AMP-activated Protein Kinase α2 Activity Is Not Essential for Contraction- and Hyperosmolarity-induced Glucose Transport in Skeletal Muscle*

To examine the role of AMP-activated protein kinase (AMPK) in muscle glucose transport, we generated muscle-specific transgenic mice (TG) carrying cDNAs of inactive α2 (α2i TG) and α1 (α1i TG) catalytic subunits. Extensor digitorum longus (EDL) muscles from wild type and TG mice were isolated and subjected to a series of in vitro incubation experiments. In α2i TG mice basal α2 activity was barely detectable, whereas basal α1 activity was only partially reduced. Known AMPK stimuli including 5-aminoimidazole-4-carboxamide-1-β-4-ribofuranoside (AICAR), rotenone (a Complex I inhibitor), dinitrophenol (a mitochondrial uncoupler), muscle contraction, and sorbitol (producing hyperosmolar shock) did not increase AMPK α2 activity in α2i TG mice, whereas α1 activation was attenuated by only 30–50%. Glucose transport was measured in vitro using isolated EDL muscles from α2i TG mice. AICAR- and rotenone-stimulated glucose transport was fully inhibited in α2i TG mice; however, the lack of AMPK α2 activity had no effect on contraction- or sorbitol-induced glucose transport. Similar to these observations in vitro, contraction-stimulated glucose transport, assessed in vivo by 2-deoxy-d-[3H]glucose incorporation into EDL, tibialis anterior, and gastrocnemius muscles, was normal in α2i TG mice. Thus, AMPK α2 activation is essential for some, but not all, insulin-independent glucose transport. Muscle contraction- and hyperosmolarity-induced glucose transport may be regulated by a redundant mechanism in which AMPK α2 is one of multiple signaling pathways.

AMPK by pharmacological stimulation and transient expression of an AMPK-active mutant increases glucose transport (3)(4)(5)(6). AMPK also seems to play a role in enhancing muscle (7) and whole body (8) insulin sensitivity and responsiveness for glucose transport. Because skeletal muscle accounts for ϳ80% of disposal of an oral glucose load (9,10) and because type 2 diabetes is associated with reduced muscle glucose disposal (11), AMPK may be critical in the control of metabolic homeostasis and perhaps the development of type 2 diabetes (12,13). Not surprisingly, AMPK is now considered a drug target for the treatment of type 2 diabetes (14).
It has long been believed that there are two major signaling mechanisms leading to glucose transport stimulation in skeletal muscle. One mechanism is insulin-activated signaling through the insulin receptor, insulin receptor substrate (IRS), and phosphatidylinositol 3-kinase. The other is insulin-independent signaling for stimuli such as exercise, hyperosmolarity, mitochondrial uncoupling, and hypoxia (23), which recent reports suggest to be regulated by AMPK. Initial evidence in support of a role for AMPK in muscle glucose transport came from studies using 5-aminoimidazole-4-carboxamide riboside (AICAR), a compound that is taken up into skeletal muscle and metabolized to ZMP, an analog of AMP. It was first shown that AICAR infusion enhances insulin-stimulated glucose transport in perfused rat hindlimb skeletal muscles (3). After this report, our group (4) and Bergeron et al. (5) showed that AICAR directly stimulates glucose transport in the absence of insulin in isolated rat muscle. Similar to contraction-induced glucose transport, AICAR-stimulated glucose transport was not affected by inhibition of phosphatidylinositol 3-kinase (4,5). In addition to studies of muscle contraction, AMPK activation closely correlates with increased glucose transport in isolated rat muscle in response to other insulin-independent stimuli such as hypoxia, inhibition of mitochondrial respiration, and hyperosmolarity (24). Similar to contraction, these stimuli also reduce cellular ATP, phosphocreatine, and glycogen levels (4). A potential role for AMPK in glucose transport induced by hyperosmolarity (6,25,26) and by reduction of ATP generation by mitochondria (25,27) has also been suggested based on findings in cultured muscle cells. Therefore, AMPK may be a key molecule that is responsible for insulin-independent glucose transport caused by cellular stress in skeletal muscle. * This work was supported by National Institute of Health Grants RO1 AR45670 and AR42238 (to L. J. G.), Individual Kirschstein National Research Service Award F32 AR049662 (to R. C. H.), and Diabetes Endocrinology Research Center Grant DK36836 (to the Joslin Diabetes Center). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  Consistent with the hypothesis that AMPK regulates glucose transport in rat skeletal muscle, transgenic mice that express a muscle-specific inactive mutant of AMPK␣ that reportedly has a dominant inhibitory effect have full inhibition of AICAR-and hypoxia-stimulated glucose transport, whereas contraction-stimulated glucose transport was reduced by 30 -40% (28). In contrast, Jorgensen et al. (29) showed that contraction-induced glucose transport in isolated muscles was not altered in either ␣1 or ␣2 whole body AMPK knock-out mice. These reports have led us to reconsider the role of the two catalytic subunits of AMPK in glucose transport regulation by muscle contraction and other non-insulin stimuli. For this purpose, we generated three types of muscle-specific transgenics; they are 1) mice expressing an inactive form of the ␣1 catalytic subunit isoform (␣1i TG mice), 2) mice expressing an inactive form of ␣2 (␣2i TG mice), and 3) a cross of ␣1i TG and ␣2i TG (␣1/2i TG mice). We show that although AMPK activity is necessary for some insulin-independent glucose transport (AICAR and mitochondrial respiratory Complex I inhibition), glucose transport induced by contraction and hyperosmolarity does not require active ␣2 protein. Therefore, we conclude that at least three distinct or overlapping intracellular signaling pathways (i.e. an insulin-dependent, an AMPK-dependent, and an insulin/AMPK-independent pathway) are present in murine skeletal muscle for the regulation of glucose transport.

