AMP Kinase-Induced Skeletal Muscle Glucose But Not Long-Chain Fatty Acid Uptake Is Dependent on Nitric Oxide

Male Sprague-Dawley rats (Harlan Industries, Indianapolis, IN) were housed individually and maintained at 23°C on a 0600–1800 light cycle. Rats were fed chow (Purina Nestlé, St. Louis, MO) ad libitum and given free access to water. The rats were housed under these conditions for ∼1 week, by which time their weights had reached ∼300 g. After weight gain, rats were randomly divided into each of the experimental groups (n = 7 per group). All procedures were approved by the Vanderbilt University Animal Care and Use Subcommittee and followed the National Institutes of Health guidelines for the care and use of laboratory animals.

Surgical procedures were performed as previously described for arterial and venous catheterizations (19). Briefly, animals were anesthetized with a 50:5:1 vol/vol/vol mixture of ketamine, rompun, and acepromazine, and the left common carotid artery and right jugular vein were catheterized with PE50. Catheters were exteriorized and secured at the back of the neck, filled with heparinized saline (150 units/ml), and sealed with a stainless steel plug. Immediately postsurgery, each animal received 75 mg/kg of ampicilliin s.c. to prevent infection. After surgery, animal weights and food intake were monitored daily, and only animals in which presurgery weight was restored were used for experiments.

After 4 days of recovery, rats were given either drinking water (n = 14) or water containing 1 mg/ml l-NAME for 2.5 days. Assessment of water consumption showed rats consumed 35–40 ml of water or ∼120–160 mg l-NAME · kg⁻¹ · day⁻¹. Previous studies have shown that this dose elicits a near maximal but reversible inhibition of NOS (20–22). This method of administration was chosen because l-NAME administration by acute intravenous injection results in increases in body temperature (23).

Glucose and LCFA tracers used in the present study were 2-deoxy[³H]glucose ([2-³H]DG) and [¹²⁵I]-15-(p-iodophenyl)-3-R,S-methylpentadecanoic acid ([¹²⁵I]-BMIPP). [2-³H]DG was purchased from New England Nuclear (Boston, MA). BMIPP was a kind gift from Russ Knapp, Oak Ridge International Laboratories (Oak Ridge, TN). Radioiodination was performed according to the manufacturer’s suggested protocol. Briefly, BMIPP was heated in the presence of Na¹²⁵I solution (740 MBq/200 μl), propionic acid, and copper (II) sulfate. Na₂S₂O₃ was then added and the organic phase ether extracted and sequentially back extracted with saturated NaHCO₃ and water. After evaporation, the [¹²⁵I]BMIPP was solubilized using sonication into ursodeoxycholic acid with a final activity of 775 μCi/ml.

Before the study, rats were fasted overnight for 10 h. On the day of the experiment, catheters were flushed with heparinized saline (10 units/ml) and connected to PE50 and silastic tubing for infusions and sampling. Throughout the experimental protocol rats were conscious and unrestrained. The experimental protocol consisted of continuous infusions of saline, AICAR (10 mg · kg⁻¹ · min⁻¹), or AICAR + Intralipid 20% (Intralipid; 0.02 ml · kg⁻¹ · min⁻¹, containing 100 units/ml heparin; Pharmacia & Upjohn, Uppsala, Sweden). All infusions lasted for 150 min. In those rats receiving AICAR, glucose (D50) was clamped to levels found in saline and l-NAME groups (∼6.5 mmol/l). Since the administration of AICAR resulted in a decline in nonesterified fatty acid (NEFA), Intralipid was infused in one protocol to maintain plasma NEFA levels and to ensure that the observed differences in LCFA uptake (R_f) were not concentration dependent. In total this resulted in five experimental groups: saline (n = 7), AICAR (n = 7), l-NAME (n = 7), AICAR + l-NAME (n = 7), and AICAR + Intralipid (n = 7). All infusions were equilibrated for 90 min in which small arterial blood samples (20 μl) were obtained every 10 min throughout the experiment for the measurement of plasma glucose. This provided feedback that was used to adjust glucose infusion rates needed to maintain glycemia. After the equilibration period, rats received an intravenous bolus of [2-³H]DG and [¹²⁵I]-BMIPP for the measurement of glucose and LCFA uptakes, respectively (t = 95 min). The time period from 90 to 120 min is referred to as the “experimental period” in the results section. At time 0 and 90 min, large arterial blood samples (150 μl) were withdrawn for the measurement of insulin, glucose, and NEFA. In addition, small samples (50 μl) were obtained at 97, 100, 110, 120, 130, 140, and 150 min for the measurement of [2-³H]DG, [¹²⁵I]-BMIPP, NEFA, and insulin. To prevent declines in hematocrit, the erythrocytes taken before the isotopic analog infusion were washed in saline and reinfused shortly after each sample was taken. At t = 150 min, rats were anesthetized with pentobarbital sodium, and their soleus (89% type I, 11% type IIa, 0% type IIb) (24), gastrocnemius (6% type I, 23% type IIa, 71% type IIb) (24), and superficial vastus lateralis (SVL) (0% type I, 1% type IIa, 99% type IIb) (24) were excised and rapidly freeze-clamped in liquid nitrogen.

