Pharmacological activators of AMP-activated protein kinase have different effects on Na⁺ transport processes across human lung epithelial cells

Abstract

Background and purpose:

AMP-activated protein kinase (AMPK) is activated by metformin, phenformin, and the AMP mimetic, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside (AICAR). We have completed an extensive study of the pharmacological effects of these drugs on AMPK activation, adenine nucleotide concentration, transepithelial amiloride-sensitive (I_amiloride) and ouabain-sensitive basolateral (I_ouabain) short circuit current in H441 lung epithelial cells.

Experimental approach:

H441 cells were grown on permeable filters at air interface. I_amiloride, I_ouabain and transepithelial resistance were measured in Ussing chambers. AMPK activity was measured as the amount of radiolabelled phosphate transferred to the SAMS peptide. Adenine nucleotide concentration was analysed by reverse phase HPLC and NAD(P)H autofluorescence was measured using confocal microscopy.

Key results:

Phenformin, AICAR and metformin increased AMPK (α1) activity and decreased I_amiloride. The AMPK inhibitor Compound C prevented the action of metformin and AICAR but not phenformin. Phenformin and AICAR decreased I_ouabain across H441 monolayers and decreased monolayer resistance. The decrease in I_amiloride was closely related to I_ouabain with phenformin, but not in AICAR treated monolayers. Metformin and phenformin increased the cellular AMP:ATP ratio but only phenformin and AICAR decreased cellular ATP.

Conclusions and implications:

Activation of α1-AMPK is associated with inhibition of apical amiloride-sensitive Na⁺ channels (ENaC), which has important implications for the clinical use of metformin. Additional pharmacological effects evoked by AICAR and phenformin on I_ouabain, with potential secondary effects on apical Na⁺ conductance, ENaC activity and monolayer resistance, have important consequences for their use as pharmacological activators of AMPK in cell systems where Na⁺K⁺ATPase is an important component.

Introduction

Recently, there has been much interest in the action of AMP-activated protein kinase (AMPK) in cellular signalling processes, because of its role as a cellular metabolic sensor and its ability to elicit a wide range of effects in diverse tissues (Hardie, 2003; Kemp et al., 2003). This kinase is activated by an increase in the intracellular ratio of AMP:ATP, making it very sensitive to changes in cellular metabolic status. Once activated, AMPK downregulates pathways that consume ATP and upregulates those that generate ATP, in order to maintain cellular energy stores (Hardie, 2003, 2004; Hardie and Sakamoto, 2006).

There are several pharmacological agents that have been used extensively to activate AMPK in intact cells. These include the biguanides metformin, (Zhou et al., 2001), its more potent analogue phenformin (Sakamoto et al., 2004) and the nucleoside 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside (AICAR) (Corton et al., 1995). Phenformin is thought to activate AMPK by inhibiting Complex I of the mitochondrial respiratory chain (El-Mir et al., 2000; Owen et al., 2000), resulting in changes in intracellular levels of AMP and/or the AMP:ATP ratio (Hawley et al., 2005). Phenformin was used in the clinical treatment of type II diabetes, but it has been withdrawn from clinical use in the UK, due the high incidence of lactic acidosis. Metformin, which has replaced the use of phenformin, also inhibits Complex I (El-Mir et al., 2000; Owen et al., 2000), but changes in intracellular AMP:ATP ratios have been difficult to determine with this drug (Hawley et al., 2002; Fryer et al., 2002b). In contrast to the biguanides, AICAR, currently being considered as a clinical alternative, is taken into the cell via adenosine transporters (Gadalla et al., 2004) and then converted by adenosine kinase to the monophosphorylated nucleotide, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranosyl (AICAR) 5′ monophosphate, or ZMP, which binds to and activates AMPK in a manner similar to AMP. While these drugs activate AMPK, they may also have other cellular effects. Therefore, the AMPK inhibitor Compound C has been widely used to attribute downstream effects specifically to the activation of this kinase (Zhou et al., 2001; Fryer et al., 2002b).

Transport of Na⁺ across the lung epithelium from the lumen to interstitium is an essential regulator of airway fluid volume. Na⁺ crosses the epithelial cell apical membrane, predominantly through the amiloride-sensitive Na⁺ channel (ENaC), and is extruded via the basolateral Na⁺K⁺ATPase pump in an energy consuming process. Misregulation of this pathway can lead to accumulation or dehydration of fluid in the airway and contribute to the pathogenesis of airway disease (Boucher et al., 1988; Kerem et al., 1999). Changes in cell metabolism and oxygen tension have been observed to decrease Na⁺ transport processes across bronchiolar epithelium and rat alveolar cell monolayers (Stutts et al., 1988; Ramminger et al., 2000; Mairbaeurl et al., 2002), implying that AMPK may have a role in the signalling pathways mediating these events.

We have shown that pharmacological activation of AMPK with phenformin and AICAR, for 1 h decreased transepithelial Na⁺ transport, measured as short circuit current (I_sc), in H441 airway epithelial cells (Woollhead et al., 2005). These findings provided support for a cellular signalling pathway that links decreased metabolic status and the consequent reduction in ion transport processes to activation of AMPK. Consistent with work in Xenopus oocytes, we found that this effect was the result of decreased apical Na⁺ entry through ENaC (Carattino et al., 2005). However, we also showed that there was a significant pharmacological effect of these agents on Na⁺K⁺ATPase function, which can also regulate transepithelial transport (Woollhead et al., 2005). Moreover, our findings raised the possibility that the widely prescribed drug metformin, also an activator of AMPK, could additionally inhibit ion transport processes in the lung.