EXPERIMENTAL PROCEDURES
Generation of Transgenic Mice-To render the catalytic subunit inactive, the aspartic acid at amino acid residue 157 of rat AMPK ␣2 subunit was substituted to alanine by a PCR-based site direct mutagenesis as described by Stein et al. (30). Complementary primers spanning the residue to be mutated (30) were synthesized and used for the mutagenesis. Mice expressing the inactive ␣2 tagged at the amino terminus with a hemagglutinin epitope (␣2i TG mice) were generated by injecting the recombinant DNA driven by a muscle creatine kinase promoter (a gift from Drs. Kohjiro Ueki and C. Ronald Kahn, Joslin Diabetes Center, Boston, MA) into fertilized FVB mouse oocytes at the Brigham and Women's Hospital Transgenic Mouse Facility (Boston, MA). Three ␣2i TG lines were confirmed, and the founders were bred to FVB mice. A similar strategy was used with the goal of generating mice that express an inactive ␣1 in skeletal muscle (␣1i TG mice). Rat AMPK ␣1 cDNA, in which the lysine at amino acid residue 45 was changed to arginine to render the catalytic subunit inactive (31), was tagged at the amino terminus with the Myc epitope. Six ␣1i TG lines were confirmed, and the founders were bred to FVB mice. Transgenic mouse founders were identified by polymerase chain reaction-based methods, and transgene product expression was confirmed by immunoblots with an antibody for each tagged epitope. Mice expressing both transgenes were generated by breeding an ␣2i TG line to an ␣1i TG line (␣1/2i TG mice). Ten-to-sixteen-week-old mice from F2 and F3 generations were used for experiments. Comparisons of the transgenic mice were made against littermate wild type mice. All procedures used were approved by the Institutional Animal Care and Use Committee of the Joslin Diabetes Center.
Muscle Incubation and Contraction in Vitro-Mice were sacrificed, and the extensor digitorum longus (EDL) muscles were rapidly removed and treated as previously described (4,24). Both ends of each EDL muscle were tied with suture (silk 4 -0) and mounted on an incubation apparatus. Muscles were preincubated in 6 ml of Krebs-Ringer bicarbonate (KRB) buffer (117 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl 2 , 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , 24.6 mM NaHCO 3 , pH 7.5) containing 2 mM pyruvate for 20 min. The muscles were then incubated in KRB buffer in the absence or presence of 2 mM AICAR (20 min), 500 M dinitrophenol (DNP; 20 min), 3 M rotenone (40 min), or 120 mM sorbitol (30 min). We have previously reported that muscle treatment with DNP, rotenone, and sorbitol decreases ATP, phosphocreatine, and glycogen in skeletal muscle (24). For the wortmannin studies muscles were preincubated in KRB buffer containing 2 mM pyruvate for 30 min in the presence or absence of wortmannin (100 nM). The maximal concentration of vehicle (Me 2 SO) was 0.1%, which did not affect any assay. The buffers were kept at 37°C throughout the experiment and gassed continuously with 95% O 2 and 5% CO 2 . For muscle contraction, muscles were transferred to a tissue support with stimulating electrodes (Harvard Apparatus, Holliston, MA), and resting tension was set to 0.5 g. Muscles were stimulated for 10 min with the following parameters: train rate ϭ 1/min; train duration ϭ 10 s; pulse rate ϭ 100 pulse/s; duration ϭ 0.1 ms; volts ϭ 100 V. Force production during contraction was monitored with an isometric force transducer (Kent Scientific, Litchfield, CT), and the converted digital signal was captured by a data acquisition system (iWorx114, CB Sciences, Dover, NH). Muscle force generation was determined with data analysis software (LabScribe, CB Sciences, Dover, NH) and represented by g ϫ 10 s. In the second series of experiments muscles from wild type mice were stimulated 2-3 times before the 10-min contraction period to find an optimum voltage to match muscle force production with the ␣2i TG mice. Prior contractions were very short (train rate ϭ 0.1/min; train duration ϭ 180 ms), and we determined that these short contractions did not affect force generation during the subsequent 10 min of contraction. The standard 10-min contraction protocol was then carried out with the determined optimum voltage (8.4 -21 V).
