External Nickel Inhibits Epithelial Sodium Channel by Binding to Histidine Residues within the Extracellular Domains of α and γ Subunits and Reducing Channel Open Probability*

Epithelial sodium channels (ENaC) are regulated by various intracellular and extracellular factors including divalent cations. We studied the inhibitory effect and mechanism of external Ni2+ on cloned mouse α-β-γ ENaC expressed inXenopus oocytes. Ni2+ reduced amiloride-sensitive Na+ currents of the wild type mouse ENaC in a dose-dependent manner. The Ni2+ block was fast and partially reversible at low concentrations and irreversible at high concentrations. ENaC inhibition by Ni2+ was accompanied by moderate inward rectification at concentrations higher than 0.1 mm. ENaC currents were also blocked by the histidine-reactive reagent diethyl pyrocarbonate. Pretreatment of the oocytes with the reagent reduced Ni2+inhibition of the remaining current. Mutations at αHis282and γHis239 located within the extracellular loops significantly decreased Ni2+ inhibition of ENaC currents. The mutation αH282D or double mutations αH282R/γH239R eliminated Ni2+ block. All mutations at γHis239eliminated Ni2+-induced inward current rectification. Ni2+ block was significantly enhanced by introduction of a histidine at αArg280. Lowering extracellular pH to 5.5 and 4.4 decreased or eliminated Ni2+ block. Although αH282C-β-γ channels were partially inhibited by the sulfhydryl-reactive reagent [2-(trimethylammonium)ethyl] methanethiosulfonate bromide (MTSET), α-β-γ H239C channels were insensitive to MTSET. From patch clamp studies, Ni2+ did not affect unitary current but decreased open probability when perfused into the recording pipette. Our results suggest that external Ni2+ reduces ENaC open probability by binding to a site consisting of αHis282 and γHis239 and that these histidine residues may participate in ENaC gating.

Epithelial Na ϩ channels (ENaCs) 1 mediate Na ϩ transport across high resistance epithelia and participate in the regula-tion of extracellular fluid volume and blood pressure. Molecular cloning has revealed that ENaC subunits (␣, ␤, and ␥) belong to a super gene family which is now often referred to as ENaC/ Deg family and consist of ENaCs, degenerins (DEG and MEC), acid-sensing ion channels (ASIC), and FMRFamide-gated Na ϩ channels (1)(2)(3)(4)(5). These channel proteins are composed of subunits that share a common membrane topology with two transmembrane domains, intracellular N and C termini and large extracellular loops (ECL). Members of the ENaC/Deg family are Na ϩ -selective and sensitive to the epithelial Na ϩ channel blocker amiloride, although differences in factors that activate, or gate, these channels have been reported (1). Combined application of site-directed mutagenesis and oocyte expression has yielded important findings regarding structure and function relationships of ENaC (6 -9). Although the exact subunit stoichiometry of ␣-␤-␥ ENaC is still controversial (10 -12), all three ENaC subunits (␣, ␤, and ␥) contribute to formation of the core structure, the ion conduction pore (13). Two amiloridebinding sites have been identified within ENaC subunits; one within the pore regions and a second within the ECL of ␣ENaC (13)(14)(15). The main selectivity filter resides at a 3-residue tract ((G/S)XS) within the putative pore region (or equivalently termed as pre-M2 region) (16 -20). Mutations that alter ENaC gating have been found in the N termini (21,22), pore regions (23,24), and nearby regions (25), M2 domains (26), and the C termini (27).
Variability in the ionic selectivity, conductance, and gating of ENaCs in both native tissues and expression systems has been observed (1). In part, the functional diversity of the Na ϩ channels may reflect the influence of intracellular and extracellular factors that regulate channel activity. Divalent cations are important regulators of many ion channels, especially voltagegated and ligand-gated channels (28). Several cations (Ca 2ϩ , Mg 2ϩ , and Ba 2ϩ ) were reported to block Na ϩ currents from the external side with low affinity and complex features in native tissues (29). Recently, it was reported that external Ni 2ϩ stimulated short circuit currents in A6 monolayers (30) and blocked whole-cell currents in oocytes expressing rat ␣-␤-␥ ENaC (31). We report the inhibitory effects of extracellular Ni 2ϩ on the whole-cell and single channel currents in Xenopus oocytes expressing ␣-␤-␥ mouse ENaC (mENaC), and the identification of two histidine residues within the extracellular domain as putative Ni 2ϩ binding sites. Preliminary description of this work was presented in abstract form (32,33).

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis and Functional Expression of ␣-␤-␥ mENaC in Xenopus oocytes-Point mutations were generated in ␣, ␤, or ␥ mENaC cDNAs (34) cloned into pBluescript SKϪ vector (Stratagene, La Jolla, CA) using two-step PCR methods as described previously (19). cRNAs were synthesized with T3 RNA polymerase (Ambion Inc., Austin, TX) and dissolved in nuclease-free water (Ambion Inc.). Stage V-VI Xenopus oocytes were treated with collagenase type IV (Sigma) to remove follicle cell layers and injected with 1-4 ng of cRNA for each mENaC subunit per cell in a total volume of 50 l. Injected oocytes were maintained at 18°C in modified Barth's saline.
Two-electrode Voltage Clamp-Two-electrode voltage clamp was performed 24 -72 h after cRNA injection at room temperature (20 -24°C) as described previously (19). The bath solution contained 110 mM NaCl, 4 mM KCl, 2 mM CaCl 2 , 10 mM HEPES, and the pH was adjusted to 7.40 with NaOH. Pipettes filled with 3 M KCl had a resistance of 0.5-2 megohms. Typically, oocytes were clamped from Ϫ140 to 60 mV in increments of 20 mV for 500 ms every 2 s. Amiloride at the concentration of 100 M was added to the bath at the end of each experiment to determine the amiloride-insensitive current that was used to calculate amiloride-sensitive currents. Data acquisition and analyses were performed with pClamp 7.1 for Windows (Axon Instruments, Union City, CA).
Patch Clamp-Vitelline membranes of oocytes were removed manually following incubation of the oocytes at room temperature in a hypertonic solution containing 200 mM sucrose. Oocytes were then transferred to a recording chamber with bath solution and allowed to recover for 30 min before clamping. The bath solution contained 110 mM NaCl, 2 mM CaCl 2 , 10 mM HEPES, pH 7. 40. For Na ϩ current recordings the pipette solution was identical to the bath solution, and for Li ϩ current recordings NaCl was replaced by LiCl. Glass pipettes with tip resistances of 5-20 megohms were used. Single channel currents were recorded in cell-attached configuration using PC ONE Patch Clamp amplifier (Dagan Corp., Minneapolis, MN), DigiData 1322A interface, and Clampex 8.1 software (Axon).
