Metabolic hyperemia requires ATP-sensitive K⁺ channels and H₂O₂ but not adenosine in isolated mouse hearts

Abstract

We have previously demonstrated that adenosine-mediated H₂O₂ production and opening of ATP-sensitive K⁺ (K_ATP) channels contributes to coronary reactive hyperemia. The present study aimed to investigate the roles of adenosine, H₂O₂, and K_ATP channels in coronary metabolic hyperemia (MH). Experiments were conducted on isolated Langendorff-perfused mouse hearts using combined pharmacological approaches with adenosine receptor (AR) knockout mice. MH was induced by electrical pacing at graded frequencies. Coronary flow increased linearly from 14.4 ± 1.2 to 20.6 ± 1.2 ml·min⁻¹·g⁻¹ with an increase in heart rate from 400 to 650 beats/min in wild-type mice. Neither non-selective blockade of ARs by 8-(p-sulfophenyl)theophylline (8-SPT; 50 μM) nor selective A_2AAR blockade by SCH-58261 (1 μM) or deletion affected MH, although resting flow and left ventricular developed pressure were reduced. Combined A_2AAR and A_2BAR blockade or deletion showed similar effects as 8-SPT. Inhibition of nitric oxide synthesis by N-nitro-l-arginine methyl ester (100 μM) or combined 8-SPT administration failed to reduce MH, although resting flows were reduced (by ∼20%). However, glibenclamide (K_ATP channel blocker, 5 μM) decreased not only resting flow (by ∼45%) and left ventricular developed pressure (by ∼36%) but also markedly reduced MH by ∼94%, resulting in cardiac contractile dysfunction. Scavenging of H₂O₂ by catalase (2,500 U/min) also decreased resting flow (by ∼16%) and MH (by ∼24%) but to a lesser extent than glibenclamide. Our results suggest that while adenosine modulates coronary flow under both resting and ischemic conditions, it is not required for MH. However, H₂O₂ and K_ATP channels are important local control mechanisms responsible for both coronary ischemic and metabolic vasodilation.

myocardial function is dependent on a constant O₂ supply from the coronary circulation to myocytes. Myocardial O₂ consumption (mvo₂) increases whenever there is enhanced cardiac work. Because the myocardium has a very limited anerobic capacity with a high O₂ extraction (75%) (46) under resting conditions, the increased O₂ delivery during exercise mostly relies on increased coronary blood flow, termed as metabolic hyperemia. Numerous mechanisms are responsible for coronary metabolic hyperemia, allowing a balance between myocardial O₂ demand and supply. Disturbance of these mechanisms results in decreased cardiac output, hypotension, arrhythmia, heart failure, and death.

Metabolic coronary vasodilation has been found to be regulated by both local metabolic feedback and adrenergic feedforward mechanisms (46). For decades, adenosine, a locally released metabolite from the myocardium, has been postulated as one of the important mechanisms responsible for metabolic coronary hyperemia (3, 4, 6, 16, 17, 19, 28). However, recent evidence suggests that adenosine is not required for metabolic coronary vasodilation (4, 14, 47–49), and its role in regulating resting coronary flow (CF) still remains controversial (14, 25–27, 49). The discrepancy may be due to different animal models, differences in species, and/or the different agonists and antagonists applied in these studies. It is well established that adenosine exerts its effect through activation of four receptor subtypes, namely, A₁ and A₃ adenosine receptors (ARs), which exert constrictive effects, and A_2A and A_2BARs, which have dilatory effects (31). The majority of these studies to characterize the functional role of adenosine in coronary hyperemia have relied heavily on adenosine-related pharmacological compounds, which may lead to misinterpretation of the data due to the mixed specificity of the agonists and antagonists for the four different ARs. In addition, the complexity of the combined intrinsic (e.g., myogenic, shear-mediated, metabolic, and endothelium-originated mediators) with extrinsic (neural hormonal regulatory mechanism) blood flow control mechanisms make whole animal study a nonoptimal model for the mechanistic exploration in local metabolic coronary blood flow regulation. Therefore, there is a need to reexamine the functional role of adenosine in coronary metabolic hyperemia using combined AR knockout (KO) mice with traditional pharmacological agents in isolated Langendorff-perfused hearts.

Increasing evidence suggests that H₂O₂, an endogenous metabolite, acts as an important mediator responsible for metabolic coronary vasodilation (37, 42, 53, 54). We (42, 57) have previously demonstrated that A_2AAR-mediated H₂O₂ production opens ATP-sensitive K⁺ (K_ATP) channels in coronary smooth muscle cells, contributing to the coronary ischemic vasodilation (42, 57). While the K_ATP channel is known to be important for coronary reactive hyperemia (1, 2, 10, 11, 42), it remains controversial regarding its role in metabolic hyperemia (13, 15, 16, 32, 34, 35, 47, 51). Therefore, the present study further examined whether H₂O₂ and/or K_ATP channels play an important role in coronary metabolic hyperemia using isolated Langendorff-perfused mouse hearts in the presence of catalase (an enzyme that decomposes H₂O₂ to water) or glibenclamide (a selective K_ATP channel blocker).

