Authors: Johnathan D. Tune (1Department of Physiology and Anatomy, University of North Texas Health Science Center), Dirk J. Duncker (2Division of Experimental Cardiology, Department of Cardiology, Thoraxcenter, Erasmus University Medical Center), Adam G. Goodwill (3Department of Biomedical Sciences, Northeast Ohio Medical University), Cooper M. Warne (1Department of Physiology and Anatomy, University of North Texas Health Science Center), Salman I. Essajee (1Department of Physiology and Anatomy, University of North Texas Health Science Center), Selina M. Tucker (1Department of Physiology and Anatomy, University of North Texas Health Science Center), Steven A. Romero (1Department of Physiology and Anatomy, University of North Texas Health Science Center), Shawn B. Bender (4Department of Biomedical Sciences & Dalton Cardiovascular Research Center, University of Missouri; 5Research Service, Harry S. Truman Memorial Veterans Hospital), Daniel A. Beard (6Department of Molecular and Integrative Physiology, University of Michigan), C. Alberto Figueroa (7Department of Biomedical Engineering, University of Michigan; 8Department of Surgery, University of Michigan), Oana Sorop (2Division of Experimental Cardiology, Department of Cardiology, Thoraxcenter, Erasmus University Medical Center), Daphne Merkus (2Division of Experimental Cardiology, Department of Cardiology, Thoraxcenter, Erasmus University Medical Center), Gregory M. Dick (1Department of Physiology and Anatomy, University of North Texas Health Science Center)
Categories: Article, Coronary blood flow, local metabolic control, conducted responses, ATP-sensitive potassium channel
Source: Basic research in cardiology
Authors: Johnathan D. Tune, Dirk J. Duncker, Adam G. Goodwill, Cooper M. Warne, Salman I. Essajee, Selina M. Tucker, Steven A. Romero, Shawn B. Bender, Daniel A. Beard, C. Alberto Figueroa, Oana Sorop, Daphne Merkus, Gregory M. Dick
This study tested the hypothesis that K^+^ serves as an in vivo signal coupling coronary blood flow with the oxidative requirements of the myocardium. Experiments were performed in swine in which coronary parameters and arterial and coronary venous [K^+^] were measured under baseline conditions, during exogenous administration of K^+^ (1–5 mM; n = 4), during increases in myocardial oxygen consumption (MVO2) to dobutamine (n = 7) and exercise (n = 6), alterations in coronary perfusion pressure (CPP; n = 8), and systemic hypoxemia (PaO2 to 30 mmHg; n = 7). Exogenous intracoronary K^+^ increased blood flow ( 20%) in direct proportion to the coronary venous [K^+^] up to the lethal limit of 10 mM. Dobutamine increased coronary flow and MVO23-fold but the coronary venous – arterial [K^+^] gradient (i.e., a surrogate index of myocardial release of K^+^ into the coronary circulation) did not change. Similarly, exercise increased coronary flow and MVO22.5-fold without a change in the coronary venous – arterial [K^+^] gradient. The coronary venous – arterial [K^+^] gradient did not change over the CPP range of 140–40 mmHg. Hypoxemia increased coronary blood flow during physiologic increases in cardiac work or when oxygen delivery is constrained.2-fold and coronary vascular resistance was weakly associated with <0.5 mM change in the coronary venous – arterial [K^+^] gradient. Intracoronary glibenclamide dose-dependently (1–3 mg/min; n = 4) increased coronary resistance but did not affect the coronary venous – arterial [K^+^] gradient. Intracoronary pinacidil dose-dependently (0.3–3.0 μg/kg/min; n = 3) increased coronary blood flow but did not affect the coronary venous – arterial [K^+^] gradient. Similarly, intravenous glibenclamide (3 mg/kg; n = 6) increased coronary resistance but did not affect the coronary venous – arterial [K^+^] gradient in exercising swine. These findings demonstrate that myocardial interstitial [K^+^] is unlikely to couple coronary blood flow to MVO2
It is well established that coronary microvascular resistance is tightly controlled to maintain adequate oxygen delivery to the myocardium [9, 10, 16, 31, 38]. Understanding precisely how coronary blood flow is coupled directly to the oxidative requirements of the heart has been the focus of much research [21, 47, 49]. However, results to date have generated far more questions than answers regarding this fundamental physiologic phenomenon [21, 22, 47]. Recently, Lederer and colleagues advanced a novel “electro-metabolic signaling” paradigm in which ATP-sensitive K^+^ (KATP) channels in cardiomyocytes convert the underlying metabolic energy state into an electrical signal (increased interstitial [K^+^]) that modulates microvascular membrane potential, vascular tone, and thus myocardial blood flow [50]. Their elegant hypothesis was supported by integrative studies which utilized an ex vivo ventricular tissue preparation and electrophysiological and imaging approaches to show cardiomyocyte KATP channel activation is coupled to hyperpolarization of capillary endothelial cells, pericytes, and vascular smooth muscle cells [50]. Importantly, such an interstitial K^+^ pathway could feasibly contribute to the balance of coronary blood flow with myocardial metabolism during increases in myocardial oxygen consumption (MVO2) and/or during reductions in myocardial oxygenation.
A role for interstitial K^+^ in local metabolic control of coronary blood flow was first proposed in the 1930’s by Katz and Lindner [25]. These and subsequent studies [2, 8, 14, 30, 37, 42] up until the late 1970’s demonstrated that K^+^ is a vasodilator over a select concentration range (~4–10 mM) [8] and that K^+^ efflux from the myocardium is elevated by increases in heart rate [34] and during conditions that promote myocardial ischemia [3, 4, 6, 23, 39]. Data from the Sparks laboratory indicated that K^+^-induced coronary vasodilation is mediated by the Na^+^/K^+^ ATPase pump (i.e., is ouabain-sensitive) and that increases in interstitial [K^+^] were sufficient to significantly diminish coronary resistance in response to tachycardia [34, 35]. However, the elevation in [K^+^] was modest and not sustained throughout the duration of increased cardiometabolic activity [34]. These findings led to the prevailing consensus that K^+^ could possibly contribute to the initiation of a vasodilator response but is likely not a mediator of steady-state local metabolic increases in coronary blood flow [16].