To investigate the effect of metformin, phenformin and AICAR in more detail, we have carried out the first extensive study of the pharmacological effects of these drugs in H441 lung epithelial cells. The data presented here indicate that although all drugs activated AMPK, they have different effects on Na⁺ transport processes. Our data indicate that changes in cellular nucleotides, independently of activation of AMPK, could have additional effects on Na⁺K⁺ATPase function and apical Na⁺ entry, which contribute to the difference in potency observed with these drugs. These findings have wide implications for the use of pharmacological activators of AMPK in other cell systems.

Methods

Test system used

The human H441 lung epithelial cell line was obtained from the American Type Culture Collection (ATCC) (Manassas, VA, USA) and maintained in Rosewell Park Memorial Institute-1640 medium supplemented with foetal bovine serum (10%) (Invitrogen, Paisley, Scotland, 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 (penicillin, 100 U ml⁻¹/streptomycin, 100 μg ml⁻¹). Cells were seeded in 25 cm² flasks and incubated in a humidified atmosphere with 5% CO₂ at 37°C. For functional studies, H441 cells were seeded on Snapwell clear membranes (Costar, VWR, Lutterworth, Leicestershire, UK) and cultured overnight. The following day (or when fully confluent) the serum in the medium was replaced with 4% charcoal-stripped serum, with the additional supplement of triiodothyronine (T3) (10 nM) and dexamethasone (200 nM). The monolayers were cultured for 6–7 days in this medium at air interface to induce polarization.

Measurements made

Identification of subunits and measurement of AMPK activity PCR. RNA was extracted from H441 cells using Trizol (Sigma, Poole, Dorset, UK) with a post-extraction treatment with 1 U amplification grade DNase1 for 15 min at room temperature (Invitrogen, UK). A 2 μg weight of RNA was reverse transcribed using avian myeloma virus reverse transcriptase, as previously described (Woollhead et al., 2005). Reaction mixtures contained 50 ng of cDNA, 10 mM dNTPs, 0.2 μl platinum taq polymerase (5 U μl⁻¹), 1.5 mM MgCl₂ and 0.8 μM of primer. α1 AMPK primers (5′–3′; forward, cggcaaagtgaaggttggcaaa and reverse caaatagctctcctcctgagac) gave an expected product size of 227 bp pairs using cycle conditions of 94°C for 1 min, 60°C for 1 min and 72°C for 1 min for 35 cycles. α2 AMPK primers, (5′–3′; forward ttgggagggtgccacaaagaag and reverse, acgggttgaagagatggaagccag) gave an expected product size of 314 bp. Cycling conditions were the same as above, except that the annealing temperature was raised to 61.1°C. Amplification of β-actin (Clontech, Saint-Germain-en-Laye, France) was used as a reaction control. Amplified products were resolved on 2% agarose gels stained with ethidium bromide and visualized with UV light.

Western blotting. Lysates prepared from H441 cells and rat liver tissue were heated to 94°C for 5 min in the presence of 100 mM dithiothreitol (DTT) and 2% w/v sodium dodecyl sulphate (SDS), to denature the proteins. A 30 μg weight of protein was subjected to electrophoresis on SDS-polyacrylamide gels. Fractionated proteins were transferred to nitrocellulose membranes by electroblotting and immunostained with either anti-α1 or -α2 AMPK antiserum, using standard techniques.

AMPK activity. H441 cell protein (100 μg) was incubated for 2 h at 4°C with 10 μg of AMPK α1 or α2 antibodies, which had been prebound to 10 μl of protein G–Sepharose. The immunoprecipitate was then washed 5 × 1 ml of ice-cold immunoprecipitation (IP) buffer ((in mM): Tris, 50 pH 7.4; NaCl, 1000; NaF, 50; Na⁺ pyrophosphate, 5; EDTA, 1; EGTA, 1; DTT, 1; benzamidine, 0.1; phenylmethylsulphonylfluoride, 0.1; 5 μg ml⁻¹ soybean trypsin inhibitor and 1% v/v Triton X-100), and then with 5 × 1 ml of ice-cold IP buffer (as described above, except containing 150 mM NaCl) then with 1 × 1 ml with assay buffer (50 mM HEPES pH 7.4, 1 mM DTT, 0.02% v/v Brij-35). Immunopurified AMPK activity was measured as the amount of radiolabelled phosphate transferred to the SAMS peptide, derived from the sequence surrounding the major AMPK site on acetyl-CoA carboxylase, as previously described (Hardie et al., 2000; Woollhead et al., 2005).

Measurement of I_sc. Monolayers were mounted in Ussing chambers, where the drug was circulated in a physiological salt solution (in mM): NaCl, 117; NaHCO₃, 25; KCl, 4.7; MgSO₄, 1.2; KH₂PO₄, 1.2; CaCl₂, 2.5; D-glucose, 11 (equilibrated with 5% CO₂ to pH 7.3–7.4). The solution was maintained at 37°C, bubbled with 21% O₂+5% CO₂ premixed gas and continuously circulated throughout the course of the experiment. Control and drug-treated monolayers were analysed in parallel. The monolayers were first maintained under open-circuit conditions, while transepithelial potential difference (V_t) and resistance (R_t) were monitored and observed to reach a stable level. The cells were then short circuited by clamping V_t at 0 mV using a DVC-4000 voltage/current clamp and the current required to maintain this condition (I_sc), was measured and recorded using a PowerLab computer interface. Every 30 s, throughout each experiment, the preparations were returned to open-circuit conditions for 3 s, so that the spontaneous V_t could be measured and R_t could be calculated.

Amiloride-sensitive transepithelial Na⁺ I_sc (I_amiloride) was measured by adding 10 μM amiloride to the solution perfusing the apical side of the monolayer. Basolateral ouabain-sensitive I_sc (I_ouabain) was measured by permeabilizing the apical membrane with 75 μM nystatin. The unrestricted access of ions to the intracellular compartment then causes basolateral I_sc to rise. Once peak levels were achieved, 1 mM ouabain was added to the basolateral compartment to inhibit Na⁺K⁺ATPase activity. The ouabain-sensitive component of I_sc (I_ouabain) was calculated as the difference in I_sc before and after application of ouabain, and gives a measure of maximal net transport capacity via Na⁺K⁺ATPase (number of pumps × turnover) in the ionic conditions used (Figures 4a-c).

