Activation of the Human Epithelial Sodium Channel (ENaC) by Bile Acids Involves the Degenerin Site*

The epithelial sodium channel (ENaC) is a member of the ENaC/degenerin ion channel family, which also includes the bile acid-sensitive ion channel (BASIC). So far little is known about the effects of bile acids on ENaC function. ENaC is probably a heterotrimer consisting of three well characterized subunits (αβγ). In humans, but not in mice and rats, an additional δ-subunit exists. The aim of this study was to investigate the effects of chenodeoxycholic, cholic, and deoxycholic acid in unconjugated (CDCA, CA, and DCA) and tauro-conjugated (t-CDCA, t-CA, t-DCA) form on human ENaC in its αβγ- and δβγ-configuration. We demonstrated that tauro-conjugated bile acids significantly stimulate ENaC in the αβγ- and in the δβγ-configuration. In contrast, non-conjugated bile acids have a robust stimulatory effect only on δβγENaC. Bile acids stimulate ENaC-mediated currents by increasing the open probability of active channels without recruiting additional near-silent channels known to be activated by proteases. Stimulation of ENaC activity by bile acids is accompanied by a significant reduction of the single-channel current amplitude, indicating an interaction of bile acids with a region close to the channel pore. Analysis of the known ASIC1 (acid-sensing ion channel) crystal structure suggested that bile acids may bind to the pore region at the degenerin site of ENaC. Substitution of a single amino acid residue within the degenerin region of βENaC (N521C or N521A) significantly reduced the stimulatory effect of bile acids on ENaC, suggesting that this site is critical for the functional interaction of bile acids with the channel.

The epithelial sodium channel (ENaC) is a member of the ENaC/degenerin ion channel family, which also includes the bile acid-sensitive ion channel (BASIC). So far little is known about the effects of bile acids on ENaC function. ENaC is probably a heterotrimer consisting of three well characterized subunits (␣␤␥). In humans, but not in mice and rats, an additional ␦-subunit exists. The aim of this study was to investigate the effects of chenodeoxycholic, cholic, and deoxycholic acid in unconjugated (CDCA, CA, and DCA) and tauro-conjugated (t-CDCA, t-CA, t-DCA) form on human ENaC in its ␣␤␥and ␦␤␥-configuration. We demonstrated that tauro-conjugated bile acids significantly stimulate ENaC in the ␣␤␥and in the ␦␤␥-configuration. In contrast, non-conjugated bile acids have a robust stimulatory effect only on ␦␤␥ENaC. Bile acids stimulate ENaC-mediated currents by increasing the open probability of active channels without recruiting additional near-silent channels known to be activated by proteases. Stimulation of ENaC activity by bile acids is accompanied by a significant reduction of the single-channel current amplitude, indicating an interaction of bile acids with a region close to the channel pore. Analysis of the known ASIC1 (acid-sensing ion channel) crystal structure suggested that bile acids may bind to the pore region at the degenerin site of ENaC. Substitution of a single amino acid residue within the degenerin region of ␤ENaC (N521C or N521A) significantly reduced the stimulatory effect of bile acids on ENaC, suggesting that this site is critical for the functional interaction of bile acids with the channel.
The epithelial sodium channel (ENaC) 3 belongs to the ENaC/degenerin superfamily of non-voltage-gated ion channels (1). The recently published crystal structure of chicken acid-sensing ion channel 1 (ASIC1) belonging to the same channel family suggests that ENaC is a heterotrimer composed of three homologous subunits: ␣, ␤, and ␥ (2-4). Atomic force microscopy data also support the assumption that ENaC is a heterotrimeric channel (5). Each subunit of ENaC consists of short intracellular N and C termini, a large extracellular domain, and two transmembrane domains (TM1 and TM2). All subunits are thought to contribute to the channel pore with their TM2 domains (1). The initial part of the TM2 domain is highly conserved among the ENaC/degenerin ion channel superfamily and contains the sites critically important for channel function including the selectivity filter (6 -10), the binding site of the channel blocker amiloride (11), and the degenerin site (12)(13)(14)(15). The degenerin site is a hallmark of channels of the ENaC/degenerin superfamily and is important for channel gating.
The well characterized ␣␤␥ENaC plays a pivotal role in sodium transport across the apical plasma membrane in the aldosterone-sensitive distal nephron, respiratory epithelia, distal colon, and sweat and salivary ducts (16 -19). In humans, an additional ␦-subunit exists that can functionally replace the ␣-subunit in heterologous expression systems (20 -24). ␦ENaC has been found in various epithelial and non-epithelial tissues (20,25). In particular, ␦ENaC mRNA is highly expressed in the brain (20,26,27). However, the functional role of the ␦-subunit remains unclear.
