AICAR activates AMPK and alters PIP₂ association with the epithelial sodium channel ENaC to inhibit Na⁺ transport in H441 lung epithelial cells

Abstract

Changes in amiloride-sensitive epithelial Na⁺ channel (ENaC) activity (NP_o) in the lung lead to pathologies associated with dysregulation of lung fluid balance. UTP activation of purinergic receptors and hydrolysis of PIP₂ via activation of phospholipase C (PLC) or AICAR activation of AMP-activated protein kinase (AMPK) inhibited amiloride-sensitive Na⁺ transport across human H441 epithelial cell monolayers. Neither treatment altered α, β or γ ENaC subunit abundance (N) in the apical membrane indicating that the mechanism of inhibition was via a change in channel open state probability (P_o). We found that UTP depleted PIP₂ abundance in the apical membrane whilst activation of AMPK prevented the binding of β and γ ENaC subunits to PIP_2. The association of PIP₂ with the ENaC subunits is required to maintain channel activity via P_o. Thus, these data show for the first time that AICAR activation of AMPK inhibits Na⁺ transport via a mechanism that perturbs the PIP₂–ENaC channel interaction to alter P_o. In addition, we show that dissociation of PIP₂ from ENaC together with activation of AMPK further reduced Na⁺ transport by a secondary effect that correlated with ENaC subunit internalization. Thus, when PIP₂–ENaC subunit interactions were compromised, ENaC protein retrieval was initiated, indicating that AMPK can modulate ENaC P_o and N.

Na⁺ is transported across the apical membrane of lung epithelial cells by the amiloride-sensitive epithelial Na⁺ channel, ENaC, down an electrochemical gradient generated by the Na⁺,K⁺-ATPase, which extrudes Na⁺ across the basolateral membrane and consumes ATP in the process (Matalon & O'Brodovich, 1999). Inhibition of Na⁺ transport in lung impairs fluid clearance in several species, including human (Berthiaume et al. 1987; O'Brodovich et al. 1990; Matalon et al. 1991; Jayr et al. 1994; Sakuma et al. 1995). Experimentally, inhibiting or increasing ENaC function results in fluid accumulation or dehydration, respectively (Hummler & Horisberger, 1999; Mall et al. 2004), and up-regulation of ENaC activity is associated with the airway pathology of cystic fibrosis (Boucher et al. 1988; Mall et al. 2004).

In the lung, constitutively active, non-voltage-dependent ENaC channels are predominantly formed from a heterotrimer of the pore-forming α subunit and accessory β and γ subunits (Canessa et al. 1994). Each subunit comprises cytosolic amino and carboxy terminals that are subject to numerous regulatory protein interactions and two transmembrane-spanning domains that are linked by a large extracellular loop (Garty & Palmer, 1997; Benos & Stanton, 1999; Fyfe et al. 1999). It has been proposed that channels composed of α, α–γ, α–β and α–β–γ combinations can also be formed (Firsov et al. 1998; Kosari et al. 1998; Snyder et al. 1998; Staruschenko et al. 2005) which may produce Na⁺ channels of differing characteristics. However, all three subunits are necessary to produce the low-conductance (5 pS), highly Na⁺-selective channel with an amiloride sensitivity of < 1 μm (Ma et al. 2004).

Apical insertion of the α subunit is rapidly increased in response to β-adrenergic agonists (Dumasius et al. 2001), oxygen (Ramminger et al. 2000), glucocorticoids (Tchepichev et al. 1995; Minakata et al. 1998), and thyroid hormones (Richard et al. 2004). Physiologically, up-regulation of ENaC is responsible for the transition of the fetal lung from net Cl⁻ secretion to net Na⁺ absorption at birth (Olver, 1986; Hummler et al. 1996) and it is involved in the clearance of pulmonary oedema fluid in the adult lung (Matalon & O'Brodovich, 1999). There is evidence from studies in polarized cortical collecting duct (CCD) epithelial cells to suggest that ENaC retrieval and recycling is controlled in part by ubiquitination by the E3–ubiquitin ligase, Nedd4-2 (Raikwar & Thomas, 2008) and de-ubiquitination by the ubiquitin carboxy-terminal hydrolase, UCH-L3 (Butterworth et al. 2007). ENaC activity is also increased by luminal proteases (Planes et al. 2005), phosphatidylinositol bisphosphate (Kunzelmann et al. 2005; Pochynyuk et al. 2007b) and casein kinase 2 (Bachhuber et al. 2008), and decreased by cellular energy sensing (Woollhead et al. 2005, 2007).

The ratio of intracellular nucleotides AMP : ATP are sensed by the AMP-activated protein kinase (AMPK) which acts to balance cellular energy by coordinating cellular energy-generating and -utilizing processes in the cell. We have previously shown that pharmacological activation of AMPK inhibits amiloride-sensitive transepithelial Na⁺ transport and amiloride-sensitive apical Na⁺ conductance in H441 lung epithelial cell monolayers (Woollhead et al. 2005, 2007; Bhalla et al. 2006; Woollhead & Baines, 2006). ENaC activity is a function of the number of channels in the membrane (N) × their open state probability (P_o). In our experiments, there was no change in membrane abundance (N) of α, β or γ ENaC subunits in response to a 1 h application of AICAR (5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside). In these experiments, AICAR activated AMPK and the inhibition was recovered by addition of the AMPK inhibitor Compound C indicating that AMPK regulates channel open state probability (P_o) not surface expression (N). Whilst mechanisms describing AMPK-mediated retreival of ENaC subunits have been described (Carattino et al. 2005; Bhalla et al. 2006), the mechanism by which AMPK reduces P_o is unknown.

A potential pathway for altering ENaC P_o involves interaction with phosphatidylinositol 4,5-bisphosphate (PIP₂). Apical extracellular nucleotides inhibit Na⁺ transport across absorptive epithelia via P2Y₂ purinergic receptors that are equally sensitive to ATP and UTP (Ramminger et al. 2000). Similar to that described in rat distal nephron epithelium, P2Y₂-induced activation of phospholipase C (PLC) was recently shown to inhibit ENaC channel activity via hydrolysis of PIP₂ without effect on surface expression (Kunzelmann et al. 2005; Tong & Stockand, 2005). The PIP₂–ENaC interaction appears to be direct since addition of exogenous PIP₂ to excised patches reversed the rapid run-down in ENaC activity in A6 distal nephron cells and mouse collecting duct (M1) cells (Ma et al. 2002; Yue et al. 2002; Kunzelmann et al. 2005). Sequence analysis has revealed a PIP₂ binding domain in the NH₃-terminal region of the β subunit of ENaC (Ma & Eaton, 2005). This led to the hypothesis that the carboxy terminus of ENaC may determine surface expression whilst the amino terminus regulates channel P_o.

