct cell line or primary cultures of human airway epithelial cells. The IC 50 for amiloride block of ENaC was not affected by the presence of ␣-26, indicating that ␣-26 does not bind to or interact with the amiloride binding site. Substitution of Arg residues within ␣-26 with Glu, or substitution of Pro residues with Ala, significantly reduced the efficacy of ␣-26. The peptide inhibits ENaC by reducing channel open probability. Our results suggest that proteolysis of the ␣ subunit activates ENaC by disassociating an inhibitory domain (␣Asp-206-Arg-231) from its effector site within the channel complex.
Epithelial sodium channels (ENaCs) 2 mediate Na ϩ entry across the apical membrane of high resistance epithelia, including the distal nephron, airway and alveolar epithelia, and distal colon. These channels are composed of three homologous subunits, termed ␣, ␤, and ␥, which have intracellular amino and carboxyl termini and two membranespanning domains connected by large extracellular loops (1). Residues preceding and within the second membrane-spanning domain constitute the channel pore (2)(3)(4)(5). ENaCs have a key role in the regulation of urinary Na ϩ reabsorption, extracellular fluid volume homeostasis, and control of blood pressure (6 -9). Epithelial Na ϩ channel gain-of-function mutations have been identified in patients with Liddle syndrome, a disorder characterized by volume expansion and hypertension (10 -15). In airway epithelia, ENaC has an important role in regulating the volume of airway surface liquids and mucociliary clearance. Increased ENaC activity is thought to contribute to poor mucociliary clearance observed in cystic fibrosis (16).
Maturation of ENaC subunits in Xenopus oocytes, Chinese hamster ovarycells,andMadin-Darbycaninekidney(MDCK)cellsinvolvesfurindependent proteolysis at two sites within the ␣ subunit (after Arg-205 and Arg-231) and at a single site within the ␥ subunit (after Arg-143) (17,18). Interestingly, both processed (i.e. cleaved) and unprocessed ENaCs are expressed on the cell surface (19). Proteolytic processing of ENaC likely occurs within the biosynthetic pathway as well as at the cell surface. Furin, a serine protease that resides primarily in the trans-Golgi network, is required both for cleavage and activation of ENaC in oocytes, Chinese hamster ovary, and MDCK cells (18). Other proteases, such as prostasin (also known as CAP-1) (20 -23), CAP-2 and CAP-3 (24), trypsin (18,20,25), and elastase (26), are also thought to have a role in the proteolytic processing and activation of ENaC. Proteolysis activates the channel by increasing channel open probability (25,27,28).
We recently reported that ␣ subunits lacking either one or both cleavage sites exhibited a marked enhancement of the Na ϩ self-inhibition response, suggesting that ␣ subunits must be cleaved at both furin consensus sites in order to activate the channel (28). We now report that ENaCs with mutant ␣ subunits lacking either one or both furin cleavage sites exhibited markedly reduced activity, confirming that cleavage at a single furin site in the ␣ subunit was not sufficient to activate ENaC. A mutant lacking both furin cleavage sites at Arg-205 and Arg-231, as well as the intervening tract Asp-206-Arg-231 within the ␣ subunit that is presumably excised following furin cleavage, was active when expressed in oocytes, suggesting that proteolytic processing of the ␣ subunit releases a 26-mer fragment that is inhibitory within non-cleaved channels. Addition of a synthetic version of this 26-mer fragment (␣-26) to the solution bathing Xenopus oocytes expressing ENaCs, or to the apical solution bathing monolayers of either a mammalian collecting duct cell line (mpkCCD c14 ) or primary cultures of human airway epithelial (HAE) cells, inhibited Na ϩ current in a dose-dependent manner. Our results suggest that multiple proteolytic cleavage events within the ␣ subunit are required to activate ENaC and that proteolytic activation of ENaC likely results through release of the inhibitory 26-mer peptide.

