ATP-Dependent K⁺ Channels Contribute to Local Metabolic Coronary Vasodilation in Experimental Diabetes

This investigation was approved by the Institutional Animal Care and Use Committee and was conducted in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health publ. no. 85-23, revised 1996). Experiments were performed on six adult mongrel dogs of either sex (weighing 25–30 kg) taught to run on a motorized treadmill. Preanesthesia (acepromazine 0.03 mg/kg intramuscularly) was administered 30 min before induction of anesthesia with thiopental sodium (5 mg/kg i.v.). After endotracheal intubation, a surgical plane of anesthesia was maintained by mechanical ventilation with 1–3% isoflurane gas with supplemental oxygen. Using sterile technique, a left lateral thoracotomy was performed in the fifth intercostal space. A custom-made, coextruded polyurethane catheter (Putnam Plastics, Dayville, CT) was implanted in the descending thoracic aorta to measure aortic blood pressure and obtain arterial blood samples (8,22). A second polyurethane catheter was placed in the coronary sinus via a purse-string suture in the right atrial appendage for coronary venous blood sampling. The circumflex coronary artery was dissected free, and a flow transducer was placed around the artery. To avoid ventricular tissue injury, no devices were implanted in the myocardium and no surgical stitches were placed in the ventricles. A chest tube was placed to evacuate the pneumothorax, and the chest was closed in layers. The catheters and the flow transducer wire were tunneled subcutaneously and exteriorized between the scapulae. The incision was infiltrated with 2.5% bupivacaine, and buprenorphine (Buprenex) (0.3 mg intramuscularly) was administered to minimize postoperative pain. Amoxicillin (Clavamox; 6.25 mg/lb) and aspirin (81 mg) were administered twice a day for 10 days after surgery. A nylon jacket (Alice King Chatham, Hawthorne, CA) was placed on the animals to protect the catheters and the flow transducer wire. A elastomeric balloon pump (Access Technologies, Skokie, IL) was connected to the coronary sinus catheter so that heparinized saline (5 units/ml) could be continuously infused at 2 ml/h. The aortic catheter was flushed daily and filled with heparinized saline (5,000 units/ml). The animals were allowed at least 7 days for recovery before the experiments were conducted.

A coextruded polyurethane catheter was implanted in the aorta so that a high-fidelity catheter-tipped pressure transducer (Mikro-tip 3F; Millar Instruments, Houston, TX) could be inserted through it at the time of the experiment to measure aortic blood pressure (23). This pressure transducer was introduced into the polyurethane aortic catheter through a hemostatic control valve (Tuohy-Borst; Mallinckrodt Medical, Hazelwood, MO), which allowed arterial blood samples to be drawn while maintaining a fluid-tight seal around the catheter.

Phasic and mean coronary blood flow were continuously measured throughout the experimental protocol with an ultrasonic, perivascular flow transducer (Transonic Systems, Ithaca, NY). After experiments were completed, the animals were killed with sodium pentobarbital, and the circumflex perfusion territory was dyed with Evans blue. The weight of the dyed tissue was used to calculate coronary blood flow per gram of perfused myocardium.

Arterial and coronary venous blood samples were collected simultaneously in heparinized syringes that were immediately sealed and placed on ice. The samples were analyzed in duplicate for pH, Pco₂, Po₂, hematocrit, and oxygen content with an automatic blood gas analyzer (Synthesis 30) and CO-oximeter (model 682) system (Instrumentation Laboratories, Lexington, MA). Plasma glucose and lactate concentrations were measured with a Radiometer analyzer (model EML105; Radiometer, Bronshoj, Denmark). Myocardial oxygen consumption (MVo₂) (μl O₂ · min⁻¹ · g⁻¹) was calculated by multiplying coronary blood flow per gram of perfused tissue by the coronary arteriovenous difference in oxygen content. Lactate uptake (μmol · min⁻¹ · g⁻¹) was calculated by multiplying coronary plasma flow per gram of perfused tissue by the coronary arteriovenous difference in plasma lactate concentration.

The hypothesis that K_ATP channels are an important mechanism of functional coronary hyperemia in diabetes was tested at rest and during graded treadmill exercise with and without the K_ATP channel inhibitor glibenclamide. Each animal served as its own control. The dose of glibenclamide (1 mg/kg i.v.) used in this investigation has been used in previous studies (7, 8,24) and was found to effectively block the vasodilating action of the K_ATP channel opener cromakalim (25). After an overnight fast, coronary blood flow, aortic pressure, and heart rate were continuously measured while the dogs were resting in a sling and then during three levels of treadmill exercise: 1) 2 mph, 0% grade; 2) 3 mph, 5% grade; and 3) 4 mph, 10% grade. The animals exercised at similar levels (i.e., speed and percent grade) before and after alloxan treatment. Arterial and coronary venous blood samples were collected when hemodynamic variables were stable at each exercise level. Each exercise period was ∼2 min in duration, and the animals were allowed to rest sufficiently between each level for hemodynamic variables to return to baseline.

