Improving Type 2 Diabetes Through a Distinct Adrenergic Signaling Pathway Involving mTORC2 That Mediates Glucose Uptake in Skeletal Muscle

The marked increase in type 2 diabetes worldwide (1) emphasizes the importance of novel treatments that rectify glucose homeostasis. Ideally, this involves signaling pathways that are not dependent on insulin. Although activation of adrenoceptors in vitro has been shown to stimulate glucose uptake in skeletal muscle (2–10), the signaling pathways involved and the potential for treating type 2 diabetes have been unclear. In vivo, sympathetic effects on glucose homeostasis are complex, being influenced by glucose outflow from the liver, insulin release from the pancreas, as well as glucose uptake into peripheral tissues such as white fat, brown fat, and muscle. Long-term activation of β₂-adrenoceptors in rats causes skeletal muscle hypertrophy (11), leading to decreased plasma insulin levels, increased insulin sensitivity, and improved glucose tolerance (12,13); in humans, there are beneficial metabolic changes (14,15), although some β-blockers exacerbate diabetes (16). Despite this circumstantial evidence (2–10), the mechanisms involved in adrenergic facilitation of glucose uptake in peripheral tissues and their potential therapeutic role are poorly understood.

Skeletal muscle is the most important organ in the body for glucose clearance and utilization (17). High sympathetic nervous system activity (e.g., exercise, fight, or flight) is associated with enhanced glucose uptake into skeletal muscle, resulting primarily from norepinephrine release from adrenergic nerve terminals (18). Catecholamines released during sympathetic activation stimulate adrenoceptors at the cell surface. Mechanisms that have been proposed to explain adrenergically mediated glucose uptake include stimulation of the transcription and translation of proteins, non–carrier mediated and unspecific effects, and translocation or activation of GLUTs (10,19,20), emphasizing the lack of a unifying concept for the mechanisms involved. Previously, we showed that β-adrenoceptors profoundly affect glucose uptake by an AMPK-independent pathway (3) that is not secondary to the lowering of ATP levels or energy deficiency, inasmuch as activation also increases glycogen formation in L6 skeletal muscle cells (4,21). Based on these studies and others (22), we have examined the mechanism used by β-adrenoceptors to increase glucose uptake in skeletal muscle and whether activation of this pathway could improve the condition of individuals with type 2 diabetes.

We describe here for the first time that specific activation of β₂-adrenoceptors in skeletal muscle, but not white adipocytes, causes translocation of GLUT4 to increase glucose uptake, by a mechanism dependent on mammalian (or mechanistic) target of rapamycin (mTOR) complex 2 (mTORC2). Stimulation of this pathway improves glucose tolerance in rat and mouse models of type 2 diabetes. These studies reveal a physiological pathway and a novel mechanism that challenges accepted views of the role of G-protein–coupled receptor (GPCR) signaling in glucose metabolism. Furthermore, it identifies novel targets that could be exploited for the treatment of type 2 diabetes.

129Sv mice (Charles River Laboratories) were bred at the Stockholm University animal facility. β₁β₂ knockout (KO) mice (The Jackson Laboratory) (23) were of mixed background (129 × 1/SvJ, C57BL/6J, DBA/2, and FVB/N). Male C57BL/6J mice (45–57 g after being fed a high-fat diet for 6 months were provided by Dr. Natasa Petrovic [Stockholm University, Stockholm, Sweden]). The high-fat diet contained 45% fat, 35% carbohydrates, and 20% proteins (catalog no. D12451; Research Diets). Sprague-Dawley rats (male) (Scanbur, Sollentuna, Sweden) and 12-month-old male Goto-Kakizaki (GK) rats (260–390 g) were provided by Professor Gustav Dallner (Stockholm University).

All studies were approved by the North Stockholm Animal Ethics committee (permission N123/09, N69/12, N51/12). Animals were killed by CO₂ under anesthesia.

The glucose tolerance test (24) was performed by intraperitoneal injection of glucose (2 g/kg) (Riedel-de Haën AG, Buchs, Switzerland) in fasted (6 h) GK rats or high-fat diet–fed C57BL/6J mice. Blood glucose levels were measured from a drop of blood obtained from the tail before and after 15, 30, 60, 90, and 120 min of glucose administration. Animals were treated with clenbuterol (Sigma-Aldrich, St. Louis, MO), 30 mg/L, in their drinking water for 4 days, and the glucose tolerance test was performed on the 5th day.

In vivo glucose uptake was performed according to the methods of Liu and Stock (5) with minor modifications. Wild-type (WT) 129Sv mice or β₁/β₂ KO mice were fasted for 5 h and anesthetized with pentobarbital (60 mg/kg i.p.). Mice were injected with KU0063794 (KU) (10 mg/kg) in DMSO or vehicle 10 min before treatment with insulin (1 mg/kg), isoproterenol (1 mg/kg), or saline solution. After 20 min, 130 μCi/kg [³H]-2-deoxy-D-glucose (8 Ci/mmol; PerkinElmer, Waltham MA) was injected intraperitoneally. Animals were killed 1 h later by CO₂, and skeletal muscle (quadriceps) was dissected and lysed in 0.5 mol/L NaOH, and radioactivity was measured by liquid scintillation counting.

