5′-AMP-activated protein kinase (AMPK) functions as an energy sensor to provide metabolic adaptation under conditions of ATP depletion, such as hypoxia and inhibition of oxidative phosphorylation. Whether activation of AMPK is critical for stimulation of glucose transport in response to inhibition of oxidative phosphorylation is unknown. Here we found that treatment of Glut1-expressing Clone 9 cells with sodium azide (5 mM for 2 h) or the AMPK activator 5′-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside (AICAR, 2 mM for 2 h) stimulated the rate of glucose transport by two- to fourfold. Use of small interference RNA (siRNA) directed against AMPKα1 or AMPKα1 + AMPKα2 (total AMPKα) resulted in a significant inhibition of the glucose transport response and the content of phosphorylated AMPKα1 + phosphorylated AMPKα2 (total p-AMPKα) and phosphorylated acetyl-CoA carboxylase (p-ACC) in response to azide. Transfection with siRNA directed against AMPKα2 did not affect the glucose transport response. The efficacy of transfection with siRNAs in reducing AMPK content was confirmed by Western blotting. Incubation of cells with compound C, an inhibitor of AMPK, abrogated the glucose transport response and abolished the increase in total p-AMPK in azide-treated or hypoxia-exposed cells. Simultaneous exposure to azide and AICAR did not augment the rate of transport in response to AICAR alone. There was no evidence of coimmunoprecipitation of total p-AMPKα with Glut1. However, LKB1 was associated with total p-AMPKα. We conclude that activation of AMPK plays both a sufficient and a necessary role in the stimulation of glucose transport in response to inhibition of oxidative phosphorylation.
- small interference RNA
- compound C
exposure of cells and tissues to conditions associated with decreased energy production, such as ischemia, hypoxia, and inhibition of oxidative phosphorylation, leads to stimulation of glucose transport and glycolytic ATP production (4, 8, 13). This response is of vital importance in cells in which the rate of glucose transport is rate limiting for glucose metabolism. For example, we previously found that inhibition of oxidative phosphorylation by hypoxia or chemical agents such as cyanide or azide in Clone 9 cells, a rat liver cell line expressing the Glut1 isoform of facilitative glucose transporters, leads to a marked early fall in cellular ATP and a rise in cellular ADP and AMP associated with increases in ADP-to-ATP and AMP-to-ATP ratios (13, 14, 25). However, within 30–60 min after the start of the exposure, cellular ATP, ADP, and AMP levels return to normal (or near-normal) levels and are maintained at these new levels for 24 h, despite the continued presence of the inhibitors (13, 14, 25). This adaptive response is associated with stimulation of glucose transport and lactate production (5, 14). More prolonged exposure to azide or CoCl2 (a stimulator of the hypoxia-responsive pathway) results in an increase in Glut1 mRNA expression and subsequent enhanced Glut1 expression; however, this effect occurs only after ∼4 h of exposure (6, 11, 23, 24).
More recently, we documented that the above-described response is associated with stimulation of 5′-AMP-activated protein kinase (AMPK) brought about by increased phosphorylation of AMPK, a finding that has been confirmed by others (1, 4, 13). Although the stimulation of AMPK by 5′-aminoimidazole-4-carboxamide-1,β-d-ribofuranoside (AICAR) or use of constitutively active AMPK leads to stimulation of glucose transport in Glut1-expressing cells (1, 31), the potential effect of inhibition of the AMPK pathway on stimulation of glucose transport has not been examined. Moreover, the signals, sequence of events, and underlying mechanisms by which the activation of glucose transport is initiated and maintained after stimulation of AMPK are incompletely understood. Although these results establish that stimulation of AMPK is sufficient for stimulation of Glut1-mediated glucose transport, they do not define whether the simulation of AMPK activity is necessary for the enhancement of glucose transport in response to inhibition of oxidative phosphorylation or, alternatively, whether activation of AMPK simply occurs in parallel with the glucose transport response.
AMPK is a heterotrimeric protein complex; on phosphorylation of its α-subunit, the enzyme's kinase activity is activated under conditions of metabolic stress, especially in response to increased cellular AMP-to-ATP ratio (1, 4, 8, 13, 27). Indeed, it has been proposed that AMPK serves as a universal energy sensor (a “switch”) that controls the metabolic pathways regulating rates of energy production and energy consumption (8, 27). Stimulation of AMPK in cells that express Glut4 is associated with translocation of this glucose transporter from intracellular vesicles to the plasma membrane, thereby leading to stimulation of glucose transport (2, 15). In cells that express the Glut1 isoform (e.g., Clone 9 cells), enhancement of AMPK activity also stimulates glucose transport, an effect that appears to be predominantly due to activation of Glut1 transporters preexisting in the plasma membrane (4, 12, 25).