Measurement of Glucose Transport in Vitro-Immediately after muscle incubation or contraction, the muscles were transferred to KRB buffer containing 1 mmol/liter 2-deoxy-D-glucose (1.5 Ci/ml) and 7 mmol/liter D-[ 14 C]mannitol (0.3 Ci/ml) (PerkinElmer Life Sciences) at 30°C and incubated for 10 min. The same concentration of AICAR, DNP, rotenone, sorbitol, or insulin was included in each buffer if present during the previous incubation period. Glucose transport was terminated by dipping muscles in KRB buffer containing 80 mol/liter cytochalasin B at 4°C, and muscles were frozen in liquid nitrogen. Muscles were weighed and digested by incubating in 250 l of 1 M NaOH at 80°C for 10 min. Digests were neutralized with 250 l of 1 M HCl, and particulates were precipitated by centrifugation at 13,000 ϫ g for 5 min. Radioactivity in aliquots of the digested muscles was determined by liquid scintillation counting for dual labels, and the extracellular and intracellular spaces were calculated as previously described (4).
In Situ Muscle Contraction-Mice that had been fasted for 12 h were anesthetized with an intraperitoneal injection of pentobarbital sodium (100 mg/kg). The sciatic nerves were bilaterally isolated, and subminiature electrodes were placed around each nerve and interfaced with a Grass model S88 electrical stimulation unit as previously described (32,33). Hindlimb muscles on one side were stimulated to induce contractions for 15 min (1 train/s, 500-ms train duration, 100 Hz, 0.1-ms duration, 1-5 V), whereas the contralateral side remained unstimulated and served as a sham-treated control.
Measurement of Glucose Uptake in Vivo-Muscle glucose uptake in vivo was measured as described previously (34). Briefly, base-line blood samples were collected from the tails of mice, the jugular vein was catheterized, and an intravenous bolus of 2-deoxy-D-[ 3 H]glucose (10 Ci/ mouse) was administered. After injection of the tracer, mice were subjected to the 15 min in situ muscle contraction protocol described above. Blood samples were obtained at 5, 10, 15, 25, 35, and 45 min for the determination of blood glucose concentrations and 2-deoxy-D-[ 3 H]glucose specific activities. After collection of the last blood sample, animals were sacrificed, and the EDL, the tibialis anterior, and gastrocnemius muscles were removed and snap-frozen in liquid nitrogen. Muscles were digested by incubating in 400 l of 1 M NaOH at 80°C for 10 min and neutralized with 400 l of 1 M HCl. Aliquots of this neutralized solution were added to either perchloric acid or Ba(OH) 2 /ZnSO 4 and centrifuged at 13,000 ϫ g for 5 min. The radioactivity of an aliquot of each of these two supernatants was determined by liquid scintillation counting. Phosphorylated 2-deoxy-D-glucose in each tissue was then calculated as the difference between the radioactivity in the perchloric acid and Ba(OH) 2 /ZnSO 4 supernatants and used to calculate rates of uptake (35).
Measurement of Isoform-specific AMPK Activity-AMPK activity was measured as previously described (4). Briefly, muscle lysates (150 g protein) were immunoprecipitated with specific antibodies against the ␣1 or ␣2 catalytic subunits (36) and protein A beads. The kinase reaction was carried out in 40 mM Hepes, pH 7.0, 0.1 mM synthetic SAMS peptide, 0.2 mM AMP, 80 mM NaCl, 0.8 mM dithiothreitol, 5 mM MgCl 2 , and 0.2 mM ATP (2 Ci of [␥-32 P]ATP) for 20 min at 30°C. Reaction products were spotted on Whatman P81 filter paper, the papers were extensively washed in 1% phosphoric acid, and radioactivity was assessed with a scintillation counter. Kinase activity was assessed by incorporated ATP (pmol) per immunoprecipitated protein (mg) per min as previously described (37).
