Syntaxin 1A Regulates ENaC via Domain-specific Interactions*

The epithelial sodium channel (ENaC) is a heterotrimeric protein responsible for Na (cid:1) absorption across the apical membranes of several absorptive epithelia. The rate of Na (cid:1) absorption is governed in part by regulated membrane trafficking mechanisms that control the apical membrane ENaC density. Previous reports have implicated a role for the t-SNARE protein, syntaxin 1A (S1A), in the regulation of ENaC current (I Na ). In the present study, we examine the structure-function rela-tions influencing S1A-ENaC interactions. In vitro pull-down assays demonstrated that S1A directly interacts with the C termini of the (cid:2) -, (cid:3) -, and (cid:4) -ENaC subunits but not with the N terminus of any ENaC subunit. The H3 domain of S1A is the critical motif mediating S1A-ENaC binding. Functional studies in ENaC expressing Xenopus oocytes revealed that deletion of the H3 domain of co-expressed S1A eliminated its inhibition of I Na , and acute injection of a GST-H3 fusion protein into ENaC expressing oocytes inhibited I Na to the same extent as S1A co-expression. In cell surface ENaC labeling experiments, reductions in plasma membrane ENaC ac-counted for the H3 domain inhibition of I Na . Individu- ally substituting C terminus-truncated (cid:2) -, (cid:3) -, or (cid:4) -ENaC subunits for their C and N Termini— Fresh bacterial BL21 colonies harboring His 6 -tagged ENaC cDNAs were cultured in 20 of LB medium containing ampicillin (cid:4) g/ml). Overnight cultures were diluted with pre-warmed LB medium at a ratio of 1:10, cultured at 37 °C for 1–1.5 h and then induced with 1 m M isopropyl- (cid:2) - D -thiogalac- topyranoside at 37 °C with for 3–3.5 h until the A 595 0.35–0.8. Bacterial pellets were resuspended in sonication (100 m M HEPES, M KCl, m M (cid:2) -mercaptoethanol, X-100, m M phenylmethylsulfonyl fluoride, and protease inhib- itor mixture, 1 tablet/50 ml). Cells were lysed by sonication and incubated at 4 °C for 30 min. Lysates were clarified by centrifugation at 12000 (cid:5) g at 4 °C and subsequently incubated with pre-equilibrated nickel-nitrilotriacetic acid beads at room temperature for 30 min. Beads were washed at least three times in Buffer A (20 m M HEPES, m M KCl, 2 m M (cid:2) -mercaptoethanol, 0.5 m M Na 2 -ATP, 10% glycerol, 30 m M imidazole). Fusion proteins were eluted by three washes with 250 m M imidazole in Buffer A. Purified His 6 -tagged ENaC C-terminal or N- terminal proteins were dialyzed in phosphate-buffered saline for 36 h at 4 °C. The purified proteins were verified using Coomassie Blue-stained SDS-PAGE or by Western blotting with monoclonal anti-polyhistidine antibodies (Sigma). Pull-down Assays— These assays were performed as described previously with modifications. Briefly, (cid:4) g of GST-S1A fusion pro- a luminometer er where the signal integrated over a 60-s in relative These and other are expressed as mean statistical differences were

The epithelial Na ϩ channel (ENaC) 1 is located at the apical membranes of aldosterone-responsive epithelial cells of the renal collecting ducts, the descending colon, salivary ducts, and other organs including the airways, where it plays a critical role in the homeostatic control of fluid and electrolyte transport (1). This highly selective pathway for Na ϩ entry is characterized by a low single channel conductance (ϳ5 picosiemens), high affinity blockade by the diuretic amiloride, and a negligible conductance to K ϩ . Functional ENaCs consist of three homologous subunits (␣-, ␤-, and ␥-ENaC) (2) that associate as a heterotrimeric channel complex; the stoichiometry of subunit associations remains unclear (3,4). Each ENaC subunit contains two transmembrane domains, one large extracellular domain and relatively short intracellular N-and C termini (2,5).
To maintain extracellular salt and water homeostasis, Na ϩ entry via ENaC is tightly regulated. A key component of this regulation involves both short and long term control over the number of functional apical membrane ENaCs, which is achieved by changes in their exocytic insertion and/or endocytic retrieval (6). Recently, significant progress has been made in elucidating the mechanisms underlying ENaC endocytosis (for review see Ref. 6). Retrieval of ENaC from the cell surface has been shown to depend upon a highly conserved PPXY sequence within the C termini of ENaC subunits, known as a PY motif. The ubiquitin ligase Nedd4 interacts primarily with the PY motifs of the ␤and ␥-subunits (7), which decreases cell surface ENaC expression by increasing channel internalization and degradation (8). In addition, recent findings (9,10) indicate that the aldosterone-induced protein, the serum-glucocorticoidinduced kinase, sgk1, phosphorylates Nedd4 to reduce its interaction with ENaC and thereby increase apical channel density. The ENaC-Nedd4 interaction plays an important role in ENaC regulation, as genetic deletion or mutation of the PY motifs in the C termini of the ␤or ␥-subunits eliminates Nedd4 interactions and produces Liddle's syndrome (11), a gain of function mutation characterized by an increased number of ENaC channels at the cell surface leading to salt retention (12). In addition, the PY motif has also been implicated as an endocytic signal, which is recognized by the AP-2 adapter complex, mediating ENaC retrieval into clathrin-coated endosomes (13).
