Cholinergic Involvement in Vascular and Glucoregulatory Actions of Insulin in Rats

All experimental procedures were approved by the animal care and handling committee of Laval University and conformed to the Guide for the Care and Use of Laboratory Animals as adopted by the National Institutes of Health. Male Sprague-Dawley rats (Charles River, St. Constant, Canada) weighing 250–300 g were housed in stainless steel cages. They were randomly divided into two groups. One group (n = 78) was used for euglycemic-hyperinsulinemic clamp studies and hemodynamic measurements in conscious rats. In this group of rats, there were two surgical procedures undertaken in all animals. The first surgical procedure was to place the flow probes around specific arteries, and 7 days later, the second surgical procedure was to place the vascular catheters. Then, a euglycemic-hyperinsulinemic clamp study was conducted 72 h after the second surgical step. The second group of rats (n = 37) was used to determine in vivo glucose utilization in individual muscles. In this group of rats, only one surgical procedure was undertaken to place the vascular catheters.

The rats were anesthetized with a mixture of ketamine and xylazine (100 and 10 mg/kg i.p., respectively) and had pulsed Doppler flow probes implanted around the left renal and superior mesenteric arteries and the lower abdominal aorta to monitor renal, mesenteric, and hindquarter blood flows, according to Gardiner and Bennett (23) and as previously described (5). After surgery, the rats were given subcutaneous injections of buprenorphine (0.05 mg/kg), returned to their cages, and allowed to recover for at least 7 days. After the recovery period, the rats were reanesthetized (as above), and the leads of the implanted probes were soldered to a microconnector (Microtech). Three separate catheters were implanted in the right jugular vein (for intravenous infusions) and one catheter in left femoral artery (for blood pressure and heart rate measurements). The catheters were tunneled subcutaneously to emerge at the same point as the probe wires. The rats were given subcutaneous injections of buprenorphine (0.05 mg/kg) and returned to their cages. Recovery after surgery was deemed satisfactory by the resumption of growth and normalization of 24-h food intake the day after this second surgical procedure. Experiments began 72 h after this last surgical step.

The rats were deprived of food for 10–12 h before the glucose clamp study. Before each experiment, blood glucose and plasma insulin were determined, and the resting heart rate, blood pressure, and regional blood flows were recorded over 15 min in quiet, unrestrained rats. After that time, the first group of rats (n = 50) received a continuous intravenous infusion of saline (0.9% NaCl), and the second group (n = 28) received atropine (1 mg/kg, administered intravenously as a bolus, followed by a continuous intravenous infusion of 1 mg · kg⁻¹ · h⁻¹). The rate of infusion was 0.3 ml/h. The two groups of rats were further divided in three subgroups each. The euglycemic-hyperinsulinemic clamp was performed in two subgroups receiving saline only (n = 34) and in two other subgroups receiving atropine (n = 18). The other subgroups of saline-treated (n = 16) or atropine-treated (n = 10) rats were infused with saline/BSA 0.2%, instead of insulin and dextrose, to approximately match the saline load delivered during the clamp studies. Sixty minutes after the beginning of intravenous infusion of atropine or saline, a second blood sample (0.3 ml) was collected for blood glucose and plasma insulin determinations. The euglycemic-hyperinsulinemic clamp was then carried out over 120 min in the designated groups as previously described (5). Insulin was infused at a rate of 4 or 16 mU · kg⁻¹ · min⁻¹ in the four subgroups of rats receiving atropine or saline (only one dose of insulin was tested per subgroup). Heart rate, blood pressure, and regional blood flows were continuously measured.

