Incorporation of the δ-subunit into the epithelial sodium channel (ENaC) generates protease-resistant ENaCs in Xenopus laevis

The epithelial sodium channel (ENaC) is a critical regulator of vertebrate electrolyte homeostasis. ENaC is the only constitutively open ion channel in the degenerin/ENaC protein family, and its expression, membrane abundance, and open probability therefore are tightly controlled. The canonical ENaC is composed of three subunits (α, β, and γ), but a fourth δ-subunit may replace α and form atypical δβγ-ENaCs. Using Xenopus laevis as a model, here we found that mRNAs of the α- and δ-subunits are differentially expressed in different tissues and that δ-ENaC predominantly is present in the urogenital tract. Using whole-cell and single-channel electrophysiology of oocytes expressing Xenopus αβγ- or δβγ-ENaC, we demonstrate that the presence of the δ-subunit enhances the amount of current generated by ENaC due to an increased open probability, but also changes current into a transient form. Activity of canonical ENaCs is critically dependent on proteolytic processing of the α- and γ-subunits, and immunoblotting with epitope-tagged ENaC subunits indicated that, unlike α-ENaC, the δ-subunit does not undergo proteolytic maturation by the endogenous protease furin. Furthermore, currents generated by δβγ-ENaC were insensitive to activation by extracellular chymotrypsin, and presence of the δ-subunit prevented cleavage of γ-ENaC at the cell surface. Our findings suggest that subunit composition constitutes an additional level of ENaC regulation, and we propose that the Xenopus δ-ENaC subunit represents a functional example that demonstrates the importance of proteolytic maturation during ENaC evolution.

Mechanisms allowing salt and water conservation are some of the most important adaptations to a terrestrial environment and key in tetrapod evolution (1). The renin-angiotensin-aldosterone system evolved during water-land transition and is the major regulatory mechanism controlling the ionic composition of the extracellular milieu in vertebrates (1). A key target for the renin-angiotensin-aldosterone system is the epithelial sodium channel (ENaC) 4 in the apical membrane of principal epithelial cells in the aldosterone-sensitive distal nephron (2). ENaC represents the rate-limiting step for apical uptake of sodium ions and, in concert with the basolateral Na ϩ /K ϩ -ATPase, transepithelial sodium and, consequently, water absorption.
ENaC is the only member of the degenerin (DEG)/ENaC superfamily of proteins, which is constitutively open (2) and is therefore one of the most stringently regulated ion channels in vertebrates. Hormones, such as aldosterone and glucocorticoids, control ENaC expression (3), whereas a complex network of intracellular kinases regulates its membrane abundance (4). ENaC open probability (P o ) is modulated by intramembrane lipids (2) and is sensitive to changes in the extracellular environment (2), including the concentration of sodium ions (5), pH (6), or mechanical stimuli (7). In addition to these regulatory mechanisms, ENaC activity is linked to its maturation state through proteolytic processing of its subunits (2).
The canonical ENaC is a heterotrimeric channel composed of three homologous subunits (␣, ␤, and ␥) (8). Each subunit has short intracellular N and C termini connected by a large extracellular loop. The extracellular loop probably represents a "handlike" structure, consisting of a "palm," "ball," "finger," "thumb," and "knuckle" domain (9). The extracellular loops of the ␣and ␥-subunits contain short inhibitory peptides in the finger domains, which are flanked by protease cleavage sites. The inhibitory peptide within the ␣-subunit is flanked by two cleavage sites that are recognized by the protease furin, whereas the inhibitory peptide within the ␥-subunit contains only one furin cleavage site and additional cleavage sites for extracellular proteases (10). Furin, which is primarily present in the trans-Golgi network, cleaves the ␣-subunit twice and thereby releases its inhibitory peptide, whereas the ␥-subunit is only cut once (2). Channels lacking this endogenous proteolytic maturation display a very low P o , whereas furin-cleaved ENaCs that are translocated to the plasma membrane have moderate activity. Extracellular proteases cut the ␥-subunit and thereby release its inhibitory peptide, resulting in ENaCs with a high P o (2, 10). Proteolytic processing not only represents an additional level of ENaC regulation but was also suggested to be crucial for the transition of a stimulus-activated DEG/ENaC ancestor to a constitutively open channel during ENaC evolution (10).
In addition to the three subunits composing canonical ENaCs, a ␦-subunit was identified, which can assemble with the ␤and ␥-subunits to form atypical ENaCs (11). The physiological importance of this subunit is not understood, but in vitro studies using human orthologues demonstrated that ␦␤␥-ENaC has distinct properties from ␣␤␥-ENaC (12). Complicating its physiological analysis, mice and rats lack a functional gene for ␦-ENaC (13). However, ␦-ENaC is present in amphibians, including the South African clawed frog Xenopus laevis, where it had originally been named ⑀-ENaC (14). Given the lack of an appropriate animal model to study this ion channel and the fact that X. laevis is an established model organism in molecular cell biology as well as physiology, this study investigated the molecular physiology of ENaC containing the ␦-subunit in this species.
We demonstrate that ␦-ENaC has a different expression pattern compared with ␣-ENaC and that ␦-containing channels display an intrinsic dynamic open probability. We show that incorporation of the ␦-subunit renders this amphibian ENaC completely insensitive to proteolytic activation by extracellular proteases due to intersubunit interactions involving ␥-ENaC. We suggest that subunit composition provides an additional level of ENaC regulation but also propose that Xenopus ␦-ENaC represents a functional example for ENaC at the interface of stimulus-activated and constitutively open DEG/ENaC ion channels, demonstrating the importance of proteolytic maturation in ENaC evolution.

