Novel Determinants of Epithelial Sodium Channel Gating within Extracellular Thumb Domains*

Activity of the epithelial Na+ channel (ENaC) is modulated by Na+ self-inhibition, an allosteric down-regulation of channel open probability by extracellular Na+. We searched for determinants of Na+ self-inhibition by analyzing changes in this inhibitory response resulting from specific mutations within the extracellular domains of mouse ENaC subunits. Mutations at γMet438 altered the Na+ self-inhibition response in a substitution-specific manner. Fourteen substitutions (Ala, Arg, Asp, Cys, Gln, Glu, His, Ile, Phe, Pro, Ser, Thr, Tyr, and Val) significantly suppressed Na+ self-inhibition, whereas three mutations (Asn, Gly, and Leu) moderately enhanced the inhibition. Met to Lys mutation did not alter Na+ self-inhibition. Mutations at the homologous site in the α subunit (G481A, G481C, and G481M) dramatically increased the magnitude and speed of Na+ self-inhibition. Mutations at the homologous βAla422 resulted in minimal or no change in Na+ self-inhibition. Low, high, and intermediate open probabilities were observed in oocytes expressing αG481Mβγ, αβγM438V, and αG481M/βγM438V, respectively. This pair of residues map to theα5 helix in the extracellular thumb domain in the chicken acid sensing ion channel 1 structure. Both residues likely reside near the channel surface because both αG481Cβγ and αβγM438C channels were inhibited by an externally applied and membrane-impermeant sulfhydryl reagent. Our results demonstrate that αGly481 and γMet438 are functional determinants of Na+ self-inhibition and of ENaC gating and suggest that the thumb domain contributes to the channel gating machinery.

Maintenance of body fluid volume homeostasis requires a collaborative interaction of many Na ϩ transport mechanisms. Na ϩ transport in epithelia that line the late distal convoluted tubule, connecting tubule, and collecting tubule relies on apical Na ϩ entry through epithelial Na ϩ channels (ENaC). 4 Gain-of-function and loss-of-function ENaC mutations in humans result in Liddle syndrome and pseudohypoaldosteronism type I, respectively (1). ENaC activity is regulated by a variety of factors that primarily alter channel density at plasma membrane, open probability (P o ), or both. ENaC gating is characterized by long open and close time in the range of up to minutes and a highly variable P o , although channels with brief open (on the order of tens of ms) and long close times have also been observed (2,3). Although previous studies have identified several regions that appear to have a role in ENaC gating, the mechanism of ENaC gating is not well understood. It is now well established that ENaC undergoes a reduction in P o as a result of Na ϩ binding to extracellular component(s) within the channel complex, a process referred to as Na ϩ self-inhibition (4 -6). Recent studies have revealed specific characteristics and have identified several sites that have a functional role in Na ϩ self-inhibition (4, 6 -8). However, detailed elements regarding its mechanism have not been revealed.
A logical place to search for structural elements associated with Na ϩ self-inhibition is the large extracellular domain (ECD) that connects the two transmembrane domains (M1 and M2) within each ENaC subunit. The ECD likely exists as well structured subdomains with 16 conserved Cys residues. We recently reported that point mutations at multiple ␣ and ␥ ECD Cys residues blunted Na ϩ self-inhibition, and certain double or triple mutations rendered ENaC insensitive to high concentration of extracellular Na ϩ . These results suggest that multiple Cys residues are required to establish the proper tertiary structure permitting this allosteric regulation (9). In addition, the N-terminal portion of ECD contains ␥His 239 , a previously identified residue critical for Na ϩ self-inhibition, as well as defined protease cleavage sites (4, 10 -12). Various proteases have been shown to regulate ENaC activity, in part, by interfering with Na ϩ self-inhibition (6,7,13).
The resolved high resolution structure of the chicken acidsensing ion channel 1 (cASIC1) revealed an extracellular domain with a highly organized structure characterized by ␤ sheets within a core embedded within peripheral ␣ helices (14). These helices form the finger, thumb, and knuckle domains, whereas two bundles of long and short ␤-sheets occupy the palm and ␤-ball domains. Given the similarities in amino acid sequences and biophysical properties among ASIC and other members of ENaC/degenerin family, it is very likely that ENaC subunits adopt a similar overall tertiary structure, although considerable differences likely exist within local structures, particularly for poorly conserved regions that likely contribute to differences in the biophysical properties of ASIC and ENaC. Additional evidence supporting the notion that ECDs of ASIC and ENaC share a similar overall structure is that the disulfide bridges proposed for the ECD of ␣ ENaC, based on mutational analyses and chemical modifications, match the disulfide bonds within the cASIC1 structure (9,14).
