5-Aminoimidazole-4-Carboxamide-1-β-d-Ribofuranoside Causes Acute Hepatic Insulin Resistance In Vivo

A total of 18 mongrel dogs of either sex (mean weight 23 ± 1 kg) were housed in a facility that met the guidelines of the American Association for the Accreditation of Laboratory Animal Care. All surgical and experimental protocols were approved by the Vanderbilt University animal care and use committee. A laparotomy was performed on animals at least 16 days before an experiment, as previously described (11). Silastic infusion catheters were inserted into the inferior vena cava for the infusion of somatostatin (SRIF), glucose, indocyanine green, and radioactive isotopes of glucose ([3-³H]glucose and l-[¹⁴C]glucose). Infusion catheters were also inserted into the splenic and jejunal veins and fed forward into the portal vein for the infusion of insulin, glucagon, AICAR, glucose, and l-[¹⁴C]glucose. Sampling catheters were inserted into the portal vein, hepatic vein, and femoral artery. Catheters were filled with saline containing heparin and knotted at the free end. Transonic flow probes (Transonic Systems, Ithaca, NY) were fitted to the hepatic artery and portal vein. Flow probe leads and the knotted catheter ends were placed in subcutaneous pockets. Animals that met the experiment inclusion criteria (good appetite [consumed entire daily ration], normal stool, white blood cell count <18,000, and hematocrit ≥0.35) were fasted 18 h before an experiment.

The experimental protocol is shown in Fig. 1. After the fasting period, the knotted catheter ends and flow probes were removed from the subcutaneous pockets with 1-cm incisions made under a local anesthetic (2% lidocaine). Dogs were then placed in a Pavlov harness, and the catheters were aspirated with saline. At t = −220 min, peripheral infusions of indocyanine green, [3-³H]glucose, and l-[¹⁴C]glucose were initiated. Indocyanine green was used for the determination of splanchnic blood flow in the event of flow probe failure. The infusion of [3-³H]glucose continued for the duration of the experiment. The peripheral l-[¹⁴C]glucose was terminated at −90 min. From t = −120 to −90 min, baseline blood samples were taken. At t = −90 min, a peripheral infusion of SRIF was initiated to suppress pancreatic insulin and glucagon secretion. Basal circulating levels of glucagon were maintained with a portal glucagon infusion (0.5 ng · kg⁻¹ · min⁻¹). Physiological hyperinsulinemia was established with 1.2 mU · kg⁻¹ · min⁻¹ intraportal insulin infusion. Also at t = −90 min, glucose was infused (3.2 mg · kg⁻¹ · min⁻¹) into the portal vein. l-[¹⁴C]glucose was added to the portal glucose infusion to establish the same tracer infusion rate as that used for the peripheral infusion before t = −90 min. A variable peripheral glucose infusion was used to clamp arterial blood glucose concentrations at 150 mg/dl. After the initiation of these infusions, 80 min was allotted for the establishment of steady-state conditions. At t = −10 and 0 min, blood samples were taken for assessment of substrate concentration and hepatic substrate balance. After the t = 0 min sample, dogs received either 1 mg · kg⁻¹ · min⁻¹ (AICAR1 group, n = 5) or 2 mg · kg⁻¹ · min⁻¹ (AICAR2 group, n = 5) portal AICAR infusion from t = 0 to 60 min. Control animals (saline group, n = 8) were treated identically to the AICAR1 and AICAR2 groups, except that AICAR was not infused. These animals served as time controls. Six of the eight control dogs were previously described (11) and did not have the −10 and 0 min samples. Blood samples were taken every 10 min after the onset of saline or AICAR. At 60 min, dogs were killed, and liver biopsies were frozen in liquid nitrogen.