MATERIALS AND METHODS

Animals.

The Institutional Animal Care and Use Committee of West Virginia University School of Medicine approved all experimental protocols. We followed guidelines set forth by the American Physiological Society and National Institutes of Health regarding the care and use of laboratory animals. A_2AAR and A_2BAR single KO mice (A_2A KO and A_2B KO mice, respectively) were generously provided by Dr. C Ledent (Universite Libre de Bruxelles, Brussels, Belgium) and Stephen Tilley (University of North Carolina, Chapel Hill, NC), respectively. A_2A and A_2B KO mice, both backcrossed 12 generations to the wild-type (WT) C57BL/6 background (Jackson Laboratory, Bar Harbor, ME), were bred to generate A_2A/A_2B double-KO (DKO) heterozygotes. Double heterozygotes were intercrossed, and 1/16 of the offspring were A_2A/A_2B DKO mice. A_2A/A_2B DKO breeding pairs were then established. Mice were caged in a 12:12-h light-dark cycle with free access to standard chow and water. The absence of A_2AAR and A_2BAR at both mRNA and protein levels in A_2A KO, A_2B KO, and A_2A/2B DKO mice has been confirmed by our previous studies using PCR, Western blot analysis, and immunohistochemistry in the aorta, mesenteric arteries, and coronary arteries (30, 41, 43–45, 57). Our functional data also confirmed the lack of a CGS-21680 (a selective A_2AAR agonist)-induced CF response in A_2A KO and A_2A/2B DKO mice, the lack of a BAY 60-6583 (selective A_2BAR agonist) response in A_2B KO and A_2A/2B DKO mice, and the lack of an adenosine response in A_2A/2B DKO mice (41, 45).

Langendorff-perfused mouse heart preparations.

Mice (14–17 wk) of either sex (equal ratio) were anesthetized with pentobarbital sodium (50 mg/kg body wt ip), and hearts were excised into heparinized (5 U/ml) ice-cold Krebs-Henseleit buffer containing (in mM) 119 NaCl, 11 glucose, 22 NaHCO₃, 4.7 KCl, 1.2 KH₂PO₄, 1.2 MgSO₄, 2.5 CaCl₂, 2 pyruvate, and 0.5 EDTA. After removal of the lung and surrounding tissue, the aorta was rapidly cannulated with a 20-gauge, blunt-ended needle and continuously perfused with 37°C buffer bubbled with 95% O₂ and 5% CO₂ at a constant perfusion pressure of 80 mmHg. The left atrium was removed, and a fluid-filled balloon made of plastic wrap was inserted into the left ventricle (LV) across the mitral valve. The balloon was connected to a pressure transducer for the continuous measurement of LV developed pressure (LVDP). The heart was then immersed in a water-jacketed perfusate bath maintained at 37°C and allowed to beat spontaneously. LV diastolic pressure was adjusted to 2–5 mmHg. CF was continuously measured with an ultrasonic flow probe (Transonic Systems, Ithaca, NY) placed in the aortic perfusion line. A Power Lab Chart data-acquisition system (AD Instruments, Colorado Springs, CO) was used for data acquisition. After stabilization for 20 min, hearts were then paced at 400 beats/min by two platinum wires connected to the stainless steel cannula and through the apical myocardium. Function was allowed to stabilize for another 10 min before baseline functional data were collected. The heart was then paced at gradually increasing frequencies (500, 550, 600, and 650 beats/min) to induce metabolic hyperemia. Hearts with persistent arrhythmias or LVDP of <80 mmHg were excluded from the study.

Metabolic hyperemia protocol.

After stabilization, hearts were subjected to a 2 min of pacing at each frequency of 500, 550, 600, and 650 beats/min, respectively. The reason why we paced the heart from 400 to 650 beats/min (1.62-fold increase) was based on the fact that heart rate (HR) has been reported to increase by ∼1.9-, 1.8-, 1.8-, and 1.7-fold in humans, swine, dogs, and mice, respectively, upon maximum exercise (13, 17, 29). At the end of each pacing, the heart was allowed to stabilize for 5 min at a HR of 400 beats/min before the heart was paced at a different frequency. 8-(p-Sulfophenyl)theophylline (8-SPT; nonselective adenosine receptor antagonist, 50 μM, Sigma), N-nitro-l-arginine methyl ester [l-NAME; nitric oxide (NO) synthase inhibitor, 100 μM, Sigma, based on our previous study (45) showing that l-NAME at ≥10⁻⁵ reached its maximum inhibition on resting CF and 5′-(N-ethylcarboxamido)adenosine (nonselective adenosine agonist)-induced CF responses], SCH-58261 (selective A_2AAR antagonist, K_i = 1.3 nM, 1 μM, Tocris Bioscience), CVT-6883 (selective A_2BAR antagonist, K_i = 22 nM, 1 μM, Tocris Bioscience), catalase (the enzyme that decomposes H₂O₂ to water, 2,500 U/ml, Sigma), or glibenclamide (K_ATP channel blocker, 5 μM, Sigma) were delivered into the aortic perfusion line for 15 min using a microinjection pump (Harvard Apparatus, Holliston, MA) as 1% of CF. In the presence of each pharmacological reagent, a second pacing-induced functional hyperemia was elicited. Time-matched control experiments with two consecutive inductions of functional hyperemia showed no differences in baseline function as well as the magnitude of pacing-induced changes in CF (n = 3 animals).