The regulation of ENaC is highly complex and involves several hormones, signaling pathways, and local mediators (28 -30). In particular, local factors affecting ENaC function are likely to be important for tissue-specific ENaC regulation. Moreover, important species differences appear to exist in ENaC regulation and need to be considered when studying the underlying regulatory mechanisms (31)(32)(33). Interestingly, rat ␣␤␥ENaC has recently been reported to be activated by bile acids (34). Bile acids have previously been shown to activate another member of the ENaC/degenerin ion channel family, the bile acid-sensitive ion channel (BASIC) (35,36), previously named intestine Na ϩ channel (INaC) in humans (37), and brain liver intestine Na ϩ channel (BLINaC) in mouse and rat (38).
The physiological role and mechanism of ENaC activation by bile acids remain to be elucidated. In particular, it has not yet been shown whether bile acids also affect the function of human ENaC, which can occur in at least two configurations (␣␤␥ or ␦␤␥) in different tissues. Therefore, the aim of this study was to investigate whether major human bile acids (chenodeoxycholic, cholic, and deoxycholic acid) modulate human ENaC in its ␣␤␥and ␦␤␥-configuration. We demonstrate that bile acids can activate human ENaC, probably by specifically interacting with the degenerin region.

Results
Bile Acids Are More Potent Activators of ENaC in the ␦␤␥than the ␣␤␥-Configuration-To test the effect of bile acids on ENaC function, human ␣␤␥or ␦␤␥ENaC was heterologously expressed in Xenopus laevis oocytes. Amiloride-sensitive whole-cell currents (I ami ) were measured using the two-electrode voltage clamp technique. Representative current traces for ␣␤␥or ␦␤␥ENaC-expressing oocytes are shown in Fig. 1, A and C and Fig. 1, B and D, respectively. Whole-cell current recordings were started in the presence of amiloride in a concentration of 2 or 100 M to inhibit ␣␤␥ENaC or ␦␤␥ENaC, respectively (22). Wash-out of amiloride revealed an ENaCmediated sodium inward current. Interestingly, the current response to superfusion with chenodeoxycholic acid (CDCA) was different in ␣␤␥ENaC-expressing oocytes when compared with that in ␦␤␥ENaC-expressing oocytes. CDCA reduced ␣␤␥ENaC currents by about 10% but resulted in 2.7-fold increase of ␦␤␥ENaC currents (Fig. 1, A and B). In contrast, tauro-deoxycholic acid (t-DCA) stimulated ENaC in both subunit configurations ( Fig. 1, C, D, and E). The effects of bile acids were reversible within about 2 min after wash-out. Re-addition of amiloride returned the whole-cell current toward the initial baseline level. The rapid onset and reversibility of the bile acid effect suggest that it is not caused by a permanent chemical modification of the channel or by a gradual accumulation of bile acids in the lipid bilayer of the plasma membrane. Control experiments demonstrated that bile acids had no effect on whole-cell currents in non-injected oocytes and that in ENaCexpressing oocytes the current stimulated by t-DCA was fully blocked by amiloride (data not shown). These control experiments confirm that the observed current responses are mediated by the effects of bile acids on ENaC activity. In addition to CDCA and t-DCA, we also tested the effect of cholic (CA), tauro-cholic (t-CA), deoxycholic (DCA), and tauro-chenodeoxycholic (t-CDCA) acid in similar experiments. Normalized data from these experiments are summarized in Fig. 1E. Interestingly, ␦␤␥ENaC currents were stimulated more than 2-fold by non-conjugated as well as tauro-conjugated bile acids. In contrast, only the tauro-conjugated forms of bile acids (t-CA, t-DCA, and t-CDCA) markedly stimulated ␣␤␥ENaC currents with t-DCA producing the largest effect (about 2-fold). The non-conjugated bile acids CA and DCA had a stimulatory effect of about 20% on ␣␤␥ENaC currents, whereas CDCA even inhibited ␣␤␥ENaC currents on average by more than 20%. Thus, a robust stimulatory effect on ␣␤␥ENaC was only observed with tauro-conjugated bile acids (Fig. 1E).
The marked difference between the effects of DCA and t-DCA on ␣␤␥ENaC may result from a lower binding affinity of DCA to the channel or from DCA binding without major effect on channel function. To address this question, we performed experiments in which we first applied DCA and subsequently applied t-DCA in the presence of DCA ( Fig. 2A) or first applied t-DCA and subsequently applied DCA in the presence of t-DCA (Fig. 2C). As summarized in Fig. 2B, DCA stimulated ENaC-mediated currents by about 30%, whereas subsequent application of t-DCA in the presence of DCA resulted in a stimulation of ENaC activity by more than 2-fold. In contrast, when ENaC was first activated by t-DCA, subsequent DCA application did not further activate the channel (Fig. 2D). Importantly, the combined stimulatory effects of DCA and t-DCA on ␣␤␥ENaC were similar in both cases (DCA followed by t-DCA: 2.18 Ϯ 0.08-fold, n ϭ 20, N ϭ 2; t-DCA followed by DCA: 2.15 Ϯ 0.09-fold, n ϭ 20, N ϭ 2) and were not significantly The groups compared with two-way ANOVA are indicated by horizontal brackets (***, p Ͻ 0.001, n.s., not significant). Values above or below the dashed line correspond to ENaC stimulation or inhibition, respectively. different from that of t-DCA applied alone (2.11 Ϯ 0.08-fold, n ϭ 20, N ϭ 2; Fig. 2, B and D). Thus, the effects of DCA and t-DCA were not additive. These data suggest that DCA and t-DCA modify channel activity by the same mechanism. The finding that an initial application of DCA does not reduce the stimulatory effect of t-DCA applied in the presence of DCA suggests that DCA does not prevent t-DCA binding to the channel. Thus, t-DCA is likely to have a higher binding affinity to the channel than DCA and binding of DCA may have a smaller stimulatory effect on ␣␤␥ENaC than binding of t-DCA.
