External Cu2+ Inhibits Human Epithelial Na+ Channels by Binding at a Subunit Interface of Extracellular Domains*

Epithelial Na+ channels (ENaCs) play an essential role in the regulation of body fluid homeostasis. Certain transition metals activate or inhibit the activity of ENaCs. In this study, we examined the effect of extracellular Cu2+ on human ENaC expressed in Xenopus oocytes and investigated the structural basis for its effects. External Cu2+ inhibited human αβγ ENaC with an estimated IC50 of 0.3 μm. The slow time course and a lack of change in the current-voltage relationship were consistent with an allosteric (non pore-plugging) inhibition of human ENaC by Cu2+. Experiments with mixed human and mouse ENaC subunits suggested that both the α and β subunits were primarily responsible for the inhibitory effect of Cu2+ on human ENaC. Lowering bath solution pH diminished the inhibition by Cu2+. Mutations of two α, two β, and two γ His residues within extracellular domains significantly reduced the inhibition of human ENaC by Cu2+. We identified a pair of residues as potential Cu2+-binding sites at the subunit interface between thumb subdomain of αhENaC and palm subdomain of βhENaC, suggesting a counterclockwise arrangement of α, β, and γ ENaC subunits in a trimeric channel complex when viewed from above. We conclude that extracellular Cu2+ is a potent inhibitor of human ENaC and binds to multiple sites within the extracellular domains including a subunit interface.

The epithelial Na ϩ channel (ENaC) 2 mediates Na ϩ transport across apical membranes of high resistance epithelia in kidney, colon, and lung. ENaC has important roles in the maintenance of extracellular fluid volume and the regulation of airway surface liquid volume (1). Alterations in ENaC activity have been associated with several human diseases. For example, enhanced ENaC activity is responsible for the hypertension seen in Liddle's syndrome, contributes to the mucociliary dysfunction seen in cystic fibrosis, and is believed to contribute to hypervolemia associated with nephrotic syndrome (2,3).
A variety of intracellular and extracellular factors regulate ENaC activity by distinct mechanisms (4). External amiloride analogs, cations, anions, nucleotides, serine proteases, and lam-inar shear stress inhibit or stimulate endogenous or exogenous ENaCs (5)(6)(7)(8)(9)(10)(11)(12). All of these extracellular regulators appear to directly alter the activity of ENaCs in plasma membranes rather than affect channel subunit trafficking (1). Their primary targets likely reside within the characteristically large extracellular domains (ECDs) of ENaC subunits. This notion is in line with the well defined subdomains within the ECDs of the chicken acid-sensing ion channel 1 (cASIC1), a member of the ENaC/ degenerin family, revealed in a crystal structure and the identification of proton binding sites within the ECDS (13).
We have previously examined the effects of the transition metals, Ni 2ϩ and Zn 2ϩ , on ENaC activity. External Ni 2ϩ inhibits and Zn 2ϩ activates mouse ENaCs in Xenopus oocytes by directly interacting with the channels and altering channel gating (6,7). Some of these metal effects are thought to be related to Na ϩ self-inhibition, a down-regulation of open probability (P o ) by extracellular Na ϩ (7,14). Yu et al. (14) have also examined the effects of several transition metals on the single channel activity of native Xenopus ENaCs in A6 cells. These metals differentially affect Xenopus ENaC Po and channel number in membrane patches without changing the single channel conductance. However, the exact binding sites and detailed mechanisms for the metal effects on ENaCs remain largely unknown.
Copper is the third most abundant trace metal in humans and has a variety of important biological functions. Excessive Cu 2ϩ is highly toxic to cells, and its content in cells is carefully maintained at low levels. Indeed, Cu 2ϩ is implicated in several human diseases such as Wilson disease, Menkes disease, neurodegenerative disorders, and cancers (15,16). The therapeutic potential of copper chelators and copper complexes is being intensively investigated (16). In addition, particulate matters contain high amounts of transitional metals including copper. Soluble metals in airborne particles contribute to pulmonary and cardiovascular toxicity (17,18). Recent studies suggest that copper nanoparticles are highly toxic (19). The underlying mechanisms for the harmful effects of Cu 2ϩ are not fully understood. Many studies have suggested that certain metals exert their toxic effects in part by altering functions of ion channels or transporters (20,21). Clearly, a better understanding of the interactions between copper and biological molecules is crucial to an elucidation of its physiological, pathological, and toxicological roles in human health.
In this report, we examined the effects of external Cu 2ϩ on amiloride-sensitive Na ϩ currents in oocytes expressing ␣␤␥ human ENaC (hENaC) and probed the structural basis by sitedirected mutagenesis. We found that external Cu 2ϩ is a potent inhibitor of hENaC. The inhibitory effect of Cu 2ϩ on hENaC depends on the ␣ and ␤ subunits. The most important site for Cu 2ϩ inhibition was identified at the ␣/␤ subunit interface.