Plasma glucose concentrations were measured by the glucose oxidase method using an automated glucose analyzer (Beckman Instruments, Fullerton, CA), and immunoreactive insulin was measured in samples obtained at t = 0, 90, 120, and 150 min using a double-antibody method (25). NEFA concentrations were determined at t = 0, 90, 110, 130, and 150 min spectrophotometrically using a kit obtained from Wako Chemicals (NEFA-C, Richmond, VA) in the saline, AICAR, l-NAME, and AICAR + l-NAME groups. In the AICAR + Intralipid group, administration of heparin would have resulted in increased in vitro lipolysis due to the presence of lipoprotein lipase. As such, samples used for NEFA analysis in this group were treated with 5 mol/l NaCl as previously described (26,27) before analysis.

[¹²⁵I]BMIPP and [2-³H]DG were measured in the same plasma samples (20 μl) as previously described (28). Briefly, 20 μl plasma was counted for [¹²⁵I]BMIPP using a Beckman Gamma 5500 counter (Beckman, Fullerton, CA). After this, the plasma sample was deproteinized in 100 μl of Ba(OH)₂ and 100 μl of ZnSO₄ and subsequently centrifuged. Supernatant (100 μl) was then diluted in 900 μl of H₂O. ³H radioactivity was counted after addition of fluor (10 ml; Ultimate Gold, Packard Bioscience, Boston, MA) using a Packard Tri-Carb 2900TR Liquid Scintillation Analyzer (PerkinElmer, Boston, MA). The relationship between γ-radioactivity and β-emissions has been established in our laboratory for that specific counter. This relationship was used to correct ³H radioactivity for β-emissions originating from [¹²⁵I] radioactivity (28).

Tissues were analyzed for accumulation of free (2-[³H]DG) and phosphorylated deoxy-[2-³H]glucose (2-[³H]DGP) as previously described (29). Briefly, muscle [¹²⁵I]BMIPP and [2-³H]DG were determined on 50 mg of soleus and 100 mg of gastrocnemius and SVL. After the determination of tissue ¹²⁵I radioactivity, tissue was homogenized in 2 ml of 0.5% perchloric acid and centrifuged for 20 min. Supernatants (1.5 ml) were then neutralized using 5 mmol/l KOH, and 250 μl was counted after the addition of fluor. This fraction provided total radioactivity [2-³H]DG (free + phosphorylated ([2-³H]DGP). In addition, 500 μl of supernatant was treated with 250 μl of Ba(OH)₂ and 250 μl of ZnSO₄ and centrifuged. After this, 500 μl of supernatant was diluted to 1 ml in ddH₂O before fluor (10 ml) was added and samples were counted. Treatment of Ba(OH)₂ and ZnSO₄ removed all but free [2-³H]DG (30). Therefore, [2-³H]DGP is calculated as the difference in radioactivity without (total) and with (free [2-³H]DG) Ba(OH)₂ and ZnSO₄. This analytical approach allows the separation of [2-³H]DGP from the fraction of [2-³H]DG that is incorporated into glycogen, inclusion of which results in the underestimation of R_g (31,32). Sample ³H radioactivity was corrected for β-emissions originating from ¹²⁵I, as described for plasma samples.

To determine the effects of AICAR on the phosphorylation of AMPK, acetyl-CoA carboxylase (ACC), and NOS, gastrocnemius muscles were homogenized in a solution containing 10% glycerol, 20 mmol/l Na-pyrophosphate, 150 mmol/l NaCl, 50 mmol/l Hepes (pH 7.5), 1% NP-40, 20 mmol/l β-glycerophosphate, 10 mmol/l NaF, 2 mmol/l EDTA (pH 8.0), 2 mmol/l phenylmethylsulfonyl fluoride, 1 mmol/l CaCl₂, 1 mmol/l MgCl₂, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 2 mmol/l Na₂VO₃, and 3 mmol/l benzamide. After centrifugation (1 h at 4,500g) the pellets were discarded while supernatants retained for protein determination using a Pierce BCA protein assay kit (Rockford, IL). Proteins (30 μg) were separated on a SDS-PAGE gel and then transferred to a polyvinylidene fluoride membrane. Membranes were blotted with rabbit phospho-AMPK α (Thr172), phosphor-acetyl CoA carboxylase (Ser79), or phosho-endothelial NOS (Ser1177) (1:1,000) (Cell Signaling, Beverly, MA) and then incubated with α-rabbit-horseradish peroxidase (Pierce, Rockford, Il). Densitometry was performed using Lab Image software (Kaplan).