In the present study, using small interference RNA (siRNA) technology to suppress AMPK activity and then to test whether the glucose transport response is negatively affected, we examined the possibility that stimulation of AMPK plays a critical role in the enhancement of glucose transport in response to inhibition of oxidative phosphorylation. AICAR was used as a positive control throughout the study. In addition, we employed compound C as an inhibitor of AMPK activity in azide-treated or hypoxia-exposed cells. Suppression of activity of AMPK and its isoforms was determined by Western blotting of total AMPKα (AMPKα1 + AMPKα2) and its phosphorylated forms (p-AMPKα1 and p-AMPKα2) and by measurement of phosphorylated acetyl-CoA carboxylase (p-ACC) content in response to azide (and AICAR). A potential additive effect of azide and AICAR on glucose transport was determined. We also examined whether Glut1 associates with AMPK in response to azide and whether Glut1 becomes phosphorylated on stimulation of AMPK activity. Finally, we examined the potential role of LKB1 (10, 22) and c-Jun NH2-terminal kinase (JNK) (30) as upstream kinases that could phosphorylate AMPK in response to inhibition of oxidative phosphorylation.
MATERIALS AND METHODS
Clone 9 cells were obtained from American Type Culture Collection (Rockville, MD). DMEM, Opti-MEM, trypsin-EDTA, calf serum, Lipofectamine 2000, and recombinant protein G-agarose were purchased from GIBCO (Grand Island, NY); cell culture dishes from Corning Glass Work (Medfield, MA); [3-O-methyl-d-3H]glucose ([3H]3-OMG, 3.4 mCi/mmol) and enhanced chemiluminescence Western blotting detection kit from Amersham Life Science (Arlington Heights, IL); peroxidase-conjugated goat anti-mouse and anti-rabbit IgG, mouse anti-β-actin, phloretin, cytochalasin B (CB), PMSF, AICAR, anisomysin, and standard chemical reagents from Sigma (St. Louis, MO); and rabbit anti-AMPK-α-pan, anti-AMPKα1, anti-AMPKα2, monoclonal anti-phosphorylated (Thr172) AMPKα1 + AMPKα2 (recognizing both phosphorylated isoforms), anti-p-ACC (Ser79), and mouse anti-phosphorylated JNK from Upstate Cell Signaling Solutions (Charlottesville, VA). Mouse anti-Glut1 monoclonal antibody (MAb 37-4) was described previously (31). Rabbit anti-Glut1 antibody was obtained from Abcam (Cambridge, MA); goat anti-total LKB1, LKB1 positive control, peroxidase-conjugated donkey anti-goat IgG, and protein A/G PLUS-agarose from Santa Cruz Biotechnology (Santa Cruz, CA); rabbit anti-phosphoserine from Zymed Laboratories (San Francisco, CA); and siRNA directed against AMPKα1 or AMPKα2, Block-iT fluorescent oligonucleotide, and si-control nonspecific targeting siRNA from Dharmacon (Lafayette, CO).
Clone 9 cells in 60-mm dishes in triplicate were maintained in DMEM containing 5.6 mM d-glucose supplemented with 10% calf serum at 37°C in a 9% CO2-humidified chamber (pH 7.4). Cells were used between passages 30 and 50. On confluence (between day 3 and day 4), the medium was replaced with serum-free DMEM for 24 h. In experiments employing compound C, cells were preincubated in the presence or absence of the reagent (20 μM dissolved in 20 μl of DMSO) for 30 min before coincubation with diluent, 5 mM azide, or 2 mM AICAR for 2 h; cells that were not treated with compound C received the diluent. Similarly, in experiments involving hypoxia and compound C, cells were preincubated in the presence and absence of compound C for 30 min before incubation in the hypoxic chamber with nominal (>0.5%) O2 for 2 h. Because the hypoxic chamber was equilibrated with 95% N2-5% CO2, cells not exposed to hypoxia were incubated in a 5% CO2-humidified chamber for 2 h before assay. To test for the additive effect of azide and AICAR, Clone 9 cells were treated with diluent, 5 mM azide, 2 mM AICAR, or azide + AICAR for 2 h before CB-inhibitable [3H]3-OMG uptake assay. After the uptake assay, cells were harvested in 100 μl of Nonidet P-40 (NP-40) lysis buffer: 5 mM NaCl, 25 mM Tris, 25 mM sodium fluoride, 5 mM sodium pyrophosphate, 0.5 mg/l leupeptin, 1 mg/l aprotinin, 1 mM sodium orthovanadate, 0.1 mM PMSF, and 0.5% NP-40 (pH 7.5). After centrifugation, 70 μl of the postnuclear lysate were used for radioactive counting and the rest for SDS-PAGE and Western blot analysis.
Application of siRNA directed against AMPK.
At 70% confluence, Clone 9 cells in six-well dishes were incubated with fresh DMEM without antibiotics for 2 h before transfection. Lipofectamine 2000 was used according to the manufacturer's protocol for transfection of cells in duplicate with siRNA “smart pool” targeting four different sections of the mRNA encoding AMPKα1, AMPKα2, or total AMPKα (100 pmol/ml). At 48 h after transfection, the medium was changed, and the cells were treated with diluent, 5 mM azide, or 2 mM AICAR for 2 h before CB-inhibitable [3H]3-OMG glucose uptake assay and Western blotting (see above).
