Short-Term Overexpression of a Constitutively Active Form of AMP-Activated Protein Kinase in the Liver Leads to Mild Hypoglycemia and Fatty Liver

AMPKα2-CA was created by truncating a full-length myc epitope–tagged rat AMPKα2 cDNA (a gift from D. Carling, London, U.K.) at residue 312, as described for the AMPKα1 isoform (11). AMPKα2-CA was amplified by PCR then subcloned into the pAdTrack shuttle vector, and recombinant adenoviral vector was obtained as previously described (12). Ad expressing β-galactosidase was used as a control. Ads were propagated in AD-293 cells and purified by cesium chloride density centrifugation as previously described (13).

Male C57BL/6j and C57BL/6j ob/ob mice (9 weeks old) were maintained under a 12-h light/12-h dark cycle with free access to water and standard mouse diet. Streptozotocin (STZ)-induced diabetic mice were obtained by two intraperitoneal injections of 125 mg/kg STZ 24 h apart. Five days after the first injection, mice with blood glucose levels up to 280 mg/dl were studied further. Animals were killed 48 h postinfection under ketamine/xylazine anesthesia. Tissues were dissected immediately, frozen in liquid nitrogen, and stored at −80°C until analysis. All procedures were performed in accordance with the principles and guidelines established by the European Convention for the Protection of Laboratory Animals.

Mice were anesthetized with isoflurane, and Ad β-gal or Ad AMPKα2-CA (1 × 10⁹ plaque-forming units) was injected into the penis vein. After Ad injection, C57BL/6j mice were divided into two groups: fasted and refed. The fasted group was fed standard diet ad libitum for 24 h then fasted for 24 h. The refed group had free access to food for 8 h after injection. They were then fasted for 24 h then refed a high-carbohydrate/low-fat (HC) diet (75% wt/wt sucrose, 15% wt/wt protein, and 3% wt/wt fat) for 16 h. Mice were killed 48 h after Ad injection. Tissue analyses and biochemical analyses of plasma were carried out as previously described (14). Plasma glucagon concentration was assessed by radioimmunoassay (Linco).

Hepatocytes were isolated from fed male C57BL/6j mice by a modified version of the collagenase method and were transfected with 10 plaque-forming units/cell of Ad β-gal or Ad AMPKα2-CA (13). Untransfected hepatocytes were treated for 6 h with 500 μmol/l AICAR (Toronto Research Chemicals). For hepatic glucose production, cells were incubated with glucose-free culture medium containing 100 μmol/l Bt₂-cAMP, 10 mmol/l lactate, and 1 mmol/l pyruvate. We collected culture medium at 4, 8, and 12 h to determine the amount of glucose released. Glucose levels were determined by evaluating the production of NADPH from NADP⁺ in the presence of hexokinase and glucose-6-phosphate dehydrogenase (Roche).

Total protein extracts from hepatocytes and liver samples were obtained as previously described (15). We subjected 50 μg of protein to SDS-PAGE and transferred the resulting bands to nitrocellulose membrane. Blots were probed with antibodies against AMPKα1, AMPKα2 (a gift from G. Hardie, Dundee, U.K.), pan-AMPKα, phospho-Thr172-AMPKα, phospho-Ser79–acetyl-CoA carboxylase (ACC) (Cell Signaling), myc tag (clone 9E10; Sigma), and β-galactosidase (Rockland). ACC was detected with streptavidin–horseradish peroxidase (Dako). Total AMPK activity was measured in polyethylene glycol protein precipitate from liver by phosphorylation of the AMARA peptide as previously described (15).

Total RNA was isolated with RNA+ (Qbiogene), and single-strand cDNA was synthesized from 5 μg of total RNA with random hexamer primers and Superscript II (Invitrogen). Real-time RT-PCRs were carried out with a LightCycler reaction kit (Eurogenetec) in a final volume of 20 μl containing 250 ng of reverse-transcribed total RNA, 500 nmol/l of primers (Table 1), 10 μl of 2 × PCR mix, and 0.5 μl of Sybr Green. The reactions were carried out in capillaries in a LightCycler instrument (Roche) with 40 cycles. We determined the relative amounts of the mRNAs studied by means of the second-derivative maximum method, with LightCycler analysis software version 3.5 and 18S RNA as the invariant control for all studies.