The ␦-ENaC subunit is predominantly expressed in urogenital tissues in X. laevis
Due to the lack of specific antibodies that allow discrimination of ␣and ␦-ENaC, expression of these subunits was investigated in adult X. laevis via RT-PCR (Fig. 1). Control reactions were performed with cDNA samples that were generated without the addition of reverse transcriptase. These reactions were always negative and did not show any DNA signals in the MidoriGreen fluorescence of the agarose gels after the PCRs (data not shown). Transcripts for the ␦-ENaC subunit were almost exclusively found in urogenital tissues (kidney, urinary bladder, testes, ovary, oocytes, and cloaca), whereas expression of the ␣-subunit was more diversely distributed (skin, lung, liver, parts of the gastrointestinal and urogenital tract, brain, and spinal cord). All ␦-ENaCexpressing tissues, except the ovary, also contained transcripts for the ␣-subunit.

The presence of the ␦-subunit alters ENaC characteristics
Functional characterization of ␣␤␥and ␦␤␥-ENaC was achieved by expression of either channel isoform in X. laevis oocytes. Although transcripts for ␦-ENaC were detected in oocytes ( Fig. 1) injection of ␤and ␥-ENaC RNA does not produce amiloride-sensitive transmembrane current (I M ) (14), excluding a potential contamination by endogenous ␦-ENaC.
Electrophysiological assessment of whole-cell I M at membrane potentials (V M ) from Ϫ100 to ϩ40 mV revealed an inward current at negative V M with a reversal potential of 9.8 mV (␣␤␥-ENaC) and 14.2 mV (␦␤␥-ENaC) and was inhibited by the ENaC-blocker amiloride (Fig. 2a). Consistent with previous reports (14), channel isoforms exhibited different kinetics of I M after washout of amiloride. Cells expressing ␣␤␥-ENaC display a slow and steady current increase that eventually reaches a plateau, whereas channels containing the ␦-subunit show a rapid, peaked increase in I M followed by a continuous current decline (Fig. 2b). Despite this current rundown, amiloridesensitive current fractions (⌬I ami ) of ␦␤␥-ENaCexpressing oocytes were 2.4-fold larger than in ␣␤␥-ENaCexpressing cells (Fig. 2c). The increased ⌬I ami mediated by ␦␤␥-ENaC was also sustained after 14 min when the current eventually approached a plateau (Fig. 2, d and e).
The increased ⌬I ami of ␦␤␥-ENaCexpressing oocytes may result from changes in the single-channel conductance, P o , or Figure 1. The ␦-ENaC subunit is predominantly expressed in urogenital tissues in X. laevis. Tissues from adult X. laevis were screened for expression of ␦and ␣-ENaC subunit mRNA via RT-PCR. Amplicons for ␦-ENaC (320 bp) were found in tissue samples from kidney, urinary bladder, testes, ovary, oocytes, and the cloaca. With the exception of the ovary, these tissues also expressed ␣-ENaC (291 bp). Amplification of ␤-actin mRNA (610 bp) served as a control. Control reactions have been performed without reverse transcriptase but are omitted from the figure for clarity. These reactions did not show any DNA signals in the MidoriGreen fluorescence of agarose gels after PCR. Numbers in parentheses indicate the quantity of positively tested samples with respect to the total amount of samples from different animals.

Subunit composition determines ENaC regulation
membrane abundance of these channels. Cell-attached patchclamp recordings of oocytes expressing ␦␤␥-ENaC revealed unitary channel characteristics similar to ␣-ENaCcontaining channels (Fig. 3a). Analysis of individual recordings lasting for 120 -180 s revealed no significant difference between ENaC isoforms in the number of visible channels or the slope conductance (G slope ), derived from V M of Ϫ100 to Ϫ40 mV (Fig. 3,  b and c). Although there was no difference in P o compared with ␣␤␥-ENaC, channels containing the ␦-subunit displayed significantly prolonged open times and a tendency for extended mean closed times that did not reach statistical significance (Fig. 3, d and e). Consistent with whole-cell currents (Fig. 2d), patch-clamp recordings of ␦␤␥-ENaC regularly showed a decline of channel activity (Fig. 3f), which impaired proper estimation of P o . Therefore, P o of ␣␤␥and ␦␤␥-ENaC was additionally determined using 2-(trimethylammonium)ethylmeth-anethiosulfonate (MTSET)-sensitive ␤ S540C -ENaC mutant channels. Covalent modification of an introduced cysteine within the conserved degenerin site of ␤-ENaC by MTSET elevates P o of ENaC containing the ␤ S540C mutation to 1.0 (16). Comparison of ⌬I ami before and after MTSET application allows estimation of channel P o . The ⌬I ami of oocytes expressing ␣␤ S540C ␥or ␦␤ S540C ␥-ENaC was not significantly different from that for cells expressing respective WT channels (Fig. 4a). P o of ␣␤ S540C ␥or ␦␤ S540C ␥-ENaC was estimated by measuring ⌬I ami before and after administration of MTSET (1 mM), which increased I M mediated by both channels (Fig. 4, b and c), assuming that MTSET shifts the P o of both channel isoforms to the same degree (close to 1.0). Although initial ⌬I ami of ␦␤ S540C ␥-ENaC was increased compared with ␣-ENaCcontaining channels, this discrepancy was lost after application of MTSET (Fig. 4d). Comparison of the initial ⌬I ami with ⌬I ami after exposure to MTSET (P o ϭ 1.0) indicated a significantly larger initial P o for ENaC containing the ␦-subunit compared with ␣␤ S540C ␥-ENaC (Fig. 4e). Overall, these data suggest that the presence of the ␦-subunit enhances the amount of current generated by ENaC due to an increased P o rather than single-channel conductance, but also changes current into a transient form.