Potential nonspecific effects of transfection with siRNA on AMPK content were monitored by transfection of nonspecific (scrambled) siRNA and immunoblotting for AMPK. In preliminary experiments, use of nonspecific siRNA resulted in no change in the cellular content of AMPKα1 or in the amount of total p-AMPKα (p-AMPKα1 + p-AMPKα2) generated in response to AICAR.
Transfection efficiency was monitored using 100 pmol/ml of Block-iT fluorescent oligonucleotide, and fluorescence was observed 24–48 h after transfection by microscopy. Transfection efficiency was 90–95% at 48–72 h. In preliminary experiments, various concentrations (25–200 pmol/ml) of siRNA were employed to determine the optimal condition for reducing AMPK expression without causing cell toxicity. The ideal concentrations of siRNA were 100 pmol/ml for each isoform used individually and 50 pmol/ml for each used in combination. Use of 150–200 pmol/ml siRNA was associated with toxicity and cell death.
Measurement of CB-inhibitable [3H]3-OMG uptake.
Cells in duplicate six-well dishes in experiments using siRNA or in triplicate 60-mm dishes in experiments using compound C, azide, hypoxia, or AICAR were incubated for 60 s in glucose uptake medium containing DMSO alone or DMSO containing CB at a final concentration of 50 μM. The uptake medium consisted of 1.0 ml of DMEM with 1 μCi of [3H]3-OMG and 1 μl of CB solution or DMSO. Uptake was stopped by addition of an ice-cold solution of 100 mM MgCl2 and 100 μM phloretin (2, 4). Cells were harvested in the NP-40 lysis buffer, and radioactivity was determined by scintillation spectrometry. [3H]3-OMG uptake was calculated in parallel assays as the difference between uptake in the absence of CB and uptake in the presence of CB. Unless noted otherwise, experiments were repeated three times, and the results were averaged. Uptake in control and treated cells was determined in parallel. We reported previously that the rate of glucose transport in control cells under basal conditions and the degree of stimulation of transport by any specific agent vary between experiments, possibly reflecting the “state” of the cells, their passage number, and the degree of confluence (4, 13, 24). Nevertheless, a two- to more than fourfold stimulation of glucose transport in response to azide is the constant finding. For this reason, control and treated cells were studied in parallel in all experiments.
SDS-PAGE and Western blotting.
NP-40 lysis buffer was used for preparation of all cell lysates. In certain experiments, a portion of the lysate was used for radioactivity counting (see above). Cell lysates from different treatment groups were centrifuged at 14,000 g for 20 min for removal of nuclei and insoluble materials. Protein samples were separated by 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes. Membranes were blocked using 5% nonfat milk and incubated with rabbit anti-AMPKα1 (63 kDa), rabbit anti-AMPKα2 (63 kDa), or rabbit anti-AMPK-α-pan (63 kDa) for measurement of total AMPKα; rabbit anti-phosphorylated (Thr172) AMPKα1 + AMPKα2 for measurement of total p-AMPKα (63 kDa); rabbit anti-p-ACC or rabbit anti-total ACC (265 kDa) for measurement of total p-ACC; and rabbit anti-Glut1 (55 kDa) for measurement of Glut1. Goat anti-LKB1 (52 kDa) was used in Western blot and immunoprecipitation experiments. Mouse anti-β-actin (42 kDa) antibody was used to verify equal loading of the gels. The secondary antibody was a 1:500–2,000 (vol/vol) dilution of horseradish peroxidase-conjugated goat anti-rabbit, anti-mouse, or donkey anti-goat antibody in Tris-buffered saline-Tween 20 (50 mM Tris, 150 mM NaCl, and 0.05% Tween 20, pH 7.5). Blots were developed using enhanced chemiluminescence reagents; immunoreactive bands were visualized on Kodak X-Omat film, and the intensity of the bands was determined by densitometry. Blots were reprobed with mouse anti-β-actin for control of protein loading, with horseradish peroxidase-conjugated goat-anti-mouse antibody used as secondary antibody. Each experiment was repeated at least three times.
For study of JNK activation, confluent cells preincubated in serum-free DMEM for 24 h were treated with 10 μM anisomycin or 2 mM AICAR for 1 h. Postnuclear cell lysates were subjected to 10% SDS-PAGE and probed with mouse anti-p-JNK (46 and 54 kDa), rabbit anti-total p-AMPKα, p-ACC, and mouse anti-β-actin.
Immunoprecipitation of Glut1, total p-AMPKα, and LKB.
Cells were treated with diluent, 5 mM azide, or 2 mM AICAR for 2 h and then lysed in the NP-40 buffer. Protein (3 mg) from the postnuclear cell lysate from each condition was immunoprecipitated using mouse monoclonal anti-Glut1 antibody (1:100) with prewashed protein G-Sepharose overnight. Samples were centrifuged at 3,000 rpm for 1 min at 4°C, and the pellets were washed three times with 10 vol of ice-cold 0.5% NP-40 buffer. The final pellet was treated with SDS sample buffer, divided into three equal parts, and analyzed in separate blots. Membranes were probed with rabbit anti-Glut1, rabbit anti-phosphorylated (Thr172) AMPKα1 + AMPKα2, and rabbit anti-phosphoserine antibody. In other experiments, 1 mg of protein from the postnuclear lysate from each of the above-described conditions was immunoprecipitated with rabbit anti-phosphorylated (Thr172) AMPKα1 + AMPKα2, and the product was probed for Glut1 and total p-AMPKα in separate blots.