The second group of rats (n = 37) was used for measurement of steady-state insulin-stimulated glucose uptake in individual muscles, during a euglycemic infusion of insulin (at 4 or 16 mU · kg⁻¹ · min⁻¹) carried out in the absence or presence of atropine, as described above. This was performed by measuring the incorporation of [³H]2-deoxy-d-glucose according to Kreagen et al. (24) and as previously described (21). Briefly, a bolus injection of 250 μCi/kg [³H]2-deoxy-d-glucose and 25 μCi/kg [¹⁴C]sucrose in a 0.5-ml saline solution were injected during the last 25 min of a euglycemic clamp performed at the doses of 4 or 16 mU · kg⁻¹ · min⁻¹ insulin and in the absence or presence of atropine. Blood samples were taken at regular intervals (2.5, 5, 7.5, 10, 15, and 20 min) after the bolus injection for determination of plasma glucose, [³H]2-deoxy-d-glucose, and [¹⁴C]sucrose. Twenty-five minutes after the bolus, the animals were rapidly killed by decapitation, and hindlimb muscles (soleus, extensor digitorum longus [EDL], gastrocnemius, and quadriceps) were rapidly excised, frozen into liquid nitrogen, and stored at −80°C. Subsequently, tissue samples (30–50 mg) were dissolved in 1 ml 0.5 mol/l ammonium hydroxide at 55°C for 16 h. Thereafter, hydrogen peroxide (30%) was added for 60 min at 55°C to decrease quenching, followed by the addition of 10 ml scintillation fluid (Amersham, Mississauga, Ontario, Canada). The samples and plasma (20 μl) were then counted in a liquid scintillation counter (Wallach 1409). The accumulation of [³H]2-deoxy-d-glucose in muscle, corrected for the extracellular space with [¹⁴C]sucrose, was used as an index of glucose uptake rates (25).

The drugs used were atropine (Sigma Aldrich, Oakville, Ontario, Canada), dextrose (Sigma Aldrich), BSA (Sigma Aldrich), insulin (Humulin R; Eli Lilly, Toronto, Ontario, Canada), [³H]2-deoxy-d-glucose (Perkin Elmer, Woodbridge, Ontario, Canada), [¹⁴C]sucrose (Mandel-NEN Lifesciences, Guelph, Ontario, Canada), buprenorphine (Schering-Plough, Welwyn Garden City, Hertfordshire, U.K.), ketamine (Rogar/STB, London, Ontario, Canada), and xylazine (Bayer, Toronto, Ontario, Canada).

Blood for plasma glucose and insulin determinations in the basal state and during insulin infusion was drawn, put in untreated polypropylene tubes, and centrifuged with an Eppendorf microcentrifuge (Minimax; International Equipment). Plasma was stored at −20°C until assay. The glucose concentration of the supernatant was measured by the glucose oxidase method (26) using a glucose analyzer (Technicon RA-XT), and plasma insulin level was measured by radioimmunoassay using porcine insulin standards and polyethylene glycol for separation (27).

Data are means ± SE; n is the number of observations. Data were analyzed for statistical significance by an ANOVA for repeated measurements. Post hoc comparisons were made using Fisher’s test. A P value <0.05 was taken to indicate a significant difference.

Table 1 illustrates the effects of continuous intravenous infusion of atropine on plasma levels of insulin and glucose and baseline values of heart rate, mean arterial blood pressure, regional blood flows, and vascular conductances in conscious, unrestrained rats. Thus, the continuous intravenous infusion of atropine had no effect on plasma glucose or insulin concentrations. However, the infusion of atropine was found to slightly but significantly increase basal heart rate and mean arterial blood pressure values and to decrease basal superior mesenteric blood flow and vascular conductance values, whereas no changes in resting values of renal or hindquarter blood flows or vascular conductances were noted.

Figure 1A and B shows that insulin infusion at a rate of 4 or 16 mU · kg⁻¹ · min⁻¹ caused significant increases in renal and hindquarter blood flows but had no effect on heart rate or mean arterial blood pressure compared with measurements after a control infusion of vehicle (saline/0.2% BSA). Furthermore, a significant decrease in superior mesenteric flow was observed at the highest dose of insulin tested (Fig. 1B). These cardiovascular responses were associated with increases in renal (Fig. 2A and B) and hindquarter vascular conductances (Fig. 2B) and decreases in superior mesenteric vascular conductance (Fig. 2B) when compared with the effects of control infusion of vehicle.

