Ca²⁺/Calmodulin-Dependent Protein Kinase Kinase-α Regulates Skeletal Muscle Glucose Uptake Independent of AMP-Activated Protein Kinase and Akt Activation

Experiments were performed in accordance with the institutional animal care and use committee of the Joslin Diabetes Center as well as the National Institutes of Health guidelines for the care and use of laboratory animals. Mice were housed in cages at 21°C with a 12-h light/dark cycle. LabDiet rodent chow (Purina, St. Louis, MO) and water were available ad libitum. Male and female ICR mice (6–8 weeks) were obtained from Taconic (Germantown, NY) or Charles River Laboratories (Wilmington, MA). Male and female muscle-specific AMPKα2 inactive (AMPKα2i) mice and their wild-type littermates (FVB background, 12–16 weeks) were generated as previously described, and a description of their phenotype has recently been published (6).

Vectors encoding CaMKKα (amino acids 1–434) and the empty vector (pCS2+) were generously donated by Thomas R. Soderling (Vollum Institute, Oregon Health and Science University, Portland, OR) (19). Plasmid DNA injections and in vivo electroporation were performed using a method adapted from Aihara and Miyazaki (20–22). Plasmid DNA (100 μg) for CaMKKα was injected into the tibialis anterior muscle of one leg, and empty vector DNA was injected into the contralateral leg. Needle electrodes were inserted 5 mm apart, and muscles were stimulated eight times (parameters: train rate = 1 train/s, train duration = 500 ms, pulse rate = 1 pulse/s, duration = 20 ms, volts = 100 V) with a stimulator (model S88; Grass Instruments, Quincy, MA). Mice were allowed 2 weeks to express the plasmid. In a previous study performed in our laboratory, this procedure resulted in >85% of skeletal muscle fibers expressing exogenous protein (21).

Skeletal muscle glucose uptake was measured in vivo as previously described (23). Mice were fasted overnight and anesthetized with nembutal sodium (100 mg/kg i.p.). After 30 min, 25 μl of blood was taken from the tail to assess basal glucose and background radioactivity. A bolus of 0.9% NaCl and [³H]-2-deoxyglucose (0.33 μCi [³H]/g body wt, retro-orbital) was administered, and blood samples were taken 5, 10, 15, 25, 35, and 45 min later for glucose and [³H]-2-deoxyglucose measurements. Mice were killed by cervical dislocation and muscles frozen in liquid nitrogen. Accumulation of muscle radioactivity was assessed using a perchloric acid precipitation procedure modified from Ferre et al. (24), and the rate of glucose uptake was calculated as previously described (22).

Mice were fasted overnight and anesthetized with nembutal sodium for 30 min. Blood (25 μl) was taken from the tail to assess basal glucose levels. A bolus of 0.9% NaCl or 20% glucose (1 mg glucose/g body wt, retro-orbital) was administered and a blood sample taken 10 min later to assess glucose levels. Mice were killed by cervical dislocation and muscles frozen in liquid nitrogen.

Frozen muscle homogenates not used for uptake measurements were mixed with 1% Nonidet P-40, rotated end-over-end at 4°C for 1 h, and centrifuged at 14,000g for 30 min. Protein concentrations were determined via the Bradford assay. Immunoblots were performed by standard methods (25). Primary antibodies were obtained from the following sources: phospho-AMPK (Thr¹⁷²) and phospho-AS160 (PAS; Thr⁶⁴²) from Biosource International (Camarillo, CA), CaMKK from BD Transduction Laboratories (Franklin Lakes, NJ), phospho-Akt (Thr³⁰⁸) and PAS (Ser³¹⁸, Ser³⁴¹, Ser⁵⁷⁰, Ser⁵⁸⁸, Thr⁶⁴², and Ser⁷⁵¹) from Cell Signaling Technology (Beverly, MA), GLUT1 and GLUT4 from Chemicon International (Temecula, CA), and Akt1/2 and AS160 from Upstate Biotechnology (Lake Placid, NY). The phospho-CaMKI (Thr^177/180) antibody was generated as previously described (26) and generously donated by Dr. Naohito Nozaki (Kanagawa Dental College, Yokosuka, Kanagawa, Japan).