Discrepancies between prior studies of K^+^-induced coronary control [34, 35] and the electro-metabolic hypothesis [50] highlight important questions that merit further study. Most notable is the need to establish that sustained increases in myocardial interstitial [K^+^] produce coronary responses at relevant concentrations and in response to appropriate (patho)physiological conditions. Furthermore, studies to examine whether inhibition of the electro-metabolic pathway attenuates coronary vasodilation in response to (patho)-physiologic perturbations are also lacking. Particularly important is that data from conscious animals during exercise-induced increases in MVO2 are not available, as this is the most physiologically-relevant stimulus for coronary vasodilation. This series of studies was designed to interrogate the hypothesis that K^+^ serves as an in vivo signal that couples coronary blood flow with the oxidative requirements of the myocardium (i.e., to test the electro-metabolic hypothesis [50]). Experiments were performed in open-chest swine in which coronary and systemic hemodynamic parameters and arterial and coronary venous [K^+^] were measured under baseline conditions, during exogenous administration of K^+^ (1–5 mM intracoronary; n = 4), during dobutamine-induced increases in MVO2 (1–30 μg/kg/min iv; n = 7), alterations in coronary perfusion pressure (CPP) from 140 mmHg to 40 mmHg (n = 8), and systemic hypoxemia (PaO2 to 30 mmHg; n = 7). Additional studies also examined the effects of the KATP~ channel antagonist glibenclamide (1–3 mg/min intracoronary; n = 4) and the KATP channel opener pinacidil (0.3–3.0 μg/kg/min intracoronary; n = 3) on the coronary venous – arterial [K^+^] gradient. Finally, arterial and coronary venous [K^+^] were measured at rest and during exercise in chronically instrumented animals (n = 6), under control conditions and in the presence of KATP channel blocker glibenclamide (3 mg/kg intravenous). Assuming the coronary venous – arterial [K^+^] gradient reflects myocardial K^+^ release [8, 17, 27], these studies provide novel insight that increases in myocardial [K^+^] are not sufficient to couple coronary blood flow to metabolic demand when MVO2 is elevated or when O2 delivery is constrained.
This investigation was performed in Domestic (Yorkshire) swine (~50 kg; n = 33) after approval by the University of North Texas Health Science Center Institutional Animal Care and Use Committee and performed in accordance with the Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85–23, Revised 2011). Following completion of experiments, anesthetized animals were euthanized by electrical fibrillation and excision of the heart.
This study was approved by the Erasmus University Medical Center Rotterdam Institutional Animal Care and Use Committee and performed in accordance with the Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85–23, Revised 2011). Domestic (Yorkshire x Landrace) swine (~22 kg at the time of surgery; n = 6; 5 neutered males, 1 female) underwent daily adaptation to laboratory conditions starting 1 week before surgery and continuing during the first week after surgery.
Swine were sedated with an intramuscular injection of Telazol (5 mg/kg), xylazine (2.5 mg/kg), and ketamine (2.5 mg/kg) prior to anesthesia with buprenorphine (0.03 mg/kg, im) and α-chloralose (60 mg/kg, iv). Additional α-chloralose (20 mg/kg, iv) was given hourly to maintain an adequate plane of anesthesia. Anesthetized swine were intubated and ventilated with room air supplemented with O2. Oxyhemoglobin saturation (maintained >95%) and end tidal CO2 (maintained ~40 mmHg) were continuously monitored by aural pulse oximetry and inline capnography, respectively. A catheter was placed in a femoral artery and advanced into the aorta to measure blood pressure and heart rate. Another catheter was placed in a femoral vein and advanced into the inferior vena cava for systemic administration of drugs (supplemental α-chloralose, heparin, dobutamine). A second femoral artery catheter was utilized to supply blood to an extracorporeal servo-controlled pump to perfuse the left anterior descending coronary artery at a constant designated CPP as previously described by our laboratory [26, 48].
Following placement of femoral lines, succinylcholine (0.5 mg/kg, iv) was administered and a left lateral thoracotomy performed in the fifth intercostal space. The pericardium was incised to expose the heart and the left anterior descending isolated distal to its first major diagonal branch. Heparin (500 units/kg, iv) was administered and the left anterior descending artery was cannulated with a steel tip cannula fed by the extracorporeal perfusion circuit (n = 19; for experiments with K^+^ infusion, changing CPP, and administering glibenclamide or pinacidil). CPP was regulated by a servo-controlled roller pump and coronary blood flow was continuously measured by an in-line flow transducer (Transonic Systems, Ithaca, NY, USA). An alternative surgical preparation was also performed in a subset of swine (n = 14; for experiments with dobutamine and hypoxia) in which a perivascular flow probe (Transonic Systems, Inc., Ithaca, NY, USA) was placed around the artery to measure coronary blood flow. Following cannulation or placement of a perivascular flow probe, the anterior interventricular vein was catheterized to sample coronary venous blood from the left anterior descending perfusion territory.