Plasma glucose levels were determined during experiments via the glucose oxidase method, using a Glucose Analyzer II (Beckman Instruments, Fullerton, CA). Plasma and tissue samples that were not immediately analyzed were stored at −70°C. Whole-blood samples were deproteinized with barium hydroxide and zinc sulfate and then centrifuged (3,000g for 30 min), and the supernatant was dried. Samples were reconstituted in 1 ml of water and 10 ml Ultima Gold scintillant (Packard, Meriden, CT), and radioactivity was determined using a Packard Tri-Carb 2900TR liquid scintillation counter. Plasma insulin and glucagon were measured using previously described radioimmunoassay techniques (12,13). Liver glycogen concentration and radioactivity (14) and liver glycogen phosphorylase and synthase activities (15) were determined as described previously. For the determination of hepatic glucose-6-phosphate and fructose-6-phosphate levels, liver samples where homogenized in 8% perchloric acid and centrifuged (3,000g for 10 min). Blood glucose and hepatic concentrations of glycogen (after digestion of precipitated glycogen with amyloglucosidase), glucose-6-phosphate, and fructose-6-phosphate were measured using enzymatic methods on a Technicon autoanalyzer (16). Hepatic adenine nucleotide and ZMP concentrations were determined from liver samples homogenized in 0.4 mol/l perchloric acid/0.5 mmol/l EGTA and neutralized with 0.5 mol/l potassium carbonate, pH 6.8. High-performance liquid chromatography analysis and UV detection was performed to assess nucleotide levels, as previously described (17,18).

Liver samples 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. Homogenized samples were assayed for protein concentration using a Pierce BCA protein assay kit (Rockford, Il). Next, 30 μg of protein was run on an SDS-PAGE gel and transferred to a polyvinylidene difluoride membrane. Membranes were treated with rabbit α-phospho-AMP kinase-α (Thr172) or α-phospho-acetyl CoA carboxylase (Cell Signaling, Beverly, MA) and then incubated with α-rabbit-horseradish peroxidase (Promega, Madison, WI). Chemiluminescence was detected with a WesternBreeze kit from Invitrogen (Carlsbad, CA). Densitometry was performed using LabImage software (National Institutes of Health).

Net hepatic glucose balance was calculated using the following equation: ([A] -[H]) × HAF + ([P] -[H]) × PVF, where [A], [H], and [P] represent blood glucose concentrations of the artery, hepatic vein, and portal vein, respectively. HAF and PVF represent the hepatic artery and portal vein blood flows, respectively. Positive values obtained from this equation represent net hepatic glucose uptake, whereas negative values represent output. Net hepatic glycogen deposition was calculated as the amount of [3-³H]glucose radioactivity incorporated into glycogen per mass tissue divided by the inflowing specific activity. Inflowing specific activity was calculated as: ([P*] × PVF + [A*] × HAF)/([P] × PVF + [A] × HAF). In this equation, [P*] and [A*] represent portal and arterial [3-³H]glucose radioactivity (dpm/ml), respectively.

Mixing of glucose in the portal vein was assessed using l-[¹⁴C]glucose (11). The appearance of l-[¹⁴C]glucose was calculated by multiplying the difference in radioactivity in the artery and portal vein by the portal vein flow. Studies were considered mixed if at least half of the time points had an l-[¹⁴C]glucose appearance rate within ±30% of the actual infusion rate.

All dogs received identical treatment before the infusion of AICAR. Data before the experimental period that were pooled after statistical analyses showed no differences between groups. Comparisons within a group and between groups were made using ANOVA and repeated-measures ANOVA. Differences were considered significant when P < 0.05.

After an 18-h fast, arterial blood glucose was 82 ± 1 mg/dl and net hepatic glucose output 1.5 ± 0.2 mg · kg⁻¹ · min⁻¹ (n = 18). Arterial plasma insulin and glucagon concentrations were typical of the short-term fasted state (Table 1). Arterial plasma cortisol, epinephrine, and norepinephrine concentrations suggest the animals were unstressed (Table 2). Arterial lactate, alanine, glycerol, and nonesterified fatty acid (NEFA) concentrations were 577 ± 59, 350 ± 28, 69 ± 10, and 773 ± 106 μmol/l, respectively.

Insulin levels rose approximately threefold compared with baseline levels during the hyperinsulinemic-hyperglycemic clamp period (Table 1). The infusion of AICAR did not alter insulin levels. During the hyperinsulinemic-hyperglycemic clamp, similar arterial blood glucose levels of 150 mg/dl were established in all groups (Fig. 2). Arterial plasma glucagon was similar to baseline during the clamp with or without AICAR. Arterial cortisol, epinephrine, and norepinephrine concentrations were also similar during the hyperinsulinemic-hyperglycemic clamp in the presence or absence of AICAR (Table 2).