In searching for additional determinants for Na ϩ self-inhibition and ENaC gating, we observed a significant suppression of this inhibitory response by a mutation of a Met residue (M438A) within the second cysteine-rich domain in ECD of ␥ENaC. The ␥Met 438 is near the 13th conserved Cys (␥Cys 440 ) whose mutation leads to significant suppression of Na ϩ selfinhibition (9). Sequence alignments placed this residue at a site corresponding to Lys 342 within cASIC1, which is located at the N-terminal part of helix ␣5, one of the two helices that comprise the thumb domain (Fig. 1). The thumb domain within cASIC1 was proposed to interact with other parts of the extracellular domains during proton gating of the channel (14). In this study we performed mutagenesis analyses to assess the functional role of ␥Met 438 and its homologous residues in ␣ and ␤ ENaC in Na ϩ self-inhibition.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis and in Vitro Transcription-Mouse ␣, ␤, and ␥ ENaC cDNAs in pBluescript SK-vector (Stratagene, La Jolla, CA) were used as templates to generate point mutations using a PCR-based method as previously described (15). Target mutations were confirmed by direct sequencing. The complementary RNAs (cRNAs) for wild type (WT) and mutant ENaC subunits were synthesized with T3 RNA polymerase (Ambion, Inc.), purified with an RNA purification kit (Qiagen Inc., Valencia, CA), and quantitated by spectrophotometry and density analyses of the RNA band in a denaturing agarose gel.
Procedures for Observing Na ϩ Self-inhibition-To examine Na ϩ self-inhibition, a low Na ϩ bath solution (NaCl-1; containing 1 mM NaCl, 109 mM N-methyl-D-glucamine, 2 mM KCl, 2 mM CaCl 2 , 10 mM HEPES, pH 7.4) was replaced rapidly by a high Na ϩ bath solution (NaCl-110; containing 110 mM NaCl, 2 mM KCl, 2 mM CaCl 2 , 10 mM HEPES, pH 7.4), whereas the oocytes were continuously clamped to Ϫ60 or Ϫ100 mV as indicated. Bath solution exchange was performed with a 6-channel Teflon valve perfusion system from Warner Instruments. At the end of the experiment, 10 M amiloride was added to the bath to determine the amiloride-insensitive component of the whole cell current. Currents remaining in the presence of 10 M amiloride were generally less than 200 nA. The results from oocytes that showed unusually large amiloride-insensitive currents (Ͼ5% of total currents) were discarded to minimize current contamination from endogenous channels and membrane leak. Given the well known variability of Xenopus oocyte expression system and our initial observation that the Na ϩ self-inhibition responses of WT ␣␤␥ mENaCs varied moderately among different batches of oocytes, despite a very small variation in the response within the same batch of oocytes, the responses of WT channels were always examined with mutants in the same batch of oocytes in an alternating manner.
The first 40 s of current decay was fit with an exponential equation by Clampfit 9.0 (Axon Instruments Inc.). The peak current (I peak ) was the measured maximal inward current immediately after bath solution exchange from low Na ϩ to high Na ϩ concentration. The steady state current (I ss ) represented the measured current at 40 s post I peak . The current ratio of I ss /I peak was calculated from amiloride-sensitive I ss and I peak obtained by subtracting amiloride-insensitive currents from I ss and I peak .
Analysis of Na ϩ Binding Affinity-To estimate the Michaelis constant (K m ) for the Na ϩ concentration-current relationship, both I peak and I ss were measured in the same cell after the bath Na ϩ concentration was raised from 1 to 3, 10, 30, 60, 90, or 110 mM. I peak and I ss were plotted against [Na ϩ ]. K m and V max (maximal current) were obtained by best fit of the current-concentration data according to the following equation, with least squares nonlinear curve fitting using Origin Pro 7.5 (OriginLab Corporation, Northampton, MA): I ϭ V max ⅐ C/(C ϩ K m ). In the equation, I is the relative I peak or I ss , and C refers to the [Na ϩ ] used to initiate self-inhibition. The apparent inhibitory constant (K i ) of Na ϩ self-inhibition was estimated by fitting I ss and [Na ϩ ] data with an equation for substrate inhibition of enzymes (16): Single Channel Recordings-The oocytes were placed in a hypertonic solution (NaCl-110 supplemented with 200 mM sucrose) for 5 min, and the vitelline membranes were removed manually. The oocytes were then transferred to a recording chamber with NaCl-110 and maintained for at least 20 min at room temperature (22-25°C) before recordings. Pipette solution was the same as bath solution (NaCl-110). Patch pipettes with a tip resistance of 5-10 M⍀ were used. Patch clamp in the cell-attached configuration was performed using PC-One Patch Clamp amplifier (Dagan Corp.) and a DigiData 1322A interface connected to a PC. The patches were clamped at membrane potentials (negative value of pipette potentials) of Ϫ40 to Ϫ100 mV. Software pClamp 8 or 10 (Molecular Devices Corporation) was used for data acquisition and analyses. Single channel recordings were acquired at 5 kHz, filtered at 300 Hz by a 4-pole low pass Bessel Filter built in the amplifier, and stored on the hard disk. Open probability was estimated by single channel search function of pClamp 10 from recordings that contained no more than three current levels at a clamping membrane potential of Ϫ100 mV and lasted for at least 2 min. The voltages were not corrected by the junction potential.