Calmodulin affinity precipitation experiments were performed using procedures adapted from Rose et al. (27). Muscle lysates containing caCaMKKα (200 μg) were diluted in precipitation buffer containing (in mmol/l): 50 Tris-HCl, pH 7.6, 1 EGTA, pH 8.0, 5 Na₄PO₇, 2 dithiothreitol, 0.1 benzamidine, 0.5 phenylsulfonyl fluoride, 0.003 soybean trypsin inhibitor, and 0.009 leupeptin in the presence or absence of 2 mmol/l CaCl₂ (∼1 mmol/l Ca²⁺). Calmodulin-agarose (20 μl, no. 14–426; Upstate Biotechnology) was added and samples rotated end-over-end at 4°C overnight. The supernatant was removed, and pellets washed four times in precipitation buffer. Laemmli's sample buffer was added and samples boiled at 95°C for 5 min. Calmodulin-affinity precipitates were diluted in double-distilled H₂O. Lysate (20 μg) was directly analyzed by immunoblot.

Isoform specific AMPK activity assays were performed as previously described (5,6). Muscle lysates (200 μg) were immunoprecipitated with antibodies generated against the AMPKα1 and -α2 subunits (28) and protein A beads. The kinase reaction was carried out in buffer containing (in mmol/l): 40 HEPES, pH 7.0, 80 NaCl, 5 MgCl₂, 0.1 synthetic SAMS peptide (HMRSAMSGLHLVKRR), 0.2 AMP, 0.8 dithiothreitol, and 0.2 ATP (2 μCi of [γ-³²P]ATP) for 20 min at 30°C. Reaction products were spotted on Whatman P81 filter paper and washed 5 × 30 min in 0.1% phosphoric acid, and 10 min in acetone. Radioactivity was assessed by liquid scintillation counting of the [³²P] label. AMPK activity was assessed by the incorporation of [γ-³²P] into the SAMS peptide.

Ex vivo muscle incubation experiments were performed as previously described (6,29). Mice were fasted overnight and killed by cervical dislocation. For contraction experiments, isolated extensor digitorum longus muscles were placed in Krebs-Ringer bicarbonate (KRB) buffer containing (in mmol/l) 117 NaCl, 4.7 KCl, 2.5 CaCl₂ · 2 H₂O, 1.2 KH₂PO₄, 1.2 MgSO₄ · 7 H₂O, and 24.6 NaHCO₃ supplemented with 2 mmol/l pyruvic acid and either DMSO (0.1%) or the CaMKK inhibitor STO-609 (1.0 μg/ml; Calbiochem, LA Jolla, CA) for 60 min. Optimal muscle length was determined for each muscle just before the 10-min tetanic contraction (parameters: train rate = 1/min, train duration = 10/s, pulse rate = 100 pulses/s, duration = 0.1 ms, and volts = 100 V). Force production was monitored using 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) and analyzed with software (Labscribe; CB Sciences). For insulin experiments, soleus muscles were placed in KRB buffer supplemented with pyruvic acid and either DMSO or STO-609 for 50 min, before the addition of insulin (50 mU/ml) for 20 min.

For glucose uptake, muscles were incubated in KRB buffer containing 1.5 μCi/ml [³H]-2-deoxyglucose, 1 mmol/l deoxyglucose, 0.45 μCi/ml [¹⁴C]-mannitol, 7 mmol/l cold-mannitol, and the appropriate amount of DMSO, STO-609, and/or insulin. Muscles were frozen in liquid nitrogen, solubilized in 1N NaOH at 80°C, and neutralized with 1 N HCl. Samples were centrifuged at 11,000g for 1 min. Aliquots (350 μl) were removed for scintillation counting of the dual [³H] and [¹⁴C] labels, and the accumulation of [³H]-2-deoxyglucose was calculated after adjusting for extracellular space.