Although the mechanisms underlying retrieval of ENaC from the cell surface are becoming more apparent, relatively little is known about the processes involved in exocytic trafficking of ENaC to the apical membrane. Previous studies demonstrating functional and physical interactions between ENaC and syntaxin 1A (S1A) (14,15) suggested that plasma membrane insertion of ENaC is mediated via functional interactions with membrane trafficking proteins of the SNARE complex. Exogenous expression of S1A, a target or t-SNARE protein that is a member of the core complex required for vesicle docking and exocytosis, inhibited Na ϩ currents expressed in Xenopus oocytes, and this inhibition was associated with a decrease in cell surface ENaC expression (14). A physical interaction between S1A and ENaC was implicated from the results of coimmunoprecipitation experiments, which demonstrated that immunoprecipitation of S1A from A6 epithelia co-precipitated ␥-ENaC (14). In a similar study, Saxena et al. (15) found that S1A co-precipitated from Xenopus oocytes with FLAG-tagged subunits of both ␤and ␥-ENaC, whereas possible interactions with ␣-ENaC were not addressed. Although these studies have identified a functional interaction of ENaC with S1A, the precise domains of both proteins involved in these interactions remain unclear.
The aim of the present study was to determine the specific domains involved in the physical and functional interactions between S1A and ENaC. We evaluated the in vitro binding of S1A with the cytosolic ENaC N and C termini in pull-down assays and examined their functional interactions using oocyte co-expression assays. Our results define the domain interactions between S1A and the ENaC subunits and demonstrate reversal of the S1A inhibitory action on I Na by elimination of these domains. These findings, coupled with the effects of S1A on ENaC cell surface expression and on ENaC bearing a PY mutation, support the concept that the S1A-ENaC interaction controls ENaC trafficking mediated by SNARE protein interactions.

EXPERIMENTAL PROCEDURES
Materials-mMESSAGE mMACHINE TM T7 or T3 complimentary RNA synthesis kits were purchased from Ambion Inc. (Austin, TX). DNA Mini-Prep kits, Midi-Prep kits, and nickel-nitrilotriacetic acid beads were obtained from Qiagen. Restriction enzymes, T4 DNA Ligase, Taq polymerase, and Pfu polymerase were purchased from New England Biolabs. Glutathione-Sepharose 4B was purchased from Amersham Biosciences. Complete Protease Inhibitor Mixture tablets were obtained from Roche. All other reagent grade chemicals were obtained from Sigma. Plasmids encoding human ENaC ␣-, ␤-, and ␥-subunits were kindly provided by Dr. Michael Welsh (University of Iowa). Mouse ENaC constructs were generously provided by the laboratory of Dr. Thomas Kleyman (University of Pittsburgh). Subunit truncations were generated by introducing a stop codon by PCR at the designated amino acid (see below).
Syntaxin 1A C Terminus Deletion Mutant (S1A  )-A syntaxin 1A construct deleting amino acids 190 -288 of S1A was amplified from the cDNA by PCR using the following primers: sense primer, 5Ј-ATAAGC-TTATGAAGGACCGAACC; antisense primer, 5Ј-CAGAATTCCTAACT-GAGGGCCTGCTTCGA and sense primer, 5Ј-CGGAATTCATGAAGG-ACCGAACC; antisense primer, 5Ј-ATCTCGAGCTAACTGAGGGCCT GCTTC. The amplicons obtained from the first primer set were cloned into a pcDNA3 vector. The product obtained from the second primer set was introduced into a pGEX-6p-1 vector. Other S1A deletion mutants (see below) were constructed in a similar manner. All constructs were verified by microsequence analysis.
Glutathione S-Transferase-S1A Fusion Proteins-Syntaxin 1A fusion proteins containing GST at the N terminus were produced in BL21 competent cells. GST-H3 contains the S1A H3 domain, S1A 191-267 , GST-H3-TM adds the transmembrane domain, S1A 194 -288 , and GST-⌬ H3 truncates both the H3 and TM domains from full-length S1A, S1A  . These constructs were employed in in vitro pull-down assays with the His 6 -tagged ENaC C termini (domain deletions are also provided with the results; see Fig. 2A). Conditions for bacterial growth and purification of GST fusion proteins have been described previously (16).
Generation of His 6 -tagged ENaC C and N Termini-Fresh bacterial BL21 colonies harboring His 6 -tagged ENaC cDNAs were cultured in 20 ml of LB medium containing ampicillin (50 g/ml). Overnight cultures were diluted with pre-warmed LB medium at a ratio of 1:10, cultured at 37°C for 1-1.5 h and then induced with 1 mM isopropyl-␤-D-thiogalactopyranoside at 37°C with shaking for 3-3.5 h until the A 595 was 0.35-0.8. Bacterial pellets were resuspended in sonication buffer (100 mM HEPES, 500 mM KCl, 8 mM ␤-mercaptoethanol, 5 mM Na 2 -ATP, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, and protease inhibitor mixture, 1 tablet/50 ml). Cells were lysed by sonication and incubated at 4°C for 30 min. Lysates were clarified by centrifugation at 12000 ϫ g at 4°C and subsequently incubated with pre-equilibrated nickel-nitrilotriacetic acid beads at room temperature for 30 min. Beads were washed at least three times in Buffer A (20 mM HEPES, 200 mM KCl, 2 mM ␤-mercaptoethanol, 0.5 mM Na 2 -ATP, 10% glycerol, 30 mM imidazole). Fusion proteins were eluted by three washes with 250 mM imidazole in Buffer A. Purified His 6 -tagged ENaC C-terminal or Nterminal proteins were dialyzed in phosphate-buffered saline for 36 h at 4°C. The purified proteins were verified using Coomassie Blue-stained SDS-PAGE or by Western blotting with monoclonal anti-polyhistidine antibodies (Sigma).