In this manuscript we have tested the hypothesis that activation of AMPK alters the PIP₂–ENaC subunit interactions leading to inhibition of ENaC activity. This manuscript is a continuation of the previous finding that ENaC is inhibited by AICAR in H441 cells and we attempt to provide a mechanistic explanation to the former observation. We are studying events at the level of the cell membrane and we have used H441 cell monolayers as a test system in which to mechanistically deduce how AMPK interacts with epithelial Na⁺ channels to affect transepithelial Na⁺ transport and apical Na⁺ entry. We firstly investigated the effect of PIP₂ depletion on amiloride-sensitive Na⁺ transport in H441 cells using nucleotide uridine triphosphate (UTP) activation of P2Y₂ receptors. We then investigated the effect of activation of AMPK before and after depletion of PIP₂. Our results indicate that AMPK and PIP₂ depletion inhibit Na⁺ channels in H441 cells by independent pathways that do not affect subunit abundance. However, simultaneous activation of both pathways leads to a further inhibition of Na⁺ transport mediated, at least in part, via ENaC subunit retrieval from the cell membrane. We conclude that AICAR activates AMPK and alters PIP₂ association with ENaC subunits to inhibit Na⁺ transport in H441 lung epithelial cells.

Methods

Cell culture

The human H441 lung epithelial cell line was obtained from the American Type Culture Collection (ATCC, USA) and maintained in Gibco-1640 medium supplemented with fetal bovine serum (10%; Invitrogen, UK), l-glutamine (2 mm), sodium pyruvate (1 mm), insulin (5 μg ml⁻¹), transferrin (5 μg ml⁻¹), sodium selenite (7 ng ml⁻¹) and antibiotics (100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin). The cells were grown in 25 cm² flasks and kept in a humidified atmosphere with 5% CO₂ at 37°C.

Drug treatments

All agents were obtained from Sigma, UK. AICAR (1 h, 2 mm) was used to activate AMPK. The inhibitor of AMPK, Compound C (80 μm, 40 min) was used to associate AICAR with AMPK activation. To activate or inhibit PLC activity, we have used UTP (100 μm, 20 min) to stimulate purinergic receptor signalling pathways that induce PLC activity and result in PIP₂ hydrolysis, and U-73122 (10 μm, 20 min) which is an inhibitor of PLC, as well as its inactive analogue, U-73343 (10 μm, 20 min). The aminoglycoside neomycin (5 mm, 3 h), was used to sequester PIP₂ and wortmannin (10 μm, 3 h) to block PIP₂ synthesis.

Ussing chamber experiments

Details of the measurement of short circuit current (I_SC) have been extensively described in Woollhead et al. (2007). Briefly, confluent non-polarized H441 cells were seeded on to permeable supports (Costar Snapwells) and cultured overnight. The following day, the serum was replaced with 4% charcoal stripped serum (CSS) containing thyroxine (T₃; 10 nm) and dexamethasone (200 nm) to polarize the monolayer. Resistive monolayers cultured at air interface for 6–7 days were used in Ussing chamber experiments.

Monolayers were mounted into an Ussing chamber in a physiological salt solution (PSS) containing (mm): NaCl 117, NaHCO₃ 25, KCl 4.7, MgSO₄ 1.2, KH₂PO₄ 1.2, CaCl₂ 2 and d-glucose 11 (pH 7.4). Experiments were performed under open circuit conditions. Once values for transepithelial voltages (V_t) and transepithelial resistance (TER) reached stability, short-circuit current (I_SC) was recorded by clamping V_t at 0 mV using a DVC 4000 voltage/current clamp and recorded via a PowerLab computer interface. The resistances were corrected for the resistance of the supports in the absence of H441 cell monolayers. Addition of 10 μm amiloride to the apical bath was used to determine amiloride-sensitive I_SC (I_SC-Amiloride). Apical amiloride-sensitive Na⁺ conductance was measured as previously described (Collett et al. 2002; Ramminger et al. 2004). The PSS was replaced with potassium gluconate solution consisting of (mm): potassium gluconate 121.7, KHCO₃ 25, MgSO₄ 1.2, KH₂PO₄ 1.2, calcium gluconate 11.5, d-glucose 11 (pH 7.4). A final dilution of PSS : potassium gluconate solution (8.1 : 91.9) and a final Na⁺ concentration of ∼11.5 mm. Na⁺,K⁺-ATPase was then inhibited with ouabain (1 mm) and the basolateral membrane permeabilized with nystatin (75 μm). The concentration of Na⁺ in the apical bath was raised to 55 mm by a sodium gluconate solution (mm): sodium gluconate 117, NaHCO₃ 25, potassium gluconate 4.7, MgSO₄ 1.2, KH₂PO₄ 1.2, calcium gluconate 2.5, d-glucose 11 (pH 7.3–7.4) (91.9 : 8.1 with PSS) creating a gradient for Na⁺ influx across the apical membrane. Amiloride-sensitive apical Na⁺ conductance (GNa⁺) was calculated from the amiloride (10 μm)-inhibitable apical current using G_Na=I_SC-Amiloride/V_Na (Woollhead et al. 2005).

Protein preparation

Monolayers were treated as desired. Proteins were prepared with extensive precautions to maintain the integrity and localization of proteins, including protection of phosphorylation status (Mace et al. 2007a,b). The trafficking of cytosolic and membrane proteins was arrested by washing monolayers with ice-cold PBS and all subsequent steps were performed at 4°C. Total protein was harvested into the following buffer (mm): Tris 50 (pH 7.4), NaCl 150, NaF 50, sodium pyrophosphate 5, EDTA 1, EGTA 1, DTT 1, 1% v/v Triton X-100 and 1% protease inhibitor cocktail (Sigma). The phosphorylation status of the proteins was maintained by addition of (mm): EDTA 10, activated sodium orthovanadate 200, NaF 20, EGTA 10 and EDTA 5 to the lysis buffer. This was centrifuged (5 min, 250 g) to remove nuclei. To isolate the plasma membranes and the cytosolic fraction, monolayers were lysed and the membranes separated from the cytosolic fraction by centrifugation (60 000 g) at 4°C for 30 min as previously described (Brot-Laroche et al. 1988). Membrane and cytosolic fractions were suspended in lysis buffer. Plasma membranes were isolated and purified by modification of the method of Woollhead et al. (2007). Apical or basolateral membranes were exposed to a 10 m excess of Sulfo-NHS-biotin (50 mg ml⁻¹). We have previously found that increasing the amount of biotin does change the amount of protein recovered and is optimal at 50 mg ml⁻¹ and does not increase further with additional biotin. Subsequently, the cells were lysed and proteins were solubilized. Solubilized proteins were incubated with excess streptavidin agarose beads overnight and biotinylated proteins bound to beads were separated from non-biotinylated proteins by centrifugation. Our elution is optimized for recovery. The samples were then prepared for Western blotting and stored at –80°C until required.