To explore the association between AMPK and LKB1, Clone 9 cells treated with diluent, 5 mM azide, or 2 mM AICAR for 2 h were lysed in the NP-40 buffer. One milligram of protein from the postnuclear cell lysate prepared from each condition was immunoprecipitated in 500 μl of NP-40 buffer overnight at 4°C with 2 μg of rabbit anti-phosphorylated (Thr172) AMPKα1 + AMPKα2. Products were prepared in duplicate blots and probed for LKB1 and total p-AMPKα. In additional experiments, 1 mg protein from the postnuclear lysate prepared from cells treated as described above was immunoprecipitated with 2 μg of goat anti-total LKB1 antibody, immunoblotted, and probed with rabbit anti-phosphorylated (Thr172) AMPKα1 + AMPKα2 and anti-total LKB1 in separate blots.
Values are means ± SE. ANOVA was used for significance. In certain analyses involving direct comparison between two groups, Student's t-test was employed. In all cases, P < 0.05 was considered significant.
Effect of siRNA directed against AMPKα2 on azide-stimulated glucose transport.
At 48 h after transfection with 100 pmol/ml of siAMPKα2 directed against four segments of AMPKα2 or treatment with Lipofectamine, the acute effect of 5 mM azide on 3-OMG-inhibitable glucose transport was measured. AICAR (2 mM for 2 h) was used as a positive control. We chose a 2-h time point for these studies, because Glut1 expression, which is increased at later times, does not occur before 4 h (24, 25). Treatment with azide or AICAR for 2 h stimulated the rate of glucose transport in mock-transfected cells by 2.8 ± 0.3- and 3.5 ± 0.4-fold but did not affect the cells treated with siRNA directed against AMPKα2 (2.9 ± 0.9- and 3.5 ± 0.4-fold, respectively, P > 0.05; Fig. 1A). Samples of lysates were assayed for the content of AMPKα2 and total p-AMPKα. Immunoblots of AMPKα2 showed an average decrease of 94%, 88%, and 98% in untreated and azide- and AICAR-treated cells, respectively. However, immunoblots using anti-phosphorylated (Thr172) AMPKα1 + AMPKα2, which recognizes phosphorylation of AMPKα1 and AMPKα2, showed that the content of total p-AMPKα was not decreased in siAMPKα2-treated cells and increased significantly in response to azide or AICAR with or without siAMPKα2 treatment (Fig. 2B). Total AMPKα content was slightly (but not significantly) reduced in cells exposed to siRNA. In addition, p-ACC content was significantly increased in azide- or AICAR-exposed cells in the presence or absence of siAMPKα2. Finally, the abundance of β-actin remained unchanged. These results suggest that AMPKα2 may be a minor component of total AMPKα in Clone 9 cells.
Effect of siRNA directed against AMPKα1 on azide-stimulated glucose transport.
We next determined the importance of AMPKα1 in the glucose transport response to azide. Again, AICAR was used as a positive control. Cells were treated with siRNA directed against four different sections of AMPKα1 mRNA. Control cells were treated with Lipofectamine. Transfection with siAMPKα1 (100 pmol/ml) caused a slight, but significant, fall in the rate of glucose transport in control (basal) cells but significantly suppressed azide- and AICAR-induced stimulation of glucose transport; however, the rate of uptake was not decreased to the level in control cells (Fig. 2A).
To determine whether the decrease in glucose transport in siAMPKα1-treated cells was associated with suppression of AMPKα1 expression and phosphorylation, we prepared immunoblots from samples from the above-described transport experiments. Immunoblots using anti-AMPKα1 demonstrated an 89%, 83%, and 78% decrease in AMPKα1 in control and azide- and AICAR-treated cells, respectively (Fig. 2B). Immunoblots using anti-phosphorylated (Thr172) AMPKα1 + AMPKα2 showed that the content of total p-AMPKα increased in azide- and AICAR-treated cells compared with Lipofectamine-treated cells. In contrast, total p-AMPKα content was significantly suppressed in siAMPKα1-transfected cells treated with azide or AICAR to 0.23 ± 0.1- and 0.25 ± 0.1-fold, respectively, compared with control cells. Under basal conditions, endogenous p-AMPKα1 was also suppressed to 0.04 ± 0.02-fold in siAMPKα1-transfected cells compared with controls. Total AMPK content was also measured. Exposure to siAMPKα1 resulted in a significant (∼70%) decrease in total AMPK content in cells under basal conditions, and the marked decrease in total AMPK content was also evident in siAMPKα1-exposed cells treated with azide or AICAR. The increase in p-ACC content was greatly suppressed in cells preexposed to siAMPKα1. Finally, no significant change in β-actin content was observed in control or treated cells. The incomplete suppression of total AMPKα and total p-AMPKα in siAMPKα1-treated cells noted above is consistent with the less-than-complete suppression of azide- and AICAR-stimulated glucose transport in these same cells (see above). These results are consistent with the previous report that AMPKα1 predominates in Clone 9 cells (4).