Taken together, these findings demonstrate that tauro-conjugated bile acids stimulate ENaC in the ␦␤␥and in the ␣␤␥configuration, whereas non-conjugated bile acids have a robust stimulatory effect only on ␦␤␥ENaC.
Tauro-deoxycholic Acid Activates Human ␣␤␥ENaC by Increasing Open Probability (P o ) of Active Channels but Not by Recruiting Additional Near-silent Channels-Our whole-cell current measurements demonstrated that the effects of bile acids on human ENaC in both ␣␤␥-subunit and ␦␤␥-subunit configurations were rapid and reversible. This makes it unlikely that insertion of new channels in the plasma membrane contributes to the stimulatory effect of bile acids on ENaC. Instead, bile acids probably stimulate ENaC by increasing the P o of channels that are already present in the plasma membrane. Two functionally distinct ENaC populations are thought to be present in the plasma membrane: active channels with an average P o of about 0.5 and so-called near-silent channels with an extremely low P o (31, 39 -41). Thus, bile acids may further increase the P o of the active channel population or may recruit additional near-silent channels by converting them into active channels. To investigate the biophysical mechanisms of ENaC activation by bile acids, we performed single-channel record-ings in outside-out patches of ENaC-expressing oocytes and tested the effect of bile acids on ENaC. Fig. 3A shows a representative recording from an outside-out patch excised from an oocyte expressing ␣␤␥ENaC. The initial wash-out of amiloride revealed single-channel activity with up to three apparent channel levels (Fig. 3A). After t-DCA application, the same number of active channels in the patch was observed. However, NP o was moderately increased from 1.04 to 1.64 (where N is the number of channels and P o is single-channel open probability), and the single-channel current amplitude (i) was reduced from 0.39 to 0.37 pA (Fig. 3A, insets 1 and 2). The observed increase in NP o is consistent with the stimulatory effect of t-DCA on ENaC whole-cell currents in oocytes expressing ␣␤␥ENaC (Fig.  1E). Upon wash-out of t-DCA, NP o and i returned approximately to their initial values (0.88 and 0.39 pA, respectively; Fig.  3A, inset 3). It is well established that proteolytic activation of ENaC is associated with the recruitment of near-silent channels (31,39,41,42). Indeed, subsequent application of the serine protease chymotrypsin led to the recruitment of additional channel levels and a strong increase of NP o from 0.88 to 5.28 In some experiments with very low initial ␣␤␥ENaC activity, the difference in the effects of t-DCA and chymotrypsin was particularly impressive (Fig. 3C). In this experiment, only a few, extremely brief single-channel openings were observed prior to the application of t-DCA (NP o Ͻ 0.01), which indicates that in this outside-out patch, only near-silent channels with a very low P o are present. Interestingly, application of t-DCA had no significant effect on channel activity in this patch, which suggests that near-silent channels are not markedly affected by t-DCA. In contrast, application of chymotrypsin resulted in the expected conversion of near-silent channels into active channels as evidenced by the appearance of up to four channel levels with long channel openings (NP o ϭ 2.52). In summary, our single-channel data indicate that t-DCA stimulates ␣␤␥ENaC activity by increasing the P o of active channels without recruiting additional near-silent channels, which distinguishes the effect of t-DCA from that of channel-activating proteases. Moreover, the small but significant reduction of single-channel current amplitude in the presence of t-DCA indicates a possible interaction of bile acids with a region close to the channel pore.
Chenodeoxycholic Acid Reduces the Single-channel Current Amplitude of ␦␤␥ENaC-In outside-out patches from oocytes expressing ␦␤␥ENaC, we confirmed our previously reported finding that for unknown reasons ␦␤␥ENaC shows a high basal P o close to 1 (Fig. 4, A and B) (22). Indeed, in outside-out patches, the channel resides almost all the time in its open state with only brief closing events (Fig. 4, inset 1 demonstrating part of the trace on an expanded time scale). Under these conditions, it is not surprising that application of CDCA had no additional stimulatory effect on P o despite the large stimulatory effect of CDCA observed on ENaC whole-cell currents (Fig. 1B). However, the single-channel current amplitude of ␦␤␥ENaC was slightly reduced from 0.95 to 0.84 pA (Fig. 4A, inset 2). The single-channel current amplitude of ␦␤␥ENaC returned to its initial value upon wash-out of CDCA (Fig. 4A, inset 3). In outside-out patches, chymotrypsin also had no detectable effect on P o and single-channel current amplitude of ␦␤␥ENaC (Fig. 4A, inset 4), which is consistent with our previously reported data (22). Importantly, the small but significant and reversible reduction of the single-channel current amplitude of ␦␤␥ENaC by CDCA (Fig. 4C) suggests that CDCA interacts with ␦␤␥ENaC in a similar way as t-DCA interacts with ␣␤␥ENaC ( Fig. 3B). This further supports the hypothesis that bile acids interact with a region in ENaC close to the channel pore.