Data are the mean ± SE of the mean. Statistical significance was defined as P < 0.05 and determined by Student's t tests or two-way ANOVA and Student-Newman-Keuls post hoc analysis. The number of muscles used to determine statistical significance is indicated in the text or figure legends.

Endogenous mouse CaMKKα is a 505–amino acid, 68-kDa protein. Truncation of the COOH-terminal 71 amino acids eliminates the autoinhibitory and calmodulin-binding domains, resulting in a constitutively active form of CaMKKα (caCaMKKα) that is Ca²⁺/calmodulin independent (30). In skeletal muscle, CaMKKα protein is detected at ∼68 kDa (Fig. 1A). Expression of caCaMKKα into mouse muscle resulted in robust expression of the ∼52 kDa protein (21.6 ± 2.5-fold) compared with endogenous CaMKKα. There was a slight (∼15%), but nonsignificant, increase in CaMKKα protein levels after caCaMKKα expression. To confirm that the transfection procedure resulted in the expression of active caCaMKKα, the phosphorylation of an established CaMKKα substrate, CaMKI, was examined. CaMKIV is also a CaMKK substrate; however, CaMKIV protein is not expressed in mouse skeletal muscle (31). caCaMKKα expression resulted in an approximately twofold increase in the phosphorylation of three bands detected by the phospho-CaMKI (Thr^177/180) antibody, all of which fall within the predicted molecular weight range for CaMKI (i.e., 37–43 kDa) (Fig. 1B). These results are consistent with studies that have detected mRNA for CaMKIα (32), CaMKIβ (33), and CaMKIδ (34) in skeletal muscle. However, a recent study suggested that skeletal muscle does not contain detectable levels of CaMKI protein (27). Thus, to confirm that the bands detected by the phospho-CaMKI antibody are indeed calmodulin binding proteins, muscle lysates were subjected to calmodulin affinity precipitation. As shown in Fig. 1C, all three bands detected by the phospho-CaMKI antibody bind calmodulin in the presence of ∼1 mmol/l Ca²⁺, providing further evidence that CaMKI protein is expressed in mouse skeletal muscle.

Studies from nonmuscle cell culture models have suggested that CaMKKs act as upstream regulators of AMPK (14–17) and Akt (35) in mammalian cells. To assess whether CaMKKα regulates AMPK and/or Akt in skeletal muscle, immunoblots for AMPK (Thr¹⁷²) and Akt (Thr³⁰⁸) phosphorylation were performed on muscles expressing caCaMKKα. As shown in Fig. 2, caCaMKKα increased AMPK (Thr¹⁷²) phosphorylation approximation twofold but did not increase Akt (Thr³⁰⁸) phosphorylation.

Numerous studies have shown that activation of AMPK by AICAR stimulates skeletal muscle glucose uptake (4–6). To determine whether caCaMKKα stimulates glucose uptake, and whether it is dependent on AMPK activation, muscles from wild-type and transgenic mice expressing an inactive form of the AMPKα2 subunit (AMPKα2i) were transfected with caCaMKKα, and [³H]-2-deoxyglucose uptake assessed in vivo without additional stimulation. Muscle glucose uptake was increased ∼2.5-fold in both wild-type mice and AMPKα2i mice after caCaMKKα expression (Fig. 3). CaMKI (Thr^177/180) phosphorylation was increased to a similar extent in both wild-type (empty vector, 1.0 ± 0.05, caCaMKKα 2.1 ± 0.09, n = 6) and AMPKα2i mice (1.13 ± 0.12, 1.84 ± 0.18, n = 4). To confirm that AMPK activity was not increased by caCaMKKα in the AMPKα2i mice, kinase activity assays were performed. In wild-type mice, caCaMKKα increased AMPKα1 and -α2 activities ∼2.5-fold (Fig. 4A). In contrast, in AMPKα2i mice caCaMKKα did not increase either AMPKα1 or -α2 activities (Fig. 4A), demonstrating that caCaMKKα can stimulate glucose uptake independent of AMPK.