Epididymal white fat depots were isolated (25) from Sprague-Dawley rats, digested in buffer (HEPES with 0.2% collagenase, 0.9 g/L glucose, and 1.5% BSA) at 37°C with gentle shaking for 30 min, and filtered (nylon mesh, 250 μm). Mature adipocytes were collected from the top of the suspension. Cells were washed with DMEM containing 3% BSA. Mature adipocytes were incubated with agonists in glucose-free DMEM containing 3% BSA for up to 1 h at 37°C. After incubation, [³H]-2-deoxy-d-glucose (50 nmol/L) was added to the cells for a further 60 s at 37°C. Cells were then separated from the incubation medium by centrifugation though a layer of silicone oil, and frozen at −20°C (26). Frozen cell cakes were removed and transferred to scintillation vials for counting.

Soleus muscle was dissected from Sprague-Dawley rats and suspended with hooks stretching the muscle very gently in water-jacketed organ baths containing 30 mL Krebs-Henseleit bicarbonate (KHB) buffer (118.5 mmol/L NaCl, 4.7 mmol/L KCl, 1.2 mmol/L KH₂PO₄, 25 mmol/L NaHCO₃, 2.5 mmol/L CaCl₂, 1.2 mmol/L MgSO₄, and 5 mmol/L HEPES, pH 7.4) containing 5 mmol/L glucose and 15 mmol/L mannitol, bubbled with 95% O₂/5% CO₂, and maintained at 37°C. Tissues were equilibrated for 5 min before KU (100 nmol/L) or vehicle was added for 30 min, followed by the addition of either insulin (100 nmol/L) or isoproterenol (100 nmol/L) for 1 h. Muscles were rinsed with KHB buffer (containing 20 mmol/L mannitol) for 10 min, then incubated in KHB buffer (containing 8 mmol/L 3-O-methylglucose, 12 mmol/L mannitol, 438 µCi/mmol 3-O-methyl[³H]glucose [PerkinElmer; 80.2 Ci/mmol] and 42 µCi/mmol [¹⁴C] mannitol [PerkinElmer; 58.8 mCi/mmol]) for 12 min. Muscles were rinsed with PBS, frozen in liquid nitrogen, weighed, and dissolved in 1 mL of 0.5 mol/L NaOH at 60°C. ³H and ¹⁴C radioactivity was measured by liquid scintillation counting. Total muscle 3-O-methylglucose and extracellular space were measured as described previously (23) and were expressed as a rate of 3-O-methylglucose transport per milliliter of intracellular water per hour.

L6 cells (ATCC, Manassas, VA) or L6-GLUT4-myc cells (Professor Amira Klip, Hospital for Sick Children, Toronto, ON, Canada) were grown in DMEM supplemented with 4 mmol/L l-glutamine, 10% FBS, 100 units/mL penicillin, 100 μg/mL streptomycin, and 10 mmol/L HEPES, in a 37°C incubator with 8% CO₂. Differentiation was induced by growing cells to ∼90% confluence and lowering of FBS to 2% for 7 days. Human primary skeletal muscle cells (SKMCs) were purchased from Karocell AB (Stockholm, Sweden), Lonza (Basel, Switzerland), and PromoCell GmbH (Heidelberg, Germany), and were grown in Hams F-10 nutrient mixture containing 20% heat-inactivated FBS, 2 mmol/L l-glutamine, 50 units/mL penicillin, and 50 μg/mL streptomycin. Differentiation was initiated by growing cells to ∼90% confluence and reducing FBS to 4% for 3 days then to 2% for 4 days. Prior to experiments, cells were kept in serum-free media overnight. All cell media were purchased from Sigma-Aldrich (St. Louis, MO).

Cells were incubated overnight in serum-free medium (0.5% BSA), and [³H]-2-deoxy-d-glucose (50 nmol/L, 8 Ci/mmol; PerkinElmer) uptake was measured as previously described (6). Cells were pretreated with inhibitors for 30 min, and then exposed to insulin, isoproterenol, or 8-Br-cAMP for 2 h, unless otherwise stated. Radioactivity was detected by liquid scintillation counting.