Analysis of the Crystal Structure of the Transmembrane Domains of ASIC1 Suggests a Putative Binding Site for Bile Acids in the Degenerin Region of ENaC-To identify a putative bile acid binding site, we analyzed the crystal structure of ASIC1 transmembrane domains. Analysis revealed the presence of hydrophobic crevices in the membrane-spanning region of the channel occupied by three maltose groups of the detergent n-dodecyl-␤-D-maltoside used for channel isolation from the plasma membrane before crystallization (Protein Data Bank (PDB) ID 2QTS (3)) ( Fig. 5A). Two of the putative detergent binding sites are located on the outer side of the transmembrane domains and are probably occupied by membrane lipids. Interestingly, the third site is located within the pore region formed by the second transmembrane domains (TM2) of all three subunits and is accessible from the extracellular solution. Bile acids are amphiphilic substances and can behave as detergents (43). Therefore, we hypothesized that maltoside detergent may mimic the effect of bile acids on ENaC function. Indeed, similar to CDCA, application of maltoside in a nonsolubilizing concentration (10 M) significantly stimulated ␦␤␥ENaC by about 1.5-fold and inhibited ␣␤␥ENaC by about 10% (Fig. 5, C-E). Thus, it is tempting to speculate that bile acids may bind to ENaC at sites corresponding to those identified for maltoside detergent co-crystallized with ASIC1. Among the three putative binding sites, the site located within the pore region appeared to be the most likely candidate for bile acid binding, because of the rapid onset and reversibility of the bile acid effect on P o and single-channel current amplitude. Estimating the energy of interaction (MolDock score (44) using Molegro Molecular Viewer 2.5.0 from CLC bio) between the maltose group and the channel reveals that the aspartate resi-due in position 433 (Asp 433 ) makes the highest contribution to binding at this site. Using protein sequence alignment, we identified residues homologous to ASIC1 Asp 433 in all four human ENaC subunits (Fig. 5B). The homologous residues are Asn 550 , Ala 527 , Asn 521 , and Asn 530 in ␣-, ␦-, ␤-, and ␥-ENaC, respectively (Fig. 5B). Interestingly, these residues belong to the degenerin region known to play an important role in ENaC gating (1,8,10,13). Thus, we hypothesized that the amino acid residues in ENaC subunits homologous to ASIC1 Asp 433 may be functionally important sites for the interaction between bile acids and ENaC.
Mutating the Asparagine Residue ␤Asn 521 to Cysteine or Alanine Reduces the Stimulatory Effect of Bile Acids on ␦␤␥ENaC-As shown above, bile acids activate ␦␤␥ENaC more potently than ␣␤␥ENaC. Therefore, we initially focused on ENaC in its ␦␤␥-configuration to investigate whether the amino acid residues homologous to Asp 433 in ASIC1 are involved in bile acidmediated activation of ENaC. The homologous amino acid residues ␦Ala 527 , ␤Asn 521 , and ␥Asn 530 were mutated to cysteines. The substitution of these residues by cysteine was chosen, because cysteine does not change substantially the size and charge of the amino acid side chain. Moreover, the cysteine residue can be covalently modified by the sulfhydryl reagent (2-(trimethylammonium)ethyl) methanethiosulfonate bromide (MTSET). Thus, treatment with MTSET can be used as a tool to alter the function of mutant channels by chemical modification of the introduced cysteine residue (1,8,10,13,22,41,45). Each mutant subunit was co-expressed in oocytes with two complementary wild-type ENaC subunits (␦ A527C ␤␥-, ␦␤ N521C ␥-, and ␦␤␥ N530C ENaC). For unknown reasons, no channel function was observed in experiments with mutant subunits ␦A527C or ␥N530C (data not shown). In contrast, co-expression of ␤N521C with wild-type ␦and ␥-subunits (␦␤ N521C ␥ENaC) resulted in measurable ENaC currents. Moreover, in oocytes expressing ␦␤ N521C ␥ENaC, application of MTSET acutely and irreversibly increased ENaC currents (Fig.  6, B and D). This stimulatory effect of MTSET was similar to that observed with the ␤S520C mutation (Fig. 6, C and D), which is consistent with previous studies (22). This indicates that both ␤Asn 521 and ␤Ser 520 are accessible to MTSET when mutated to cysteine and form part of the degenerin region in the ␦␤␥-configuration of ENaC.