Data analysis and statistics.

All data collected were analyzed by LabChart 7.0 software. Baseline CF, LVDP, and dP/dt were calculated as mean values after each heart was paced at 400 beats/min for 10 min. Mean CF was calculated during the 2-min pacing at each frequency. Because absolute CF rates changed proportionally with heart mass, CF was normalized by heart mass and presented as milliliters per gram of wet heart weight per minute. Because mvo₂ is a function of HR, cardiac contractility, ventricular wall tension, and muscle shortening (9, 12, 20), it is represented by the rate-pressure product (RPP). To examine the relationship between mvo₂ and RPP, Po₂ of the inflow (Pi{O₂}; an indication of arterial O₂ tension) and outflow perfusate (Po{O₂}; an indication of venous O₂ tension collected through a tube inserted in the right atrium close to the coronary sinus) were measured using a GEM Premier 4000 gas analyzer (n = 2). Because of the absence of hemoglobin in the perfusion buffer, O₂ content is dependent solely on the O₂ solubility in the buffer at 37°C, which was 0.023 ml O₂·ml buffer⁻¹·atm⁻¹ (39). mvo₂ was calculated using the following equation: mvo₂ = CF × (Pi{O₂} − Po{O₂}) × 0.023/760, as previously described (39). As expected, mvo₂ linearly correlated with RPP as hearts were paced at graded frequencies (mvo₂ =1.49 × RPP + 101.5, R² = 0.9596, P < 0.01), indicating that mvo₂ is a function of the HR and LVDP product.

Least-square linear regression analysis was conducted between HR or RPP, and CF to calculate the slope of the relationship. The slope of each curve was compared before and after drug treatment or between groups. All data are presented as means ± SE; n represents the number of animals unless otherwise indicated. Paired t-test was used for paired data analysis. One-way ANOVA was used to compare the data between groups. The effect of treatment on the relation between two variables was analyzed by analysis of covariance using GraphPad Prism5. P values of <0.05 were considered as statistically significant.

RESULTS

Baseline function from WT, A_2A KO, A_2B KO, and A_2A/2B DKO mouse hearts.

Baseline functional parameters of isolated mouse hearts were recorded at 10 min of pacing (400 beats/min) after 20-min stabilization. There were no statistically significant differences in the heart-to-body weight ratio and maximum −dP/dt among all animal strains, although +dP/dt had a trend to be less in A_2A KO, A_2B KO, or A_2A/2B DKO mice compared with WT mice. Baseline CF as well as LVDP were significantly lower in KO mice compared with WT mice (P < 0.05 vs. WT mice; Table 1).

Table 1.

Baseline functional data in isolated WT, A_2A KO, A_2B KO, and A_2A/2B DKO mouse hearts

	WT	A2A KO	A2B KO	A2A/2B DKO
No. of animals	23	12	6	6
Age, wk	14.8 ± 0.17	14.8 ± 0.20	14.3 ± 0.25	15.0 ± 0.10
Body weight, g	24.3 ± 0.82	25.1 ± 1.14	22.3 ± 0.96	25.0 ± 1.49
Heart weight, mg	112.4 ± 2.74	109 ± 3.9	103 ± 4.7	113 ± 8.1
Heart-to-body weight ratio, %	0.46 ± 0.005	0.43 ± 0.010	0.46 ± 0.001	0.45 ± 0.020
Coronary flow, ml·min−1·g−1	17.1 ± 0.54	14.4 ± 0.56*	14.1 ± 1.78*	13.4 ± 0.52*
Left ventricular developed pressure, mmHg	105 ± 4.0	90 ± 3.9*	93 ± 5.3	88 ± 2.5*
+dP/dt, mmHg/s	4,704 ± 233.7	4,158 ± 278.6	3,878 ± 313.9	4,078 ± 249.0
−dP/dt, mmHg/s	3,582 ± 169.5	3,743 ± 272.5	3,167 ± 164.1	3,245 ± 404.4

All values are means ± SE. WT, wild type; A_2A KO, A_2A adenosine receptor knockout; A_2B KO, A_2B adenosine receptor knockout; A_2A/AB DKO, A_2A/2B adenosine receptor double knockout.

^*P < 0.05 compared with WT.

Pacing-induced coronary hyperemia in the absence or presence of 8-SPT.

