Small Molecule Activator of the Human Epithelial Sodium Channel*

The epithelial sodium channel (ENaC), a heterotrimeric complex composed of α, β, and γ subunits, belongs to the ENaC/degenerin family of ion channels and forms the principal route for apical Na+ entry in many reabsorbing epithelia. Although high affinity ENaC blockers, including amiloride and derivatives, have been described, potent and specific small molecule ENaC activators have not been reported. Here we describe compound S3969 that fully and reversibly activates human ENaC (hENaC) in an amiloride-sensitive and dose-dependent manner in heterologous cells. Mechanistically, S3969 increases hENaC open probability through interactions requiring the extracellular domain of the β subunit. hENaC activation by S3969 did not require cleavage by the furin protease, indicating that nonproteolyzed channels can be opened. Function of αβG37Sγ hENaC, a channel defective in gating that leads to the salt-wasting disease pseudohypoaldosteronism type I, was rescued by S3969. Small molecule activation of hENaC may find application in alleviating human disease, including pseudohypoaldosteronism type I, hypotension, and neonatal respiratory distress syndrome, when improved Na+ flux across epithelial membranes is clinically desirable.

Epithelial sodium channels (ENaCs) 2 are members of the ENaC/degenerin family of ion channels that includes acidsensing ion channels (ASIC) in mammals, mechanosensitive degenerin channels in worms, and FMRF-amide peptide-gated channels in mollusks (1). ENaC mediates apical membrane Na ϩ transport across high resistance epithelia in numerous tissues, including kidney, colon, and lung. ENaC, a heterotrimeric channel composed of ␣, ␤, and ␥ subunits, is involved in the maintenance of extracellular volume and blood pressure, absorption of fluid from the lungs during late stages of gestation, and electrolyte homeostasis (2,3). Gain of function human ENaC (hENaC) mutations result in hypertension because of increased renal Na ϩ reabsorption in Liddle's syndrome (4,5), whereas loss of function hENaC mutations result in salt wasting because of decreased renal Na ϩ reabsorption in pseudohypoaldosteronism type I (PHA1) (6,7).
The diuretic amiloride inhibits transepithelial Na ϩ transport across reabsorptive epithelia by blocking apical membrane ENaC function (8). The concentration of amiloride yielding 50% ENaC inhibition (IC 50 ) is ϳ100 nM, whereas the more potent and specific derivatives phenamil and benzamil exhibit IC 50 values ϳ10 nM. Thus, high affinity ENaC blockers have been identified (9).
Conversely, potent and specific small molecule ENaC activators have not been previously described. Molecules reported to increase ENaC Na ϩ transport, such as glybenclamide, CPT-cAMP, capsazepine, and icilin, all suffer from low efficacy, low specificity, and/or lack of effect on wild-type ␣␤␥ hENaC. Glybenclamide, an inhibitor of members of the ATP-binding cassette superfamily, including the cystic fibrosis transmembrane conductance regulator (CFTR) and the sulfonylurea receptor, increased amiloride-sensitive ENaC current by 50 -100% (apparent concentration yielding 50% ENaC activation (EC 50 ) of 45 M) in Xenopus oocytes and A6 kidney cells; glybenclamide activation of ENaC required the extracellular and transmembrane domains of the ␣ subunit and doubled the open probability of individual channels (10,11). It is unclear if glybenclamide effects were indirect and attributable to primary regulation of ATP-binding cassette transporters or other glybenclamide receptors that secondarily regulate ENaC activity. CPT-cAMP increased amiloride-sensitive current by ϳ300% (no EC 50 determined) in Xenopus oocytes expressing chimeric ENaC channels composed of guinea pig ␣ subunits in combination with rat ␤␥ subunits; however, CPT-cAMP had no effect on wild-type, nonchimeric ENaC channels (12,13). Openers of ␦ ENaC, another sodium-selective pore-forming channel, have recently been reported. Capsazepine, a nonspecific inhibitor of TRPV1, and icilin, an activator of TRPM8 temperature-sensitive nonselective cation channels, increased amiloride-sensitive current by 100 -200% (apparent EC 50 of 8 and 33 M, respectively) in Xenopus oocytes expressing ␦ or ␦␤␥ hENaC but not ␣ or ␣␤␥ hENaC (14,15). In addition, agents that activate ENaC by removing Na ϩ self-inhibition, the rapid, transient increase in inward current followed by a decline to a lower steady state level after removal of amiloride or switch from a low to high extracellular Na ϩ solution, have been reported, including Zn 2ϩ ions, the mercurial agent p-chloromercuribenzoate, and trypsin-like as well as furin serine proteases (16 -21).
This study describes the first compound, S3969, that fully and reversibly activates hENaC in heterologous cells. Mechanistically, S3969 requires the extracellular domain of the ␤ subunit and increases hENaC open probability, thereby augmenting flux of Na ϩ ions into cells. Accordingly, S3969 activated ␣␤G37S␥ hENaC, a channel variant that exhibits reduced open probability and leads to the salt-wasting disease PHA1. Promoting hENaC-dependent Na ϩ and fluid transport across epithelia may find utility in modulating extracellular volume and electrolyte homeostasis in both normal and diseased states.
␤ hENaC and ␤ mENaC chimeras were constructed using overlapping PCR with Pfu Turbo DNA polymerase (Stratagene) to amplify individual fragments and PCR SuperMix High Fidelity (Invitrogen) to amplify overlapping products. Chimeras, summarized in Table 1, were sequenced to confirm desired junctions and the absence of other mutations introduced during amplification.
In Vitro Transcription-ENaC cRNA was generated from linearized plasmids using the mMESSAGE mMACHINE kit with T7 RNA polymerase according to the manufacturer's instructions (Ambion, Austin, TX). cRNA quality was checked by denaturing agarose gel electrophoresis and spectrophotometric absorbance readings at 260 and 280 nm to ensure that fulllength, nondegraded cRNA was generated.