AS160 is a recently discovered protein that regulates insulin-stimulated GLUT4 translocation in L6 cells (36), and it is phosphorylated in response to both insulin and contraction in skeletal muscle (29,37). Recent evidence from our lab has shown that AMPK phosphorylates AS160 on phospho-Akt substrate (PAS) sites in response to AICAR and contraction in skeletal muscle (29). As shown in Fig. 4B, caCaMKKα increased AS160 (PAS) phosphorylation ∼60% (P = 0.06) in muscles from wild-type but not AMPKα2i mice. Collectively, these results demonstrate that caCaMKKα can activate skeletal muscle glucose uptake independent of the activation of AMPK and the phosphorylation of AS160 (PAS).

Although our initial experiments demonstrated that caCaMKKα does not increase Akt (Thr³⁰⁸) phosphorylation (Fig. 2), it was important to determine whether caCaMKKα alters the insulin-induced phosphorylation of Akt and/or AS160. For these experiments, mice were injected with glucose to elicit a physiological insulin response, and muscles were taken 15 min later for immunoblot analysis. In vivo insulin stimulation increased Akt (Thr³⁰⁸) and AS160 (Thr⁶⁴²) phosphorylation, and caCaMKKα expression did not enhance this insulin-induced phosphorylation of Akt or AS160 (Thr⁶⁴²) (Fig. 5). caCaMKKα increased AMPK (Thr¹⁷²) phosphorylation, and in vivo insulin stimulation had no additional effect on AMPK (Fig. 5). The combination of caCaMKKα and in vivo insulin enhanced AS160 (PAS) phosphorylation above that elicited by either treatment alone (Fig. 5), consistent with previous studies suggesting convergence of Akt- and AMPK-dependent phosphorylation on AS160 (PAS) residues (29).

To evaluate the possibility that caCaMKKα alters the expression of proteins involved in regulating glucose uptake, immunoblot analyses were performed to examine AMPK, Akt, AS160, GLUT1, and GLUT4 protein levels. Expression of caCaMKKα did not alter the protein expression of AMPK, Akt, AS160, GLUT1, or GLUT4 (supplementary Fig. 1, which can be found in an online appendix [available at http://dx.doi.org/10.2337/db06-1230]).

To determine whether CaMKK signaling regulates contraction-stimulated glucose uptake, isolated muscles were treated with or without the CaMKK inhibitor STO-609 (1.0 μg/ml) and then electrically stimulated to contract. As shown in Fig. 6A, STO-609 slightly inhibited contraction-induced glucose uptake in muscles from wild-type mice, but this was not statistically significant. In muscles from AMPKα2i mice, contraction-induced glucose uptake was significantly decreased (∼24%) by STO-609. STO-609 had no effect on muscle force production (Fig. 6B). To assess whether STO-609 elicits nonspecific effects on glucose uptake, insulin-induced glucose uptake was examined. STO-609 had a slight, but nonsignificant, effect on insulin-stimulated glucose uptake in both wild-type and AMPKα2i mice (Fig. 6C). Collectively, these results suggest that CaMKK-dependent signaling is involved in the regulation of contraction-stimulated muscle glucose uptake.

In vitro studies have demonstrated that STO-609 (1.0 μg/ml) can directly bind to and inhibit AMPK activity by up to 60% (38). If this also occurs in skeletal muscle in vivo, then treatment with STO-609 should directly inhibit both CaMKK- and AMPK-dependent signaling regardless of possible physiological interactions. To determine whether STO-609 directly inhibits AMPK-dependent signaling in muscle from wild-type mice, we assessed AMPK (Thr¹⁷²) phosphorylation and the phosphorylation of an established AMPK substrate, acetyl-CoA carboxylase (ACC; Ser²²¹). STO-609 had no effect on the basal or contraction-induced phosphorylation of AMPK or ACC (Fig. 7). These results demonstrate that STO-609 (1.0 μg/ml) does not directly inhibit AMPK-dependent signaling, providing more evidence for the lack of a decrease in contraction-induced glucose uptake in the muscles from wild-type mice (Fig. 6A).