EXPERIMENTAL PROCEDURES
Oocyte Expression-cRNAs for ␣, ␤, and ␥ mouse ENaC (wild-type and mutant) subunits and mouse prostasin were synthesized with T3 or T7 mMessage mMachine TM (Ambion, Austin, TX). Subunits carrying furin cleavage mutations and the corresponding wild-type subunits have amino-terminal HA and carboxyl-terminal V5 tags (18). Stage V-VI Xenopus laevis oocytes were pretreated with 1.5 mg/ml of type IV collagenase and injected with 0.5-2 ng of cRNA/subunit. Injected oocytes were maintained as previously reported (29).
Two-electrode Voltage Clamp-Two-electrode voltage clamp (TEV) was performed as previously described (29). The extracellular solution (TEV solution) was (in mM): 110 NaCl, 2 KCl, 1.54 CaCl 2 , 10 HEPES, pH 7.4, or otherwise as indicated. The ENaC-mediated component of the whole cell Na ϩ current was determined by bath perfusion with TEV solution supplemented with 10 M amiloride. ENaC-mediated whole cell Na ϩ currents, at Ϫ60 mV, were defined as the amiloride-sensitive component of the current. Experiments with oocytes co-expressing ENaC and prostasin and the corresponding controls were performed in the presence of aprotinin (1 M) to impede any potential degradation of ␣-26 by prostasin. mpkCCD c14 Cell Culture-mpkCCD c14 cells were grown as previously reported (30). mpkCCD c14 cells were subcultured onto permeable filter supports coated with collagen (0.4-m pore size, 1-cm 2 surface area, Snapwell filters; Corning Inc., Acton, MA). Cells were kept on filters for at least 4 days in defined medium (30) that was changed every second day. At least 24 h prior to the experiments, the culture medium was replaced with a minimal medium (without drugs or hormones) that contained only Dulbecco's modified Eagle's medium and Ham's F-12.
Primary Cultures of HAE Cells-HAE cells were obtained from excess pathological tissue remaining after lung transplantation or from organ donor lungs deemed not suitable for transplantation under a protocol approved by the University of Pittsburgh Investigational Review Board. All cells were isolated from second through sixth generation bronchi and grown on permeable Transwell supports (Corning Inc.) as previously described (31).
Short Circuit Current Measurements-Cell culture inserts were mounted in modified Ussing chambers (Harvard Apparatus, Holliston, MA), and the monolayers were continuously short-circuited with a voltage/current clamp system (Physiologic Instruments, San Diego, CA). A 2-mV bipolar pulse was applied periodically to measure the resistance of the monolayer. As previously described by Bridges et al. (32), the bath solution contained (in mM) 120 NaCl, 25 NaHCO 3 , 3.3 KH 2 PO 4 , 0.8 K 2 HPO 4 , 1.2 MgCl 2 , 1.2 CaCl 2 , and 10 glucose. The pH of this solution was 7.4 when gassed with a mixture of 95% O 2 -5% CO 2 at 37°C.
Single Channel Studies-Bath and pipette solutions contained (in mM) 110 LiCl, 2 CaCl 2 , 2 KCl, and 10 Hepes, pH 7.4. Experiments were performed as previously described (33). Single channel currents were recorded in the cell-attached configuration of patch clamp, acquired at 5 kHz and filtered at 1 kHz by a 4-pole low pass Bessel filter. Single channel currents were further filtered at 100 -200 Hz with a Gaussian filter for display and analysis. Recordings were performed at an applied pipette potential of ϩ60 mV.
Western Blot-ENaC was immunoprecipitated from extracts of transiently transfected MDCK cells with goat anti-V5 antibody conjugated to agarose from Novus Biologicals (Littleton, CO) and immunoblotted for the ␣ subunit with mouse anti-V5 monoclonal antibody from Invitrogen as previously described (17).