The ␦-subunit is not proteolytically processed and does not change membrane abundance compared with ␣-containing channels
Proteolytic processing of ENaC subunits by intra-and extracellular proteases impacts channel P o (10). To monitor proteolytic cleavage of both ␣␤␥and ␦␤␥-ENaC, epitope-tagged ␣ HA/V5 -and ␦ HA/V5 -ENaC subunits containing an N-terminal HA tag and a C-terminal V5 tag were generated (Fig. 5a). Immunoblots from whole-cell lysates of oocytes expressing either ␣ HA/V5 ␤␥or ␦ HA/V5 ␤␥-ENaC targeted against the HA or V5 tag revealed a discrepancy in the endogenous cleavage pattern of both ENaC subunits (Fig. 5b). Blots targeting the N-terminal HA tag of the ␣ HA/V5 -subunit displayed a fulllength peptide at ϳ75 kDa and two short peptides at ϳ18 and ϳ13 kDa, probably corresponding to cleavage products resulting from proteolysis by intracellular furin, which cuts the ␣-subunit twice (Xenopus ␣-ENaC: RVKR 145 and RVSR 168 ) (17). In contrast, only the full-length peptide of the ␦ HA/V5subunit at ϳ75 kDa was detected, consistent with the lack of a minimal consensus sequence for furin-mediated cleavage (RXXR2 (18)) in this ENaC subunit. These findings were confirmed by blots targeting the C-terminal V5 tags, which yielded two signals for ␣ HA/V5 -ENaC, representing full-length (ϳ75 kDa) and cleaved (ϳ60 kDa) peptides but only one signal (ϳ75 kDa) representing the full-length ␦ HA/V5 -ENaC. Proteolysis by furin would occur in the trans-Golgi network. Therefore, glycosylation of the ␦ HA/V5 -subunit, which is also associated with this route for post-translational modification, was examined. Treatment of whole-cell lysates from oocytes expressing ␦ HA/V5 ␤␥-ENaC with PNGase F induced a downward migration shift of the signal in blots targeting the V5 tag (Fig. 5c). This indicates glycosylation of the ␦-ENaC subunit and suggests passage through the trans-Golgi network. Epitope-tagged ␣ HA/V5and ␦ HA/V5 -ENaC subunits were subsequently employed to compare membrane abundance of both channel isoforms by . c, amiloride-sensitive current fractions (⌬I ami ) in both ENaC isoforms derived from the second application of amiloride (Student's unpaired t test with Welch's correction). d and e, representative current traces (d) and corresponding mean values of I M (e) of ␣␤␥or ␦␤␥-ENaCexpressing oocytes, depicting the enhanced current rundown of ␦-ENaCcontaining channels over an extended time period. At any given time point, I M mediated by ␦-containing channels was significantly increased compared with ␣␤␥-ENaCmediated currents (*, p Ͻ 0.0003, Mann-Whitney U test). Lines and error bars, mean and S.E.

Subunit composition determines ENaC regulation
surface biotinylation. Similar to WT channels, the ⌬I ami mediated by ␦ HA/V5 ␤␥-ENaC was increased compared with ␣ HA/V5 ␤␥-ENaC (Fig. 5, d and e). Only a single signal (ϳ60 kDa) was detected in membrane fractions from ␣ HA/V5 ␤␥-ENaCexpressing cells, indicating the absence of unprocessed ␣ HA/V5 -ENaC peptides at the cell surface, whereas only unprocessed ␦ HA/V5 -ENaC peptides (ϳ75 kDa) could be detected in membrane and cytosolic fractions (Fig. 5f). Membrane abundance was determined by densitometric analysis of signal intensities from both membrane and cytosolic protein fractions in V5-targeted immunoblots. There was no difference in surface expression between ␣ HA/V5 ␤␥and ␦ HA/V5 ␤␥-ENaC (Fig. 5g), supporting the notion that an enhanced P o rather than membrane abundance contributes to the increased I M mediated by ENaC containing the ␦-subunit. In sum, these data suggest that unlike ␣-ENaC, the ␦-subunit does not undergo proteolytic maturation by the endogenous protease furin, and incorporation of this subunit does not affect channel membrane abundance.

ENaCs containing the ␦-subunit are insensitive to activation by extracellular proteases
Because the ␦-ENaC subunit is not processed by furin, the activation of ␣␤␥and ␦␤␥-ENaC by extracellular proteases was compared. Oocytes expressing either ENaC isoform were exposed to chymotrypsin, which activates ENaC at the cell surface (19). The ⌬I ami of both channel isoforms was determined before and after perfusion of the cells with 2 g/ml ␣-chymotrypsin (from bovine pancreas, TLCK-treated; Sigma) in ORS (90 mM NaCl, 1 mM KCl, 2 mM CaCl 2 , 5 mM HEPES, pH 7.4) containing amiloride for 5 min (Fig. 6, a and d). Administration of chymotrypsin significantly increased ⌬I ami of oocytes expressing ␣␤␥-ENaC (Fig. 6b). This activation was also present in cells incubated with furin inhibitor I (40 M) before recordings. Inhibition of furin potentiated the proteolytic activation of ␣␤␥-ENaC from a 4-fold to a 9-fold increase of ⌬I ami , whereas no change of ⌬I ami (-fold activation ϭ 1) was observed in control groups that were perfused with ORS containing amiloride without chymotrypsin (Fig. 6c). Application of chy-

Subunit composition determines ENaC regulation
motrypsin did not increase currents mediated by ␦␤␥-ENaC, regardless of preceding incubation with the furin inhibitor (Fig.  6, d-f). The same results were observed using trypsin (data not shown). These data indicate that currents generated by ␦␤␥-ENaC are insensitive to activation by extracellular serine proteases, such as chymotrypsin and trypsin.