Frog Surgery, Oocyte Isolation, and Microinjection-Female Xenopus laevis South African clawed frogs greater than or equal to 9 cm in length were obtained from NASCO (Fort Atkinson, WI). Frogs were anesthetized in 0.15% ethyl-3-aminobenzoate methanesulfonate (Sigma) in distilled water and placed on ice. Using sterile surgical tools, sequential 1-2-cm incisions were made in the abdomen through both the outer skin layer and the inner peritoneal layer to reveal the ovaries. Excised ovarian lobes (containing 1000 -2000 oocytes) were placed in OR-2 calcium-free media (82.5 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 5 mM HEPES, pH 7.5, with NaOH) and sequentially digested with 2 mg/ml collagenase type IA (Sigma), prepared immediately before use, for 45 min followed by 1 mg/ml collagenase type IA for 15 min on a rocking platform at room temperature. After enzymatic digestion, at which point the majority of oocytes are released from the ovarian lobes, oocytes were thoroughly rinsed in OR-2 without collagenase and transferred to a Petri dish containing Barth's saline (88 mM NaCl, 2 mM KCl, 0.82 mM MgSO 4 , 0.33 mM Ca(NO 3 ) 2 , 0.41 mM CaCl 2 , 2.4 mM NaHCO 3 , and 5 mM HEPES, pH 7.5; Specialty Media, Phillipsburg, NJ) supplemented with 2.5 mM sodium pyruvate. Mature stage V or VI oocytes (ϳ1 mm diameter) were selected for microinjection. Frog skin was closed using a C6 needle with a 3-0 black braid suture (Harvard Apparatus, Holliston, MA), and frogs were reused for subsequent oocyte isolations following a 2-3-month recovery period. Oocytes were microinjected in the animal pole with 10 -15 nl containing 1 ng of each ENaC subunit cRNA. Following microinjection, oocytes were incubated in Barth's solution supplemented with 2.5 mM sodium pyruvate at 18°C overnight.
Two-electrode Voltage Clamping-ENaC function was measured using the two-electrode voltage clamp technique on an OpusXpress 6000A parallel oocyte voltage clamp system (MDS Analytical Technologies) 24-h post-microinjection unless noted otherwise. The OpusXpress system is an integrated work station that allows electrophysiological recordings to be made from up to eight oocytes simultaneously. This system has been used previously to examine the function of ion channels, including nicotinic acetylcholine and serotonin 5HT3 receptors as well as CFTR (23)(24)(25). Oocyte impalement is automated, and compound delivery is performed by a computer-controlled fluid handler; compounds are removed from 96-well plates using disposable pipette tips and applied to individual oocytes. Oocytes were placed in the OpusXpress system and perfused with ND-96 solution (96 mM NaCl, 2.5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 5 mM HEPES, pH 7.5, with NaOH). Oocytes were then impaled with voltage-sensing and current-passing electrodes back-filled with 3 M KCl. Electrodes exhibited resistances between 2 and 10 megohms for voltage-sensing electrodes and between 0.5 and 2 megohms for current-passing electrodes. Following impalement, oocytes were voltage-clamped to Ϫ60 mV, and experimental recordings were initiated. Data were acquired at 50 Hz and low pass-filtered at 5 Hz.  H-M  N-term, TMD1  1-71  ECL, TMD2, C-term  72-638  H-M-H  N-term, TMD1, TMD2, C-term  1-71; 534-640  ECL  72-531  M-H  ECL, TMD2, C-term  72-640  N-term, TMD1  1-71  M-H-M  ECL  72-533  N-term, TMD1, TMD2, C-term  1-71, 532-638 Cell Culture and Transfection-HEK293 cells were maintained in high glucose Dulbecco's modified Eagle's medium with glutamine (Invitrogen) containing 10% bovine calf serum (Hyclone, Logan, UT) and penicillin/streptomycin (Invitrogen) in 6% CO 2 -balanced air at 37°C. CHO cells were maintained as above in Kaighn's modified FK12 Nutrient Mixture (Invitrogen) containing 10% bovine calf serum and penicillin/streptomycin. Cells were subcultured by trypsinization (0.25% trypsin, 1 mM EDTA in Hanks' balanced salt solution; Invitrogen) and seeded on number 1 glass coverslips (Warner Instrument Corp., Hamden, CT) coated with growth factor-reduced Matrigel (1:200; BD Biosciences) for HEK293 cells or BD BioCoat coverslips pre-coated with poly-D-lysine (BD Biosciences) for CHO cells. Transient transfection was carried out using the Whole-cell Patch Clamp Electrophysiology-Whole-cell currents were acquired from cells 24 -72 h post-transfection at room temperature (21-24°C) using an Axopatch200B amplifier and DIGIDATA 1322A with pCLAMP9.2 software (MDS Analytical Technologies). Currents were sampled at 2 kHz and filtered at 1 kHz. Patch pipettes were pulled from borosilicate glass and had resistances of 3-6 megohms. The pipette solution contained the following (in mM): 105 CsF, 35 NaCl, 10 EGTA, 2 Na-ATP and 10 HEPES, pH 7.4. The bath solution contained the following (in mM): 150 NaCl, 2 KCl, 1.5 CaCl 2 , 1 MgCl 2 , and 10 HEPES, pH 7.4. After establishment of the whole-cell configuration, currents were measured at a holding potential of Ϫ80 mV.