L6 cells were harvested in 1 mL Ultraspec RNA isolation reagent (Biotecx Laboratories, Houston, TX), and RNA was isolated according to the manufacturer’s instructions. Total RNA was dissolved in DEPC-water (Invitrogen, Carlsbad, CA) and quantified in a DU-50 Beckman (Fullerton, CA) spectrophotometer. Total RNA was reverse transcribed (Applied Biosystems, Carlsbad, CA) using random hexamers for first-strand cDNA synthesis. Primers for GLUT1 (forward: AGAACCGGGCCAAGAGT; reverse: GAACAGCTCCAAGATGGTGAC), GLUT4 (forward: TTGCAGTGCCTGAGTCTTCTT; reverse: CCAGTCACTCGCTGCTGA), and general transcription factor IIB (forward: GTTCTGCTCCAACCTTTGCCT; reverse: TGTGTAGCTGCCATCTGCACTT) as endogenous controls were designed by Universal Probe Library (Roche Applied Science, Penzberg, Germany) and were purchased from Invitrogen. SYBR Green PCR Master Mix (Applied Biosystems) was mixed with reference dye and primers, and loaded onto 96-well MicroAmp plates (Applied Biosystems). A 2-μL cDNA template was added in duplicate. General transcription factor IIB was used as a control.

L6 myotubes were used in all immunoblotting experiments. Cells were serum starved overnight before the experiment. Immunoblotting was performed as previously described (27,28). Primary antibodies (Akt, phospho-Akt threonine 308 [T308], phospho-Akt serine 473 [S473], phospho-AS160 threonine 642 [T642], phospho-TBC1D1 [Thr590], GLUT4, GLUT1, Myc-tag, phospho-mTOR S2448, phospho-mTOR S2481, phospho-p70-S6 kinase (p70S6K), Raptor, and Rictor, diluted 1:1,000) were all from Cell Signaling Technology (Danvers, MA) and were detected using a secondary antibody (horseradish peroxidase–linked anti-rabbit IgG; Cell Signaling Technology) diluted 1:2,000 and measured using enhanced chemiluminescence (Amersham Biosciences).

L6 or SKMCs were grown in 4- or 8-well culture chamber glass-bottom slides (BD Biosciences, Franklin Lakes, NJ), treated with drugs, washed in DMEM, fixed with 4% formaldehyde in PBS for 5 min, quenched with 50 mmol/L glycine in PBS for 10 min, blocked with 5% BSA in PBS for 1 h, and incubated with GLUT4 (Santa Cruz Biotechnology) or Myc-tag antibody (Cell Signaling Technology) (1:200 dilution in 1.5% BSA in PBS) overnight at 4°C. Cells were washed in PBS and incubated with Alexa Fluor 488–conjugated goat anti-rabbit IgG, Alexa Fluor 555–conjugated goat anti-rabbit IgG, or Alexa Fluor 488–conjugated rabbit anti-goat IgG secondary antibody (1:500 dilution, 1.5% BSA in PBS) for 1 h. Phosphatidylinositol 3,4,5-trisphosphate (PIP₃) was visualized in permeabilized L6 cells (0.5% Triton X-100 in water, 15 min) and blocked with 10% rat serum in Tris-buffered saline (TBS) overnight at 4°C. Cells were incubated for 1 h at 37°C with mouse anti-PIP₃ conjugated to a biotinylated goat anti-mouse IgM (diluted 1:100) and washed three times (10% rat serum in TBS), and a secondary antibody was added (streptavidin-Alexa Fluor 488, 1:2,000 in TBS, 30 min, 37°C). Slides were mounted in ProLong Gold antifade reagent (Invitrogen), and images were observed in an inverted laser-scanning microscope (LSM 510 META; Carl Zeiss Microscopy, Jena, Germany). Laser power, offset, and gain including other conditions were the same for each experiment. Stained cells were quantified using ImageJ (National Institutes of Health).

L6 cells were transfected with GLUT4-myc-GFP constructs (from Professor Jeffrey E. Pessin, Albert Einstein College of Medicine, New York, NY) or small interfering RNA (siRNA) (Qiagen, Hilden, Germany) against rat GLUT4, Rictor, or Raptor. Scrambled siRNAs were used as control. Myotubes or myoblasts were detached using trypsin/EDTA, transferred to Eppendorf tubes and centrifuged at 1,000g for 3 min, resuspended in 20 μL SE Cell Line Nucleofector solution for L6 cells and P1 primary Nucleofector solution for SKMCs (Lonza), and 100 pmol siRNA (L6 myotubes) or 0.8 μg GLUT4-myc-GFP construct (L6 myoblasts and human SKMCs) were added. L6 cells were nucleofected by using the DS-137 program and SKMCs with the DS-138 program in the Lonza 4D-Nucleofector. Cells were electroporated in 16-well microcuvette plates (Lonza), prewarmed RPMI 1640 medium was added (80 μL), and then cells were transferred to 1.5-mL microcentrifuge tubes or glass-bottom slides (BD Biosciences) containing DMEM with 10% FBS for 8 h. Medium was changed to serum-free medium (0.1% BSA), and cells were incubated overnight. The transfection did not affect cell viability assessed by trypan blue staining (data not shown). Because myotubes do not easily reattach, these cells were kept in suspension in microcentrifuge tubes, whereas myoblasts (used for confocal experiments) were seeded on slides. Twenty-four hours after transfection, glucose uptake, immunohistochemistry, or Western blotting was performed.