Examination of Effects of Inhibitors-Inhibitors used in two-electrode voltage clamps (NiCl 2 , MgCl 2 , amiloride, diethyl pyrocarbonate, and sulfhydryl reagents) were prepared in the bath solution and delivered to the bath by gravity perfusion. To avoid hydrolysis, diethyl pyrocarbonate (DEPC) and sulfhydryl reagents were prepared immediately prior to use. Whole-cell Na ϩ currents were recorded before and after external application of an inhibitor. The effects of a certain inhibitor on ENaC currents were defined by comparison of the amiloridesensitive currents before and after perfusion of the inhibitor in the same oocyte. Dose-response relationship for Ni 2ϩ inhibition of mENaC currents was obtained by plotting the relative currents measured at Ϫ100 mV and in increasing concentrations of NiCl 2 (0.01, 0.1, 1, 10, 25, and 50 mM) against Ni 2ϩ concentrations using a semi-logarithmic scale. The relative currents represent the ratios of whole-cell amiloride-sensitive Na ϩ currents in the presence of Ni 2ϩ in bath solutions relative to the current measured immediately before application of Ni 2ϩ . The osmolarity of the bath solution containing NiCl 2 was not adjusted. Non-linear least squares curve fitting was used to obtain the dose-response parameters. In most cases dose-response data were fitted with both one-site and two-site equations using software OriginPro 7.0 for Windows (Orig-inLab Corp., Northampton, MA). The one-site Equation 1 is as follows: where I R is the relative current in the presence of a specific concentration of inhibitor; K i is the inhibition constant; C is the concentration of an inhibitor, and n represents Hill coefficient. The two-site Equation 2 is as follows: where I R and C are as defined for Equation 1; K 1 and K 2 are the K i values for two components in the dose-response curve; A and B are the fractions of each component, and their sum equals the total drop of I R in the dose-response curve. The goodness of fitting of curves obtained from the one-site or two-site equation was compared by an F test designed to assess changes in both the sum-of-squares and degrees of freedom using Prism 3.0 (GraphPad Software Inc., San Diego). Correlation coefficients (R 2 ) calculated from curve fittings were also used as a parameter of goodness of fitting. Pipette perfusion of Ni 2ϩ in patch clamps was performed by applying either positive pressure to the perfusion vial or negative pressure to the pipette using 2PKϩ Whole-cell/Patch Perfusion Kit (ALA Scientific Instruments Inc., Westbury, NY). Success of pipette perfusion was judged by examining the increase in pipette solution and additional noise in the recordings.
Statistical Analyses-Data are expressed as mean Ϯ S.E. in most figures. Student's t test was used to assess significance of difference between WT and mutant channels or between mutant channels. Nonlinear curve fitting was performed as above with Origin 7.0 Pro.

RESULTS
External Ni 2ϩ Inhibited Amiloride-sensitive Na ϩ Currents in Oocytes Expressing ␣-␤-␥ mENaC-The effect of externally applied Ni 2ϩ on whole-cell amiloride-sensitive Na ϩ currents was studied in Xenopus oocytes expressing ␣-␤-␥ mENaC. Ni 2ϩ inhibited amiloride-sensitive Na ϩ currents in a dose-dependent manner in the concentration range of 0.01-50 mM (Fig. 1A). At 50 mM, Ni 2ϩ inhibited about 80% of the amiloride-sensitive current. The dose-response data were fitted reasonably well with the one-site equation described under "Experimental Procedures" (Fig. 1C, dashed line). The estimated inhibitory constant (K i ) and Hill coefficient were 0.58 Ϯ 0.09 (n ϭ 6) and 0.37 Ϯ 0.01 mM (n ϭ 6), respectively (Table I). However, the dose-response curve appeared to contain more than one component. Data were fitted with the two-site equation as described under "Experimental Procedures," assuming that two classes of binding sites with different apparent affinities for Ni 2ϩ account for the inhibitory effect of Ni 2ϩ on ENaC currents (Fig. 1C, solid line). The non-linear least square fitting resulted in a higher correlation coefficient with the two-site equation (0.9958) than that obtained from fitting with the one-site equation (0.9545). According to the F test, the two-site equation was significantly better than the one-site equation to describe the data (p Ͻ 0.05). Two K i values were 0.05 Ϯ 0.00 and 11.00 Ϯ 1.60 mM (n ϭ 6).
The current reduction by Ni 2ϩ was accompanied by a moderate inward rectification as the concentration of Ni 2ϩ exceeded 0.1 mM (Fig. 1, A and B). The current-voltage relationship curves (I-V) indicate that Ni 2ϩ inhibition of ENaC currents was enhanced when oocyte membranes were depolarized. The Ni 2ϩ -induced current rectification is different from that induced by amiloride, which is generally considered as a pore blocker for ENaC and results in outward current rectification (35,36).
The Ni 2ϩ inhibition of mENaC currents was fast with a maximal effect observed around 1 min after initiation of Ni 2ϩ perfusion (Fig. 1D). Removal of Ni 2ϩ from the bath solution partially restored the Na ϩ currents when Ni 2ϩ was applied at 1 mM. However, at a high concentration (50 mM) Ni 2ϩ inhibition was not reversed by removal of Ni 2ϩ from the bath for a period of 2 min (Fig. 1E).
It was reported that external Ca 2ϩ , Mg 2ϩ , Ba 2ϩ , and Sr 2ϩ blocked Na ϩ channels in toad urinary bladder with complex voltage dependence. The blocking effects were rather weak with calculated concentrations for half-maximal inhibition of greater than tens of millimolar (29). To test whether the relatively high affinity Ni 2ϩ inhibition of ␣-␤-␥ mENaC was unique among divalent cations, we examined several other divalent cations for their effects on ␣-␤-␥ mENaC. As shown in Fig. 2, external Mg 2ϩ produced a weak inhibition of amiloride-sensitive Na ϩ currents with an estimated K i of 73.5 Ϯ 10.9 mM (n ϭ 5) and a Hill coefficient of 0.47 Ϯ 0.05 (n ϭ 5). Furthermore, current rectification was not observed in the presence of external Mg 2ϩ . The inhibition potency of external Mg 2ϩ on mENaC is similar to that observed in toad bladder (29). We previously reported that external Cd 2ϩ moderately increased amiloridesensitive Na ϩ currents at a concentration of 5 mM (24). Schild et al. (13) reported that inward Li ϩ currents in oocytes expressing ␣-␤-␥ rat ENaC were insensitive to external Zn 2ϩ even in the millimolar range, and a small current reduction (ϳ20%) was shown with 10 mM Zn 2ϩ . We observed no inhibition of amiloride-sensitive Na ϩ currents with 5 mM Zn 2ϩ in the bath solution. However, at low concentrations (0.01-1 mM), Zn 2ϩ produced a moderate potentiation of the Na ϩ currents, similar in amplitude to the increase in Na ϩ currents observed with Cd 2ϩ . 2 Therefore, among divalent cations (Ca 2ϩ , Mg 2ϩ , Ba 2ϩ , Sr 2ϩ , Zn 2ϩ , Cd 2ϩ , and Ni 2ϩ ), Ni 2ϩ appears to be the most potent external inhibitor of ENaC.
DEPC Inhibited ␣-␤-␥ mENaC Currents and DEPC Pretreatment Rendered the Channels Less Sensitive to Ni 2ϩ -Nickel is a transition ion with polarizability between those of hard ions and soft ions. Nitrogen is considered as its most preferred ligand, although sulfur and oxygen can also coordinate Ni 2ϩ . Nickel is often coordinated by nitrogen atoms from histidine residues in nickel enzymes (37)(38)(39), and it is conceivable that Ni 2ϩ binds to histidine residue(s) within ENaC subunits to exert its inhibitory effect on the channel. We therefore examined if the histidine-modifying reagent DEPC had any effect on ␣-␤-␥ mENaC expressed in oocytes and whether DEPC-modified channels exhibited an altered response to Ni 2ϩ . Perfusion  of oocytes with bath solutions containing DEPC at concentrations of 0.1, 1, and 10 mM blocked amiloride-sensitive Na ϩ currents in a dose-dependent manner. The dose-response relationship was satisfactorily fitted with the one-site equation (R 2 ϭ 0.998) (Fig. 3A). The estimated K i was 0.45 Ϯ 0.07 mM and the Hill coefficient was 0.99 Ϯ 0.05 (n ϭ 4). The Na ϩ currents were almost (ϳ94%) completely blocked by 10 mM DEPC. DEPC block was essentially irreversible. In contrast to Ni 2ϩ -induced inward rectification, the amiloride-sensitive Na ϩ currents in the presence of DEPC were linear in the range of Ϫ140 to 60 mV (Fig. 3B). After perfusion of 1 mM DEPC, the remaining currents were essentially insensitive to Ni 2ϩ (Fig.  3C). The currents from oocytes pretreated with DEPC and perfused with Ni 2ϩ did not exhibit rectification (Fig. 3D). These results suggest that histidine residues within ENaC, accessible from extracellular space, participate in Ni 2ϩ inhibition of ENaC.
