Thumb domains of the three epithelial Na+ channel subunits have distinct functions

The epithelial Na+ channel (ENaC) possesses a large extracellular domain formed by a β-strand core enclosed by three peripheral α-helical subdomains, which have been dubbed thumb, finger, and knuckle. Here we asked whether the ENaC thumb domains play specific roles in channel function. To this end, we examined the characteristics of channels lacking a thumb domain in an individual ENaC subunit (α, β, or γ). Removing the γ subunit thumb domain had no effect on Na+ currents when expressed in Xenopus oocytes, but moderately reduced channel surface expression. In contrast, ENaCs lacking the α or β subunit thumb domain exhibited significantly reduced Na+ currents along with a large reduction in channel surface expression. Moreover, channels lacking an α or γ thumb domain exhibited a diminished Na+ self-inhibition response, whereas this response was retained in channels lacking a β thumb domain. In turn, deletion of the α thumb domain had no effect on the degradation rate of the immature α subunit as assessed by cycloheximide chase analysis. However, accelerated degradation of the immature β subunit and mature γ subunit was observed when the β or γ thumb domain was deleted, respectively. Our results suggest that the thumb domains in each ENaC subunit are required for optimal surface expression in oocytes and that the α and γ thumb domains both have important roles in the channel's inhibitory response to external Na+. Our findings support the notion that the extracellular helical domains serve as functional modules that regulate ENaC biogenesis and activity.

Epithelial Na ϩ channel (ENaC) 3 -mediated Na ϩ transport regulates extracellular fluid volume and K ϩ homeostasis, as well as blood pressure (1). Three homologous subunits typically constitute a functional trimeric ENaC complex (1)(2)(3)(4). Distinct from most ion channels composed of subunits with two transmembrane domains, ENaC/Degenerin family members possess a large extracellular domain (ECD) composed of over 400 residues per subunit. The resolved structure of an acid-sensing ion channel (ASIC1), a member of the ENaC/Degenerin family, revealed a highly organized ECD with five distinct subdomains. These include a central core formed by ␤ strand-structured palm and ␤-ball domains, and peripheral ␣-helical thumb, finger, and knuckle domains. Sequence similarities suggest that the ECDs of other members of the ENaC/Degenerin family have a similar structure (2)(3)(4)(5). Previous studies also suggest that the well-organized and tightly packed ECDs sense a variety of external cues that modulate channel open probability (6 -18).
ENaCs in both native tissues and heterologous expression systems are constitutively open, albeit with considerably variable open probability. The latter can be attributed to factors that regulate ENaC gating, including ions, pH, proteases, temperature, mechanic forces, acidic phospholipids, and palmitoylation (4). ENaC has been suggested to function as a ligandregulated ion channel (9,19). ENaC open probability is reduced by extracellular Na ϩ , a process referred to as Na ϩ self-inhibition (7,8,18,20). Many of the identified residues where substitutions either enhance or suppress Na ϩ self-inhibition are located within the ECDs, including a region in the ␣ subunit containing a putative Na ϩ -binding site that mediates the inhibitory effect of external Na ϩ (18). We hypothesized that the independently folded helical domains (thumb, finger, and knuckle) serve as functional modules to regulate gating and mediate the interactions of ENaC with various factors, including extracellular Na ϩ . In support of this hypothesis, we reported that deletion of the ␣ subunit knuckle domain leads to hyperactive channels with a loss of the Na ϩ self-inhibition response, whereas deletion of either the ␤ or ␥ subunit knuckle domain diminished channel expression and activity (21,22). In this study, we sought to determine specific functional roles of the helical thumb domains within ENaC subunits. Our data indicate that the thumb domain supports various ENaC-specific activities. Surprisingly the thumb domain in each subunit exhibits distinct functional roles. These results emphasize the complex inter-and intra-protein interactions to which this regulated channel is exposed.

Deletion of the thumb domain alters ENaC activity in a subunit-specific manner
The thumb domain in ASIC1 is composed of two ␣ helices (␣4 and ␣5) connected by a loop and by a series of five disulfide bridges. Two additional loops connect the ␣ helices to adjacent ␤ strands of the palm domain (␤9 and ␤10) (2). The disulfidebridged Cys residues are strictly conserved among ENaC/Degenerin family members. In this study, we deleted all 86 to 88 residues between (and including) the most proximal and distal Cys residues in the thumb domain in each mouse ENaC subunit, leaving an 8-residue loop that provides a transition between the ␤9 and ␤10 strands of the palm domain (Fig. 1). To examine the effects of thumb domain deletions, we expressed either wildtype (WT) ENaC subunits (␣, ␤, and ␥) or two WT subunits together with a mutant subunit lacking a thumb domain (␣⌬T, ␤⌬T, or ␥⌬T) in oocytes. Amiloride-sensitive whole cell Na ϩ currents measured in oocytes expressing channels with either ␣⌬T or ␤⌬T were dramatically less than in cells expressing WT ENaC (normalized currents, 0.02 Ϯ 0.02 for ␣⌬T, 0.07 Ϯ 0.03 for ␤⌬T, mean Ϯ S.D., p Ͻ 0.001 versus WT, n ϭ 29 -30, Fig. 2). Currents in oocytes expressing channels with ␥⌬T were similar to WT channels (0.89 Ϯ 0.39, p Ͼ 0.05, n ϭ 30, Fig. 2).
As the thumb deletion constructs retained an 8-residue loop to link the ␤9 and ␤10 strands, it was possible that these linking residues contribute to the functional changes described above. We generated a set of thumb deletions in which all thumb domain residues, including the 8-residue linking tracts, were deleted. Channels with an individual subunit bearing a thumb deletion lacking an 8-residue linker exhibited levels of activity in oocytes similar to channels with a thumb deletion bearing an 8-residue linker. Normalized currents were 0.03 Ϯ 0.04 (p Ͻ 0.001 versus WT, n ϭ 39) for ␣ mutant, 0.07 Ϯ 0.06 (p Ͻ 0.001 versus WT, n ϭ 38) for ␤ mutant, and 1.12 Ϯ 0.93 (p Ͼ 0.05 versus WT, n ϭ 75) for ␥ mutant. These observations suggest that both the ␣ and ␤ subunit thumb domains are required for maximal functional ENaC expression in oocytes.