Measurement of intracellular adenine nucleotide concentration. Adenine nucleotides—ATP, ADP, AMP and TAN (total adenine nucleotides)—were determined in untreated monolayers or those treated with phenformin, AICAR or metformin for 4 h. Cells were washed with cold phosphate-buffered saline (PBS), and then cold (0.4 M) perchloric acid was added to extract the nucleotides. Cell extracts were neutralized with 3 M K₃PO₄ and were analysed with high-performance liquid chromatography, according to the reverse-phase procedures described previously in detail (Smolenski et al., 1990; Kalsi et al., 1999). The equipment used was the Hewlett-Packard 1100 series linked to a diode array detector. Protein content was determined in the perchlorate precipitate by using the Bradford assay.

Measurement of nicotinamide adenine dinucleotide (phosphate) (NAD(P)H) autofluorescence. Cells seeded on coverslips were grown to subconfluence prior to treatment for 1–4 h with phenformin, AICAR or metformin and mounted in a perfusion chamber containing phenol red free culture medium (Invitrogen, UK) at room temperature. Patches of cells were imaged using a Zeiss LSM 510 laser scanning confocal microscope (Carl Zeiss, Jena, Germany) with a blue diode laser and excitation at 405 nm. Emitted fluorescence was captured using LSM 510 software (release 3.2, Carl Zeiss, Jena, Germany) via a Zeiss Apochromat × 63 oil-immersion objective (numerical aperture 1.4). There was no discernable effect on cell confluence or viability from treatment with the drugs.

Experimental design

For short circuit current measurement, sets of snapwells (up to 12, plated at the same time and cultured under similar conditions for a similar length of time) supporting resistive monolayers of H441 cells (⩾300 Ω cm²) were treated in culture with 5 mM phenformin, 2 mM AICAR or 2 mM metformin for 1–48 h. The concentrations of phenformin and AICAR have been shown to elicit a similar elevation of AMPK activity in H441 cells (Woollhead et al., 2005), and 2 mM metformin has been shown to similarly elevate AMPK activity in other cell systems (Fryer et al., 2002a). We also used 0.5 mM phenformin, which we have previously shown to elevate AMPK activity twofold after a 1 h treatment (Woollhead et al., 2005). Cells were pretreated with Compound C (80 μM) (Calbiochem, Nottingham, UK) for 40 min before treatment with phenformin, AICAR or metformin as above. All agents were suspended in culture medium or as a 1000 × concentrated stock in dimethyl sulphoxide. Untreated cells were overlaid with culture medium alone or treated with vehicle control. At the end of each functional experiment, the cell monolayers were rinsed in ice-cold PBS and harvested into lysis buffer for analysis of AMPK activity (in mM): Tris, 50 (pH 7.4); NaCl, 150; NaF, 50; Na pyrophosphate, 5; EDTA, 1; EGTA, 1; DTT, 1; 1% v/v Triton X-100 and 1% protease inhibitor cocktail (Sigma, UK). H441 cell suspensions were then stored at −80°C until required. Cell viability in untreated monolayers and those treated with metformin, phenformin and AICAR for 1, 4, 8 and 24 h was assessed by Trypan Blue exclusion. Adherent (and any non-adherent cells) present in the transwell chamber were released from the filter by incubation with trypsin/EDTA solution (Sigma, UK) (200 μl) for 15 min at 37°C. Cells were resuspended in full culture medium and aliquots were mixed with an equal volume of Trypan Blue (Sigma, UK). White (live) and blue (blue) cells were counted (cells ml⁻¹) using fast read counting chambers (ISL Ltd., Paignton, Devon, UK) and the live/dead cell ratio (indicative of cell viability) was calculated.

Data analysis and statistical procedures

Since the resistance and basal levels of ion transport across the monolayers varied between batches of cells, control experiments and treatments were carried out on monolayers plated on the same day. Results are compiled from at least three independent sets of cells. Data are presented as mean±s.e.m. in Table 1. In Figures 3, 4 and 5, I_sc is presented graphically as percentage of control I_sc to facilitate comparison between the pharmacological agents and effects on transepithelial amiloride-sensitive short circuit current (I_amiloride) and I_ouabain.

Table 1.

Effects of phenformin, AICAR and metformin on H441 cellular nucleotide abundance

	ATP	ADP	AMP	TAN	AMP:ATP	ADP:ATP
Control	26.3±1.1	1.39±0.17	0.23±0.04	27.96±1.3	0.009±0.002	0.05±0.004
Phenformin	21.9±0.7*	3.08±0.14*	1.08±0.0.24**	25.58±0.3*	0.049±0.025*	0.14±0.006***
Metformin	25.3±1.8	1.96±0.12*	0.31±0.02*	27.57±1.6	0.012±0.001*	0.08±0.001**
AICAR	18.7±2.3*	0.88±0.11*	0.17±0.02	19.7±2.4*	0.009±0.002	0.05±0.003

Abbreviations: AICAR, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside;

Amounts of cellular adenine nucleotides (ATP, ADP, AMP) and total adenine nucleotides (TAN) are shown in nmol mg⁻¹ cellular protein. Results are shown as mean±s.e.m.

^*significantly different from control, P<0.05

^**significantly different from control, P<0.01

^***significantly different from control, P<0.001 for n=3 independent experiments apart from phenformin/AMP, n=6.