Swine were sedated with ketamine (20 mg/kg i.v.) and midazolam (1 mg/kg i.m.), anesthetized with thiopental (10 mg/kg, iv), intubated and ventilated with a mixture of O2 and N2O (1:2) to which 0.2–1.0% (vol/vol) isoflurane was added [7, 11]. Anesthesia was maintained with midazolam (2 mg/kg, iv) and fentanyl (10 μg/kg per hour, iv). Under sterile conditions, the chest was opened via the fourth left intercostal space. A fluid-filled polyvinylchloride catheter was inserted into the aortic arch for measurement of aortic pressure, and blood sampling for determination of PO2, PCO2 and pH (ABL-505, Radiometer), and O2 saturation and hemoglobin concentration (OSM2, Radiometer). Polyvinylchloride catheters were also inserted into the pulmonary artery for administration of drugs. A Doppler flow probe was placed around the proximal left anterior descending artery for measurement of coronary blood flow [11]. Finally, two small angio-catheters were inserted into the anterior interventricular vein for coronary venous blood sampling [11]. Electrical wires and catheters were tunneled subcutaneously to exit at the back. Then, the chest was closed, and animals were allowed to recover, receiving analgesia (0.3 mg buprenorphine, im) and a slow-release fentanyl patch (12 μg/h) for 3 days and antibiotic prophylaxis (25 mg/kg amoxicillin and 5 mg/kg gentamycin, iv) for 5 days. All catheters were flushed daily with heparinized saline (1000–5000 IU/ml saline) to prevent the formation of blood clots and secure catheter patency.
To examine the physiologic effects of sustained increases in [K^+^] on coronary blood flow, exogenous K^+^ (calculated to increase the coronary arterial plasma concentration by 1–5 mM) was infused directly into the left anterior descending artery perfusion line at a constant CPP of 100 mmHg (n = 4; male n = 2; female n = 2). Arterial and coronary venous blood samples were obtained once coronary blood flow was stable at each dose of exogenous K^+^ (~1 min infusion). Intracoronary bolus doses of exogenous K^+^ (40 μmol) were also administered in select studies (n = 3; male n = 1; female n = 2).
To determine the effects of graded increases in MVO2 on the relationship between the coronary venous – arterial [K^+^] gradient and coronary blood flow, dobutamine (1–30 μg/kg/min) was continuously infused intravenously in swine instrumented with perivascular flow probes (n = 7; male n = 3; female n = 4). Hemodynamic parameters were recorded once blood pressure, heart rate and coronary blood flow plateaued at each dose (~15 min at each dose). Arterial and coronary venous blood samples were collected once hemodynamics stabilized at each concentration of dobutamine.
Effects of alterations in CPP on the coronary venous – arterial [K^+^] gradient and coronary blood flow autoregulation were assessed in swine in which the left anterior descending artery was cannulated by reducing CPP in increments of 20 mmHg from 140 to 40 mmHg (n = 8; male n = 3; female n = 5). Arterial and coronary venous blood samples were collected simultaneously once hemodynamic parameters stabilized at that CPP (~5 min at each CPP).
Experiments to examine the effects of graded hypoxemia on the coronary venous – arterial [K^+^] gradient and coronary blood flow were performed by supplementing ventilated air with increasing amounts of N2 gas to reduce PaO2 to approximately 70, 50, 40, and 30 mmHg (n = 7; male n = 3; female n = 4). Arterial and coronary venous blood samples were obtained once hemodynamic variables stabilized at each PaO2.
Experiments to determine the effects of KATP channel blockade on the coronary venous – arterial [K^+^] gradient in the absence and presence of systemic hypoxemia (PaO2 = 22 ± 1 mmHg) were also performed (n = 4; male n =2, female n = 2). Utilizing a cannulated preparation, the extracorporeal perfusion system was set to a baseline constant flow which maintained CPP at an average of 100 ± 1 mmHg. Coronary flow was subsequently held constant at this baseline level during intracoronary infusion of increasing concentrations of the KATP channel antagonist glibenclamide (0.3–3.0 mg/min). Arterial and coronary venous blood samples were obtained once CPP and hemodynamic variables stabilized at each concentration of glibenclamide. Reverting to constant pressure mode and continuing glibenclamide infusion (3.0 mg/min; intracoronary), ventilated air was supplemented with N2 gas to reduce PaO2 to ~ 20–25 mmHg in order to measure hypoxic vasodilation. Blood samples were again collected once hemodynamic variables were stable at reduced PaO2.
Experiments to determine the effects of KATP channel activation on the coronary venous – arterial [K^+^] gradient were performed (n = 3; male n = 1, female n = 2). Pinacidil (0.3–3.0 μg/kg/min was infused directly into the left anterior descending artery perfusion line at a constant CPP of 100 mmHg. Hemodynamic parameters were recorded once blood pressure, heart rate and coronary blood flow plateaued at each dose. Arterial and coronary venous blood samples were collected once hemodynamics stabilized at each concentration of pinacidil.
With swine lying quietly on the treadmill, resting hemodynamic measurements were obtained and arterial and coronary venous blood samples were collected. Hemodynamic measurements were repeated, and rectal temperature was measured, with animals standing on the treadmill. Subsequently, a five-stage exercise protocol (1–5 km/h) began with each stage lasting 3 min. Hemodynamic variables were continuously recorded and blood samples collected during the last 30 seconds of each exercise stage, at a time when hemodynamics had reached a steady state.
Arterial and coronary venous blood samples were collected, immediately sealed, and placed on ice. The samples were analyzed for pH, PCO2, PO2, hematocrit, lactate, K^+^, and oxygen content with an automatic blood gas analyzer and CO-oximeter system (Stat Profile Prime+ Vet Blood Gas Analyzer, Nova Biomedical, Waltham, MA, USA). Left anterior descending artery perfusion territory was estimated to be 30% of total heart weight, as previously described [13]. MVO2 was calculated by multiplying coronary blood flow by the arterial-coronary venous difference in oxygen content. Lactate uptake was calculated by multiplying coronary blood flow by the arterial-coronary venous difference in the lactate concentration.
Arterial and coronary venous blood samples were collected, immediately sealed, and placed on ice. Samples were analyzed for PO2, PCO2, pH and K^+^ (ABL-505, Radiometer), and for O2 saturation and hemoglobin concentration (OSM2, Radiometer). Myocardial O2 delivery (MDO2) was computed as the product of left anterior descending coronary blood flow and arterial blood O2 content. MVO2 in the region of myocardium perfused by the left anterior descending artery was calculated as the product of coronary blood flow and the difference in O2 content between arterial and coronary venous blood. Myocardial oxygen extraction (MEO2) was computed as the ratio of MVO2 and MDO2.