On the day of the experiment, transfected cells were stimulated with insulin (1 μmol/L) or isoproterenol (1 μmol/L) for 2 h in microcentrifuge tubes at 37°C. Cells were centrifuged and pelleted (1,000g, 2 min), media were carefully removed and changed to glucose-free media, and insulin and isoproterenol were added again. After incubation for 15 min, 50 nmol/L 2-deoxy-d-[1-³H] glucose (8 Ci/mmol) was added for 10 min. Uptake of 2-deoxy-d-[1-³H] glucose was terminated by washing with ice-cold PBS twice with centrifugation (1,000g, 30 s), and 400 µL of 0.2 mol/L NaOH was added to lyse the cells. Radioactivity was measured by liquid scintillation counting.

Experiments were performed in duplicate, and the results are expressed as the mean ± SEM. The statistical significance of differences between groups was analyzed by Student paired two-tailed t test, or, for more than two groups, one-way or two-way ANOVA was used (*P < 0.05, **P < 0.01, ***P < 0.001). Data are presented as the increase expressed as the percent of basal levels, if not otherwise stated.

GK rats and high-fat diet–fed C57BL/6J mice are two established models of type 2 diabetes (Fig. 1A and B) that were treated with the β₂-adrenoceptor agonist clenbuterol (30 mg/L in the drinking water, 4 days) or normal drinking water prior to a glucose tolerance test. Clenbuterol-treated GK rats had greatly improved glucose tolerance after 60, 90, and 120 min, and C57BL/6J mice after 30 and 60 min of glucose administration compared with untreated animals (Fig. 1A and B), demonstrating that β₂-adrenoceptor stimulation greatly improves glucose tolerance in two different models of type 2 diabetes. Clenbuterol, like isoproterenol and insulin, also increased glucose uptake in L6 myotubes (Supplementary Fig. 1A). We then investigated whether this effect results from actions in glucose uptake in tissues such as white fat and skeletal muscle. The general β-adrenoceptor agonist isoproterenol was used to rule out effects on α₁-adrenoceptors that increase glucose uptake via an AMPK-mediated mechanism (3). Isoproterenol increased glucose uptake in vivo in WT mouse skeletal muscle to a greater extent than insulin, and this effect was absent in β₁/β₂-adrenoceptor KO mice (Fig. 1C), indicating that the effect of isoproterenol is mediated by β-adrenoceptors (almost certainly β₂-adrenoceptors, see 25Discussion). Glucose uptake was also stimulated by isoproterenol ex vivo in intact rat soleus muscle (Fig. 1D), but not in rat isolated white adipocytes, showing that the pathway is present in skeletal muscle but not in white adipocytes (Fig. 1E).

The potential relevance of these findings to humans was tested in human primary SKMCs, where isoproterenol stimulation increased glucose uptake to a greater extent than insulin (Fig. 1F). L6 cells were used to further investigate the mechanism of β₂-adrenoceptor–mediated glucose uptake. Isoproterenol increased glucose uptake in both L6 myoblasts and myotubes (Fig. 1G and H), with the effect in myotubes being more than twice that in myoblasts (41.5 vs. 18 fmol/well). The effect of isoproterenol was concentration dependent, and displayed the same efficacy and potency as insulin (Supplementary Fig. 1B), with the effect measureable by 30 min and peaking by 2 h (Supplementary Fig. 1C).

Insulin-mediated glucose uptake occurs by activation of a well-characterized signaling pathway involving phosphoinositide 3-kinase (PI3K), Akt, and AS160 (29). To investigate whether β₂-adrenoceptor–mediated glucose uptake requires PI3K, we used the PI3K-phosphatidyl-inositol-3-kinase–like (PIKK) inhibitor PI-103. A role for PI3K or related kinases was suggested by the inhibition of both insulin- and isoproterenol-stimulated glucose uptake by PI-103 (half-maximal inhibitory concentration 0.7 μmol/L) (Fig. 2A). However, Akt, a downstream target of PI3K activation, was clearly not phosphorylated at T308 or S473 after β₂-adrenoceptor stimulation (in contrast to insulin that increased Akt phosphorylation at both sites at all time points examined) (Fig. 2B). The Akt inhibitor X (10 μmol/L) or SH-6 (1 μmol/L), which prevent Akt activation by interfering with its phosphatidyl-binding domain (30,31), both inhibited insulin-stimulated glucose uptake, Akt phosphorylation, and AS160 phosphorylation, but had no effect on β₂-adrenoceptor–mediated glucose uptake (Fig. 2C–E and Supplementary Fig. 2A and B). TBC1D1 phosphorylation at Thr590 was increased only after insulin, but not isoproterenol, stimulation (Supplementary Fig. 2C). Insulin, but not isoproterenol, increased PIP₃ content, and this effect was inhibited by LY294002, a widely used PI3K inhibitor (32) (Fig. 2F). These results show that β₂-adrenoceptor–stimulated glucose uptake does not use major components of the insulin pathway.