Indexes of glucose and LCFA uptake and clearance: tissue glucose clearance (K_g) and metabolic (R_g) indexes were calculated from the accumulation of [2-³H]DGP in muscle and the integral of the plasma [2-³H]DG concentration after a [2-³H]DG bolus (33,34). The relationships are defined as:

where the subscript “p” refers to mean arterial plasma from t = 90 to 120 min. The measurement of R_g has been described earlier (33,34). In an analogous manner, LCFA clearance (K_f) and metabolic (R_f) indexes were calculated after a [¹²⁵I]BMIPP bolus from the accumulation of [¹²⁵I]BMIPP in muscle and the integral of the plasma [¹²⁵I]BMIPP concentration after the bolus.

[[¹²⁵I]BMIPP]_m is the [¹²⁵I]BMIPP in the cell, [[¹²⁵I]BMIPP]_p is the [¹²⁵I]BMIPP in plasma. The measurement of R_f and K_f has been described earlier (28,35)

To compare differences, ANOVA was performed. To establish differences within the ANOVA, a Student’s Newman Keuls post hoc test was used. Significance levels of P ≤ 0.05 were used, and data are reported as means ± SEM.

Preexperimental values for animal weight, arterial plasma glucose, insulin, and NEFA concentrations are reported in Table 1. Rats in the AICAR + l-NAME groups were lighter compared with the AICAR group with mean values of 281 ± 7 vs. 317 ± 10 g. This was not due to a treatment effect but rather lower presurgery weights. No differences were apparent between groups for plasma glucose, insulin, and NEFA. Hematocrit values were not different between groups before (44 ± 0.4%, t = 0) or after (42 ± 0.7%, t = 150 min) the experimental protocol.

Mean arterial plasma values during the experimental period (t = 90–120 min) are reported in Table 2. No differences in plasma glucose were apparent between groups with time or treatment. In addition, no differences were seen for glucose infusion rates between AICAR, AICAR + l-NAME, and AICAR + Intralipid. Plasma insulin levels were not significantly different among the saline, AICAR, l-NAME, and AICAR + l-NAME groups during the experimental period. In AICAR + Intralipid, plasma insulin levels were higher compared with all other groups. This was not unexpected, as previous studies have shown Intralipid infusion to increase plasma insulin levels due to a decrease in insulin clearance (36). Plasma NEFAs in the AICAR and AICAR + l-NAME groups were significantly lower at all time points in the experimental period compared with baseline values (t = 0). In contrast, NEFAs were stable, and no differences were apparent with time or treatment for saline and l-NAME. In AICAR + Intralipid, baseline NEFAs (t = 0) were lower compared with 150 min; however, no differences were present among saline, l-NAME, and AICAR + Intralipid treatments. NEFA values were lower in AICAR and AICAR + l-NAME compared with saline, l-NAME, and AICAR + Intralipid in the experimental period at all time points measured.

Results of K_g and R_g are depicted in Figs. 1 and 2. Results for R_g were generally similar to those for K_g in that there were no differences in arterial plasma glucose concentrations between groups. AICAR treatment alone increased K_g and R_g in all muscles. l-NAME treatment decreased K_g in all muscles, while R_g decreased in soleus and gastrocnemius with reductions of ∼50 and ∼54%, respectively. The combination of AICAR + l-NAME resulted in no change in R_g or K_g compared with saline in all muscles. However, R_g values were significantly higher compared with l-NAME. In AICAR + Intralipid, R_g and K_g values were not different from AICAR in soleus and SVL. However, AICAR + Intralipid resulted in a greater R_g and K_g versus AICAR alone in the gastrocnemius.