Statistical analysis was carried out using one-way analysis of variance (ANOVA) with a post hoc Gabriel's pairwise test (Kendall and Stuart, 1968) or Student's t-tests, where applicable; P-values of < 0.05 were considered significant. Results are presented as mean±s.e.m.

Drugs

All drugs and reagents were purchased from Sigma (Poole, UK), unless otherwise stated.

Results

AMPK isoforms in H441 cells

PCR analysis using primers to amplify sequences derived from α1 and α2 catalytic subunits of AMPK showed that mRNA products encoding both isoforms were present in H441 cells. However, amplified products from α1 AMPK were more abundant than those of α2 from similar starting quantities of cDNA. This was not due to inefficiency of the α2 primers, because amplification of both isoforms was similar from RNA extracted from our positive control (human ovarian tissue). The α1 subunit was also more readily detectable than the α2 subunit, in H441 cells, by western blotting (Figures 1a and b). Treatment of H441 cells with either 5 mM phenformin or 2 mM AICAR for 1 h induced a significant increase in AMPK activity of the α1 isoform. There was almost undetectable activity associated with the α2 isoform of AMPK (0.004±0.001 nmol min⁻¹ mg⁻¹, n=6), and this was not significantly changed by treatment with either phenformin or AICAR (Figure 1c).

Figure 1.

Expression and activity of AMPK α1 and α2 in H441 cells. (a) Reverse transcriptase PCR of 50 ng mRNA extracted from H441 cells (H441) and ovarian tissue (+control) using primers to amplify α1-AMPK (227 bp) or α2-AMPK (314 bp) or β-actin (296 bp). (b) Representative western blot of 100 μg protein extracted from H441 cells (H441) or rat liver tissue (liver) immunostained with anti α1 or anti α2 AMPK (both ∼63 kDa). Blots were reprobed with anti-β-actin to control for protein loading. (c) Activity of immunopurified AMPK α1 and α2 from H441 cells that were untreated (control) or treated with 5 mM phenformin (phen) or 2 mM AICAR (AICAR). ^**P<0.01 significantly different from control; mean±s.e.m., n=3. AMPK, AMP-activated protein kinase.

Time course of effect of phenformin, AICAR and metformin on AMPK activity

Treatment of H441 cells with 5 mM phenformin, 2 mM AICAR and 2 mM metformin significantly increased AMPK activity over time (one-way ANOVA, F_5,12=5.45, P=0.0076; F_5,12=9.34, P=0.0008 and F_4,10=49.83, P=0.0001). Phenformin and AICAR increased AMPK activity approximately × twofold after 0.5 h, and elevated activity was sustained at 4, 8 and 24 h (n=3). There was no significant difference between phenformin- and AICAR-induced AMPK activity at any time point, up to 24 h. The metformin-induced increase in AMPK activity was significantly less than that induced by phenformin or AICAR, at 4 and 8 h (P<0.01 and P<0.05, n=3, respectively), but reached similar levels by 24 h (Figure 2).

Figure 2.

Effect of phenformin, AICAR and metformin on AMPK activity in H441 cells. H441 cells were treated in culture for 30 min to 24 h with 5 mM phenformin, 2 mM AICAR or 2 mM metformin. AMPK activity was measured as the rate of phosphorylation of the SAMS peptide (nmol min⁻¹ mg⁻¹) and is expressed as the change in AMPK activity of treated cells compared to control. Results are shown as means±s.e.m. (n=3). Data were analysed using one-way ANOVA followed by a pairwise Gabriel's test. ^*P<0.05; ^**P<0.01, significantly different from corresponding values after treatment with phenformin or AICAR. AICAR, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside; AMPK, AMP-activated protein kinase; ANOVA, analysis of variance.

Effect of phenformin, AICAR and metformin on transepithelial resistance

To further investigate the differences in pharmacological effect of these drugs, we investigated their effect on transepithelial resistance (R_t) of H441 monolayers used in this study. This was lower at ∼300 Ω cm² than we have previously published, and we have found this to be a consequence of passage number. However, spontaneous I_sc and the component of current that was amiloride sensitive remained similar at 34±2 μA cm⁻² and 85±5%, respectively. Compared to untreated monolayers, there was no significant change in R_t of metformin-treated monolayers at any of the time points studied. However, we did observe a decrease in resistance in monolayers treated with 5 mM phenformin after 8 h (P<0.01, n=4) (Figure 3). We were unable to measure a significant R_t across monolayers 24 h after treatment with 5 mM phenformin and 2 mM AICAR (Figure 3). Compared to controls, these snapwell membranes were flooded with medium on their apical surface.

Figure 3.

Effect of phenformin, AICAR and metformin on transepithelial resistance of H441 monolayers. Monolayers were treated with 5 mM phenformin, 2 mM AICAR or 2 mM metformin for 1, 4, 8, 24 and 48 h. Transepithelial resistance (R_t) was expressed in Ωcm². Data are shown as mean±s.e.m., for the number of experiments given in Table 1. R_t could not be measured at 24 or 48 h after 5 mM phenformin or 2 mM AICAR treatment, and are assigned a value of zero. AICAR, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside.

To exclude the possibility that effects on R_t were the result of a drop in cell viability, we used Trypan Blue exclusion to assess the viability of cells contained within the transwell. We found that there was neither a significant elevation in the number of dead cells nor a decrease in the ratio of live/dead cells, which would indicate a decrease in cell viability of monolayers treated for 1, 4 and 8 h with metformin, phenformin or AICAR, compared to controls (n=3 for each time and treatment). We did however, find that there was a significant increase in the number of dead cells (P<0.05, n=3 respectively) and a decrease in the live/dead cell ratio of phenformin- and metformin-treated mononolayers after 24 h; live/dead cell ratio (control) 8±1 to metformin (4±0.5, P<0.05, n=3) and phenformin (2±0.2, P<0.01, n=3). There was a similar trend in monolayers treated with AICAR for 24 h, but we could not attribute significance to these effects.