Immunoprecipitation

Following the desired treatment, cell lysates were harvested in the lysis buffer described above and plasma membranes were prepared. These were incubated with the protease and phosphatase inhibitor cocktail as described. Plasma membrane proteins were immunoprecipitated with an anti-PIP₂ antibody (Cambridge Biosciences, UK) for 30 min using the Catch and Release v2.0 Reversible Immunoprecipitation System (Millipore, UK). The samples were then prepared for Western blotting and stored at –80°C until required.

Western blotting

Total, cytosolic and biotinylated membrane proteins (50 μg) were separated on 4–12% Bis-Tris NuPAGE (Invitrogen, UK), transferred on to PVDF and incubated with anti-PIP₂ (Cambridge Biosciences, UK; g15–062; 1 : 200 dilution), anti-β-actin (Abcam, UK; ab8229; 1 : 500 dilution), anti-α1-Na⁺,K⁺-ATPase (Developmental Studies Hybridoma Bank, University of Iowa, USA; 1 : 1000 dilution), and C-terminal antibodies raised to α ENaC (Santa Cruz; sc22239; 1 : 500) and β ENaC (Santa Cruz; sc22242; 1 : 500) or γ ENaC (Alpha Diagnostics; ENACg31-A; 1 : 200 dilution). Cytosolic proteins were immunoblotted with total and phospho-specific antibodies for AMPK and its downstream substrate, acetyl-CoA carboxylase (ACC) to measure AMPK activity. Total protein antibodies were anti-AMPKα (Cell Signalling Technology; no. 2532; 1 : 500 dilution) and anti-ACC (Cell Signalling Technology; no. 2531; 1 : 500 dilution) and phospho-specific species were phospho-AMPKα (Thr172) (Cell Signalling Technology; no. 3662; 1 : 500 dilution) and phospho-ACC (Ser79) (Cell Signalling Technology; no. 3661; 1 : 500 dilution). Immunoprecipitates were resolved by 4–12% Bis-Tris NuPAGE, transferred on to PVDF (Amersham) and incubated with rabbit anti-α-, β- and γ-ENaC antibodies. Following primary antibody incubation, proteins were visualized using either biotinylated anti-rabbit IgG (1 : 1000) followed by a streptavidin HRP conjugate (1 : 1000) or an anti-mouse IgG HRP (1 : 1000) conjugated secondary antibody as appropriate. Proteins were visualized using an ECL Advance Detection Kit (Amersham Pharmacia Biotech). The Western blot analysis system (NEN Life Science Products, USA; Western lightning, PerkinElmer, USA) was used to quantify band intensity. Specificity of antibodies was confirmed by pre-incubation of antibody with excess antigenic peptides.

Statistical analysis

Data are presented as mean ±s.e.m. and statistical significance was determined using paired or unpaired Students t test as appropriate. The AMPK activity was calculated as the ratio of the total protein to phosphorylated species as measured by densitometry for both AMPK and ACC.

Results

UTP but not AICAR depletes PIP₂ in H441 cell monolayers

Treatment of monolayers with UTP caused a 93% decrease in PIP₂ abundance determined by Western blotting from control of 1.0 ± 0.05 to 0.07 ± 0.05 (P < 0.001, n = 3) (Fig. 1). The abundance of PIP₂ recovered following 60 min of UTP washout, the time scale of which is in keeping with receptor-induced phosphoinositide hydrolysis and resynthesis (Ford et al. 2003) (n = 4, Fig. 3B). Treatment of cells with U-73122 (an inhibitor of PLC) caused a 44% increase in PIP₂ abundance to 1.44 ± 0.12 (P < 0.05, n = 1) whereas the inactive analogue, U-73343, did not produce any significant change (Fig. 1). These data indicate that under normal conditions, there is a basal PLC-mediated hydrolysis of PIP₂ in these cells. Neomycin sequesters PIP₂ with high affinity (Gabev et al. 1989) and eliminates the possibility that diacylglycerol or intracellular Ca²⁺ signalling may inhibit ENaC activity. Although neomycin (5 mm) is relatively impermeable to plasma membranes, a 3 h treatment did not significantly affect abundance of PIP₂ in the membrane which remained at 1.05 ± 0.12 (P = 0.16, n = 3; Fig. 1). Previous investigators have used a similar approach to show that PIP₂ could regulate the Na⁺–H⁺ exchanger, NHE1 (Aharonovitz et al. 2000). AICAR also had no effect on PIP₂ abundance in the membrane (P = 0.23, n = 3; Fig. 1).

Figure 1. Membrane PIP₂ levels in treated monolayers.

H441 monolayers were treated as control (vehicle), or with AICAR, neomycin, U-73343, UTP, UTP + AICAR or U-73122. The cells were kept on ice, lysed and protein solubilized. A, 20 μg of cell lysate was separated by 4–12% Bis-Tris NuPAGE, transferred on to PVDF membrane and immunoblotted with anti-PIP₂ antibody. B, band signal was determined by densitometry from three separate experiments performed in duplicate from 3 Western blots and the data were normalized to control (as 1.0). Statistical significance was determined by Student's t test versus control where **P < 0.01 and ***P < 0.001.

Figure 3. Transepithelial amiloride-sensitive Na⁺ transport after UTP inhibition parallels PIP₂ levels.