Results are expressed as means ± SE. We used Student’s t test for unpaired data. Differences were considered significant if P < 0.05.

Cellular AMPK activity was increased by transfection with an adenoviral vector encoding a constitutively active form of AMPKα2 (Ad AMPKα2-CA). This protein is a truncated form of AMPKα2, lacking the autoinhibitory domain and the region interacting with the regulatory subunits β/γ (Fig. 1A). The infection of primary cultures of mouse hepatocytes with Ad AMPKα2-CA led to strong expression of myc epitope–tagged AMPKα2-CA of the expected molecular weight and no change in endogenous AMPKα expression (Fig. 1B). The increase in AMPK activity in hepatocytes resulted in a fourfold increase of ACC phosphorylation (Fig. 1B). The magnitude of this increase is similar to that observed in hepatocytes incubated with AICAR, a known pharmacological activator of AMPK. Incubation of the cells with AICAR increased the amount of the active Thr-172 phosphorylated form of endogenous AMPK. Anti–phospho-AMPK antibodies also recognized the AMPKα2-CA protein, suggesting that the truncated AMPKα2 form is phosphorylated, accounting for its constitutive activity. The pharmacological activation of AMPK has been shown to decrease the level of expression of gluconeogenic genes such as those encoding PEPCK and glucose-6-phosphatase (G6Pase) in hepatoma cells (16,17). The induction of PEPCK and G6Pase gene transcription by Bt₂-cAMP was inhibited by the infection of hepatocytes with Ad/AMPKα2-CA (Fig. 1C). Consequently, AMPKα2-CA expression decreased cumulative glucose production by 30% in primary cultures of mouse hepatocytes stimulated for 12 h with Bt₂-cAMP (Fig. 1D). Thus, glucose production in hepatocytes is directly inhibited by activated AMPK.

The injection of Ad AMPKα2-CA led to expression exclusively in the liver (Fig. 2A). We checked for the presence of AMPKα2-CA or β-galactosidase protein in the liver of infected mice by Western blotting (Fig. 2B), and the efficiency of gene transfer was assessed by determining the number of cells positive (>80%) for AMPKα2-CA and β-gal expression (data not shown). Ad AMPKα2-CA infection resulted in levels of AMPK activity twice those observed following Ad β-gal infection (Fig. 2C), with no significant change in endogenous AMPKα1 and -α2 protein levels (Fig. 2B).

We investigated the impact of hepatic AMPKα2-CA on glucose metabolism in vivo by injecting Ad β-gal or Ad AMPKα2-CA into C57BL/6j mice and determining plasma glucose concentration. Ad AMPKα2-CA injection triggered a decrease in plasma glucose concentration in the fasted and fed states, 24 and 48 h after infection (Fig. 3A). This effect on blood glucose levels was associated with a physiological response of the pancreatic hormones, a decrease in plasma insulin, and an increase in plasma glucagon concentrations (Table 2). The expression of genes encoding proteins involved in gluconeogenesis, such as PEPCK and G6Pase, in Ad AMPKα2-CA–infected mice was decreased by nearly twofold relative to Ad β-gal–infected mice (Fig. 3B). Similar experiments were conducted in insulin-resistant ob/ob mice. Twenty-four hours after infection with Ad AMPKα2-CA, blood glucose concentrations had decreased significantly (Fig. 3C), and plasma insulin levels were lower than those in mice infected with Ad β-gal (10,914 ± 473 pg/ml in Ad β-gal vs. 6,774 ± 1,112 pg/ml in Ad AMPKα2-CA, n = 5, P < 0.05). Finally, Ad AMPKα2-CA was injected into STZ-induced diabetic mice to assess whether AMPKα2-CA expression was also effective in a murine model of type 1 diabetes. Forty-eight hours after infection with Ad AMPKα2-CA, blood glucose levels had decreased significantly (Fig. 3E), and plasma insulin levels remained low (42 ± 10 pg/ml in Ad β-gal vs. 41 ± 8 pg/ml in Ad AMPKα2-CA, n = 5, NS). Both diabetic ob/ob and STZ-treated mice infected with Ad AMPKα2-CA displayed reduced expression levels for the PEPCK and G6Pase genes (Fig. 3D and F). Thus, the activation of AMPK in the liver was sufficient to normalize glucose homeostasis not only in insulin-resistant diabetic mice but also in severe insulinopenic mouse models by reducing the expression of genes encoding the key enzymes of gluconeogenesis.