Statistical analysis was performed in Sigma Plot (version 14.0) or GraphPad Prism (version 10.0) and comparisons were made using paired t-test or one-way repeated measures analysis of variance (ANOVA) as appropriate. When significance was found with ANOVA, a Student-Newman-Keuls multiple comparison test was performed to identify differences relative to the baseline-control condition. Linear regression analysis of all data points was used to compare slopes of response variables and regression lines provided for statistically significant relationships. The effect of exercise and glibenclamide on hemodynamic and metabolic parameters was tested using two-way repeated measures analysis of variance (ANOVA), followed by post-hoc testing with Tukey’s multiple comparisons test. Values in figures and tables are presented as mean ± SEM. For all statistical comparisons, P < 0.05 (two-tailed) was considered statistically significant.
Figure 1a shows a representative trace of the coronary blood flow response to a bolus infusion (200 μl) of K^+^ (40 μmol) directly into left anterior descending artery perfusion line. Consistent with prior data of Murray and Sparks [35], we found bolus administration of K^+^ resulted in a transient ~80% increase in coronary flow (Fig 1b; P = 0.04). A representative trace of the coronary response to continuous intracoronary infusions of K^+^ (1–5 mM) is shown in Figure 1c. Sustained infusion of K^+^ over a range from 1 to 5 mM at a constant CPP of 100 mmHg produced concentration-dependent increases in coronary blood flow from an average baseline value of 0.73 ± 0.03 ml/min/g to 0.95 ± 0.03 ml/min/g at the highest [K^+^] (Fig 1d; P < 0.01). The K^+^-induced increases in coronary flow were directly related to graded increases in coronary venous [K^+^] from an average of 4.1 ± 0.1 mM at baseline to 8.6 ± 0.6 mM at the highest [K^+^] achieved (4–5 mM; Fig 2a; P < 0.001). Increases in coronary venous [K^+^] over this concentration range decreased coronary vascular resistance 22 ± 2% in a concentration-dependent manner (Fig 2b; P < 0.001). Infusion of intracoronary [K^+^] ≥ 5 mM (producing coronary venous [K^+^] > 10 mM) resulted in ventricular fibrillation (Fig 1c).
Hemodynamic and metabolic responses to systemic dobutamine infusion (1–30 μg/kg/min, iv) are provided in Table 1. Dobutamine increased heart rate from 84 ± 12 beats/min at baseline to 135 ± 14 beats/min at the highest dose of dobutamine (30 μg/kg/min; P < 0.05). Mean blood pressure remained relatively stable throughout the protocol (P = 0.22). Administration of dobutamine produced concentration-dependent increases in MVO2 (Table 1; P <0.001) that were directly related to increases in coronary blood flow (Fig 3a; P < 0.0001). Arterial and coronary venous [K^+^] both averaged 3.8 ± 0.3 mM under baseline conditions and were not significantly affected by dobutamine infusion (Table 1). The coronary venous – arterial difference in [K^+^] (with positive numbers representing myocardial K^+^ production) was calculated. Regression analyses of all data points failed to show any relationship between the coronary venous – arterial [K^+^] gradient and dobutamine-mediated increases in MVO2 (Fig 3b), between coronary blood flow and the coronary venous – arterial [K^+^] gradient (Fig 3c), or between coronary vascular resistance and the [K^+^] gradient (Fig 3d). Using the Fick principle, the coronary venous – arterial [K^+^] gradient was multiplied by the flow rate to calculate myocardial production (positive value) or uptake (negative value) of K^+^. Dobutamine stimulation tended to cause net uptake of K^+^, as the value went from 0.01 ± 0.02 μmol/min/g at rest to −0.30 ± 0.17 μmol/min/g with the highest dose of dobutamine (P = 0.07).
Table 2 provides hemodynamic and metabolic data in response to changes in CPP. Mean blood pressure decreased from 105 ± 5 to 93 ± 4 mmHg as CPP was reduced from 140 to 40 mmHg (P < 0.001), as ischemia impaired cardiac function. Heart rate increased from 83 ± 7 beats/min to 103 ± 13 beats/min over this range of CPP (Table 2). Coronary blood flow decreased linearly with CPP (Fig 4a; P < 0.001) as did myocardial O2 delivery and MVO2 (Table 2). Myocardial lactate uptake was also significantly decreased at CPP = 40 mmHg. Reductions in CPP did not influence the coronary venous – arterial [K^+^] gradient (Fig 4b). Neither changes in coronary blood flow (Fig 4c) nor decreases in coronary vascular resistance (Fig 4d) were related to the coronary venous – arterial [K^+^] gradient.
Table 3 contains hemodynamic and metabolic data observed in response to graded reductions in PaO2. Mean blood pressure decreased from 88 ± 5 to 73 ± 6 mmHg as PaO2 was reduced from 166 ± 9 to 32 ± 5 mmHg (P < 0.001). Heart rate increased from 77 ± 15 beats/min to 112 ± 8 beats/min over this range of PaO2 (Table 3). Coronary blood flow increased exponentially as PaO2 fell (Fig 5a) such that myocardial O2 delivery and MVO2 were preserved over the entire range of hypoxemia (Table 3). Systemic hypoxemia increased both the arterial and coronary venous [K^+^] by ~15% (Table 3) and the coronary venous – arterial [K^+^] gradient was modestly related to the degree of hypoxia (Fig. 5b; P = 0.04). Similarly, the myocardial production of K^+^ (coronary venous – arterial gradient multiplied by flow; Fick principle) also tended to increase at during from 0.01 ± 0.10 to 0.18 ± 0.18 μmol/min/g (P = 0.06). Hypoxemia-induced increases in coronary blood flow were not related to the coronary venous – arterial [K^+^] gradient (Fig. 5c; P = 0.72). However, changes in coronary resistance in response to hypoxemia were modestly related to the coronary venous – arterial [K^+^] gradient (Fig 5d; P < 0.01).