Thumb domain deletions reduce ENaC surface expression
To determine whether a thumb domain deletion in the ␣ or ␤ subunit reduced whole-cell currents by reducing channel expression at the plasma membrane, we examined surface protein levels of WT and mutant channels with a chemiluminescence-based assay, as previously described (22,23). As shown in Figure 1. Thumb domains in monomeric and trimeric ASIC1 models. A, a monomer ASIC1 structural model is shown as ribbons with individual domains colored and labeled (TM, transmembrane domain). Seven disulfide bridges are numbered and displayed as space-filled spheres. The beginnings and ends of the two sets of thumb domain deletions are identified by a red knife and a black arrowhead. B, a trimeric ASIC1 model is shown as three colored ribbons with the deletion targets as surface rendering. Both monomeric and trimeric ASIC1 models were built from the coordinates of the closed state structure of chicken ASIC1 (PDB entry code 5WKV (55)), using PyMol (59). C, sequence alignments showing the targeted residues in the two sets of thumb domain deletions. Alignments were done with Vector NTI 11.0 (Invitrogen) with manual adjustment from protein sequences with GenBank TM accession numbers AAD21244 (␣mENaC), AAD21245 (␤mENaC), AAD21246 (␥mENaC), and AY956393 (cASIC1). The first residue number in each line of sequence is shown in parentheses. Identical and conserved residues are shown in red letters on yellow background and black letters on green background, respectively. The percentage of identical residues between a mouse ENaC subunit and cASIC1 is 22-24 for the full-length subunits and 17-24 for the thumb domains alone. The black arrowhead and red knife show the beginnings of two sets of thumb domain deletion.

Figure 2.
␣ or ␤ subunit thumb domain deletions significantly reduce ENaC activity. Whole cell Na ϩ currents were measured in oocytes injected with WT or mutant mouse ␣␤␥ cRNAs (2 ng/subunit) at a clamping voltage of Ϫ100 mV (intracellular potential). Mutant channels contained one subunit with a thumb domain deletion (⌬T) and two WT subunits. A, representative current recordings in the absence or presence of 10 M amiloride from oocytes expressing ␣␤␥ (WT), ␣⌬T␤␥, ␣␤⌬T␥, or ␣␤␥⌬T ENaCs. The traces are superimposed with the same scale bars. B, scatter plots of normalized currents in groups. Normalized currents represent amiloride-sensitive currents from individual oocytes, divided by the mean of the amiloride-sensitive currents in oocytes expressing WT channels from the same batch of oocytes. Data were collected in two batches of oocytes. The amiloride-sensitive currents from the two batches of oocytes expressing WT ENaC were Ϫ3.8 Ϯ 1.6 A (mean Ϯ S.D., n ϭ 15) and Ϫ5.6 Ϯ 1.6 A (n ϭ 15). Numbers of oocytes for each group are shown below data points. Horizontal bars are mean values. Statistical significances between mutant and WT channels were obtained from one-way ANOVA followed by a Tukey post hoc test (***, p Ͻ 0.001, n.s., no significant difference). Fig. 3, channels with ␣⌬T or ␤⌬T had levels of surface expression that were markedly lower compared with WT channels (relative expression, 0.06 Ϯ 0.04 for ␣⌬T, 0.09 Ϯ 0.11 for ␤⌬T, p Ͻ 0.001 versus WT, n ϭ 46 -53). In contrast, channels with ␥⌬T had levels of surface expression that were moderately, but significantly lower than WT channels (0.70 Ϯ 0.46, p Ͻ 0.05 versus WT, n ϭ 46).