Single Channel Patch Clamp Electrophysiology-Single channel currents were measured at room temperature in oocytes using the outside-out patch clamp configuration. Oocytes were incubated for 5-10 min in ND-96 supplemented with 100 mM NaCl, and the vitelline membrane was peeled away using fine forceps. Single channel currents were recorded using an Axopatch200B amplifier and DIGIDATA 1322A as with wholecell recordings. Currents were acquired at 4 kHz and low passfiltered at 1 kHz. Patch pipettes were pulled from borosilicate glass and fire-polished and had resistances of 8 -12 megohms. Patches were voltage-clamped to a pipette potential of Ϫ80 mV. Bath solution contained 115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl 2 , and 10 mM HEPES, pH 7.5. Pipette solution contained 88 mM KCl, 9.6 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 11 mM EGTA, 2 mM Na-ATP, and 5 mM HEPES, pH 7.5. For data analysis and presentation, currents were subsequently filtered at 200 Hz in pCLAMP10.2. NP o was determined over 1-2-min intervals from idealized traces using the single channel search feature. Similar values were obtained by fitting a Gaussian to an allpoints current amplitude histogram. Open probability (P o ) was estimated by dividing NP o by the number of channels in the patch.
Compounds-Amiloride and trypsin (salt-free, lyophilized powder, minimum 10,000 benzoyl-L-arginine ethyl ester units/mg protein, L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated) were from Sigma. MTSET was from Toronto Research Chemicals (North York, Ontario, Canada) and Biotium (Hayward, CA). S3969 was diluted to appropriate concentrations in ND-96 from 100 mM stock solutions in DMSO. The final concentration of DMSO in experiments was Ͻ0.1%; this level of vehicle had no effect on ENaC function.
Synthesis of S3969-A potassium iodide/iodine solution, prepared by mixing iodine (9.7 g, 38.2 mmol in 75 ml water) with potassium iodide (19.0 g, 114.5 mmol in 225 ml water), was added dropwise to a solution of 4-methylindole (5.0 g, 38.1 mmol) and thiourea (2.9 g, 38.1 mmol) in methanol (80 ml). The reaction mixture was stirred at room temperature for 2 h and concentrated to dryness by rotary evaporation. The crude solid was redissolved in boiling water, treated with decolorizing carbon, and filtered using filter paper. The solution was cooled to room temperature and was stored at 4°C overnight. The resulting pale yellow crystals were collected by filtration, washed with cold water (2 times, 25 ml), and dried under vacuum to give a 69% yield of 4-methyl-1H-indol-3-yl carbamimidothioate hydroiodide. This product (1.0 g, 3.0 mmol) was mixed with 2 ml of 1.5 N NaOH, heated to 90°C, and stirred for 30 min. 2-Bromo-4-methylpentanoic acid (0.6g, 3.1 mmol) was added, and the reaction mixture was stirred for 2 h. The solution was cooled to room temperature, acidified to pH 2 with 1 N HCl, and extracted with ethyl acetate (three times, 10 ml). The combined organic layers were dried with sodium sulfate, filtered, and concentrated under reduced pressure. Purification by flash chromatography (dichloromethane, 5% MeOH) afforded 4-methyl-2-(4-methyl-1H-indol-3-ylthio)pentanoic acid in 75% yield as a colorless oil. This product (305.1 mg, 1.1 mmol) in dimethylformamide (4 ml) was combined with N-hydroxybenzotriazole (135.1 mg, 1.0 mmol), 1-ethyl-3-(3Ј-dimethylaminopropyl) carbodiimide hydrochloride, (211.0 mg, 1.1 mmol), and 2-aminoethanol (61 l, 1.0 mmol). The reaction mixture was stirred at room temperature for 16 h, quenched with 1 ml of water and extracted with dichloromethane (three times, 10 ml) and water. The combined organic layers were collected, dried with sodium sulfate, and concentrated under reduced pressure. Purification by preparative thin layer chromatography (dichloromethane/MeOH) afforded S3969 in 67% yield as an off-white powder. The identities of intermediate reaction products were confirmed by 1 H NMR and liquid chromatography/mass spectrometry, and the identity of S3969 was confirmed by 1 H NMR, 13 C NMR, liquid chromatography/mass spectrometry, and elemental analysis.
Statistics and Measurements-Data represent the mean Ϯ S.E. Unless otherwise noted, experiments were performed on two to four batches of independently injected oocytes harvested from different frogs. Statistical significance between different groups was determined using an unpaired, two-tailed Student's t test. Dose-response curves were plotted, and both EC 50 values and Hill coefficients were determined using GraphPad Prism version 3.02 (GraphPad Software, San Diego). Values for per-cent hENaC activation were calculated by dividing the magnitude of the inward current induced by S3969 by the magnitude of the inward current blocked by amiloride in the same oocyte and multiplying the ratio by 100%. The activity of S3969 was expressed as a percentage of the paired amiloride response because of variability in hENaC expression between different batches of oocytes. Amiloride was used at 1 M for ␣␤␥ hENaC and 10 M for ␦␤␥ hENaC in oocytes; these concentrations yield greater than 90% hENaC inhibition. Similar percent of hENaC activation ratios were obtained in oocytes expressing amiloride-sensitive currents ranging from 0.1 to 1.0 A that were typically observed in these studies. For whole-cell patch clamp experiments, mammalian cells with amiloride-sensitive inward currents less than 1 nA were studied. Cells with amiloride-sensitive currents greater than 1 nA, suggestive of channels opened by proteolysis, responded poorly to S3969.