Chemicals-2-(Trimethylammonium) ethyl methanethiosulfonate bromide (MTSET) was from Toronto Research Chemicals Inc. All of the other reagents were from Sigma. MTSET was aliquoted in powder and dissolved in bath solution immediately prior to application to avoid hydrolysis.
Statistical Analysis-The data are presented as the means Ϯ S.E. Significance comparisons between groups were performed with Student's t tests. A p value of less than 0.05 was considered statistically different.

RESULTS
Residue ␥Met 438 is not well conserved in ENaC subunits or other ENaC/degenerin members, although its neighboring residues are conserved at a similar degree to the overall homology among ENaC subunits (ϳ35%; Fig. 1B). Its corresponding residues are Gly in ␣, Ala in ␤, and Lys in most ASIC isoforms. The region harboring ␥Met 438 aligns to a helix (␣5) within the thumb domain of cASIC1 (Fig. 1A), where its homologous residue is Lys 342 (Fig. 1A). We examined the potential role of ␥Met 438 and its counterparts in the ␣ and ␤ subunits in Na ϩ self-inhibition and ENaC gating by analyzing ␣␤␥ mouse ENaCs with introduced point mutations.
Mutations at ␥Met 438 Alter the Na ϩ Self-inhibition Response in a Residue-dependent Manner-␥Met 438 was mutated to 18 different amino acids. Co-expression of all mutant mouse ␥ENaC subunits with WT mouse ␣ and ␤ subunits in oocytes resulted in measurable amiloride-sensitive currents permitting examination of Na ϩ self-inhibition. As shown in Figs. 2 and 3, most substitutions of ␥Met 438 (Gln, Arg, Glu, Ile, Ser, Cys, Asp, Tyr, Phe, Thr, Pro, Ala, and Val) greatly suppressed Na ϩ self-inhibition as determined by the magnitude (I ss / I peak ) and rate (tau) of this process (p Ͻ 0.01). A His residue at this site slightly reduced the magnitude of Na ϩ self-inhibition (p Ͻ 0.05), whereas a Leu slightly enhanced the inhibition (p Ͻ 0.05). Two mutations (␥M438G and ␥M438N) significantly enhanced Na ϩ self-inhibition (p Ͻ 0.01). The only mutation  that did not significantly alter the speed and magnitude of Na ϩ self-inhibition was ␥M438K. The data from 19 different channels offered an opportunity to identify residue factors dictating the degree of Na ϩ self-inhibition using correlation analysis. A plot of I ss /I peak against side chain lengths revealed a linear relationship ( Fig. 4A; r ϭ Ϫ0.58, p ϭ 0.01, n ϭ 19). Excluding Gly that lacks a side chain gave a slightly better fit (r ϭ Ϫ0.68, p Ͻ 0.01, n ϭ 18). The values for side chain lengths (shown in Fig. 4 legend) were estimated as the maximal lengths between the ␣-carbons and the most distal non-hydrogen atoms of the side chains using HyperChem 8.0 Pro (Hypercube, Inc.), which are in close agreement to the values used by Hirel et al. (17). For the same 18 channels (WT and mutants), I ss /I peak values were also found to correlate with Van der Waals' volumes (18) (Fig. 4B; r ϭ Ϫ0.54, p Ͻ 0.05, n ϭ 18). The correlation became insignificant when Gly was included in the analysis (r ϭ Ϫ0.43, p ϭ 0.06, n ϭ 19).
A weak correlation between I ss /I peak and accessible area (19) (r ϭ Ϫ0.49, p Ͻ 0.05, n ϭ 18) was also identified. No correlation between I ss /I peak and hydrophobicity (20) was found (r ϭ Ϫ0.04, p ϭ 0.86, n ϭ 18). These correlation analyses suggest that the residue volume or side chain length may dictate, in part, the functionality of ␥Met 438 in Na ϩ self-inhibition.