The CF response in isolated hearts of WT mice was measured before and after the addition of 8-SPT (50 μM, a nonselective adenosine antagonist, n = 6). Under control conditions, mean CF increased significanlty from 14.4 ± 1.2 to 16.8 ± 1.1, 18.5 ± 1.3, 19.5 ± 0.9, and 20.6 ± 1.2 ml·min⁻¹·g⁻¹ when hearts were paced from 400 to 500, 550, 600, and 650 beats/min, respectively (Fig. 1A). The increased CF linearly correlated with the enhanced mvo₂, indicated as RPP, which increased from 41 ± 5.1 to 56 ± 5.1, 65 ± 6.7, 70 ± 6.1, and 75 ± 7.0 ×10³ beats·min⁻¹·mmHg, respectively (P < 0.05; Fig. 1B). LVDP increased significanlty from 102 ± 12.7 to 112 ± 10.0, 117 ± 12.1, 117 ± 10.0, and 115 ± 10.7 mmHg, respectively (P < 0.05; Fig. 1C). The linear relationship between CF and RPP suggests a balance between O₂ delivery and consumption in pacing-induced metabolic hyperemia in isolated Langendorff-perfused mouse hearts.

Fig. 1.

Effect of 8-(p-sulfophenyl)theophylline (8-SPT; 50 μM) on the pacing-induced coronary flow (CF) response. A–C: relationship between heart rate [HR; in beats/min (bpm)] and CF (A), rate-pressure product (RPP) and CF (B), and HR and left ventricular developed pressure (LVDP; C) before and after the addition of 8-SPT (n = 6). D: adenosine (10⁻⁹–10⁻⁵M)-induced coronary response before and after the infusion of 8-SPT (50 μM, n = 3). *P < 0.05, significant difference between the curves.

In the presence of 8-SPT, resting CF significantly decreased from 14.4 ± 1.20 to 10.8 ± 0.91 ml·min⁻¹·g⁻¹ (P < 0.05 by paired t-test). The decreased CF was associated with a correlated decrease in baseline LVDP from 102 ± 12.7 to 80 ± 6.5 mmHg (P < 0.05; Fig. 1C). However, the pacing-induced increase in CF was not prevented by 8-SPT, although the CF at each given HR or RPP was decreased (Fig. 1, A and B). CF increased from 10.8 ± 0.91 to 16.0 ± 0.61 ml·min⁻¹·g⁻¹ when hearts were paced from 400 to 650 beats/min, and the linear curve of the relationship between HR or RPP and CF was shifted downward in a parallell manner without significant changes in the slope before or after 8-SPT treatment (0.187 ± 0.001 vs. 0.185 ± 0.002, P > 0.05; Fig. 1, A and B).

To confirm the antagonistic ability of 8-SPT against adenosine, we examined the effect of 8-SPT at the same concentration (50 μM) on adenosine (10⁻⁹∼10⁻⁵ M)-induced CF responses. 8-SPT significantly decreased both baseline CF (from 13.2 ± 0.55 to 10.8 ± 0.52 ml·min⁻¹·g⁻¹, n = 3, P < 0.05) as well as the adenosine-induced CF increase (maximal coronary flow was reduced from 35 ± 1.4 to 15.4 ± 1.5 ml·min⁻¹·g⁻¹, P < 0.05; Fig. 1D).

Combined effect of l-NAME and 8-SPT on pacing-induced coronary hyperemia.

Baseline CF was significantly decreased by l-NAME (10⁻⁴ M) from 18.0 ± 0.86 to 14.8 ± 0.78 ml·min⁻¹·g⁻¹ (P < 0.05, n = 8; Fig. 2A). However, the pacing-induced CF increase was not affected by l-NAME (CF increased linearly from 14.8 ± 0.78 to 22.9 ± 1.53 ml·min⁻¹·g⁻¹ when hearts were paced from 400 to 650 beats/min; Fig. 2, A and B). The slope of the relationship between HR or RPP and CF was not significantly changed before or after the addition of l-NAME (P > 0.05), and the linear curve was shifted downward in a parallell manner. Concomitant with the decreased baseline flow, LVDP was signficantly diminshed by l-NAME from 110 ± 5.9, 130 ± 4.9, 136 ± 3.5, 135 ± 3.5, and 134 ± 3.4 to 88 ± 6.1, 104 ± 6.4, 112 ± 44.8, 112 ± 4.8, and 116 ± 4.1 mmHg when hearts were paced at 400, 500, 550, 600, and 650 beats/min, respectively (P < 0.05; Fig. 2C).

Fig. 2.

Effect of N-nitro-l-arginine methyl ester (l-NAME) on the pacing-induced CF response in the absence or presence of 8-SPT. A–C: relationship between HR and CF (A), RPP and CF (B), and HR and LVDP (C) under control conditions (n = 8), in the presence of l-NAME (nitric oxide synthase inhibitor, 100 μM, n = 4), and in the combined presence of 8-SPT (50 μM) and l-NAME (100 μM, n = 4). *P < 0.05, significant difference from the control curve (A and B) and versus control (C).

In four of eight animals, 8-SPT (50 μM) was added in the presence of l-NAME. 8-SPT did not further decrease baseline CF, nor did it affect pacing-induced hyperemia (P > 0.05 vs. l-NAME; Fig. 2, A–C).

Effect of A_2AAR blockade or deletion on pacing-induced coronary hyperemia.