To further investigate β-adrenoceptor–mediated glucose uptake, we used cAMP analogs because β₂-adrenoceptor activation increased cAMP levels in L6 cells (Supplementary Fig. 2D). Glucose uptake was increased by the cell-permeable cAMP analogs 8-bromoadenosine 3′,5′-cAMP (8-Br-cAMP, 1 mmol/L), N6,2′-O-dibutyryladenosine 3′,5′-cAMP (db-cAMP, 1 mmol/L), and cAMP (2 mmol/L), whereas the cell-impermeable analog 8-hydroxyadenosine- 3′,5′-cAMP (8-OH, 1 mmol/L) had no effect (Fig. 3A). This suggested that increases in cAMP levels were associated with increased glucose uptake in L6 cells. Because the PI3K inhibitor PI-103 also inhibits other PIKK family kinases (33,34), including mTOR, we examined more selective inhibitors. The mTOR inhibitors Torin-1 and KU blocked β₂-adrenoceptor–mediated and cAMP-mediated glucose uptake in L6 myotubes (Fig. 3B and C). KU also inhibited isoproterenol-mediated glucose uptake in intact rat soleus muscle ex vivo and in mouse skeletal muscle in vivo (Fig. 3D and E).

mTOR exists as two complexes: mTORC1 that is associated with Raptor and is phosphorylated at S2448 and mTORC2 that is associated with Rictor and is phosphorylated at S2481, with these two phosphorylation sites reflecting mTOR activity (35). In L6 cells, isoproterenol increased phosphorylation only at S2481 in a time-dependent manner (Supplementary Fig. 3A), whereas insulin phosphorylated mTOR at both S2448 and S2481. While S2448 was phosphorylated within 5 min of insulin stimulation, isoproterenol did not affect phosphorylation at this site at any time point examined (Supplementary Fig. 3A). mTOR phosphorylation in response to isoproterenol or insulin was blocked by KU (Fig. 4A). A downstream target of mTORC1 (p70S6K) was phosphorylated slightly after isoproterenol, although to a lesser extent than insulin (Supplementary Fig. 3B). Akt inhibitor X decreased insulin-induced phosphorylation of mTOR at S2448, but not isoproterenol- or insulin-induced phosphorylation of mTOR at S2481 (Supplementary Fig. 3C), indicating that Akt is not upstream of mTOR S2481. 8-Br-cAMP (1 mmol/L) or the cAMP-dependent protein kinase (PKA)-selective cAMP analog 6-Benz-cAMP (1 mmol/L) also caused phosphorylation of mTOR at S2481 (Fig. 4A and B), mimicking the effect of isoproterenol and indicating that this site is regulated by the cAMP-PKA pathway. The PKA inhibitor (PKI) 14–22 partially blocked the isoproterenol effect and completely blocked the effect of 8-Br-cAMP on S2481, while not affecting insulin-mediated S2481 phosphorylation (Fig. 4C). Because mTOR is a key factor in the β₂-adrenoceptor pathway, we examined the complex involved. mTORC1 contains the regulatory-associated protein of mTOR (Raptor), whereas mTORC2 contains the rapamycin-insensitive companion of mTOR (Rictor). Short-term (30 min) treatment with the mTORC1 inhibitor rapamycin had no effect on isoproterenol and 8-Br-cAMP–stimulated glucose uptake (Fig. 4D), but long-term exposure (24 h), which inhibits the mTORC2 assembly, inhibited isoproterenol-stimulated glucose uptake, suggesting the involvement of mTORC2 (Fig. 4E). Rictor siRNA treatment markedly depleted Rictor (Fig. 4F) and abolished isoproterenol-stimulated glucose uptake (Fig. 4G), whereas Raptor siRNA or control siRNAs had no effect. This clearly demonstrated that mTORC2 is a key factor involved in β₂-adrenoceptor–mediated glucose uptake in skeletal muscle.

Glucose uptake can be increased by de novo synthesis of GLUTs associated with increases in transcription and/or translation. The general transcription inhibitor actinomycin D (2 μg/mL) did not significantly affect basal glucose uptake or cell viability (not shown), or insulin- or isoproterenol-mediated glucose uptake at 2 h (Fig. 5A), indicating that changes in transcription are not involved. Accordingly, GLUT1 and GLUT4 mRNA levels were not changed by isoproterenol (1 μmol/L) (Fig. 5B and C). The general translational inhibitor cycloheximide (50 μmol/L) did not affect basal glucose uptake or cell viability or either insulin- or isoproterenol-mediated glucose uptake at 2 h (Fig. 5D). Neither GLUT1 nor GLUT4 protein levels were altered by isoproterenol (Fig. 5E and F), suggesting that β₂-adrenoceptor–stimulated glucose uptake is not dependent upon de novo GLUT1 and GLUT4 synthesis at time points up to 2 h.