Deletion of either the ␣ or ␥ subunit thumb domain reduces the Na ؉ self-inhibition response
ENaC lacking a ␥ subunit thumb domain had current levels similar to the WT channel, but there were also lower levels of surface expression. This discrepancy suggests that the mutant channel has a higher channel open probability, compared with WT. A key factor regulating ENaC open probability is Na ϩ self-inhibition. We have previously shown that, for selected mutants, channel open probability correlates with the magnitude of the Na ϩ self-inhibition response (24). Therefore, we hypothesized that channels with a thumb domain deletion in the ␥ subunit have a reduced Na ϩ self-inhibition response. Indeed, ␥⌬T channels showed a significantly reduced Na ϩ selfinhibition response, as shown by an increase in the ratio of the steady state to peak current (I ss /I peak , 0.93 Ϯ 0.04 (␥⌬T) versus 0.65 Ϯ 0.08 (WT), p Ͻ 0.001 versus WT, n ϭ 24, Fig. 4, C and F).
It was formally possible that oocytes injected with ␣␤␥⌬T cRNAs express both channels with three subunits (i.e. ␣␤␥⌬T) as well as channels lacking the ␥⌬T subunit (i.e. ␣␤). Rat ␣␤ channels have been reported to have a very high open probability in the presence of Ͼ100 mM extracellular Na ϩ or Li ϩ (25,26), indicating that ␣␤ channels lack a Na ϩ self-inhibition response. When we expressed only ␣ and ␤ subunits in oocytes, amiloride-sensitive currents were generally too low to permit an examination of the Na ϩ self-inhibition response. Na ϩ currents in cells injected with ␣ and ␤ cRNAs were 0.01 Ϯ 0.01, relative to WT (i.e. ␣␤␥) channels (n ϭ 65, p Ͻ 0.001 versus WT, n ϭ 65-74). Nevertheless, a small number of ␣␤ cRNAinjected oocytes expressed currents at levels sufficient to examine the Na ϩ self-inhibition response. Not surprisingly, Na ϩ self-inhibition was absent (I ss /I peak , 0.98 Ϯ 0.01, p Ͻ 0.001 versus WT, n ϭ 7-18, Fig. 5). In contrast, cells injected with ␣␥ cRNAs showed a robust Na ϩ self-inhibition response that was similar to that in cells injected with cRNAs corresponding to ␣, ␤, and ␥ (I ss/ I peak , 0.74 Ϯ 0.06, p Ͼ 0.05 versus WT, n ϭ 18 -22, Fig. 5). These results suggest that loss of the Na ϩ self-inhibition response seen with ␣␤␥⌬T channels directly reflects the response of this channel, and not contributions from ␣␤ channels.
To examine whether the ␣ and ␤ thumb domains also play a role in the Na ϩ self-inhibition response, we examined the response in channels expressing either ␣⌬T␤␥ or ␣␤⌬T␥. As oocytes expressing these mutants exhibited very low Na ϩ currents, we injected 4 ng of cRNA/subunit for the mutant channels, rather than 1 ng/subunit for the WT channel. A Na ϩ selfinhibition response was absent in the recordings corresponding to the mutant channels lacking the ␣ thumb domain (I ss /I peak , 0.98 Ϯ 0.03, p Ͻ 0.001 versus WT, n ϭ 14 -17, Fig. 4, A and D). These data suggest that the ␣ thumb domain also plays a role in  , and in C for ␣␤␥⌬T (␥⌬T) ENaCs. Bath Na ϩ concentrations and amiloride (10 M) addition are shown above the traces. I peak and I ss in A show where peak current and steady state currents were measured. D-F, scatter plots of individual datum points of I ss /I peak (representing the magnitude of Na ϩ self-inhibition) for WT and mutant channels. The ratio was determined from amiloride-sensitive I ss and I peak . Number of oocytes in each group is shown in parentheses. Data were collected from 2 to 5 batches of oocytes. For WT and ␣␤␥⌬T, 1 ng/subunit of cRNAs was injected, whereas for ␣⌬T␤␥ and ␣␤⌬T␥, 4 ng/subunit of cRNAs were injected to increase expressed currents to levels that permit analyses of the Na ϩ self-inhibition response. Bar, mean for each group. Statistical comparisons were analyzed with two group Student's t test (p Ͻ 0.001, or n.s., no significant difference, as indicated).

ENaC thumb domain
the Na ϩ self-inhibition response. As ␤␥ channels lack function (27,28), the loss of Na ϩ self-inhibition likely reflects the assembled ␣⌬T␤␥ channels. Mutant channels lacking the ␤ thumb domain (␣␤⌬T␥) had a Na ϩ self-inhibition response that was similar to that of ␣␤␥ WT channels, suggesting that the ␤ thumb domain is dispensable for Na ϩ self-inhibition. However, as ␣␥ channels exhibit modest currents when expressed in oocytes (0.03 Ϯ 0.03, relative to WT, p Ͻ 0.001 versus WT, n ϭ 69) (27), it is possible that oocytes injected with ␣␤⌬T␥ cRNAs express both ␣␥ as well as ␣␤⌬T␥ channels. As noted above, ␣␥ channels exhibited a Na ϩ self-inhibition response similar to WT channels (Fig. 5). To confirm that ␤⌬T subunits reside in the functional channel complex, we introduced a S518C mutation into the ␤⌬T construct. The sulfhydryl reactive reagent MTSET specifically activates ␣␤S518C␥ channels, increasing channel open probability to near 1 (29,30). As shown in Fig. 6A, ␣␤S518C␥ channels responded to 1 mM MTSET with an increase in channel current and a loss of Na ϩ self-inhibition, consistent with a transition to a fully open state. In a majority of cells expressing ␣␤⌬T-S518C␥, MTSET increased the current and reduced Na ϩ selfinhibition, suggesting that currents are mediated predominately by the ␣␤⌬T-S518C␥ channels. The effect of MTSET on ␣␤⌬T-S518C␥ was similar to that on ␣␤S518C␥ (I MTSET /I, 1.73 Ϯ 0.38 ((␣␤⌬T-S518C␥) versus 1.85 Ϯ 0.32 (␣␤S518C␥), p Ͼ 0.05, n ϭ 9 -10). Therefore, most channels harbored a ␤ subunit. In contrast, MTSET had no effect on ␣␤␥ channels (31,32). There was also no statistically significant difference in the Na ϩ self-inhibition responses of ␣␤⌬T-S518C␥ and ␣␤S518C␥ prior to MTSET treatment, and following MTSET treatment (Fig. 6, A-C). Taken together, our observations suggest that a ␤ thumb domain deletion does not significantly affect Na ϩ self-inhibition.