Experiments were performed where coronary flow (and thus oxygen delivery) was maintained constant while intracoronary glibenclamide was infused. This was to determine whether myocardial KATP channels serve as a source for increases the coronary venous – arterial [K^+^] gradient, while preventing any possible ischemia. CPP was measured and coronary vascular resistance was calculated. Glibenclamide, predictably, produced concentration-dependent increases in coronary vascular resistance from 208 ± 19 to 269 ± 27 mmHg/ml/min/g (Fig. 6a; P < 0.05), as KATP channels are expressed in vascular smooth muscle and are important regulators of vascular tone [43]. The coronary venous – arterial [K^+^] gradient, however, did not fall as postulated when glibenclamide was added (Fig. 6b; P = 0.31); i.e. glibenclamide-mediated increases in coronary vascular resistance were not related to the coronary venous – arterial [K^+^] gradient (Fig. 6c; P = 0.39). When hypoxic coronary vasodilation was studied in constant pressure mode, glibenclamide did not affect the coronary venous – arterial [K^+^] gradient when PaO2 was reduced (< 30 mmHg; Fig 6d; P = 0.25).
The next experiment had two goals and involved using pinacidil to open KATP channels in the coronary circulation and myocardium. The first was to further examine whether opening myocardial KATP channels increases the coronary venous – arterial [K^+^] gradient. The second was to determine whether coronary endothelial cells function as a “sink” for K^+^ and thereby buffer the appearance of this ion in coronary venous blood. Endothelial cells express KATP channels [33]; therefore, opening them with pinacidil should decrease the ability of these cells to accumulate K^+^ and buffer changes in the coronary venous – arterial [K^+^] gradient. Pinacidil increased coronary blood flow in a dose-dependent manner (Fig. 7a; P < 0.05). Blood flow increased from 0.51 ± 0.14 to 1.52 ± 0.24 ml/min/g. In contrast, coronary venous K^+^ was 3.9 ± 0.1 mM under control conditions and 3.9 ± 0.2 during the highest dose of pinacidil (Fig. 7b). Myocardial KATP channels were indeed opened by pinacidil, as the Q-T interval (QTc) was shortened 10 ± 2%. There was no relationship between coronary blood flow (Fig. 7c) or coronary vascular resistance (Fig. 7d) and the coronary venous – arterial [K^+^] gradient.
Exercise increased heart rate from 121 ± 7 at rest to 259 ± 7 beats/min during the highest intensity of exercise (p < 0.05), with no significant changes in mean aortic blood pressure (Table 4). These hemodynamic changes were accompanied by increases in myocardial oxygen consumption from 166 ± 20 to 492 ± 41 μmol/min (p < 0.05), which was principally met by an increase in coronary blood flow from 49 ± 5 to 128 ± 16 ml/min (p < 0.05), as myocardial oxygen extraction (77–80%) and hence coronary venous PO2 (23–25 mmHg) remained constant (Table 4). Exercise produced small increases in arterial [K^+^] from 4.1 ± 0.1 mM at rest up to 4.5 ± 0.1 mM at the highest level of exercise, and in coronary venous [K^+^] from 4.0 ± 0.1 mM at rest to 4.5 ± 0.1 mM (Table 4). Importantly, there was no net coronary venous – arterial [K^+^] gradient either at rest or during exercise (Fig. 8). Moreover, there was no relation between the coronary venous – arterial [K^+^] gradient and either coronary blood flow or coronary vascular resistance (Fig. 8). These findings fail to provide evidence for a coronary venous – arterial [K^+^] gradient over a wide range of exercise intensities and values for coronary flow, coronary resistance, and MVO2.
Glibenclamide resulted in elevations in mean aortic blood pressure and produced marked increases coronary vascular resistance up to 45% (Table 4). However, while glibenclamide decreased both arterial and coronary venous [K^+^] (P<0.05 by two-way ANOVA, Table 4), it had no effect on the coronary venous – arterial [K^+^] gradient either at rest or during exercise (Fig. 8). The myocardial production of K^+^ (coronary venous – arterial gradient multiplied by flow; Fick principle) was not altered from the transition from rest to maximal exercise under control conditions (−9 ± 15 to 6 ± 19 μmol/min; P = 0.39) or with intravenous glibenclamide treatment (−5 ± 4 to 13 ± 24 μmol/min; P = 0.15). Nor was there an effect of KATP channel blockade (control vs. glibenclamide P = 0.40). These findings are inconsistent with the hypothesis that K^+^ released from myocardial KATP channels plays a physiological role in regulation of myocardial perfusion. Alternatively, the K^+^ concentration might be controlled by multiple factors which buffer changes when only a single type of K^+^ channel is manipulated.