ENaC Expression and Two-electrode Voltage Clamp-ENaC expression in Xenopus oocytes and current measurements by two-electrode voltage clamp were performed as previously reported (6). Stage V and VI oocytes with the follicle cell layer removed were injected with 50 nL/cell of mixed cRNAs composed of 2 ng of each hENaC subunit or 1 ng of each mENaC subunit. Injected oocytes were incubated at 18°C in modified Barth's solution (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO 3 , 15 mM HEPES, 0.3 mM Ca(NO 3 ) 2 , 0.41 mM CaCl 2 , 0.82 mM MgSO 4 , 10 g/ml of sodium penicillin and streptomycin sulfate, 100 g/ml gentamycin sulfate, pH 7.4). All of the experiments were performed at room temperature (20 -24°C) 20 -30 h following injection. The oocytes were placed in a recording chamber from Warner Instruments (Hamden, CT) and perfused with bath solutions at a constant flow rate of 12-15 ml/min. Voltage clamp was performed using Axoclamp 900A amplifier and DigiData 1440A interface controlled by pClamp 10 (Molecular Devices Corporation, Sunnyvale, CA). The oocytes were either continuously clamped to Ϫ100 mV to monitor current change over a period of time or stepwise clamped to a series of voltages (Ϫ140 to 60 mV) to determine the currentvoltage relationship.
Examination of the Effects of Cu 2ϩ on ENaCs in Oocytes-The oocytes were perfused with normal bath solution (NaCl-110, containing 110 mM NaCl, 2 mM KCl, 2 mM CaCl 2 , pH 7.4) while clamped to Ϫ100 mV. Bath solution was buffered with either 10 mM HEPES or 5 mM MES and 5 mM MOPS. Although HEPES reportedly forms complex with Cu 2ϩ (24), we observed similar responses to Cu 2ϩ using bath solution buffered with either HEPES or MES and MOPS that do not complex Cu 2ϩ . Inward currents were continuously recorded, whereas bath solutions supplemented with or without Cu 2ϩ were exchanged. Following Cu 2ϩ washout, bath solution was switched to NaCl-110 with 10 M amiloride to determine the amiloride-insensitive current. The effects of Cu 2ϩ on ENaCs were analyzed by comparison of the amiloride-sensitive currents prior to and after Cu 2ϩ application. A Cu 2ϩ stock solution of 1 M was prepared by dissolving CuSO 4 ⅐5H 2 O (purity of 99.999%; Sigma-Aldrich) in deionized water and was diluted to its final concentration in NaCl-110. The highest concentration of Cu 2ϩ solution that could be made in this bath solution and remain relatively stable was 100 M, and precipitates appeared at higher concentration. Bath solutions containing Cu 2ϩ were prepared fresh prior to experiments to minimize the reduction of Cu 2ϩ concentration because of the slow formation of insoluble Cu(OH) 2 at neutral pH. Nominal concentrations of Cu 2ϩ were used. Oocytes with unstable currents were not used in these experiments.
Examination of Na ϩ Self-inhibition-The Na ϩ self-inhibition responses were examined as previously reported (25,26). Briefly, Na ϩ self-inhibition was examined by rapidly replacing a low [Na ϩ ] bath solution (NaCl-1; containing 1 mM NaCl, 109 mM N-methyl-D-glutamine (NMDG), 2 mM KCl, 2 mM CaCl 2 , 10 mM HEPES, pH 7.4) with a high [Na ϩ ] bath solution (NaCl-110), whereas the oocytes were continuously clamped to Ϫ100 mV. Bath solution exchange was done with a Teflon valve perfusion system controlled by computer (AutoMate Scientific Inc, Berkeley, CA). Upon completion of the experiment, 10 M amiloride was added to the bath solution so as to determine the amiloride-insensitive portion of the whole cell current. To avoid complications from the observable variability in the Na ϩ self-inhibition response of WT ENaCs among different batches of oocytes (27), the response of WT channels was always tested in an alternating manner with mutants in the same batch of oocytes. Steady state current (I ss ) was measured at 40 s after I peak . The amiloride-insensitive currents were subtracted from I ss and I peak currents to determine the amiloride-sensitive current ratio of I ss /I peak , the index for the magnitude of Na ϩ self-inhibition.
Examination of the Effect of Cu 2ϩ on hENaCs in Human Airway Epithelial Monolayer-Primary human airway epithelial cells were cultured from excess pathological tissue following lung transplantation and organ donation under a protocol approved by the University of Pittsburgh Investigational Review Board. Human airway epithelial cells were cultured on human placental collagen-coated Costar Transwell filters (0.33 cm 2 ) as described previously (28) and used for experimentation following 4 -6 weeks of culture at an air-liquid interface. Short circuit currents (I sc ) were measured as previously described (28). In brief, cells cultured on filter supports were mounted in modified Ussing chambers, and the cultures were continuously short circuited with an automatic voltage clamp (Physiologic Instruments). The bathing Ringer's solution was composed of 120 mM NaCl, 10 mM HEPES, 3.3 mM KH 2 PO 4 , 0.8 mM K 2 HPO 4 , 0.6 mM MgCl 2 , 0.6 mM CaCl 2 , and 10 mM glucose (pH, 7.4). Chambers were constantly gassed with a mixture of 95% O 2 and 5% CO 2 at 37°C, which maintained a pH of 7.35. Simultaneous transepithelial resistance was recorded by applying a 10-mV pulse/s via an automated pulse generator. Acquire and Analyze 2.3 (Physiological Instruments) was used to control the voltage clamp and analyze the I sc data. I sc recordings included a 30-min equilibration period, followed by the addition of Cu 2ϩ from a 1 M Cu 2ϩ stock solution in water, followed by the addition of 20 M amiloride.