Thumb domain deletions differentially affect channel protein maturation and degradation in a subunit-specific manner
ENaC matures inefficiently, which leads to robust targeting via the endoplasmic reticulum-associated degradation pathway (33, 34). Given the low level of surface expression of channels lacking the ␣ or ␤ subunit thumb domain, we examined whether these individual subunits underwent accelerated endoplasmic reticulum-associated degradation. To test this possibility, we performed cycloheximide chase experiments in yeast, because this model has served as a readout for inefficient channel folding and chaperone-mediated selection of orphaned ENaC subunits (34 -37). As shown in Fig. 7, there was no significant difference in rates of degradation between mutant and WT subunits when individual subunits were expressed in Saccharomyces cerevisiae.
To determine the stabilities of the subunits in a system in which the trimeric ENaC channel can assemble and transit beyond the ER, we performed cycloheximide chase experiments in Fisher rat thyroid (FRT) cells expressing either WT ␣␤␥ or mutant ENaCs lacking a thumb domain in a single subunit. Only the mutant subunit or the corresponding WT control contained an HA epitope tag at the N terminus and V5 tag at the C terminus (Fig. 8). Our previous studies showed that ENaC subunits processed in the biosynthetic pathway acquire mature N-linked glycans, and that the ␣ and ␥ subunits are cleaved at specific sites in their extracellular domains (38 -41). As a result of these maturation steps, the ␣, ␤, and ␥ subunits appear on immunoblots as bands at the following molecular weights using antibodies against the C-terminal V5 tag: immature ␣ (95 kDa) and mature ␣ (65 kDa); immature ␤ (96 kDa)

ENaC thumb domain
and mature ␤ (110 kDa); immature ␥ (93 kDa) and mature ␥ (75 kDa). We observed that deletion of the ␣ subunit thumb domain had no effect on the degradation rate when compared with that of the immature, noncleaved ␣ subunit. We also failed to detect a cleaved, mature ␣ subunit when the thumb domain was deleted (Fig. 8, A-C). Deletion of the thumb domain in the ␤ subunit also prevents its maturation, but in this case the ␤ subunit thumb domain deletion resulted in significantly faster degradation of the immature ␤ subunit (Fig. 8, D-F). In turn, a modest degree of proteolytic processing of the ␥ subunit was detectable in the ␥ subunit thumb deletion mutant. Although the degradation of the immature ␥ subunit did not differ between WT and mutant channels with a ␥⌬T subunit, degradation of the mature cleaved ␥⌬T subunit was significantly faster at the 2-h time point compared with the degradation of the cleaved WT ␥ subunit (75 kDa, Fig. 8, G-I). Whole cell protein expression levels of both ␤⌬T and ␥⌬T at the start of the chase (i.e. t ϭ 0) were also significantly lower than those of the respective WT subunits (p Ͻ 0.05, Fig. 8J, n ϭ 3), consistent with faster degradation rates of the ⌬T subunits. In contrast, whole cell expression of ␣⌬T was similar to the WT ␣ subunit, consistent with its WT-like degradation rate.
Finally, we examined whether the ␥ thumb domain deletion affects the protein levels of the ␣ or ␤ subunit. As shown in Fig.  9, ␥⌬T reduced whole cell ␤ subunit expression levels. No effect on whole cell ␣ subunit expression was evident. However, there was a loss of ␣ (65 kDa) and ␤ subunit (110 kDa) maturation in the presence of ␥⌬T. Overall, our results suggest that the presence of a thumb domain on each of the three subunits is required for channel maturation, and the absence of the ␤ or ␥ subunit thumb domain enhances ␤ (immature) or ␥ (mature) subunit degradation, respectively.