The role of A_2AARs in coronary functional hyperemia was examined in WT mice (n = 6) in the presence of SCH-58261 (1 μM, a selective A_2AAR antagonist). SCH-58261 significantly decreased baseline CF from 17.5 ± 1.54 to 11.3 ± 0.70 ml·min⁻¹·g⁻¹ (P < 0.05; Fig. 3, A and B) and LVDP from 101 ± 3.5 to 70 ± 5.8 mmHg, respectively (P < 0.05; Fig. 3C). However, after A_2AAR blockade, CF still increased from 11.3 ± 0.70 to 16.3 ± 0.54 ml·min⁻¹·g⁻¹ when hearts were paced from 400 to 650 beats/min. The slope of the relationship between HR or RRP and CF was not different before or after A_2AAR blockade (P > 0.05; Fig. 3, A and B).

Fig. 3.

Effect of A_2A adenosine receptor (AR) blockade or deletion on pacing-induced coronary hyperemia. A–C: relationship between HR and CF (A), RPP and CF (B), and HR and LVDP (C) in control wild-type (WT) mice (n = 6), A_2AAR knockout (A_2A KO) mice (n = 9), and WT mice treated with SCH-58261 (selective A_2AAR antagonist, 1 μM, n = 6). *P < 0.05, significant difference between the curves or significant difference versus control; #P < 0.05, significant difference versus A_2A KO mice.

We further examined the effect of A_2AAR deletion on pacing-induced coronary hyeremia using A_2A KO mice (n = 9). Baseline CF and LVDP were 14.4 ± 0.56 ml·min⁻¹·g⁻¹ and 90 ± 3.9 mmHg, respectively, which were significantly lower than those in WT mice (Table 1 and Fig. 3) but were higher than those in WT mice treated with SCH-58261 (P < 0.05; Fig. 3, A and C). Consistent with the effect of SCH-58261, A_2AAR deletion did not affect the pacing-induced CF increase, although receptor deletion resulted in a parallel downward shift of the relationship between RPP or HR and CF (Fig. 3, A and B).

Effect of double blockade or deletion of A_2AAR and A_2BAR on pacing-induced coronary hyperemia.

Both A_2AARs and A_2BARs are involved in coronary vasodilation (8, 41, 45). Therefore, we examined the effect of double blockade or deletion of both ARs on pacing-induced coronary hyperemia. In A_2A KO mice (n = 4), CVT-6883 (1 μM) further decreased baseline CF from 13.5 ± 1.03 to 11.6 ± 0.75 ml·min⁻¹·g⁻¹ (P < 0.05) and LVDP from 91 ± 2.3 to 71 ± 5.7 mmHg (P < 0.05; Fig. 4D). Similarly, in A_2B KO mice, which have lower baseline CF (14.1 ± 1.78 ml·min⁻¹·g⁻¹, n = 6) and LVDP (93 ± 5.3 mmHg) than WT mice (P < 0.05; Table 1 and Fig. 4D), combined blockade of A_2AARs by SCH-58261 (1 μM) further decreased baseline CF and LVDP to 10.6 ± 0.65 ml·min⁻¹·g⁻¹ and 74 ± 3.3 mmHg, respectively (P < 0.05; Fig. 4D). However, in both cases, pacing-induced hyperemia was not affected (CF increased from 11.0 ± 0.58 and 10.6 ± 0.65 to 16.0 ± 0.81 and 15.7 ± 0.77 ml·min⁻¹·g⁻¹, respectively when hearts were paced from 400 to 650 beats/min; Fig. 4, A–C). The slope of the relationship between HR and/or RPP and CF under each experimental condition was not significantly different from WT mice (Fig. 4, A–C).

Fig. 4.

Effect of double blockade or deletion of A_2AARs and A_2BARs on the pacing-induced CF response. A–D: relationship between HR and CF (A), RPP and CF (B and C), and HR and LVDP (D) in WT mice (n = 6), A_2A KO mice (n = 4), A_2BAR knockout (A_2B KO) mice (n = 6), A_2A KO mice with CVT-6883 (selective A_2BAR antagonist, 1 μM, n = 4), and A_2B KO mice with SCH-58261 (1 μM, n = 6). E and F: the relationship between HR and CF (E) and the relationship between RPP and CF (F) were compared between WT mice (n = 6) and A_2A/2BAR double-knockout (A_2A/2B DKO) mice (n = 6). *P < 0.05, significant difference between the curves or significant different versus control.

To confirm the role of A_2AARs and A_2BARs in coronary metabolic hyperemia, we further examined the hyperemic response in A_2A/2B DKO mice (n = 6). Baseline CF (13.4 ± 0.52 ml·min⁻¹·g⁻¹) and LVDP (88 ± 2.5 mmHg) were lower than those in WT mice (P < 0.05) and were less than those in either A_2A KO or A_2B KO mice, although they did not reach statistical significance (Table 1 and Fig. 4, E and F). However, the pacing-induced hyperemic response was not affected (CF increased from 13.3 ± 0.52 to 19.9 ± 0.73 ml·min⁻¹·g⁻¹, linearly correlated with the increased HRs from 400 to 650 beats/min; Fig. 4E). The slope of the relationship between HR and/or RPP and CF in A_2A/2B DKO mice was not significantly different from that in WT mice, suggesting that neither A_2AARs nor A_2BARs are required for the pacing-induced increase in CF.