Identification of a Small Molecule
Opener of ␣␤␥ hENaC-A high throughput screen was conducted using a fluorescent membrane potential dye to identify novel compounds that open hENaC channels expressed in heterologous cells (26). When a compound opens hENaC, sodium flows into the cell resulting in cell depolarization and an increase in dye fluorescence. This study describes in detail one of the more potent compounds, termed S3969 (Fig. 1A). Chemistry optimization and structure activity relationships of S3969 functional groups will be described elsewhere. Treatment of oocytes expressing ␣␤␥ hENaC with 1 M S3969 activated hENaC by 300.1 Ϯ 12.2% (n ϭ 85 determinations on separate oocyte batches using three independent synthetic preparations of S3969), and the time required for S3969 to achieve half-maximal hENaC activation was 33.1 Ϯ 1.6 s (Fig. 1B). The maximal stimulatory effect of S3969 was reached ϳ3 min following compound application, and currents returned to base line following compound washout (Fig. 1B), suggesting that interaction of S3969 with hENaC was noncovalent. Repeated application of S3969 did not result in hENaC desensitization; the ratio of hENaC activation from two successive S3969 applica-tions was 95.0 Ϯ 3.0% (n ϭ 11). Amiloride inhibited current activation by S3969 demonstrating that S3969 effects were hENaC-dependent (Fig. 1C). S3969 failed to activate inward currents in sodium-free solution, indicating that S3969 responses in sodium-replete solution were attributable to sodium influx (Fig. 1D). In sodium-free solution, small outward currents, attributable to sodium efflux from oocytes, were observed; the S3969-activated outward current in sodium-free solution was 25.5 Ϯ 5.8% (n ϭ 11) of the inward amiloridesensitive current in sodium-replete solution (Fig. 1D). Doseresponse experiments, performed to determine the potency of S3969, revealed that the threshold for S3969 hENaC activation was ϳ30 nM, and maximal hENaC activation was achieved at ϳ30 M (Fig. 1E). The apparent EC 50 for S3969 activation of hENaC was 1.2 Ϯ 0.1 M (n ϭ 46 determinations on separate oocyte preparations using three independent synthetic preparations of S3969) with a Hill coefficient of 1.0 Ϯ 0.1 (Fig. 1F). Current-voltage (I-V) curves from hENaC-expressing oocytes were linear, and S3969 increased amiloride-sensitive hENaC currents at both negative and positive holding potentials (Fig.  1G). By contrast, S3969 had no effect on currents at any holding potential in uninjected oocytes, illustrating that S3969-induced currents required hENaC expression (Fig. 1H). Treatment of HEK293 or CHO cells transiently expressing ␣␤␥ hENaC with 10 M S3969, applied by bath perfusion, activated hENaC by 357.2 Ϯ 82.4% (n ϭ 24) and 148 Ϯ 50.7% (n ϭ 10) respectively (Fig. 1, I and J), indicating that S3969 enhances hENaC in both amphibian and mammalian expression systems. The time required for S3969 to achieve half-maximal hENaC activation was 18.9 Ϯ 1.8 s in HEK293 cells and 20.0 Ϯ 2.3 s in CHO cells, qualitatively similar to the kinetics of hENaC activation by S3969 in oocytes. Amiloride inhibited current activation by S3969 (Fig. 1J), and S3969 did not affect currents in mocktransfected cells (data not shown), demonstrating that S3969 effects were hENaC-dependent. These results demonstrate that S3969 is a novel, potent, and reversible activator of hENaC.
S3969 Activates hENaC following Removal of Amiloride-Experiments were next performed to begin to elucidate the mechanism of hENaC activation by S3969. Application of amiloride only (1 M; Fig. 2A, treatment a) inhibited hENaC, and application of S3969 only (1 M; Fig. 2A, treatment b) activated hENaC. Coapplication of 1 M amiloride (a dose that yields greater than 90% hENaC inhibition) and 1 M S3969 (a dose that yields ϳ50% maximal hENaC activation) inhibited hENaC ( Fig. 2A, treatment c) similar to subsequent application of 1 M amiloride alone ( Fig. 2A, treatment d). Under these conditions, . G, I-V curves plotting amiloride-sensitive currents for control (untreated) and S3969 (1 M)-treated oocytes expressing ␣␤␥ hENaC. Currents were measured from Ϫ100 to ϩ60 mV in 20-mV increments 3 min following application of 1 M S3969 (n ϭ 8). Reversal potentials, the points where I-V curves cross the x axis, were between 0 and 10 mV, consistent with previous reports (17,38), and attributable to increased intracellular sodium concentrations following ENaC expression in oocytes. H, I-V curves plotting whole-cell currents for control (untreated) and S3969 (1 M)-treated uninjected oocytes (n ϭ 8). Reversal potentials were highly negative, indicating currents were attributable to endogenous potassium and chloride channels (77). S3969 had no effect on measured currents in the absence of hENaC expression.  , without amiloride priming, induced slow hENaC activation. C, quantitation of % hENaC activation (magnitude of the inward current induced by S3969 divided by the magnitude of the amiloride-sensitive current, multiplied by 100%) for control (no amiloride priming; treatment d in B) and amiloride priming (treatments b and c in B) protocols (n ϭ 7). In the amiloride-primed group, base-line currents for S3969 activation were measured before amiloride treatment a; this eliminates an additional 100% hENaC enhancement that would be obtained by measuring base-line currents for S3969 activation at treatment b, when all hENaC channels are nonconducting. If this correction is not performed, S3969 activates hENaC by 442.7 Ϯ 26.0% compared with 355.6 Ϯ 27.7% with correction. D, apparent time required to reach half-maximal hENaC activation (t1 ⁄2 ) is faster following amiloride priming (n ϭ 7). *, p Ͻ 0.001 compared with control. the effect of the hENaC blocker amiloride was dominant to the effect of the hENaC activator S3969. Following washout of amiloride and S3969 ( Fig. 2A, immediately after treatment c), current rapidly increased to levels observed when oocytes were treated with S3969 alone (Fig. 2A, treatment b) before returning to near control levels. These results suggest that amiloride dissociates from hENaC faster than S3969, resulting in immediate, albeit transient, hENaC activation following removal of both compounds.