Statistical Analysis-The data are presented as the means Ϯ S.E. Significance comparisons between groups were performed using Student's t test. Curve fittings were performed with Origin Pro 8.0 (OriginLab Corporation, Northampton, MA).

RESULTS
External Cu 2ϩ Inhibits Human ␣␤␥ ENaC-We examined the effect of extracellular Cu 2ϩ on whole cell currents in Xenopus oocytes expressing ␣␤␥ hENaC. Oocytes were continuously clamped at Ϫ100 mV to monitor inward current changes before, during (10 min), and after Cu 2ϩ application in bath solution (NaCl-110 buffered with 5 mM MES and 5 mM MOPS). Because ENaC activity in oocytes typically experiences a rundown (i.e. a spontaneous decline of current over time) (29,30), control experiments were performed the same way except with no Cu 2ϩ addition. The rundown was nearly linear but best fit with an exponential equation. The estimated time constant was 736.6 Ϯ 67.2 s (n ϭ 13). In response to external application of 0.1 M Cu 2ϩ , the current increased slightly at first and then declined in a similar way to the control recording (Fig. 1A), indicating a lack of inhibitory effect. At higher concentrations, Cu 2ϩ reduced the currents in a dose-dependent manner. The time course of the current changes in the absence and presence of Cu 2ϩ are shown in Fig. 1B, and the estimated time constants are listed in Table 1.
We generated a dose-response curve by plotting the relative currents against Cu 2ϩ concentrations (Fig. 1C). Measurements were taken at times equivalent to approximately three time constants after the addition of Cu 2ϩ (Table 1 and black arrowheads in Fig. 1B). Time constant for 0.1 M Cu 2ϩ could not be determined; values at 10 min were used. The data for 30 M Cu 2ϩ were taken at 2 min from experiments with Cu 2ϩ applied FIGURE 1. External Cu 2؉ inhibits human ENaC. A, representative recordings of whole cell currents from oocytes expressing ␣␤␥ hENaCs before, during, and after Cu 2ϩ applications. Each trace represents at least five independent observations. The oocytes were clamped at Ϫ100 mV. Cu 2ϩ was added to the bath solution (NaCl-110 with 110 mM Na ϩ , buffered with 5 mM MES and 5 mM MOPS) at 0.1, 0.3, 1, 3, 10, and 30 M. Control oocytes were clamped in the same way except no Cu 2ϩ application. Inward currents were shown in negative values by convention. The durations of Cu 2ϩ and amiloride (10 M) applications are indicated by black and open bars, respectively. Black arrows point to the transient increases in currents after initiation of Cu 2ϩ applications. B, time courses of the current changes in the absence (control) or presence of Cu 2ϩ . The relative currents were the ratios of the amiloride-sensitive currents measured every minute after the beginning of the Cu 2ϩ applications and the amiloride-sensitive currents prior to Cu 2ϩ additions (at time 0). The data were from 17 oocytes for control and 9 or 10 oocytes for Cu 2ϩ applications. The black arrowheads indicate the times when datum points were chosen for dose-response analysis (2 min for 3 and 10 M Cu 2ϩ , 5 min for 1 M Cu 2ϩ , and 10 min for 0.1 and 0.3 M Cu 2ϩ ). C, dose response of Cu 2ϩ on ␣␤␥ hENaCs. The relative currents from B were adjusted to deduct rundown contribution from the observed current decreases at the corresponding time with the formula: I R (corrected) ϭ I R ϩ (1 Ϫ I R control ). I R and I R control were the relative currents with and without Cu 2ϩ , respectively. The line was from a best fit with the Hill equation. The parameters were: IC 50 , 0.31 M; n, 1.28, B, 0.23; and R 2 , 0.99.

TABLE 1 Time constants for current decreases in the absence (control) and presence of various concentrations of Cu 2؉ in oocytes expressing ␣␤␥ hENaC
Time constants were obtained from best fit of the current changes within 10 min using built-in exponential function in Clampfit 10. The data were from Fig. 1

Copper Inhibits Human ENaC
for 3 min (Fig. 1A). The relative currents were adjusted to deduct rundown contribution from the observed current decreases at the corresponding time as described in details in the figure legend. Fitting the data with the Hill equation (I Cu / I ϭ IC 50 n /(C n ϩ IC 50 n ) ϩ B, n for Hill coefficient, C for Cu 2ϩ concentration, and B for bottom plateau) yielded the following parameters: IC 50 , 0.31 M; n, 1.28; and B, 0.23 with the coefficient of determination (R 2 ) of 0.98.
Prior to the inhibitory effect of Cu 2ϩ on hENaC, a small transient increase in current was typically seen (arrows in Fig. 1A). The transient change in current was not an artifact, although its magnitude varied among different batches of oocytes. It likely resulted from a rapid activation of the channel prior to the inhibitory effect of Cu 2ϩ .