Effect of catalase on pacing-induced coronary hyperemia.

The role of H₂O₂ in pacing-induced hyperemia was examined in WT mice (n = 6) before and after the addition of catalase (2,500 U/ml). Catalase significantly decreased CF from 16.5 ± 1.22 to 13.8 ± 1.40 ml·min⁻¹·g⁻¹ (Fig. 5A) and significantly decreased LVDP from 96 ± 7.4 to 88 ± 7.1 mmHg (Fig. 5C), respectively (P < 0.05 by paired t-test). Additionally, catalase significantly attenuated the pacing-induced increase in CF by ∼37% (the net increase in CF of hearts paced from 400 to 650 beats/min was decreased from 7.4 ± 0.12 to 4.6 ± 0.49 ml·min⁻¹·g⁻¹, P < 0.05). The slope of the relationship between HR and/or RPP and CF in the presence of catalase was significantly lower than that under control conditions (P < 0.05 by paired t-test; Fig. 5, A and B).

Fig. 5.

Effect of catalase on pacing-induced coronary hyperemia. A–C: relationship between HR and CF (A), RPP and CF (B), and HR and LVDP (C) before and after the addition of catalase (the enzyme that decomposes H₂O₂ to water, 2,500 U/ml, n = 6). *P < 0.05, significant difference in the slope of the linear curves (A and B) and significant difference versus control (C, paired t-test).

Effect of glibenclamide on pacing-induced coronary hyperemia.

The role of K_ATP channels in pacing-induced coronary hyperemia was determined in isolated WT mouse hearts (n = 6) in the absence or presence of glibenclamide (K_ATP channel blocker, 5 μM). In the absence of glibenclamide, CF increased by 6.2 ml·min⁻¹·g⁻¹ after hearts were paced from 400 to 650 beats/min (Fig. 6, A and B). After glibenclamide infusion, CF decreased significantly by ∼ 45% (from 20 ± 1.6 to 11 ± 0.9 ml·min⁻¹·g⁻¹, P < 0.05; Fig. 6, A and B) and LVDP was reduced by ∼36% (from 112 ± 13.3 to 72 ± 7.0 mmHg, P < 0.05; Fig. 6C). Moreover, pacing-induced hyperemia was dramatically decreased by ∼94% (the net CF increase before and after glibenclamide was 6.1 and 0.39 ml·min⁻¹·g⁻¹, respectively; Fig. 6, A and B). The dramatic reduction in functional hyperemia was concomitant with a significant decrease in LVDP during pacing (LVDP decreased from 72 ± 7.0 to 63 ± 9.0, 57 ± 4.8, and 59 ± 9.2 mmHg when hearts were paced at 550, 600, and 650 beats/min, respectively; Fig. 6C), suggesting compromised heart function as a result of an imbalance between O₂ supply and demand after K_ATP channel blockade.

Fig. 6.

Effect of ATP-sensitive K⁺ channel blockade on pacing-induced coronary hyperemia. A–C: relationship between HR and CF (A), RPP and CF (B), and HR and LVDP (C) before and after the addition of glibenclamide (ATP-sensitive K⁺ channel blocker, 5 μM, n = 6). *P < 0.05, significant difference in the slope of the linear curves (A and B) and significant difference versus control; #P < 0.05 vs. LVDP at a HR of 400 beats/min.

DISCUSSION

In the present study, using targeted AR KO mice combined with pharmacological approaches, we further examined the role of adenosine, H₂O₂, and K_ATP channels in coronary metabolic hyperemia. In isolated Langendorff-perfused mouse hearts from A_2A KO, A_2B KO, and A_2A/2B DKO mice, we demonstrated that adenosine and its receptor subtypes (A_2A and A_2B) are not required for pacing-induced coronary metabolic hyperemia, although they exert important roles in maintaining CF under resting conditions. Similarly, broad pharmacological blockade of ARs by 8-SPT in the absence or presence of a NO synthase inhibitor or selective A_2AAR and A_2BAR blockade did not affect pacing-induced increase in CF, although baseline flow was reduced. However, K_ATP channel blockade by glibenclamide remarkably reduced both resting flow and pacing-induced hyperemia, resulting in a dramatic decrease in LVDP. Scavenging of H₂O₂ by catalase also decreased resting flow and pacing-induced coronary hyperemia but to a lesser extent than glibenclamide. Our results suggest that while adenosine modulates coronary vascular tone under both resting and ischemic conditions, it is not required for metabolic hyperemia even when the NO pathway is inhibited. However, H₂O₂ and K_ATP channels are important local control mechanisms responsible for both coronary ischemic and metabolic vasodilation.