The data presented in this study implicate CaMKKα in the regulation of contraction-induced skeletal muscle glucose uptake independent of AMPK. These novel data provide new insight into the mechanisms regulating contraction-induced glucose uptake, especially those mediated by Ca²⁺/calmodulin-dependent signaling.

caCaMKKα expression increased skeletal muscle glucose uptake independent of changes in AMPKα1 or -α2 activity (Figs. 3 and 4). Our results are consistent with data obtained from rat epitrochlearis muscles, in which the sarcoplasmic reticulum Ca²⁺ store releasing agent, caffeine, and the AMPK activator, AICAR, had additive effects on glucose uptake (11). Other evidence for an AMPK-independent regulation of non–insulin-dependent glucose uptake comes from studies using AMPK transgenic and knockout mice, in which inactivation of muscle AMPK signaling had little to no effect on contraction-induced increases in glucose uptake (6–8). Thus, our data provide further in vivo evidence for a Ca²⁺-dependent, AMPK-independent mechanism that can regulate skeletal muscle glucose uptake.

In muscles from AMPKα2i mice, caCaMKKα did not increase AMPKα1 activity (Fig. 4). This finding was not surprising because the exogenous AMPKα2i subunit would be expected to replace both the endogenous AMPKα1 and -α2 subunits. Consistent with this, AMPKα1 expression is decreased in muscles from AMPKα2i mice (6). Because basal AMPKα1 activity is not decreased in AMPKα2i mice, this suggests that the AMPKα1 protein still present in the muscle is already mostly activated and that there is not enough nonactivated AMPKα1 remaining to detect a significant increase in AMPKα1 activity in response to caCaMKKα expression.

Akt (Thr³⁰⁸) phosphorylation was not increased by caCaMKKα expression, even when Akt phosphorylation was increased by in vivo insulin stimulation (Figs. 2 and 5). This result was surprising because in vitro studies using recombinant proteins, and studies using COS-7 cells, have demonstrated that CaMKKα can regulate Akt (35). The reason behind the lack of Akt regulation by CaMKKα in skeletal muscle is currently unknown, although possible explanations include tissue-specific CaMKKα function, protein localization, and a lack of necessary cofactors. We do not believe that the lack of Akt phosphorylation was caused by inadequate CaMKKα activity because significant CaMKI and AMPK phosphorylation were detected in our model.

In mouse skeletal muscle, insulin increases AS160 (Thr⁶⁴²) phosphorylation via activation of Akt, whereas contraction increases AS160 (PAS) phosphorylation via activation of AMPK (29). In line with those results, caCaMKKα expression did not increase AS160 (Thr⁶⁴²) phosphorylation, but it did increase AS160 (PAS) phosphorylation in wild-type mice (Fig. 5). In muscles from AMPKα2i mice, a change in AS160 (PAS) phosphorylation was not detected, demonstrating that the increased AS160 (PAS) phosphorylation seen in the wild-type mice expressing caCaMKKα was attributable to AMPK activation (Fig. 4). We did not detect CaMKKα-dependent, AMPK-independent AS160 phosphorylation.

The role of CaMKK signaling in the regulation of contraction-induced skeletal muscle glucose uptake was directly assessed using the CaMKK inhibitor STO-609. STO-609 was used at 1.0 μg/ml (∼2.7 μmol/l), a concentration that inhibits recombinant CaMKKα and CaMKKβ activity by >80% in vitro, with only minor inhibition of CaMKI and -II isoforms (38). Thus, we do not believe that STO-609 is directly inhibiting CaMKI or -II in our study. Interestingly, STO-609 (1.0 μg/ml) directly inhibits AMPK activity by ∼60% in vitro (14). To assess this possibility in skeletal muscle in vivo, we examined the phosphorylation of the AMPK substrate ACC (Ser²²¹) (Fig. 7). STO-609 had no effect on basal or contraction-induced ACC (Ser²²¹) phosphorylation, demonstrating that STO-609 was not directly inhibiting AMPK in skeletal muscle in vivo.