Results of K_f and R_f are depicted in Figs. 3 and 4. AICAR treatment increased K_f in all muscles. R_f was decreased in AICAR due to the decline in NEFA levels. When NEFA levels were maintained in the AICAR + Intralipid group, K_f remained elevated and R_f was increased. This indicates that the effects of AICAR on K_f are not dependent on concentration. Unlike its effect on K_g, l-NAME treatment did not lower K_f compared with saline in any of the muscles studied. The combination of AICAR + l-NAME resulted in K_f values that were greater than those for saline and l-NAME in all muscles. Although there was no statistical difference between AICAR and AICAR + l-NAME, values were slightly lower in both the soleus and gastrocnemius. In contrast, the combination of AICAR + l-NAME resulted in a K_f intermediate between AICAR and l-NAME, showing that AICAR-induced LCFA uptake in the gastrocnemius was not as dependent on NO activation as the soleus and SVL. AICAR + Intralipid had a significantly greater K_f and R_f than AICAR in gastrocnemius, but not in soleus and SVL.

The effects of AICAR on AMPK phosphorylation (AMPK Thr¹⁷⁹), ACC phosphorylation (ACC Ser⁷⁹), and endothelial NOS phosphorylation (NOS Ser¹¹⁷⁷) in gastrocnemius muscle are shown in Fig. 5. AMPK Thr¹⁷² and ACC Ser⁷⁹ phosphorylation was increased in all groups treated with AICAR, demonstrating that the dosage (10 mg · kg⁻¹ · min⁻¹) activated AMPK. NOS Ser¹¹⁷⁷ phosphorylation was increased in AICAR but not AICAR + NAME, l-NAME, or AICAR + Intralipid compared with saline.

As previously mentioned, AICAR resulted in increases in K_g and K_f in all muscles compared with saline (Figs. 2 and 4). When percentage increases in AICAR-induced K_g and K_f were compared between muscle fiber types, results showed AICAR-induced glucose uptake to be greatest in fast-twitch muscles, while AICAR-induced LCFA uptake was greater in muscles comprised of slow-twitch muscle fibers.

The primary objectives of the present study were to determine the fiber type-specific effects of AMPK activation on muscle LCFA and glucose uptake in vivo and the NO dependence of the observed increases. Using isotopic analogs to measure indexes of glucose and LCFA uptake in the presence or absence of the NO synthase inhibitor l-NAME and the AMPK activator AICAR, we found that AMPK activation increased glucose and LCFA clearance. Only muscle glucose uptake was inhibited by l-NAME. This demonstrates that the mechanism by which AMPK increases glucose uptake is NO dependent, while AMPK-induced LCFA uptake occurs independent of this pathway. In addition, we demonstrate that in vivo, the magnitude of the AICAR activation of both glucose and LCFA uptake is fiber-type specific. Soleus that has the greatest percentage of slow-twitch fibers exhibits the greatest increase in LCFA uptake and the smallest increase in glucose uptake.

The finding that l-NAME impedes AMPK-induced glucose uptake is consistent with the hypothesis that NO may act as signal that couples tissue energy demands to glucose delivery. This has been clearly demonstrated for both insulin- and contraction-stimulated glucose uptake. Studies using hyperinsulinemic-euglycemic clamps have shown l-NAME to reduce whole-body glucose disposal by ∼15–30% (14,17). Similarly, moderate exercise in humans in the presence of l-NMMA reduces skeletal muscle glucose uptake by ∼30% in normal subjects and ∼75% in subjects with type 2 diabetes (15). Taken together, these data imply that a percentage of both insulin- and exercise-stimulated glucose uptake can be attributed to NO. Since l-NAME reduced basal and AICAR-stimulated glucose uptake with no effect on fatty acid uptake, the mechanism by which NO regulates glucose uptake must be to some extent distinct from that of LCFA. Indeed, there is in vitro evidence that NO directly regulates glucose transport. Isolated muscle preparations treated with SNP have elevated glucose transport and GLUT-4 translocation via a phosphatidylinositol 3-kinase-independent mechanism (13,37,38). This is thought to occur by an NO-activated guanosine 3′-5′-cyclic monophosphate (cGMP) pathway, although the exact signaling pathway of cGMP-induced glucose trafficking remains to be established (39). Alternative NO targets that could potentially influence glucose uptake include creatine kinase, cytochrome-c oxidase, sacroplasmic reticulum Ca²⁺ ATPase, and glyceraldehyde-3-phosphate (39).