Since Berne (6) proposed that adenosine is an important metabolite that links the changes in CF to mvo₂, there have been numerous studies that have examined the adenosine hypothesis in coronary metabolic hyperemia (3, 4, 7, 14, 21, 47–49). Early studies supported this hypothesis in that the augmentation in cardiac interstitial adenosine concentration was observed to be correlated with increased mvo₂ in chronically instrumented dogs (3, 19, 21) and pharmacological blockade of ARs was shown to decrease coronary metabolic hyperemia in isolated guinea pig hearts (21). However, more recent studies in awake dogs have reported that there was no significant increase in cardiac interstitial adenosine levels that was sufficient to cause coronary vasodilation (49, 50, 52) and that AR blockade (49, 50) or triple blockade of adenosine, K_ATP channels, and NO (47, 48) did not affect exercise or norepinephrine-induced coronary hyperemia, suggesting that adenosine is not required in coronary functional hyperemia. Interestingly, Duncker and colleagues (16, 17) suggested that adenosine is an important mediator for coronary metabolic hyperemia only when the K_ATP channel is blocked or under conditions where coronary perfusion pressure is reduced. The discrepancies among these studies may be due to the mixed drug specificity or to the complexity of mixed local and neuronal control mechanisms responsible for coronary blood flow regulation under in vivo conditions. Using targeted AR KO mice combined with an isolated heart preparation to solve the specificity issue as well as to exclude the neurohormonal effect on CF regulation, the present study showed that A_2AARs and A_2BARs are important in modulating coronary vascular tone under resting conditions because baseline CF in A_2A KO, A_2B KO, and A_2A/2B DKO mice was reduced (by ∼16%, ∼18%, and ∼24%, respectively) compared with WT mice (Figs. 3 and 4). Of interest, although double deletion of A_2AARs and A_2BARs further reduced baseline CF from either A_2A KO or A_2B KO mice, the extent of reduction in A_2A/2B DKO mice (by ∼22%) was much less than the summed effect of A_2AAR and A_2BAR deletion (∼34%), suggesting that other unknown compensatory mechanisms might be turned on to maintain resting CF when both ARs are deleted. Consistent with the results observed in AR KO mice, broad AR blockade by 8-SPT and selective A_2AAR or A_2BAR blockade by SCH-58261 (1 μM) or CVT-6883 (1 μM) also resulted in a significant reduction (by ∼25%, ∼35%, and ∼18%, respectively) in resting CF (Figs. 1, 3, and 4). In agreement with our findings, other studies in humans (18), pigs (14), dogs (49), and isolated guinea pig hearts (21) have also reported that adenosine is involved in modulating CF under resting condition. Of note, a lower reduction in CF was observed in A_2A KO mice (by ∼18%) versus SCH-58261-treated animals (by ∼35%; Fig. 3, A and C). The lower reduction in baseline CF might be due to upregulated A_2BARs in A_2A KO mice compared with WT mice, as previously reported (45). Regarding the downstream mechanisms of adenosine in modulating resting CF, our results (45) and those of others (56) suggest that A_2AAR-mediated NO production may be responsible for resting tone regulation because 8-SPT (nonselective adenosine antagonist; Fig. 2) or SCH-58261 (selective A_2AAR antagonist) (56) failed to further decrease baseline CF in the presence of a NO synthase inhibitor. Concomitant with the decreased baseline flow, a significant reduction in LVDP (Table 1 and Figs. 1C, 3C, and 4D) and a relatively but not statistically lower +dP/dt (Table 1) were observed upon blockade of adenosine either pharmacologically or genetically. The correlated decrease in cardiac contraction with the decreased CF suggest that adenosine is an important mediator involved in modulating resting CF and that the decreased cardiac function is associated with the decreased CF.

Despite the decrease in resting CF, pacing-induced hyperemia was not affected after AR blockade or deletion, in as much as the linear curve of the relationship between HR and/or RPP and CF was downward shifted in a parallel manner and the slope of the curve was not altered before or after AR blockade or deletion (Figs. 1, 3, and 4). These results, obtained from targeted AR KO mice, add further evidence to the previous findings that an AR blocker failed to inhibit exercise-induced coronary hyperemia in chronically instrumented dogs, although it lowered the balance between O₂ supply and delivery under resting conditions (47, 49). The approach of using targeted AR KO mice allowed us to rule out the possibility that increased endogenous adenosine overcomes the competitive receptor blockade, thus masking the direct AR blockade effect. Additionally, in the presence of a NO synthase inhibitor, 8-SPT did not further decrease resting flow as well as pacing-induced coronary hyperemia, indicating both NO and adenosine act as tonic vasodilator but are not required for coronary metabolic hyperemia.