AICAR Tolerance Test-AICAR (0.25 g/kg) was injected intraperitoneally to wild type and ␣2i TG mice. Blood was collected from the tail before (0 min) and after AICAR injection (20, 40, and 60 min). Blood glucose was determined using a glucometer (ONE TOUCH Ultra, LifeScan, PA).
Noninvasive Physiological and Behavioral Characterization-Fourteen-week-old wild type and ␣2i TG mice were subjected to noninvasive physiological and behavioral characterization using the Comprehensive Lab Animal Monitoring System (CLAMS; Columbus Instruments, Columbus) at the Physiology Core Laboratory of the Joslin Diabetes Center. The mice were monitored for 24 h to assess oxygen consumption (ml/kg/h), carbon dioxide generation (ml/kg/h), heat generation calculated from the gas exchange data (kcal/h), food consumption (g/24 h), water consumption (ml/24 h), and locomotive activity, evaluated by three-dimensional fixed point observation (counts/h). Monitoring started at 10:00 h, and CLAMS assessment was made during both the light cycle (07:00 to 19:00 h) and dark cycle (19:00 to 7:00 h).
Statistics-Statistical evaluation was performed by two-way analysis of variance or Student's two-tailed t test. When analysis of variance revealed significant differences, the Bonferroni t test was used as a post hoc test for multiple comparisons. . AMPK activity and expression of AMPK subunits in muscle from wild type and ␣2i TG mice. A, representative immunoblots for hemagglutinin (HA), AMPK ␣2, ␣1, ␤2, and ␥3 in EDL muscle lysates from wild type and ␣2i TG mice. The upper arrow shows exogenously expressed ␣2 inactive form, and the lower arrow shows the endogenous ␣2 protein. B, isolated EDL muscles were incubated with or without AICAR (2 mM) for 20 min in vitro, and then the muscles were used for isoform-specific AMPK activity assays. The left panel shows ␣2 activity, and the right panel shows ␣1 activity. C, isolated EDL muscles were incubated in KRB buffer for 20 min with or without DNP (500 M), and the muscles were used for isoform-specific AMPK activity assays. The left panel shows ␣2 activity, and the right panel shows ␣1 activity. Data are the means Ϯ S.E. n ϭ 4 -6/group. *, significantly different from basal of the same genotype (*, p Ͻ 0.05; **, p Ͻ 0.01). †, significantly different from wild type with the same intervention ( †, p Ͻ 0.05; † †, p Ͻ 0.01).

RESULTS
AMPK Activity and Subunit Expression in ␣2i TG Mice-Immunoblot analysis using an isoform-specific AMPK ␣2 antibody showed that the ␣2 protein in the ␣2i TG mice possessed a slower mobility (Fig. 1A, upper arrow) compared with wild type mice (lower arrow), likely due to the hemagglutinin tag. In ␣2i TG mice, endogenous ␣2 was only detected with very long exposure times, suggesting that almost all endogenous ␣2 protein was replaced by the mutated ␣2 in ␣2i TG mice. Band densities revealed that expression of the ␣2 mutant protein was 1.6-fold higher than the endogenous ␣2 expressed in wild type mice (Fig. 1A, left). To characterize the effects of ␣2 transgene expression on AMPK activity, EDL muscles were isolated from ␣2i TG mice and their wild type littermate controls. Muscles were incubated in vitro in the absence or presence of AICAR. Both basal and AICAR-stimulated ␣2 activities were severely reduced in ␣2i TG mice compared with wild type mice (Fig. 1B, left). On the other hand, basal levels of ␣1 activity were only slightly decreased in the muscle of the ␣2i TG mice (Fig. 1B,  right). AICAR-stimulated ␣1 activity was reduced by 50% in ␣2i TG mice, but this still represented a significant 2-fold increase above basal. We also characterized AMPK activation in muscles of ␣2i TG mice using DNP, a chemical mitochondrial uncoupler which reduces cellular ATP levels and activates AMPK. Similar to AICAR, DNP-induced ␣2 activation was abolished, and ␣1 activation was partially inhibited in ␣2i TG mice (Fig. 1C).