To assess the activation of AMPK and its effects on NOS, immunoblotting for AMK Thr¹⁷², ACC Ser⁷⁹, and NOS Ser¹¹⁷⁷ was performed on gastrocnemius muscle. Results demonstrated that the AICAR dose (10 mg · kg⁻¹ · min⁻¹) was sufficient to activate both AMK Thr¹⁷² and ACC Ser⁷⁹ activity. However, results of NOS Ser¹¹⁷⁷ are not as clearly defined. Corroborating previous findings, we show that AMPK stimulation increases NOS Ser¹¹⁷⁷ compared with saline (40,41). However, no increase in NOS Ser^1,177 was observed in either the AICAR + l-NAME or the AICAR + Intralipid groups. This finding reflects the complex nature of NOS regulation. Numerous kinases, including protein kinase G, protein kinase A, protein kinase B (Akt), and/or mitogen-activated protein kinases are known to act at this site (40,42–46). In addition, AMPK can also phosphorylate an alternate inhibitory site on the NOS protein (Thr⁴⁹⁵) (40). Therefore, a lack of effect of AICAR on NOS Ser¹¹⁷⁷ in these groups may reflect phosphorylation of ACC Thr⁴⁹⁵ or alterations in the ability of AMPK to phosphorylate this site in the presence of l-NAME or Intralipid.

In the present study, lack of effect of l-NAME on basal and AMPK-stimulated LCFA clearance indicates that the coupling of LCFA delivery to uptake is not dependent on NO and must, therefore, be mediated by an alternative signaling intermediate. The finding that AICAR-induced increases in K_f were sustained, even in the presence of l-NAME, was unexpected considering that NO-dependent vascular perfusion was likely to be considerably reduced. To ensure that the increase in K_f was not caused by AICAR-induced declines in NEFA, AICAR was infused with Intralipid to maintain NEFA levels. Results demonstrated that the AICAR-induced increase in K_f was not dependent on NEFA concentration (Fig. 3). Given this, it is reasonable to speculate that LCFA uptake seen with AICAR was a result of LCFA protein-mediated transport. Evidence for the direct recruitment of LCFA transporters by AMPK has been demonstrated by Luiken et al. (47) who have shown that AICAR stimulation results in translocation of the CD36/FAT in cardiac myocytes. Another possibility is that fatty acids are “pulled”into the cell by AMPK-stimulated increases in LCFA oxidation. Stimulation of skeletal muscle with AICAR concurrently phosphorylates and inhibits ACC, the rate-limiting enzyme in malonyl-CoA synthesis, and also phosphorylates and activates the enzyme that deactivates malonyl-CoA, malonyl-CoA decarboxylase (48,49). Together, these targets effectively reduce malonyl-CoA levels, an allosteric inhibitor of carnitine palmitoyltransferase-1 transporter in the mitochondrial membranes allowing increased fat oxidation, which in itself may stimulate translocation of fatty acid transporters.

In addition to determining the NO dependence of AMPK’s actions, this study demonstrates that both AMPK-induced glucose and LCFA uptake are dependent on muscle fiber-type composition. Our results demonstrate that muscles comprised of a greater percentage of fast-twitch fibers are more responsive to AMPK-stimulated glucose uptake than muscles comprised of more slow-twitch fibers. This finding is consistent with previous in vitro studies showing AMPK protein content and AMPKα2 activity to be greater in fast-twitch versus slow-twitch muscle fibers (50,51). As previously mentioned, AICAR treatment also increased LCFA clearance in a muscle-specific manner, although the fiber-type dependency was opposite that seen for glucose. The effect of AICAR on LCFA uptake was actually greater in muscle with lower AMPK activity. LCFA clearance was greatest in muscle comprised of slow-twitch fibers compared with fast-twitch fibers, a finding that is likely explained by increased LCFA transporter number and LCFA oxidative capacity of slow-twitch muscle (52).

In summary, we found that AMPK activation simultaneously increased glucose and LCFA clearance in vivo. Only muscle glucose uptake, however, was inhibited by l-NAME. This demonstrates that the mechanism by which AMPK increases glucose uptake is NO dependent, while AMPK-induced LCFA uptake occurs independent of this pathway in the majority of muscles studied. In addition, the magnitude of AICAR activation of both glucose and LCFA uptake in vivo is fiber-type specific. Soleus, which has the greatest percentage of slow-twitch fibers, exhibits the greatest increase in LCFA uptake and the smallest increase in glucose uptake. Taken together, our findings provide insight into the signaling mechanisms by which AMPK results in increased glucose and LCFA uptake in skeletal muscle.

This work was supported by NIDDK DK54902 (to D.W.). J.S. is supported by a mentor-based fellowship from the American Diabetes Association.

The authors acknowledge the contributions of Paul Martin, Angela Slater, Freyja James; the scientific input of Dr. Owen McGuinness; and the laboratories of Erik Richter and Jorgen Wojtaszewski at the Copenhagen Muscle Research Centre, Copenhagen, Denmark, for their instruction and enthusiasm in AMPK research.