The presence of the ␦-subunit prevents cleavage of the ␥-subunit by extracellular proteases
Intracellular cleavage of the ␥-ENaC subunit renders the channel susceptible to activation by extracellular proteases (20). Cleavage of ␥-ENaC at the cell surface by extracellular proteases induces the release of its inhibitory peptide and acti-vates the channel (21). Because the ␦-subunit prevented ENaC activation by extracellular chymotrypsin, the cleavage states of the ␣-, ␦-, and ␥-subunits at successive stages of proteolytic ENaC processing were examined. Epitope-tagged subunits were expressed in oocytes together with WT ENaC subunits (␣ HA/V5 ␤␥, ␦ HA/V5 ␤␥, ␣␤␥ HA/V5 , and ␦␤␥ HA/V5 ), and ⌬I ami of individual oocytes from each group with or without prior incubation in chymotrypsin (2 g/ml; 5 min) was measured. After surface biotinylation, membrane fractions were isolated, and cleavage states of individual ENaC subunits were analyzed in immunoblots targeted at the C-terminal V5 epitope. Incubation with chymotrypsin enhanced ⌬I ami of ␣ HA/V5 ␤␥-ENaC-  . The ␦-subunit is not subject to proteolytic maturation, and ␦␤␥-ENaC membrane abundance is not changed compared with ␣-containing channels. a, schematic depiction of epitope-tagged ENaC subunits.
The numbers indicate approximate molecular mass of peptides with/without furin cleavage, detected by immunoblots, as presented in b. b, immunoblots of whole-cell lysates from oocytes expressing ␣ HA/V5 ␤␥or ␦ HA/V5 ␤␥-ENaC using anti-HA and anti-V5 antibodies. c, a migration shift of the ␦ HA/V5 -subunit due to treatment with PNGase F indicates glycosylation of the protein. d, representative I M recordings of oocytes expressing ␣ HA/V5 ␤␥or ␦ HA/V5 ␤␥-ENaC. e, amiloride-sensitive current fractions (⌬I ami ) of ␣ HA/V5 ␤␥and ␦ HA/V5 ␤␥-ENaC (Student's unpaired t test). f, immunoblot using an anti-V5antibody of membrane (mem.) and cytosolic fractions from oocytes expressing ␣ HA/V5 ␤␥or ␦ HA/V5 ␤␥-ENaC. The bottom blot shows the results from reprobing of the same membrane with an anti-pan-actin antibody to verify proper fractionation. g, membrane abundance of ␣ HA/V5 ␤␥and ␦ HA/V5 ␤␥-ENaC derived from densitometric analysis of immunoblots as shown in f (Student's unpaired t test). Lines and error bars, mean and S.E.

Subunit composition determines ENaC regulation
expressing oocytes, whereas this activation was absent in ␦ HA/V5 ␤␥-ENaCexpressing cells (Fig. 7a). In corresponding immunoblots, two signals at ϳ75 and ϳ60 kDa were visible in cytosolic fractions for the ␣ HA/V5 -ENaC subunit that represent uncleaved and furin-cleaved peptides, whereas only processed subunits were present in membrane fractions (Fig. 7b). Analogous stages of subunit maturation were not observed for the ␦ HA/V5 -subunit , for which only one signal at ϳ75 kDa was visible in cytosolic and membrane fractions. No additional cleavage products of ␣ HA/V5 -and ␦ HA/V5 -ENaC were found in membrane fractions of chymotrypsin-treated oocytes, indicating no further subunit cleavage at the cell surface. However, examination of ␥ HA/V5 -ENaC cleavage stages revealed further processing of this subunit at the cell surface, which depended on its co-expression with ␣or ␦-ENaC. Whereas ⌬I ami mediated by ␣␤␥ HA/V5 -ENaC was increased in oocytes treated with chymotrypsin, activity of ␦␤␥ HA/V5 -ENaC was unaffected by the protease (Fig. 7c). Immunoblots of either ␣or ␦-ENaCcontaining channels revealed two signals in cytosolic fractions that represent uncleaved (ϳ100-kDa) and cleaved (ϳ75-kDa) ␥ HA/V5 -ENaC peptides (Fig. 7d). Only one signal was present in the membrane fraction of these cells (ϳ75 kDa), indicating the absence of uncleaved ␥ HA/V5 -ENaC subunits at the cell surface. An additional cleavage product (ϳ70 kDa) appeared in the membrane fraction of ␣␤␥ HA/V5 -ENaC-expressing oocytes after incubation with chymotrypsin. However, chymotrypsin did not induce an additional cleavage product in membrane fractions of cells expressing ␦␤␥ HA/V5 -ENaC. The secondary cut of ␥ HA/V5 -ENaC by chymotrypsin is prevented in the presence of the ␦-subunit, which probably contributes to the insensitivity of ␦␤␥-ENaC to extracellular proteases. These data indi-cate that the presence of the ␦-subunit prevents cleavage of the ␥-subunit at the cell surface and activation by extracellular proteases.

Discussion
This study revealed that the presence of the ␦-subunit changes properties of transmembrane currents generated by X. laevis ENaC. In contrast to the rather stable currents mediated by ␣␤␥-ENaC, channels containing the ␦-subunit mediated larger currents that had a more transient nature. These observations are consistent with findings from a previous study and can be attributed to differences in the speed of ENaC sodium self-inhibition (SSI) (14), which reflects the allosteric reduction of ENaC P o in the presence of extracellular sodium ions (5). SSI kinetics are reduced in Xenopus ENaCs containing the ␦-subunit (14). Human ␦␤␥-ENaC also displays a slow progression of SSI (22) and reduced gating kinetics (11), suggesting a generally decreased gating velocity of ENaC containing the ␦-subunit. This is supported by single-channel recordings of both ENaC isoforms presented in this study, which revealed prolonged mean open times and a tendency for longer closed times of ␦-containing ENaCs. However, whereas the magnitude of SSI is decreased in human ␦␤␥-ENaC (22), it is markedly pronounced in the Xenopus orthologue (14) and probably accounts for the rather transient current form observed for ␦␤␥-ENaC in this study.
Despite this rundown, whole-cell currents of oocytes expressing ␦␤␥-ENaC were consistently larger compared with ␣␤␥-ENaCexpressing cells. Similarly, substitution of ␣for the ␦-subunit in human ENaC results in elevated whole-cell currents, which have been suggested to be caused by a larger