Forced expression of an inactive ␣ subunit can act as a dominantnegative by preventing the native ␣ subunit from binding to the regulatory subunits (39). This in turn leads to instability and degradation of the ␣ monomer subunit (39,40). The expression level of ␣1 was partially reduced (ϳ50%) in ␣2i TG mice (Fig. 1A, right), indicating that the expressed ␣2 inactive form exchanged to greater extent with endogenous ␣2, and this resulted in dominant inhibition of ␣2 activity in ␣2i TG mice. The level of expression of ␤2 and ␥3 did not change in ␣2i TG mice (Fig. 1A). All results shown in Fig. 1 were identical in two other lines of ␣2i TG mice (data not shown).
␣1i TG Mice Have Ablated ␣2 but Not ␣1 Activity-The predominant inhibition of AMPK␣2 activity observed in ␣2i TG mice led us to determine whether we could generate a mouse with predominant inhibition of AMPK ␣1 activity in skeletal muscle (␣1i TG). To directly compare wild type, ␣1i TG, ␣2i TG, and ␣1/2i TG crosses, ␣1i TG and ␣2i TG mice were bred, and littermates from the breeding were used to characterize AMPK subunit expression and AMPK activity. For the ␣1i TG mice, the Myc tag did not clearly change mobility of the ␣1 protein ( Fig. 2A), so we could not distinguish between the endogenous and mutant ␣1. However, in the ␣1 TG mice, the Myc tag was plainly evident (not shown), and immunoblotting with our ␣1 antibody that recognizes both endogenous and mutant protein revealed that ␣1 expression was Ͼ25-fold higher compared with wild type mice ( Fig. 2A). Interestingly, the ␣1 TG mice had reduced endogenous ␣2 protein, similar to the ␣2i TG mice (Fig. 2A).
Basal levels of ␣1 activity were not reduced in the ␣1i TG mice (Fig.  2B). Furthermore, AICAR-stimulated ␣1 activity was decreased by only 25% in the muscles of these animals, making the magnitude of decrease in ␣l activity comparable for the ␣1i and ␣2i TG mice. Although expression of the ␣1i transgene had minimal effects on ␣1 activity, the ␣1i TG mice had marked reductions in both basal and AICAR-stimulated ␣2 activity (Fig. 2C). Taken together, the decreases in endogenous ␣2 protein expression in the ␣1i TG mice and the greater decreases in ␣2 activity compared with ␣1 activity suggest that the expressed ␣1 inactive form predominantly displaces endogenous ␣2 rather than ␣1 and forms ␣1i/␤/␥ heterotrimer complexes. In ␣1/2i TG mice the expres-sion pattern of the ␣2 protein was similar to that in ␣2i TG mice, and the expression pattern of the ␣1 protein was similar to that in ␣1i TG mice ( Fig. 2A). This crossing of the two lines did not further decrease ␣1 and ␣2 activities (Fig. 2C). Due to the similarities of the three types of mice, only data from the ␣2i TG mice are presented for the studies of glucose transport and physiological characterization.
Physiological and Behavioral Parameters in ␣2i TG Mice-Body weight, monitored in male and female mice for up to 35 weeks of age, was not significantly different between wild type and ␣2i TG mice (not shown). As summarized in TABLE ONE, expression of the ␣2i transgene in muscle did not significantly alter locomotive activity, oxygen consumption, carbon dioxide generation, heat generation, or food and water consumption.
AICAR Effects in ␣2i TG Mice-Basal rates of glucose transport measured in vitro were not altered in the ␣2i TG mice (Fig. 3A). The ablation of AICAR-stimulated ␣2 activity (Fig. 1B) was associated with full inhibition of AICAR-stimulated glucose transport in incubated EDL muscles from the muscle-specific ␣2i TG mice (Fig. 3A). This result corroborates that AMPK ␣2 activity is necessary for AICAR effects on muscle glucose transport and that the remaining AMPK ␣1 activity is not sufficient for this effect. To determine whether muscle-specific ablation of ␣2 activity changes whole body sensitivity to AICAR, we performed AICAR tolerance tests. Consistent with our glucose transport data, the blood glucose lowering effect of AICAR was diminished in the ␣2i TG mice (Fig. 3B). The slight decrease in blood glucose concentrations in AMPK ␣2i TG mice during the AICAR tolerance test is probably due to the well known effects of AICAR to inhibit glucose production in the liver and stimulate glucose transport in adipose tissue.