The inhibitory effect of Cu 2ϩ on hENaCs was not readily reversible. Washout of Cu 2ϩ for 1 min only moderately restored the inhibited currents (Figs. 1 and 2A). This slow reversal could result from either tight binding of Cu 2ϩ to the channel complex or permanent inactivation of channels. To distinguish these two possibilities, we utilized a high affinity Cu 2ϩ chelator, tetrathiomolybdate (TTM), to facilitate Cu 2ϩ removal from its binding site(s). Inhibition of hENaC by 10 M Cu 2ϩ was nearly completely reversed following the addition of 10 M TTM (Fig. 2, B and D). TTM alone did not change the current (Fig. 2C). The results suggested that the slow reversibility of Cu 2ϩ inhibition reflects tight binding of Cu 2ϩ to hENaC.
External Cu 2ϩ Inhibits hENaC by Acting on Sites outside of the Pore-Because Cu 2ϩ was applied in bath solution in our study, its action site is likely within either the ECD or the transmembrane domain. We carried out experiments to examine these two possibilities. First, we examined the effect of Cu 2ϩ on the current-voltage (I-V) relationship of hENaC. As shown in Fig. 3A, the I-V curve remained linear following a 3-min application of 10 M Cu 2ϩ . The voltage independence of Cu 2ϩ inhibition is inconsistent with a pore blocking effect. A positively charged pore blocker such as amiloride preferentially blocks the inward current causing an outward rectification in the I-V curve (Fig. 3A, curve for amiloride). Second, we tested whether pore blocker amiloride could protect the channels from inhibition by Cu 2ϩ . ENaC currents completely recovered after washout of 10 M amiloride (Fig. 3, B and D). However, when oocytes were pretreated with 10 M amiloride followed by 10 M Cu 2ϩ , currents only partially recovered after washout of both Cu 2ϩ and amiloride (Fig. 3, C and D). Apparently, Cu 2ϩ inhibited hENaC current after the pore had been occupied by amiloride. These results suggest that Cu 2ϩ acts at a site that is external to the amiloride-binding site within the pore. Based on the above observations, we conclude that external Cu 2ϩ inhibits human ENaCs likely by binding to sites outside of the pore.
Human ␣ and ␤ Subunits Are Necessary to Confer the Response of hENaC to Cu 2ϩ -We tested whether Cu 2ϩ had a similar effect on mouse ENaC and observed that 10 M Cu 2ϩ did not significantly inhibit the current in oocytes expressing ␣␤␥ mENaC (Fig. 4E). To determine the hENaC subunits (␣, ␤, and/or ␥) that are necessary for Cu 2ϩ inhibition, we expressed hybrid channels containing mixed human and mouse subunits in Xenopus oocytes. Replacing human ␣ or ␤ subunits with their mouse counterpart significantly reduced the inhibitory effect of 10 M Cu 2ϩ , whereas replacing the ␥ subunit did not alter the effect of 10 M Cu 2ϩ (Fig. 4). These results suggest that both the ␣ and ␤ subunits are required for the inhibitory effect of Cu 2ϩ on hENaC activity.
Mutations of Multiple His Residues within the ECDs Reduce Cu 2ϩ Inhibition-We hypothesized that the ECDs of the ␣ and ␤ hENaCs contain residues that are required to confer the inhibitory response to Cu 2ϩ . Because Cu 2ϩ is often coordinated by side chains of His, Cys, Met, Glu, and Asp (31), we replaced His, Met, Glu, and Asp residues of ␣ or ␤ hENaC to their counterparts seen in mENaC at equivalent sites. All 16 Cys residues within ␣ ECDs and 18 Cys residues within ␤ ECDs are conserved between human and mouse ENaCs and therefore were not included in this analysis. Among nine ␣ mutants, ␣H468S significantly reduced Cu 2ϩ inhibition (Fig. 5A). Substitution of ␣His 468 with Cys, Asp, Glu, Lys, or Ala similarly attenuated Cu 2ϩ inhibition (Fig. 5, C and D). Moreover, these mutations also rendered the Cu 2ϩ inhibition reversible. Certain mutations at ␣His 468 also decreased the rates of Cu 2ϩ inhibition or increased the rates of current recovery following washout (Table 2). Interestingly, ␣H468C dramatically increased indicates that the relative current after washout with TTM was significantly higher than that after washout with NaCl-110 (p Ͻ 0.001). To minimize rundown of ENaC currents typically seen under continuous hyperpolarization, 10 M Cu 2ϩ was applied for only 1 min and washed out for 1 min. The current was approaching a plateau at 1 min of Cu 2ϩ application, and the I Cu /I was slightly greater than that with 3 min of application. Bath solution was buffered with 5 mM MES and 5 mM MOPS.
both rates, suggesting that the introduced Cys side chain might retain certain capability of this residue interacting with Cu 2ϩ . After His, Cys is the second most frequent Cu 2ϩ coordinating residue (31). One of the 13 ␤ mutants, ␤H159D, showed a significantly reduced response to 10 M Cu 2ϩ but did not alter its reversibility or rate (Fig. 5, B and D). Double mutant ␣H468S-␤H159D showed a minimal inhibitory response to 10 M Cu 2ϩ (I Cu /I ϭ 0.85 Ϯ 0.01, n ϭ 6, p Ͻ 0.0001 versus WT; Fig. 5).