Discussion
In this study we sought to dissect the functional roles of the extracellular thumb domains within each of the ENaC subunits. Our results suggest that thumb domains in each subunit are required for optimal ENaC surface expression in Xenopus oocytes. Channels with an ␣ or ␤ subunit thumb domain deletion had a very low level of surface expression, whereas the absence of the ␥ subunit thumb domain resulted in only a moderate reduction in channel surface expression. Furthermore, deletions of the ␣ or ␥ subunit thumb domain were associated with a loss of the Na ϩ self-inhibition response, whereas channels with a ␤ subunit thumb domain deletion had a Na ϩ selfinhibition response that was similar to the WT channel. These results indicate that the thumb domains within the three ENaC subunits have distinct and nonequivalent roles for ENaC activity and surface expression. Although deletion of the ␥ subunit thumb domain did not significantly change macroscopic Na ϩ currents in oocytes, we found that ␥⌬T channels had reduced surface expression as well as a loss of the Na ϩ self-inhibition response that should increase channel open probability.
Previous studies analyzing channels with specific point mutations in the ␣ or ␥ subunit thumb domain suggested that these domains are involved in the Na ϩ self-inhibition response. A reduction in the Na ϩ self-inhibition response was seen with channels bearing individual mutations in six of the 10 Cys residues in the ␥ subunit thumb domain (42). Mutation of a Met residue (␥Met-438) within ␣5 helix was also associated with a loss of the Na ϩ self-inhibition response (24). Although oocytes expressing ␣⌬T␤␥ channels had very low currents in oocytes, reflecting markedly reduced ENaC surface expression, we observed that these channels lost Na ϩ self-inhibition. Previous findings have also suggested that the ␣ subunit thumb domain has a role in this response. Specifically, channels with an individual mutation in five of the 10 ␣ subunit thumb domain Cys residues exhibited altered responses to extracellular Na ϩ (42). Interestingly, mutations at four sites led to a reduction of the Na ϩ self-inhibition response, whereas one mutation enhanced the response (42). Other work indicated that mutation of a Cys residue (␣C479R) in the ␣ subunit thumb domain of human ENaC was associated with Liddle syndrome in a sibling pair ENaC thumb domain (43). As a mutation of the homologous residue in mouse ␣ENaC (␣C506A) eliminated Na ϩ self-inhibition (42), the phenotype seen with the human ␣C579R mutant activity likely reflects suppressed Na ϩ self-inhibition and an increase in channel open probability. Interestingly, most mutations in the region linking ␣ helices in the thumb domain of the ␣ subunit dramatically enhanced Na ϩ self-inhibition and reduced ENaC activity (42,44). In contract to channels with an ␣⌬T or ␥⌬T subunit, channels with a ␤⌬T subunit had a Na ϩ self-inhibition response that was similar to WT channels. This result is also in agreement with several previous studies where ␤ subunit mutations did not affect Na ϩ self-inhibition (8, 15, 16, 22, 24, 42). Epitope-tagged subunits were either wildtype (WT) or a thumb domain deletion mutant (⌬T) as indicated. Cycloheximide (100 g/ml) was added to cells growing on plastic (12-well size) at t ϭ 0. Cells lysates were prepared at varying times after cycloheximide addition (t ϭ 0, 1, 2, 4, or 6 h) in the presence of protease inhibitors. ENaC was immunoprecipitated with anti-V5 antibodies conjugated to beads and immunoblotted with anti-V5 antibodies. Aliquots of the initial cell extracts were subjected to immunoblotting with anti-␤-actin antibodies. The relative levels of proteins (compared with t ϭ 0) are presented as the mean Ϯ S.D. (A, B, D, E, G, and H) from three experiments, and a representative immunoblot is shown in each case (C, F, and I). Molecular weights of the immature ␣ (95,000), ␤ (96,000), and ␥ (93,000) subunits, and the mature ␣ (65,000), ␤ (110,000), and ␥ (75,000) subunits are indicated to the left of the immunoblots for the WT subunits. The mobility of the mutant subunits in each case were slightly faster than for the WT subunits as expected for the thumb deletions (ϳ10 kDa). The mature form of the ␥⌬T subunit was observed (H and I), but mature forms of the ␣⌬T and ␤⌬T subunits were absent. Degradation rates of the immature ␣ and ␥ subunits were similar between WT and ⌬T (A and G, p Ͼ 0.05, two-way ANOVA). The immature ␤⌬T subunit had significantly faster degradation than the WT ␤ subunit (D, *, p Ͻ 0.001, two-way ANOVA with Tukey post hoc test). The mature ␥⌬T level was significantly less than the mature WT ␥ subunit level at t ϭ 2 h (H, *, p Ͻ 0.01, two-way ANOVA with Tukey post hoc test). J, relative levels of mutant versus WT subunit proteins at t ϭ 0 (⌬T/WT, n ϭ 3). Protein levels were normalized to ␤-actin expression. Although there was no significant difference in ␣ subunit level in cells transfected with WT ␣␤␥ and ␣⌬T␤␥ (0.83 Ϯ 0.25, n.s., no significant difference, Student's t test) cDNAs, both ␣␤⌬T␥and ␣␤␥⌬Ttransfected cells had significantly less ␤ subunit (0.33 Ϯ 0.04, ***, p Ͻ 0.001, Student's t test) and ␥ subunit (0.76 Ϯ 0.14, *, p Ͻ 0.05, Student's t test) protein levels than WT-transfected cells, respectively. Both mature and immature forms were included in the quantification. K, representative examples (n ϭ 2) of immunoblots are shown for anti-V5 immunoprecipitates from 10 to 100% cell extract expressing WT ␣␤␥ (for comparison to 100% cell extract expressing ⌬T subunits), and for relevant amounts of extract for immunoblotting ␤-actin.