In rats receiving atropine, the same infusions of insulin were found to cause slight but significant increases in heart rate (Fig. 1B) and mean blood pressure (Fig. 1A and B), when compared with the effects of control infusion of vehicle in atropine-treated rats. These responses did not differ from those observed in untreated rats, although in those animals, the heart rate and blood pressure responses to insulin did not reach the level of significance. However, the increases in renal and hindquarter flows and vascular conductances previously noted in untreated rats were completely abolished by the treatment with atropine (Figs. 1 and 2). A marked and long-lasting hindquarter vasoconstriction was observed at the dose of 4 mU · kg⁻¹ · min⁻¹ insulin (Fig. 2A). Moreover, there were significant decreases in superior mesenteric blood flows and vascular conductances (Figs. 1 and 2), which did not differ from those observed in rats receiving insulin alone.

During the euglycemic-hyperinsulinemic clamp, which was performed at an insulin infusion rate of 4 or 16 mU · kg⁻¹ · min⁻¹ and in the presence or absence of atropine, we found that for each dose of insulin tested, fasting plasma insulin levels rose acutely and achieved similar plateaus, according to the dose of insulin, whereas normal plasma glucose levels were maintained in every group of rats (Table 2). However, the average glucose infusion rate required to maintain euglycemia during the last hour of the clamp, the conditions of which closely approximated a steady-state insulin concentration and which represented the whole-body glucose utilization, was significantly smaller in the atropine-treated group than in the group receiving only insulin at the dose of 4 mU · kg⁻¹ · min⁻¹ (Table 2). This difference was not observed when insulin was infused at a rate of 16 mU · kg⁻¹ · min⁻¹ in both experimental groups.

Figure 3 shows that in untreated rats, the euglycemic infusion of insulin at a rate of 4 or 16 mU · kg⁻¹ · min⁻¹ elicited significant increases in glucose uptake in soleus, quadriceps, gastrocnemius, and EDL muscles when compared with that measured after a control infusion of vehicle. However, atropine treatment was found to markedly and significantly reduce (∼50%) in vivo insulin-stimulated glucose uptake in every studied muscle, when insulin was infused at a rate of 4 mU · kg⁻¹ · min⁻¹. However, at the rate of 16 mU · kg⁻¹ · min⁻¹, a slight but significant reduction in glucose uptake was only observed in EDL and quadriceps muscles. In the absence of insulin infusion, no effect of atropine was noted on basal glucose uptake in any muscles, when compared with that measured after a control infusion of vehicle only.

The present study indicates that at the whole-animal level acute treatment with atropine significantly reduces insulin sensitivity, as determined using a euglycemic-hyperinsulinemic clamp performed in conscious rats at a low dose of insulin. Thus, at the dose of 4 mU · kg⁻¹ · min⁻¹ insulin, we found that the glucose infusion rate required to maintain euglycemia in the face of physiological hyperinsulinemia was significantly lower (∼25%) in the group of rats receiving atropine than in the control group. However, atropine had no effect on basal plasma glucose or insulin concentrations, suggesting that muscarinic cholinergic antagonism has no net systemic glucoregulatory effect in terms of the plasma glucose concentration. Moreover, during the euglycemic infusion of insulin, atropine had no effect on steady-state plasma insulin concentrations and variability of plasma glucose concentrations. However, the reduction in insulin sensitivity was no longer observed at supra-physiological levels of insulin. Thus, it is likely that the apparent insulin resistance noted at a low dose of insulin was overcome at higher insulin levels (16 mU · kg⁻¹ · min⁻¹) (5). These latter findings contrast with those of Xie and Lautt (28), who used a higher dose of insulin than us and showed a reduction in insulin sensitivity in fed anesthetized rats intravenously treated with atropine. Part of the discrepancy could be explained by differences in the prandial status of the animals, because fed rats would be anticipated to be more responsive to insulin action and to have a higher degree of basal parasympathetic tone than overnight-fasted rats (29).