After nondiabetic control experiments were conducted, alloxan monohydrate (40–60 mg/kg) was administered intravenously over 1 min to induce diabetes (21). Alloxan was prepared as a 5% solution in citrate buffer (pH 4.0–4.5). The dogs were fasted at least 12 h before the injection of alloxan and were then fed immediately. A dog was considered diabetic if its plasma glucose concentration was >11 mmol/l. The experimental protocol described above was repeated in the diabetic animals at least 7 days after the alloxan injection with and without glibenclamide. In addition, three experiments were also conducted at rest and during exercise with glibenclamide in nondiabetic dogs to confirm the previous findings that K_ATP channels are not required for exercise-induced coronary vasodilation (3–5,7,8).

Changes in myocardial oxygen supply must be evaluated in relation to MVo₂, the primary determinant of coronary blood flow. Thus, the present data were analyzed by plotting changes in myocardial oxygen delivery and coronary venous Po₂ directly with changes in MVo₂. Coronary venous Po₂ is a sensitive index of tissue oxygenation, and the plot of coronary venous Po₂ versus MVo₂ is a sensitive way to determine whether the relationship between coronary blood flow and myocardial metabolism has been altered. On this plot, a vasodilator influence will either shift the regression line upward or make the slope less negative. A vasoconstrictor influence will shift the regression line downward or make the slope more negative. Influences on baseline flow will shift the relationship in a parallel manner, whereas an intervention that affects only exercise vasodilation will change the slope. Importantly, by plotting coronary response variables as a function of MVo₂, these plots account for any drug-induced changes in heart rate, blood pressure, or contractility that may significantly affect myocardial oxygen demand. Many studies have used this approach to analyze mechanisms of coronary blood flow control (3–5,8,20, 22,24).

Data are presented as means ± SE. Hemodynamic variables were recorded with a Hewlett Packard 7758A recorder and analyzed with a Sonometrics Sonolab 3.1.4 data acquisition system. Statistical testing was directed to detect overall treatment effects, i.e., diabetes-control versus diabetes-glibenclamide. A repeated-measures ANOVA was used to test for differences between coronary and systemic hemodynamic variables at rest and during exercise. When significance was found with ANOVA (P < 0.05), a Student-Newman-Keul’s multiple comparison test was performed. Linear regression analysis was used to compare the slopes of response variables (oxygen delivery and coronary venous Po₂) plotted versus MVo₂.

Hemodynamic, blood gas, and metabolic variables at rest and during exercise are given in Table 1. Average blood urea nitrogen and creatinine levels after 1 week of alloxan treatment were 34.5 ± 11.9 and 1.75 ± 0.60 mg/dl, respectively. (Reference values for dogs range from 7 to 27 mg/dl for blood urea nitrogen and from 0.5 to 1.8 mg/dl for creatinine.) Body weight was reduced from 28.7 ± 3 to 27.5 ± 3 kg after alloxan treatment. Average arterial plasma glucose concentration increased from 4.8 ± 0.3 mmol/l in nondiabetic control dogs to 21.5 ± 2.2 mmol/l ∼1 week after alloxan injection. Glibenclamide did not significantly alter the arterial plasma glucose concentration in the alloxan-treated dogs. In the nondiabetic control dogs, exercise significantly increased MVo₂ 3.4-fold, myocardial oxygen delivery 3.0-fold, and heart rate 2.4-fold (Table 1). Coronary venous Po₂ decreased from 19.9 ± 0.8 mmHg at rest to 14.8 ± 0.8 mmHg at the highest level of exercise. Diabetes significantly reduced the exercise-induced increase in myocardial oxygen delivery to 2.5-fold and lowered the coronary venous Po₂ to 16.3 ± 0.8 mmHg at rest and to 13.3 ± 0.8 mmHg during exercise.

Blockade of K_ATP channels with glibenclamide in the diabetic dogs tended to lower coronary blood flow (Fig. 1A) and MVo₂ (Fig. 1B) at rest and during exercise. K_ATP channel blockade decreased myocardial oxygen delivery at the highest level of exercise (Fig. 1C) and significantly reduced coronary venous Po₂ to 14.3 ± 0.7 mmHg at rest and to 9.4 ± 0.9 mmHg at the highest level of exercise (Fig. 1D).