Effect of siRNA directed against AMPKα1 and AMPKα2 on azide-stimulated glucose transport.
Given the above observations, we reasoned that the effects of simultaneous administration of siAMPKα1 and siAMPKα2 should mimic or slightly exceed the effects of siAMPKα1 alone. Combined transfection with siRNA directed against AMPKα1 and AMPKα2 (each at the submaximal level of 50 pmol/ml) resulted in a significant decrease in the rate of glucose transport in azide- and AICAR-treated cells (compared with control cells under basal conditions) to approximately the rate observed under basal conditions (Fig. 3A); stimulation of glucose transport in response to azide or AICAR was significantly suppressed in cells preexposed to siAMPKα1 (P < 0.05, by t-test). Treatment of cells with siAMPKα1 + siAMPKα2 also reduced the glucose transport rate in control cells, perhaps because of suppression of endogenous AMPK activity or, in part, nonspecific effects.
Immunoblots of samples from the above-described experiments showed a highly significant suppression of total AMPKα expression by 86%, 85%, and 89% in control and azide- and AICAR-treated cells, respectively (Fig. 3B). Total p-AMPKα was inhibited by 93%, 81%, and 88%, respectively, in the three groups. The content of p-ACC in response to azide (or AICAR) was markedly reduced in cells preexposed to the combination of the siRNAs, whereas β-actin, used as a control for loading, remained constant.
Effect of compound C on AMPK activation and on the glucose transport response to azide.
Clone 9 cells preexposed to 20 μM compound C for 30 min were treated with diluent, azide, or AICAR for 2 h before measurement of glucose transport. Exposure to compound C resulted in a slight reduction in the rate of glucose transport in control cells. Glucose uptake was stimulated by azide and AICAR by 3.7 ± 0.3- and 3.8 ± 0.4-fold, respectively, as expected, in cells that were not exposed to compound C. In contrast, pretreatment with compound C completely abolished the glucose transport response to azide and AICAR (Fig. 4A). Compound C profoundly suppressed total p-AMPKα in azide- and AICAR-treated cells to well below the level in control cells (Fig. 4B).
Effect of compound C on AMPK activation and on the glucose transport response to hypoxia.
The purpose of these experiments was twofold: 1) to verify that the stimulation of glucose transport occurred in response to hypoxia and is associated with stimulation of AMPK and 2) to test the effect of compound C on both responses. After 2 h of incubation under hypoxic conditions (>0.5% O2), the rate of glucose transport was augmented to a degree similar to that in cells exposed to azide in the presence of oxygen (Fig. 5A). The stimulation of glucose transport in response to hypoxia was completely prevented by compound C. Similar to experiments with azide described above, compound C decreased the content of total p-AMPKα in cells exposed to hypoxia, whereas total AMPKα remained constant (Fig. 5B).
Effect of azide and/or AICAR on glucose transport and total p-AMPKα.
The results described above suggest that activation of AMPK plays an important role in azide-stimulated glucose transport. Hence, we examined whether the stimulation of glucose transport in response to azide is additive to the response to AICAR. Cells were treated with diluent, azide, AICAR, or azide + AICAR for 2 h before measurement of glucose transport (Fig. 6A). Exposure to azide and AICAR alone resulted in a 2.6 ± 0.3- and a 4.5 ± 0.4-fold increase in the rate of transport, respectively. Exposure to azide + AICAR resulted in no further stimulation of glucose transport compared with cells treated with AICAR alone.
Immunoblots prepared from the above-described transport experiments showed that the content of total p-AMPKα increased in azide- and AICAR-treated cells (Fig. 6B). The content of total AMPKα was unchanged. Interestingly, exposure to azide + AICAR resulted in a higher abundance of total p-AMPKα than exposure to either agent alone, although the rate of transport was not further augmented.
Potential association of total p-AMPKα with Glut1 and phosphorylation of Glut1.