Substitutions at ␣Gly 481 Enhance Na ϩ Self-inhibition-A functional ENaC complex typically contains all three homologous subunits and, similar to cASIC1, is likely a heterotrimer (14). We examined the role of residues in ␣ and ␤ ECDs at the position homologous to ␥Met 438 in Na ϩ self-inhibition. Residue ␤Ala 422 was substituted with Met and Cys, whereas ␣Gly 481 was replaced with Met, Ala, and Cys (Fig. 5). All three ␣ mutations (␣G481A, ␣G481C, and ␣G481M) greatly enhanced Na ϩ self-inhibition. The ␤A422C mutant showed a WT-like Na ϩ self-inhibition response. Although the tau and I ss /I peak values for ␤A422M mutant were statistically greater than WT (p Ͻ 0.05), the changes were small compared with ␣ and ␥ mutants. These results suggest that ␣Gly 481 plays a role in restricting Na ϩ self-inhibition, and ␤Ala 422 is not important for this process.
Enhanced Na ϩ Self-inhibition of ␣G481M Does Not Appear to Originate from Increased Na ϩ Binding Affinity-To determine whether the altered Na ϩ self-inhibition responses resulted from changed Na ϩ binding affinity or allosteric steps following Na ϩ binding, we examined the apparent Na ϩ affinity   for Na ϩ self-inhibition. As previously described (4), we analyzed the Na ϩ self-inhibition responses at 3, 10, 30, 60, 90, and 110 mM Na ϩ (Fig. 6). Because Na ϩ self-inhibition is a low affinity regulatory process with reported K i over 100 mM for human and mouse ENaCs (4, 7), we chose a mutant with greatly enhanced Na ϩ self-inhibition, ␣G481M, for this analysis. Although WT channels showed clear current decay starting from 30 mM Na ϩ , the ␣G481M displayed robust self-inhibition at Na ϩ concentrations as low as 10 mM. The peak currents of both WT and mutant channels followed Michaelis-Menten enzyme kinetics very well (r 2 of 0.99; Fig. 6C). The estimated K m values were slightly different (44.3 Ϯ 0.9 mM, n ϭ 5, for WT and 37.3 Ϯ 1.9 mM, n ϭ 8, for ␣G491M, p Ͻ 0.05). The steady state currents also fit well to the Michaelis equation (r 2 of 0.98 for WT and 0.97 for the mutant (Fig. 6D)). The estimated K m for I ss of ␣G481M was 15.8 Ϯ 1.5 mM (n ϭ 8), moderately lower than that of WT (25.7 Ϯ 1.3 mM (n ϭ 5, p Ͻ 0.001). We estimated the K i of Na ϩ by fitting I ss against [Na ϩ ] with the substrate inhibition equation (16) as described under "Experimental Procedures." The average K i for ␣G481M mutant was 273 Ϯ 41 mM (n ϭ 8, r 2 of 0.92), not significantly different (p Ͼ 0.05) from that of WT (272 Ϯ 27 mM, n ϭ 5, r 2 of 0.99). Our data are consistent with the view that the enhanced Na ϩ self-inhibition observed with the ␣G481M mutant is not due to a large change in Na ϩ binding affinity but primarily reflects allosteric changes subsequent to Na ϩ binding.
Mutations Affecting Na ϩ Self-inhibition Alter Open Probability-Because Na ϩ self-inhibition is considered an intrinsic gating event (6,21,22), changes in Na ϩ self-inhibition observed with ␣Gly 481 and ␥Met 438 mutant channels are expected to be associated with parallel changes in channel P o . Unitary Na ϩ currents were obtained from cell-attached patches in oocytes expressing ␣G481M␤␥, ␣␤␥M438V, or ␣G481M/␤␥M438V mENaCs. Both bath and pipette solutions contained 110 mM Na ϩ . As shown in Fig. 7, ␣G481M␤␥ channels reside primarily in the closed state with infrequent short openings, resulting in a very low mean P o (0.05 Ϯ 0.01, n ϭ 6). In contrast, ␣␤␥M438V channels spend most of the time in the open state with rare short closures, leading to a very high P o (0.95 Ϯ 0.02, n ϭ 6). The double mutants behave like WT mENaC with the estimated P o of 0.29 Ϯ 0.02 (n ϭ 4), which is close to the value we reported recently under similar conditions (0.37 Ϯ 0.06) (11,23). All of the mutants had unitary currents (0.53 Ϯ 0.02 pA, n ϭ 6 for ␣G481M␤␥; 0.50 Ϯ 0.01 pA, n ϭ 6 for The numbers below each current decay are the Na ϩ concentrations (in mM) used for inducing self-inhibition. Averaged data are shown in C-E with circles for ␣␤␥ and triangles for ␣G481M mutant. Solid (␣␤␥) and dashed (␣G481M) lines were from best fitting of the relative I peak (C) or relative I ss (D and E) with equations indicated in parentheses (K m , Michaelis equation, and SI, substrate inhibition equation, see "Experimental Procedures"). Relative I peak represents the individual peak current at a specific Na ϩ concentration that was normalized to the maximal I peak observed in the same cell. Relative I ss represents the individual steady state current normalized to maximal I ss measured in the same cell. ␣␤␥M438V and 0.47 Ϯ 0.01 pA, n ϭ 4 for ␣G481M/␤␥M438V) comparable with WT (ϳ0.5 pA) (24). A plot of P o versus I ss /I peak values for WT (P o for WT was obtained from Ref. 11) and the three mutants revealed a significant correlation between the two parameters (r ϭ 0.98, p Ͻ 0.05; Fig. 8).