A, summary of experiments showing the recovery of transepithelial amiloride-sensitive Na⁺ transport (I_SC-Amiloride) following application of UTP to H441 cell monolayers. Monolayers were treated with vehicle (control) or UTP which was washed out for 20, 40, 60 and 80 min. B, cell lysates were prepared from monolayers that were exposed to UTP, and washed out for 20, 40 and 60 min. 20 μg of cell lysate was separated by 4–12% Bis-Tris NuPAGE, transferred on to PVDF membrane and immunoblotted with anti-PIP₂ and anti-β actin. Band signal was quantified by densitometry and expressed relative to β-actin. The data were normalized to the 60 min washout. C, cell lysates were prepared from monolayers that were treated with vehicle (control), UTP, UTP and washout (recovery) for 60 min, and UTP and washout (recovery) for 60 min in the presence of wortmannin. 20 μg of cell lysate was separated by 4–12% Bis-Tris NuPAGE, transferred on to PVDF membrane and immunoblotted with anti-PIP₂ and anti-β actin. Band signal was quantified by densitometry and expressed relative to β-actin. The data were normalized to the vehicle-treated monolayers. Statistical significance was determined by Student's t test where **P < 0.01 and ***P < 0.001, n = 3.

AICAR and UTP inhibit transepithelial amiloride-sensitive Na⁺ transport in H441 cells

After mounting monolayers in Ussing chambers, the spontaneous I_SC generated during control experiments was 39.4 ± 4.9 μA cm⁻² (n = 8), transepithelial voltage (V_t) was –5.7 ± 0.4 mV (n = 8) and transepithelial resistance (TER) was 147 ± 6.8 Ωcm²(n = 8). The amiloride-sensitive I_SC (I_SC-Amiloride) was 38.0 ± 4.5 μA cm⁻² (n = 8).

A time course for the inhibition of I_SC-Amiloride by UTP application was performed and correlated with the level of PIP₂ (Fig. 2). The level of PIP₂ was significantly depleted after 5 min of UTP stimulation that was concordant with a fall in the I_SC-Amiloride by ∼45% (P < 0.01, n = 3, Fig. 2). Total PIP₂ depletion and maximal I_SC-Amiloride inhibition was obtained following 20 min of UTP stimulation. Purinergic receptor stimulation by luminal UTP (20 min) resulted in a 44% reduction in spontaneous I_SC to 20.6 ± 2.8 μA cm⁻² (P < 0.01, n = 8 compared with that of control monolayers, Fig. 4A). Transepithelial I_SC-Amiloride was also diminished by 49% to 19.5 ± 2.7 μA cm⁻² (P < 0.01, n = 8, Fig. 4B). This was not the result of changes in the TER of treated monolayers which was unchanged at 132 ± 9.4 Ωcm²(n = 8). Correlation of the I_SC-Amiloride to the level of PIP₂ was high and the correlation coefficient (R²) was determined to be 0.9939 (Fig. 2D). I_SC-Amiloride was not totally inhibited by the depletion of PIP₂ by UTP; following depletion, a component of I_SC-Amiloride remained showing that only a proportion of the I_SC-Amiloride was dependent on PIP₂ levels. Approximately 50% of the amiloride-sensitive component was inhibited by UTP and depletion of PIP₂.

Figure 2. Time course depicting the relationship between I_SC-Amiloride and PIP₂ levels.

A, experiments showing the inhibition of transepithelial amiloride-sensitive Na⁺ transport (I_SC-Amiloride) following application of UTP to H441 cell monolayers for 30 min at 5 min intervals. Monolayers were treated with UTP and amiloride-sensitive Na⁺ transport measured by application of amiloride (10 μm). B, cell lysates were prepared from monolayers that were characterized functionally as above. 20 μg of cell lysate was separated by 4–12% Bis-Tris NuPAGE, transferred on to PVDF membrane and immunoblotted with anti-PIP₂ and anti-β actin. C, the band signal was quantified by densitometry and expressed relative to the control as 1.0. D, correlation of I_SC-Amiloride (shown in A) and relative level of PIP₂ measured by densitometry and normalized to control (as 1.0; shown in C). The correlation coefficient was calculated as R²= 0.9939. Statistical significance was determined by Student's t test where **P < 0.01 and ***P < 0.001, n = 3.

Figure 4. AICAR and PIP₂ individually modulate amiloride-sensitive transepithelial and apical Na⁺ currents.

Monolayers were treated with vehicle (control), UTP, AICAR, neomycin, U-73122 and U-73343. A, spontaneous I_SC was measured at the start of each experiment. B, amiloride-sensitive I_SC (I_SC-Amiloride) was calculated. C, apical Na⁺ conductance was calculated as described in experimental procedures. D–F, continuous I_SC traces from monolayers treated with vehicle (control), UTP or U-73122. Amiloride (10 μm) was added to the apical reservoir to measure I_SC-Amiloride. The application of each drug is shown. Statistical significance was determined using Student's t test versus control where *P < 0.05 and ***P < 0.01, n = 6.

Following 60 min of UTP washout, transepithelial I_SC-Amiloride recovered to 35 ± 2.1 μA cm⁻² which was 92% of the control value (P = 0.34, n = 8, Fig. 5A). Inhibition of PIP₂ re-synthesis by pre-treatment with wortmannin (a PI3 and PI4 kinase inhibitor) prevented recovery of PIP₂ as determined from Western blot (Fig. 3C) and prevented recovery of transepithelial I_SC-Amiloride following 60 min of UTP washout (Figs 3A and 5A). I_SC-Amiloride remained at 15.8 ± 3.7 μA cm⁻² (P < 0.05, n = 8, Figs 3A and 5A). These data indicate that PIP₂ is required for channel activity.

Figure 5. AICAR and PIP₂ effects on transepithelial and apical amiloride-sensitive Na⁺ currents are additive.

A, monolayers were treated with vehicle (control), UTP and AICAR following 60 min washout in the presence and absence of wortmannin, and amiloride was added to the apical reservoir to determine amiloride-sensitive I_SC. B, monolayers were treated with vehicle (control) or with AICAR in the presence and absence of neomycin, U-73122 or U-73343, and amiloride was added to the apical reservoir to determine amiloride-sensitive I_SC. C, apical Na⁺ conductance (GNa⁺) was determined as described in the experimental procedures. Statistical significance was determined using Student's t test versus control where *P < 0.05 and **P < 0.01, n > 6.

In concordance with our previous data, treatment of monolayers with AICAR (2 mm, 1 h) diminished spontaneous I_SC by 46.2% of the control to 19.6 ± 1.7 μA cm⁻² (P < 0.01, n = 8, Fig. 4A). Transepithelial I_SC-Amiloride was reduced to 19.0 ± 1.4 μA cm⁻², a value ∼50% of the control (P < 0.01, n = 8, Fig. 4B). TER of the treated monolayers was unchanged compared to control monolayers at 149 ± 8.7 Ωcm² (P = 0.37, n = 8).