Data and Statistical Analyses-IC 50 and mean time of inhibition are expressed as the mean with a 95% confidence interval (CI); otherwise, data were expressed as the mean Ϯ S.E. (n), where "n" equals the number of independent experiments analyzed. IC 50 was estimated from normalized currents plotted as a function of the peptide concentration fitted, as shown in Equation 1 where y is the response variable, X is the concentration of peptide, and IC 50 is the concentration of peptide that provokes a response half way between the baseline (t) and maximum response. The time course of inhibition in oocytes was estimated with the Equation 2, which described an exponential decay from a plateau where y is the response variable, p and m are the plateau and maximal responses, k is the rate constant, T 0 is the beginning of exponential decay, and T is time. For statistical analyses the normality and equality of the standard deviations of the data were tested. Based on these results a parametric or a nonparametric test was used. Fitting and statistical comparisons were performed with Sigmaplot 8.02 (SPSS Inc., Chicago, IL) and GraphPad 3.0 (GraphPad Software, San Diego, CA).

The Lack of Furin Cleavage Destabilizes the Open State of ENaC-We
previously reported that the ␣ subunit of ENaC is cleaved twice, immediately following two furin consensus cleavage sites, at Arg-205 and Arg-231 (18). A third potential furin consensus cleavage site at Arg-208 within the ␣ subunit is not processed by furin (18). The introduction of Ala mutations at all three Arg in the ␣ subunit (␣R205A,R208A,R231A (␣RtripleA)) prevented ␣ subunit cleavage and significantly reduced the expression of ENaC whole cell currents in Xenopus oocytes (18). Whole cell currents measured in oocytes expressing wild-type or furin-insensitive mutant channels (␣RtripleA␤␥) increased to similar levels in response to external trypsin, suggesting that ENaC subunit proteolysis increases channel open probability rather than channel number (18) Previous studies have shown that ENaC gating is modulated, in part, via a regulatory domain that has been localized to the region preceding the second membrane-spanning (pre-M2) domain (5,34). Channels with a specific mutation at a key site (the degenerin site) within the pre-M2 region of the ␤ subunit (␤S518K) have a very high open probability (35,36), whereas channels that lack proteolysis of the ␣ subunit have a very low open probability (28). We used these mutants to examine whether the lack of proteolysis, or the introduction of a mutation at the degenerin site, exerted a dominant effect on ENaC activity when expressed in oocytes. Oocytes were co-injected with either (i) wild-type ␣␤␥, (ii) wild-type ␤ with ␣ (␣RtripleA) and ␥ (␥R143A) subunits carrying mutations in the furin cleavage sites, (iii) wild-type ␣ and ␥ with a ␤ subunit degenerin mutation (␤S518K), or (iv) three mutant subunits (␣RtripleA␤S518K␥R143A). We found that oocytes expressing ␣RtripleA␤S518K␥R143A exhib-ited whole cell Na ϩ currents that were similar in magnitude to currents measured in oocytes expressing wild-type ␣␤␥ ( p Ͼ 0.05). These currents were significantly greater than the currents measured in oocytes expressing ␣RtripleA␤␥R143A ( p Ͻ 0.001) and significantly less than currents measured in oocytes expressing ␣␤S518K␥ ( p Ͻ0.05, n ϭ 24 -26) (Kruskal-Wallis test (non-parametric ANOVA) followed by Dunn's multiple comparisons test) (Fig. 1A). At the single channel level, ␣RtripleA␤S518K␥R143A channels had a single channel conductance of 6.2 Ϯ 0.1 pS (n ϭ 5-6) similar to what we described previously for wild-type ␣␤␥ (33). However, analysis of ␣RtripleA␤S518K␥R143A at a single channel level showed frequent short openings and closures (Fig. 1C ), whereas the ␣␤S518K␥ mutant had a high open probability with long openings and brief closures (Fig. 1B). These data provide additional evidence in support of a major role for proteolytic processing of ENaC extracellular domains in the regulation of channel gating.