Histidine has an ionizable imidazole ring with an average pK a of 6.6 Ϯ 1.0 in folded proteins (32). If His residues are indeed involved in Cu 2ϩ inhibition of hENaC, lowering the pH of the bath solution containing Cu 2ϩ should reduce Cu 2ϩ inhibition because of increased protonation of imidazole nitrogens. We therefore examined the effect of 10 M Cu 2ϩ in pH 6.0 bath solution on hENaC currents. Changing the pH of the bath solution from 7.4 to 6.0 moderately increased currents in oocytes expressing ␣␤␥ hENaCs (relative current ϭ 1.19 Ϯ 0.01, n ϭ 7, p Ͻ 0.001; Fig. 6A) as reported by Collier and Snyder (8). At pH 6.0, we observed a minimal inhibitory effect of 10 M Cu 2ϩ (I Cu /I ϭ 0.86 Ϯ 0.01, n ϭ 7, p Ͻ 0.001), in contrast to the large inhibitory effect at pH 7.4 (p Ͻ 0.001; Fig. 6B). These results demonstrate that Cu 2ϩ inhibition of hENaC is pH-dependent and are consistent with His side chains mediating the interaction of Cu 2ϩ with hENaC.
There are 10, 13, and 14 His residues in the ECDs of ␣, ␤, and ␥ hENaC, respectively. We individually mutated each His residue to identify other sites for Cu 2ϩ interaction with the chan- . Cu 2؉ binding site is likely not located within hENaC pore. A, I-V curves before and after 10 M Cu 2ϩ . Oocytes expressing ␣␤␥ hENaCs (n ϭ 5) were clamped to a series of voltages (Ϫ140 to 60 mV) for 500 ms, and the currents were measured at 400 ms. B, recording trace showing current changes before, during, and after washout of 10 M amiloride (80 s). Oocytes were clamped at Ϫ100 mV. The current decay from a peak current following amiloride washout reflects Na ϩ self-inhibition response (25). C, recording from a similar experiment to the one in B except that 10 M Cu 2ϩ was co-applied with 10 M amiloride for 60 s. Cu 2ϩ was added 10 s after amiloride and withdrawn 10 s before amiloride to ensure that Cu 2ϩ was only applied in the presence of amiloride. Bath solution was buffered with 5 mM MES and 5 mM MOPS. D, current recovery after amiloride washout with (open bar) or without (black bar) Cu 2ϩ . The values were the ratios of the amiloride-sensitive currents measured after amiloride washout for 60 s and before amiloride application. The current recovery following co-application of amiloride and Cu 2ϩ was significantly lower than that following amiloride application alone (p Ͻ 0.001, n ϭ 10 for amiloride and 7 for amiloride ϩ Cu 2ϩ ). nel. In addition to the ␣His 468 mutants (Fig. 5), substitution of ␣His 255 with Ala, Cys, Asp, or Arg significantly reduced channel inhibition by 10 M Cu 2ϩ (Fig. 7A). For the ␤ subunit, two mutations (␤H159A and ␤H160A) significantly reduced Cu 2ϩ inhibition (Fig. 7B). The change in Cu 2ϩ inhibition by ␤H159A was similar to that of ␤H159D (Fig. 5B). Two ␥ His mutants (␥H88A and ␥H277A) showed a significantly reduced inhibitory effect of 10 M Cu 2ϩ (Fig. 7C). We also noted that substitutions at two ␥ECD sites (␥His 233 and ␥His 332 ) increased the inhibitory effect of 10 M Cu 2ϩ (p Ͻ 0.01).
A Cu 2ϩ Binding Site Is Located at a Subunit Interface-In certain ligand-gated channels, transition metal-binding sites are located at subunit interfaces (33,34). We reasoned that external Cu 2ϩ might bind at a contact site between two subunits. Residue ␣His 468 aligned with Lys 355 of cASIC1. The cASIC1 structure shows that the side chain of Lys 355 , located at the carboxyl end of an ␣ helix (␣5) in the thumb domain interacts with side chain of Glu 178 within the ␤3-␤4 loop of the palm domain of an adjacent subunit (Fig. 9A) (13). The distance between NZ of Lys 355 and OE2 of Glu 178 is 3.2 Å, well within the range of distance for ion pairs in proteins (35). Lys 355 of cASIC1 aligns with hENaC ␣His 468 , ␤Arg 437 , and ␥Glu 446 , whereas Glu 178 of cASIC1 aligns with hENaC ␣Val 287 , ␤Glu 254 , and ␥Val 265 (Fig. 9C). Of these six hENaC residues, only ␣His 468 and ␤Glu 254 could form an ion pair. If a His residue resides at the position of Lys 355 in subunit A of cASIC1, the distances between an imidazole nitrogen of Lys 355 and a carboxyl oxygen of Glu 178 in subunit C of cASIC1 would be 3.4 -5.5 Å, close to the sum of average distances for Cu 2ϩ -N-His (2.0ϳ2.1 Å) and Cu 2ϩ -O-Glu (2.2 Å) in proteins (36). We therefore hypothesized that ␣His 468 and ␤Glu 254 coordinate the same Cu 2ϩ at the ␣ and ␤ subunit interface. To test the hypothesis, we mutated ␤Glu 254 to Ala, Arg, Gln, His, Asp, and Cys. All of the substitutions except ␤E254D led to a modest but significant reduction of inhibition by 10 M Cu 2ϩ (Fig. 9D). The double mutant ␣H468A/␤E254A nearly eliminated the inhibitory effect of 10 M Cu 2ϩ (I Cu /I ϭ 0.83 Ϯ 0.01, n ϭ 5, p Ͻ 0.0001 versus WT, ␣H468A or ␤E254A; Fig. 9E). Our observations are consistent with the above hypothesis that ␣His 468 and ␤Glu 254 contribute to coordination of the same Cu 2ϩ ion. The close proximity of ␣His 468 and ␤Glu 254 requires an obligate counterclockwise subunit arrangement of ␣-␤-␥ when viewed from above (Fig.   