Next, we tested whether the substitution of ␤Asn 521 by cysteine modulates the stimulatory effect of bile acids on ␦␤␥ENaC. As shown in Fig. 7A, application of CDCA to control oocytes expressing ␦␤␥ENaC without a mutation resulted in a more than 2-fold increase of ENaC-mediated currents consistent with the findings shown in Fig. 1. Subsequent application of chymotrypsin (2 g/ml) in the presence of CDCA did not  cause a further current increase. We have previously shown that chymotrypsin in this concentration maximally activates ␦␤␥ENaC by proteolytic cleavage, increasing channel P o from about 0.5 to almost 1 consistent with a 2-fold increase of ENaCmediated whole-cell currents (22). We confirmed in the present study that chymotrypsin alone caused an ϳ2-fold increase of ENaC-mediated whole-cell currents (data not shown) similar to the effect observed with CDCA alone or with CDCA in combination with chymotrypsin. Thus, our findings suggest that both chymotrypsin and CDCA can increase the P o of wild-type ␦␤␥ENaC from ϳ0.5 to almost 1. Importantly, substitution of ␤Asn 521 by cysteine (␦␤ N521C ␥ENaC) reduced the stimulatory effect of CDCA by about 50%. Subsequent application of chymotrypsin to ␦␤ N521C ␥ENaC resulted in an additional stimulatory effect (Fig. 7, B and I). These findings indicate that the ␤N521C mutation partially prevents the stimulatory interaction of CDCA with the channel, whereas proteolytic channel activation is preserved.
The significant inhibitory effect of the ␤N521C mutation on ENaC stimulation by CDCA was confirmed by regression anal-ysis of normalized current data (Fig. 8A) taken from similar experiments as shown in Fig. 7, A and B. This analysis demonstrates that ENaC currents measured in the presence of CDCA are proportional to those measured in the presence of both CDCA and chymotrypsin. Importantly, the proportionality coefficient was close to 1 for the wild-type channel (0.94 Ϯ 0.01; N ϭ 3; n ϭ 38), confirming that chymotrypsin has no additional stimulatory effect on ␦␤␥ENaC after channel activation by CDCA. In contrast, the proportionality coefficient for ␦␤ N521C ␥ENaC was significantly smaller (0.48 Ϯ 0.01; p Ͻ 0.001; N ϭ 3; n ϭ 29). This supports the conclusion that the ␤N521C mutation reduces the stimulatory effect of CDCA by more than 50%. In additional experiments, we tested the effect of five other bile acids on ␦␤ N521C ␥ENaC. Using similar regression analysis as illustrated in Fig. 8A (data not shown), we demonstrated that the ␤N521C mutation significantly reduced the stimulatory effects of all bile acids tested (Fig. 8B).
Next, we tested whether substitution of ␤Asn 521 by other amino acids also affected the stimulatory effects of bile acids. Substituting ␤Asn 521 by alanine appeared to decrease basal P o , because the maximal current reached in the presence of CDCA and chymotrypsin was almost 6-fold larger than the baseline current (Fig. 7C). Under the assumption that the mutant channel reaches a P o of almost 1 in the presence of CDCA and chymotrypsin, it can be concluded that the baseline P o of the channel containing the ␤N521A mutation is considerably lower than that of wild-type ␦␤␥ENaC. This conclusion is consistent with the finding that basal I ami in oocytes expressing ␦␤ N521A ␥ENaC was significantly lower than in matched oocytes expressing wild-type ␦␤␥ENaC (0.28 Ϯ 0.03 A, n ϭ 33 versus 0.84 Ϯ 0.08 A, n ϭ 20, N ϭ 2; p Ͻ 0.01). The lower initial P o of ␦␤ N521A ␥ENaC could explain why the relative stimulatory effect of CDCA on the mutant channel was larger than that on the wild-type channel (Fig. 7I). Importantly, the stimulation achieved by CDCA did not reach the level obtained by subsequent exposure to chymotrypsin. Thus, similar to the ␤N521C mutation, the ␤N521A mutation also resulted in a reduced stimulatory effect of CDCA on ENaC in the ␦␤␥configuration (Fig. 8B). In contrast, the mutant channel ␦␤ N521S ␥ENaC behaved like the wild-type channel (Figs. 7, D and I, and 8B), and basal I ami of ␦␤ N521S ␥ENaC was similar to that of ␦␤␥ENaC in matched oocytes (0.81 Ϯ 0.10 A, n ϭ 24 versus 0.84 Ϯ 0.08 A, n ϭ 20, N ϭ 2; n.s.), arguing against a nonspecific effect of mutating ␤Asn 521 . Furthermore, substitution of ␤Asn 521 by aspartate (␤N521D) resulted in a large reduction of the stimulatory effect of both CDCA and chymotrypsin (Fig. 7, E and I). The decreased responsiveness of this mutant channel to two different stimuli suggests that introducing a negative charge at this site increased baseline P o close to 