Amiloride is thought to occlude the pore of ␣ENaC subunits in an open conformation; thus, amiloride functions as an openstate blocker (27). Experiments were performed to synchronize hENaC in the open state prior to stimulation with S3969. Oocytes were first treated with amiloride to prime hENaC in the open but occluded state (Fig. 2B, treatment b). Next, oocytes were treated with S3969 in the continued presence of amiloride (Fig. 2B, overlap of treatments b and c); no activation of hENaC was observed, consistent with data in Fig. 2A. Removal of amiloride, in the continued presence of S3969, resulted in an immediate and sustained activation of hENaC (Fig. 2B, treatment c), similar in magnitude to current activation with S3969 alone (Fig. 2B, treatment d). Quantitation of these data revealed no difference in hENaC activation when oocytes were treated with S3969 before or after amiloride priming (Fig. 2C). However, amiloride priming significantly decreased the apparent time required for S3969 to achieve half-maximal hENaC activation by a factor of 7 (from ϳ35 to ϳ5 s; Fig. 2D). Taken together, these data indicate that S3969 activates ENaC following removal of amiloride.
S3969 Increases hENaC Open Probability-Macroscopic hENaC current is dependent on the number, single channel current, and open probability of functional hENaC channels in the plasma membrane. S3969 could activate hENaC by increasing one or more of these parameters. The immediate stimulatory effect of S3969 on hENaC function following amiloride priming (Fig. 2B) is temporally inconsistent with a mechanism whereby S3969 increases the number of hENaC channels in the plasma membrane by promoting vesicular trafficking of hENaC from an intracellular pool to the cell surface; agonist-stimulated trafficking of ion channels, including ENaC, to the plasma membrane in oocytes requires tens of minutes or more (28,29).
To determine whether S3969 promotes hENaC function by increasing single channel current or open probability, experiments were performed with ␣␤S520C␥ hENaC and the sulfhydryl reagent MTSET. Substitution of serine by cysteine at the degenerin site near the extracellular end of the second transmembrane domain of ␤ hENaC, or at the equivalent site in rodent ENaC, generates a channel that can be covalently modified by MTSET; MTSET selectively reacts with cysteine 520 in the ␤ subunit of conducting ␣␤S520C␥ hENaC and locks channels in a constitutively open conformation (open probability 0.96 -1.0) (30,31). Accordingly, S3969 should not further activate ␣␤S520C␥ hENaC following MTSET treatment if S3969 functions to increase hENaC open probability; conversely, S3969 should activate ␣␤S520C␥ hENaC following MTSET treatment if S3969 functions to increase hENaC single channel current. This rationale has been used to determine the mecha-nism of activation of rat ENaC by the serum-and glucocorticoid-inducible kinase SGK1 (32).
To independently determine whether S3969 increases the open probability of wild-type hENaC channels, experiments were performed with the serine protease trypsin. Trypsin irreversibly increases macroscopic ENaC current by increasing the open probability of near electrically silent plasma membrane channels (18 -20). These channels are incorporated in the plasma membrane but reside predominantly in the closed state. Similar to MTSET treatment of oocytes expressing ␣␤S520C␥ hENaC, trypsin treatment of oocytes expressing wild-type ENaC stabilizes channels in an open conformation. Accordingly, S3969 should not activate hENaC following trypsin treatment if S3969 functions to increase hENaC open probability. As shown in Fig. 3C, trypsin treatment resulted in rapid activation of a transient inward current, formerly identified as the endogenous oocyte Ca 2ϩ -activated Cl Ϫ current (33), followed by slow activation of an amiloride-sensitive inward hENaC current, as described previously (19). Trypsin treatment increased macroscopic hENaC current more than 5-fold (Fig. 3D). S3969 failed to activate hENaC following trypsin application (Fig. 3E). Amiloride-sensitive currents and levels of hENaC activation were similar before and after cotreatment of oocytes with trypsin plus soybean trypsin inhibitor, indicating that trypsin effects were attributable to protease function. 3 These data indicate that locking hENaC channels in a constitu-  MAY  To validate that S3969 increases hENaC open probability, single channel patch clamp experiments were performed using the outside-out configuration in oocytes. Application of hENaC activator (30 M) to the bath, contacting the extracellular patch surface, increased open probability (Fig. 3, F and G) but did not affect single channel conductance. Open probability was 0.2 Ϯ 0.1 before and 0.6 Ϯ 0.1 after compound (p Ͻ 0.05), whereas single channel conductance was 4.8 Ϯ 0.3 pS before and 5.1 Ϯ 0.2 pS after compound (p Ͼ 0.05, n ϭ 3). Collectively, macroscopic and single channel recordings indicate that S3969 activates hENaC by increasing open probability and not by increasing single channel current or the number of channels at the plasma membrane.

Identification of a Human ENaC Opener
S3969 hENaC Activation Requires ␣, ␤, and ␥ Subunit Coexpression-Studies were conducted to determine the subunit requirements for hENaC activation by S3969. To this end, all possible combinations of ␣, ␤, and ␥ hENaC were expressed, and currents blocked by amiloride and activated by S3969 were measured. As shown in Fig. 4A and consistent with previous reports on human and rodent ENaC (34,35), ␣ hENaC, ␣␤ hENaC, or ␣␥ hENaC generated significantly smaller amiloride-sensitive currents compared with ␣␤␥ hENaC, whereas ␤ hENaC, ␥ hENaC, or ␤␥ hENaC failed to generate any detectable amiloride-sensitive currents. The only subunit combination activated by S3969 was ␣␤␥ hENaC, demonstrating that expression of all three channel components is required for the stimulatory effect of S3969 on hENaC function (Fig. 4B).