Fig. 2A. In experiments with both pH 6.0 and 7.4, 10 M Cu 2ϩ was applied for only 1 min to minimize potential complication from intracellular acidification, which inhibits ENaCs (47). The I Cu /I obtained with 1 min of application of 10 M Cu 2ϩ at pH 7.4 was slightly greater than that with 3 min of application. 9F). A clockwise arrangement of ␣, ␤, and ␥ subunits would place ␥Glu 446 and ␤Glu 254 in close proximity (Fig. 9G). Charge reversal mutation ␥E446R did not alter the inhibitory effect of 10 M Cu 2ϩ (I Cu /I ϭ 0.08 Ϯ 0.01, n ϭ 4, p Ͼ 0.7 versus WT). The result was not consistent with an involvement of ␥Glu 446 in Cu 2ϩ binding and a pairing between ␥Glu 446 and ␤Glu 254 that would otherwise require a clockwise organization of ␣, ␤, and ␥ subunits. Cu 2ϩ Inhibition and Na ϩ Self-inhibition-We studied the effect of Cu 2ϩ on hENaC in a bath solution containing 110 mM Na ϩ , which is typically used by investigators so as to produce easily measurable currents. At this concentration, extracellular Na ϩ reduces the P o of ENaC via a process of Na ϩ self-inhibition (25,26,37,38). The degree of Na ϩ self-inhibition dramatically affects the effects of other extracellular regulators on ENaCs such as Zn 2ϩ , H ϩ , and Cl Ϫ (7-9). We suspected that some of the six His mutations might alter the Na ϩ self-inhibition response, and as a secondary effect Cu 2ϩ inhibition appeared to be reduced. We therefore examined the Na ϩ self-inhibition responses of these mutant ENaCs.
As shown in Fig. 10, ␣H255A greatly enhanced Na ϩ selfinhibition, whereas ␤E254Q and ␥H277A moderately increased the magnitude of Na ϩ self-inhibition. In contrast, ␣H468A, ␤H159A, ␤H160A, and ␥H88A did not significantly A, a structural model of chicken ASIC1. The trimeric cASIC1 structure (45) was rendered as three colored ribbons (subunit A, B, and C in red, green, and blue, respectively) with PyMol (version 1.3) using coordinates from Protein Data Bank (identification code 3HGC). The ECDs on the top part of the structure are linked to the transmembrane (TM) domain via six short coiled segments termed the wrist (13). The boxed area is enlarged on the right to show the contact between the thumb domain Lys 355 of subunit A and the palm domain Glu 178 of subunit C in the cASIC1 structure. For clarity the area is shown with semitransparent surface and ribbon rendering. The side chains of Lys 355 and Glu 178 are shown as sticks with carbon, oxygen, and nitrogen atoms colored in cyan, red, and blue, respectively. B, nontransparent surface rendering of the same region as in A with subunit A in red and subunit C in blue to highlight the subunit interface. Lys 355 was mutated to a His. Homologous hENaC residues to K355H and Glu 178 are shown in parentheses. The surface near the two residues are omitted. C, sequence alignments of human, mouse ENaCs, and cASIC1 in regions surrounding ␤Glu 254 (Ϫ) and ␣His 468 (ϩ). Secondary structures are shown according to the cASIC1 structure (13). D, I Cu /I from oocytes expressing ␤ mutant and WT ␣ and ␥ hENaCs. Black bars indicate that the values were significantly different from that of WT in the same batches of oocytes (p Ͻ 0.01, n ϭ 5). The dashed line shows the average value from pooled WT-expressing oocytes in this experiment (0.16 Ϯ 0.01, n ϭ 10). Bath solution was buffered with 10 mM HEPES. E, superimposed traces showing the responses of WT (black), ␤E254A (blue), ␣H468A (red), and ␣H468A/␤E254A (purple) and I Cu /I values in corresponding colors. All of the mutant values were significantly greater than that of WT (p Ͻ 0.001), and the asterisk shows that the value of the double mutant was significantly greater than that of the either single mutant (p Ͻ 0.001, n ϭ 5). F, an illustration showing a counterclockwise arrangement of three ENaC subunits in a top view. Three rectangles represent the approximate spaces occupied by ␣ (red), ␤ (blue), and ␥ (green) subunits. Overlapped regions highlight intersubunit contacts including three-way interaction at the center and two-way interactions at more distal regions. Subunit-subunit interactions happen primarily along the two inner laterals. We designated the long inner sides as ϩ and the short inner side as Ϫ by convention used in other ligand-gated channels. G, an alternative (clockwise) arrangement of ␣, ␤, and ␥ ENaC subunits. change the Na ϩ self-inhibition response. Na ϩ self-inhibition was largely eliminated in the ␥H233A mutant, as previously reported (8). These observations suggest that ␣His 468 , ␤His 159 , ␤His 160 , and ␥His 88 have specific roles in Cu 2ϩ inhibition of hENaC.