Ni 2ϩ Inhibition of mENaC Currents Was Attenuated by Mutations at ␣His 282 and ␥His 239 within the Extracellular Domains of ␣ and ␥ mENaC-Our results suggest that Ni 2ϩ is not a pore blocker of ENaC as it produced inward current rectification rather than outward rectification, and the inhibition of currents was only partially reversible following removal of Ni 2ϩ from the bath. The experiments with DEPC suggested that histidine residues might be involved in Ni 2ϩ block of ENaC currents. There are numerous histidine residues within the ECLs of each mENaC subunit, 6 in ␣ mENaC, 11 in ␤ mENaC, and 10 in ␥ mENaC. However, only one histidine is conserved within the ECLs of the three subunits (␣His 381 , ␤His 319 , and ␥His 338 ). These conserved histidine residues were individually mutated to an arginine, and Ni 2ϩ sensitivity of ␣-␤-␥ mENaCs containing a mutation within one subunit was examined and compared with wild type (WT) mENaC. Significant changes in Ni 2ϩ sensitivity were not observed with channels containing a mutation of these conserved histidine residues, suggesting that these conserved histidine residues are not essential for Ni 2ϩ inhibition of ENaC (Table I).
We previously identified an amiloride-binding domain within the ECL of ␣ rat ENaC based on homology with an amiloride-binding site within an anti-amiloride antibody (14,40). Selected mutations within a six-residue tract (residues 278 -283, WYRFHY), including ␣His 282 , altered the amiloride sensitivity of channels composed solely of ␣-subunits reconstituted in planar lipid bilayers, although the amiloride sensitivity of ␣-␤-␥ ENaC containing an ␣His 282 mutation was unchanged (11,17). We examined whether ␣His 282 served as one of the Ni 2ϩ -binding ligands by generating point mutations at this site in ␣ mENaC. As shown in Fig. 4A, all four mutations (␣H282C, ␣H282R, ␣H282W, and ␣H282D) dramatically decreased Ni 2ϩ inhibition as evidenced by the shift in the Ni 2ϩ dose-response curves of the mutant channels to the right with respect to that of WT. The changes in Ni 2ϩ inhibition observed with these mutant channels were ␣H282D Ͼ ␣H282W Ͼ ␣H282R Ͼ ␣H282C. Only the dose-response data for ␣H282C-␤-␥ mENaCs were fitted reasonably well with the one-site equation that yielded parameters as follows: K i , 7.46 mM; Hill coefficient, 0.45; and R 2 , 0.968. No inhibition of amiloridesensitive Na ϩ currents by Ni 2ϩ was observed in oocytes expressing ␣H282D-␤-␥ mENaCs. On the contrary, external Ni 2ϩ induced a significant increase in ␣H282D-␤-␥ currents at concentrations of 0.01, 0.1, and 1 mM, with peak stimulation at 0.1 mM. At higher Ni 2ϩ concentrations, no change in Na ϩ currents was observed when compared with control currents. The data suggest that ␣His 282 participates in the binding of Ni 2ϩ to ENaC. Interestingly, inward current rectification was observed in the presence of Ni 2ϩ with all four mutant channels despite significantly reduced Ni 2ϩ inhibition of the Na ϩ currents (Fig.  4, B-E). For ␣H282D-␤-␥ mENaC, current rectification was evident in the absence of Ni 2ϩ (Fig. 4C).
The above results encouraged us to examine whether an equivalent histidine residue in ␥mENaC (␥His 239 ) was also involved in Ni 2ϩ inhibition. The mutations ␥H239C, ␥H239R, and ␥H239D significantly attenuated Ni 2ϩ inhibition (Fig. 5). Surprisingly, Na ϩ currents in oocytes expressing the mutant channels did not exhibit rectification in the presence of Ni 2ϩ (Fig. 5, B-D), suggesting that ␥His 239 is required for Ni 2ϩinduced current rectification. The double mutations (␣H282R and ␥H239R) eliminated Ni 2ϩ inhibition of Na ϩ currents (Fig. 6A); these channels did not display current rectification (Fig.  6B). These results provided strong evidence that ␣His 282 and ␥His 239 provide Ni 2ϩ -binding sites and are primarily responsible for Ni 2ϩ inhibition of ENaC currents. Introduction of a histidine residue at the corresponding site in ␤ mENaC (␤Q220H) resulted in a modest increase in Ni 2ϩ K i , suggesting that this residue does not have an important role in Ni 2ϩ inhibition of ENaC (Table I).
Introduction of Additional Histidine Residues at Sites Neighboring ␣His 282 Altered Ni 2ϩ Inhibition of ENaC Currents-In nickel-binding proteins, a Ni 2ϩ can be coordinated by one (41), two (42)(43)(44), three (45,46), or four (47) histidine residues, and two histidine residues are often separated by one or three residues (HXH or HXXXH) (37, 48 -50). A sequence "HEXXH" located in an ␣-helix has been considered as a signature metal-binding motif for metalloproteins (45,46,51,52). To gain more knowledge about the putative Ni 2ϩ -binding site, we examined whether the introduction of additional histidine residue(s) in the vicinity of ␣His 282 enhanced Ni 2ϩ inhibition of ENaC currents, presumably by altering Ni 2ϩ binding affinity. Histidine residues were individually introduced at positions ␣Trp 278 , ␣Tyr 279 , ␣Arg 280 , ␣Phe 281 , ␣Tyr 283 , ␣Ile 284 , and ␣Asn 285 . In addition, two ␣-subunit mutants with consecutive 4-histidine tracts, ␣Y279H/R280H/F281H/ H282 (referred to as ␣His 279 -282 ) and ␣H282/Y283H/I284H/ N285H (referred to as ␣His 282-285 ), were also generated. The effects of Ni 2ϩ on these mutant channels are shown in Fig. 7. The mutant channel ␣R280H-␤-␥ showed a significant increase in Ni 2ϩ sensitivity, greatly shifting the high affinity component in the dose-response curve to the left with less effect on the low affinity component compared with WT dose response (Table II). Interestingly, double mutations (␣R280H and ␥H239R) shifted the high affinity component in the doseresponse curve to the right compared with ␣R280H-␤-␥ but to the left compared with WT mENaCs. The low affinity component was shifted to the right compared with both ␣R280H-␤-␥ and WT mENaCs ( Fig. 7E and Table II). Like other mutations at ␣His 282 , ␣R280H did not affect the Ni 2ϩinduced current rectification (Fig. 7F). Three other mutations (␣W278H, ␣Y283H, and ␣N285H) slightly decreased Ni 2ϩ block of the Na ϩ currents, whereas three other mutations (␣Y279H, ␣F281H, and ␣I284H) did not significantly change Ni 2ϩ block (Fig. 7A). As observed with ␣R280H-␤-␥ channels, the mutant channels with a consecutive 4-histidine tract including ␣R280H exhibited a large decrease in the Ni 2ϩ K i (␣His 279 -282 , K i ϭ 0.12 Ϯ 0.03, n ϭ 8). In contrast, the ␣-subunit mutant with a consecutive 4-histidine tract Cterminal to (and including) ␣His 282 exhibited a large increase in the Ni 2ϩ K i (␣His 282-285 , K i ϭ 10.37 Ϯ 0.92 mM, n ϭ 5). These data suggest that ␣R280H, in addition to ␣His 282 and ␥His 239 , participates in Ni 2ϩ binding. Fig. 7G provides three models illustrating the coordination of Ni 2ϩ by these residues, generated based on the changes in Ni 2ϩ inhibition of ENaC currents observed with mutant channels and the preferred coordination geometries of Ni 2ϩ complexes and Ni 2ϩbinding peptides and proteins (see "Discussion").