Pull-down Assays-These assays were performed as described previously (16), with modifications. Briefly, 10 g of GST-S1A fusion protein was immobilized on glutathione-Sepharose 4B and incubated with 10 g of His 6 -tagged ENaC proteins in 200 l of a modified DIGNAM D buffer (20 mM HEPES, 50 mM KCl, 2.5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 0.2% Triton X-100, 20 M CaCl 2 , 1 tablet of protease inhibitor mixture/50 ml of buffer, pH to 7.0, with KOH) at 4°C with shaking. The next day, samples were pelleted by centrifugation, washed three times with DIGNAM D buffer at 4°C, and resolved using 15% SDS-PAGE. His 6 -tagged ENaC proteins were detected by use of polyhistidine antibodies (see above). The effect of altered Na ϩ or Ca 2ϩ on ENaC-S1A interactions was assessed by varying the ion concentrations of the binding buffer; solutions containing 0.1 or 10 M free Ca 2ϩ were generated by EGTA buffering, as described (17).
Complimentary RNA (cRNA) Transcription and Oocyte Injection-Vectors containing mouse ENaC, human ENaC, and rat syntaxin 1A inserts were linearized, and cRNAs were synthesized in vitro by use of T7 or T3 cRNA synthesis kits. Oocyte isolation and RNA injection were performed as described previously (18). Briefly, 0.5 ng of cRNA of each ENaC subunit was injected into stage V or VI oocytes. For experiments investigating the effect of S1A co-expression on ENaC function, oocytes were co-injected with 5 ng of S1A. Expression proceeded at 18°C for 16 -24 h in sodium-free ND96 solution before ENaC current recordings. GST fusion proteins were injected in a volume of 50 nl (estimated final concentration, 50 ng/l) into oocytes expressing ENaC, and currents were recorded 1 h after injection. Protein binding studies (above) were performed using cytoplasmic domains derived from human ENaC, whereas the functional studies generally employed mouse ENaC subunit expression. Amino acid identity within the C-terminal cytoplasmic domains of human and mouse ENaC is Ͼ70% (this region is important in S1A-ENaC interactions; see below). In addition, functional experiments with co-expressed S1A, performed using C terminus human ENaC truncation, yielded results identical to those performed with mouse ENaC (see Fig. 3), confirming that these effects are not species-specific.
Electrophysiology-Two-electrode voltage clamp recordings were performed as described (18) using 3 M KCl-filled micropipettes connected to a GeneClamp 500 amplifier (Axon Instruments, Foster City, CA). Oocytes were bathed continuously in ND96 solution as follows (in mM): 96 NaCl, 1 KCl, 1.8 CaCl 2 , 1 MgCl 2 , and 5 HEPES, pH 7.4. In sodium-free ND96, equimolar N-methyl-D-glucamine chloride replaced NaCl. After impalement, membrane potentials were allowed to stabilize before voltage clamping at Ϫ100 mV; amiloride-sensitive Na ϩ currents were recorded as the difference in current before and after addition of 10 M amiloride.
Cell Surface ENaC Labeling-The general approach was based on that of Zerangue et al. (19). Briefly, oocytes expressing mouse ␣-, ␤-FLAG, and ␥-ENaC subunits for 2 days were blocked for 30 min in MBS supplemented with 1 mg/ml of bovine serum albumin (MBS-BSA) and then exposed to MBS-BSA plus 1 g/ml of a mouse monoclonal anti-FLAG antibody (M2; Sigma) at 4°C for 1 h. ␤-ENaC contained the FLAG epitope (DYDKKKD) at the extracellular loop position defined by Firsov et al. (20), which did not alter I Na relative to wt ENaC expression. This was confirmed in parallel current measurements performed prior to antibody labeling. After primary antibody labeling the oocytes were washed six times with MBS-BSA at 4°C and then incubated with MBS-BSA supplemented with 1 mg/ml horseradish peroxidase-coupled secondary antibody for 1 h at 4°C (peroxidase-conjugated AffiniPure F(abЈ 2 ) fragment goat anti-mouse IgG; Jackson Immunoresearch Laboratories, West Grove, PA). After 12 additional washes, individual oocytes were placed in 100 l of SuperSignal ELISA Femto solution (Pierce, Rockford, IL) and incubated at room temperature for 1 min. Chemiluminescence was quantitated in a TD-20/20 luminometer (Turner Designs, Sunnyvale, CA) where the signal was integrated over a 60-s interval and is provided in relative light units. These and other results are expressed as mean Ϯ S.E.; statistical differences were assessed by analysis of variance. (14) and those of Saxena et al. (15) demonstrated that S1A co-expression reduced ENaC currents expressed in Xenopus oocytes. A physical interaction of S1A with ENaC subunits was also demonstrated in co-immunoprecipitation assays (14,15). We reasoned that the ENaC cytoplasmic domains would be the potential binding sites for S1A, a type 2 plasma membrane protein that has no significant structure within the extracellular compartment. To evaluate the cytoplasmic domains of ENaC as potential binding sites for S1A, GST-S1A 4 -267 protein was employed in pull-down experiments with His 6 -tagged ENaC C or N termini. As in similar protein interaction studies, a truncated syntaxin 1A (S1A 4 -267 ) lacking the transmembrane domain was employed for its enhanced solubility and purification properties and because the last 21 amino acids of the S1A C terminus constitute the transmembrane domain, which is not expected to be involved in cytosolic protein-protein interactions. In addition, previous studies (14) have demonstrated that ENaC currents are inhibited to the same extent by co-expression of either full-length S1A or a soluble S1A 4 -267 , lacking the transmembrane domain. As shown in the upper panel of Fig. 1, the C termini of ENaC ␣-, ␤-, and ␥-subunits interacted with GST-S1A. In contrast with these results, no interaction could be detected between S1A and the N terminus of any ENaC subunit (Fig. 1, lower panel). These findings demonstrate that S1A binds to the ENaC subunit cytoplasmic C termini.