We next examined whether β₂-adrenoceptor–stimulated glucose uptake involved GLUTs. Both of the GLUT inhibitors cytochalasin B (10 μmol/L) and phloretin (100 μmol/L) (36) reduced basal glucose uptake (Supplementary Fig. 4A) and inhibited glucose uptake to isoproterenol and insulin (Supplementary Fig. 4B and C), suggesting that GLUTs were involved. The selective GLUT4 inhibitor indinavir (100 μmol/L) blocked basal glucose uptake by ∼40% (Supplementary Fig. 4A), which is consistent with other studies in L6 myotubes (37), and markedly reduced both β₂-adrenoceptor–stimulated and insulin-stimulated glucose uptake, after both 45 and 120 min of stimulation (Supplementary Fig. 4D and E), suggesting a role for GLUT4.

To confirm that β₂-adrenoceptor–stimulated glucose uptake was carrier mediated, kinetic studies were performed with increasing concentrations of 2-deoxyglucose (38). Insulin increased glucose uptake (Fig. 5G) with K_m values (Eadie-Hofstee plot) in agreement with previously reported values (10,39). β₂-Adrenoceptor stimulation increased glucose uptake V_max following classical saturable Michaelis-Menten kinetics with a K_m value (2.5 ± 0.3 mmol/L) similar to that of insulin (Supplementary Fig. 4F), suggesting that β₂-adrenoceptor–stimulated glucose uptake involves transporter proteins (40). To confirm the involvement of GLUT4, siRNA directed against GLUT4 was used and markedly reduced β₂-adrenoceptor–mediated glucose uptake compared with a scrambled siRNA (Fig. 5H). Thus, GLUT4 is the transporter responsible for β₂-adrenoceptor–mediated glucose uptake. Because GLUT4 is involved, we investigated the role of actin and cytoskeletal rearrangement using latrunculin B (Fig. 5I), which inhibited both insulin- and isoproterenol-mediated glucose uptake, indicating that actin polymerization was involved and suggesting a role for GLUT4 translocation.

Direct visualization of GLUT4 translocation to the plasma membrane was examined by immunohistochemistry in nonpermeabilized L6 cells and was observed after β₂-adrenoceptor stimulation in native L6 myoblasts, and in L6 myoblasts transiently transfected with GLUT4-myc-GFP or stably transfected with GLUT4-myc (Fig. 6A and B and Supplementary Fig. 5A). This was confirmed in L6 myotubes and in stably transfected GLUT4-myc L6 myotubes (Supplementary Fig. 5B and C). β-Adrenoceptor stimulation also increased GLUT4 translocation in human native primary SKMCs (Fig. 6C) and in SKMCs transiently transfected with GLUT4-myc-GFP (Fig. 6D). GLUT4 translocation to isoproterenol was abolished by mTOR inhibition (Fig. 6E). β-Adrenoceptor–stimulated glucose uptake is therefore dependent on mTORC2 and on GLUT4 translocation.

There is increasing recognition that glucose homeostasis can be regulated independently of insulin receptors by the activation of GPCRs. Typical GPCRs such as adrenoceptors facilitate glucose uptake in skeletal muscle, and we have demonstrated that activation of α₁- and β₂-adrenoceptors in skeletal muscle increases glucose uptake, although the mechanisms are not fully understood (6,7). This has led us to determine the signaling pathways used by β₂-adrenoceptors in skeletal muscle to increase glucose uptake because their activation improves glucose tolerance in diabetic GK rats and obese C57BL/6J mice (Fig. 1). Although this could result from a variety of mechanisms in different tissues, our combined results suggest that the major part of the improvement emanates from β₂-adrenoceptor activation of glucose uptake in skeletal muscle (with β-adrenoceptors not increasing glucose uptake in mature white adipocytes). We suggest that sympathetic stimulation shunts glucose from liver to skeletal muscle to provide energy for muscular action rather than for storage in white fat. We believe that this event has high physiological and pathophysiological importance, given the major role of skeletal muscle in glucose uptake.

We ruled out a role for β₁- or β₃-adrenoceptors, because the in vivo effect of β-adrenergic agonists on glucose uptake in skeletal muscle is absent in β₁/β₂ KO mice and the selective β₂-adrenoceptor agonist clenbuterol displayed the same efficacy as isoproterenol in L6 muscle cells. Previous studies show that β₂-adrenoceptors are the primary target for catecholamines in skeletal muscle (6,41), and studies in L6 cells demonstrate that β-adrenoceptor agonist effects are mediated by β₂-adrenoceptors (6), clearly suggesting that the effects of β-adrenergic agonists on glucose uptake are mediated primarily by β₂-adrenoceptors.