MTSET Inhibited Amiloride-sensitive Na ϩ Currents in Oocytes Expressing ␣H282C-␤-␥ mENaCs-The above results suggested that a Ni 2ϩ -binding site consisting of ␣His 282 and ␥His 239 has a key role in Ni 2ϩ inhibition of ENaC currents. These histidine residues may also have an important functional role. To explore further the role of ␣His 282 in channel activity, we examined the response of the mutant channels (␣H282C-␤-␥) to external sulfhydryl reagents. The positively charged reagent MTSET irreversibly inhibited amiloride-sensitive Na ϩ currents by 45% when applied externally at the concentration of 1 mM (Fig. 8, A and B). As observed with Ni 2ϩ , inward rectification was observed following MTSET modification of the channel (Fig. 8B). The remaining currents following MTSET modification were less sensitive to Ni 2ϩ (Fig. 8C). MTSET at 1 mM did not alter amiloride-sensitive Na ϩ currents in oocytes expressing WT ␣-␤-␥ mENaCs (24) or ␣-␤-␥H239C mENaCs (Fig. 8A), and pretreatment of the oocytes with the reagent did not alter Ni 2ϩ inhibition of these channels. The negatively charged reagent MTSES [sodium (2-sulfonatoethyl) methanethiosulfonate] at 5 mM also inhibited about 40% of the Na ϩ currents of ␣H282C-␤-␥ channels, whereas it did not change the currents of WT channels (24). A possible explanation of why MTSET failed to inhibit ␣-␤-␥H239C mENaC is illustrated in Fig. 8D. We propose that ␥His 239 is located deeper within a putative Ni 2ϩ -binding pocket than ␣His 282 and is not accessible to MTSET that requires a cylindrical space of 6 ϫ 10 Å (53).
In summary, we observed that amiloride-sensitive Na ϩ currents were blocked by three different reagents. Ni 2ϩ and DEPC blocked WT ENaC. MTSET and MTSES blocked ␣H282C␤␥. Moreover, pretreatment of channels with either DEPC or MTSET resulted in a reduced response to Ni 2ϩ , suggesting that these three reagents interact with a common site within the channel. Furthermore, these results suggest that ␣His 282 and ␥His 239 have an important role in channel function.
Ni 2ϩ Block Was Reduced by Extracellular Acidification but Not by Amiloride-The imidazole rings from histidine residues have pK a values of 6.0 and 14 and may exist in different protonation states including a protonated imidizolium, a neutral imidazole, and a negatively charged imidazolate. Protonated imidizolium cannot coordinate a metal ion (54). Therefore, we examined whether Ni 2ϩ inhibition of ENaC currents was affected by acidification of the bath solution. The Ni 2ϩ inhibitory effect on Na ϩ currents was examined with bath solutions buffered to pH 7.4 with HEPES (buffer range 6.8 -8.2), 5.5 with MES (buffer range 5.5-6.5), or 4.4 with citric acid (buffer range 2.2-6.5). Fig. 9 shows that lowering the bath pH reduced (at pH 5.5) or eliminated (at pH 4.4) the Na ϩ current inhibition by Ni 2ϩ . We observed a biphasic effect of extracellular acidification on amiloride-sensitive Na ϩ currents, a brief stimulation (10% for pH 5.5 and 70% for 4.4) with peak effect at 1-2 min followed by a return to basal current or below basal current. 2 These results are consistent with a recent report (55) and inconsistent with others (56,57). Because of the biphasic pH effects, a Ni 2ϩ dose response could not be accurately obtained. Instead, we compared the effects of 1 mM Ni 2ϩ on the Na ϩ currents at different pH values. As different effects of extracellular acidification were observed in native tissues and oocytes expressing cloned ENaC, the physiological significance of these pH effects is not clear (55).
Residue ␣His 282 is located within a previously identified amiloride-binding domain in the ECL of the ␣ subunit (14,40). We have demonstrated that ␣His 282 , together with ␥His 239 , is involved in Ni 2ϩ inhibition of ENaC. We performed two experiments to examine whether Ni 2ϩ and amiloride compete for the same binding site when both inhibitors were present in the bath solution. First, we determined the response of WT ␣-␤-␥ mENaC to increasing concentrations of amiloride in the presence or absence of an intermediate concentration of Ni 2ϩ (1 mM). Ni 2ϩ did not significantly alter the sensitivity of the channel to amiloride. 2 Second, the response of WT channels to increasing concentrations of Ni 2ϩ was examined in the presence or absence of an intermediate concentration (100 nM) of amiloride. Amiloride did not significantly alter the sensitivity of the channel to Ni 2ϩ . 2 These results suggest that Ni 2ϩ and amiloride do not compete for the same site in ␣-␤-␥ ENaC.
Ni 2ϩ Reduced ENaC Open Probability without Altering Single Channel Current Levels-Reduction of whole-cell Na ϩ currents by external Ni 2ϩ could be due to a decrease in single channel conductance, open probability (P o ) of individual channels, number of functional channels on the cell surface, or electrochemical driving force, or by a combination of these factors. We performed single channel recordings to distinguish these possibilities. The unitary Na ϩ currents were not altered by including 10 mM Ni 2ϩ in the pipette solution (Fig. 10, A and  B). As ENaC open probability is highly variable from patch to patch, it is difficult to determine whether Ni 2ϩ reduces ENaC P o by comparing the P o values from patches performed in the absence and presence of Ni 2ϩ in the pipette solution. We therefore examined channel transitions while Ni 2ϩ was being added to the patch pipette by pipette perfusion. The results from pipette perfusion experiments indicated that Ni 2ϩ reduced NP o of mENaC without effect on unitary Li ϩ current (Fig. 10, C-F). Single channel conductances before and after pipette perfusion of Ni 2ϩ were 8.3 Ϯ 0.4 pS (n ϭ 4) and 8.1 Ϯ 0.3 pS (n ϭ 4), respectively. Since Ni 2ϩ inhibition was not accompanied by a decrease in the reversal potential in oocytes, it was unlikely that external Ni 2ϩ in the millimolar range significantly altered the driving force for Na ϩ influx. Moreover, the density of channel proteins was not likely reduced by external Ni 2ϩ within 1 min. Therefore, our results suggest that Ni 2ϩ inhibition of ENaC currents is due to a reduction of channel P o .