1. This conclusion is consistent with the finding that basal I ami in oocytes expressing ␦␤ N521D ␥ENaC was significantly larger than in matched oocytes expressing wild-type ␦␤␥ENaC (2.04 Ϯ 0.20 A, n ϭ 23 versus 0.84 Ϯ 0.08 A, n ϭ 20, N ϭ 2; p Ͻ 0.001). In addition, we introduced a cysteine residue in the neighboring position (␦␤ S520C ␥ENaC). The mutant ␦␤ S520C ␥ENaC behaved like the wild-type channel in response to CDCA (Figs. 7, F and I, and 8B), supporting the specific functional importance of the position ␤Asn 521 for the interaction of bile acids with the channel in the ␦␤␥-configuration. This interaction may be only partially prevented by mutating ␤Asn 521 to cysteine or alanine, which may explain the incomplete inhibition of the CDCA effect by these mutations. Alternatively, the incomplete inhibition may be due to an interaction of CDCA with additional sites. Candidate sites in the ␦and ␥-subunits are ␦Ala 527 and ␥Asn 530 , respectively, which correspond to ␤Asn 521 in the ␤-subunit. As stated above, no channel function was observed when ␦Ala 527 or ␥Asn 530 were mutated to cysteine. In contrast, functional channel expression was preserved when mutating ␦Ala 527 and ␥Asn 530 to asparagine and alanine, respectively. However, in oocytes expressing ␦ A527N ␤␥or ␦␤␥ N530A ENaC, the stimulatory effect of CDCA was similar to that observed in oocytes expressing wild-type ␦␤␥ENaC, and an additional stimulation by chymotrypsin was also absent (Figs. 7, G-I, and 8B). These findings do not support the hypothesis that ␦Ala 527 or ␥Asn 530 contribute to the functional interaction of bile acids with the channel. However, this possibility and the existence of additional interaction sites cannot be ruled out.
Substitution of Asparagine Residue ␤Asn 521 by Alanine Decreases the Relative Stimulatory Effect of t-DCA on ␣␤␥ENaC-We wondered whether ␤Asn 521 also plays a role in bile acid-mediated activation of ␣␤␥ENaC. Co-expression of the ␤N521A and ␤N521S with wild-type ␣and ␥-subunits resulted in measurable ENaC currents, whereas in this configuration, no functional channels were detected with the ␤N521C subunit. At first inspection, the stimulatory effect of t-DCA on ␣␤ N521A ␥ENaC appeared to be similar to that on wild-type ␣␤␥ENaC (Fig. 9). However, subsequent application of chymotrypsin in the presence of t-DCA revealed a larger additional increase of ␣␤ N521A ␥ENaC currents when compared with that of wild-type ␣␤␥ENaC currents (Fig. 9, B and  C). Thus, the increased stimulatory effect of chymotrypsin on ␣␤ N521A ␥ENaC after pre-stimulation with t-DCA may be interpreted as a reduced relative stimulatory effect of t-DCA on the mutant channel when compared with its effect on the wild-type channel (Fig. 9D). This interpretation is consistent with a functional role of the ␤Asn 521 residue in mediating the stimulatory effect of bile acids on ENaC in the ␣␤␥-configuration of the channel. In contrast to the ␤N521A mutation, the ␤N521S mutation did not reduce the relative stimulatory effect of t-DCA on ENaC in the ␣␤␥-configuration (Fig. 9D), which is consistent with the finding that this mutation also failed to reduce the stimulatory effect of CDCA on ␦␤␥ENaC currents (Figs. 7, D and I, and 8B). To investigate whether ␥Asn 530 , which corresponds to ␤Asn 521 , may be involved in mediating the stimulatory effect of t-DCA on ENaC in the ␣␤␥-configuration, we mutated this residue to an alanine (␥N530A). In oocytes expressing ␣␤␥ N530A ENaC, the relative stimulatory effect of t-DCA when compared with that of chymotrypsin was not reduced but was minimally increased as indicated by a slightly larger proportionality coefficient (Fig. 9D). However, this subtle effect cannot be taken as evidence for a role of ␥Asn 530 in the interaction of bile acids with ␣␤␥ENaC. Finally, we also had no indication that the ␣N550A mutation altered the effect of t-DCA on ENaC in the ␣␤␥-configuration (Fig.  9D).
In summary, these findings support the concept that the ␤Asn 521 residue in the degenerin region is involved in mediating the stimulatory effect of bile acids on ␣␤␥ENaC as well as on ␦␤␥ENaC.

Discussion
In this study, we demonstrated that physiologically relevant bile acids activate human ENaC in the ␣␤␥and ␦␤␥-configuration. Moreover, we found that substitution of a single amino acid residue within the degenerin region of ␤ENaC (Asn 521 ) significantly reduced this stimulatory effect. This indicates that the degenerin region is critical for the functional interaction of bile acids with ENaC.