Rotenone and Sorbitol Effects on Glucose Transport in ␣2i TG Mice-We next determined if AMPK ␣2 plays a role in mediating skeletal muscle glucose transport by other non-insulin stimuli. EDL muscles were isolated and incubated in KRB buffer with or without rotenone, a mitochondrial respiratory Complex I inhibitor. As shown in Fig. 4A, rotenone increased glucose transport 1.5-fold above basal levels in wild type mice, and this increase in glucose transport was completely abolished in ␣2i TG mice. The changes in glucose transport were associated with full inhibition of AMPK ␣2 activation and partial reduction (ϳ40%) in AMPK ␣1 activation in ␣2i TG mice (data not shown). Thus, similar to AICAR-induced glucose transport, AMPK ␣2 activity is necessary for rotenone-induced glucose transport.
High concentrations of sorbitol induce hyperosmolarity that results in increases in glucose transport and AMPK activity in rat skeletal muscle (24) and C2C12 muscle cells in culture (6). To determine whether AMPK is necessary for hyperosmolarity-induced glucose transport, isolated EDL muscles from ␣2i TG and wild type mice were incubated in KRB buffer with or without sorbitol. In contrast to AICAR-and rotenone-induced glucose transport, sorbitol-induced glucose transport was not affected by the lack of ␣2 activity in skeletal muscle (Fig. 4B). Sorbitol-stimulated ␣1 activity was only slightly inhibited in ␣2i TG mice (data not shown).
Contraction-induced Glucose Transport in a2i TG Mice-Mu et al. (28) reported that contraction-induced glucose transport was impaired in muscles from AMPK transgenic mice (28), whereas Jorgensen et al. (29) reported that contraction-induced glucose transport was normal in both ␣1 and ␣2 whole body knock-out mice. The reason for these discrepant findings is not clear. Here, we used multiple muscles from wild type and ␣2i TG mice to measure contraction-stimulated glucose transport both in vitro and in vivo. First, EDL muscles were isolated from wild type and ␣2i TG mice and used for contraction experiments followed by glucose transport measurements in vitro. Fig. 4C shows that contraction-induced glucose transport is slightly reduced in ␣2i TG mice (by 13.6%). In these mice ␣2 activation in response to muscle contraction was severely diminished, and ␣1 activation was reduced by 50% (data not shown). These data suggest that AMPK activity may be necessary for full activation of glucose transport with contraction. However, by performing detailed analysis of muscle force generation recorded during contraction, we found that force production was reduced in ␣2i TG mice compared with wild type mice (Fig. 5A). This raised the possibility that the decreases in glucose transport were due to decreases in force production and not impaired AMPK ␣2 activity. To address this question, we compared glucose transport rates over a wide range of force production. Because we found a linear relationship between muscle   force generation and glucose transport (Fig. 5B), the decrease in contraction-induced glucose transport in ␣2i TG mice is likely due to the reduced muscle force generation. To test this more directly, we decreased voltage of the electrical stimulation in wild type mice to match muscle force generation of the ␣2i TG mice. When muscle force generation in wild type mice was matched to ␣2i TG mice (Fig. 5C), there were no differences in contraction-induced glucose transport (Fig. 5D). We next measured glucose transport in a number of muscles using 2-deoxy-D-[ 3 H]glucose infusion after in situ muscle contraction produced by electrical stimulation of the sciatic nerve. In wild type mice contraction increased glucose uptake into EDL, tibialis anterior, and gastrocnemius muscles by 5-10-fold above sham-treated muscles (Fig.  6, left panels). Consistent with our in vitro muscle contraction data, contraction-induced glucose uptake measured in vivo was normal in all muscles in ␣2i TG mice (Fig. 6, left panels). This occurred despite blunted activation of AMPK ␣2 in the three muscles from the ␣2i TG mice (Fig. 6, center panels). AMPK ␣1 activity was not significantly increased with in situ contraction, consistent with work from our group and others using this type of contraction protocol (41,42). GLUT1 and GLUT4 protein abundance assessed by immunoblotting were not altered in EDL muscles from the ␣2i TG mice (Fig. 7), indicating there was no compensatory adaptation at the level of glucose transporter expression in the transgenic animals. We also examined the possibility that components of the insulin signaling pathway compensate for ablated AMPK ␣2 activity in the transgenic mice. IRS-1 and Akt protein determined by immunoblotting and insulin-stimulated glucose transport measured in vitro were not different between wild type and ␣2i TG mice (data not shown). In addition, EDL muscles from wild type and ␣2i TG mice were contracted in vitro in the absence or presence of the phosphatidylinositol 3-kinase inhibitor wortmannin. Wortmannin had no affect on contraction-induced glucose transport in both wild type (minus wortmannin; 4.09 Ϯ 0.27 versus plus wortmannin; 4.11 Ϯ 0.39 mol/g/min) and ␣2i TG mice (minus wortmannin; 3.04 Ϯ 0.45 versus plus wortmannin; 3.47 Ϯ 0.26 mol/g/min, mean Ϯ S.E., n ϭ 6 -8). Taken together, these results demonstrate that there is no compensation by insulin-dependent signaling molecules in ␣2i TG mice.