This systematic series of experiments was designed to test the electro-metabolic hypothesis for the local control of coronary blood flow [32, 50]. This novel paradigm (Fig. 9) proposes that increases in interstitial [K^+^] and cardiomyocyte hyperpolarization are produced by the activity of KATP channels and serve to electrically communicate reductions in the cardiometabolic energy state with coronary microvascular membrane potential and vascular resistance [32]. Extending the prior study of Zhao et al., who reported a reliable and robust electro-metabolic vasodilation in vitro [50], we examined whether K^+^ and KATP channels produce an in vivo signal to couple coronary blood flow with the oxidative requirements of the myocardium. We measured coronary blood flow as well as coronary arterial and venous [K^+^]. There are many differences between in vitro and in vivo assessments of coronary flow that may affect conclusions made about the roles of KATP channels and K^+^ as a vasodilator (e.g., presence of erythrocytes and their effects on K^+^ dynamics, oxygen content, and whether the heart or isolated cardiac tissue is performing work); therefore, it is often difficult to make comparisons. Further, it must be emphasized that we did not measure interstitial K^+^ directly. While the interpretation of venous [K^+^] as a metric of interstitial [K^+^] is strictly valid only in the limit of equilibrium conditions with no net [K^+^] gradient, concentrations of K^+^ in the interstitial space and capillary plasma equilibrate over a time scale of less than one minute [44]. Data were collected with exogenous K^+^ added to coronary arterial blood, during dobutamine- and exercise-induced increases in MVO2, with alterations in CPP, and during systemic hypoxemia. Exogenous K^+^ caused modest (30%) vasodilation and increases in coronary venous [K^+^] (i.e., a positive value for the coronary venous – arterial [K^+^] gradient). However, no physiological stimulus applied (dobutamine-induced increases in inotropy and chronotropy, variations in CPP, or exercise) revealed an association between coronary flow and the coronary venous – arterial [K^+^] gradient. According to the electro-metabolic signaling hypothesis [50], glibenclamide-and pinacidil-sensitive KATP~ channels of cardiomyocytes are the source for proposed increases in interstitial K^+^, but our experiments show that coronary venous – arterial [K^+^] gradient does not change with glibenclamide or pinacidil treatment. Collectively, analyzing our results with 3 key criteria required to satisfy the definition of a local metabolic “metabolite” (see detailed discussion below) [12, 16] demonstrate that KATP channels and myocardial interstitial [K^+^] are unlikely to couple coronary blood flow to MVO2 during physiologic increases in cardiac work or pathophysiologic reductions in CPP. Our data do suggest that there may be a modest role for the coronary venous – arterial [K^+^] gradient when PaO2 is reduced, which is a stimulus associated with pathophysiological conditions and extreme environments and relies on mechanisms different from those involved in physiological metabolic vasodilation.
An essential aspect of any metabolic mediator is that exogenous infusion elicits physiologically relevant, concentration-dependent increases in coronary blood flow [12, 16]. We determined that exogenous administration of K^+^ directly into coronary arterial blood as either a bolus or steady-state infusion increased coronary blood flow. Our bolus K^+^ experiments validate the findings of Murray and Sparks [35], as we both demonstrate that a bolus injection of 40 μmoles of K^+^ produces a transient doubling of coronary blood flow (Fig 1). There have been no previous studies of prolonged, steady-state infusions of K^+^. Sustained intracoronary infusion of K^+^ increases coronary flow 30% (Fig 2), which produced coronary venous [K^+^] of 2.1 current in the study of Zhao et al. [50]. Hein et al. [19], show that the addition of just 2 mM extracellular K^+^ results in statistically significant dilation of porcine coronary arterioles; therefore, there is credence to the idea that lower concentrations of K^+^ produce important physiological effects. However, most studies of reactivity use K^+^ concentrations which would produce fatal arrhythmias and are clearly not physiologically relevant. We acknowledge that our studies do not assess local K^+^ concentrations in microenvironments (nor do the results presented by Zhao et al. [50] or Murray and colleagues [34, 35]). When exogenous K^+^ is added, it is the elevation of the interstitial concentration that causes arrythmia (i.e., the [K^+^] surrounding the cardiomyocytes). Because we assume that coronary venous and interstitial [K^+^] are in equilibrium, our measurements suggest that interstitial [K^+^] cannot exceed 10 mM. In fact, if some diffusion barrier for K^+^ were to exist between the coronary venous and interstitial compartments, this would imply that interstitial [K^+^] can never safely approach the 10 mM we measured in coronary venous blood and that even lower concentrations of interstitial K^+^ would be fatal. Beyond this relatively narrow physiologic concentration range of myocardial [K^+^] is recognition that coronary blood flow typically increases 2–4 fold in response to conditions such as exercise [9, 16] or hypoxemia [15, 46], whereas sustained/stead-state K^+^-induced vasodilation is of small magnitude (8–10 mM. Concentrations over this range predictably and reliably resulted in ventricular fibrillation. It is important to point out that prior in vitro studies demonstrated that a [K^+^] of 16 mM was required to elicit statistically significant dilation from rat coronary arterioles [29] and 15 mM [K^+^] was utilized to activate Kir30%). Based on our results regarding the relationship between coronary blood flow and coronary venous [K^+^] (Fig 2a), a doubling of coronary blood flow (assuming a monotonic relationship) would require elevating the steady state coronary venous [K^+^] to for smooth muscle contraction. While artificial infusion of K^+^ produces concentration-dependent increases in coronary blood flow, the degree of vasodilation over the range of exogenous [K^+^] that is physiologically relevant is relatively modest and insufficient to account for increases of coronary flow over ~30%.26 mM, well beyond the lethal limit of [K^+^] and into the range associated with the EC50
Another important consideration for any metabolic factor is whether the metabolite can be recovered from the tissue or its venous outflow under specific (patho-)physiological conditions [12, 16]. The electro-metabolic hypothesis proposes that a decrease in cardiomyocyte [ATP] produces an efflux of K^+^ through KATP channels during increases in metabolic activity or when myocardial oxygenation becomes limited [32, 50]. Accordingly, we postulated that conditions that augment MVO2 (dobutamine or exercise) or reduce tissue PO2 (decreased CPP, hypoxemia) should increase the coronary venous – arterial [K^+^] gradient. Dobutamine and exercise are two stimuli that evoke coronary vasodilation, in part, by feed forward mechanisms, while hypoxia may rely more on negative feedback (the effects of those stimuli on myocardial ATP levels are discussed below). Previous studies by Sparks and colleagues support the idea that increasing cardiac metabolic activity increases K^+^, as accelerating heart rate from 125 to 200 beats per min in anesthetized dogs was associated with a 0.53 mM increase in coronary venous [K^+^] [34].