AMPK has been implicated in the regulation of hepatic gene transcription (15,16,18–20). We therefore thought it likely that AMPKα2-CA expression would modify the pattern of gene expression in the livers of infected mice. The expression of genes encoding proteins involved in glycolysis and lipid synthesis and their upstream regulators, SREBP-1 (sterol regulatory element–binding protein-1) and ChREBP (carbohydrate response element–binding protein), was induced during the transition between fasting and refeeding in control Ad β-gal–infected mice (Fig. 4). In contrast, the activation of SREBP-1 and ChREBP was considerably reduced in refed AMPKα2-CA–infected mice. This resulted in a lower level of transcriptional activation of downstream target genes, including the genes encoding GLUT2, GK, L-PK, ACC, FAS, S14, and GPAT in response to refeeding (Fig. 4). Protein levels for GK, ACC, and GLUT2 were low, as shown by Western blotting (Fig. 5E and online appendix [available at http://diabetes.diabetesjournals.org]). Interestingly, the expression of SCD-1, a leptin target in the liver (21), was strongly inhibited in both fasted and refed animals (Fig. 4). The genes encoding the LDLr and HMG-CoA synthase, involved in cholesterol uptake and synthesis, were also downregulated. Finally, the abundance of HKII and GAPDH mRNAs was higher in fasted and refed animals. The activation of glycolytic and lipogenic genes in response to refeeding was decreased by AMPKα2-CA and could explain the twofold decrease of hepatic glycogen content in the liver of refed mice and the fourfold decrease in plasma triglyceride concentrations (Table 2). In contrast, plasma total cholesterol and HDL cholesterol concentrations were unaffected (Table 2).

Fasting is known to stimulate the release of free fatty acids (FFAs) from adipose tissue, triggering their uptake and oxidation by the liver. The livers of Ad β-gal–infected normal mice contained more triglycerides after 24 h of fasting than during the fed period (Table 2). Lipid deposition, as indicated by oil red O staining, was concentrated in the area around the portal vein (Fig. 5A). The livers of Ad AMPKα2-CA–infected mice contained three times the amount of triglycerides present in Ad β-gal–infected normal mice (Table 2), with diffuse lipid droplets throughout the liver (Fig. 5B). Liver cholesterol content was also significantly higher in these mice (Table 2). Consistent with this fatty liver, AMPKα2-CA–infected mice had significantly heavier livers than Ad β-gal–infected normal mice in fasting conditions (Table 2). We investigated this fat accumulation further by monitoring expression of hepatic lipoprotein lipase (LPL). LPL hydrolyzes the triglycerides present in circulating plasma lipoproteins, and the FFAs and monoglycerides released are taken up by tissues for reesterification (for storage) or use as a metabolic fuel. LPL is weakly expressed in the liver but is upregulated in conditions of low carbohydrate availability (22). LPL gene expression was strongly increased by Ad AMPKα2-CA infection (Fig. 5C), probably accounting for lipid uptake by the liver and the decrease in plasma triglyceride concentration. A significant decrease in epididymal fat pad weight and body weight was observed after Ad AMPKα2-CA infection (Table 2), probably due to an increase in fat mobilization from the adipose tissue to the liver. This was expected to result in an increase in plasma FFA and glycerol levels in response to the increase in adipose tissue lipolysis. However, plasma FFA and glycerol levels were slightly lower in fasted Ad AMPKα2-CA–infected animals (Table 2), suggesting that the FFA released by adipose tissue were rapidly utilized by peripheral tissues.