S1A Interacts with ENaC Subunit C Termini-Our previous results
H3 Domain of S1A Binds the ENaC C Termini-As the H3 helical region of S1A is an important domain for protein-pro-tein interactions within the SNARE fusion complex (21) and has been shown to interact with other ion channels (22,23), we determined whether the H3 domain of S1A was involved in interactions with ENaC. Three deletion constructs ( Fig. 2A) containing N terminus GST fusions were derived from a fulllength S1A clone for comparison with the data from the GST-S1A used in Fig. 1. These constructs were employed in in vitro pull-down assays with the His 6 -tagged ENaC C termini. The results (Fig. 2B) show that GST fusions containing the syntaxin 1A H3 domain (GST-H3 and GST-H3-TM) generally exhibited a stronger interaction with the ENaC subunit C termini than did the same amount of GST-S1A. This is not unexpected, because prior work (24) has demonstrated that regions Nterminal to the H3 domain have an autoinhibitory function and can reduce H3 domain interactions with substrates. Deletion of the H3 domain abolished the interactions of syntaxin with the ENaC C termini (Fig. 2B, lane 5). As a negative control, GST alone had no significant affinity for the ENaC C termini. These results indicate that the H3 domain of S1A is the critical site for ENaC C terminus binding.
C Terminus Truncations Reduce S1A Inhibition of ENaC Currents-To determine whether the physical interaction between S1A and the ENaC C termini is involved in the inhibition of ENaC current observed with S1A co-expression, we evaluated the effect of S1A on Na ϩ currents expressed from subunits with C terminus truncations. In agreement with previous studies (14,15), S1A co-expression significantly decreased the amiloride-sensitive Na ϩ current in oocytes expressing wt ENaC (Fig. 3, A-D). This inhibition averaged 75%. Substitution of the wt ␣-subunit by ␣-H613X had no significant effect on the magnitude of the amiloride-sensitive I Na , in agree-FIG. 1. Syntaxin 1A interacts with ENaC C termini. Pull-down assays were performed as described under "Experimental Procedures." 10 g of GST or GST-S1A was immobilized on 20 l of glutathione beads and incubated with 10 g of His 6 -tagged C-or N-terminal fragments of the ␣-, ␤-, and ␥-ENaC subunits (designated by subscripts) in binding buffer at 4°C overnight. After three washes, samples were resolved on 15% SDS-PAGE and blotted with poly-His antibodies (1:2000). The experiment was performed three additional times with similar results. Lane 1, GST; lane 2, GST-S1A; lane 3, 10% sample input of His 6 peptides.
FIG. 2. Domain-specific interactions between S1A and ENaC C termini. A, schematic diagram of S1A proteins, fused to GST at their N terminus. B, in vitro pull-down assays were performed as described for Fig. 1, using GST or GST-S1A constructs incubated with His 6 -tagged ENaC subunit C termini (␣ C , ␤ C , and ␥ C ). Blots were probed with anti-His antibodies (1:2000). The experiment was performed two additional times with similar results.
Qualitatively similar results were obtained from individual C terminus truncations of the ␤and ␥-subunits. In contrast to the ␣ truncation, expression of the ␤-R583X significantly increased I Na relative to controls, also consistent with prior findings and the stimulation of I Na observed with Liddle's mutations that truncate ␤-ENaC (12). Fig. 3B shows that, when co-expressed with ␤-R583X ENaC, S1A did not significantly inhibit I Na , indicating that S1A inhibition may be reversed by truncation of the ␤-ENaC C terminus alone. Truncation of the ␥-ENaC C terminus resulted also in a gain of function consistent with mutations producing Liddle's disease. S1A co-expression did not inhibit ␥-R583X ENaC I Na as potently as in wildtype controls (Fig. 3C); the inhibition was reduced to 51% with ␥-ENaC truncation. These results suggest that S1A can interact with each ENaC subunit C terminus to produce an inhibi-tion of I Na , because truncation of individual subunit C termini fully or partially reverse the S1A inhibitory effect. The elimination of S1A inhibition by ␤-ENaC truncation alone suggests that a selective interaction at this site may be sufficient to account for the inhibitory effect of S1A. Finally, we expressed a functional ENaC comprised of three C terminus truncated subunits and determined the influence of S1A co-expression on I Na (Fig. 3D). S1A inhibition was reversed completely under these conditions, as there was no significant difference between the combined ␣␤␥ truncated ENaC currents and those with coexpressed S1A. These findings suggest that the S1A inhibition of I Na involves functional interactions of syntaxin with the C termini of all ENaC subunits, which is consistent with the protein binding data.