Both ␥Met 438 and ␣Gly 481 Are Solvent-accessible-To further explore functional roles of ␣Gly 481 and ␥Met 438 in the mechanism of Na ϩ self-inhibition and gating of ENaC, we investigated whether they are solvent-accessible by examining the responses of the ␣G481C and ␥M438C mutants to externally applied sulfhydryl reagents. Membrane-impermeant and positively changed MTSET was chosen to avoid modifications of intracellular sulfhydryl groups (25) and because it does not significantly alter the current of WT ␣␤␥ mENaC (9). MTSET (1 mM) applied in the bath solution slowly inhibited currents in oocytes expressing ␣␤␥M438C mENaC with a maximal inhibition of 56 Ϯ 2% (n ϭ 6) within 2 min and a time constant of 40 Ϯ 2 s (n ϭ 6; Fig. 9). The inhibition persisted following washout of the reagent from the bath and was reversed by 10 mM dithiothreitol, indicating that the inhibition was due to covalent modification of the introduced Cys at ␥Met 438 . The ␣G481C␤␥ mutant was also inhibited by 1 mM MTSET with the maximal inhibition of 47 Ϯ 4% (n ϭ 4) within 2 min and a time constant of 3.0 Ϯ 0.5 s (n ϭ 4). MTSET caused a greater inhibition of ␣G481C␤␥M438C (83 Ϯ 1%, n ϭ 6) than observed with either single mutant. The inhibition of ␣G481C␤␥M438C by MTSET displayed a biphasic time course that was best described by a double exponential decay with 1 of 3.2 Ϯ 0.5 s (n ϭ 6) and 2 of 73 Ϯ 7 s (n ϭ 6). The magnitude and time course of the inhibition of ␣G481C␤␥M438C by MTSET suggested that both introduced Cys residues were modified by the reagent. In contrast to ␣G481C and ␥M438C, ␤A422C was insensitive to external MTSET. The currents prior to MTSET, during perfusion with 1 mM MTSET, and following washout were 3.8 Ϯ 0.2, 3.9 Ϯ 0.2, and 3.3 Ϯ 0.2 A (n ϭ 5), respectively (p Ͼ 0.05). A considerable delay following MTSET application was observed prior to current inhibition in all recordings. The estimated latencies were 3.4 Ϯ 1.0 s (n ϭ 4) for ␣G481C␤␥; 7.5 Ϯ 0.5 s (n ϭ 6) for ␣␤␥M438C; and 3.2 Ϯ 0.8 s (n ϭ 5) for ␣G481C␤␥M438C. These delays were not due to slow reagent delivery, because the oocyte perfusion rate was very high (ϳ15 ml/min) in an oocyte chamber with unidirectional flow and small volume (ϳ0.5 ml). In addition, the pore blocker amiloride (10 M) had an estimated latency of 0.6 Ϯ 0.0 s (n ϭ 15, pooled from oocytes expressing all mutants).
Chemical Modification of ␥M438C Inhibits the Channel by "Restoring" Na ϩ Self-inhibition-Current inhibition of the Cys mutants by MTSET likely reflects a decrease in channel P o . Because P o is correlated with the magnitude of Na ϩ self-inhibition (Fig. 8), MTSET-modified ␣␤␥M438C should show greater Na ϩ self-inhibition than unmodified ␣␤␥M438C. This is what we observed (Fig. 10). Following MTSET treatment, the Na ϩ self-inhibition response of ␣␤␥M438C was greatly enhanced (I ss /I peak (ϩMTSET), 0.57 Ϯ 0.03, n ϭ 5; I ss /I peak (ϪMTSET), 0.86 Ϯ 0.02, n ϭ 5) such that the modified mutant channels behaved like WT. The modification of Cys by MTSET results in a new side chain -CH 2 -S-S-CH 2 CH 2 N(CH 3 ) 3 ϩ with a length of ϳ9 Å (estimated using HyperChem 8.0 Pro). The linear relationship between I ss /I peak and side chain length (Fig. 4A) was preserved to a moderate degree with MTSET-modified ␣␤␥M438C (r ϭ Ϫ0.59, p Ͻ 0.01, n ϭ 20), supporting the notion that an increasing side chain length of residue ␥438 generally favors Na ϩ self-inhibition.