Proteolytic Activation of ENaC Requires Cleavage at Multiple Sites within the ␣ Subunit-As there are two furin cleavage sites within the ␣ subunit, we examined whether the presence of a single furin cleavage site was sufficient for activation of ENaC. Xenopus oocytes expressing ENaCs with mutations at the proximal (␣R205A␤␥), distal (␣R231A␤␥), or at both furin cleavages sites (␣R205A/R208A/ R231A␤␥ (␣RtripleA)) exhibited significantly reduced amiloride-sensitive currents when compared with oocytes expressing wild-type channels ( Fig. 2A). Subsequent treatment of these oocytes expressing either wild-type or mutant ENaCs with trypsin (2 g/ml) increased amiloridesensitive currents to similar levels ( Fig. 2A), suggesting that similar numbers of channels were present on the plasma membrane in each case. These data support our previous finding that the ␣ subunit must be Experiments were performed 20 -24 h after injection of oocytes with cRNAs for wild-type or mutant ENaCs. The ␣ and ␥ subunits had amino-(HA) and carboxyl-(V5) terminal epitope tags. A, mutation of furin cleavage sites in both the ␣ and ␥ subunits reduced the activity of channels with a degenerin site mutation (␤S518K). Whole cell ENaC currents were measured at Ϫ60 mV in oocytes expressing either ␣␤␥, ␣RtripleA␤␥R143A, ␣␤S518K␥, or ␣RtripleA␤S518K␥R143A. Significant differences were observed between ␣␤␥ controls versus ␣RtripleA␤␥R143A or ␣␤S518K␥ ( p Ͻ 0.001). Significant differences were also observed between ␣RtripleA␤S518K␥R143A versus ␣RtripleA␤␥R143A ( p Ͻ 0.001) and ␣␤S518K␥ ( p Ͻ 0.05, Kruskal Wallis test (nonparametric ANOVA) followed by Dunn's multiple comparisons post-test). Experiments were performed with 24 -26 oocytes for each group. B and C, mutations of furin consensus cleavage sites in the ␣ and ␥ subunits affect the gating of ENaCs with a degenerin site mutation. Single channel tracings were obtained in the cell-attached mode as described under "Experimental Procedures." The closed state is indicated by "C". Recordings were performed at an applied pipette potential of ϩ60 mV. Upper tracings show representative recordings of ␣␤S518K␥ (n ϭ 11) (B) or ␣RtripleA␤S518K␥R143A (n ϭ 5) (C ) channels. Normalized amplitude histograms are presented at the right side of each recording. A, cleavage at both furin consensus sites within the ␣ subunit of ENaC is required for expression of active channels. ENaC currents were recorded before and 5 min following treatment with 2 g/ml of trypsin (closed bars). Experiments were performed with a solution containing (in mM) 100 Na ϩ gluconate, 1.54 CaCl 2 , 5 BaCl 2 , 10 tetraethylammonium chloride, 10 Hepes, pH 7.4. A gray bar indicates statistically significant differences in amiloride-sensitive currents between ␣␤␥ control and ␣R205A␤␥, ␣R231A␤␥, or ␣RtripleA␤␥ ( p Ͻ 0.001, Kruskal Wallis test (nonparametric ANOVA) followed by Dunn's multiple comparisons post-test). Amiloride-sensitive currents following treatment with trypsin were not significantly different between the four groups (black bars). Experiments were performed with 15-18 oocytes for each group. B, ENaCs lacking ␣Asp-206-Arg-231 are active in the presence or absence of furin cleavage. Whole cell currents in oocytes expressing ␣⌬206 -231␤␥ were similar to currents measured in oocytes expressing wild-type ␣␤␥ ( p Ͼ 0.05). Whole cell currents were statistically different between ␣␤␥ control and ␣RtripleA␤␥ ( p Ͻ 0.01) and ␣R205A,⌬206 -231␤␥ ( p Ͻ 0.05, Kruskal Wallis test (nonparametric ANOVA) followed by Dunn's multiple comparisons post-test). Experiments were performed with 15 oocytes for each group. C, characterization of ␣ subunit processing. ENaC was immunoprecipitated with anti-V5 antibodies from extracts of MDCK cells transiently expressing green fluorescent protein control (GFP), wild-type ␣␤␥, ␣⌬206 -231␤␥, ␣R205A,⌬206 -231␤␥, or ␣RtripleA␤␥ and immunoblotted for the carboxyl-terminal V5-tagged ␣ subunit in each case. Numbers to the left of the gel represent mobility of Bio-Rad molecular weight markers. cleaved twice during ENaC maturation for channels to exhibit normal activity (28).