UTP and AICAR inhibit apical amiloride-sensitive Na⁺ conductance

We examined how activation of AMPK and depletion of PIP₂ modified amiloride-sensitive apical Na⁺ conductance (GNa⁺). Addition of AICAR to monolayers suppressed apical GNa⁺ by 49.4% from a control value of 310 ± 33 to 157 ± 15 μS cm⁻² (P < 0.01, n = 5, Fig. 4C). Application of UTP or neomycin also significantly diminished the amiloride-sensitive apical GNa⁺ of monolayers to 137 ± 17 μS cm⁻² and 133 ± 19 μS cm⁻², respectively (both P < 0.01, n = 5, Fig. 4C). The PLC inhibitor, U-73122, increased apical GNa⁺ by 30% to 394 ± 39 μS cm⁻² (P < 0.05, n = 5, Fig. 4C) compared to control, whilst its inactive analogue failed to change GNa⁺.

AICAR and UTP have an additive effect on I_SC-Amiloride and GNa⁺

We explored whether the PIP₂ interaction was necessary for AMPK-mediated inhibition of I_SC-Amiloride. PIP₂ abundance in H441 cell monolayers was modified with UTP followed by a 60 min washout period in the presence or absence of wortmannin prior to the application of AICAR. In the absence of wortmannin, AICAR inhibited the recovered I_SC-Amiloride by 53.5% from 35.3 ± 3.2 to 16.4 ± 2.5 μS cm⁻² (P < 0.01, n = 8; Fig. 5A). In the presence of wortmannin, the I_SC-Amiloride did not recover after 60 min of UTP washout and the addition of AICAR inhibited the I_SC-Amiloride by a further 52.3% from the non-recovered value of 15.8 ± 3.7 to 7.6 ± 1.2 μA cm⁻² (P < 0.01, n = 8). In comparison to the untreated control value, this represented a reduction of 79% (Fig. 5A). Sequestering PIP₂ with neomycin followed by addition of AICAR also diminished transepithelial I_SC-Amiloride a further 57.6% from 20.6 ± 2.7 to 8.7 ± 2.5 μA cm⁻² (P < 0.01, n = 8); a 76.4% reduction from the control value (Fig. 5B). Addition of U-73122 augmented transepithelial I_SC-Amiloride, consistent with its ability to elevate plasma membrane PIP₂ levels (Fig. 1). The addition of AICAR to monolayers pre-treated with U-73122 caused a 49.3% reduction in I_SC-Amiloride from 46.5 ± 3.8 to 23.6 ± 2.8 μA cm⁻² (P < 0.01, n = 8; Fig. 5B). Pre-treatment with the inactive analogue, U-73343, did not augment I_SC-Amiloride and addition of AICAR inhibited the I_SC-Amiloride by 49.6% from 34.7 ± 2.8 to 17.5 ± 1.1 μA cm⁻² (P < 0.01, n = 8; Fig. 5B). PIP₂ levels did not change when PIP₂ was sequestered with neomycin, presumably because PLC is not able to reach its substrate. Supporting this observation, no changes to the I_SC-Amiloride were observed when UTP was applied in the presence of neomycin.

AICAR and UTP had a similar additive effect on apical GNa⁺. Treatment with UTP or neomycin in the presence of AICAR further decreased apical conductance by 55 and 52%, respectively (Fig. 5C). The AICAR-treated control value for GNa⁺ was 157 ± 15 μS cm⁻² and was reduced to 72 ± 7 and 75 ± 9 μS cm⁻² (both P < 0.01, n = 5, Fig. 5C) by UTP and neomycin, respectively. Increasing PIP₂ by using U-73122 enhanced apical Na⁺ conductance by 17.8% above the control value to 195 ± 10 μS cm⁻² (P < 0.05, n = 6, Fig. 5C). Taken together, these data indicate that UTP and AICAR inhibit apical amiloride-sensitive Na⁺ channels via independent but additive pathways.

AICAR but not modulators of PIP₂ activate AMPK

Treatment of monolayers with AICAR significantly increased the ratio of phosphorylated AMPK to total AMPK from a control value of 0.20 ± 0.05 to 1.1 ± 0.1 (P < 0.001, n = 3, Fig. 6). That AICAR activated AMPK phosphorylation activity was supported by the fact that the phospho-ACC to total-ACC ratio was also enhanced 4-fold from 0.25 ± 0.1 to 1.1 ± 0.2 (P < 0.001, n = 3, Fig. 6). The ratio of phospho-AMPK to total-AMPK was not altered by UTP, neomycin, U-73122 or U-73343 which remained at 0.3 ± 0.05, 0.09 ± 0.04, 0.11 ± 0.06 and 0.13 ± 0.05, respectively (all n = 3, Fig. 6). The ratio of phospho-ACC to total-ACC also remained unchanged by UTP, neomycin, U-73122 or U-73343 at 0.4 ± 0.1, 0.18 ± 0.2, 0.20 ± 0.2 and 0.21 ± 0.2, respectively (all n = 3, Fig. 6). The addition of UTP in the presence of AICAR did not activate AMPK further than AICAR alone; the ratio of phospho- to total AMPK was 0.9 ± 0.1 (P = 0.13, n = 3), and the ratio of phospho- to total-ACC was 1.0 ± 0.2 (P = 0.21, n = 3, Fig. 6).

Figure 6. AMPK activity after treatment with AICAR and PIP₂ modifying agents.

H441 monolayers were treated with vehicle (control), AICAR, UTP, neomycin, U-73122, U-73343 or UTP + AICAR and neomycin + AICAR. Monolayers were assessed for functional AMPK activity. Following treatment, the cells were kept on ice, lysed and protein solubilized. A, 20 μg of protein was separated by 4–12% Bis-Tris NuPAGE, transferred on to PVDF membrane and immunoblotted with anti-αAMPK and anti-ACC antibody. Subsequently, the blots were stripped and re-probed with anti-phospho αAMPK (Thr172) antibody and anti-phospho ACC (Ser79) antibody. B, densitometry was used to quantify band signal and the phospho-specific to total protein ratio was used to express the data. Statistical significance was determined by Student's t test versus control and ***P < 0.001, n = 6.