Our results are compatible with the involvement of cholinergic input on the effect of insulin on whole-body clearance of glucose. This contention is supported by several pieces of evidence implicating the parasympathetic nerves on glucoregulation and indicating that parasympathetic neuropathy could be contributory to insulin resistance. Insulin resistance is commonly observed in clinical conditions that are associated with autonomic dysfunction, such as hypertension (30), diabetes (31), and obesity (32). Furthermore, electrical stimulation of the lateral hypothalamic nucleus (parasympathetic area) was shown to reduce blood glucose level in rabbits (33), and vagal nerve stimulation (34) and selective stimulation of hepatic efferent parasympathetic nerves (35,36) were found to decrease hepatic glucose output and increase hepatic glucose uptake and glycogen synthesis. In recently published studies, and using a modified euglycemic technique to quantify insulin sensitivity, Lautt (14) described a novel neurohormonal mechanism by which postprandial elevation in insulin levels activates a hepatic parasympathetic reflex release of a putative hepatic insulin–sensitizing substance, which sensitizes skeletal muscle to the effects of insulin. The reflex is mediated by acetylcholine, which binds to muscarinic receptors and stimulates the production of NO within the liver (14). Surgical denervation of the liver, cholinergic blockade by using atropine, or prevention of hepatic NO liberation by the use of NO synthase antagonists were all found to cause severe insulin resistance that was attributed to the blockade of hepatic insulin–sensitizing substance release (14). Although the present findings are unable to support or deny the hepatic autonomic reflex control of peripheral insulin action, as proposed by Lautt (14), our results agreed, however, with the presence of a cholinergic mediation of the action of insulin on glucose metabolism. Furthermore, in the experiments in which we have combined atropine treatment with in vivo administration of [³H]2-deoxy-d-glucose during insulin clamps, which permits maintenance of insulin at a steady-state level during investigation of its effect on glucose uptake in muscles, we found a marked reduction (∼50%) in insulin-mediated glucose uptake by a physiological dose (4 mU · kg⁻¹ · min⁻¹) in every tested muscle. At the dose of 16 mU · kg⁻¹ · min⁻¹ insulin, a significant reduction in glucose uptake in the presence of atropine was only noted in EDL and quadriceps muscles. These results highly suggest the existence of a cholinergic mediation of the actions of insulin on glucose uptake and blood flow regulation (see below) at the skeletal muscle level. Furthermore, the demonstration here that atropine treatment alone (i.e., in the absence of insulin) did not affect basal glucose uptake in any tested muscles excluded the possibility that glucose uptake may have already been substantially inhibited, before insulin addition (i.e., as a result from a redistribution of blood flow in the leg muscles to the detriment of glucose uptake).

In untreated rats, we found that the euglycemic infusion of insulin elicited vasodilations in renal and hindquarter vascular beds and a transient vasoconstriction in the superior mesenteric vascular bed but no changes in mean arterial blood pressure or heart rate. These hemodynamic responses are consistent with those we previously reported in normal Wistar, Wistar-Kyoto, and Sprague-Dawley rats (5,21). In contrast, in the group of rats receiving atropine, we found that the physiological effects of insulin to vasodilate skeletal muscle and renal vasculatures were completely abolished, whereas the superior mesenteric vasoconstrictor response to insulin was not affected by the treatment. These observations highly suggest that cholinergic vasodilation could play an important role in the mediation of the vascular response to insulin, particularly at the skeletal muscle level. Our results are consistent with those of Mahajan et al. (37), showing that local methacholine enhances insulin-mediated muscle glucose uptake in vivo by augmenting capillary recruitment in anesthetized rats, but our results contrast with those of Randin et al. (38), demonstrating an absence of effect of atropine on the vascular response to euglycemic hyperinsulinemia in lean healthy humans. The reason for the conflicting results is not readily apparent. Differences in the experimental design, such as the dose of insulin, amount of atropine infused, and the site and method used to measure blood flow, may have contributed to these seemingly disparate results. Furthermore, we must consider that the extrapolation of the data in rats to humans may be difficult, because evidence for cholinergic vasodilator nerves in human skeletal muscle (in contrast to many animal species) has not been established (although the presence of local non-neuronal cholinergic mechanisms is not excluded) (39). It appears, therefore, that differences between rats and humans, in the effects of atropine on insulin stimulation of whole-body glucose uptake and muscle blood flow could be related to anatomical rather than to methodological differences.