In contrast to pacing-induced increases in MVO2, we found that coronary venous [K^+^] tended to decrease from an average of 3.8 ± 0.3 mM under baseline conditions to an average of 3.2 ± 0.2 mM at the highest dose of dobutamine. That is a 0.6 mM decrease in [K^+^] when MVO2 was increased 150% and coronary blood flow was increased (Figs. 3 and 8). One would predict a strong monotonic relationship between these variables if myocardial interstitial [K^+^] significantly contributed to local metabolic coronary vasodilation. No measures of arterial or venous [K^+^] or calculations of interstitial [K^+^] were made in the study of Zhao et al. [50]; therefore, it is difficult to determine the relationship between flow and K^+^ they might have observed.125% (Table 1). This is opposite to what Sparks and colleagues observed and is the reverse of what the electro-metabolic hypothesis predicts. However, this decrease in coronary venous [K^+^] is not unexpected, as β adrenergic stimulation of the heart increases activity of the Na^+^/K^+^-ATPase [45]. The fact that Sparks and coworkers used pacing as their stimulus rather than a catecholamine likely explains the differences in coronary venous [K^+^] between our studies. This is further supported by our observations that in awake swine there was no myocardial release of [K^+^] either at rest or during exercise (Fig. 8). Thus, the trend towards an increase in arterial [K^+^] (due to K^+^ release from exercising skeletal muscle [28]), was parallelled by a similar increase in coronary venous [K^+^]. That is, there is no measurable coronary venous – arterial [K^+^] gradient during exercise. Regardless, the change in coronary venous [K^+^] is physiologically negligible and thus consistent with the lack of a relationship between either coronary blood flow or coronary resistance and coronary venous [K^+^] during dobutamine- or exercise-induced increases in MVO2
Due to the capillary endothelium’s high permeability to K^+^, interstitial and plasma K^+^ concentrations remain equal under steady-state conditions [44]. Thus, venous plasma [K^+^] reflects interstitial [K^+^] in our steady-state infusion experiments. The observation that the level of interstitial [K^+^] required to elicit a relatively minor increase in coronary flow is equivalent to the level of interstitial [K^+^] that induces fibrillation casts serious doubt on the viability of the proposition that physiological increases in coronary flow are mediated by increases in interstitial [K^+^]. Moreover, if interstitial (and venous) [K^+^] were elevated during exercise to levels required to cause meaningful increases in myocardial perfusion, the myocardium would be depleted of K^+^ within several minutes of sustained exercise. Consistent with this interpretation, increases in MVO2 do not lead to sustained coronary venous – arterial [K^+^] differences.
A reason for the lack of a marked increase in the coronary venous – arterial [K^+^] gradient to increases in MVO2 is consistent with prior studies that support essentially no change in myocardial energetics ([Phosphocreatine]/[ATP]) in response to increases in MVO2 [1, 20, 24]. In other words, there is no reduction of high energy phosphates to act as an error signal to activate myocardial KATP channels. Alternatively, high energy phosphates decrease within seconds of limitations of myocardial perfusion [5, 18, 41]. Thus, we sought to further interrogate the electro-metabolic hypothesis during reductions in CPP and PaO2. Diminishing CPP from 140 mmHg to 40 mmHg significantly decreased coronary blood flow 54% and reduced myocardial lactate uptake by decreased from ~166 mmHg to ~32 mmHg) increased coronary blood flow (~100%), and only modestly increased the coronary venous – arterial [K^+^] gradient (Fig. 5; Table 3). Thus, our findings do not support the contribution of myocardial [K^+^] to the control of coronary flow during ischemia (reduced CPP) but are potentially consistent with a modest role for this mechanism when tissue oxygenation (myocardial energetics) is limited by hypoxemia. However, the small increase in K^+^ associated with this level of hypoxia assayed falls far short of the calculated increase in interstitial [K^+^] required to elicit the observed increase in flow. More likely other mechanisms, such as myocardial adenosine production, are much more important contributors to vasodilation under this condition.70% (Table 2). Despite these appreciable reductions, coronary venous – arterial [K^+^] remained unchanged over the entire range of CPP (Fig 4b) and thus was not associated with pressure-induced reductions in coronary flow (Fig 4c) or coronary resistance (Fig 4d). Systemic hypoxemia (PaO2
A crucial test of the relevance of a proposed metabolic pathway is determining whether blocking steps in its signaling pathway lessen the physiologic response [12, 16]. The electro-metabolic hypothesis of Zhao et al. [50] suggests that the source of interstitial K^+^ in the heart is KATP channels of cardiac myocytes. These myocardial KATP channels are sensitive to blockade by glibenclamide [40], as are the KATP channels of arterial smooth muscle [36]. A clear prediction from this is that glibenclamide-induced coronary vasoconstriction could be due a) a reduction in myocardial interstitial [K^+^] (i.e., an electro-metabolic effect) or b) depolarization of the coronary smooth muscle (i.e., a direct vascular effect). To examine this, we performed dose-response studies to intracoronary glibenclamide (0.3–3.0 mg/min) in our cannulated swine preparation at constant coronary blood flow. The rationale for this approach was that inhibition of the electro-metabolic pathway would result in concentration-dependent reductions in the coronary venous – arterial [K^+^] gradient, independent of any effect of glibenclamide on coronary blood flow. Furthermore, direct vascular smooth muscle effects of glibenclamide would produce increases in CPP without confounding effects on overall perfusion. Utilizing this approach with a constant oxygen delivery, we determined that glibenclamide dose-dependently increased coronary resistance (Fig 6a) but did not reduce the coronary venous – arterial [K^+^] gradient as predicted by the electro-metabolic hypothesis (Fig 6b). Additional studies also showed that glibenclamide (3.0 mg/min, ic) failed to attenuate hypoxemia-induced increases in the coronary venous – arterial [K^+^] gradient (Fig 6d), suggesting that myocardial KATP channels may not be a meaningful source of interstitial K^+^ even during hypoxia. Thus, our in vivo inhibition studies demonstrate that concentrations of glibenclamide sufficient to increase coronary resistance by 30% do not influence the coronary venous – arterial [K^+^] gradient. Accordingly, the effects of glibenclamide are best explained by direct effects on vascular smooth muscle to cause vasoconstriction [43] while myocardial KATP~ channels do not appear to be a significant and quantifiable source of myocardial interstitial K^+^. Finally, our observations in exercising swine also fail to support a role for myocardial KATP channels as a source of K^+^ release, as no change in coronary venous – arterial [K^+^] gradient was observed following glibenclamide, administered in a dose (3.0 mg/kg intravenous) that produced a 30–40% increase in coronary vascular resistance (Fig. 8). Accordingly, the effects of glibenclamide are consistent with direct (i.e., smooth muscle) vasoconstriction of coronary circulation [43], while myocardial KATP channels do not appear to be a significant and quantifiable source of a coronary venous – arterial [K^+^] gradient during exercise. Conversely, this idea is reinforced by experiments with pinacidil, which opens KATP channels on cardiac myocytes, vascular smooth muscle, and endothelium, but failed to increase the coronary venous – arterial [K^+^] gradient (Fig. 7).