Subunit composition determines ENaC regulation
single-channel conductance and an enhanced P o (19). Channels resulting from co-expression of Xenopus ␦-ENaC with the human ␤and ␥-ENaC subunits also exhibit a high spontaneous P o , whereas their unitary conductance equaled those of ␣-ENaCcontaining channels (23). Cell-attached patch-clamp recordings of Xenopus ␣and ␦␤␥-ENaC in the present study did not reveal differences in the conductance or P o of these ENaC isoforms, supporting the concept that incorporation of the ␦-ENaC subunit from X. laevis does not affect ENaC singlechannel conductance. Without apparent differences in the number of visible channels and P o of ␣␤␥and ␦␤␥-ENaC obtained from patch-clamp recordings, the observed differences in amiloride-sensitive whole-cell currents could not be explained. A continuous decrease of channel activity was observed in patches containing ␦␤␥-ENaC, which corresponds to the macroscopic decline of whole-cell transmembrane currents due to SSI and suggests an SSI-dependent temporal decrease in channel activity.
Alternative determination of P o at the whole-cell level using mutant ␤ S540C -ENaC in combination with MTSET displayed a markedly increased P o of ␦-ENaC compared with ␣-ENaCcontaining channels. The generated Xenopus ␤ S540C -ENaC mutation is analogous to the previously utilized rat ␤ S518C (24) and human ␤ S520C (19) constructs that target the conserved "degenerin" site of ENaC subunits, which corresponds to a domain essential for the gating of related DEG channels in Caenorhabditis elegans (25). Administration of MTSET to ENaC containing this mutation sets channel P o to nearly 1.0 through the covalent addition of a bulky residue to the intro-duced cysteine near the channel gate (16), and we assumed that MTSET shifts the P o of ␣and ␦-containing ENaCs to the same degree. Introduction of this mutation did not change baseline ENaC current, as initial ⌬I ami of channels containing the mutant ␤ S540C -subunit were not significantly different from respective WT channels. MTSET increased ⌬I ami of ␣␤ S540C ␥-ENaC and, to a lower extent, ⌬I ami of ␦␤ S540C ␥-ENaC. Assuming a similarly effective activation of both ENaC isoforms by MTSET and a negligible proportion of closed channels that are insensitive to chemical modification (16), the estimated P o of ␣␤ S540C ␥-ENaC (0.2) correlates well to patch-clamp results in this (0.24) and a previous study (7). ENaC containing the ␦-subunit displayed a P o that was 3 times larger (0.6), which explains the enhanced whole-cell currents mediated by Xenopus ␦␤␥-ENaC. Furthermore, equal levels of ⌬I ami in ␣and ␦␤ S540C ␥-ENaC after application of MTSET indicate a similar membrane expression of both channel isoforms. This is further supported by the equal number of visible channels observed in patchclamp recordings of ␣and ␦␤␥-ENaC as well as the similar membrane abundance of epitope-tagged ␣ HA/V5 -or ␦ HA/V5 -ENaC subunits. Hence, an elevated P o is the major determinant for increased whole-cell currents mediated by ␦␤␥-ENaC, but P o is not stable and decreases over time.
Interestingly, introduction of the ␤ S540C mutation abolished the transient nature of whole-cell currents mediated by ␦-containing ENaC. This supports the conclusion that the transient activity of Xenopus ␦␤␥-ENaC reflects its P o , which is inhibited in the presence of extracellular sodium due to SSI. As a release from SSI and a resulting increase of ENaC P o are essential for Figure 7. The presence of the ␦-ENaC subunit prevents cleavage of the ␥-subunit by extracellular proteases. a, comparison of ⌬I ami from oocytes expressing epitope-tagged ␣ HA/V5 ␤␥or ␦ HA/V5 ␤␥-ENaC with (ϩ) or without (Ϫ) prior incubation with chymotrypsin (2 g/ml, 5 min). b, V5-targeted immunoblots of cytosolic and membrane fractions from oocytes expressing ␣ HA/V5 ␤␥or ␦ HA/V5 ␤␥-ENaC that were treated with (ϩ) or without (Ϫ) chymotrypsin. Blots were reprobed for pan-actin to verify proper isolation of membrane fractions. c, ⌬I ami of oocytes expressing epitope-tagged ␣␤␥ HA/V5 -or ␦␤␥ HA/V5 -ENaC with (ϩ) or without (Ϫ) prior incubation with chymotrypsin. d, V5-targeted immunoblots of cytosolic and membrane fractions from oocytes expressing ␣␤␥ HA/V5 -or ␦␤␥ HA/V5 -ENaC, which were treated with (ϩ) or without (Ϫ) chymotrypsin. Lines and error bars, mean and S.E.