An ENaC PY Mutant Is Inhibited by S1A-The PY motif located in the C terminus of the ␤and ␥-ENaC subunits is involved in important interactions with the ubiquitin ligase Nedd4 (7). To examine whether the PY motif may also influence S1A binding and subsequent ENaC inhibition, we measured I Na in oocytes co-expressing a ␤-ENaC PY mutant, FIG. 3. ENaC C terminus truncations reverse syntaxin inhibition. Oocytes were injected with ␣-, ␤-, and ␥-ENaC subunit cRNAs with or without S1A cRNA. Channels formed from C terminus truncation mutants are named by the single mutated subunit, expressed with complementary wild-type subunits. Amiloride-sensitive Na ϩ currents were measured as the difference before and after addition of 10 M amiloride at a holding potential of Ϫ100 mV. Data are expressed as the fraction of the amiloride-sensitive current from each oocyte (n) relative to the wild-type ENaC control mean (I/I wt ) calculated for each animal (N), n ϭ 12; N ϭ 4. The S.E. for wt ENaC currents was calculated from the individual oocyte control values, each taken as a fraction of the wt mean for each animal. A, S1A effect on wt and ␣-H613X ENaC. B, S1A effect on wt and ␤-R564X ENaC. C, S1A effect on wt and ␥-R583X ENaC. D, S1A effect on wt and ENaC formed entirely from the above truncated subunits.
␤-Y618A, together with S1A. As described above and shown in Fig. 4, S1A inhibited wt ENaC current, which was reversed upon truncation of the ␤-ENaC C terminus. Amiloride-sensitive currents were approximately doubled in ␤-Y618A ENaCexpressing oocytes relative to wt ENaC controls (Fig. 4), as has been demonstrated previously for ␤or ␥-ENaC PY mutations (25). However, in contrast to the ␤-ENaC C terminus truncation, co-expression of S1A with ␤-Y618A ENaC resulted in a 76% inhibition of the I Na (Fig. 4). This value is quantitatively similar the inhibition observed for S1A co-expression with wt ENaC (Fig. 3), indicating that the PY motif itself is not involved in the functional interaction with S1A.
The H3 Domain Is Necessary for Inhibition of ENaC Current-The H3 domain of S1A has been shown to regulate other ion channels, and our results demonstrate that it interacts physically with the ENaC C termini (Fig. 2). To test the hypothesis that the H3 domain interaction with ENaC regulates channel function, we co-expressed a S1A H3 deletion mutant (⌬H3) together with ENaC in oocytes and measured amiloridesensitive I Na . Fig. 5A shows that wt S1A significantly inhibited I Na whereas S1A ⌬H3 had no effect. These results indicate that functional inhibition of ENaC requires the H3 domain of S1A. To verify this, we examined the effect of acutely injecting various S1A fusion proteins on I Na ; current recordings were obtained ϳ1 h after injection. As shown in Fig. 5B, injection of GST alone had no effect on I Na . However, injection of GST-H3 significantly attenuated the amiloride-sensitive I Na ; the inhibition obtained from acute injection of GST-H3 was quantitatively similar to that resulting from co-expression of full-length S1A. This acute effect of the H3 domain suggests that S1A overexpression is not generally interfering with protein production, which is in agreement with prior control experiments in which S1A was expressed with other plasma membrane receptors or transporters (26). Finally, injection of a S1A fusion protein lacking the H3 domain, GST-⌬H3, did not significantly affect I Na relative to cells expressing ENaC alone (Fig. 5B). These results support the hypothesis that the H3 domain of S1A is important in its regulation of ENaC activity. (14), we found that co-expression of S1A reduced cell surface ENaC localization without changing total protein expression levels. This finding suggests that S1A may interfere with ENaC trafficking to the plasma membrane. In light of the fairly rapid effect of H3 domain injection on I Na shown above, we sought to determine whether this short term action of GST-H3 also affected the amount of ENaC expressed at the cell surface; alternately, H3 injection may decrease I Na by influencing channel gating. FLAG-tagged surface ENaC expression was assessed using an enzyme-linked luminescence assay developed previously (19) for cell surface K ϩ channel expression; the results are summarized in Fig. 6. The cell surface signal from ␤-FLAG ENaC was about eight times the background level, and the FIG. 4. Mutation of the PY motif does not affect S1A inhibition of ENaC. Oocytes were injected with ␣-, ␤-, and ␥-ENaC subunit cRNAs with or without S1A cRNA. Channels formed from ␤ C terminus mutants are designated by the single mutated subunit and were expressed with complementary wild-type subunits. Wt and ␤-R564X ENaC were used as positive and negative controls, respectively. Amiloride-sensitive Na ϩ currents were measured as described for Fig. 3. Data are expressed as the fraction of amiloride-sensitive current relative to wt ENaC controls (I/I wt ), n ϭ 13-15; N ϭ 4 (see legend for Fig. 3).
latter was determined using oocytes expressing wt ENaC lacking FLAG epitope. Injection of GST-H3 ϳ1 h prior to surface labeling reduced this signal by 70% (corrected for background), whereas prior injection of S1A fusion protein lacking the H3 domain had no effect. This reduction in surface labeling correlates with the 75% reduction in I Na observed in similar H3 injection experiments (Fig. 5B) or when S1A was co-expressed with ENaC (Fig. 3). The data suggest that, even within this time frame, the primary action of the H3 domain is on expression of ENaC at the cell surface.