DISCUSSION
Ni 2ϩ Is an External Inhibitor of ENaC-In this study, we found that external Ni 2ϩ inhibited amiloride-sensitive wholecell Na ϩ currents from oocytes expressing ␣-␤-␥ mENaC in a dose-dependent manner. The blocking effect of Ni 2ϩ is in good agreement with that observed by Segal and colleagues (31) who reported that 2.5 mM Ni 2ϩ inhibited 60% of ␣-␤-␥ rENaC currents. Another group recently reported (58) that Ni 2ϩ and other Ca 2ϩ channel blockers partially blocked amiloride-sensitive short circuit currents in fetal rat alveolar type II epithelia. Apical membranes of this type of epithelia are believed to have two types of Ca 2ϩ -activated and amiloride-sensitive channels (28 pS nonselective cation channel and 12 pS Na ϩ channel) whose properties are clearly different from those of ␣-␤-␥ ENaC (59,60). Therefore, the identity of the Ni 2ϩ -blocked channels is not clear, although under certain cultural conditions the predominant channels expressed in these cells show single channel properties similar to those of ENaCs (61). It appears that Ni 2ϩ is the only divalent cation that externally blocks ENaC at submillimolar concentrations. Why is Ni 2ϩ a high affinity blocker of ENaC? A simple explanation is that metal ion-binding site(s) are located in functional domains within the ENaC complex that exhibit higher affinity for Ni 2ϩ than other divalent cations. In metalloproteins, formation of a metal-ligand complex is dependent on many factors including the size, relative charge, polarizability (softness and hardness), preferential coordination geometry of the ion, the liganding donor atoms, and solvent effects (62,63). The ionic radius of the divalent cation does not appear to be the most important determinant in blocking ENaC. A smaller cation Mg 2ϩ (0.65 Å, Pauling (64)) is a much weaker inhibitor of ENaC than Ni 2ϩ (0.72 Å), and a larger cation Cd 2ϩ (0.97 Å) does not inhibit ENaC at a concentration of 5 mM (24). Moreover, Zn 2ϩ (0.74 Å) with very close radius to Ni 2ϩ has no inhibitory effect on ENaC at millimolar concentrations (13). As transition metal ions, Ni 2ϩ and Zn 2ϩ have a number of common properties including size, charge, intermediate polarizability (not as "hard" as Ca 2ϩ and Mg 2ϩ and not as "soft" as Cd 2ϩ and Hg 2ϩ ), and favor nitrogen and sulfur atoms as their binding ligands (62). It is not clear why Zn 2ϩ and Ni 2ϩ do not have similar inhibitory effects on ENaC. It was recently reported that Zn 2ϩ might be a physiological co-activator of the acid-sensing ion channels (111). Therefore, the observed ENaC current stimulation by low concentrations of external Zn 2ϩ suggests that Zn 2ϩ might be a Open bars indicate no significant difference between mutant and WT channels (p Ͼ 0.05), and filled bars indicate the values of mutant channels that are statistically different from those of WT channels (p Ͻ 0.01). D, inward amiloride-sensitive Na ϩ currents measured at Ϫ100 mV in oocytes expressing WT or mutant mENaCs are shown as mean Ϯ S.E. (number of oocytes). The negative sign indicating inward currents was removed for ease of display. The currents were not measured in the same batch of oocytes. E, dose responses of Ni 2ϩ on WT (E), ␣R280H-␤-␥ (Ⅺ), and ␣R280H-␤-␥H239R (‚) mENaCs were examined with nine concentrations of Ni 2ϩ . Relative currents represent mean Ϯ S.E. from 6 to 8 oocytes. Solid lines are from fitting the data with two-site equation. F, I-V curves of amiloride-sensitive Na ϩ currents in the absence (q) and presence of 0.001 (छ), 0.003 (‚), 0.01 (OE), 0.03 (E), 0.1 (ϫ), 1 (ϩ), 10 (ࡗ), 25 (Ⅺ), 50 (f), and 100 (no symbol) mM Ni 2ϩ in the bath solution from an oocyte expressing ␣R280H-␤-␥ mENaCs were made as above. They represent the results from six oocytes. G, proposed Ni 2ϩ coordination models for WT (left), ␣R280H (middle), and ␣R280H/␥H239R mENaCs (right). For WT channels, Ni 2ϩ is coordinated by three nitrogen atoms from two ␣His 282 and one ␥His 239 and the fourth ligand (water oxygen). For ␣R280H-␤-␥ mENaCs, the introduced histidines provide two additional ligands besides the four ligands in WT. A solvent (water) replaces the eliminated ␥His 239 in ␣R280H-␤-␥H239R channels. high affinity stimulator of ENaC. It has been known for a long time that sulfhydryl group-reactive reagents, including Cd 2ϩ , stimulate epithelial Na ϩ transport, possibly by relieving Na ϩ self-inhibition; however, its physiological significance and mechanism are still unknown (1). The inhibition of ENaC currents by external Ni 2ϩ could be due to a direct blocking mechanism (pore plugging) or an indirect one, often referred to as an allosteric mechanism. The inward current rectification observed in the presence of Ni 2ϩ does not fit with a typical voltage-dependent block of a cation channel from the external side. For example, current-voltage relationship curves in the presence of submaximal concentrations of amiloride show outward rectification, indicating a reduced block of ENaC when the membrane is depolarized and consistent with the idea that amiloride is an external pore blocker of ENaC (1). The poor reversibility of Ni 2ϩ inhibition of the ENaC currents suggests a complex feature of the blocking mechanism and is inconsistent with the notion of a direct pore plugging. We favor the idea that Ni 2ϩ is an external inhibitor of ENaC but does not directly interact with the pore.
It has been known that external Ni 2ϩ affects a variety of ion channels including voltage-gated Ca 2ϩ channels (65-68), voltage-gated K ϩ channels (69 -71), voltage-gated Na ϩ channels (72), voltage-gated H ϩ channels (73-75), CNG channels (76), P2X receptors (77, 78), ␥-aminobutyric acid type A receptors FIG. 9. Ni 2؉ inhibition of ENaC was reduced by extracellular acidification. The effects of 1 mM external Ni 2ϩ on amiloridesensitive Na ϩ currents at three external pH values (7.4, 5.5, and 4.4) were examined in oocytes expressing ␣-␤-␥ mENaCs. Bars represent relative currents calculated by normalizing the amiloride-sensitive Na ϩ currents in the absence (open bars) and presence (solid bars) of 1 mM Ni 2ϩ to the current levels in the absence of Ni 2ϩ in each bath solution with the corresponding pH value. Data are mean Ϯ S.E., and the numbers of observations are indicated in the parentheses. The * indicates the absolute currents in the presence of 1 mM Ni 2ϩ are significantly lower than the absolute control currents with pH 7.4 and 5.5 (p Ͻ 0.001). The difference between the control currents and currents in the presence of 1 mM Ni 2ϩ at pH 4.4 is not statistically significant (p Ͼ 0.05). Triangles indicate that the relative currents with 1 mM Ni 2ϩ at pH 5.5 and 4.4 are significantly higher than the relative current with 1 mM Ni 2ϩ at pH 7.4.

FIG. 8. MTSET reduced Na ؉ currents in oocytes expressing ␣H282C-␤-␥ mENaC and further reduced Ni 2؉ inhibition of the remaining currents.