Previous studies have clearly demonstrated that the insulin-mediated skeletal muscle vasodilation is mediated by NO, because this effect is abolished by N^G-monomethyl-l-arginine (11,12), and by inhibition of tetrahydrobiopterin synthesis, a cofactor essential for NO synthesis (40). It was suggested that insulin stimulates NO release by a direct, local effect at the vascular endothelium (13). However, the present findings raise the possibility that a cholinergically mediated stimulation of NO-dependent vasodilatory mechanisms could contribute, at least in part, to the endothelium-dependent vasodilator response to insulin. Alternatively, our results could also indicate that part of the vasodilator response to insulin is related to inhibition of adrenergic neurotransmission by stimulation of inhibitory prejunctional cholinergic receptors located on vascular adrenergic nerve endings (41). Thus, in the presence of atropine, the vasodilator response to euglycemic infusion of insulin might have been limited by stimulation of sympathetic vasoconstrictor outflow. This contention is supported by previous studies carried out in humans and animals and demonstrating that insulin administration stimulates sympathetic nerve activity as determined by measurements of plasma catecholamine levels or by direct microneurographic recordings of sympathetic nerve action potentials targeted specifically at the skeletal musculature (42–45). Thus, in the presence of atropine, it is likely that the endothelium-dependent vasodilator response to insulin was overcome by unopposed sympathetic vasoconstrictor influences.

Impairment in the vascular responses to insulin is thought to contribute to deficient uptake of glucose by peripheral tissues (presumably because of less delivery of glucose) (3). Therefore, according to that concept, it is likely that cholinergic blockade during the infusion of insulin (at a rate of 4 mU · kg⁻¹ · min⁻¹) has contributed to reduce insulin sensitivity by altering the hemodynamic responses to insulin, thus restricting insulin and glucose distribution through insulin-sensitive tissues and then reducing insulin action on glucose extraction at the level of skeletal muscle. Furthermore, the present findings raise the possibility that interruption of the cholinergic contribution to the blood flow response to insulin may have contributed, at least in part, to the state of insulin resistance previously reported by Buse and Buse (19) in denervated skeletal muscles, although this is not supported by a recent study using a different approach and technique to measure insulin sensitivity and blood flow in anesthetized rats (46). At the highest dose of insulin tested (16 mU · kg⁻¹ · min⁻¹), we found that the reduction in insulin sensitivity previously noted at the low dose of insulin was no more observed, despite a persistent defect in the vascular responses to insulin in the atropine-treated rats and a slight but significant reduction in glucose uptake in some but not all muscles in vivo (EDL and quadriceps only). The failure of atropine to achieve the same outcome at a higher dose of insulin might suggest that at a supra-physiological concentration of insulin (∼2,700 pmol/l), the “atropine-sensitive” insulin blood flow effect is not rate-limiting on glucose uptake, as previously described for physiological insulin and that steps in insulin action beyond any effect on blood flow are important determinants of muscle glucose uptake.

In conclusion, the overall concept presented in this study is novel and original. Taken together, the results of our experiments carried out in vivo strongly suggest that in rats, a high percentage of insulin-mediated muscle glucose uptake by a physiological dose involves a cholinergic mechanism. This atropine-sensitive effect is largely attributable to the vascular action of insulin in rats.

H.B. has received grants from the Heart and Stroke Foundation of Quebec and the Canadian Institutes of Health Research.

We thank Marie Tremblay for her expert assistance.