ENaC thumb domain
ENaCs are cleaved and activated by proteases within the biosynthetic pathway and at the cell surface (4,45). ENaC proteolysis appears to have a role in activating the channel under certain physiologic conditions, including volume depletion and hyperkalemia (46,47). Studies of ENaC activation by proteases have provided clues regarding a mechanism by which the ␣ and ␥ thumb domains influence the Na ϩ self-inhibition response. Channel activation by proteases involves cleavage at sites flanking imbedded inhibitory tracts in the ␣ and ␥ subunit finger domains (4, 45, 48 -50). Release of these inhibitory tracts transitions channels to a higher open probability state with an associated loss of Na ϩ self-inhibition (48 -50). Peptides corresponding to the released tracts are reversible channel inhibitors, and a putative inhibitory binding region for the ␣ subunitbased peptide mapped to sites in ␣ helices (␣1 and ␣2) in the finger domain, and the loop connecting the thumb domain ␣ helices in the ␣ subunit (5,51). It was proposed that the ␣ subunit thumb-finger domain interface is dynamic, and that the ␣ subunit-imbedded inhibitory tract stabilizes the thumb-finger domain interface, conferring a low activity channel state (5,44,51). In support of this model, cross-linking the finger domain ␣1 helix and the loop connecting the thumb domain in the ␣ subunit stabilized ENaC in a low activity state (51). As this model was based on functional studies, a resolved ENaC structure including the finger and thumb domains is needed to confirm key components of the model. We predict that the ␥ subunit imbedded inhibitory tract stabilizes the channel in a lower activity channel state by a similar mechanism.
Our model suggests that the ␣ and ␥ subunit thumb domains help restrain the activity of noncleaved channels. As ENaC activation by proteases occurs in association with reductions in (or a loss of) the Na ϩ self-inhibition response (7,48), the model also predicts that channels lacking an ␣ and ␥ subunit thumb domain will exhibit a reduced Na ϩ self-inhibition response. This is exactly what was observed (Fig. 4). We also found that the Na ϩ self-inhibition response of ␤⌬T channels was similar to WT channels. This is notable as ␤ subunits are not processed by proteases (38 -40), and there is no evidence that ␤ subunits have imbedded inhibitory tracts.
Thumb domain interactions with other extracellular components likely impact channel gating. For example, intersubunit cross-linking between thumb and palm domains alters ENaC activity in a length-dependent manner (52). Furthermore, there is evidence that ASIC1 thumb domain interactions with the ␤-ball and palm domains influence channel gating (53,54). The recently published closed state structure of ASIC1 highlights a prominent role of the thumb domains in channel gating (55). Although the reduced Na ϩ self-inhibition response in channels with an ␣⌬T or ␥⌬T subunit should result in an increase in channel activity, thumb domain deletions were also associated with significant reductions in ENaC surface expression. Expression of WT or mutant channels in FRT cells suggested that the ␣ and ␤ subunit thumb domains are required for subunit maturation. Cycloheximide-chase studies suggested that the ␤ subunit thumb domain has a role in stabilizing the immature ␤ subunit (Fig. 7). In addition, the ␥ thumb domain appears to have a role in processing leading to the mature forms of the ␣ and ␤ subunits, or stabilizing the mature form of the ␥ subunit ( Figs. 8 and 9). These findings are in agreement with previous studies of ENaCs with deletions of specific extracellular domains, where reductions in subunit maturation and channel surface expression were seen. For example, deletion of either the ␤ or ␥ subunit knuckle domain, but not the ␣ subunit knuckle domain, dramatically reduced ENaC surface expression and prevented subunit maturation (22).
In summary, we found that ENaC extracellular thumb domains are important regulators of ENaC gating by extracellular Na ϩ , steady state levels of channels at the cell surface, and channel subunit maturation. Our observations support the notion that specific extracellular domain structures serve as key functional modules (3,21,22).

Site-directed mutagenesis
Point mutations and deletions were introduced into mouse ␣, ␤, and ␥ ENaC cDNAs using the QuikChange II XL Sitedirected Mutagenesis Kit (Stratagene, La Jolla, CA). One set of Figure 9. Deletion of ␥ thumb domain reduces ␤ subunit expression. To examine the effect of the ␥ thumb domain deletion on stability of the ␣ subunit, FRT cells were transfected with WT ␣ or HA-␣-V5 and WT ␤ cDNAs with either WT ␥ or ␥⌬T cDNA (A). To examine the effect of ␥ thumb domain deletion on ␤ subunit stability, cells were transfected with WT ␤ or HA-␤-V5, and WT ␣ cDNAs with either WT ␥ or ␥⌬T cDNA (B). ENaC was immunoprecipitated with anti-V5 antibodies conjugated to beads and immunoblotted with anti-V5 antibodies. Representative blots are shown with mobility of immature (␣95K and ␤96K) and mature (␣65K and ␤110K) subunits shown to the left of the panels, and molecular mass markers shown to the right. Aliquots of the initial cell extracts were subjected to immunoblotting with anti-␤-actin antibodies. Cells transfected with nontagged ␣ and ␤ cDNAs were used as negative control in both experiments. C, relative levels represent the protein levels of either ␣ or ␤ subunit in cells expressing ␥⌬T, relative to their levels in the presence of the WT ␥ subunit (n ϭ 4). Protein levels were normalized to ␤-actin expression. There was no significant difference in relative ␣ subunit level when expressed with WT ␥ or ␥⌬T (n.s., no significant difference, Student's t test). However, ␤ subunit protein level was significantly lower when expressed with ␥⌬T than with the WT ␥ subunit (p Ͻ 0.001, Student's t test).