Mutations that altered the Na ϩ self-inhibition response also changed the magnitude of the transient activation of current that appeared before the inhibitory effect of Cu 2ϩ (Figs. 1A and  7D). The transient activation was absent in mutants with Na ϩ self-inhibition eliminated (for example, ␥H233D; Fig. 7D). It suggests that the transient activation by Cu 2ϩ was caused by a relief of Na ϩ self-inhibition, in a manner similar to that of ENaC activation by external Zn 2ϩ and H ϩ (7, 8). Accordingly, ␣H255D and ␥H277A that enhanced the Na ϩ self-inhibition response increased the magnitude of the transient Cu 2ϩ activation by 5.3-and 3.1-fold, respectively, compared with WT (p Ͻ 0.001). These observations suggest that the reduced Cu 2ϩ inhibition in ␣His 255 and ␥His 277 mutants (i.e. greater I Cu /I; Fig. 7) was related to the augmented stimulatory effect of Cu 2ϩ , which resulted from enhanced Na ϩ self-inhibition. Other mutations including ␣His 468 mutations (Fig. 5D), ␤H159A (Fig. 7D), ␤H160A (Fig. 7D), ␤Glu 254 mutations (Fig. 9E), and ␥H88A (Fig. 7D) did not significantly change the transient Cu 2ϩ stimulation, consistent with their specific role in the inhibitory effect of Cu 2ϩ . , and ␥H277A mutants. The oocytes were clamped at Ϫ100 mV, and the whole cell currents were continuously recorded, whereas bath [Na ϩ ] was rapidly increased from 1 mM (open bar) to 110 mM (gray bar). Both bath solutions were buffered with 10 mM HEPES. The traces are representative of at least five independent observations. The current decay following an increase in [Na ϩ ] represents Na ϩ self-inhibition. B, I ss /I peak of mutant hENaCs. I peak was the peak current, and I ss was the current 40 s after the peak current. The ratio is inversely proportional to the magnitude of Na ϩ self-inhibition. The black bars indicate that values were significantly different from that of WT obtained in the same batches of oocytes (p Ͻ 0.01, n ϭ 5). The dashed line indicates the average value of WT hENaCs pooled from all oocytes in this experiment (0.50 Ϯ 0.01, n ϭ 84) for reference only.

Cu 2ϩ Inhibits Short Circuit Current in Human Airway
Epithelial Monolayers-We studied whether external Cu 2ϩ had an inhibitory effect on natively expressed human ENaCs similar to that on cloned ENaCs expressed in oocytes by examining the effect of apically applied Cu 2ϩ on the short circuit currents in human airway epithelial monolayers. The portion of I sc that was blocked by 10 M amiloride was considered ENaC-mediated current. As shown in Fig. 11, Cu 2ϩ at doses above 10 M reduced the I sc in a dose-dependent manner.

DISCUSSION
In this study, we found that external Cu 2ϩ inhibited human ␣␤␥ ENaC in both Xenopus oocytes and human airway epithelia. External Cu 2ϩ at 10 M does not inhibit mouse (Fig. 4) or rat ENaC. 3 Native Xenopus ENaCs in A6 cells are activated by extracellular Cu 2ϩ (14). Therefore, in the context of previous reports and our current observations, external Cu 2ϩ appears to be a specific inhibitor of human ENaC among cloned ENaCs. The inhibitory effect of Cu 2ϩ on the hENaC current in the human airway epithelia appeared to be smaller and weaker than that in oocytes (Fig. 11). We do not know the exact cause for the different responses to Cu 2ϩ . They could be related to certain experimental conditions utilized in the two systems, such as temperatures (i.e. 20 -24°C in oocytes and 37°C in epithelia) and oxygen tensions, both of which regulate ENaC activity (25,39,40). The differences in the response to Cu 2ϩ could also reflect the inherent differences between native and heterologously expressed channels. It has been reported that ENaCs in oocytes and epithelial monolayers display different sensitivities to the peptide inhibitors derived from the inhibitory domains of ␣ and ␥ mouse ENaCs (41,42).
The estimated IC 50 of 0.31 M for Cu 2ϩ inhibition suggests that external Cu 2ϩ is a potent hENaC inhibitor. A high affinity ENaC inhibitor may be useful in treating diseases associated with elevated ENaC-mediated Na ϩ absorption such as Liddle syndrome and cystic fibrosis (43,44).