The ENaC-S1A Interaction Is Salt-sensitive-Increased cellular Na ϩ concentration elicits a decrease in ENaC-mediated apical membrane conductance in epithelial cells. This cellular protective mechanism balances apical Na ϩ entry with basolateral Na ϩ extrusion (1). To determine whether the S1A-ENaC interaction may be involved in feedback regulation of Na ϩ entry, we determined the effect of ambient Na ϩ and K ϩ concentrations in the binding buffer on the physical interaction between syntaxin and ENaC determined in vitro. As shown in Fig. 7, the binding of the ␤-ENaC C terminus to GST-S1A decreased with increasing Na ϩ concentration in the binding buffer, such that 100 mM Na ϩ virtually abolished this interaction. There was also a marked reduction in binding at 50 mM Na ϩ . Increasing buffer K ϩ concentration also reduced the interaction of S1A with the ␤-ENaC C terminus, but elevated K ϩ was less disruptive than Na ϩ (Fig. 7, lower panel). Similar results were obtained for the interaction of S1A with the ␣and ␥-ENaC C termini (data not shown). These data suggest that ionic forces are involved in the ENaC C terminus interaction with S1A and that binding is particularly sensitive to ambient Na ϩ .
Ca 2ϩ Sensitivity of S1A-ENaC Interactions-Apical Na ϩ conductance is also inhibited when cellular free Ca 2ϩ concentration rises (1). We evaluated the Ca 2ϩ and ATP dependence of the ENaC-S1A interactions using in vitro pull-down assays performed in the presence of 0.1 or 10 M free Ca 2ϩ (EGTAbuffered). We also tested the effect of ATP on the S1A-ENaC interaction in vitro. The results in Fig. 8 show that 10 M Ca 2ϩ abolished the S1A-ENaC interaction. The presence of 2.5 mM ATP in the low Ca 2ϩ binding buffer did not influence the interaction of ENaC with S1A. DISCUSSION The results of this study demonstrate a physical interaction between ENaC and syntaxin 1A that modulates amiloridesensitive Na ϩ entry by influencing the density of plasma membrane ENaC channels. Syntaxin 1A binds directly to the C termini of the ␣-, ␤-, and ␥-ENaC subunits in vitro and does not associate with the N terminus of any ENaC subunit. Functional data confirm these sites of interaction, because C-terminal truncations of individual subunits reversed the inhibition associated with S1A co-expression, and a channel formed entirely from truncated subunits was not inhibited by S1A. The   FIG. 6. The S1A H3 domain decreases cell surface ENaC. Oocytes were injected with wt or ␣-, ␤-FLAG-, or ␥-ENaC cRNAs and assayed for cell surface FLAG epitope expression. 1 h prior to surface labeling, oocytes expressing FLAG-ENaC were injected with GST-H3 or GST-⌬H3 as indicated. Amiloride-sensitive I Na was measured prior to labeling in parallel experiments and did not differ between wt and ␤-FLAG-ENaC. Data are expressed as relative light units from luminometry of individual oocytes, n ϭ 12; N ϭ 2.
FIG. 7. Na ؉ and K ؉ concentration dependence of ENaC-S1A interactions. Pull-down assays were performed using the C terminus of ␤-ENaC (as in Fig. 1), except the binding buffers contained the indicated concentrations of NaCl or KCl. Samples were resolved using 15% SDS-PAGE, and blots were probed with poly-His antibodies (1: 2000). Gel locations of the C-terminal ENaC fragments are shown by arrows. The experiment was performed two additional times with similar results. The lower panel provides quantitation of the gel data; see text for other details.
FIG. 8. The ENaC-S1A interaction is sensitive to Ca 2؉ but insensitive to ATP. In vitro pull-down assays were performed to test the interaction between S1A and the C terminus of ␣-ENaC in the presence of 0.1 (L) or 10 M (H) Ca 2ϩ , with or without 2.5 mM ATP. In these experiments, the binding buffer (17) contained 1.2 mM EGTA and sufficient CaCl 2 to yield the free Ca 2ϩ level shown. Samples were separated on 15% SDS-PAGE, transferred to polyvinylidene difluoride membrane, and probed with poly-His antibodies (1:2000). The experiment was performed twice with similar results. H3 domain of syntaxin was critical for this functional effect. Its deletion restored I Na to values not significantly different from expression of wild-type ENaC alone, and acute injection of a GST-H3 fusion protein, but not a S1A GST-⌬H3 fusion protein, inhibited ENaC currents and cell surface ENaC to the same degree as S1A co-expression. Previous studies (14) showed that co-expression of a soluble S1A lacking the transmembrane domain was as inhibitory for ENaC currents as the full-length protein, in agreement with the inhibitory effect of the soluble H3 domain shown here. Increasing the Na ϩ or K ϩ concentrations of the binding buffer reduced ENaC-S1A binding, suggesting that electrostatic forces are involved in these interactions. Although the ENaC C terminus was the structural target for the inhibitory effect of S1A, a Liddle's disease-related mutation in the C terminus PY motif did not obviate the S1A inhibition of ENaC current. This result, together with previous findings (14), has implications for the mechanism of S1A inhibition that will be discussed below.