ENaC thumb domain
thumb domain deletions was generated where all residues between (and including) the first and last Cys residues within the thumb domains were deleted leaving an 8-residue linking tract to allow proper transition between the ␤9 and ␤10 strands of the palm domain (referred to as ⌬T). There 8 residue tracts were SLGGNYGD for ␣⌬T, RKGEPYSP for ␤⌬T, and KLSEPYSQ for ␥⌬T. Deleted residues are identified in Fig. 1C. A second set of thumb domain deletions was generated where all thumb domain residues, including the 8-residue tracts, were deleted. Target mutations were verified by DNA sequencing. WT and mutant ENaC cRNAs were synthesized by in vitro transcription using T3 RNA polymerase (Ambion, Inc.), purified by an RNA purification kit (Qiagen), and quantified by spectrophotometry.

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 (13). Stage V and VI oocytes with the follicle cell layer removed were injected with between 0.

Na ؉ self-inhibition
Na ϩ self-inhibition was examined as previously reported (8). Briefly, a low [Na ϩ ] bath solution (NaCl-1; containing 1 mM NaCl, 109 mM N-methyl-D-glucamine, 2 mM KCl, 2 mM CaCl 2 , 10 mM HEPES, pH 7.4) was rapidly replaced with a high [Na ϩ ] bath solution (NaCl-110; containing 110 mM NaCl, 2 mM KCl, 2 mM CaCl 2 , 10 mM HEPES, pH 7.4). Oocytes expressing ENaCs were continuously clamped to a membrane potential of Ϫ60 or Ϫ100 mV. Bath solution exchange was done with a Teflon valve computer-controlled perfusion system (AutoMate Scientific Inc., Berkeley, CA). At the end of each experiment, 10 M amiloride was added to the bath solution to determine the amiloride-insensitive portion of the whole-cell current. High perfusion rate (10 ml/min) and small bath volume (0.1 ml) in a linear flow chamber permitted rapid solution exchange (Ͻ1 s for a full exchange). The magnitude of Na ϩ self-inhibition was represented by the ratio of the steady state current (I ss ) to peak current (I peak ). I peak was obtained by visual inspection as the maximal inward current after bath [Na ϩ ] increase, typically within seconds. I ss was determined as the current at 40 s follow-ing the I peak . Both I ss and I peak were amiloride-sensitive currents, obtained by subtracting the current in the presence of 10 M amiloride. To minimize complications due to variability in the Na ϩ self-inhibition response between batches of oocytes, the responses of WT channels were always tested in an alternating manner with mutants in the same batch of oocytes.

Surface expression in oocytes
Surface expression of mouse ENaC in oocytes was determined using a ␤ or ␥ subunit with an extracellular FLAG epitope tag, as previously reported (23). Briefly, oocytes were injected with 4 ng/subunit of the three ENaC cRNAs with either the ␤ or ␥ subunit containing an extracellular FLAG epitope tag (DYKDDDDK). The FLAG tag was placed immediately following residue Thr 137 in the ␤ subunit, as previously described (56). The ␥ subunit tag was generated by replacing residues Val 131 -Thr 138 with a FLAG sequence. Robust amiloride-sensitive Na ϩ currents confirmed that ENaCs containing a FLAG-tagged ␤ or ␥ subunit were functional. Two days after cRNA injection, a surface expression assay was performed on ice, except for the last step that was at room temperature (20 -24°C). Following a 30-min incubation in antibiotic-free MBS supplemented with 1% BSA (MBS/BSA), oocytes were incubated for 1 h with MBS/ BSA supplemented with 1 g/ml of a mouse monoclonal anti-FLAG antibody (M2, Sigma). Oocytes were then washed in MBS/BSA and incubated with MBS/BSA supplemented with 1 g/ml of a horseradish peroxidase coupled to a secondary antibody (peroxidase-conjugated AffiniPure F(abЈ) 2 fragment goat anti-mouse IgG, Jackson ImmunoResearch, West Grove, PA) for 1 h. Cells were washed and transferred to MBS without BSA. Individual oocytes were placed in a 96-well plate, and 100 l of SuperSignal ELISA Femto Maximum Sensitivity Substrates (Thermo Scientific, Rockford, IL) was added to each well. Cells were then incubated at room temperature for 1 min, and chemiluminescence in relative light units was quantified in a GloMax-Multi ϩ Detection System (Promega, Madison, WI).
ENaC degradation in yeast. Yeast S. cerevisiae strains were grown at 26°C using standard methods, and media preparation and transformation were performed as described (57). The WT yeast strain used was BY4742 from Open Biosystems (Thermo Scientific). Construction of pRS426GPD-␣ENaC-HA, pRS426GPD-␤ENaC-HA, and pRS426GPD-␥ENaC-HA was previously described (34). The cloning strategy and restriction enzymes used to construct pRS426GPD-␣ENaC⌬T-HA, pRS426GPD-␤ENaC⌬T-HA, and pRS426GPD-␥ENaC⌬T-HA were similar to the strategy used to generate knuckle deletion mutants (22). Cycloheximide chase analyses of the ENaC subunits were performed as published (34), and cell lysates from chase samples were generated using alkaline lysis, followed by TCA precipitation (58). Proteins were resolved by SDS-PAGE before Western blot analysis. The ENaC subunits were detected using horseradish peroxidase-conjugated anti-HA antibodies (clone 3F10; Roche Applied Science) at a dilution of 1:5000. Western blots were also probed with anti-glucose-6-phosphate dehydrogenase (Glc-6-P; Sigma) rabbit antisera, which served as a loading control. The Glc-6-P primary antibody was detected with a donkey horseradish peroxidase-conjugated anti-rabbit IgG secondary antibody (GE Healthcare). Immuno-ENaC thumb domain blot signals were imaged using enhanced chemiluminescence (Pierce) and visualized on a Bio-Rad ChemiDoc XRSϩ system. Quantitation was carried out using ImageJ software (version 1.44, NIH).