The effect of AICAR on I_SC-Amiloride but not modulators of PIP₂ is reversed by the AMPK inhibitor Compound C

Transepithelial I_SC-Amiloride was measured in the presence and absence of Compound C and the data were normalized to the control value (100%). Pre-treatment with Compound C had no significant effect on TER (132 ± 16 Ωcm², n = 8), spontaneous I_SC (37.1 ± 5.8 μA cm⁻², n = 8) or on the transepithelial I_SC-Amiloride (36.4 ± 6.2 μA cm⁻², n = 8, Fig. 7). AICAR diminished transepithelial I_SC-Amiloride by 50.1 ± 8.2% and this was recovered by Compound C to 102.3 ± 10.3% of the control value (P < 0.05; n = 8, Fig. 7). Following pre-treatment with AICAR, UTP further reduced I_SC-Amiloride to 71.4% with respect to vehicle-treated control monolayers (Fig. 7) and Compound C rescued transepithelial I_SC-Amiloride to reach 53.2% of the control, an inhibition similar to that of UTP alone (Fig. 7). Similarly, Compound C reversed the additive effect of neomycin and AICAR to that of neomycin alone and reversed the effect of AICAR in the presence of U-73122 whilst its inactive analogue was concordant with the untreated control. PIP₂ levels were measured in the presence of Compound C but no changes were observed.

Figure 7. Effect of Compound C on I_SC-Amiloride of monolayers treated with AICAR and PIP₂ modifying agents.

Monolayers were pre-incubated with Compound C (40 min, 80 μm) or vehicle (control) prior to the application of AICAR. Subsequently, these monolayers were treated with either UTP, neomycin, U-73122 or U-73343 in the presence and absence of AICAR. I_SC-Amiloride was calculated following apical application of 10 mm amiloride. Values are expressed relative to the control (as 100%). Statistical significance was determined using the Student's t test versus control (–Compound C) where *P < 0.05, *P < 0.01, and versus–Compound C where ^$P < 0.05, n > 6.

AMPK interferes with the interaction of ENaC subunits with PIP₂

The ability of PIP₂ to bind the ENaC subunits in the presence of AICAR and UTP was examined by immunoprecipitation using an anti-PIP₂ antibody and Western blotting using subunit-specific antibodies. Immunoprecipitation of α(95 kDa and 65 kDa), β (100 kDa) and γ (90 kDa) subunits were observed in untreated cells using the PIP₂ antibody. In immunoprecipitates prepared from monolayers treated with AICAR, both the 95 kDa and 65 kDa proteins of α ENaC were clearly detected (Fig. 8A). In contrast, PIP₂ did not immunoprecipitate β and γ ENaC proteins (Fig. 8A). In monolayers treated with UTP (which depletes PIP₂ in the membrane) the 65 kDa α ENaC protein was barely detectable and neither β or γ ENaC proteins were observed. These data indicate that PIP₂ binds all three ENaC subunits and that AICAR interrupted the binding of PIP₂ to the β and γ subunits.

Figure 8. The effect of PIP₂ and AMPK on PIP₂–ENaC interactions and surface-expressed ENaC subunits.

A, coimmunoprecipitation of α, β and γ ENaC subunits with PIP₂. Cell lysates were immunoprecipitated with an anti-PIP₂ antibody and the immunoprecipitates resolved on 4–12% NuPAGE gels, transferred on to PVDF and immunoblotted with subunit-specific ENaC antibodies. B, Western blot analysis of total (T) protein and non-bound (NB) fraction (20 μg), or biotinylated apical (A) and basolateral (B) proteins isolated from 1 mg of total protein extracted from polarized H441 cell monolayers. Immunostained proteins corresponding to the α1 subunit of the Na⁺,K⁺-ATPase (∼113-kDa) was present predominantly in the total protein and basolateral fractions, β-actin (42 kDa) was also present in the total and non-bound protein fractions. The 65 kDa mature α ENaC is shown in the total and biotinylated apical protein fractions (for simplicity, the immature 95 kDa α ENaC subunit band was not included). The numbers shown in brackets in B correspond to the total number of biotinylation experiments for which the blots are representative. C, H441 cell monolayers were treated with either vehicle (control), UTP, AICAR, neomycin, UTP + AICAR, neomycin + AICAR, U-73122, U-73122 + AICAR or UTP + neomycin + AICAR. Pure biotinylated apical fractions were prepared and proteins resolved on 4–12% Bis-Tris NuPAGE gels, transblotted on to PVDF and immunostained with subunit-specific ENaC antibodies. D, band signal was quantified by densitometry and normalized to the control (as 1.0). Statistical significance was determined using Student's t test versus control where **P < 0.01 and ***P < 0.001, n = 5.

The recycling of ENaC subunits by AICAR is dependent on PIP₂

The effect of AICAR and UTP on the surface expression of the ENaC subunits was explored to determine whether inhibition of transepithelial I_SC-Amiloride and apical GNa⁺ correlated with membrane abundance of ENaC subunits. The 65 kDa α ENaC protein was detected in the apical biotinylated fraction and the total protein fraction consistent with its localization to the apical membrane (Fig. 8B). The α ENaC subunit protein was not detected in the biotinylated basolateral fraction showing that there was no contamination of the basolateral protein fraction by apical membrane proteins.

In the majority (40/45) of our biotinylation experiments, the α1 subunit of the Na⁺,K⁺-ATPase was predominantly detected in the basolateral membrane and total protein fraction consistent with its basolateral distribution in polarized cells. However, in five of our biotinylation experiments, a small amount of Na⁺,K⁺-ATPase was observed in the biotinylated apical protein fraction. Quantification of immunoblots indicated < 5% ATPase in the apical fraction (4.7 ± 4.4%; n = 45). There was no significant change to TER in any of the experiments performed so we are confident that tight junction permeability to biotin was not compromised by treatment. Neither were there any significant changes to the protein concentration of the biotinylated apical fractions as determined by the Bradford assay (5.2 ± 1.4 mg ml⁻¹; n = 45) implying that apical biotin did not have significant access to further basolateral proteins. This was not the result of biotin saturation since biotin was in excess of protein. The cytosolic marker protein, β-actin, was only detected in the total and non-bound protein fractions and was absent in the biotinylated protein fractions, showing that they were not contaminated with cytosolic proteins (Fig. 8B).