The introduction of a bulky side chain at the degeneration site in the ENaC pore, such as ␤S518K, locks the channel in an open state with a P o close to 1 (26 -31). If MTSET treatment of ␣␤␥M438C reduces the P o of the channel, the channel must be able to alter transitions between open and closed states in response to MTSET. This should not occur with a mutation that locks ENaC in an open state. As shown in Fig. 11, MTSET did not alter the current in oocytes expressing ␣␤S518K/ ␥M438C, consistent with the notion that MTSET inhibits ␣␤␥M438C by reducing P o .
Mutations of Selected Aromatic Residues near ␣Gly 481 / ␥Met 438 Alter Na ϩ Self-inhibition-To examine whether residues in proximity to ␣Gly 481 and ␥Met 438 contribute to the Na ϩ self-inhibition response, we mutated four aromatic residues in the vicinity of these sites. Based on sequence alignments, these aromatic residues reside in the same helix (␣5) as ␣Gly 481 and ␥Met 438 (Fig. 1). Mutations immediately before and after ␣Gly 481 (␣W480A and ␣Y482A) had no significant effect on Na ϩ self-inhibition (Fig. 12). However, both ␣Y484A and ␣Y485A greatly enhanced Na ϩ self-inhibition. Substitutions immediately before and after ␥Met 438 (␥W437A and ␥Y439A) significantly reduced and slowed Na ϩ self-inhibition. WT-like Na ϩ self-inhibition was observed in oocytes expressing the ␥Y441A or ␥Y442A mutant. The results suggest that ␣Tyr 484 and ␣Tyr 485 have similar functional roles as ␣Gly 481 in Na ϩ self-inhibition. Similarly, ␥Trp 437 and ␥Tyr 439 may have a functional role similar to ␥Met 438 .

DISCUSSION
The tight correlation between ENaC P o and the magnitude of Na ϩ self-inhibition validates the concept that Na ϩ self-inhibition is an intrinsic gating event that modulates channel P o (6). The observed changes in the Na ϩ self-inhibition response of FIGURE 8. Correlation between P o and I ss /I peak . The P o values from Fig. 7 were plotted against the I ss /I peak obtained from the same mutants. The data points are labeled with mutation names. The P o of WT was from published data (11).
ENaCs with mutations at ␥Met 438 and/or ␣Gly 481 suggest that both residues have functional roles in Na ϩ self-inhibition. The inhibitory effect of MTSET on channels with Cys introduced at either ␣Gly 481 or ␥Met 438 , as well as the effects of ␣Gly 481 and ␥Met 438 mutations on channel P o , also supports the notion that these residues function as gating modifiers.
Our current findings provide new sites in addition to those that have previously been implicated in the allosteric regulation of ENaC by extracellular Na ϩ (4,6,8,9). Additional sites are expected to be identified in future studies as investigators continue to examine the mechanism of Na ϩ self-inhibition, because this inhibitory response to Na ϩ has a high activation energy (Q 10 of ϳ8), implying that large scale conformational changes are associated with this allosteric process (7,32). It is possible that these sites are spread out along the ECD of a subunit, contributing to one or more putative Na ϩ receptors as well as to structures linking the receptors to the gate within the pore.
The changes in Na ϩ self-inhibition by the homologous mutations at ␣Gly 481 and ␥Met 438 suggest that these residues have different roles in the regulation of ENaC by external Na ϩ . All three ␣Gly 481 mutations greatly enhanced Na ϩ self-inhibition, whereas most ␥Met 438 mutations greatly suppressed Na ϩ selfinhibition. We previously identified another pair of homologous residues, ␣His 282 and ␥His 239 , where ␣His 282 substitutions either enhanced (H282D, H282R, and H282C) or accelerated (H282W) the Na ϩ self-inhibition response, whereas ␥His 239 mutations (H239D, H239R and H239C) eliminated the response (4). These observations raise the possibility that certain subdomains within the ␣ subunit restrict Na ϩ self-inhibition, whereas subdomains within the ␥ subunit favor Na ϩ self-inhibition. Drawing parallels between the contributions of specific residues and the functional roles of different subunits should be performed with caution. For example, we recently reported that mutations at eight of sixteen extracellular Cys residues within the ␣ subunit significantly suppressed Na ϩ self-inhibition (9).