We investigated the effect of Ad AMPKα2-CA infection on lipid utilization in the liver by assessing ketone bodies production, using the surrogate measure of plasma β-hydroxybutyrate concentration. A twofold increase in plasma β-hydroxybutyrate concentration was observed in the fasted state following infection with Ad AMPKα2-CA (Table 2), suggesting that β-oxidation was activated in these animals. Among the genes encoding proteins involved in fatty acid oxidation and ketogenesis, those encoding CPT-1, MCAD, AOX, ketothiolase, and HMG-CoA lysase showed no change or a slight decrease in expression following AMPKα2-CA expression (Fig. 5D). The rate-limiting step in the β-oxidation of fatty acids is the transport of fatty acyl-CoA into the mitochondrion by the CPT-1 shuttle system. The key allosteric inhibitor of CPT-1 is malonyl-CoA, which is synthesized from acetyl-CoA by ACC. The levels of ACC protein in the liver of Ad AMPKα2-CA–infected mice were significantly lower than those in normal mice (Fig. 5E). The resulting decrease in malonyl-CoA levels abolishes inhibition of the rate-limiting transport of fatty acyl-CoA into the mitochondrion for β-oxidation. Finally, levels of uncoupling protein 2 (UCP-2) mRNA were higher in Ad AMPKα2-CA–infected animals than in normal mice, suggesting that fatty acids may be used to produce heat by uncoupling electron transport from oxidative phosphorylation (Fig. 5F).

An important hallmark of diabetes is hyperglycemia, which is associated with an increase in hepatic glucose production. In this study, we showed that short-term hepatic expression of a constitutively active form of AMPKα2 leads to mild hypoglycemia in normal mice and abolishes hyperglycemia in diabetic ob/ob and STZ-induced diabetic mice. These results demonstrate that AMPK activation may have an effect under conditions of severe insulinoresistance or insulinopenia and that insulin signaling is therefore not required for this effect. The hypoglycemic effect of AMPKα2-CA is consistent with the abolition of endogenous glucose production, as suggested by the weaker expression of the gluconeogenic genes PEPCK and G6Pase in AMPKα2-CA–infected animals. As hepatic glucose production is the primary mechanism regulating glucose flux in the basal state, a decrease in blood glucose levels in the absence of a concomitant rise in basal insulin levels indicates that hepatic AMPKα2-CA expression probably decreases hepatic glucose production. This correlates with the inhibitory effect of Ad AMPKα2-CA on glucose output in primary mouse hepatocytes. Moreover, acute exposure to AICAR in vivo has been shown to decrease endogenous glucose production (8). Previous reports (7,23–25) have also reported that AICAR decreases blood glucose levels in various rodent models of type 2 diabetes. However, the relative roles played by skeletal muscle and liver in mediation of the hypoglycemic effect of AICAR remain unclear. In muscle-specific AMPK dominant-negative mice, AICAR has a weaker hypoglycemic effect than in control animals (26), suggesting that this compound lowers glycemia at least partly by increasing muscle glucose uptake. Mice with a liver-specific AMPKα knockout (KO) also display an impaired hypoglycemic response to AICAR, indicating that the liver is involved in the hypoglycemic effect of AICAR in vivo (B.V., F.A., unpublished results). The importance of AMPK in the control of glucose output by the liver was recently highlighted by the potent effects of circulating adipocyte-derived hormones on whole-body glucose metabolism. A physiological link has been established between circulating resistin levels and hepatic AMPK activity in the maintenance of blood glucose (4). Short-term adiponectin infusion is known to inhibit both the expression of genes encoding hepatic gluconeogenic enzymes and the rate of endogenous glucose production (27). Similarly, the glucose-lowering effect of metformin has been attributed to a decrease in hepatic glucose production due to a decrease in gluconeogenesis (28). Our results therefore confirm the physiological importance of hepatic AMPK for whole-body glucose homeostasis.