DISCUSSION
Multiple agents and metabolic perturbations can result in increased rates of glucose transport into skeletal muscle fibers. AMPK has been proposed to be the key signaling molecule mediating insulin-independent glucose transport. However, the role of AMPK in glucose transport regulation in skeletal muscle has proven to be ambiguous. Full inhibition of AICAR-and rotenone-induced glucose transport in ␣2i TG mice clearly demonstrates a role for AMPK ␣2 in glucose transport regulation in skeletal muscle. On the other hand, both muscle contraction and hyperosmolarity strongly activate AMPK ␣2, but inhibition of ␣2 activity does not diminish glucose transport increased in response to these stimuli. These results suggest that there is a redundant regulatory mechanism that consists of multiple signaling pathways within skeletal muscle for stimuli such as contraction and hyperosmolarity.
Hyperosmolarity has long been documented as a potent stimulator of glucose transport in isolated skeletal muscle (43). We found previously that hyperosmolar concentrations of sorbitol decrease intracellular ATP, phosphocreatine and glycogen content and concomitantly increase AMPK ␣1 and ␣2 activities in rat skeletal muscle (24). Therefore, we hypothesized that AMPK might be involved in the signaling mechanism of hyperosmolarity-induced muscle glucose transport. However, in the current study dominant inhibition of AMPK ␣2 activity had no effect on sorbitol-induced muscle glucose transport, indicating that AMPK ␣2 activation is not essential for hyperosmolarity-induced glucose transport in skeletal muscle. In contrast to our data in intact adult skeletal muscle using a mouse H-2K b muscle cell line, Fryer et al. (6) showed that adenoviral expression of an AMPK dominant-negative FIGURE 5. Reduced contraction-induced glucose transport in vitro in skeletal muscle of ␣2i TG mice is related to reduced muscle force generation. A, isolated EDL muscles were subjected to electrical stimulation in vitro using conditions described under "Experiment Procedures" to induce contractions. Force production during the contraction was monitored with an isometric force transducer, the converted digital signal was captured by a data capture system, and muscle force generation was analyzed with data recording software. *, significantly different between wild type (WT) and ␣2i TG (p Ͻ 0.05). n ϭ 5/group. B, relationship between muscle force generation and increase in glucose transport over basal. n ϭ 29 from 3 independent experiments. C, muscle force generation during contractions in wild type was matched to ␣2i TG. Muscles from wild type mice were stimulated 2-3 times before the 10-min contraction period to find an optimum voltage to match muscle force production with the ␣2i TG mice. The standard 10-min contraction protocol was then carried out with the determined optimum voltage (8.4 -21 V). D, contraction-induced glucose transport was measured in the muscles used in C. Data are as means Ϯ S.E. n ϭ 5/group. CTXN, contraction. mutant fully inhibited sorbitol-induced increases in glucose uptake. We suspect that the divergent results may be a function of different glucose transport machinery in the cell line versus the intact adult skeletal muscle.