A, effects of MTSET on ␣H282C-␤-␥ and ␣-␤-␥H239C mENaCs. Amiloride-sensitive Na ϩ currents were measured at Ϫ100 mV before, 2 min after perfusion of 1 mM MTSET prepared in the bath solution, and 2 min after washout of MTSET from the bath. Relative currents were obtained by normalizing the currents to the levels before application of MTSET. Open, filled, and shaded bars are the relative currents (mean Ϯ S.E., n ϭ 5 for ␣H282C-␤-␥ and n ϭ 13 for ␣-␤-␥H239C) obtained before and after MTSET and after washout of MTSET, respectively. B, I-V curves of amiloride-sensitive Na ϩ currents before (E) and after (Ⅺ) MTSET and after washout of MTSET (‚) were generated as before. Data are mean Ϯ S.E. from five oocytes. C, Ni 2ϩ dose responses on ␣H282C-␤-␥ mENaC currents without (E) and with (q) treatment of the oocytes with 1 mM MTSET. The dose response of Ni 2ϩ on ␣H282C-␤-␥ mENaCs without MTSET is the same as in Fig. 5A. Relative currents in solid circles (mean Ϯ S.E., n ϭ 5) were obtained by normalizing the currents in the presence of Ni 2ϩ to the current levels following 1 mM MTSET perfusion for 3 min and washout of the reagent for 2 min, and immediately prior to Ni 2ϩ application. Dashed lines are from fitting the data with the one-site equation. D shows one of the possible reasons why MTSET inhibited ␣H282C-␤-␥ mENaC without effect on ␣-␤-␥H239C mENaC. The shaded area shows part of the ENaC ECLs with a semicircle pocket designated as a Ni 2ϩ -binding site. The three open rectangles represent three Ni 2ϩ -binding ligands out of a possible total number of four or six ligands. Side chains of ␣His 282 and ␥His 239 provide three of the ligands. A circle inside the pocket represents a bound Ni 2ϩ ion to WT channels. One molecule of MTSET is drawn with space-filled model, and its thiol head is placed at the entrance of the pocket. Attachment of MTSET to sulfhydryl group of ␣H282C would prevent Ni 2ϩ from entering the binding site to cause inhibition of the channel currents. The ϪSH group of ␥H239C may be too deep for interaction with MTSET. (79), and glutamate receptors (80,81). These channels display different sensitivities to Ni 2ϩ with inhibitory constants in the range of micromolar to millimolar. The mechanism of Ni 2ϩ effects on these channels is not clear, although both direct and allosteric blocking mechanisms have been suggested (67,69,70,77,82). Xenopus oocytes express several kinds of endogenous channels whose activity is normally small compared with that of overexpressed foreign channels (83). However, expres-sion of selected exogenous proteins has been reported to activate certain endogenous channels that are normally rather "silent" (84,85). Therefore, it is possible that the expressed whole-cell currents in oocytes may contain a significant level of endogenous channel activity that is sensitive to both external Ni 2ϩ and amiloride if ENaC expression activates these channels. This is unlikely the case under our experimental conditions. First, there is no clear evidence that ENaC expression in Xenopus oocytes activates endogenous channels. When extracellular Na ϩ is replaced by K ϩ or Ca 2ϩ , no inward currents were detected in oocytes expressing ENaC that are clamped at Ϫ100 mV (16,18,19). These data indicated that endogenous channels do not provide a significant contribution to the wholecell currents measured in ENaC-expressed oocytes under our typical clamping protocols (i.e. protocols that are not designed to elicit voltage-dependent currents). Second, we evaluated the Ni 2ϩ effect on ENaC currents at a clamp potential of Ϫ100 mV, a potential at which many voltage-gated channels should not be activated. The linear I-V relationship in the range of Ϫ140 to 60 mV of the currents measured in ENaC-expressed oocytes indicated a lack of voltage-gated current. Moreover, the observed Ni 2ϩ inhibition of ENaC currents was specifically eliminated by ␣H282D mutation and by the double (␣H282R/␥H239R) mutation, which strongly suggests that the whole-cell current reduction is due to Ni 2ϩ interaction with ENaC rather than with other channels. It was recently reported that external Ni 2ϩ (less potent than Cd 2ϩ and Zn 2ϩ ) blocked a slowly activating Na ϩ current elicited by sustained depolarization of the Xenopus oocyte membrane, a current that is clearly different from ENaC currents (86).
Whether Ni 2ϩ has any physiological role in channel regulation in humans is unclear, although it is known that Ni 2ϩ is essential for bacteria and plant growth (87)(88)(89). As nickel is considered a hazardous ion to human, the inhibitory effect of Ni 2ϩ on ENaC activity may contribute to its toxic effects on kidneys, lungs, and digestive system where ENaC is expressed. We anticipate that Ni 2ϩ will be a useful tool to study the structure and function of epithelial Na ϩ channels. Nickel has been used successfully as a pharmacological tool to distinguish different types of voltagegated Ca 2ϩ channels and to gain useful information regarding gating mechanisms of CNG channels (90,91).
Histidine Residues ␣His 282 and ␥His 239 within the Extracellular Loops of ENaCs Form the Ni 2ϩ -binding Site-The histidinereactive reagent DEPC inhibited amiloride-sensitive Na ϩ currents with an estimated K i of 0.45 mM, suggesting that histidine residue(s) in ENaC complex are located in a functional domain. The potential role of histidyl groups in amiloride-sensitive Na ϩ channels was proposed nearly 20 years ago based on the identification of a titratable group with a pK a of 6.7 and DEPC inhibition of short circuit current in toad urinary bladder (92). Moderate DEPC treatment (1 mM) of oocytes expressing WT mENaCs significantly reduced Ni 2ϩ inhibition on the remaining currents, suggesting that Ni 2ϩ and DEPC act at a common site.
Our mutagenesis studies suggest that ␣His 282 and ␥His 239 are the primary structural determinants for Ni 2ϩ inhibition of mENaC currents. However, there were distinct changes associated with mutations at these two sites. The ␥-mutation-induced changes in Ni 2ϩ inhibition followed the order: ␥H239R Ͼ ␥H239D Ͼ ␥H239C, consistent with the preferential coordination of Ni 2ϩ . Cysteine residues coordinate Ni 2ϩ in hydrogenase, dehydrogenase, and deformylase enzymes (38,39,42,44). Aspartic acid participates in Ni 2ϩ coordination in urease (37). The relationship between substituted residues at ␣His 282 and the degrees of alteration in Ni 2ϩ inhibition did not follow the order observed with ␥His 239 . The largest change in Ni 2ϩ dose response on mENaC currents was observed with ␣H282D (Fig. 4A). The most FIG. 11. Sequence alignments, structure prediction, and working models for the Ni 2؉ -binding domain and mechanism of Ni 2؉ inhibition. A, sequence alignments were performed with Vector NTI 7.0 (InforMax Inc., Bethesda) from sequences as follows: ␣, ␤, and ␥ mouse ENaCs (GenBank TM accession numbers AF112185, AF112186, and AF112187); ␣, ␤, and ␥ rat ENaCs (GenBank TM accession numbers X70497, X77932, and X77933); ␣, ␤, and ␥ human ENaCs (GenBank TM accession numbers L29007, L36593, and L36592); ␣ bovine ENaC (GenBank TM accession number U14944); and ␦ human ENaC (GenBank TM accession number U38254). The first amino acid residue number for each sequence is listed in parentheses. Homology is presented in the following colors codes: identical, red background; similar, yellow background; and block of similar residues, green background. The histidine residues studied in this report are in boldface type. B, a linear model is shown to indicate the location of the histidine residues (␣His 282 and ␥His 239 , as a red sphere) in one ENaC subunit. Secondary structural predictions were performed on the entire amino acid sequence of ␣mENaC (C) or ␥mENaC (E) with five different methods and only predictions of ␣Cys 263 -Pro 291 and ␥Cys 216 -247 striking difference observed with the mutations ␣His 282 and ␥His 239 was that all channels with mutations at ␥His 239 showed no Ni 2ϩ -induced current rectification (Fig. 5, B-D), whereas all channels with ␣-subunit mutations (␣His 282 or ␣Arg 280 ) showed inward rectification in the presence of Ni 2ϩ (Fig. 4, B-E, and Fig.  7F). These results suggest that the interaction of Ni 2ϩ with ␥His 239 is required for Ni 2ϩ -induced current rectification.