The ␤ Subunit Is Critical for S3969 Activation of hENaC-Differential activation of hENaC and mENaC by S3969 prompted subunit mixing experiments to determine which specific channel subunit(s) were important for the S3969 response. To this end, all possible heterotrimeric combinations of mENaC and hENaC subunits were coexpressed and both amiloride-sensitive currents and activation by S3969 were measured. Subunit mixing experiments have been performed to examine ENaC function from different species (10,12,34). All human-mouse hybrid channels generated amiloride-sensitive currents, indicating the production of functional channels at the plasma membrane (Fig. 6A). S3969 activated heterotrimeric hybrids containing ␤ hENaC but not ␤ mENaC (Fig. 6B). Most importantly, ␣ H ␤ M ␥ H channels (␣ and ␥ subunits from human and ␤ subunit from mouse) lost the ability to respond to S3969, whereas ␣ M ␤ H ␥ M channels (␣ and ␥ subunits from mouse and ␤ subunit from human) gained the ability to respond to S3969 (Fig. 6B). Collectively, these results demonstrate that the human ␤ subunit is critical for S3969 activation of hENaC.
S3969 Maps to the Extracellular Domain of ␤ hENaC-ENaC subunits have two transmembrane domains, intracellular N and C termini, and a large extracellular loop comprising ϳ75% of protein sequence. Alignment of human and mouse ␤ENaC protein sequences revealed that the two subunits exhibit 84% identity. Because most of the amino acid differences between human and mouse ␤ENaC localize to the large extracellular domain and because the effect of S3969 on hENaC function is rapidly reversible following compound washout (Fig. 1B), we hypothesized that S3969 required the extracellular domain of ␤ hENaC to activate channel function. To test this hypothesis, both amiloride-sensitive currents and S3969 activation of heterotrimeric ENaC channels, consisting of ␣ and ␥ mENaC subunits with ␤ subunit chimeras composed of intracellular, transmembrane, and extracellular domains derived from ␤ mENaC All mouse-human hybrids generated amiloride-sensitive currents, demonstrating the production of functional ENaC channels; no significant differences existed between amiloride-sensitive currents when comparing ␣␤␥ hENaC to hybrid channels. ENaC activation by S3969 was strictly dependent on expression of ␤ hENaC. S3969-induced currents were not observed in oocytes expressing hybrid channels containing ␤ mENaC. n ϭ 8 -13 oocytes/ group. *, p Ͻ 0.001 compared with ␣␤␥ hENaC. Amiloride (1 M)-sensitive currents (C) and % ENaC activation by S3969 (10 M) (D) in oocytes expressing heterotrimeric ENaC channels, consisting of ␣ and ␥ mENaC subunits with ␤ subunit chimeras composed of intracellular, transmembrane, and extracellular domains derived from ␤ mENaC (thin lines) or ␤ hENaC (thick lines), are as indicated. ⌬V348 indicates deletion of valine 348 (small gray circle). All ␤ subunit chimeras generated amiloride-sensitive currents, demonstrating the production of functional ENaC channels; no significant differences existed between amiloride-sensitive currents when comparing chimera H to other chimeras. ENaC activation by S3969 was strictly dependent on the extracellular loop of ␤ hENaC. S3969-induced currents were not observed in oocytes expressing ␤ subunit chimeras containing the mENaC extracellular loop. Similar results were obtained when ␤ subunit chimeras were expressed with ␣ and ␥ hENaC subunits. 3 n ϭ 10 -17 oocytes/group. *, p Ͻ 0.001 compared with ␤ chimera H. or ␤ hENaC, were measured. ENaC subunit chimera experiments have been performed to examine the functional contribution of individual domains to channel biology (11,13,36). All chimeras generated amiloride-sensitive currents, indicating the production of functional channels (Fig. 6C). S3969 activated chimeras containing the extracellular loop of ␤ hENaC (chimeras M-H and M-H-M) but not the extracellular loop of ␤ mENaC (chimeras H-M and H-M-H) (Fig. 6D). A ␤ hENaC splice variant lacking valine 348 in the extracellular loop, because of use of an alternative splice acceptor site, was identified in tongue tissue. Expression of hENaC lacking valine 348 in the ␤ extracellular loop generated amiloride-sensitive channels that were not activated by S3969 (Fig. 6, C and D). Therefore, S3969 opening of hENaC requires the ␤ hENaC extracellular loop, especially valine 348 within this region, and is not critically dependent upon the ␤ hENaC transmembrane and cytosolic domains.
S3969 Opens ENaC Channels Not Cleaved by Furin Proteolysis-Transport of sodium by ENaC requires proteolytic cleavage of ␣ and ␥ subunits in the extracellular loop distal to the first transmembrane domain by the serine endoprotease furin (41). Furin cleaves ENaC subunits, expressed in mammalian cells or oocytes, at the C-terminal side of conserved dibasic RXXR motifs (where R represents arginine and X represents any amino acid) (41,42). Thus, experiments were performed to determine whether S3969 activation of hENaC requires channel proteolysis by furin. To this end, ␣R178A/R181A/ R204A␤␥R138A hENaC, a channel in which all potential arginine residues targeted by furin proteolysis in ␣ and ␥ subunits were substituted with alanine, was generated. Homologous substitutions made in mENaC abolish furin proteolysis and dramatically decrease sodium transport in oocytes (41,43).