We were not able to detect changes in CaMKI (Thr^177/180) phosphorylation by immunoblot analysis after ex vivo muscle contraction experiments (data not shown). However, we currently believe that this is because of the complex structural/functional properties of CaMKKα, CaMKI, and CaMK phosphatase, and not because CaMKKα and CaMKI are inactive during muscle contraction. Directly assessing CaMKKα or CaMKI under physiological conditions is extremely challenging because of the lack of Ca²⁺/calmodulin-independent activity of these enzymes (39,40). Thus, immediately after the removal of a stimulus, and a decline in intracellular Ca²⁺ levels, CaMKKα is not able to maintain kinase activity (40) and stops phosphorylating CaMKI. In addition, in vitro studies have shown that CaMKI can be rapidly (≤1 min) dephosphorylated by ≥80% by CaMK phosphatase (41), a phosphatase with a high degree of specificity for the CaMK family (41,42), which is expressed in skeletal muscle (43). Thus, novel techniques will need to be developed to directly examine the possible activation of CaMKKα or CaMKI by contraction in skeletal muscle.

STO-609 only inhibited contraction-induced glucose uptake in muscles from AMPKα2i mice (Fig. 6A). This finding is consistent with the hypothesis that contraction-mediated glucose uptake is regulated by multiple redundant pathways and that inhibition of more than one pathway is required to blunt the effect. STO-609 did not completely inhibit contraction-induced glucose uptake in the muscles from AMPKα2i mice. Thus, there are likely one or more other kinases involved in this process. Of the possible candidates, we cannot exclude a role for CaMKII. Data obtained using the Ca²⁺/calmodulin competitive inhibitor KN-93 has shown ≥50% inhibition of caffeine- and contraction-induced glucose uptake in rat epitrochlearis muscles (11). However, because KN compounds also inhibit insulin-dependent muscle glucose uptake (12,13), the role of CaMKII in the regulation of both contraction- and insulin-dependent glucose uptake is controversial and warrants further investigation.

During the revision of this manuscript, Jensen et al. (44) reported a role for CaMKK-dependent signaling in the regulation of muscle glucose uptake. However, in contrast to our study, their study suggests that CaMKK regulates glucose uptake via an AMPK-dependent pathway. This aspect of the Jensen et al. study is not supported by our current work showing that caCaMKKα can increase glucose uptake in the absence of changes in AMPKα1 or -α2 activity (Figs. 3 and 4).

In summary, this study is the first to demonstrate that CaMKKα-dependent signaling regulates skeletal muscle glucose uptake independent of AMPK, Akt, and AS160 (PAS) phosphorylation. In addition, we show that treatment with the CaMKK inhibitor STO-609 significantly decreases contraction-induced glucose uptake in muscles lacking AMPKα2 activity, consistent with the hypothesis that contraction-mediated glucose uptake is mediated via multiple signaling pathways. Collectively, these novel findings represent a significant step toward the elucidation of the signaling pathways regulating non–insulin-dependent skeletal muscle glucose uptake.

This work was supported by National Institutes of Health Grants R01AR42238 and R01AR45670 (to L.J.G.), F32AR051663 (to C.A.W.), and T32DK07260-27 (to the Joslin Diabetes Center) and Diabetes Endocrinology Research Grant DK36836.

We thank Thomas R. Soderling for his generous donation of the expression vectors, Dr. Naohito Nozaki for his generous donation of the phospho-specific CaMKI antibody, and Eric S. Silver for technical assistance.