As an allosteric process, Na ϩ selfinhibition likely involves three major sequential steps: Na ϩ binding to its "receptor(s)," a series of local and remote conformational changes, and gate closure. ␣Gly 481 and ␥Met 438 could contribute to one or more above steps. Because the gate of ENaC likely resides within the channel pore and the residue in cASIC1 (Lys 342 ) homologous to ␣Gly 481 and ␥Met 438 is located at least 40 Å away from the pore domain (Fig.  1), the likelihood that either ␣Gly 481 or ␥Met 438 is located within the gate is very low. The other two possibilities (Na ϩbinding site versus allosteric regulatory site) are difficult to distinguish. To address this question, we estimated the apparent Na ϩ affinity for Na ϩ self-inhibition of WT and ␣G481M channels (Fig. 6). Our observations suggest that ␣G481M does not increase Na ϩ binding affinity. Both I peak and I ss from WT and ␣G481M mutant showed similar saturation behavior that was well described by Michaelis kinetics. Because I peak is observed prior to appearance of Na ϩ self-inhibition, it likely represents the high P o state of the channel, and its saturation with increasing [Na ϩ ] largely reflects Na ϩ binding affinity for the conduct- The oocytes were clamped at Ϫ60 mV for ␣␤␥M438C and Ϫ100 mV for ␣G481C␤␥ and ␣G481C␤␥M438C. B, summary data. White, black, and gray bars show the averaged amiloride-sensitive currents prior to MTSET, after MTSET treatment, and following washout of MTSET, respectively. For display purposes, the inward currents (negative values) were inverted and shown as positive currents. The numbers of oocytes were 4, 7, and 6 for ␣G481C, ␣␤␥M438C, and ␣G481C-␥M438C, respectively. Asterisks indicate that the currents were significantly different from baseline currents (prior to MTSET) (p Ͻ 0.05, paired Student's t test). C, percentage of MTSET-induced inhibition of the three mutants. An asterisk indicates that the percentage of inhibition in the single mutant was significantly lower than the double mutant (p Ͻ 0.001). D, latency. The time between the digital tag for opening the valve of the perfusion channel for MTSET and the earliest visible change in current was considered latency. An asterisk indicates that the value was significantly lower than that of ␣␤␥M438C (p Ͻ 0.01). E, time constants. The time constants for ␣G481C␤␥ and ␣␤␥M438C were obtained by fitting data with a single exponential equation, whereas the values for ␣G481C␤␥M438C were from fitting with a double exponential equation.
ing pore. Indeed, our estimated K m for I peak of WT mouse ␣␤␥ENaC (44 mM) is very close to the K m (48 mM) from [Na ϩ ] dependence of single channel current in rat cortical collecting tubule (33). On the other hand, I ss saturation is influenced by Na ϩ binding affinity for both the pore and the site for selfinhibition. The similar relationships of I peak and I ss with [Na ϩ ] suggest that the mutant and WT channels have similar Na ϩ affinities for permeation and self-inhibition. If the ␣G481M mutant had higher Na ϩ binding affinity for self-inhibition, a decline of I ss at [Na ϩ ] larger than 30 mM would be expected, as we previously reported for ␣H282 mutants and an ␣ furin site mutant (4,6). Rather as seen in Fig. 6, I ss for the ␣G481M mutant continued to increase with increasing [Na ϩ ]. As a result, the I ss /[Na ϩ ] data fit very well with the Michaelis equation (Fig. 6D). We suggest that ␣G481M "sensitizes" ENaC such that the conformational changes resulting from Na ϩ binding are "amplified" because of a reduced energy barrier for pore closure. We did not determine the apparent Na ϩ binding affinity for a ␥Met 438 mutant that showed suppressed Na ϩ selfinhibition, because these experiments would require a very high Na ϩ concentration (Ͼ Ͼ100 mM). We speculate that ␥Met 438 mutations do not alter Na ϩ binding affinity based on the following observations. First, the changes in Na ϩ self-inhibition with mutations of ␥Met 438 are not dependent on the charge of the side chain. Second, MTSET modification of ␥M438C enhanced Na ϩ self-inhibition. We propose that amino acid substitutions of ␥Met 438 either increase or decrease the energy barrier for shutting off the channels in a residue-dependent manner. This notion is consistent with the rapid inhibition of ␣G481C (lowered energy barrier for closing channels) and the slow inhibition of ␥M438C (increased energy barrier; Fig. 9) in response to MTSET.