Subunit composition determines ENaC regulation
channel activation by proteolytic processing (26), the regulation of Xenopus ␦␤␥-ENaC by proteases was investigated.
Initial proteolytic processing of ENaC occurs by selective cleavage of the ␣and ␥-ENaC subunits by furin (27). During this proteolytic maturation in the trans-Golgi network, furin cleaves the channel at three distinct sites. Two cleavage sites located in the finger domain of ␣-ENaC enclose a small inhibitory peptide that is released upon furin-mediated cleavage (28), whereas proteolysis at a single site on the ␥-ENaC subunit primes the channel for further activation at the cell surface (20). Secondary proteolysis of the ␥-ENaC subunit distal to the furin cleavage site entails the dissociation of another inhibitory peptide and thus facilitates activation of ENaC at the cell surface by further increasing P o . Additional cleavage of ␥-ENaC may be facilitated by various serine proteases like trypsin, channel-activating protease 1, or neutrophil elastase (2).
The present study demonstrates that the ␦-subunit of Xenopus ENaC is not subject to proteolytic maturation by furin, and channels containing this subunit are insensitive to activation by extracellular proteases. Immunoblots targeting epitope-tagged ␣ HA/V5 -, ␦ HA/V5 -, or ␥ HA/V5 -ENaC subunits revealed ␣and ␥-ENaC cleavage fragments. These fragments probably represent furin cleavage products, because both ENaC subunits contain conserved furin consensus sites (RXXR2) (2), and channel activation by extracellular chymotrypsin is potentiated after inhibition of furin. By contrast, the absence of ␦-ENaC cleavage fragments in oocyte whole-cell lysates and membrane fractions implies a lack of intracellular proteolytic processing and is consistent with the absence of a furin consensus site in the peptide sequence of this subunit. Furthermore, the stimulatory effect of extracellular proteases as observed for ␣␤␥-ENaC was completely lost in ␦␤␥-ENaC. Immunoblots targeting the V5 epitope of ␥ HA/V5 -ENaC revealed two signals, corresponding to uncleaved and furin-cleaved peptides, in cytosolic fractions, whereas the membrane fraction only contained one signal corresponding to the furin-cleaved peptide. This was observed for ␥ HA/V5 -ENaC after co-expression with the ␣or ␦-subunit, suggesting that the presence of the ␦-subunit does not prevent intracellular maturation of ␥-ENaC. After exposure to chymotrypsin, an additional immunoblot signal occurred in membrane fractions of ␣␤␥ HA/V5 -ENaC, representing a second cleavage event in the ␥-subunit by the extracellular protease. In contrast, this was not observed in ␦␤␥ HA/V5 -ENaC, suggesting that the presence of the ␦-subunit prevents cleavage of ␥-ENaC by extracellular proteases. Extracellular cleavage of the ␥-subunit is considered rate-limiting for maximal ENaC activation at the cell surface, thus explaining the lack of functional current stimulation of ␦␤␥-ENaC by chymotrypsin.
The lack of an ENaC crystal structure complicates the mechanistic interpretation of this observation. The crystal structure of the homologous acid-sensing ion channel suggests that the finger domains play a prominent role in intersubunit interactions (9). Based on the location of the inhibitory peptides within the finger domains of the ␣and ␥-subunit (2), it could be hypothesized that cleavage of the ␣-subunit triggers structural changes that affect the neighboring ␥-subunit. The lack of cleavage in the ␦-subunit might prevent such a structural rearrangement and prevent extracellular proteases from accessing corresponding cleavage sites in the ␥-subunit. This hypothesis is challenged by the fact that the presence of the inhibitory sequence rather than cleavage per se is responsible for low ENaC activity (2). However, the peptide sequence of the ␦-subunit does not contain homologies to an inhibitory tract, and activity of ␦␤␥-ENaC is generally higher than that of ␣␤␥-ENaC. The lack of an inhibitory peptide in ␦-ENaC might therefore lead to an enhanced ENaC P o, but also prevent further increase in P o by impeding the access of extracellular proteases to the ␥-subunit.
The lack of proteolytic processing might also be reflected in the observed expression pattern of ␣and ␦-ENaC in Xenopus tissues. Protease/anti-protease balance is suggested as a mechanism regulating the volume of surface liquid lining pulmonary epithelia (29). This might explain the need for a protease-sensitive ENaC in lung epithelia and the lack of ␦-ENaC in this organ. The presence of ␦-ENaC in osmoregulatory organs, such as the kidney, might constitute an additional level of ENaC control, bypassing proteolytic regulation. Future studies will have to identify the precise cellular localization of ␣and ␦-ENaC as well as the mechanisms regulating the expression and P o of ␦␤␥-ENaC. Interestingly, ␦-ENaC mRNA was detected in testes and oocytes/ovaries. Because experimental expression of the Xenopus ENaC ␤and ␥-subunits does not increase transmembrane currents in oocytes (14), which would occur due to complementation of the channel by endogenous ␦-ENaC, it is likely that translation of (maternal) ␦-ENaC mRNA is blocked by masking proteins (30). This, however, might suggest a role for ␦-ENaC in early embryonic development. Extracellular serine proteases play an important role in Xenopus early embryonic development (31); a functional role for ENaC channels in embryogenesis, however, remains to be proven.
Finally, the data reported herein may provide insights into ENaC evolution. Kleyman's group suggested that proteolytic processing was a crucial step in the evolution of ENaC from a stimulus-activated DEG/ENaC ancestor to a constitutively open channel (10) and hence adaptation to terrestrial life. Unlike any other vertebrate ENaC characterized to date, Xenopus ␦-ENaC is the only channel that is completely insensitive to proteolytic processing and also the only vertebrate ENaC that displays a strong current rundown that is almost comparable with transient currents triggered by acid-sensing ion channels, which are evolutionarily older than ENaC. Furthermore, Xenopus ␦-ENaC is the only ␣-like subunit that does not generate currents when expressed alone in oocytes (14). Xenopus ␦-ENaC might therefore provide functional evidence for Kleyman's hypothesis of proteolytic activation as a prerequisite for constitutively open ENaCs. Interestingly, cloning of three ENaC subunits from the Australian lungfish, Neoceratodus forsteri, revealed the presence of furin cleavage sites in the ␥-subunit, but not in the ␣-subunit (32). Because lungfish are the oldest living ancestors of tetrapods, proteolytic sensitivity of the ␥-subunit might have been a prerequisite for further evolution of proteolytic maturation in ␣-like ENaC subunits in higher tetrapods.
Overall, our findings suggest that subunit composition constitutes an additional level of ENaC regulation, and we propose that the Xenopus ␦-ENaC subunit represents a functional Subunit composition determines ENaC regulation example that demonstrates the importance of proteolytic maturation during ENaC evolution.