The mechanism by which AMPK activation leads to simulation of glucose transport is only partially understood. Studies in the past few years have shown that activation of AMPK by AICAR or other stimuli leads to translocation of Glut4 to the plasma membrane and, consequently, stimulation of glucose transport in skeletal muscle, adipose tissue, and heart (2, 15). In the case of cells such as Clone 9, which express only the Glut1 isoform of facilitative glucose transporters, the mechanism by which stimulation of AMPK augments glucose transport is not understood, and activation of Glut1 preexisting in the plasma membrane has been proposed (1, 4). Here we explored the possibility that p-AMPK associates with Glut1 and leads to its phosphorylation. Lysates prepared from cells treated with diluent, 5 mM azide, or 2 mM AICAR for 2 h were immunoprecipitated using anti-Glut1 antibody, and cell lysates as well as the immunoprecipitation products were assayed for Glut1 and total p-AMPKα (Fig. 7A). The content of total p-AMPKα increased in lysates of cells exposed to azide or AICAR. However, no total p-AMPKα was detected in Glut1 immunoprecipitates (Fig. 7A, top). Similar blots were developed using anti-phosphoserine antibody (Fig. 7A, bottom). Multiple phosphorylated bands in lysates from control and treated cells showed no obvious difference (Fig. 7A, bottom). In samples of Glut1 immunoprecipitates, a few phosphoserine-containing protein bands (∼250, ∼120, and 60–55 kDa) were present; none of these proteins showed a differential change in their phosphorylation status in response to exposure to azide or AICAR. In addition, no phosphoserine was detected in the region of Glut1 itself in immunoprecipitates from control and stimulated cells (Fig. 7A, bottom). In addition, we immunoprecipitated total p-AMPKα from control cells and cells exposed to azide or AICAR (Fig. 7B). Treatment with azide and AICAR increased the content of total p-AMPKα in cell lysates and in total p-AMPKα immunoprecipitates. However, no Glut1 was detected in immunoprecipitates of total p-AMPKα.
Potential role of LKB1 and JNK in activation of AMPK.
Results of recent studies indicate that increased binding of AMP to the regulatory γ-subunit of AMPK enhances the phosphorylation of the α-catalytic subunit of the enzyme by an upstream kinase, LKB1 (22, 28). Hence, we examined whether LKB1 is expressed in Clone 9 cells and whether it can be found in association with total p-AMPKα in response to azide exposure; AICAR was employed as a positive control. Cells were treated as described above. As shown in Fig. 8, LKB1 was expressed in Clone 9 cells. As expected, the content of total p-AMPKα increased in cells treated with azide or AICAR (Fig. 8). Immunoprecipitates of total p-AMPKα contained LKB1, and the abundance of LKB1 was higher in immunoprecipitation products from azide- and AICAR-treated cells (Fig. 8). Similarly, immunoprecipitates of LKB1 assayed for the presence of total p-AMPKα showed significant amounts of total p-AMPKα in the immunoprecipitation products from azide- and AICAR-treated cells (Fig. 8). Interestingly, virtually no total p-AMPKα was present in immunoprecipitates of LKB1 in control unstimulated cells.
JNK has been identified as an upstream component of the AMPK signaling cascade in glucose-deprived DU145 prostate carcinoma cells (30). To determine whether activation of JNK leads to phosphorylation of AMPK in Clone 9 cells, we employed anisomycin to stimulate JNK phosphorylation and activity (3). Cells were treated with diluent, 10 μM anisomycin, or 2 mM AICAR for 1 h, and lysates were assayed for p-JNK, total p-AMPKα, p-ACC, and β-actin (Fig. 9). AICAR stimulated the phosphorylation of AMPK and ACC, as expected. Although anisomycin increased p-JNK, as expected, there was no discernible increase in total p-AMPKα or p-ACC. Moreover, treatment with AICAR did not increase phosphorylation of JNK.
Under conditions of metabolic stress, such as inhibition of oxidative phosphorylation with reduced energy production, glycolytic ATP synthesis becomes the principal energy-producing pathway. This is especially true in cells in which glucose transport is rate limiting for glucose metabolism, where stimulation of glucose transport becomes a critical component of the adaptive response (12, 14). Despite the physiological importance of this general response, the molecular mechanisms underlying the regulation of glucose transport under these conditions are poorly understood.
Results of our previous studies have shown that exposure to hypoxia, cyanide, or azide causes a sharp decrease in cellular ATP concentration and an increase in ADP and AMP concentrations within 15 min of treatment, with the nucleotide concentrations and ratios returning to normal or near-normal values by 1 h, despite the continued presence of the inhibitory conditions (13, 14, 25). As predicted by the increase in the AMP-to-ATP ratio, AMPK becomes phosphorylated and activated under the above-mentioned conditions (1, 13). However, whether the stimulation of AMPK activity that is observed in conjunction with the enhanced glucose transport plays a critical role, a permissive role, or no role in the stimulation of glucose transport is not known.
In the present study, we determined whether stimulation of AMPK in response to inhibition of oxidative phosphorylation is both sufficient and necessary for the increased rate of glucose transport. Although previous studies by us and others have shown that stimulation of AMPK is associated with an increase in the rate of glucose transport (i.e., establishing sufficiency of the premise) (1, 4, 31), the effect of inhibition of AMPK on the glucose transport response to inhibition of oxidative phosphorylation has not been reported. Hence, the necessity of the stimulation of AMPK in the glucose transport response to inhibition of oxidative phosphorylation has not been established. Here, we utilized two independent approaches to suppress AMPK activity to determine the effect of such inhibition on the glucose transport response to azide. Throughout these studies, we have employed AICAR as a positive control. We also tested for a potential additive effect of azide and AICAR on AMPK activation and on the rate of glucose transport. The results reported here place stimulation of AMPK activity directly in the pathway of the glucose transport response to inhibition of oxidative phosphorylation, signifying that the stimulation of AMPK is a necessary step in the glucose transport response.