This led us to conclude that apical biotin only significantly biotinylated apical proteins. The apical membrane ENaC and total ENaC protein in these cells was calculated by determining the mean band intensities of the biotinylated apical signal to the total signal from the biotinylation experiments. Comparing the non-biotinylated α ENaC in the supernatant to the protein bound to the streptavidin beads showed that the proportion of α ENaC exposed at the apical membrane represented ∼70% of the total α ENaC protein in the cells. This estimate assumes complete biotinylation of apically expressed α ENaC and full recovery of the biotinylated protein from the beads. The ratio of apical ENaC to total ENaC was 0.71 ± 0.06 (Fig. 8B) showing a large proportion of ENaC is concentrated within the plasma membrane, which is consistent with the fact that the cell culture conditions used in our laboratory promote apical ENaC protein expression and activity.

The 95 kDa and 65 kDa bands of the α ENaC subunit were detected together with the 100 kDa β and 90 kDa γ ENaC subunit bands in the biotinylated apical membrane proteins from control treated monolayers (Fig. 8C and D). Consistent with our previous observations, the abundance of all three ENaC subunit proteins did not change significantly following treatment with AICAR. Neither were they significantly altered by PIP₂ modification via UTP, neomycin, U-73122 or the inactive analogue U-73343 (Fig. 8C and D). These data support the notion that inhibition of I_SC-Amiloride and apical GNa⁺ by AICAR and UTP as well as the augmentation of I_SC-Amiloride and apical GNa⁺ by U-73122 correlated with changes in P_o. Dual treatment with AICAR and UTP diminished α and β ENaC subunit abundance by 65 ± 8% and 61 ± 8%, respectively (both P < 0.001, n = 3; Fig. 8C and D). Treatment with AICAR plus neomycin also caused a similar diminution in α and β subunit abundance; the 65 kDa α subunit was reduced by 71 ± 8% and the β subunit diminished by 79 ± 8% (both P < 0.001, n = 3; Fig. 8C and D). The α subunit was reduced by 72.1 ± 6.7% and the β subunit was reduced by 73.4 ± 7.1% when monolayers were treated with AICAR, neomycin and UTP (both P < 0.001, n = 3; Fig. 8C and D). Neither enhancement of PIP₂ by U-73122 or inhibition by AICAR in the presence of U-73122 altered subunit expression. Despite coordinate changes to α and β ENaC subunits, we did not observe a change in the membrane expression of the γ subunit throughout these experiments (Fig. 8C and D).

Discussion

ENaC activity is dependent on PIP₂

PIP₂ is a minority component of the inner leaflet of the plasma membrane. It is the substrate for PLC cleavage and produces the classical secondary messengers, soluble inositol triphosphate (IP₃) and membrane-bound diacylglycerol (DAG); these function to mobilize intracellular Ca²⁺ and activate the trafficking of protein kinase C (PKC) from the cytosol to the plasma membrane, respectively. Classically, PIP₂ itself is recognized to act as an anchor for proteins that initiate endo- and exoctyosis, GTPase proteins and cytoskeletal proteins. However, more recently PIP₂ has been demonstrated to directly activate multiple ion channels and transporters including the inwardly rectifying K⁺ (Kir) channels (Hilgemann & Ball, 1996), KCNQ channels (Delmas & Brown, 2005), the transient receptor potential (TRP) channels (Rohacs, 2007) and the ion transporters such as the Na⁺–Ca²⁺ exchanger (Suh & Hille, 2005).

We investigated whether AICAR activation of AMPK inhibited amiloride-sensitive Na⁺ transport via a mechanism that modulated PIP₂ in the H441 cell membrane. We show for the first time that abundance of PIP₂ was depleted by activation of P2Y₂ receptors with UTP which hydrolyses PIP₂ via PLC. UTP was chosen in preference to ATP to avoid adrenergic receptor stimulation of Na⁺ currents (Chambers et al. 2006). Furthermore, we were able to inhibit PIP₂ re-synthesis with the PI3 and PI4 K inhibitor wortmannin. Activation of PLC activity by UTP stimulation of P2Y₂ receptors and inhibition of PLC activity by U-73122 (but not the functionally inactive U-73343) decreased and increased PIP₂ abundance, respectively, indicating that there is active synthesis and hydrolysis of PIP₂ in these cells, the time scale of which is in keeping with receptor-induced PLC-mediated phosphoinositide cycling (Ford et al. 2003; Xu et al. 2003). The time scale for UTP depletion of PIP₂ was consistent with a receptor-mediated event, and immediate hydrolysis of PIP₂ by PLC. In our experiments the time course of PIP₂ resynthesis following receptor stimulation appears to take 40–60 min for global cellular PIP₂ resynthesis events that we observed. Others have shown that PIP₂ resynthesis is mostly complete within 2 min. The differences between these data are most likely due to the fact that PIP₂ resynthesis and reactivation of ion channels in microenvironments as measured via patch clamp techniques are likely to be observed on a much faster time scale. Sequestering PIP₂ with neomycin, which binds PIP₂ with high affinity, did not alter membrane PIP₂ abundance. PIP₂ is principally found in the plasma membrane (McLaughlin et al. 2002; Xu et al. 2003; Hilgemann, 2007) and our data indicate that neomycin complexes with PIP₂ in the membrane without altering PIP₂ abundance.

We show that amiloride-sensitive Na⁺ transport via ENaC in H441 cells requires PIP₂ for maximal channel activity and that signalling pathways which deplete or sequester PIP₂ mediate inhibition of channel currents. This is consistent with reports that removal of PIP₂ inhibits ENaC activity in A6 cells and principal cells from the mouse cortical collecting duct (CCD) (Ma et al. 2002; Yue et al. 2002; Tong et al. 2004; Pochynyuk et al. 2005). The application of PIP₂ to inside-out patches enhanced ENaC open probability (P_o) whilst antibodies raised towards PIP₂ augment channel rundown (Ma et al. 2002; Yue et al. 2002). Purinergic receptor stimulation of PLC and sequestration of PIP₂ with neomycin also inhibited Na⁺ transport in mouse CCD cells and trachea (Kunzelmann et al. 2005). PIP₃ has also been reported to underlie Na⁺ transport activation in the kidney by aldosterone and mouse CCD cells (Tong et al. 2004; Pochynyuk et al. 2007a) but we have not investigated PIP₃ in this study.