S3969 Activates PHA1 ENaC Channels-Loss of function hENaC mutations result in the salt-wasting genetic disease PHA1 (6,7). Clinical symptoms include hyponatremia, hyperkalemia, dehydration, elevated serum aldosterone, and mineralocorticoid unresponsiveness, and current treatments include sodium supplementation as well as dialysis to reduce elevated serum potassium (44). Experiments were performed to determine whether S3969 activates two PHA1 hENaC channels previously characterized in the Xenopus oocyte expression system. ␣␤G37S␥ hENaC has a glycine to serine substitution at amino acid 37 in the N-terminal intracellular domain and decreases open probability without affecting channel assembly or surface expression (6,7). ␣R508X␤␥ hENaC has an arginine to stop codon substitution at amino acid 508 in the extracellular loop before the second transmembrane domain and decreases expression at the plasma membrane (7,45,46). Both PHA1 variants and wild-type ENaC exhibit the same affinity for amiloride (6,46).

DISCUSSION
Small molecule ion channel modulators regulate the flux of cations and anions across biological membranes. Compounds that positively regulate ion channel function have been identified and termed activators for CFTR Cl Ϫ channels (47,48) and L-type Ca 2ϩ channels (49), openers for KCNQ (50,51) and maxi-K ϩ channels (52,53), potentiators/correctors for mutant ⌬F508 CFTR Cl Ϫ channels (54,55), or enhancers for ␥-aminobutyric acid, type A, Cl Ϫ channels (56) and cardiac Na ϩ channels (57). These modulators generally function to increase the time channels reside in an open conformation by stabilizing the open state, inhibiting inactivation, or increasing the frequency of channel opening events (49,57). Despite abundant literature identifying modulators for many ion channel families, potent and specific small molecule activators of ENaC have not been previously reported. This report describes the first compound, S3969, that fully and reversibly activates hENaC in heterologous cells. Mechanistically, S3969 increases the open probability of hENaC channels through interactions requiring the extracellular domain of the ␤ subunit. By increasing hENaC open probability, S3969 permits increased flux of Na ϩ ions across biological membranes into cells.
The maximal stimulatory effect of S3969 on hENaC function in oocytes required ϳ3 min (t1 ⁄ 2 ϳ 30 s), and currents returned to base line following S3969 washout. Slow, yet reversible, hENaC activation suggested that S3969 interacted noncovalently with hENaC channels in the plasma membrane. Activation of hENaC by S3969 exhibits qualitatively similar kinetics in mammalian cells and oocytes, suggesting that the time course of hENaC activation is not unique to the oocyte expression system but an intrinsic property of S3969 interaction with the hENaC channel. Amiloride priming experiments, performed to synchronize hENaC in an open conformation with an occluded pore, revealed that S3969 activation of hENaC was rapid (t1 ⁄ 2 ϳ 5 s) following amiloride removal. The majority of hENaC channels in the oocyte plasma membrane are closed or in a quiescent, nonconducting state (18,19,58), consistent with reports that ENaC exhibits long closed times on the order of seconds (27,59,60). Closed or quiescent channels may be poorly accessible to S3969; following transitions from the closed to the open state, S3969 may interact with hENaC to stabilize an open conformation, thereby increasing open prob- , or ␤␥ hENaC channels, or no hENaC channels (uninjected) are shown. S3969 strongly activated wild-type ␣␤␥ and ␣␤G37S␥ hENaC, weakly activated ␣R508X␤␥ hENaC, and failed to activate ␤␥ hENaC or uninjected oocytes. Oocytes were tested 48 -72 h post-injection to allow sufficient time for PHA1 hENaC channel functional expression (46). The lower activation of wild-type ␣␤␥ hENaC by 10 M S3969 compared with previous figures is likely because of a decreased driving force for sodium entry (increased intracellular sodium concentration) in oocytes expressing wild-type ␣␤␥ hENaC for 48 -72 h. n ϭ 10 -15 oocytes/group. *, p Ͻ 0.05 compared with wild-type ␣␤␥ hENaC.
ability. Additional, single channel patch clamp studies are required to further elucidate the mechanism of S3969 activation of hENaC. In addition, S3969 may function to relieve hENaC from sodium self-inhibition, similar to Zn 2ϩ ions, the mercurial agent p-chloromercuribenzoate, and trypsin-like as well as furin serine proteases (16 -21).
An alternative interpretation of the slow kinetics of hENaC activation is that S3969 interacts with an intracellular and/or transmembrane domain of hENaC; the time required to activate hENaC would simply reflect the necessity for S3969 to diffuse across the oocyte plasma membrane or into a hydrophobic cavity within the transmembrane domains. We argue against this contention based on the following points. First, positively charged analogs of S3969 exhibit similar potencies and kinetics as S3969, 3 which is uncharged at physiological pH. Charged derivatives should not readily diffuse into the oocyte cytoplasm and achieve relevant concentrations necessary to activate hENaC over the time course of our experiments. Second, activation of hENaC specifically required the extracellular domain of the ␤ subunit. Replacement of the extracellular domain of ␤ hENaC with the homologous region of ␤ mENaC did not support activation of chimeric channels by S3969; similarly, the intracellular and transmembrane domains of ␤ hENaC were not sufficient to recapitulate ENaC activation by S3969. Finally, activation of guinea pig ␣ and rat ␤␥ hybrid ENaC channels by CPT-cAMP was slow and required, in part, the extracellular domain of the ␣ subunit (12,13), indicating that slow activation of ENaC is consistent with compound interaction with an extracellular region.
Rapid activation of macroscopic hENaC current following amiloride washout, the failure to activate constitutively open hENaC (␣␤S520C␥ hENaC) following MTSET treatment, and the inability to activate wild-type hENaC following trypsin treatment suggest a mechanism whereby S3969 increases hENaC open probability without significantly affecting the number of functional channels incorporated into the plasma membrane or the single channel current of individual channels. Single channel patch clamp experiments validated that the hENaC activator increased open probability without affecting single channel current. Immediate S3969 activation of hENaC following amiloride priming is inconsistent with a mechanism whereby S3969 increases macroscopic hENaC current by stimulating trafficking of hENaC from a vesicular pool to the plasma membrane, a process requiring tens of minutes or more in the oocyte system (28,29).