To study the mechanism by which bile acids stimulate human ENaC activity, we used outside-out patch clamp recordings. We could demonstrate that t-DCA stimulated human ␣␤␥ENaC by increasing single-channel P o without recruiting additional near-silent channels in the patch. Our findings confirm recent results reported for rat ␣␤␥ENaC (34). In outsideout patches, baseline P o of ␦␤␥ENaC was close to 1 and therefore could not be further stimulated by CDCA. The reason for this high baseline P o of ␦␤␥ENaC in our patch clamp recordings is presently unknown but is in good agreement with previous findings (22). Importantly, our whole-cell recordings demonstrated that ␦␤␥ENaC was activated by bile acids to a similar extent as by chymotrypsin, which is known to increase P o of ␦␤␥ENaC to about 1 (22). Thus, our findings suggest that the stimulatory effect of bile acids on ENaC in the ␦␤␥-configuration is also mediated by an increase in single-channel P o .
The increase of ENaC P o may be caused by direct interaction of bile acids with a specific binding site of the channel or by changing the plasma membrane properties, thereby modifying the interaction of the channel protein with membrane lipids. Changes in physical properties of the plasma membrane, such as plasma membrane thickness, intrinsic monolayer curvature, or elastic properties of the lipid bilayer, are known to regulate various membrane proteins including ion channels (46). Indeed, it has been proposed that activation of rat bile acidsensitive ion channel by bile acids is likely to be caused by an alteration of the membrane environment (47). Interestingly, a significant portion of ENaC has been reported to be associated with lipid rafts, which are thought to be important for channel function and regulation (48 -51). In model lipid systems such as liposomes (68) or giant plasma membrane vesicles (52), bile acids bind more efficiently to non-raft (lipid-disordered) than to raft (ordered) membrane fractions. Thus, the ability of bile acids to modify the lipid environment of ENaC localized in lipid rafts may be limited. However, at present, the possibility that bile acids modulate ENaC activity indirectly by modifying the properties of the plasma membrane in the vicinity of the channel cannot be ruled out.
Alternatively, bile acids may interact directly with the channel protein through specific binding sites. Modification of protein function by bile acid binding to specific sites is a known phenomenon. The list of examples includes the specific nuclear receptor farnesoid X receptor (FXR), G protein-coupled receptor TGR5, bile salt export pump (BSEP), and Na ϩ /taurocholate co-transporting polypeptide (NTCP) (53). We demonstrated that ENaC in the ␣␤␥-configuration is activated by tauro-conjugated bile acids, whereas application of non-conjugated CA and DCA only moderately stimulated ␣␤␥ENaC and CDCA even had an inhibitory effect. Thus, slight differences in the chemical structure of bile acids are sufficient to modulate the effect of bile acids on ␣␤␥ENaC. This favors the interpretation that bile acids interact with specific binding sites of the channel protein, which may have a preference for certain types of bile acids. Based on the reported co-crystallization of the detergent maltoside within the degenerin region of ASIC1 (3), we hypothesized that bile acids may interact with the degenerin region of ENaC. Our findings, that maltoside mimicked the effect of bile acids on ENaC and that mutations in the degenerin region (␤N521C and ␤N521A) significantly reduced the stimulatory effect of bile acids on ENaC, strongly support this hypothesis.
The discovery of the functional importance of the degenerin site for channel gating was initially made in the context of identifying MEC-4 and DEG-1 mutations, which cause neurodegeneration in Caenorhabditis elegans (12,14,15). The introduction of bulky amino acid residues to homologous sites in other ENaC/degenerin channels mimics the gain-of-function effect of naturally occurring mutations and results in hyperactivity of the mutant channels (13,37,38,54,55). The gain-offunction effect of degenerin mutations on ENaC can be reproduced by introducing a cysteine residue at the degenerin site of the ␤-subunit. Subsequent covalent modification of this cysteine residue by sulfhydryl reagents, e.g. MTSET, activates the mutant channel and has been used as a tool to experimentally increase ENaC P o close to 1 (1,10,13,41). Under these conditions, MTSET increases P o of active ␣␤␥ENaC without recruiting additional near-silent channels (41). Moreover, the stimulatory effect of MTSET on P o is accompanied by a small reduction in the single-channel current amplitude (1,10,13). Interestingly, these effects of MTSET are similar to the bile acid effects on ENaC observed in the present study. In the light of these findings, it is tempting to speculate that MTSET and bile acids share a similar mechanism of action. Thus, binding of bile acids to the degenerin site may stabilize the open state of the channel by inducing a conformational change in the outer vestibule of ENaC, mimicking the effect of covalent modification of the degenerin site by MTSET. The concomitant small reduction of the single-channel current amplitude is consistent with the interpretation that bile acids bind to a region in close proximity to the channel pore. As discussed above, we observed some differential effects of bile acids on ENaC function in the ␣␤␥versus the ␦␤␥-configuration of the channel. These findings may be explained by subtle structural differences in the degenerin regions of ␣␤␥ versus ␦␤␥ENaC. However, at present this remains a speculation, because a crystal structure of ENaC would be needed to analyze this further.