ENaC degradation in FRT cells
FRT cells, provided by Dr. Michael Myerburg (University of Pittsburgh), were cultured in Dulbecco's modified Eagles's medium/F-12 medium supplemented with 7.5% decomplimented fetal bovine serum (FBS heated to 50°C for 50 min). Cells were plated in plastic wells (12-well size from Costar, Corning, NY) to form a uniform monolayer. The next day (18 -20 h), cells were transfected using Lipofectamine 2000 with 0.5 g of plasmids encoding ␣, ␤, and ␥ mouse ENaC subunits, with or without a thumb domain, as described in the figure legends. Thumb domains were deleted as indicated in subunits containing N-terminal HA and C-terminal V5 epitope tags, and co-expressed with the other two WT subunits with no epitope tag. 20 -24 h post-transfection cells were washed with Dulbecco's PBS with Mg 2ϩ and Ca 2ϩ (DPBS) (Corning Cellgro, Manassas, VA), and fresh media containing cycloheximide (diluted from ϫ1000 stock in DMSO) was added to cells at a final concentration of 100 g/ml (Calbiochem EMD Millipore, Billerica, MA). At 0, 1, 2, 4, or 6 h following the addition of cycloheximide, the cells were washed with DPBS, and then solubilized at 4°C in a detergent solution (0.4% sodium deoxycholate, 1% Nonidet P-40, 63 mM EDTA, and 50 mM Tris-HCl, pH 8) with 1% protease inhibitor mixture III (EMD Millipore, Billerica, MA; pH 7.4). ENaC was immunoprecipitated with anti-V5 antibodies conjugated to beads, as described previously (38). Supernatant (10%) from the immunoprecipitation was saved and used for immunoblotting for ␤-actin. Immunoprecipitated proteins were eluted from beads with Laemmli buffer followed by SDS-PAGE and immunoblotting with anti-V5 antibodies. To confirm that the bands quantified by scanning film were within a linear range with regard to band density, we examined band intensities of anti-V5 immunoprecipitates from a dilution series (between 5 and 100%) of extracts of cells expressing WT ␣␤␥ where only one subunit had N-terminal HA and C-terminal V5 epitope tags (Fig. 8K). The average band intensities of ␣⌬T, ␤⌬T, and ␥⌬T (Fig. 8K) were within the band intensities of the dilution series for the WT subunits.
To study the effect of the ␥ subunit thumb domain deletion on maturation and processing of the ␣ and ␤ subunits, FRT cells were plated in 12-well plastic plates (Costar, Corning, NY) to form a uniform monolayer. The next day (18 -20 h) cells were transfected using Lipofectamine 2000 with 0.5 g of plasmids encoding mouse ␣ and ␤ subunits, and a ␥ subunit with or without the thumb domain, as described in the figure legends. ␣ or ␤ subunit processing and maturation were examined with N-terminal HA and C-terminal V5-tagged ␣ or ␤ subunits, respectively, where the other two subunits had no epitope tag. 18 -20 h post-transfection, cells were solubilized at 4°C in a detergent solution (0.4% sodium deoxycholate, 1% Nonidet P-40, 63 mM EDTA, and 50 mM Tris-HCl, pH 8) with 1% protease inhibitor mixture III (EMD Millipore, Billerica, MA; pH 7.4). ENaC was immunoprecipitated with anti-V5 antibodies conjugated to beads, as described previously (38). Supernatant (10%) from the immunoprecipitation was saved and used for immunoblotting for ␤-actin. Immunoprecipitated proteins were eluted from beads with Laemmli buffer, followed by SDS-PAGE and immunoblotting with anti-V5 antibodies.

Statistical analyses
Data are presented as mean Ϯ S.D. Significance comparisons between WT and mutant groups were performed with a Student's t test for two group data, a one-way or two-way ANOVA with Tukey post hoc test for multiple group comparisons.