The rate of Na ϩ entry across the apical membranes of absorptive epithelial cells is determined by the number and open probability of apical ENaC channels. Control over apical ENaC density is a key component in the regulatory actions of vasopressin, aldosterone, and other factors that govern this ratedetermining step in Na ϩ absorption (see above and Ref. 1 for review). Cognate interactions between specific SNARE proteins mediate the fusion of vesicle and target membranes, and they are undoubtedly a key step in the insertion of ENaCcontaining vesicles into the apical membranes during these acute regulatory responses (14,27). Syntaxin 1A is a t-SNARE protein that plays a critical role in formation of the core fusion complex in neurosecretory vesicles, together with soluble Nethylmaleimide-sensitive factor attachment protein-25/23 and vesicle-associated membrane protein-2. Together, these proteins mediate the insertion of membrane vesicles into the plasma membranes at nerve terminals, and in other cell types they act in the final steps that mediate the regulated exocytic events involved in protein secretion or the regulated insertion of integral membrane proteins like ENaC (28). Structural studies have shown that the N-terminal amino acids, 28 -144, of syntaxin 1A form an independent folded domain comprised of three ␣-helices that can interact to form a twisted, left-handed, up-down bundle (29). This closed conformation of S1A is stabilized by the syntaxin regulatory proteins, munc-13 (30) and munc-18 (31); stabilization of this bundled structure by munc-18 regulates the ability of S1A to interact with other SNAREs (24). The inhibitory effect of S1A overexpression on ENaC currents and its reversal by munc-18 (14) suggested that these proteins play a functional role in the insertion of ENaCcontaining vesicles into the plasma membrane.
The bundled structure with which munc-18 interacts occludes the H3 helical domain (amino acids 185-267) and prevents its interaction with soluble N-ethylmaleimide-sensitive factor attachment protein-25/23 and vesicle-associated membrane protein-2. The H3 domain contains the "SNARE motif " that forms the stable tertrameric coiled-coil structure necessary for membrane fusion (21,32,33). This H3 region was found to be responsible for the interaction of syntaxin 1A with ENaC. Deletion of the H3 domain abolished the physical interaction of the ENaC C termini with syntaxin (Fig. 2) and eliminated the inhibition of ENaC current observed with S1A expression (Fig. 3). Co-expression of an H3-deleted S1A (⌬H3) had no effect on ENaC currents, and injection of the H3 domain alone was as inhibitory as the co-expression of full-length S1A (Fig. 4). The H3 domain binds to CFTR and Ca 2ϩ channels, which are also down-regulated by S1A overexpression, and this has been suggested as a general mechanism for syntaxin reg-ulation of ion channel activity (34). As observed in the present studies of ENaC, elimination of or interference with the S1A binding site on CFTR, or on the voltage-dependent Ca 2ϩ channel, abolishes the inhibitory effect of syntaxin 1A (22,35). In the case of CFTR, however, truncation of the S1A binding N terminus interferes with channel processing and reduces CFTR currents to about 10% of the level of wt CFTR. This is not observed for ENaC, because truncation of the S1A binding domains either has no effect (␣-subunit) or augments (␤-and ␥-subunits) ENaC currents, permitting a clear evaluation of the functional consequences of S1A on the truncated channel.
Previous studies of the functional effect of co-expressed syntaxin 1A on ENaC currents in Xenopus oocytes demonstrated an inhibition of amiloride-sensitive Na ϩ entry that could not be ascribed to a nonspecific effect on ENaC protein expression (14,15). The study of Qi et al. (14), attributed the decrease in ENaC current to a reduction in the number of cell surface channels, detected by antibody labeling of non-permeabilized oocytes expressing extracellular FLAG-tagged ENaC subunits. Nevertheless, in a conceptually similar study Saxena et al. (15) found that co-expression of S1A increased cell surface ENaC. This result would require a proportionately greater decrease in channel open probability to override the apparent increase in channel number. However, the surface labeling conditions employed in these experiments are questionable. After fixation of the oocytes in 3% formaldehyde, they were labeled sequentially with primary and secondary antibodies, each for 1 h at 37°C. Given these conditions, the blotchy ENaC staining detected at low magnification may reflect access of the antibody to intracellular epitope. In the experiments of Qi et al. (14), antibody labeling was performed without fixation or permeabilization at 4°C, and the resulting ENaC staining pattern was uniform. In the present study, we used luminometry to detect ENaC at the cell surface. The S1A H3 domain suppressed ENaC I Na by 75% approximately 1 h after its injection (Fig. 5), and this relatively rapid effect could be attributed to a quantitatively similar reduction in ENaC expression at the cell surface (Fig. 6).
Previous studies (7,13) have presented evidence for a relatively rapid turnover of channel protein at the plasma membrane, and they have identified the mechanisms responsible for endocytic retrieval of ENaC. Studies in Xenopus oocytes examining the decay of I Na following inhibition of ENaC traffic to the plasma membrane estimate the half-life of cell surface ENaC to be about 1 h. Accordingly, the 70% reduction in surface ENaC expression detected about 1 h after H3 domain injection would be in keeping with this rapid turnover of channels at the surface. An H3 domain-mediated block of ENaC insertion, together with an inherently rapid rate of channel retrieval, would reduce I Na through a primary effect on channel number.