Whilst crystal structures of channels are largely unknown, the phosphoinositol head group, consisting of myo-inositol, extends into the cytoplasm (Sansom et al. 2005) so that ENaC has been hypothesized to bind from the cytoplasmic side. The N-termini cytoplasmic loop of β and γ ENaC subunits in human, rat and mouse show clusters of basic residues with nearby hydrophobic residues that are candidates for hydrostatic and electrostatic interactions with acidic lipids such as PIP₂ (Ma & Eaton, 2005). In support of this hypothesis, immunoprecipitation and Western blot demonstrated that PIP₂ bound the α, β and γ ENaC subunits. Furthermore, compromising the PIP₂ interaction by depleting or sequestering PIP₂ in the membrane did not affect ENaC subunit surface expression (N) supporting the concept that PIP₂ binding to the ENaC subunits regulates channel P_o.

AMPK inhibits ENaC activity by compromising the PIP₂–β and PIP₂–γ ENaC subunit interaction

In this and previous studies, AICAR treatment for 1 h activated AMPK and diminished amiloride-sensitive transepithelial Na⁺ transport and apical Na⁺ conductance (Woollhead et al. 2005, 2007). The effect of AICAR was reversed by the AMPK inhibitor Compound C. As expected, none of the modulators of PIP₂ had an effect on AMPK activation or on the component of Na⁺ transport inhibited by AMPK and reversed by Compound C. That Compound C, an inhibitor of AMPK, reversed the AICAR-mediated inhibition implies that AMPK caused the inhibition of Na⁺ transport activity. We are aware that a more convincing effect of AMPK on ENaC would be to directly alter AMPK by introduction of a constitutively active kinase and by suppression (siRNA) but this is technically very challenging in a monolayer scenario in which all of the cells would need to be transiently transfected. In concordance with previous studies, AICAR activation of AMPK inhibited I_SC-Amiloride and GNa⁺ by approximately 50% without effect on membrane abundance of α, β or γ ENaC subunits. This indicates that the effect must be via changes in P_o. As the component of I_SC-Amiloride and GNa⁺ inhibited by AICAR and depletion/sequestration of PIP₂ were similar and neither affected membrane abundance we postulated that they may operate through a related mechanism.

AICAR activation of AMPK did not alter membrane abundance of PIP₂. However, immunoprecipitation of the β and γ ENaC subunits by the anti-PIP₂ antibody was not observed in the presence of AICAR, indicating that AICAR interfered with the binding of these subunits with PIP₂. In support of this idea, we also found that interfering with the PIP₂–ENaC interaction by complexing PIP₂ to neomycin had no effect on PIP₂ abundance but similarly decreased I_SC-Amiloride and GNa⁺. Furthermore, deletion of residues in the amino tract of β ENaC hypothesized to interact with PIP₂ significantly depress channel activity without change to surface population (Chalfant et al. 1999; Kunzelmann et al. 2005). Taken together, we suggest that activation of AMPK inhibits P_o by modulating the interaction of PIP₂ with β and γ ENaC. How AMPK mediates this effect has yet to be explored. It seems unlikely that AMPK phosphorylation of channel subunits alters this interaction. There are putative consensus sequences for AMPK phosphorylation on α, β and γ ENaC but AMPK did not associate with ENaC in CCD cells and did not phosphorylate ENaC subunits (Carattino et al. 2006). An endocytic pathway such as the one involving Nedd4-2 may still work independently from the changes in the PIP₂–ENaC interaction, possibly highlighting the case that binding of Nedd4-2 may potentially also cause the release of PIP₂ from ENaC prior to its internalization, immediately affecting the P_o of ENaC rather than N.

PIP₂ affects ENaC retrieval by AICAR

There was an additive inhibitory effect of AICAR together with UTP or neomycin on I_SC-Amiloride and GNa⁺. Interestingly, this additional effect appeared to be correlated with a change in membrane abundance (N) of α and β ENaC proteins. Why we did not observe changes to the level of γ cannot be explained and is an unusual observation in the light of many studies which indicate that insertion and retrieval of this subunit is highly regulated (Barker et al. 1998; Stockand et al. 2000; Bruns et al. 2003; Snyder, 2005). Nevertheless, a reduction in α and β ENaC protein in the membrane would result in a reduction in channel activity as α is the main pore forming subunit and γ ENaC does not form functional channels alone (Eaton et al. 2004).

The mechanism for this is undoubtedly complex. We speculate that this may involve a multistage process involving selective uncoupling of PIP₂ from the ENaC subunits. AMPK alone uncouples PIP₂ from β and γ ENaC to decrease P_o. UTP or neomycin alone uncouple α, β and γ ENaC from PIP₂ to decrease P_o. When α, β and γ ENaC are uncoupled in the presence of AMPK, then a retrieval pathway is initiated in which α and β ENaC subunits are removed from the membrane. This assumes that PIP₂ binding to α ENaC is a critical step. Such a pathway would integrate the effects we describe with those of Carattino et al. (2005) and Bhalla et al. (2006). ENaC retrieval is mediated by the binding of the ubiquitin ligase Nedd4-2 to proline-rich regions (PY) in the C-termini of the ENaC subunits (Hershko & Ciechanover, 1998; Malik et al. 2005). Ubiquitination of the ENaC proteins then targets them for retrieval from the membrane and proteosomal/lysosomal degradation. These authors showed that activation of AMPK inhibited ENaC activity in Xenopus oocytes by a mechanism involving subunit retrieval. Further work indicated that activation of AMPK phosphorylated the ubiquitin ligase Nedd4-2 in vitro and increased its association with β ENaC in HEK-293 cells (Bhalla et al. 2006). Such a mechanism may underlie retrieval of α and β subunits post release from PIP₂.

Conclusion

In conclusion, this is the first study to our knowledge to demonstrate that AICAR activation of AMPK inhibits transepithelial amiloride-sensitive Na⁺ transport and apical Na⁺ entry via a mechanism that perturbs the PIP₂–ENaC channel interaction to alter P_o. In addition, we show that dissociation of PIP₂ from ENaC together with activation of AMPK further reduced Na⁺ transport by a secondary effect that correlated with ENaC subunit internalization (N). Thus, when PIP₂–ENaC subunit interactions were compromised, ENaC protein retrieval was initiated, indicating that AMPK can modulate ENaC P_o and N to reduce Na⁺ transport. These data indicate that activation of AMPK by AICAR compromises the ENaC subunit–PIP₂ interaction and we hypothesize that this interaction is important for subunit surface expression. We propose that channel subunit retrieval by AMPK is dependent on the PIP₂ concentration being low, as would be the case in physiological situations in which PLC signalling mechanisms are operative. It would appear that using PIP₂ to regulate channel activity as well as trafficking means that ENaC would only be active at the membrane and not during insertion into or retrieval from the cell surface.