Mammalian AMPK is a heterotrimeric enzyme consisting of a catalytic α-subunit and regulatory β- and γ-subunits (21). The α-subunit is expressed as two isoforms (α1 and α2), and both contain a threonine at position 172, which, on phosphorylation, leads to a ∼50- to 100-fold increase in the enzyme's activity (29). The two isoforms are differentially expressed in various tissues. AMPK complexes containing the α2-isoform predominate in skeletal and cardiac muscle (26), whereas approximately equal levels of the two isoforms are expressed in liver (27). In contrast, β-cells in pancreatic islets largely express the α1-isoform (20). In Clone 9 cells, which were used in the present study, we previously detected both α-isoforms (1). Further detailed studies by Barnes et al. (4) utilizing immunoprecipitation of each AMPK in Clone 9 cells followed by measurement of enzyme activity in the immunoprecipitates showed that AMPKα1 is responsible for the bulk of the enzyme's activity in these cells.
To investigate the potential mediating and, perhaps, critical role of AMPK in the stimulation of glucose transport in response to inhibition of oxidative phosphorylation, we employed the following two independent strategies: the first involved suppression of AMPKα1 and AMPKα2 expression (alone and in combination) using siRNA technology, and the second employed compound C as an inhibitor of AMPK (19). Both strategies would be predicted to result in a significant suppression of the glucose transport response if the stimulation of AMPK activity plays an important role in mediating the response. Use of siRNA directed against AMPKα2 resulted in a large decrease in AMPKα2 abundance. However, on stimulation of these cells with AICAR, no significant decrease in total p-AMPKα was observed, suggesting that AMPKα2 may be a minor component of total AMPK expressed in these cells. This inference is consistent with the findings of Barnes et al. (4) in reference to the expression and function of AMPK isoforms in Clone 9 cells. We then examined whether the azide-stimulated glucose transport was affected in cells with suppressed AMPKα2 content. The results showed that neither azide- nor AICAR-induced stimulation was affected by siAMPKα2. In addition, total AMPKα and p-ACC contents in response to azide (and AICAR) were not affected.
In contrast were the results following the use of siRNA directed against AMPKα1. Treatment of control cells with siAMPKα1 slightly decreased basal transport, perhaps due to the dependence of basal glucose transport on AMPK activity. Treatment with siAMPKα1 resulted in a dramatic suppression of azide-stimulated glucose transport. Western blot analysis showed that the abundance of AMPKα1 was significantly depressed in siRNA-treated cells, as was the abundance of total AMPKα, as well as the content of total p-AMPKα, in response to azide (and AICAR), verifying that the α1-isoform is the predominant form of the enzyme in these cells (4). The content of p-ACC in response to azide (and AICAR) was significantly depressed in cells exposed to siAMPKα1. The partial suppression of glucose transport in response to azide could be interpreted to mean that AMPK plays a significant, but partial, role in the response or, alternatively, that siRNA-induced suppression of AMPK was incomplete. The finding that AICAR-stimulated glucose transport was also partially suppressed strengthens the latter possibility.
Next, we used siRNAs against both AMPK α-isoforms in combination. In these experiments, 50 pmol/ml of each siRNA was used to limit cell toxicity. The effect of the mixture of the siRNAs on azide-stimulated glucose transport was similar to that seen with use of siAMPKα1 alone, a finding that would be expected if the α2-isoform were a minor component of total AMPK activity. Western blot analysis showed a dramatic (but incomplete) suppression of total cellular AMPKα content and total p-AMPKα abundance in response to azide and AICAR. Importantly, the content of p-ACC was markedly depressed in cells exposed to azide (and AICAR), consistent with great suppression of total AMPKα activity.
We also used a complementary strategy to define the mediating role of AMPK activation in azide- and hypoxia-stimulated glucose transport. Compound C has been reported to be a “selective” inhibitor of AMPK, in that it prevents phosphorylation of the α-subunit by competing with AMP binding to the γ-subunit of the enzyme (19). Use of this reagent resulted in a near-complete suppression of the total p-AMPKα content of cells treated with AICAR used as a positive control. Similarly, the content of total p-AMPKα in azide-treated cells and in hypoxia-exposed cells was suppressed to low levels by compound C, whereas the cellular content of total AMPKα remained unaltered. Under these conditions, we observed a near-complete suppression of the stimulation of glucose transport in response to azide, hypoxia, and the positive control AICAR. In contrast to the glucose transport response following exposure to azide, cyanide, or hypoxia, which occurs within ∼10 min of exposure, stimulation of the hypoxia-responsive pathway per se by CoCl2 (which also induces Glut1 mRNA expression and, subsequently, results in increased Glut1 expression and stimulation of glucose transport) occurs after a delay of ∼4 h (11, 14, 24). Hence, the stimulation of glucose transport in response to hypoxia at 2 h and its suppression by compound C reflect the inhibition of oxidative phosphorylation and stimulation of AMPK, leading to stimulation of glucose transport with the constant cellular Glut1 content. On the basis of the results derived from the siRNA and compound C experiments, we conclude that stimulation of AMPK activity plays a critical role in the stimulation of glucose transport in response to inhibition of oxidative phosphorylation.