Effect of Compound C on phenformin, AICAR and metformin inhibition of I_amiloride and I_ouabain

Treatment of monolayers with phenformin or AICAR for 1 h significantly inhibited transepithelial I_amiloride (Figure 4a). Metformin did not significantly inhibit I_amiloride after 1 h, but did inhibit I_amiloride after 4 h (Figure 4a). Therefore, these treatment times were used in subsequent experiments. Phenformin and AICAR also inhibited I_ouabain, but metformin after 4 h had no significant effect on I_ouabain (Figure 4b). Pretreatment of monolayers with the AMPK inhibitor, Compound C, did not prevent the inhibitory effect of phenformin on I_amiloride or I_ouabain, but did prevent the inhibitory effect of AICAR and metformin on I_amiloride, and the effect of AICAR on I_oubain (Figures 4c and d).

Figure 4.

Effect of Compound C on phenformin or AICAR inhibition of I_amiloride and I_ouabain. Monolayers were preincubated with 80 μM Compound C (+Compound C, c and d) or vehicle (−Compound C, a and b) for 40 min, before treatment with 5 mM phenformin (phenformin), 2 mM AICAR (AICAR) for 1 h or 2 mM metformin for 4 h (metformin). I_amiloride (a and c) and I_ouabain (b and d) were calculated as the change in I_sc after apical application of 10 μM amiloride and basolateral application of 100 mM ouabain after apical permeabilization. Data are plotted as percent control and shown as mean±s.e.m. ^* Significantly different from control, P<0.05, n=5; ^** significantly different from control, P<0.01, n=5; ^*** significantly different from control, P<0.001, n=5. AICAR, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside; I_amiloride, amiloride-sensitive short circuit current; I_ouabainouabain-sensitive short circuit current.

Time course of effect of phenformin, AICAR or metformin on amiloride-sensitive transepithelial Na⁺ transport (I_amiloride)

All the pharmacological agents significantly decreased transepithelial I_amiloride but with differing potency order of phenformin>AICAR>metformin. A typical example of an I_sc trace showing I_amiloride for one time point (4 h post-drug treatment) is given in Figure 5. The complete time course is summarized in Figure 6. Phenformin (5 mM) induced a significant decrease in amiloride-sensitive transepithelial I_sc (I_amiloride) over a time course of 8 h (one-way ANOVA, F_5,16=9.23, P=0.0003). A decrease in I_amiloride from 37.9±4.5 to 24.2±1.0 μA cm⁻², P=0.05, n=4 was observed at 15 min. Such a rapid decrease in I_amiloride was not seen with 2 mM AICAR or 2 mM metformin. I_amiloride decreased to minimal levels (∼17% of control levels) at 4–8 h. Further changes in I_amiloride could not be determined after 8 h, because of loss of resistance of the monolayer (see above) Inhibition of I_amiloride by 5 mM phenformin followed an exponential decay curve (r²=0.88) and the time required to reduce I_amiloride to half-maximal values was 31 min (Figures 5a and 6a). Interestingly, a 10-fold lower concentration of phenformin (0.5 mM) also induced a significant exponential reduction in I_amiloride over time (one-way ANOVA, F_5,12=8.21, P=0.0014) (r²=0.97) (data not shown). AICAR inhibition of I_amiloride was slower than that observed with phenformin, half-maximal values were achieved after 60 min and minimal plateau levels were 13 % of control, (one-way ANOVA, F_3,14=6.99, P=0.0042) (r²=0.97) (Figures 5b and 6b). Metformin also induced a significant change in I_amiloride over time (one-way ANOVA, F_4,16=6.15, P=0.003), but it had the slowest effect of all the pharmacological drugs tested. Compared to control, an apparent reduction in I_amiloride was observed 1 h post-treatment, but this did not reach statistical significance until 4 h (P<0.06 and P<0.01, respectively, n=4). Half-maximal reduction of I_amiloride was not achieved until 358 min (∼6 h) post-treatment. In addition, I_amiloride remained higher at every corresponding time point compared to all other treatments (Figures 5c and 6c).

Figure 5.

Effect of phenformin, AICAR and metformin on amiloride-sensitive transepithelial (I_amiloride) and basolateral ouabain-sensitive (I_ouabain) short circuit current. Representative I_sc traces from H441 monolayers before treatment (control; black lines) and after treatment with 5 mM phenformin (a), 2 mM AICAR (b) or 2 mM metformin (c), all for 4 h (grey lines). I_amiloride was calculated as the change in I_sc after apical application of 10 μM amiloride. I_ouabain was measured by apical permeabilization of monolayers with 75 μM nystatin followed by basolateral blockade with 1 mM ouabain. AICAR, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside; I_amiloride, amiloride-sensitive short circuit current; I_ouabain, ouabain-sensitive short circuit current; I_sc, short circuit current.

Figure 6.

Time course of effect of phenformin, AICAR and metformin on amiloride-sensitive transepithelial (I_amiloride) and basolateral ouabain-sensitive (I_ouabain) short circuit current. Monolayers were treated with 5 mM phenformin (a), 2 mM AICAR (b), or 2 mM metformin (c) for 1, 4, 8, 24 and 48 h. Effects on I_amiloride are shown as percent of I_amiloride before treatment (% of control), to allow comparison between the drugs. Data are shown as mean±s.e.m. (n=4, except AICAR 1 h when n=5). One-phase exponential decay curves were fitted to the data using GraphPad Prism4. Best-fit r² values to the data were 0.88 (phenformin), 0.97 (AICAR) and 0.83 (metformin). (b) Effects on I_ouabain are shown as percent of I_ouabain before treatment (% control), to allow comparison between the drugs. Data are shown as mean±s.e.m. (n=4, except AICAR 1 h when n=5). One-phase exponential decay curves were fitted to the data as above for 5 mM phenformin and AICAR and metformin, with r² values of 0.85, 0.65 and 0.1, respectively. AICAR, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside; I_amiloride, amiloride-sensitive short circuit current; I_ouabain, ouabain-sensitive short circuit current; I_sc, short circuit current.