There are two major limitations to our study that bear mentioning. First, our experiments in swine, while producing generally consistent results, are based on relatively small sample sizes. Basing conclusions on a small sample size increases the chance of a false negative finding or type II error. However, as our findings were consistent across multiple experiments where we did not observe positive coronary venous – arterial K^+^ gradients, a type II error would seem unlikely. Second, our study assumes that the coronary venous – arterial K^+^ gradient can serve as a surrogate measure for the interstitial concentration of K^+^. This is an indirect assessment and relies on the expectation that, in the steady state, the [K^+^] concentrations of the interstitial and coronary venous compartments are in equilibrium. This approach is routinely used to measure the production or consumption of a substance across the coronary circulation (e.g., MVO2, lactate uptake). The most direct assessment of the interstitial K^+^ concentration would be measurements of dialysate sampled from myocardial interstitium; however, this approach has its own complicating factors such as tissue damage and subsequent leakage of K^+^ from ruptured cells.
Because it satisfies none of the three major criteria described, we conclude that myocardial interstitial [K^+^] is unlikely to couple coronary blood flow to MVO2 during physiologic increases in cardiac work or pathophysiologic reductions in CPP (which mimics ischemia). There may be a small, but significant role when PaO2 is reduced (i.e., during hypoxemia). This electro-metabolic signaling hypothesis is intriguing and merits further research; however, one of the primary issues to address is the proposed interstitial K^+^ concentrations involved, as they are seemingly far too high to be compatible with normal cardiac electrical activity. Peaked T-waves are one of the first electrophysiogical indications of hyperkalemia (starting at around 6 mM K^+^) and yet peaked T-waves are not observed during exercise in healthy individuals; therefore, it is unlikely that the required increase in myocardial interstitial K^+^ occurs during normal physiological coronary metabolic vasodilation. Further, it would not appear helpful to argue that there is some restricted space between cardiac myocytes and vascular endothelium where the local K^+^ concentration may be higher, as this would worsen the effects on cardiac electrical activity, rhythm, and mechanical function. It will, however, be important to more directly measure the interstitial [K^+^] and model the degree to which K^+^ efflux from myocardial KATP channels could increase interstitial [K^+^].
Discrepancies between the present investigation and those supporting the electro-metabolic signaling hypothesis [32] remain and could be related to in vivo vs. in vitro preparations, as Zhao et al. [50] studied ventricular/septal and papillary muscle preparations perfused with crystalloid buffer solution. Specifically, the lack of oxygen content in crystalloid buffers (despite the high PO2) may exaggerate effects attributable to KATP channels, as such preparations are more likely to be hypoxic and have difficulty generating ATP from oxidative metabolism. The absence of erythrocytes in the in vitro preparation may also affect the K^+^ concentration, as they could be as source and sink for this ion. Additional difficulties with the electro-metabolic signaling paradigm relate to membrane potential and the transfer of that voltage from endothelial cells to smooth muscle cells. The study of Zhao et al. [50] determined that microvascular endothelial cells have a membrane potential of −28 mV, which is hyperpolarized to −33 mV by the addition of 10 mM K^+^. This hyperpolarization is blocked by Ba^2+^, which is a clear indicator that it is mediated by Kir channels [50]. Importantly, however, this endothelial cell hyperpolarization cannot be expected to cause smooth muscle hyperpolarization, as the membrane potential of coronary smooth muscle is significantly hyperpolarized (−45 mV) relative to endothelial cells [29]. Finally, the electro-metabolic hypothesis includes another proposed mechanism to hyperpolarize the vasculature through effects on cardiac KATP hyperpolarization of cardiac myocytes by KATP channel activation is suggested to pass through electrical connections to microvascular endothelial cells and elicit a propagated vasodilation. This seems unlikely, as the resting potential of cardiac myocytes is essentially at the equilibrium potential for K^+^, leaving little room for KATP-induced hyperpolarization. KATP channel activation clearly shortens the cardiac action potential, which by itself would reduce the time averaged depolarization due to cardiac electrical activity. Importantly, however, physiological increases in cardiac metabolic activity are associated with increases in heart rate, which would negate the effect of shorter action potential duration and produce a time averaged depolarization of microvascular endothelial cells. It will be important to model cardiac action potentials and more closely inspect the electrical effect this would have on vascular endothelial cells.
Zhao et al. [50] have proposed an intriguing idea, forwarded an elegant hypothesis, and demonstrated convincing data from their in vitro preparations. This electro-metabolic paradigm, however, does not appear to be of great physiological significance under the conditions of our experiments in vivo. Our data indicate that K^+^-induced changes in coronary vascular resistance are unlikely to be critical for physiological metabolic vasodilation in the steady-state. The conclusion of Murray and Sparks is still valid in that K^+^ might play a role in the onset of coronary metabolic dilation, and we demonstrate a potential for role K^+^ for in hypoxic vasodilation. It may be in the initiation of responses and in hypoxic mechanisms where the in vitro work of Zhao et al. intersects in vivo studies.