We show here that β₂-adrenoceptor agonists acutely stimulate glucose uptake with similar efficacy, potency, and time course to insulin in L6 myotubes (Supplementary Fig. 1B and C). Initially, our assumption was that β-adrenoceptors increase glucose uptake by pathways similar to those used by insulin, because responses were blocked by PI3K inhibitors. However, unlike insulin, β₂-adrenoceptor stimulation failed to phosphorylate Akt at either S473 or T308 at all time points examined, selective Akt inhibitors failed to inhibit β₂-adrenoceptor–mediated glucose uptake, and β₂-adrenoceptor stimulation failed to increase PIP₃ levels in L6 cells, suggesting that PI3K was not involved in the β-adrenoceptor pathway. Examination of the selectivity of these inhibitors shows that they also inhibit other PIKK family kinases (33,34), including mTOR, which also plays an important role in insulin signaling (Fig. 7). Specific inhibition of mTOR by KU confirmed its role in insulin signaling but also revealed that it was necessary for β-adrenoceptor–stimulated glucose uptake. However, the pattern of mTOR phosphorylation in response to β-adrenoceptor stimulation differed from that of insulin, with phosphorylation only at S2481, whereas insulin caused phosphorylation at both S2448 and S2481. Phosphorylation at S2481 is associated with mTORC2, whereas phosphorylation at S2448 is associated with mTORC1 (35), suggesting that insulin regulates both mTOR complexes, whereas β-adrenoceptor activation regulates only mTORC2. Treatment with rapamycin and siRNA knockdown of Raptor or Rictor reveal the involvement of mTORC1 and mTORC2, and we show that knockdown of Rictor, but not Raptor, markedly inhibits both insulin-mediated and β-adrenoceptor–mediated glucose uptake, confirming mTORC2 as a key regulator of glucose uptake in skeletal muscle. Thus, although the β-adrenoceptor pathway differs from the insulin pathway at the level of Akt and PIP₃, both pathways involve mTORC2. The involvement of mTOR was confirmed ex vivo and in vivo where KU inhibited both insulin-stimulated and β-adrenoceptor–stimulated glucose uptake. Recent studies indicate a role for mTORC2 (Rictor) in glucose homeostasis, because either skeletal muscle or adipose-specific ablation of Rictor depresses insulin-stimulated glucose uptake in skeletal muscle (42) or adipose tissue (43), respectively, and in both mouse models impairs glucose tolerance in vivo.

The regulation and cross-talk among insulin, mTORC1 (Raptor), mTORC2 (Rictor), and Akt is complex (Fig. 7). Insulin mediates increases in PI3K activity, PIP₃ recruits inactive Akt, and phosphoinositide-dependent kinase-1 (PDK1) to the plasma membrane via their NH₂-terminal PH domain, allowing Akt phosphorylation at T308 by PDK1, and in parallel PI3K phosphorylates mTORC2 at S2481. The subsequent conformational change in Akt allows mTORC2 to phosphorylate Akt at S473, thereby fully activating Akt, resulting in mTORC1 phosphorylation at S2448 and subsequent AS160 and GLUT4 translocation. In contrast, β₂-adrenoceptors activate mTORC2 in a PI3K-Akt–independent manner. Because β₂-adrenoceptors fail to activate PI3K (observed as no increases in PIP₃ levels), PDK1 and Akt cannot, as discussed for insulin above, be recruited to the plasma membrane. This prevents phosphorylation of Akt at T308, which is absolutely required for mTORC2-mediated phosphorylation of Akt at S473, and subsequent phosphorylation of AS160. Our results demonstrate that β₂-adrenoceptor activation leads to phosphorylation of mTOR at S2481 via a PKA-mediated mechanism that is independent of PI3K-Akt (Fig. 7).

This is the first demonstration that GPCRs can specifically stimulate mTORC2 in mammalian cells. The only other evidence that GPCRs can activate mTORC2 is found in the soil-living amoeba Dictyostelium discoideum, where chemotactic activation of GPCRs and mTORC2 (44–46) occurs, suggesting that GPCR stimulation of mTORC2 could be of ancient origin.

There is, however, some evidence that other GPCRs can activate mTORC1 independently of Akt/PI3K, including the Gαq/11-coupled α₁-adrenoceptors in cardiomyocytes (47), the Gαq/11-coupled prostaglandin F2α receptor, and the Gαs-coupled luteinizing hormone receptor in the corpus luteum (48). There has been great attention focused on the regulation and downstream targets of mTORC1 that can be pharmacologically inhibited by rapamycin. mTORC1 regulates protein translation and synthesis by activating proteins such as p70S6K. Although our study shows, both pharmacologically and by gene knockdown, that β₂-adrenoceptors do not use mTORC1 for glucose uptake, they may activate mTORC1 to some degree (as measured as the phosphorylation of p70S6K). The activation of mTORC1 might reflect the long-term hypertrophic effects of β₂-adrenergic agonists on skeletal muscle, which was not a focus of this study.