The liver switched from net output of glucose during the baseline period to net uptake during the hyperinsulinemic-hyperglycemic clamp (Fig. 3). Before AICAR, net hepatic glucose uptake was 2.4 ± 0.5 mg · kg⁻¹ · min⁻¹ at t = 0 min. In the saline group, net hepatic glucose uptake was sustained over the course of the clamp (1.9 ± 0.4 mg · kg⁻¹ · min⁻¹ at t = 60 min). Net hepatic glucose uptake was completely suppressed in the AICAR1 group (−1.0 ± 1.7 mg · kg⁻¹ · min⁻¹ at t = 60 min). Net hepatic glucose uptake was not only completely suppressed in the AICAR2 group, but it showed marked output (−6.1 ± 1.9 mg · kg⁻¹ · min⁻¹ at t = 60 min), despite the presence of both elevated insulin and glucose.

Before the AICAR infusion period, the glucose infusion rate required to maintain the hyperglycemic clamp was ∼9 mg · kg⁻¹ · min⁻¹ (sum of all groups). The change in the glucose infusion rate required to maintain the clamp after the onset of saline or AICAR infusion is shown in Fig. 4. In saline-infused dogs, the glucose infusion rate gradually rose during the experimental period. After the onset of AICAR infusion, the glucose infusion rate fell significantly by 1.8 ± 0.6 mg · kg⁻¹ · min⁻¹ in the AICAR1 group and 3.3 ± 1.4 mg · kg⁻¹ · min⁻¹ in the AICAR2 group by t = 60 min.

During the hyperinsulinemic-hyperglycemic clamp period, lactate levels were 729 ± 119 μmol/l in the saline group and were not altered by AICAR infusion (707 ± 86 and 689 ± 74 μmol/l in the AICAR1 and AICAR2 groups, respectively). Likewise, alanine levels were similar in the saline group (285 ± 20 μmol/l) and in AICAR-infused dogs (385 ± 76 and 253 ± 36 μmol/l in the AICAR1 and AICAR2 groups, respectively). In the presence of the hyperinsulinemic-hyperglycemic clamp, arterial glycerol levels fell to similar levels in both the saline group (27 ± 7 μmol/l) and the AICAR-infused animals (20 ± 6 and 10 ± 7 μmol/l in the AICAR1 and AICAR2 groups, respectively). In a manner similar to glycerol, arterial NEFAs fell to similar levels in the saline group (82 ± 16 μmol/l) and the AICAR-infused dogs (139 ± 44 and 163 ± 65 μmol/l in the AICAR1 and AICAR2 groups, respectively). During the hyperinsulinemic-hyperglycemic clamp, net hepatic balances (μmol · kg⁻¹ · min⁻¹; positive values reflecting uptake) of lactate (−5.5 ± 1.5, −12.5 ± 6, and −8.3 ± 0.9), alanine (1.9 ± 0.41, 2.7 ± 1.4, and 2.5 ± 0.6), glycerol (0.73 ± 0.27, 0.23 ± 0.04, and 0.29 ± 0.14), and NEFA (0.15 ± 0.22, 0.41 ± 0.22, and 0.69 ± 0.5) were similar in the saline, AICAR1, and AICAR2 groups, respectively.

Net glycogen deposition was reduced by ∼45% in the AICAR1 compared with the saline group (Table 3). In the AICAR2 group, net hepatic glycogen deposition was reduced by ∼75% compared with the saline group and was significantly lower than that seen in the AICAR1 group. Insignificant trends reflecting increases in glycogen phosphorylase activity were present with AICAR infusions. AICAR had no effect on hepatic glycogen synthase activity. Liver glucose-6-phosphate and fructose-6-phosphate levels were elevated in the AICAR1 and AICAR2 groups.

ZMP levels were undetectable in the absence of AICAR (Table 4). AICAR infusion resulted in dose-dependent increases in hepatic ZMP. Hepatic AMP levels were significantly reduced in the AICAR1 group. The AICAR2 group showed an insignificant reduction in hepatic AMP levels (P = 0.145). Hepatic ADP and ATP levels were also reduced by low- and high-dose AICAR infusions.