Isolation of X. laevis tissues
All procedures and experimental protocols were approved by the Animal Welfare Officer of the University of Giessen (registration number M_478/M_549) as well as the Animal Welfare and Ethical Review Body at Newcastle University (project ID 630). The animal housing facility (Giessen, Germany) was licensed by local authorities (Az: FD 62- §11 JLU Tierphysiologie). The methods used to humanely euthanize the animals were consistent with the recommendations of the AVMA Guidelines for the Euthanasia of Animals. Adult X. laevis were anesthetized for 15 min in 0.2% MS-222/H 2 O (Pharmaq, New Hampshire, UK) at pH 6 and euthanized by decapitation and sounding of the spinal cord. Tissue samples used for RNA isolation were removed and immediately transferred into RNAlater solution (Qiagen, Hilden, Germany). Oocytes were isolated exactly as described previously (7)

RNA isolation and RT-PCR
Total mRNA pools were isolated from X. laevis tissue samples employing the RNeasy extraction system (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The mRNA was reverse-transcribed in cDNA using a poly(A) primer together with the ImProm-II reverse transcription system (Promega, Mannheim, Germany). For controls, identical reactions were performed without reverse transcriptase. The cDNA served as a template in a PCR (GoTaq DNA polymerase kit, Promega) using primers listed in Table 1. PCR conditions were an initial denaturation at 95°C for 5 min, followed by 35 cycles of denaturation at 95°C for 45 s, annealing at 64°C for 30 s, and elongation at 72°C for 45 s. The reaction was terminated with a final extension at 72°C for 10 min. Amplicons were separated on a 1% agarose gel and visualized by MidoriGreen (Nippon Genetics, Dueren, Germany) fluorescence.

Plasmids and cRNA synthesis
Coding sequences for X. laevis ␦-, ␣-, ␤-, and ␥-ENaC were present in the pTNT expression vector (Promega). A MTSETsensitive ␤ S540C -ENaC mutant was generated by site-directed mutagenesis (primers listed in Table 2) using the QuikChange Lightning Kit (Agilent Technologies, Waldbronn, Germany) according to the manufacturer's instructions. The ␦-, ␣-, and ␥-ENaC subunits were tagged with an HA epitope (YPYDVP-DYA) at the N terminus and a V5 epitope (GKPIPNPLL-GLDST) at the C terminus by PCR, using extended primers ( Table 2) and Platinum Pfx DNA polymerase (Invitrogen, Karlsruhe, Germany). PCR was performed by initial denaturation for 5 min at 95°C, followed by 35 cycles of denaturation for 45 s at 95°C, annealing for 30 s at 70°C, and elongation for 180 s at 72°C. The reaction was terminated by elongation for 10 min at 72°C, and amplicons were separated on 0.8% agarose gels. Bands of appropriate sizes were purified using a Gelextract minikit (5Prime, Darmstadt, Germany), and A-overhangs (3Ј) were generated employing the GoTaq DNA polymerase system (Promega). Amplicons were ligated (T4 Ligase, Promega) into the pGEM-T easy vector (Promega) according to the manufacturer's protocol and transformed into competent Escherichia coli (K12, DH5␣). Transformed clones were identified by ampicillin resistance and blue/white selection. Plasmids were isolated employing the QIAprep Spin Miniprep kit (Qiagen) and cut using restriction enzymes (XhoI/XbaI for ␦-ENaC, SalI/NotI for ␣and ␥-ENaC; Promega). Fragments were separated on 0.8% agarose gels, and bands of appropriate sizes were purified (Gelextract Mini Kit; 5Prime) and ligated (T4 Ligase, Promega) into pTNT. All plasmids were sequenced (SeqLab, Goettingen, Germany) to verify successful mutagenesis and epitope tagging.
Plasmids encoding ␦and ␣-ENaC were linearized with BamHI, and plasmids encoding ␥-ENaC were linearized with NaeI (both from Promega). Plasmids containing ␤-ENaC were not linearized due to the presence of BamHI and NaeI restriction sites in the DNA sequence. m 7 G-capped cRNA was generated by in vitro transcription using T7 RNA polymerase (Ribo-MAX large-scale RNA production system, Promega) according to the manufacturer's protocol. cRNAs were diluted in diethylpyrocarbonate-treated H 2 O to a final concentration of 10 ng/l per subunit for two-electrode voltage-clamp recordings, 20 ng/l for patch-clamp recordings, and 300 ng/l for immunodetection. Oocytes were injected with 32 nl of cRNA and incubated for 1-2 days at 16°C in a low-sodium CulORS (10 mM NaCl, 80 mM N-methyl D-glucamine, 1 mM KCl, 2 mM CaCl 2 , 5 mM HEPES, 2.5 mM sodium pyruvate, 0.06 mM penicillin G, and 0.02 mM streptomycin sulfate at pH 7.4). Patchclamp recordings were performed 2-7 days after injection. To inhibit the endoprotease furin, furin inhibitor I (40 M; Cayman Chemical, Hamburg, Germany) was added to the low-sodium CulORS.

Two-electrode voltage-clamp recordings
Recording of I M was performed as described previously (7) using a Turbo Tec-03X amplifier (NPI, Tamm, Germany). If not stated otherwise, the V M was clamped to Ϫ60 mV, and current signals were low-pass-filtered at 1 kHz. ENaC-mediated fractions of I M were determined with the ENaC inhibitor amiloride (100 M; Sigma-Aldrich, Taufkirchen, Germany) and

Subunit composition determines ENaC regulation
defined as amiloride-sensitive currents (⌬I ami ). Unless stated otherwise, ⌬I ami was consistently determined after the oocyte was perfused with ORS for 3 min. All recordings were performed at room temperature.