Time course of effect of phenformin, AICAR and metformin on ouabain-sensitive basolateral I_sc (I_ouabain)

A typical example of an I_sc trace showing I_ouabain for one time point (4 h post-drug treatment) is given in Figure 5. The complete time course is summarized in Figure 6. Phenformin (5 mM) induced a significant decrease in ouabain-sensitive basolateral I_sc (I_ouabain), which could be attributed to net Na⁺K⁺ATPase transport capacity, over a time course of 8 h (one-way ANOVA, F_5,16=4.89, P=0.007). A significant decrease in I_ouabain from 59.2±12.6 to 30.4±2.6 μA cm⁻², P<0.05, n=4 was observed at 15 min. This was not observed with 2 mM AICAR or 2 mM metformin. I_ouabain reached minimal levels at 4–8 h. Inhibition of I_ouabain followed an exponential decay curve (r²=0.85) and the time required for 5 mM phenformin to reduce I_ouabain by half was 12 min, more rapid than that observed to reduce I_amiloride. Minimum I_ouabain plateau levels were 26% of control levels (Figures 5a and 6a). A concentration of 2 mM of AICAR significantly decreased I_ouabain at 1 and 4 h, compared to control (P<0.05, n=5 and n=4, respectively). An exponential curve fitted to the experimental data (r²=0.65) indicated that 50% maximal values occurred at 40 min. Minimal I_ouabain levels were achieved by 4 h, but were only reduced to 66% of control levels (Table 1; Figures 5b and 6b). There was an apparent decline in I_ouabain after metformin treatment. Fitting a similar exponential curve to this data indicated that metformin reduced pump activity to 89% of control levels, with 50% maximal values occurring at 70 min (r²=0.1). However, we were unable to attribute significance to the changes in these experiments (one-way ANOVA, F_4,16=0.94, P=0.5) (Figures 5c and 6c).

Relationship between I_amiloride and I_ouabain

The relationship between I_amiloride and I_ouabain was different between the pharmacological treatments. I_ouabain reflects the basolateral transport capacity of Na⁺K⁺ATPase, which is critical for the generation and maintenance of the intracellular gradient for Na⁺ movement into the cell via the amiloride-sensitive channel ENaC. Thus, changes in I_ouabain can directly influence I_amiloride. Phenformin induced a similar rapid exponential decrease in both I_amiloride and I_ouabain (Figure 6a), indicating that inhibition of Na⁺K⁺ATPase transport capacity could mediate the decrease in I_amiloride. AICAR also induced a decrease in I_amiloride, but the inhibitory effect on I_ouabain was much less than that on I_amiloride (see above) (Figure 6b). Furthermore, we could not attribute a significant effect of metformin on I_ouabain, yet it significantly reduced I_amiloride (Figure 6c). Plotting values obtained for I_amiloride against I_ouabain from the 0, 1, 4 and 8 h time points (phenformin and AICAR) and the 0, 1, 4, 8 and 24 h points (metformin) obtained in the time course study showed that there was a significant linear relationship between the two parameters only when monolayers were treated with phenformin (r²=0.97, slope=0.68±0.08, P=0.01, n=4) (Figure 7a). These data indicate that the inhibition of I_amiloride is closely related to the inhibition of I_ouabain in cells treated with phenformin, but not in cells treated with AICAR and metformin (Figures 7b and c).

Figure 7.

Relationship between I_amiloride and I_ouabain in cell treated with phenformin, AICAR and metformin. Values obtained for I_amiloride are plotted against I_ouabain from cells treated with 5 mM phenformin (a), 2 mM AICAR (b) or 2 mM metformin (c). Data are shown as mean±s.e.m. Lines were fitted to the data using GraphPad Prism4 linear regression analysis. Best-fit r² values were 0.97 (phenformin), 0.71 (AICAR) and 0.2 (metformin). There was a significant correlation between I_amiloride and I_oubain cells treated with phenformin (P=0.01, n=4), but not in cells treated with AICAR or metformin. AICAR, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside; I_amiloride, amiloride-sensitive short circuit current; I_ouabain, ouabain-sensitive short circuit current; I_sc, short circuit current.

Measurement of intracellular nucleotide concentration

As maximal inhibition of I_ouabain occurred 4 h after treatment with phenformin and AICAR, we investigated the amount of cellular nucleotides at this time, as we hypothesized that alteration of cellular nucleotides could have additional effects on ENaC and Na⁺K⁺ATPase. Treatment with phenformin decreased cellular ATP (Table 1). This was associated with significant increases in cellular ADP and AMP, and a fivefold increase in the AMP:ATP ratio compared to control (Table 1). Metformin also increased ADP, AMP and the AMP:ATP ratio (1.3-fold compared to control), but these changes were much smaller than that seen with phenformin, and we could not detect any changes in cellular ATP levels. These data are consistent with their action as inhibitors of Complex 1 of the respiratory chain. Interestingly, AICAR significantly decreased cellular ATP and ADP. This was associated with a decrease in the total adenine nucleotide pool and did not change the AMP:ATP or ADP:ATP ratios compared to controls (Table 1).

Effect of phenformin, metformin and AICAR on NAD(P)H autofluorescence

There was a significant increase in NAD(P)H autofluorescence in cells treated with phenformin for 1 h, compared to controls. No autofluorescence was detected in cells treated with AICAR (1 h) or metformin (4 h) (Figure 8). These data indicate that phenformin is a more potent inhibitor of Complex 1 than metformin, and that AICAR has no effect on mitochondrial respiration.