Our results suggest that ␣His 282 and ␥His 239 are involved in Ni 2ϩ binding to the channel complex. However, other residues may be needed to complete the Ni 2ϩ coordination geometry as Ni 2ϩ is coordinated by 4, 5, or 6 ligands in most Ni 2ϩ -binding proteins (89). In order to identify other residues involved in Ni 2ϩ binding and to probe the secondary structure of residues near ␣His 282 , one or three additional histidines were engineered in the tract ␣WYRFHYIN (residues 278 -285) of mENaC. Among 7 single histidine substitutions in this tract, ␣R280H was the only mutation associated with enhanced Ni 2ϩ inhibition of channel currents, suggesting that the introduced histidine at ␣280 contributed to Ni 2ϩ coordination (Fig. 7G). The enhanced Ni 2ϩ inhibition by ␣R280H was attenuated by mutation of ␥His 239 (␣R280H-␤-␥H239R). Although these data suggest that ␣Arg 280 and ␣His 282 might be located in a ␤-sheet structure, other secondary structures can also place these two residues in close proximity favoring metal coordination by both residues, such as helix, 3 10 helix, and random coil. An ␣-helical structure containing both residues would place ␣Arg 280 and ␣His 282 on opposing faces of the helix (Fig. 11D) and is not consistent with the observed enhancement of Ni 2ϩ inhibition by ␣R280H. Another possibility is that ␣Arg 280 and ␣His 282 are not located in the same secondary structure, in agreement with structural predictions that place ␣Arg 280 within an ␣-helix and ␣His 282 immediately adjacent to the ␣-helix (Fig 11C). We generated a structural model of a Ni 2ϩ -binding site (Fig. 11G) based on consensus structural predictions of the regions flanking ␣His 282 and ␥His 239 (Fig. 11, C and E) and our results from mutagenesis studies. The Ni 2ϩ -binding domain in the ␣-subunit was modeled as an N-terminal ␣-helix followed by a ␤-sheet with ␣His 282 at the transition point. Our model places ␣R280H and ␣His 282 at positions that allow both residues to participate in Ni 2ϩ coordination. This proposed arrangement of ␣R280H and ␣His 282 is strikingly similar to the arrangement of a pair of metal-coordinating histidine residues, His 112 within an ␣-helix and His 110 adjacent to the helix in the Ni 2ϩ -binding metallochaperone UreE (48). The corresponding region in ␥ mENaC was modeled as an ␣-helix based on secondary structure predictions.
The dose-response relationship of Ni 2ϩ inhibition of Na ϩ currents of WT mENaC was best described with an equation providing two classes of binding sites for Ni 2ϩ (Fig. 1). For several mutations, such as ␣R280H, ␣H282C, and ␥H239C, the high affinity component of the Ni 2ϩ dose-response curve appeared to be shifted more than the low affinity component (Figs. 4A, 5A, and 7). Ni 2ϩ dose-response curves for other mutant channels (except ␣H282D) showed similar shifts in both the high and low affinity binding sites. Although these data do not allow us to distinguish whether ␣His 282 and ␥His 239 form independent Ni 2ϩ -binding sites or form a single binding site, we favor the view that these histidine residues contribute to a common Ni 2ϩ -binding site. Substitution of ␣His 282 or ␥His 239 with cysteine or arginine produced similar shifts in the Ni 2ϩ dose-response curves, and channels with mutations at both sites (␣H282R-␤-␥H239R) were Ni 2ϩ -insensitive. These data suggest collaborative coordination of Ni 2ϩ binding by these histidine residues. The elimination of Ni 2ϩ inhibition by point mutation ␣H282D, the enhanced Ni 2ϩ inhibition by ␣R280H, and partial reversal of the enhancement by a mutation in ␥ENaC (␥H239R) also support this view. Intersubunit coordination of Ni 2ϩ has been demonstrated in CNG channels and the Ni 2ϩ -binding metallochaperone UreE (47,48,76,91). Alternatively, the high and low affinity sites for Ni 2ϩ may be formed by ␣His 282 (or more than one ␣His 282 ) and ␥His 239 , respectively. The relative shift of the high affinity component of the dose-response relationship by mutations at ␥His 239 appeared to be smaller than shift of the low affinity component (Fig. 5A), whereas a greater shift of the high affinity component of the dose-response relationship was observed with ␣His 282 mutations (Fig. 4A). As the osmolarity of Ni 2ϩ -containing bath solutions was not adjusted in this study and hypertonicity may affect ENaC currents (93,94), we cannot rule out the possibility of osmotic effects on ENaC or endogenous channels at high concentrations of Ni 2ϩ (25 and 50 mM). The observations that ␣H282D or ␣H282R/␥H239R eliminated both high and low affinity inhibition are not consistent with the possibility.
Based on above results, preferences of Ni 2ϩ coordination, and assuming that the channel complex contains more than one ␣-subunit (10 -12), we propose a Ni 2ϩ -binding site within WT ENaC where Ni 2ϩ is coordinated by four ligands, three nitrogen atoms from two ␣His 282 residues and one from ␥His 239 . A fourth unknown ligand could be an oxygen atom from solvent (water) or an oxygen-bearing residue (Figs. 7G and 11G). Although we do not have direct evidence for the involvement of two ␣His 282 residues in Ni 2ϩ coordination, the loss of Ni 2ϩ block observed with ␣H282D-␤-␥ channels, compared with the modest change in Ni 2ϩ sensitivity observed with ␣-␤-␥H239D channels, is consistent with the notion that more than one ␣-subunit participates in the coordinated binding of Ni 2ϩ . Our proposed Ni 2ϩ coordination pattern is similar to that of the endonuclease domain of the bacterial toxin colicin E9 and the are shown. GOR4 (115), PHD (116,117), and Predator (118) predictions were done at the web site "Network Protein Sequence Analysis" (npsa-pbil.ibcp.fr). The predictions are presented as the following single letters: h for ␣-helix; e for extended strand; and c for random coil. NNPredict was done at the web site (www.cmpharm.ucsf.edu/ϳnomi/nnpredict.html) (119,120). A dash means no prediction by NNPredict. The four lines above the "Consensus" are predictions by DNASis software 2.6 for Windows (Hitashi Software Engineering Co., Ltd., South San Francisco, CA) using Chou-Fasman algorithm (121). The letters H (h), S (s), and T represent ␣-helix, ␤-strand, and turn, respectively. Uppercase letters indicate probability, and lowercase letters indicate possibility. Helical wheel analyses of ␣ and ␥ of mENaC were performed on ␣Val 272 -His 282 (D) and ␥Asn 230 -His 239 (E). Residue ␣His 282 is predicted beyond the helical region by most methods. It is included in the analysis to show that ␣His 282 and ␣Arg 280 would be on the opposite faces of the helix if ␣His 282 were considered within the helix. G, a structural model for Ni 2ϩ -binding site was generated based on the secondary structure predictions and mutagenesis results from this study (see text under "Discussion"). The model was built with HyperChem 6.03 (Hypercube, Inc., Gainesville, FL). Backbones are displayed as 9 thin green lines, and residues are shown in sticks. Colors for carbon, oxygen, nitrogen, and sulfur atoms are gray, red, blue, and yellow. For clarity, only selected residues are labeled with the one-letter symbol and sequence number. Two ␣ENaC domains on the top were modeled as ␣-helix (Val 272 -Phe 281 ) and ␤-sheet (His 282 -Asn 285 ), and the ␥ ENaC domain was built as a long helix on the bottom (Asn 230 -Gln 246 ). A Ni 2ϩ is shown as green sphere and a solvent (water) is shown as a red sphere. H, working models for Ni 2ϩ inhibition of ENaC. The "gate" model (left) places the Ni 2ϩ -binding site (Ni 2ϩ as green sphere) within a putative gate that swings into the outer vestibule of the pore during channel closing. The transduction model (right) proposes that Ni 2ϩ binding (green sphere) causes a local conformational change near the binding site that is remotely transmitted to the outer vestibule of the channel pore through regions connecting the binding site and the pore. Yellow arrows indicate movement directions, and the red sphere represents one Na ϩ entering the ENaC pore from outside. The pore is shown in blue, and the portions of ECLs are drawn in green or orange. zinc endopeptidase astacin. Both proteins utilize three nitrogen atoms from histidine residues and one oxygen atom from either a solvent or a tyrosine residue as Ni 2ϩ ligands (45,49).