The above-described results strongly suggest that stimulation of AMPK is necessary for stimulation of glucose transport in response to inhibition of oxidative phosphorylation. To extend the accepted premise that stimulation of AMPK is sufficient for the glucose transport response to azide, we measured the effect of simultaneous exposure to azide and AICAR. If azide stimulates glucose transport by pathways that are entirely, or partly, independent of AMPK, then addition of azide to cells exposed to AICAR should further augment the rate of transport observed in response to AICAR alone. We found no evidence for any additive effect of azide to the stimulation of glucose transport in response to AICAR. The total p-AMPKα content was highest in cells treated with both agents, whereas the transport response was not further enhanced. This finding, which has not been explained, suggests that activation of AMPK stimulates the rate of glucose transport only to a certain limiting extent. Nevertheless, the above-described results strongly suggest that activation of AMPK is associated with azide-stimulated glucose transport.
Having found that stimulation of AMPK is critical for the glucose transport response to azide (and AICAR), we explored the possibility that Glut1 becomes phosphorylated by the direct action of AMPK. Such a modification could itself mediate the presumed activation of Glut1 and stimulate glucose transport. This possibility gains credence because of the presence of two potential consensus phosphorylation sequences for AMPK in the intracellular segments of Glut1 at Ser95 and Ser226 that are conserved in rat, human, and mouse (7, 17, 18). We thus performed assays for the presence of phosphoserine in Glut1 immunoprecipitates from control cells and cells treated with azide or AICAR. No phosphoserine was detected in Glut1 immunoprecipitates under any of the above-described conditions. However, although a number of phosphoserine-containing protein bands were present in Glut1 immunoprecipitates, their abundance showed no differential change in control vs. treated cells. Identification of these proteins could be a subject of future studies. We also did not find any total p-AMPKα in Glut1 immunoprecipitates, nor did we find any Glut1 in total p-AMPKα immunoprecipitates. On the basis of these findings, we conclude that AMPK probably does not directly phosphorylate Glut1 and that the action of AMPK to stimulate glucose transport is probably indirect and mediated by unknown intermediary steps. Also the increase in Glut1 expression and content in response to azide is not observed before 4 h of exposure to the inhibitor (24, 25).
We also examined the potential role of LKB1 and JNK in the phosphorylation and activation of AMPK in response to azide and AICAR. LKB1 has been identified as an upstream kinase that phosphorylates the α-subunit of AMPK at Thr172, especially on binding of 5′-AMP to the γ-subunit of AMPK (29). LKB1 is one of the two described modes of activation of AMPK, the other being mediated by increased cellular calcium concentration and activation of calmodulin-dependent protein kinase kinase (22). It is possible that other means of activation of AMPK exist, and these potential mechanisms are under investigation (9). LKB1 has been identified as the site of action of metformin, a medication used in the treatment of type 2 diabetes (22). We found that LKB1 was immunoprecipitated in association with total p-AMPKα, and vice versa. Interestingly, the degree of association between the two proteins was augmented in cells treated with AICAR (as expected), as well as in cells treated with azide. The latter finding suggests, but does not prove, that the stimulation of AMPK and its phosphorylation in azide-treated cells are mediated by LKB1. For example, we previously reported that the intracellular free calcium concentration increases after inhibition of oxidative phosphorylation (16). We additionally examined the potential activation of AMPK in response to phosphorylation and activation of JNK. This latter possibility was examined because of a previous report that, in glucose-deprived DU145 prostate carcinoma cells, JNK phosphorylation and activation were associated with phosphorylation of AMPK (30). We found that JNK phosphorylation in response to anisomycin did not result in phosphorylation of AMPK and that treatment with AICAR was not associated with phosphorylation of JNK in Clone 9 cells. The divergent results probably reflect the use of different cells and stimuli. Our results suggest that LKB1 might play a mediating role in the azide-induced activation of AMPK. Our previous results showed that intracellular free calcium concentration rises in azide-treated cells (16). This raises the possibility that activation of calmodulin-dependent protein kinase kinase (see above) might also play a role in the stimulation of AMPK in response to azide. Further work is necessary to better define the upstream pathway(s) leading to stimulation of AMPK in response to inhibition of oxidative phosphorylation.
Finally, the steps and sequence of events that mediate the stimulation of glucose transport in response to activation of AMPK remain unknown. Our finding that AMPK plays a critical role in the stimulation of glucose transport in response to inhibition of oxidative phosphorylation implies the involvement of downstream phosphorylation events, kinases, and, perhaps, phosphatases in the response. Our results also demonstrate that AMPK is not associated with Glut1, nor is Glut1 itself phosphorylated in response to exposure to azide or after stimulation of AMPK activity. Further work is necessary to delineate the sequence of events to identify the mediating mechanisms underlying the stimulation of glucose transport in response to inhibition of oxidative phosphorylation.
This study was supported by National Institute of Diabetes and Digestive and Kidney Disease Grant RO1-DK-61994.
We thank Li Song for excellent technical support.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society