The relationship between myocardial oxygen delivery and MVo₂ is shown in Fig. 2. Glibenclamide treatment significantly decreased the slope of this relationship (P = 0.03), indicating that K_ATP channels are important to functional exercise coronary hyperemia in alloxan-induced diabetes. Because of the normally high oxygen extraction by the left ventricle (8,22,24), the data points and regression lines lie extremely close to the line of 100% oxygen extraction (oxygen consumed = oxygen delivery) illustrated in Fig. 2. Thus, significant differences in oxygen delivery at a given MVo₂ are hard to visualize. The difference between these data points and the line of 100% oxygen extraction represents the oxygen extraction reserve of the myocardium, which is shown in Fig. 3. Glibenclamide significantly reduced the percent myocardial oxygen reserve at rest and at each level of exercise, indicating that K_ATP channel blockade impaired the balance between oxygen delivery and consumption and forced the heart to use more of its limited oxygen extraction reserve.

A more sensitive index of the relationship between coronary blood flow and myocardial metabolism is shown in the plot of coronary venous Po₂ versus MVo₂ (Fig. 4) (4,5,8,20,22, 24). K_ATP channel blockade with glibenclamide made the slope of this relationship more negative (P = 0.01), indicating that K_ATP channels contribute to local metabolic coronary vasodilation in experimental diabetes.

Additional experiments (n = 3) were also conducted to confirm earlier findings of the effects of K_ATP channel blockade on coronary blood flow control during exercise. Results from these experiments, plus a replotting of the data from Richmond et al. (8), are shown in Fig. 5. Consistent with previous studies (3,8), K_ATP channel blockade with glibenclamide in the normal (control) dogs resulted in a significant parallel shift downward in the relationship between coronary venous Po₂ and MVo₂ (P < 0.0001). However, the slope of this relationship did not become more negative (P = 0.66), indicating that K_ATP channels are not required for exercise-induced coronary vasodilation in normal, nondiabetic dogs.

The present study is the first to examine whether K_ATP channels contribute to local metabolic coronary vasodilation during exercise in experimental diabetes. Relative to diabetes alone, K_ATP channel blockade with glibenclamide significantly reduced myocardial oxygen delivery at the highest level of exercise (Fig. 1C) and reduced the slope of the relationship between myocardial oxygen delivery and MVo₂ (Fig. 2), thereby decreasing the oxygen reserve of the myocardium (Fig. 3). In addition, glibenclamide made the slope of the relationship between coronary venous Po₂ and MVo₂ more negative (Fig. 4). These findings indicate that K_ATP channels are an important mechanism of functional coronary hyperemia in experimental diabetes.

Whether K_ATP channels contribute to functional coronary hyperemia in diabetic subjects has not been previously investigated. This hypothesis was examined in the present study by relating changes in myocardial oxygen delivery and coronary venous Po₂ directly with changes in MVo₂. Thus, changes in myocardial oxygen supply are related directly to changes in myocardial oxygen demand. When K_ATP channels were inhibited in the diabetic dogs with glibenclamide, the slope of the coronary venous Po₂ vs. MVo₂ relationship (Fig. 4) was steepened, i.e., became more negative. This finding indicates that K_ATP channels normally contribute to local metabolic coronary vasodilation in the diabetic animals. In addition, glibenclamide significantly decreased the slope of the relationship between myocardial oxygen delivery and MVo₂ (Fig. 2). In other words, the oxygen delivery to the myocardium was impaired by glibenclamide as MVo₂ was increased during exercise. It should be appreciated that the slope of this relationship can only decrease to a limited extent, since the myocardial oxygen extraction is so high at rest (>80% in the diabetic group). Because K_ATP channel blockade limited the delivery of oxygen to the myocardium, the heart was forced to use more of its limited extraction reserve to meet the metabolic requirements of the tissue. This fact is also evidenced by the reduction in the oxygen reserve of the myocardium (Fig. 3).

The present findings that K_ATP channels contribute to local metabolic coronary vasodilation in alloxan-induced diabetic dogs are supported by the earlier findings of Kersten et al. (14), who found that the coronary vasodilatory response of coronary arterioles (<100 μm) to the K_ATP channel agonist aprikalim was significantly enhanced in alloxan-induced diabetic dogs relative to euglycemic (control) dogs. The coronary arteriolar response to aprikalim was also enhanced in nondiabetic hyperglycemic dogs, suggesting that high circulating glucose levels may be responsible for the increase in K_ATP channel activity. The results of Shimoni et al. (15), showing that the IC₅₀ for ATP-dependent inhibition of K_ATP channels was approximately twofold higher for channels from diabetic rat hearts, also support the present findings. Further studies are needed to determine whether hyperglycemia per se is responsible for the apparent increase in K_ATP channel activity detected in this investigation.