The fatty liver caused by short-term hepatic expression of AMPKα2-CA was unexpected. Fat accumulation in the liver may be multifactorial, involving an increase in uptake from peripheral fat stores and/or an increase in de novo fatty acid synthesis by the hepatocytes. However, de novo synthesis is unlikely because lipogenic gene expression was found to be repressed. Lipid uptake from peripheral tissue is demonstrated by the decrease in adipose tissue mass and the upregulation of LPL gene expression. LPL is normally weakly expressed in adult liver and may be induced by limited carbohydrate availability, as occurs dur-ing the neonatal period (22). In this situation, LPL expression facilitates the shunting of circulating triglycerides to the liver and the generation of ketone bodies, which can supply energy to peripheral tissues such as the brain and muscle, replacing glucose. Thus, by routing triglycerides to the liver, Ad AMPKα2-CA–infected mice trigger the mobilization of lipids for energy, thereby decreasing the metabolic stress imposed by the scarcity of glucose. Furthermore, the decrease in FFA levels and the increase in β-hydroxybutyrate levels in the plasma of Ad AMPKα2-CA–infected mice suggest that the FFA released by the adipose tissue is rapidly used by the liver. Interestingly, Ad AMPKα2-CA–infected mice had lower fat masses and higher liver weights, reflecting a redistribution of body fat from the adipose tissue to the liver. This finding is consistent with previous reports on long-term AICAR treatment, demonstrating a decrease in abdominal fat mass in normal and obese Zucker rats (25,29). SCD-1 deficiency, which leads to the activation of AMPK, also results in a decrease in adiposity (30). Interestingly, infection with Ad AMPKα2-CA led to the repression of SCD-1 expression in both fasted and refed conditions. A similar decreased body fat phenotype was also observed in ACC2 KO mice, in which fatty acid oxidation was increased, indicating that high levels of mitochondrial fatty acid oxidation can reduce fat storage in adipose tissue (31). Thus, in Ad AMPKα2-CA–infected mice, there seems to be a compensatory increase in the use of released FFA for activated β-oxidation. This results in a switch from glucose to fatty acids for energy production. This switch probably involves an increase in basal lipolysis activity due to high plasma concentrations of glucagon and low plasma concentrations of insulin in Ad AMPKα2-CA–infected animals. AMPKα2-CA expression upregulated UCP-2 gene expression in the livers of fasted animals. Thus, the UCP-2 protein may be involved in adapting lipid metabolism to an excessive supply of fatty acids for the control of energy status (32). This increase in UCP-2 levels should support higher rates of β-oxidation and may increase fat degradation and the use of fatty acids for energy production. The long-term expression of AMPKα2-CA may therefore be required to achieve a significant and consistent decline in liver triglyceride content. Whether chronic AMPKα2-CA expression would lead to an increase in energy expenditure and resistance to diet-induced obesity is now under investigation.

In conclusion, the short-term overexpression of AMPKα2-CA in the liver results in a metabolic switch from glucose to lipid metabolism (Fig. 6). The lower plasma glucose concentrations in Ad AMPKα2-CA–infected mice lead to an increase in hepatic lipid utilization, resulting in a decrease in white adipose mass. The concomitant accumulation of hepatic triglycerides leads to the generation of ketone bodies, which are required as alternative substrates to supply energy to peripheral tissues in conditions of low glucose availability. The data reported here indicate that activation of hepatic AMPK is sufficient for controlling hyperglycemia in murine models of diabetes and could lead to decreased adiposity.

This work was supported by INSERM (Institut National de la Santé et de la Recherche Medicale), the French Ministry for Research, the European Commission (Grant QLG1-CT-2001-01488), ALFEDIAM, and Institut Benjamin Delessert. M.F. was supported by postdoctoral fellowships from the European Commission and GlaxoSmithKline.

We thank Véronique Fauveau, Wanda Dolci, and Myriam Bennoun for technical assistance. We also thank Catherine Postic and Michèle Cauzac for providing reagents.