The mechanism by which proteolysis increases ENaC open probability has not been elucidated. It is possible that non-cleaved subunits have an overall constrained structure that increases residency time in the closed state and/or reduces residency time in the open state. Alternatively, as the ␣ subunit must be cleaved twice during ENaC maturation for channels to exhibit normal activity, the tract between both putative furin cleavages sites might be functioning as an inhibitor that acts by stabilizing the closed conformation of the channel. To discriminate between these two possibilities, we deleted the tract of amino acids that are putatively excised from the ␣ subunit following furin processing (Asp-206-Arg-231). This mutant (␣⌬206 -231) still contains the first furin consensus site after Arg-205 and should be cleaved once. In addition, we generated an ␣ subunit mutant with both the R205A mutation and deletion of the tract Asp-206-Arg-231 (R205A,⌬206 -231) that should lack furin-dependent processing. ENaC currents were not significantly affected by the ⌬206 -231 mutation in the ␣ subunit ( p Ͼ 0.05) and were even modestly increased in the absence of ␣ subunit furin cleavage (␣R205A,⌬206 -231 mutant) versus wild-type ␣␤␥ ( p Ͻ .005, Kruskal-Wallis test (nonparametric ANOVA) followed by Dunn's multiple comparisons test). When co-expressed in MDCK cells with ␤ and ␥ subunits, ␣⌬206 -231 migrated slightly faster on SDS gels than wildtype ␣ subunit as expected with the deletion of 26 amino acid residues (Fig. 2C ), whereas both wild-type ␣␤␥ and ␣⌬206 -231␤␥ exhibited evidence of ␣ subunit cleavage (i.e. appearance of a carboxyl-terminal polypeptide of 65 kDa) consistent with previously described furin-dependent processing (17,18). In contrast, no evidence of ␣ subunit cleavage was observed when ␣R205A,⌬206 -231␤␥ was expressed in MDCK cells (Fig. 2C ). Thus, in channels lacking the ␣ subunit tract Asp-206-Arg-231, proteolytic processing by furin is not required to activate the channel, indicating that this tract may function as an inhibitor that stabilizes the channel in the closed conformation. Our data suggest that a double cleavage event is required to dissociate a putative blocker (tract ␣Asp-206-Arg-231) from the channel. In the absence of this tract, the ␣ subunit does not need to be cleaved for the channel to display normal activity.

ENaC Is Inhibited by a Short Peptide Derived from Its Proteolytic
Processing-As consequence of proteolytic processing of the ␣ subunit, a peptide of 26 residues, corresponding to Asp-206-Arg-231, is predicted to be cleaved from the ␣ subunit. We synthesized this 26-mer peptide (␣-26) and examined whether it altered channel activity. Wildtype ␣␤␥ expressed in oocytes were inhibited by ␣-26 with an IC 50 of 2.8 ϫ 10 Ϫ6 M (CI, 2.5 to 3.2 ϫ 10 Ϫ6 M) (Fig. 3A), but not by a control peptide with the sequence scrambled (Fig. 3B, SCR). IC 50 s determined at holding potentials of Ϫ20, Ϫ60, and Ϫ100 mV were 2.9 ϫ 10 Ϫ6 M (CI, 2.5 to 3.3 ϫ 10 Ϫ6 M), 2.8 ϫ 10 Ϫ6 M (CI, 2.5 to 3.2 ϫ 10 Ϫ6 M) and 2.9 ϫ 10 Ϫ6 M (CI, 2.6 to 3.4 ϫ 10 Ϫ6 M), respectively, indicating that the block of ENaC was not voltage dependent. The time required for 1 M ␣-26 to achieve half maximal inhibition was 42 s (CI, 36.8 -48.8 s) (Fig. 3C ), suggesting that ␣-26 has a slow association rate when compared with amiloride, a prototypic voltage-dependent channel blocker that achieves half maximal inhibition within seconds (37). ␣-26 was a reversible inhibitor of the channel (Fig. 3D). The IC 50 for amiloride block of ENaC was not affected by the presence of ␣-26. The amiloride IC 50 was 152 nM (CI, in the presence of 2.5 M ␣-26 and 150 nM (CI, 136 -165 nM) in the