Densitometry and representative Western blots of phosphorylated acetyl CoA carboxylase (ACC) are shown in Fig. 5. AICAR did not cause a significant increase in phospho-AMP kinase at either dose (data not shown). However, the downstream target of AMP kinase, ACC, showed a significant increase in phosphorylation in both the AICAR1 and AICAR2 groups.

This work demonstrates that portal venous AICAR infusion potently suppresses net hepatic glucose uptake in the presence of elevated insulin and glucose levels. Indeed, with the high-dose AICAR infusion (2 mg · kg⁻¹ · min⁻¹), net hepatic glucose uptake was not only suppressed, but the liver switched to marked net output. In fact, the rate of output (∼6 mg · kg⁻¹ · min⁻¹) was as high as rates evident during stress or exercise (19). The results of the current study are consistent with AICAR impacting the regulation of liver glycogen. Low-dose AICAR infusion caused a ∼45% reduction in net liver glycogen deposition, whereas the higher-dose AICAR infusion resulted in a ∼75% reduction. Considering that these reductions occurred with a rise in liver glucose-6-phosphate and fructose-6-phosphate and an increase in liver glucose production, it is likely that glycogen synthesis was inhibited by increasing glycogenolysis. The glycogenolytic response to AICAR infusion could have been mediated by the allosteric effects of ZMP on glycogen phosphorylase, activation of AMP kinase, or a combination of both (10). These findings are consistent with the observation that hepatic glycogen stores were reduced by 50% 2 h after a bolus of AICAR (8), reflecting a large net glycogenolytic response. The findings reported here are also consistent with the observation that the glucose infusion rate during an insulin clamp is decreased after administration of a primed AICAR infusion (5 mg/kg bolus, 3.75 mg · kg⁻¹ · min⁻¹ infusion) in the rat (20).

The effect of AICAR infusion on gluconeogenesis is also of important consideration. AICAR infusion inhibits hepatic fatty acid synthesis and stimulates fatty acid oxidation, which would seemingly promote gluconeogenesis (21). However, work in dominant-negative AMP kinase-expressing rats has shown that the activation of AMP kinase by adiponectin can lower plasma glucose (22). The work goes on to describe experiments in hepatocytes that implicate downregulation of PEPCK and glucose-6-phosphatase expression as a likely mechanism for the improvement in plasma glucose. Additionally, work in glycogen-reduced liver cells has shown that AICAR treatment reduces hepatic glucose production (23). This finding indicates that AICAR treatment has a suppressive effect on gluconeogenesis; however, it is unclear whether such an effect on hepatic glucose production occurs in the presence of substantial hepatic glycogen stores. In the current study, there were no differences in arterial concentrations or net hepatic balances of NEFA, lactate, alanine, or glycerol between saline- and AICAR-infused dogs, which suggests that increased gluconeogenesis did not contribute significantly to the switch from net hepatic glucose uptake to net hepatic glucose output seen with AICAR infusion.

Although there was no apparent increase in phospho-AMP kinase, the downstream target of AMP kinase, ACC, showed a significant increase in phosphorylation. Given that AMP kinase activity is regulated by both allosteric interactions as well as phosphorylation, phosphorylation of ACC is a more sensitive marker of AMP kinase activity (24–26). This notion is supported by work in myocardial tissue showing that AICAR treatment resulted in a clear increase in the phosphorylation of ACC, with little or no detectable change in phosphorylation of AMP kinase (7).

Muscle AMP kinase activation facilitates the uptake and oxidation of substrates as well as mobilizes muscle glycogen stores, and it may play a role in the postexercise sensitization of muscle to insulin stimulation (4,7,27,28). These actions of AMP kinase are consistent with a role in coordinating the metabolic response in working muscle. Interestingly, work has shown that during exercise, AMP kinase activity increases in the liver as well (9). During exercise, liver glucose output increases to help meet the increasing energy demands of working muscle (19). This increase in hepatic glucose output results from increased glycogenolysis mediated by a fall in portal vein insulin and a rise in portal vein glucagon (29,30). One may postulate that increases in AMP and/or AMP kinase signaling mediate this effect. Considering that glycogenolysis in muscle and liver are both increased during exercise and AMP kinase activity is increased in both tissues, it is reasonable to postulate that AMP kinase activation may play a role in the glycogenolytic response to exercise seen in both tissues.