It has been proposed that bile acids may play a role in ENaC regulation under physiological and pathophysiological conditions. For example, ␣␤␥ENaC is expressed in human and mouse cholangiocytes (56), indicating that bile acids may be physiologically relevant modulators of ENaC function in bile ducts. Moreover, ␦ENaC is expressed abundantly in different brain regions (26,27,57), and CDCA may also be accumulated or locally produced in the brain (58). Thus, it is conceivable that CDCA can affect ␦␤␥ENaC in the brain. Our finding that bile acids and the detergent maltoside have similar effects on ENaC raises the possibility that other endogenous amphiphilic substances may modulate ENaC activity in a similar way. Such endogenous substances, capable of binding to the degenerin region of the channel, may act as local modulators of ENaC function in a tissue-specific manner, but they remain to be identified. In conclusion, our results highlight the potential role of the degenerin region as a regulatory site involved in the functional interaction of bile acids and possibly other naturally occurring amphiphilic substances with ENaC.
Isolation of Oocytes and Two-electrode Voltage Clamp Experiments-Defolliculated stage V-VI oocytes were obtained from ovarian lobes of adult female X. laevis in accordance with the principles of German legislation, with approval by the animal welfare officer for the University of Erlangen-Nürnberg, and under the governance of the state veterinary health inspectorate. Animals were anesthetized in 0.2% ethyl 3-aminobenzoate methanesulfonate (MS-222) (Sigma), and ovarian lobes were obtained by a small abdominal incision. Isolation of oocytes and two-electrode voltage clamp experiments were performed essentially as described previously (22, 45, 60 -63). Oocytes were injected with cRNA using the same amount of cRNA per ENaC subunit per oocyte in the range of 0.1 to 5 ng. Injected oocytes were incubated in ND96 solution (in mM: 96 NaCl, 2 KCl, 1.8 CaCl 2 , 1 MgCl 2 , 5 HEPES, pH 7.4, with Tris) supplemented with 100 units/ml sodium penicillin and 100 g/ml streptomycin sulfate. Unless stated otherwise, oocytes were studied 48 h after cRNA injection. I ami were determined by subtracting the current values recorded in the presence of amiloride (2 and 100 M for ␣␤␥ENaCand ␦␤␥ENaC-expressing oocytes, respectively) from those recorded in the absence of amiloride.
Single-channel Recordings in Outside-out Patches-Oocytes injected with ␣␤␥ENaC or ␦␤␥ENaC cRNA were stored in ND96 solution. Single-channel recordings in outside-out membrane patches of ENaC-expressing oocytes were performed 48 h after cRNA injection essentially as described previously (35,41,45) using conventional patch clamp technique. Patch pipettes were pulled from borosilicate glass capillaries and had a tip diameter of about 1-1.5 m after fire polishing. Pipettes were filled with potassium gluconate pipette solution (in mM: 90 potassium gluconate, 5 NaCl, 2 Mg-ATP, 2 EGTA, and 10 HEPES, pH 7.2, with Tris). Seals were routinely formed in a low sodium NMDG-Cl bath solution (in mM: 95 NMDG-Cl, 1 NaCl, 4 KCl, 1 MgCl 2 , 1 CaCl 2 , 10 HEPES, 7.4 pH, with Tris). In this bath solution, the pipette resistance averaged about 7 megaohms. After seal formation, the NMDG-Cl solution was switched to a NaCl bath solution in which NMDG-Cl (95 mM) was replaced by NaCl (95 mM). For continuous current recordings, the holding potential was set to Ϫ70 mV. Using a 3 M KCl flowing boundary electrode, the liquid junction potential occurring at the pipette/NaCl bath junction was measured to be 12 mV (bath-positive) (35). Thus, at a holding potential of Ϫ70 mV, the effective trans-patch potential was Ϫ82 mV. This value is close to the calculated equilibrium potential of Cl Ϫ (E Cl ϭ Ϫ77.4 mV) and K ϩ (E K ϭ Ϫ79.4 mV) under our experimental conditions. Experiments were performed at room temperature (ϳ23°C). Single-channel current data were initially filtered at 1.25 kHz and sampled at 5 kHz. The current traces were refiltered at 250 Hz to resolve the single-channel current amplitude (i) and channel activity. The latter was derived from binned amplitude histograms as the product NP o (35,41,45,64). The current level at which all channels are closed was determined in the presence of 2 M amiloride for ␣␤␥ENaC and of 10 M amiloride for ␦␤␥ENaC. The apparent number of active channels (apparent N) in a patch was determined by visual inspection of the current traces. Singlechannel data were analyzed using the program Nest-o-Patch written by Dr. V. Nesterov (Institut für Zelluläre und Molekulare Physiologie, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany).
Statistical Methods-Data are presented as mean Ϯ S.E., individual data points, and bar diagrams as appropriate. Statistical significance was assessed by the appropriate version of ANOVA (with Bonferroni post hoc test) or Student's t test. N indicates the number of different batches of oocytes, and n indicates the number of individual oocytes studied. Statistical and regression analysis was performed using GraphPad Prism 5.04.
Author Contributions-A. I. and A. D. performed the experiments, analyzed the data, and prepared the figures. A. I., A. D., C. K., and S. H. designed the study, interpreted the data and wrote the paper. All authors approved the final version of the manuscript.