absence of peptide (Fig. 3E), suggesting that ␣-26 does not bind to or interact with the amiloride binding site. ␣-26 is rich in Pro and Arg residues. Substitution of all Pro with Ala, or alternatively all Arg with Glu, led to a significant reduction in the block of ENaC by 10 M peptide (Fig. 3B). Channels that lack furin-dependent cleavage of the ␣ subunit (␣RtripleA␤␥), as well as channels lacking both furin-dependent cleavage of the ␣ subunit and the tract ␣Asp-206-Arg-231 (i.e. ␣R205A,⌬206 -231␤␥) were not blocked by 1 M ␣-26 (Fig. 3F), indicating that the peptide does not have access to its effector binding site within the channel in the absence of ␣ subunit cleavage.
Single channel analyses using a patch pipette backfilled with ␣-26 (10 M) demonstrated a reduction in the product of the number of channels and open probability (NP o ) compared with controls (Fig. 4). For experiments performed with ␣-26, the tip of the pipette was filled with standard patch solution. Pipette resistances were similar in control (7.8 Ϯ 0.4 M⍀) and ␣-26-treated patches (6.9 Ϯ 0.6 M⍀) ( p ϭ 0.30, unpaired t test). Control patches that did not contain ␣-26 in the pipette remained active over the recording period (Fig. 4B). Unitary currents, at an applied pipette potential of ϩ60 mV, were similar in both control (0.42 Ϯ 0.2 pA, n ϭ 5) and ␣-26-containing patches (0.39 Ϯ 0.2 pA, n ϭ 7) ( p ϭ 0.32, unpaired t test). Fig. 4C shows a representative tracing recorded with ␣-26 in the pipette. At the beginning of the experiment open times were similar to that observed in controls (Fig. 4B). Over the course of the experiment openings decreased and were shorter in duration compared with control until channels were fully closed. Assuming no changes in the number of channels during the recording, the changes observed in NP o due to the presence of ␣-26 in the pipette largely reflect a change in channel P o .
Prostasin Reduces the Efficacy of ␣-26-The differences in the IC 50 s of ␣-26 for ENaCs expressed in oocytes versus HAE and mpkCCD c14 monolayers may represent differential proteolytic processing of   6 -8). G, amiloride-sensitive currents were recorded in oocytes expressing wild-type ENaC with or without prostasin, before and after 3 min of perfusion with ␣-26. Significant differences in the current response to ␣-26 between oocytes expressing wild-type ENaC and co-expressing ENaC and prostasin were observed at 1 M ( p Ͻ 0.001, n ϭ 10 -11) and 10 M ( p Ͻ 0.05, n ϭ 7) (Kruskal-Wallis test (nonparametric ANOVA) followed by Dunn's multiple comparisons test).
ENaC subunits. ENaC is cleaved and activated by furin in Xenopus oocytes and MDCK and Chinese hamster ovary cells (18), whereas other serine proteases, in addition to furin, are likely involved in the proteolytic processing and activation of ENaC in other mammalian epithelial cells (32,38). The mechanisms for ENaC activation by proteases other than furin are unknown. Kunitz-type protease inhibitors that do not block furin activity reduced amiloride-sensitive currents in some cultured epithelial cells (20,32,38). Prostasin is a serine protease that is blocked by Kunitz-type protease inhibitors and activates ENaC when co-expressed in oocytes (20). cRNAs for wild-type ENaC subunits (0.5 ng/subunit) were co-injected with or without prostasin cRNA (1 ng). Co-expression of prostasin and ENaC in oocytes led to an increase in amiloride-sensitive currents (Ϫ9.6 Ϯ 1.3 A, n ϭ 28) compared with oocytes expressing wildtype ENaC alone (Ϫ5.3 Ϯ 0.6 A, n ϭ 26) ( p Ͻ 0.005, unpaired t test with Welch correction). Interestingly, co-expression of prostasin and ENaC significantly reduced the efficacy of ␣-26-dependent inhibition of ENaC (Fig. 5G).