It should be acknowledged that glibenclamide also has nonspecific effects on other coronary vasodilator mechanisms that may have contributed to the present findings. Most notably, glibenclamide also significantly attenuates adenosine-mediated coronary vasodilation (2,4,5,11,13, 24,25). Adenosine does not contribute to local metabolic coronary vasodilation during exercise in nondiabetic animals (22), and furthermore, adenosine-induced coronary vasodilation is attenuated in diabetic animals (21, 26–28). Therefore, it is unlikely that glibenclamide-induced attenuation of adenosine vasodilation was directly responsible for the findings of this study. However, our laboratory recently reported that the balance between coronary blood flow and myocardial metabolism is significantly impaired at rest and during exercise in alloxan-induced diabetic dogs (20). Data from the present study support this finding (Table 1). This imbalance could lead to a compensatory increase in cardiac adenosine release, in which case the present treatment with glibenclamide would have blunted vasodilation of both K_ATP channels and adenosine. Either way, this study has demonstrated that sulfonylurea treatment with glibenclamide significantly attenuates functional coronary hyperemia in diabetic dogs.

Previous investigations have examined the role of K_ATP channels in modulating coronary vascular resistance in nondiabetic animals. The consensus of these studies is that K_ATP channels contribute to control of resting coronary blood flow (3–10) but are not required for coronary hyperemia during increases in MVo₂ (3– 5,7,8). Data from control (nondiabetic) animals studied in this investigation support these conclusions. At rest, glibenclamide decreased resting coronary flow and venous Po₂ (Fig. 5). During exercise, glibenclamide did not compromise coronary vasodilation in these animals, since the K_ATP channel blockade resulted in a parallel shift of the coronary venous Po₂ versus MVo₂ relationship but did not significantly alter the slope (Fig. 5).

The present experiments were conducted in alloxan-induced diabetic animals whose fasting plasma glucose concentration averaged ∼20 mmol/l (360 mg/dl). Acute glibenclamide administration did not significantly reduce the arterial plasma glucose concentration (Table 1), indicating that an increase in the release of insulin from the pancreas was not a confounding effect of glibenclamide treatment in this study. This effect is most likely due to the fact that alloxan destroys the islets of Langerhans in pancreas, and thus insulin release could not be significantly altered (29). The animals were not treated with insulin and thus represent a model of poorly controlled type 1 diabetes, i.e., chronic hypoinsulinemia and hyperglycemia. Whether K_ATP channels contribute to exercise-induced coronary vasodilation in type 2 (insulin-resistant) diabetes merits future investigation.

Sulfonylurea-derivative drugs such as glibenclamide have been used for many years to increase insulin release in patients with non-insulin-dependent diabetes. However, studies showing that blockade of K_ATP channels increases coronary vascular resistance in experimental nondiabetic animals (3–10) has led to an unresolved debate as to whether sulfonylurea drugs should be used in patients with diabetes, especially those with ischemic heart disease (30,31). In earlier studies designed to examine the effect of diabetes on coronary flow reserve, some patients had received oral hypoglycemic therapy (18, 32). Because the drug was withheld on the day of the investigation, its effect on coronary flow control was not determined. However, other investigations have reported a significant decrease in forearm vascular blood flow (33) and in calf reactive hyperemia (34) in nondiabetic humans after oral glibenclamide treatment.

It is also important to point out that K_ATP channels have different properties in different tissues. The IC₅₀ for glibenclamide is ∼4 nmol/l in pancreatic β-cells and ∼27 nmol/l in cardiac myocytes (35,36). Its lower potency in cardiac myocytes suggests that glibenclamide could possibly affect pancreatic insulin release without influencing coronary vasomotor tone at clinically administered doses. However, the fact that oral glibenclamide reduced peripheral blood flow control in nondiabetic patients argues against this hypothesis (33,34). To date, no study has directly examined whether oral hypoglycemic drugs significantly impair coronary vasodilation in diabetic humans.

This study is the first to show that K_ATP channels contribute to local metabolic coronary vasodilation in experimental diabetic animals. This finding is important because it is well established that K_ATP channels are not required for functional coronary hyperemia in normal (nondiabetic) animals (3–5,7,8). Future clinical studies are needed to determine whether oral hypoglycemic therapy with agents such as glibenclamide compromises coronary vasodilation in patients with diabetes.

This study was supported by a Career Development Award to J.D.T. from the American Diabetes Association.

The authors wish to thank Arthur G. Williams, Jr., and Abraham Heymann for expert technical assistance.