Contraction of isolated skeletal muscle activates AMPK in a dose-dependent manner that matches precisely with the increase in glucose transport (41). Similar to contraction, AICAR-stimulated glucose transport occurs in the absence of insulin, is wortmannin-insensitive, and is additive with the effects of maximal insulin. These results along with several other findings suggested that AMPK is an intermediate in the signaling pathway leading to contraction-induced glucose transport. However, because of the non-specificity of AICAR on AMPK activation (44 -46), definitive evidence for AMPK regulation of glucose transport has been lacking. Reports based on transgenic and knock-out mouse models have partially overcome this problem. Mu et al. (28) generated AMPK transgenic mice and found that muscle glucose transport measured in isolated muscles after in vitro or in situ contraction was 30 -40% reduced in the transgenic mice compared with wild type mice. On the other hand, Jorgensen et al. (29) showed that muscle glucose transport with in vitro contraction was essentially unaffected by knockout of either AMPK ␣ isoform. Because the ␣2 knock-out mice had a 2-3-fold increase in ␣1 expression and a 2-fold increase in contraction-stimulated ␣1 activity, as the authors acknowledge, it is possible that the up-regulation of ␣1 compensated for the loss of ␣2 function with regard to glucose transport. In our study, muscle glucose transport measured in vitro was slightly reduced in the ␣2i TG mice compared with wild type mice (Fig. 4C), probably due to reduced muscle force generation FIGURE 6. Contraction-induced glucose transport in vivo is normal in skeletal muscle of ␣2i TG mice. Wild type (WT) and ␣2i TG mice were anesthetized, and the sciatic nerve was exposed. One leg was electrically stimulated for 15 min to induce muscle contractions, and the other leg was served as sham controls. EDL (A), tibialis anterior (B), and gastrocnemius muscles (C) were isolated after the contraction. Muscles were used for the measurement of glucose transport assessed in vivo by 3 H-two dimensional glucose phosphorylation (left panels), AMPK ␣2 activity (center panels), and AMPK ␣1 activity (right panels). Open bars show sham control, and hatched bars show contracted. Data are means Ϯ S.E., n ϭ 4 -6/groups. *, significantly different from basal of the same genotype (*, p Ͻ 0.05; **, p Ͻ 0.01). †, significantly different from wild type with the same intervention ( † †, p Ͻ 0.01). during the in vitro contraction (Fig. 5A). In support of this hypothesis, there was a linear relationship between muscle force generation and glucose transport (Fig. 5B), and the finding that when force production was matched there was no difference in glucose transport between wild type and ␣2i TG mice (Fig. 5D). These results suggest that AMPK ␣2 activation is not essential for contraction-induced glucose transport measured in vitro. Moreover, we also show that contraction-induced glucose transport measured in vivo was normal in multiple muscles of the ␣2i TG mice (Fig. 6). Therefore, there is now considerable evidence using both knock-out and transgenic models that ␣2 activation is not essential for contraction-induced glucose transport in skeletal muscle.
We do not know the mechanism for reduced muscle force generation with in vitro contraction in the ␣2i TG mice. One possibility is that there are reduced intracellular energy stores due to loss of AMPK ␣2 function. We have shown that transgenic mice expressing an inactive AMPK ␣2 in heart have markedly accelerated rates of ATP depletion under ischemic conditions measured by 31 P NMR spectroscopy (47), and a similar phenomenon may occur in skeletal muscle. Contraction-induced glucose transport measured in vivo was normal in the ␣2i TG mice. Although we could not measure muscle force generation during contraction in vivo, this result suggests that contraction force was not reduced during contraction in the in vivo situation. It is possible that intracellular energy reduction during contraction is milder in vivo compared with in vitro.
We cannot completely exclude the possibility that a small fraction of ␣1 activation may be enough to induce normal glucose transport caused by muscle contraction. In our opinion, however, this seems unlikely for the following reasons. First, in situ muscle contraction increases glucose transport with no activation of ␣1 in both wild type and ␣2i TG mice (Fig. 6), and Jorgensen et al. (29) showed that contraction-induced muscle glucose transport is unaffected in ␣1 knock-out mice. These observations may suggest no role for ␣1 in contraction-induced glucose transport. Second, both AICAR-and rotenone-induced glucose transport are abolished in ␣2i TG mice, although ␣1 activity increases similarly to that seen with in vitro contractions, suggesting that ␣1 activation cannot compensate for ␣2 in regulating glucose transport. Instead, we hypothesize that mature skeletal muscle cells express low levels of ␣1 protein and that a considerable amount of the ␣1 activity detected in lysates of whole muscle tissues comes from other cell types such as vascular endothelial and smooth muscle cells, adipocytes, nerves, and fibroblasts. In fact, dorsal aorta and inferior vena cava blood vessels, visceral fat, and peroneal and sciatic nerves isolated from wild type mice have 3-4.5-fold higher ␣1 expression compared with skeletal muscle tissue. 3 This could explain why both the ␣1 and ␣2 inactive mutants exchange to a greater extent with endogenous ␣2 in the skeletal muscle. In addition, normal muscle glucose transport in the whole body ␣1 knock-out mice might also be explained by the localization of ␣1 protein to non-muscle cells.
In summary, we found that ablation of AMPK ␣2 activity inhibits AICAR-and rotenone-induced glucose transport in skeletal muscle, demonstrating that AMPK ␣2 is necessary for regulation of some insulin-independent glucose transport in skeletal muscle. In contrast, AMPK ␣2 activity is not essential for muscle contraction-and hyperosmolarity-induced glucose transport, although both stimuli are strong activators of AMPK ␣2. These results suggest that muscle contractionand hyperosmolarity-induced glucose transport are regulated by a redundant mechanism in which AMPK ␣2 is one of multiple signaling pathways.