External Ni 2ϩ Inhibits ENaC by Reducing Channel Open Probability-Our single channel results indicate that external Ni 2ϩ decreased the open probability of ␣-␤-␥ mENaC without affecting single channel currents. This notion is consistent with the recent observation that Ni 2ϩ decreased apparent open channel density without changing the single channel current as determined by noise analysis (31). A decrease in channel number is less likely a mechanism for Ni 2ϩ inhibition of ENaC currents, given the rapidity of the current response to Ni 2ϩ (Fig. 1D) and the lack of change in oocyte capacitance in response to Ni 2ϩ (31).
The open probability of ENaC has been shown to be highly variable in different systems (1,95). ENaCs display slow gating with long open and close times, and in some cases two or more distinct gating modes with high and low P o values were observed. ENaC may naturally transition among different open states that can be regulated by various factors (96,97). We observed that Ni 2ϩ inhibition was saturated at the concentrations near 50 mM (ϳ100-fold greater than the estimated K i for Ni 2ϩ ) and did not eliminate amiloride-sensitive Na ϩ currents. These data are consistent with the view that Ni 2ϩ binding to ENaC results in a conformational change that shifts channels to a gating mode with a lower P o or, alternatively, that Ni 2ϩ stabilizes a closed conformation and thus reduces ENaC P o . Ni 2ϩ -induced conformational changes in proteins have been observed in Ni 2ϩ -dependent enzymes and other native proteins or synthetic peptides (41,98,99). Fig. 11H illustrates two potential mechanisms by which Ni 2ϩ binding leads to a decrease in ENaC P o , based on our presumption that the Ni 2ϩ -binding site is located within the extracellular region outside of the conduction pore. One model (the "gate" model) places the Ni 2ϩ -binding site within a putative gate that swings into the outer vestibule of the pore during channel closure. Ni 2ϩ binding stabilizes the gate in the closed channel state. An extracellular gate has been proposed to explain state-dependent accessibility of an introduced cysteine within the pore region to sulfhydryl reagents (23). The second model (the "transduction" model) proposes that Ni 2ϩ binding induces a local conformational change near the binding site that is transmitted to the outer vestibule of the channel pore through a "linker" region connecting the Ni 2ϩ -binding site and the pore. Ni 2ϩ binding to the ECLs ultimately results in conformational changes in the outer pore that favor the closed state, which is associated with a decreased P o . The second model is similar to the gating mechanism proposed for ionotropic glutamate receptors (100,101).
␣His 282 and ␥His 239 May Be Located within a Gating Domain in the ECLs of ENaC Subunits-A number of observations suggest that ␣His 282 and ␥His 239 are located within domains that participate in the control of ENaC gating. Ni 2ϩ decreased ENaC open probability (see above and Ref. 31) through interactions with ␣His 282 and ␥His 239 . MTSET inhibition of ␣H282C-␤-␥ mENaC may be due to a decrease in P o , although we have not determined channel P o following MTSET treatment. Furthermore, we observed that ␣H282R-␤-␥H239R mENaC has a high P o , 2 and Applebury et al. (102,103) have reported that mutations within the WYRFHY tract (residues 278 -283 in rat ␣ENaC) altered gating kinetics of ␣-subunit channels expressed in Chinese hamster ovary cells. Primary structure analyses also support the notion that ␣His 282 and ␥His 239 are located in important functional domains. These tracts are among the most conserved domains within ENaC subunits (Fig. 11A) but are not found in the ECLs of other members of the ENaC/Deg family. Secondary structural pre-dictions suggest that these domains exhibit a high degree of organization, including regions that are predicted to form ␣-helices or ␤-sheets (Fig. 11, C and E). Helical wheel analyses of the predicted ␣-helical regions revealed that charged or highly polar residues are arranged on one face of the helix, whereas hydrophobic residues are located on the opposing face (Fig. 11, D and F), a structure capable of interacting with different environments (i.e. aqueous or lipophilic) or domains.
The potential role of histidine residues in ENaC gating is consistent with the role of histidine residues in mediating conformational changes in enzymes (104 -107) or ion channels. Okada et al. (108) proposed protonation of a histidine (His 37 ) in the M2 ion channel protein from influenza A virus, and its interaction with Trp 41 is a key step in channel activation. The role of histidine residues in CNG gating has been well characterized (91).
The mechanism by which domains containing ␣His 282 and ␥His 239 affect ENaC gating are unclear and deserve further investigation with methods involving structural approaches. Information regarding ENaC gating mechanisms is emerging through mutagenesis studies. Previous studies have shown that ENaC gating is affected by mutations in several different domains in epithelial Na ϩ channel subunits including the N-terminal domain (21,22,109), the pore region (23,24), the C-terminal part of the ECL (25), the M2 domain (26), and the intracellular Cterminal domain (27,110). However, it is unclear whether these different regions exert effects on channel gating in an independent manner or whether interactions between these domains influence channel gating through a common pathway.
The proposed role of ECLs in ENaC gating is consistent with the notion that ECL domains within ENaC-related channels are involved in channel gating. ASIC are activated by extracellular acidification, and a histidine (His 72 ) in the ECL of ASIC-2a has been identified as a putative H ϩ sensor (111). FaNaCh is a peptide-gated Na ϩ channel, and the peptide (FM-RFamide) binding domain was proposed to reside within the ECL (112). Residues or domains within the ECLs of subunits of mechanotransducing channels in Caenorhabditis elegans (i.e. Deg-1 and Mec-4) were proposed to modulate channel gating (113,114). ENaC may share a similar design in gating machinery with other members in the ENaC/DEG family, although it is clear that they are gated by different factors.
In summary, we have characterized Ni 2ϩ inhibition of ENaC activity at whole-cell and single channel levels, identified ␣His 282 and ␥His 239 as primary Ni 2ϩ -binding residues, and demonstrated that Ni 2ϩ reduces channel P o . Our results also suggest that these two residues are present in domains within the extracellular loops that are associated with channel gating.