Compared with mTORC1, little is known of the downstream targets of mTORC2, but they do include pathways involved in actin cytoskeleton organization. There is evidence for a Rac-dependent (Akt-independent) arm in skeletal muscle that regulates cytoskeletal rearrangement involved in insulin signaling to glucose uptake (49,50). Such pathways should be investigated further in connection with mTORC2 and adrenergic signaling. Both β₂-adrenoceptor (51) and mTORC2 activation cause acute actin cytoskeleton reorganization (52), and the inhibition of this process by latrunculin B blocks β₂-adrenoceptor–stimulated glucose uptake. Taken together, our findings identify a pathway involving the activation of mTORC2 by a mammalian GPCR, independently of PI3K/Akt, and show that mTORC2 is an important master regulator of adrenergic-mediated glucose uptake, likely through cytoskeletal rearrangement.

Glucose uptake in skeletal muscle is the rate-limiting step for whole-body glucose metabolism (17). Glucose uptake in response to β-adrenoceptor stimulation has an efficacy similar to insulin, is saturable, and is blocked by GLUT inhibitors and by pretreatment with GLUT4 siRNA. These results show for the first time that GLUT4 is required for β-adrenoceptor–stimulated glucose uptake, occurs independently of changes in transcription or translation (again suggesting that mTORC1 is not involved), and involves translocation of GLUT4 to the plasma membrane, a process generally considered to be predominantly insulin regulated. Thus, the adrenergic pathway uses two crucial components of the insulin pathway: mTORC2 and GLUT4. However, in contrast to insulin signaling, β₂-adrenoceptor activation does not lead to activation of mTORC1, Akt, or the downstream phosphorylation of the Rab-GTPase–activating protein AS160.

Glucose uptake in type 2 diabetes is associated with defects in PI3K activity, insulin receptor tyrosine, insulin receptor substrate, and Akt phosphorylation, causing impairment of GLUT4 trafficking and glucose uptake. Thus, there is normally no defect per se with GLUT4 trafficking, rather defects in the insulin signaling mechanisms regulating GLUT4 trafficking (53) and these defects are bypassed by the β₂-adrenoceptor-mTORC2 pathway. This was illustrated in two different diabetic muscle models in vivo, where β₂-adrenoceptor agonists improved glucose tolerance. However, current β₂-adrenoceptor agonists normally display a number of side effects (including myocardial hypertrophy) (54), making them unsuitable for the treatment of type 2 diabetes. Insights obtained from the β₂-adrenoceptor–mediated pathway to mTORC2 and glucose uptake opens up new avenues in the search for novel approaches and drugs to treat type 2 diabetes.

In conclusion, we describe a novel physiological pathway involving β₂-adrenoceptors, PKA, mTORC2, and GLUT4 that promotes glucose uptake in both healthy and type 2 diabetic skeletal muscle. This study reveals new and exciting approaches for our understanding of the role of adrenergic mechanisms in metabolic health and disease.

Acknowledgments. The authors thank Ashwini Kodavali and Anna-Stina Höglund (Stockholm University, Stockholm, Sweden) for technical help. The authors also thank Professor Jan Nedergaard and Professor Barbara Cannon (Stockholm University) and Dr. Bronwyn Evans (Monash University, Parkville, VIC, Australia) for valuable discussions and help. In addition, the authors thank Professor Jeffrey E. Pessin at the Albert Einstein College of Medicine (New York, NY) for GLUT4 constructs, and Professor Amira Klip at the Hospital for Sick Children (Toronto, ON, Canada) for providing L6-GLUT4-myc cells. Torin1 was kindly provided by D.M. Sabatini (Whitehead Institute for Biomedical Research, Cambridge, MA). The authors thank Professor Gustav Dallner (Stockholm University) for providing them with the GK rats and Dr. Natasa Petrovic (Stockholm University) for providing them with high-fat diet–induced diabetic mice.

Funding. M.S. is supported by the Wenner-Gren Foundation, Australian Research Council Linkage International Fellowship LX0989791, and National Health and Medical Research Council (NHMRC) CJ Martin Overseas Biomedical Fellowship 606763. N.D. is supported by the Swedish Society for Medical Research. A.I.Ö. is supported by the Henning and Johan Throne-Holst Foundation. R.J.S. is supported by NHMRC Program Grant 519461. D.S.H. is supported by NHMRC Career Development Fellowship 545952. T.B. is supported by the Ventenskapsrådet-Medicin (VR-M) from the Swedish Research Council, Novo Nordisk Fonden, Stiftelsen Svenska Diabetesförbundets Forskningsfond, the Magnus Bergvall Foundation, and the Carl Tryggers Foundation.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. M.S., N.D., and A.I.Ö. wrote the manuscript and researched data. O.S.D., A.L.S., J.M.O., and R.I.C. researched the data. R.J.S. and T.B. reviewed and edited the manuscript. D.S.H. reviewed and edited the manuscript and researched the data. T.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.