Elevations in ZMP cause phosphorylation and activation of AMP kinase in a manner identical to that of AMP, i.e., by both allosteric mechanisms and inducing the phosphorylation of AMP kinase (31). Additionally, ZMP mimics the allosteric effects of AMP in the regulation of phosphorylase activity at least in myocardial tissue (7). Work performed in lean and obese Zucker rats using a peripheral AICAR infusion indicated that an infusion of 10 mg · kg⁻¹ · min⁻¹ (with a primer bolus of 100 mg/kg) suppresses endogenous glucose production (3). In the Zucker diabetic fatty rat, a peripheral venous bolus of AICAR resulted in an improved ability of insulin to suppress hepatic glucose output 24 h later (8). In the present study, AICAR was infused directly into the portal vein at a rate that was 10–20% of that used in rodent models. Portal venous AICAR infusion allows for more direct targeting of the drug to the liver while minimizing peripheral effects of the drug.

The allosteric actions of ZMP binding activate glycogen phosphorylase in a manner identical to that of AMP (7). Elevations in AMP within the physiological range have been shown to activate liver glycogen phosphorylase (32) and override the inhibitory effect of glucose on glycogen phosphorylase (33). Additionally, recent work in cultured hepatocytes has shown that incubation with 0.2–0.3 mmol/l AICAR causes marked activation of glycogen phosphorylase (10). In the current study, the infusion of AICAR produced hepatic ZMP levels 4- and 10-fold higher than the levels of hepatic AMP seen during the saline infusion. This elevation in ZMP was enough to counterbalance the inhibitory effect of hyperglycemia on glycogen phosphorylase. In this experiment there was an 18% increase in phosphorylase activity in AICAR2. However, the change was insignificant (P = 0.07). It should be noted that this assay takes into account the activity of the enzyme and not the subcellular localization. It is possible that the 18% increase in phosphorylase activity combined with potential changes in the distribution of the phosphorylase enzyme is sufficient to stimulate glycogenolysis. Support for the notion that glycogen phosphorylase plays a critical role in the regulation of glycogen synthesis comes from metabolic control analysis, where it was shown that phosphorylase activity has a strong control coefficient with regard to glycogen synthesis (34).

These findings show that portal venous AICAR infusion for 60 min, at 1 or 2 mg · kg⁻¹ · min⁻¹, causes marked resistance to insulin-stimulated net hepatic glucose uptake. Insulin-stimulated net hepatic glucose uptake was completely alleviated, and at the higher dose, AICAR actually increased liver glucose output to 6 mg · kg⁻¹ · min⁻¹. This insulin resistance is characterized by a reduction in the glucose infusion rate to maintain the hyperglycemic clamp. The results of this work are consistent with the work of Camacho et al. (34a), which showed that portal AICAR infusion (1 mg · kg⁻¹ · min⁻¹) stimulated hepatic glucose production in the presence of basal intraportal glucagon, hyperinsulinemia, and either euglycemia or hypoglycemia. Previous work has shown that AICAR causes marked improvements in muscle glucose uptake and oxidation (3,8,35–37). Additionally, AICAR increases the production of the antidiabetic adipokine adiponectin (38). Although these improvements have been well documented, it is important to note that these improvements occurred in animals receiving acute or chronic AICAR treatment via either injections or peripheral infusions. It is possible that human work will show similar improvements using injections or peripheral infusions. However, the route of drug delivery will be a very important consideration if AICAR, or drugs with similar actions, are developed for human use. Oral administration, the preferred method for chronic use, designed to produce a given arterial concentration would result in much higher hepatic concentrations, which likely would result in hepatic effects closer to the marked increases in hepatic glucose production that were seen in the present study, provided that the liver contains appreciable glycogen stores.

In conclusion, the results of this work show that AICAR infusion causes a marked suppression of net hepatic glucose uptake in the presence of elevated insulin and glucose, and it can even cause the liver to become a net producer of glucose. These findings have important implications. First, they add new emphasis to a possible role of purine nucleotides in the regulation of hepatic glycogen metabolism and hepatic glucose output. Second, the results of these studies are important to consider when evaluating the potential use of AMP analogs in the treatment of diabetes.