Our observations and insights garnered from the structure of cASIC1 allow us to address the potential functional roles of ␣Gly 481 and ␥Met 438 . ␣Gly 481 appears to restrict the extent of Na ϩ self-inhibition, perhaps functioning as a "brake" in the response to external Na ϩ . The functional role of ␣Gly 481 may be related to the unique properties of Gly within ␣-helices. Gly and Pro have the lowest helical propensity. Gly may destabilize a helical conformation by reducing the available hydrogen bonds (34, 35). In contrast, Gly residues have an important role in helix-helix packing interactions (36). For example, Gly zipper motifs (GXXGXXXG or GXXXGXXXG) are found in membrane proteins, including ion channels, and may facilitate helix packing (37). Therefore, Gly can be either a destabilizing force within a helix or serve as a stabilizing component in helical  . MTSET has no effect on channels with a high intrinsic P o . A recording trace representing 4 independent experiments is shown in A. The oocyte was clamped at Ϫ60 mV 1 day after injection of cRNAs for ␣␤S518K/ ␥M438C mENaC. MTSET at 1 mM was applied as in Fig. 9. Amiloride-sensitive currents prior to, during (2 min) and after (1-min washout) application of MTSET are shown in B. packing. Based on the location of homologous residues Lys 342 within cASIC1 (Fig. 1A), it is unlikely that ␣Gly 481 is involved in packing against an adjacent helix (i.e. ␣4 in the thumb domain or ␣3 in the finger domain). We speculate that ␣Gly 481 may destabilize the helical structure of ␣5 in the ␣ subunit, reducing Na ϩ -induced movement of the thumb domain that leads to pore closure. Replacement of Gly by a residue that stabilizes ␣5 (Met, Cys, or Ala) would tend to enhance Na ϩ -induced movement of thumb domain.
The effects of substitutions of ␥Met 438 on Na ϩ self-inhibition correlated, in some degree, with residue volume and the length of the side chain (Fig. 4). This relationship is consistent with a "chain length effect" that has been found in the inactivation of certain enzymes and receptors (38). A long and flexible side chain (Met, Lys, or Leu) generally facilitated the functional response to external Na ϩ , whereas a shorter side chain (Val, Ala, Thr, Asp, Cys, or Ser) hindered the functional response. It is important to recognize that this correlation is weak, given the low correlation coefficient, and there is a clear outliner, Gly. Because Gly does not have a side chain and an introduced Gly has the possibility to alter the backbone and secondary structure, it is often difficult to predict the outcome of a Gly substitution. Clearly the side chain length (or volume) is only one of the factors that define the functional role of ␥438 in the Na ϩ self-inhibition response. Side chain chemical properties such as hydrophobicity and charge do not seem to be important. For instance, substitution of ␥Met 438 with negatively charged Glu or positively charged Arg suppressed Na ϩ self-inhibition to a similar degree.
In contrast to the sensitivity of ␣Gly 481 and ␥Met 438 to substitutions, the ␤A422M mutant had a modestly suppressed Na ϩ self-inhibition response, and ␤A422C caused no change in the response. In addition, ␣␤A422C␥ was not sensitive to external MTSET. This is reminiscent of previous observations that mutations of ␤ residues often resulted in little or no effect on the Na ϩ self-inhibition response, whereas large changes in Na ϩ self-inhibition were seen with substitutions of the corresponding residues in the ␣ or ␥ subunit (4,9). These results suggest that the ␤ subunit has, at best, a limited role in Na ϩ self-inhibition of channels composed of all three subunits. However, the ␤ subunit clearly contributes to the gating mechanism of ENaC, because dramatic changes of P o have been reported with mutations or modifications of an introduced Cys at the degeneration site in the ␤ subunit (26,28,29,31).
One intriguing feature of the thumb domain ␣5 is that there are four well conserved aromatic residues in the neighborhood of ␣Gly 481 /␥Met 438 and the ␣Cys 483 /␥Cys 440 (the 13 th Cys within the ECDs). Previous studies using crystallographic data have suggested that there exists a strong sulfur-aromatic interaction that may contribute to protein stability (39). Our mutational analyses indicate that ␣G481A, ␣Y484A, and ␣Y485A have a similar phenotype (i.e. enhancement on Na ϩ self-inhibition), suggesting a shared functional role. Equivalently, ␥W437A, ␥M438A, and ␥Y439A may contribute to Na ϩ selfinhibition in a similar manner.
In summary, we have identified two homologous residues (␣Gly 481 /␥Met 438 ) in the extracellular thumb domains of ␣ and ␥ ENaC subunits as novel structural determinants for Na ϩ self-inhibition and gating. A long side chain (or no side chain (i.e. Gly)) at position ␥438 appears to preserve the functional role of Met 438 , whereas Gly at position ␣481 restricts the degree of Na ϩ self-inhibition. A strong correlation between P o and the magnitude of Na ϩ self-inhibition was observed. Experiments from sulfhydryl modifications demonstrated that both residues are accessible to extracellular solvent. Our results support the notion that the putative helices of ENaC homologous to the ␣5 of the thumb domains in the cASIC1 structure are important components of the allosteric pathway for ENaC gating and its regulation by extracellular Na ϩ .