Figure 8.

Effect of phenformin, AICAR and metformin on NAD(P)H autofluorscence. Cells were pretreated in culture for 1–4 h with either 5 mM phenformin, 2 mM AICAR or 2 mM metformin. (a) Panels on the left show representative transmission images of untreated (control) or treated cells (phenformin) (AICAR) (metformin) within the field of view with the corresponding emitted light after 405 nm excitation shown on the right. (b) Fluorescence intensity measurements were made from 20 cells within fields of view acquired from three independent experiments using Zeiss LSM 510 software and data plotted as arbitrary units and shown as mean±s.e.m. ^*** significantly different from control, P<0.001, n=20. AICAR, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside. NAD(P)H, nicotinamide adenine dinucleotide (phosphate).

Discussion and conclusions

We previously showed that both isoforms of AMPK, α1 and α2, were present in H441 cells (Woollhead et al., 2005). In this study, we show that it is the α1 catalytic subunit of AMPK that is predominantly expressed and pharmacologically activated in H441 lung cells. Consistent with our findings, α1-AMPK protein has been localized to the apical membrane of polarized human bronchiolar epithelial cells and catalytic activity was associated with the α1 subunit in Calu-3 lung epithelial cells (Crawford et al., 2005; Hallows et al., 2006). The functional implications of different isoform activation is not yet fully understood, but could underlie tissue-specific differences in AMPK regulation and distribution (Kemp et al., 2003).

We show for the first time that metformin activated AMPK in H441 cell monolayers, although its effect was less potent than that of phenformin and AICAR (Woollhead et al., 2005). The biguanides inhibit Complex 1 of the mitochondrial respiratory chain. Consistent with results from a number of other cell types (Hawley et al., 2002; Fryer et al., 2002a), metformin was a less effective metabolic inhibitor in H441 cells, inducing a much smaller change in the cellular AMP:ATP ratio compared to phenformin and with no significant effect on NAD(P)H autofluorescence. This difference in efficacy between the two biguanide drugs most likely reflects the difference in their chemical structure and the increased lipophilic nature of phenformin (Jalling and Olsen, 1984; Holstein and Egberts, 2006). AMPK is reported to be activated by very small changes in the AMP:ATP ratio, because of its ultrasensitive mechanism of activation (Hardie et al., 1999). However, we suggest that the level of AMPK activity induced by the biguanides in H441 cells is related to their effect on the AMP:ATP ratio.

As expected, AICAR did not change the cellular AMP:ATP ratio or NAD(P)H levels, consistent with its direct action on AMPK. However, it did decrease the cellular content of ATP, ADP and AMP in H441 cells. Why this occurred is unclear. In short-term incubations (up to 1 h) of rat hepatocytes and cultured mouse muscle cells, 0.5 mM AICAR has been reported not to affect cellular ATP, ADP or AMP levels (Corton et al., 1995; Fryer et al., 2002a) and in Chinese hamster ovary (CHO) cells incubated for up to 10 h AICAR transiently increased ATP and its diphosphate and monophosphate forms (Sabina et al., 1985).

I_amiloride

We show that metformin decreased transepithelial amiloride-sensitive Na⁺ transport (I_amiloride) across H441 cells. The effect of metformin was slower than that of AICAR or phenformin and mirrored the lower level of α1-AMPK activity. Thus, together with our previous findings, these data indicate that activation of AMPK was the mechanism by which these drugs inhibited I_amiloride. In support of this, we found that the AMPK inhibitor Compound C prevented the AICAR- and metformin-induced reduction in I_amiloride. In contrast, Compound C did not prevent the inhibitory action of phenformin. This finding contradicts our previous suggestion that phenformin acted via AMPK to decrease I_amiloride. It is important to note that the specificity of Compound C has not been fully reported (Zhou et al., 2001), and some caution is required in the interpretation of its effect. However, this new finding raises the possibility that phenformin elicited an effect on I_amiloride that was either independent of, or additional to, pharmacological activation of AMPK.

Transepithelial I_amiloride can be decreased by inhibition of apical ENaC channels and/or by downregulating basolateral Na⁺K⁺ATPase transport capacity, which would reduce the Na⁺ gradient and the driving force for Na⁺ transport via ENaC. The most likely explanation for our new observation that metformin elevated AMPK activity and inhibited I_amiloride, but had no significant effect on Na⁺K⁺ATPase, would be that activation of AMPK must primarily inhibit apical Na⁺ entry, via the amiloride-sensitive Na⁺ channel ENaC, to decrease I_amiloride independently of effects on Na⁺K⁺ATPase function. This hypothesis would also explain why the fall in I_amiloride was greater than that of I_ouabain when cells were treated with AICAR, and why there was no significant relationship between the fall in I_amiloride and I_ouabain over time with metformin or AICAR. Although we have not yet fully investigated the effect of metformin on apical Na⁺ conductance (G_Na+), we have previously shown that treatment of H441 cells with AICAR decreased apical amiloride-sensitive G_Na+ (Woollhead et al., 2005). Furthermore, Carattino et al. (2005) showed that injection of ZMP (the active form of AICAR) and constitutively active AMPK inhibited the activity of ENaCs expressed in Xenopus oocytes.

The rapid effect (within 4 h) of AICAR on I_amiloride determined in this study and our previous study would support a post-transcriptional mechanism of action of AMPK on ENaC-mediated Na⁺ transport. This could include changes in ENaC open probability (Po) and/or the number of channels in the membrane (n) by disruption of the normal ENaC retrieval and recycling processes (Carattino et al., 2005). However, the further decline in I_amiloride associated with AICAR and metformin between 8 and 24 h means that we cannot exclude additional effects of AMPK on ENaC gene transcription or RNA stability in these cells. Both possibilities are the subject of further study.