DISCUSSION
Proteolytic processing of ENaC subunits has a key role in the regulation of channel gating. Caldwell et al. showed that ENaCs with a low P o and brief mean open times were converted by external trypsin or elastase to channels that had characteristically long open and closed times on the order of seconds (25,26). We have shown that channels with non-cleaved ␣ subunits, or with ␣ subunits with a mutation of even a single furin cleavage site, have significantly reduced activity ( Fig. 2 and Ref. 28) that reflects enhanced inhibition of the channel by external Na ϩ (28). The degenerin mutation (␣S518K) and the furin site mutations (␣RtripleA and ␥R143A) exert opposing effects on ENaC gating. Channels with both mutations (i.e. ␣RtripleA␤S518K␥R143A) flicker between open and closed states (Fig. 1), suggesting that the tract ␣Asp-206-Arg-231 influences channel gating, at least in part, by destabilizing the open conformation of ENaC or stabilizing the closed conformation. Channels that lacked both ␣ subunit furin cleavage sites as well as the tract between these sites (␣R205A,⌬Asp206-Arg231) exhibited activity similar to wild type, suggesting that the tract ␣Asp-206-Arg-231 is a channel blocker. The corresponding synthetic peptide ␣-26 inhibited wild-type ENaCs in a concentration-dependent and reversible manner, whereas channels with mutations that prevent cleavage of the ␣ subunit were not blocked by ␣-26, indicating that the binding site for ␣-26 is exposed by furin cleavage. Higher concentrations of ␣-26 were required to block channels co-expressed in oocytes with prostasin or channels expressed in HAE or mpkCCD c14 cells. These results suggest that differential proteolytic processing of ENaC subunits has an important role in determining the efficacy of ␣-26 as a channel blocker. Substitution of Arg residues within the ␣ peptide with Glu, or substitution of Pro residues with Ala, significantly reduced the affinity of block by ␣-26, indicating that electrostatic interactions and structural constraints have important roles in ␣-26 interaction with ENaC. The IC 50 of amiloride was not altered in the presence of ␣-26. Amiloride is proposed to bind to ENaC at a site adjacent to and including the selectivity filter of the channel (2, 3, 39 -41). Although amiloride block of ENaC is voltage dependent, ␣-26 block is not voltage dependent. These results suggest that amiloride and ␣-26 bind at different sites within ENaC.
ENaCs have a key role in the regulation of urinary Na ϩ reabsorption, extracellular fluid volume homeostasis, and control of blood pressure (6 -8,42). Increased ENaC-dependent Na ϩ transport across airway epithelia is thought to contribute to the poor mucociliary clearance observed in cystic fibrosis (43). Overexpression of the mouse ENaC ␤ subunit in mouse airway produced a phenotype similar to cystic fibrosis with airway surface liquid volume depletion, mucus obstruction, goblet cell metaplasia, neutrophil inflammation, and poor bacterial clearance (43). It has been proposed that inhibition of airway ENaC activity may be of therapeutic benefit in cystic fibrosis. Amiloride is a relative high affinity ENaC inhibitor and is used clinically as a pharmacologic inhibitor of renal Na ϩ channels and a potassium-sparing diuretic. Inhaled amiloride is rapidly cleared from airways and has been ineffective in the treatment of cystic fibrosis lung disease (16,44). Although ␣-26 is a relatively low affinity blocker of ENaC in HAE cells, analysis of ␣-26 derivatives may provide a basis to design higher affinity blockers of ENaC that could provide an alternative to amiloride and related compounds as inhibitors of airway Na ϩ channels.