Patch-clamp recordings
Patch-clamp recordings were performed in the cell-attached configuration using mechanically devitellinized oocytes. Patchpipettes (6 -9-megaohm resistance) were pulled from borosilicate glass capillaries employing a two-stage puller, PP-83 (Narishige, Tokyo, Japan), heat-polished, and filled with pipette solution (145 mM NaCl, 1.8 mM CaCl 2 , 10 mM HEPES, 2 mM MgCl 2 , and 5.5 mM glucose at pH 7.4). Chlorided silver wires served as recording electrodes and were either inserted into the patch pipette or connected to the bath solution (145 mM KCl, 1.8 mM CaCl 2 , 10 mM HEPES, 2 mM MgCl 2 , and 5.5 mM glucose at pH 7.4) using a 1 M KCl, 3% agar bridge. Current signals were amplified using an LM-PC patch-clamp amplifier (List-Medical, Darmstadt, Germany), low-pass filtered at 100 Hz (Frequency Devices, Haverhill, IL), and recorded at 2 kHz with Axon Clampex software (Axon Instruments, Foster City, CA) using an Axon 1200 interface. Single-channel analysis was performed with Clampfit version 10.7 (Axon Instruments). The amount of channels in each recording was stochastically estimated by comparing the number of visible channels with a theoretical amount derived from binomial distribution as described previously (7).

Immunodetection of epitope-tagged ENaC subunits
Oocyte expression of HA/V5-tagged ENaCs was confirmed by measuring V M and excluding cells with a V M Յ 0 mV. Using a 26-gauge needle, 20 -30 oocytes were lysed in 10 l/oocyte of lysis buffer (83 mM NaCl, 10 mM HEPES, 1 mM MgCl 2 , and 1% Triton X-100 at pH 7.4) supplemented with protease inhibitors (Complete mini EDTA-free protease inhibitor mixture tablets, Roche Applied Science). Lysates were centrifuged (1500 ϫ g, 10 min, 4°C), and supernatants were collected, avoiding transfer of the lipid-enriched layer. Lysates were mixed with SDS protein-loading buffer (Roti-Load 1, Roth, Karlsruhe, Germany), boiled for 5 min at 95°C, and loaded (25 l) on 4 -20% SDSpolyacrylamide gels. Some protein samples were treated with PNGase F (New England Biolabs, Frankfurt am Main, Germany) according to the manufacturer's instructions. Following separation in a tank blot system (Mini-Protean Tetra Cell, Bio-Rad, Munich, Germany), proteins were transferred onto polyvinylidene difluoride membranes. Membranes were blocked for 60 min in TBST (150 mM NaCl, 15 mM Tris-HCl, 4.6 mM Tris-base, and 1% Tween 20, pH 7.6) containing 5% skim milk at room temperature. Membranes were incubated with mouse IgG anti-HA (1:10,000; Thermo Fisher Scientific) or anti-V5 (1:5000; BioLegend, San Diego, CA) antibodies at 4°C overnight. After washing with TBST, membranes were incubated with a peroxidase-conjugated rabbit anti-mouse IgG antibody (1:5000; Thermo Fisher Scientific) in blocking buffer for 1 h at room temperature. After washing the membranes in TBST, detection of target proteins was performed using enhanced chemiluminescence solution and photoreactive films (Amersham Biosciences Hyperfilm ECL, GE Healthcare, Freiburg, Germany). Band intensities were quantified using ImageJ (National Institutes of Health, Bethesda, MD) as described previously (15).

Detection of ENaC cleavage fragments at the cell surface
Before biotinylation, ⌬I ami was measured, and oocytes were placed back into low-sodium ORS. All biotinylation procedures were conducted on ice. Oocytes (20 -30 cells/experimental group) were washed in buffer A (90 mM NaCl, 5 mM triethanolamine, 3 mM KCl, 1 mM CaCl 2 , pH 8.0) before incubation in buffer A containing 1 mg/ml biotin (EZ-Link Sulfo-NHS-SS-Biotin, Thermo Fisher Scientific) for 20 min with gentle agitation. Biotinylation was quenched by incubating the cells in buffer A supplemented with 50 mM glycine for 10 min. Using a 26-gauge needle, oocytes were lysed in 1 ml of lysis buffer (90 mM NaCl, 20 mM Tris, and 1% Triton X-100, pH 7.4) containing protease inhibitors (Roche Applied Science). Lysates were vortexed (20 s) and incubated for 20 min on ice. After centrifugation (8000 ϫ g, 10 min, 4°C), the aqueous phase of the supernatants was collected, avoiding transfer of the lipid-enriched layer. Immobilized neutravidin-agarose beads (100 l; Pierce) were washed with 500 l of lysis buffer and centrifuged (1500 ϫ g, 3 min, 4°C). Protein samples were added to the beads and incubated with gentle agitation at 4°C overnight. Samples were centrifuged (1500 ϫ g, 10 min, 4°C), and the supernatants (cytosolic fraction) were collected. The beads were washed three times with 500 l of lysis buffer (1500 ϫ g, 3 min, 4°C) and three times with 500 l of lysis buffer supplemented with 300

Data evaluation
Data are presented in scatter plots with lines and error bars representing mean Ϯ S.E. Oocytes for each experiment were derived from at least three different donors, and the number of individual experiments is noted in parentheses. Immunoblots are representative of at least three independent experiments. Statistical analyses were performed using GraphPad Prism version 7 (GraphPad Software Inc., La Jolla, CA). Gaussian distribution of individual data was assessed with the D'Agostino and Pearson omnibus normality test. Normally distributed data were analyzed by Student's unpaired t test (two-tailed). A Welch's correction was additionally performed if variances were different. Data that did not follow Gaussian distribution were compared using the Mann-Whitney U test (two-tailed). Multiple groups were compared by one-way ANOVA followed by a post hoc Tukey's multiple-comparison test (two-tailed) for normally distributed data or a Kruskal-Wallis test followed by Dunn's multiple-comparison test (two-tailed) for data not following Gaussian distribution. A p value Յ 0.05 was considered significant. All graphics were assembled and finalized using Inkscape.