I_ouabain

Phenformin and AICAR both inhibited I_ouabain, although the effect of AICAR was less potent and was reversible with Compound C. Interestingly, an effect common to phenformin and AICAR, but not metformin, was that they evoked a significant decrease in monolayer resistance (R_t) after 8 h. As a reduction in the live/dead cell ratio was only observed after a 24 h treatment with phenformin and metformin but not AICAR, we reasoned that the effect on R_t was unlikely to be due to a reduction in cellular viability. Recent reports have indicated that Na⁺K⁺ATPase is important in regulating tight junction formation (Kaplan, 2005; Rajasekaran et al., 2005). Thus, the inhibition of its transport capacity by phenformin and AICAR may be the mechanism by which epithelial resistance is lost.

Reduction of the intracellular Na⁺ concentration intracellular Na⁺ ([Na⁺]_i) by inhibition of apical Na⁺ entry has been shown to decrease Na⁺K⁺ATPase surface expression and activity in cortical collecting duct cells (Vinciguerra et al., 2003). However, our new data show that the inhibition of I_ouabain preceded inhibition of I_amiloride and there was no significant correlation between the inhibition of I_amiloride and I_ouabain with AICAR or metformin. We previously suggested that activation of AMPK with phenformin and AICAR could inhibit Na⁺K⁺ATPase directly. The finding that AICAR inhibition of I_ouabain was reversed by Compound C would support our suggestion. However, Compound C did not reverse the effect of phenformin and metformin activated AMPK, but had no effect on I_ouabain. These findings raise the possibility that other mechanisms are involved. Both phenformin and AICAR decreased cellular ATP levels (metformin did not) and Na⁺K⁺ATPase function is dependent on ATP concentration. We previously suggested that only a drastic depletion of ATP would alter Na⁺K⁺ATPase activity (Boldyrev et al., 1991; Woollhead et al., 2005). However, more recent evidence indicates that the relationship between intracellular ATP concentration and Na⁺K⁺ATPase function is non-hyperbolic and involves binding of ATP to both high- and low-affinity sites on the protein (Ward and Cavieres, 1998; Ward et al., 2006). Thus, a reduction in ATP could lead to a decrease in Na⁺K⁺ATPase function. Certainly, a diminution of ATP in renal cortical tubular cells (Coux et al., 2001) and reduced ATP levels in rat cerebellar granule cells (Petrushanko et al., 2006) were associated with a reduction of Na⁺K⁺ATPase function. Moreover, Compound C has been shown to inhibit AICAR uptake into cells and subsequent accumulation of ZMP. It could, therefore, prevent the effect on cellular nucleotides, in addition to any direct inhibition of AMPK activity (Fryer et al., 2002b). In vitro, Na⁺K⁺ATPase transport activity is also decreased by increased ADP concentration (Boldyrev et al., 1991). As cellular ADP levels were increased 200% by phenformin; this could be an additional factor contributing to the greater inhibition of I_ouabain by this drug compared to AICAR. Finally, Luly et al. (1977) showed that phenformin rapidly inhibited Na⁺K⁺ATPase activity in isolated liver plasma membranes. Their suggestion that phenformin could alter the physico-chemical properties of the membrane and change the activities of membrane bound enzymes could provide another AMPK-independent mechanism of action.

Whatever the mechanism, the powerful inhibitory effect of phenformin on I_ouabain could underlie the AMPK-independent action and the more potent effect on I_amiloride compared to AICAR. Inhibition of Na⁺K⁺ATPase would elevate [Na⁺]_i, reducing the driving force for Na⁺ entry. Increases in [Na⁺]_i have also been shown to decrease ENaC activity by ‘feedback inhibition' (Garty and Palmer, 1997) in a number of cell systems, including salivary duct cells (Dinudom et al., 1998), rat cortical collecting tubule (Frindt et al., 1995) and Xenopus oocytes expressing ENaC (Abriel and Horisberger, 1999). The observation that there was a very close relationship between I_amiloride and I_ouabain in cells treated with phenformin, coupled with our previous finding that phenformin decreased apical G_Na+, indicates that the AMPK-independent action of phenformin could involve such a mechanism.

In conclusion, we show for the first time that metformin decreases transepithelial amiloride-sensitive Na⁺ transport across H441 lung cells. This finding has significant implications for its future therapeutic use. We also show that the three pharmacological activators of AMPK we have used have different effects on Na⁺ transport processes across lung epithelium. Importantly, our data show that while pharmacological activation of α1-AMPK is associated with inhibition of amiloride-sensitive Na⁺ channels (ENaC), additional changes in intracellular nucleotide levels (AICAR and phenformin) and potential effects on membrane fluidity (phenformin) could also affect basolateral Na⁺K⁺ATPase transport capacity, with ensuing effects on apical Na⁺ conductance, ENaC activity and monolayer resistance. More work is now required to elucidate the cellular mechanisms by which AMPK and cellular nucleotide concentrations evoke such changes. In the interim, these findings have important consequences for the clinical use of pharmacological activators of AMPK, and the interpretation of their effect in other cell systems where Na⁺K⁺ATPase is an important component.

Abbreviations

AICAR

5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside

AMPK

adenosine monophosphate-activated protein kinase

DTT

dithiothreitol

G_Na+

Na⁺ conductance

I_amiloride

amiloride-sensitive short circuit current

I_ouabain

ouabain-sensitive short circuit current

I_sc

short circuit current

NAD(P)H

nicotinamide adenine dinucleotide (phosphate)

[Na⁺]_i

intracellular Na⁺

SDS

sodium dodecyl sulphate

ZMP

5-aminoimidazole-4-carboxamide-1-β-D-ribofuranosyl (AICAR) 5′monophosphate

Conflict of interest

The authors state no conflict of interest.