Recent studies have established a critical role for ENaCs in the regulation of airway surface liquid volume and excessive activity of ENaCs in airways contributes to the pathogenesis of cystic fibrosis (44). Impaired Na ϩ transport in alveoli leads to pulmonary edema (40). Airborne particles contain a considerate amount of transition metals including copper. Upon contact with biological fluids, free metal ions can be released from the particles and cause local and even remote toxic effects (17,18). We speculate that the inhibitory effect of Cu 2ϩ on human ENaCs in lung epithelia may contribute to the toxicological symptoms caused by inhaled particulate matters. Cu 2ϩ , released from particulate matters, may worsen particle-induced pulmonary edema by inhibition of Na ϩ absorption in airways and alveoli.
The main goal of this study was to probe the structural basis for hENaC inhibition by external Cu 2ϩ . Experiments with mixed human and mouse ENaC subunits demonstrate that ␣ and ␤ hENaC subunits are necessary for the specific response of hENaCs to Cu 2ϩ (Fig. 4). Initial mutational screening of the hENaC-specific residues within ␣ and ␤ ECDs identified ␣His 468 and ␤His 159 as residues involved in Cu 2ϩ inhibition (Fig. 5). The double mutant (␣H468S-␤H159D) converted the response to Cu 2ϩ of the human channel to that of the mouse channel (Figs. 4 and 5), suggesting that these two hENaC-specific residues are primarily responsible for the distinct response to 10 M Cu 2ϩ of human versus mouse ENaC. Subsequently, systematic screening of His residues within ECDs of ␣, ␤, and ␥ subunits identified additional four sites (␣His 255 , ␤His 160 , ␥His 88 , and ␥His 277 ) where mutations significantly reduced the inhibitory effect of 10 M Cu 2ϩ (Fig. 7). However, further analyses showed that ␣His 255 and ␥His 277 mutations significantly increased the magnitudes of Na ϩ self-inhibition and Cu 2ϩ -induced transient activation preceding the inhibitory effect (Figs. 7D and 10), suggesting indirect roles for both His residues in the Cu 2ϩ inhibition. Another residue, ␥His 88 , does not appear to have an essential role in Cu 2ϩ inhibition, given the small effect of its mutation on Cu 2ϩ inhibition (Fig. 7). On the contrary, ␣His 468 , ␤His 159 , and ␤His 160 mutations specifically reduced Cu 2ϩ inhibition, without affecting the Na ϩ self-inhibition response and the transient activation by Cu 2ϩ . We conclude that these three His residues are involved in Cu 2ϩ inhibition.
Our data suggest that ␣His 468 has a key role in mediating Cu 2ϩ inhibition. Taking advantage of the structural information for cASIC1 (13,45), we predicted that ␣His 468 and ␤Glu 254 could contribute to a Cu 2ϩ -binding site at the ␣/␤ subunit interface. Mutational analyses confirmed the prediction (Fig.  9). The identification of ␣His 468 /␤Glu 254 pair suggests a counterclockwise configuration of ␣, ␤, and ␥ subunits when viewed 3 J. Chen and S. Sheng, unpublished observations. FIGURE 11. Cu 2؉ inhibits native human ENaCs in human airway epithelial cells. A, superimposed short circuit recordings showing the responses of the currents to 1, 10, 30, and 100 M Cu 2ϩ applied in apical chamber. B, I Cu /I, ratio of amiloride-sensitive current in the presence of Cu 2ϩ and amiloride-sensitive current prior to Cu 2ϩ application. The asterisks indicate that the values in the presence of 30 and 100 M Cu 2ϩ were significantly less than that of base-line currents (p Ͻ 0.001, n ϭ 6). from above the channel (Fig. 9F). This subunit arrangement is in agreement with a recent report by Collier and Snyder (46). However, we cannot rule out the presence of both the counterclockwise and clockwise subunit arrangements. Another limitation of the notion is that it is based on the assumption that ENaC, like ASIC1, has a trimeric architecture, which remains to be established experimentally.
ENaC P o is regulated by a variety of extracellular factors that may share common pathways in their regulation of ENaC gating. Indeed, the effects of external Zn 2ϩ , H ϩ , and Cl Ϫ on ENaC activity rely on the existence of Na ϩ self-inhibition (7)(8)(9). However, Cu 2ϩ inhibition of hENaC does not depend on the existence of Na ϩ self-inhibition. In fact, the magnitude of Cu 2ϩ inhibition was increased by Na ϩ self-inhibition eliminating mutation (␥H233A) and reduced by Na ϩ self-inhibition enhancing mutation (␣H255A). Therefore, Cu 2ϩ likely inhibits the hENaC via a pathway distinct from that of Na ϩ self-inhibition. In contrast, the transient activation of hENaC by Cu 2ϩ appears to result from a relief of Na ϩ self-inhibition, because Na ϩ self-inhibition eliminating (␥H233A) or enhancing (␣H255A) mutations diminished or enhanced its magnitude accordingly.
In summary, we found that external Cu 2ϩ is a high affinity inhibitor of human ENaC and identified a Cu 2ϩ -binding site at subunit interface within the extracellular domains. Structureassisted mutational analyses suggest that a thumb domain His residue of ␣ subunit and a palm domain Glu residue of the ␤ subunit interact with a Cu 2ϩ ion. This pairing (␣His 468 / ␤Glu 254 ) requires a counterclockwise arrangement of ␣, ␤, and ␥ ENaC subunits when viewed from above.