Inferences regarding the mechanism whereby S1A would reduce ENaC surface expression can be made also from the ENaC mutation analysis. Studies of amiloride-sensitive, genetic hypertension (reviewed in Ref. 6) have implicated structures in the C termini of ENaC subunits in the control of plasma membrane channel number, perhaps by two mechanisms. First, the PPPXYXXL motif within the subunit C termini may serve as an internalization or endocytic motif (13). Second, this motif participates in a physical interaction with the ubiquitin ligase, Nedd4 -2, which binds to PY motifs in the ENaC C termini and decreases ENaC current by reducing cell surface channel number (7,8). Nedd4-mediated ubiquitination of the ENaC N termini promotes channel internalization by endocytosis and its degradation in lysosomes (38).
Syntaxin 1A binds to the ENaC C termini, and its inhibition of ENaC current was eliminated by C terminus truncations.
These findings could be compatible with the concept that S1A promotes ENaC retrieval by a mechanism that depends on Nedd4-mediated ENaC endocytosis/degradation. Similar to C terminus truncations, mutants in the PY motif augment ENaC currents (see Figs. 3 and 4) by increasing cell surface ENaC; Nedd4 binding to ENaC is disrupted by C terminus truncation or by mutation of the PY motif (7,25). However, a mutant that disrupts the PY motif in ␤-ENaC, Y618A, was strongly inhibited by S1A co-expression. The persistence of S1A inhibition in the PY mutant therefore suggests that the action of S1A is not related to Nedd4 binding or to the role of the PY motif in ENaC endocytosis, although we cannot formally rule out stimulation of an endocytic process by S1A that does not involve PY. Nevertheless, it seems unlikely that the influence of S1A on cell surface ENaC is because of stimulation of ENaC removal from the cell surface, because the effects of truncation and PY mutation would be expected to affect S1A inhibition similarly. It is more likely that the reduction in cell surface channel number detected here, and in the studies of Qi et al. (14), arises from inhibition of the insertion of ENaC channels into the plasma membrane.
Recent findings have implicated Nedd4 in the inhibition of Na ϩ entry that is associated with increased intracellular Na ϩ concentration (39). In addition, truncation of the ENaC C termini attenuates this feedback effect of intracellular Na ϩ (40). These findings led us to evaluate the influence of increased salt concentrations and Ca 2ϩ on the physical interaction between S1A and the ENaC C termini. Increasing Na ϩ or K ϩ concentration of the binding buffer decreased this interaction, and Na ϩ was a more potent disruptor of S1A binding than K ϩ , as would be expected from a Na ϩ -selective feedback event. However, inasmuch as the interaction of ENaC with expressed S1A is itself inhibitory, it seems difficult to infer a role for Na ϩmediated disruption of this interaction in the process of feedback inhibition. Disrupting an inhibitory interaction should increase Na ϩ transport. However, this reasoning assumes that the physiological action of endogenous S1A is inhibitory. Rather, if endogenous syntaxin 1A normally mediates apical ENaC insertion, or if it positively regulates trafficking reactions that deposit more ENaC in the plasma membrane, then a Na ϩ -induced disruption of the S1A-ENaC interaction could decrease cell surface ENaC and reduce Na ϩ currents. Similar conclusions would apply to the inhibition of S1A-ENaC binding observed at physiologically high Ca 2ϩ concentrations (10 Ϫ5 M), because increased intracellular Ca 2ϩ is also inhibitory to ENaC currents (1). Our findings concerning the cation dependence of S1A-ENaC interactions indicate that electrostatic forces are involved in their physical association. However, elucidating a potential role for S1A in ENaC regulatory effects that involve changes in cellular composition will require a better understanding of the physiological role of endogenous syntaxin 1A in regulating apical ENaC density.
How does the effect of overexpressed S1A relate to its physiological action on sodium transport? On one hand, the inhibition of ENaC current associated with S1A overexpression may result from disruption of normal SNARE-mediated ENaC trafficking mechanisms that are responsible for channel insertion into the plasma membrane. The influence of overexpressed syntaxins on specific trafficking pathways is commonly used to infer a physiological role for a specific syntaxin isoform in a specific trafficking pathway. For example, the selective inhibition of endoplasmic reticulum to Golgi traffic observed with exogenous syntaxin 5 expression reflects its physiological role as the t-SNARE in this step of protein secretory pathway traffic (36). Syntaxin 5 overexpression is sometimes used as an experimental means of interfering with vesicle-mediated protein transport between endoplasmic reticulum and Golgi (37). Other examples of selective syntaxin isoform inhibition are provided in Qi et al. (14). However, according to this view, the effect of exogenously expressed syntaxin would arise from disruption of the stoichiometric protein interactions necessary for membrane fusion, a general mechanism that should lack specificity for vesicle cargo, in this case, ENaC. Conversely, the protein binding studies and the elimination of S1A inhibition by ENaC subunit truncation argue for a more selective phenomenon, related to the physical presence of ENaC in the trafficking vesicles. The present findings suggest that domain-specific interactions between the ENaC C-terminal cytoplasmic tails and S1A are involved in regulating plasma membrane channel number. It remains to be determined whether apical S1A is the principal t-SNARE that determines the insertion of ENaC containing vesicles into the epithelial cell apical membrane or whether it may compete with another syntaxin isoform that mediates the insertion of ENaC channels.