Previous reports indicate that furin and trypsin-like serine proteases modulate the function of ENaC and ASIC channels in the ENaC/degenerin family (20,41,61,62). Similar to S3969, proteases increase macroscopic ENaC current by increasing ENaC open probability (18,19). However, unlike S3969, protease activation of ENaC is irreversible, in that ENaC currents do not return to base-line values after removal of protease because of covalent modification of ENaC channels (19,63). Because S3969 activation of hENaC is reversible and requires the ␤ subunit, whereas known proteases target the ␣ and ␥ subunits (41,(63)(64)(65), and because S3969 equally activates wild-type hENaC and hENaC lacking all known furin proteolysis sites, we consider it unlikely that S3969 activates hENaC indirectly through primary modulation of protease function. Reduced activation of hENaC by S3969 in CHO cells compared with HEK293 cells or oocytes may be attributable to hENaC proteolysis by furin in CHO cells resulting in channels with larger initial open probabilities (41).
Although activation of hENaC by S3969 required the extracellular domain of the ␤ subunit, S3969 did not activate channels composed of ␣␤ hENaC subunits, and S3969 only weakly activated ␣R508X␤␥ hENaC channels. Indeed, all three wildtype ␣␤␥ hENaC subunits were necessary to elicit a robust S3969 response. These data suggest that the binding site(s) for S3969 on hENaC is only formed upon coexpression of all three channel subunits. In addition to the ␤ extracellular loop, S3969 likely requires additional domains present in ␣ and ␥ subunits to stabilize open hENaC channels. These domains may be sufficiently conserved between human and mouse channels such that no differential functional effect was observed in subunit mixing experiments. In fact, weak activation of hENaC channels composed of ␣, ␣␤, or ␣␥ subunits was observed upon challenge with high concentrations of S3969 (ϳ50% activation at 100 M, which is slightly less than the activation observed for ␣␤␥ mENaC in Fig. 5B), 3 suggesting that nonheterotrimeric channels containing the pore-forming ␣ subunit are low affinity receptors for S3969. We identified valine 348 as a critical amino acid within the ␤ extracellular loop required for S3969 activation of hENaC. Valine 348 in ␤ hENaC is predicted to localize to ␤-sheet ␤9 in the palm domain of ASIC1a, a member of the ENaC/degenerin family that was recently crystallized (66). This region may include part of an S3969-binding site or be necessary for transmitting conformational changes between extracellular to transmembrane domains during hENaC gating. The degenerin site at the extracellular end of the second transmembrane domain of ␤ hENaC is likely not critical because substitution of serine 520 by cysteine did not affect activation of ␣␤S520C␥ hENaC by S3969.
Molecules previously reported to increase Na ϩ transport through ENaC channels in the Xenopus oocyte expression system suffer from low efficacy, low specificity, and/or lack of effect on wild-type ␣␤␥ hENaC (10 -15). In contrast, S3969 exhibits high efficacy (600 -700% hENaC activation at 30 M) as well as potency on wild-type hENaC (apparent EC 50 ϳ1 M for ␣␤␥ hENaC and ␦␤␥ hENaC but not ␣␤␥ mENaC). Further medicinal chemistry efforts have led to the discovery of S3969 derivatives with potencies ϳ100 nM. 4 Although detailed testing of S3969 on diverse ion channels was not performed, S3969 had no effect on endogenous oocyte channels, exogenous G-protein activated inwardly rectifying K ϩ channels (GIRK1 and GIRK2), and exogenous cold and menthol-activated nonselective cation channels (TRPM8). 3 Our findings provide indirect support for the hypothesis that ENaC may function as a ligand-gated ion channel, similar to other members of the ENaC/degenerin family (67). For example, ␦␤␥ ENaC and ASICs are activated by protons, and FaNaC is activated by the peptide FMRF-amide (1, 68 -70); in addition, ASICs are modulated by neuropeptides (71). It is conceivable that an endogenous peptide ligand exists that structurally resembles S3969 and modulates hENaC function in vivo. In support of this contention, S3969 has several peptide-like characteristics, including an indole ring, found in the amino acid tryptophan, as well as a peptide bond flanked by leucine-and serine-like side chains. Thus, S3969 may function as a peptidomimetic for an endogenous hENaC ligand.
Small molecule activation of ENaC may be applicable for blood pressure regulation, airway fluid reabsorption, rodent salt taste, and renal electrolyte homeostasis. Whereas inhibition of hENaC function with amiloride may decrease blood pressure in hypertensive individuals, including those expressing a T594M polymorphism in the ␣ subunit of hENaC (72), activation of hENaC may prove useful in increasing blood pressure in individuals with hypotension by increasing hENaC-dependent renal Na ϩ reabsorption. Similarly, activation of hENaC in the apical membrane of distal airway epithelia could promote lung function in neonatal respiratory distress syndrome or pulmonary edema by increasing hENaC-dependent Na ϩ and lung fluid absorption (73,74). Activation of ENaC in rodent tongue papillae could enhance salt sensation by promoting Na ϩ transport into taste bud cells (75,76). Finally, activation of residual hENaC function in individuals with PHA1 could correct dehydration and salt wasting accompanying mineralocorticoid resistance (44). Indeed, S3969 rescued ␣␤G37S␥ ENaC, a PHA1 variant with compromised channel gating (6), highlighting the potential to use hENaC channel openers in situations where improved Na ϩ flux across epithelial membranes is clinically desirable. In conclusion, promoting hENaCdependent Na ϩ and fluid transport across epithelial membranes may find utility in modulating extracellular volume and electrolyte homeostasis in both normal and diseased states.