Increasing evidence suggests that H₂O₂, an endogenous metabolite, acts as a pivotal mediator responsible for metabolic coronary vasodilation (37, 42, 53, 54). In support of this idea, our results showed that catalase significantly decreased resting CF (by ∼16%) as well as the pacing-induced increase in CF (by ∼37%; Fig. 5, A and B). It should be pointed out that although A_2AAR-mediated H₂O₂ production in the coronary endothelium and smooth muscle cells has been previously reported to play an important role in coronary ischemic hyperemia (42, 57), the failure of AR blockade or deletion to reduce metabolic hyperemia (discussed above) suggest that other mechanisms than adenosine-mediated signaling are responsible for H₂O₂ production, thus contributing to coronary metabolic vasodilation. Regarding the source of H₂O₂, we speculate that mitochondrial electron transport in cardiomyocytes might be an important source for pacing-induce H₂O₂ production (37).

The K_ATP channel has been reported to be involved in coronary ischemic vasodilation (24, 41, 42, 57), although its role in metabolic vasodilation remains controversial (13, 15, 23, 35, 47, 55). In isolated mouse hearts, the present study showed that glibenclamide (5 μM) significantly reduced CF (by ∼45%; Fig. 6A) and LVDP (by ∼36%; Fig. 6C) at rest. In agreement with our results, studies in dogs have reported that intracoronary infusion of glibenclamide (50–80 μg·kg⁻¹·min⁻¹) significantly decreased baseline flow (by ∼20–50%) (38) and was associated with cardiac contractile dysfunction (systolic wall thickening decreased by ∼43%) (15). Importantly, recovery of the CF rate to preglibenclamide levels after the addition of sodium nitroprusside reversed the cardiac dysfunction, suggesting that K_ATP channels exert essential roles in maintaining resting CF and that the decreased CF upon K_ATP channel blockade in the coronary vasculature causes ischemia-induced cardiac contractile dysfunction (15). It is important to note, however, that K_ATP channels at concentrations of >10 μM have been reported to inhibit other K⁺ and Ca²⁺ channels (5, 36). The concentration we used in the present study (5 μM) is within its selectivity range (5, 36), thus strengthening our conclusion that the K_ATP channel is important in modulating resting CF. Interestingly, in disagreement with the majority of studies showing that blockade of K_ATP channels did not attenuate the increase in CF during treadmill exercise or paired cardiac pacing in dogs and pigs (13, 15, 34, 35, 47), the present study on isolated mouse hearts demonstrated that blockade of K_ATP channels not only reduced resting flow but also markedly inhibited pacing-induced functional hyperemia by ∼94% (Fig. 6, A and B). The discrepancy among the studies could be due to 1) species differences (mice vs. pigs and dogs) or 2) the difference in animal models (isolated hearts vs. whole animals). In isolated hearts from mice deficient in K_ir6.1 (one of the K_ATP subunits), isoprenaline-induced coronary vasodilation was significantly attenuated, indicating an important role of K_ATP channels in coronary metabolic hyperemia (55). Additionally, the redundant mechanisms responsible for CF regulation in whole animals, including local and neurohormonal control mechanisms, may complicate the explanation of the data, e.g., the reduction of coronary metabolic hyperemia by the K_ATP channel blocker might be compensated by an enhanced neurohormonal effect or other unknown mechanisms to maintain the balance between O₂ supply and demand, thus resulting in unchanged metabolic hyperemia in vivo even after K_ATP channel blockade. Concomitant with the reduced metabolic hyperemia in isolated mouse hearts, we also observed dramatic cardiac contractile dysfunction. In contrast to significant increases in LVDP upon pacing in control condition, K_ATP channel blockade resulted in a significant reduction in LVDP upon pacing (Fig. 6C), suggesting potential ischemic cell damage related to a decrease in CF.

It is well established the K_ATP channels play important roles in adenosine-mediated coronary reactive hyperemia (1, 8, 22, 41, 42, 57). Our recent studies (42, 58) have demonstrated that A_2AAR-mediated H₂O₂ production opens K_ATP channels, partially contributing to coronary reactive hyperemia. In these studies, although adenosine-induced K_ATP current was reduced in A_2A KO and A_2A/2B DKO animals, pinacidil (K_ATP opener)-induced current was not affected after deletion of both A_2AARs and A_2BARs, suggesting that A_2AARs and A_2BARs are not required for K_ATP activation and that the absence of ARs does not affect channel expression, as we previously reported (33, 42). Additionally, we demonstrated that K_ATP channel blockade inhibited coronary reactive hyperemia (57) to a higher extent than A_2AAR blockade, indicating that other factors, including adenosine, can also activate K_ATP channels, thus contributing to coronary vasodilation. In the present study, K_ATP channel blockade dramatically decreased metabolic hyperemia, whereas only a parallel downward shift of the relationship between CF and RPP was observed upon A_2AAR and/or A_2BAR blockade or ablation, suggesting that other unknown mediators than adenosine released during increased cardiac metabolic demand can activate K_ATP channels, contributing to the increased flow during metabolic hyperemia. One of the mediators appears to be H₂O₂, because scavenging of H₂O₂ by catalase significantly decreased metabolic hyperemia by ∼37% (Fig. 5, A and B) and to a lesser extent than K_ATP channel blockade (by ∼94%; Fig. 6, A and B). Taken together, our results suggest that while adenosine-mediated H₂O₂ production and K_ATP channel opening are important for coronary reactive hyperemia, adenosine appears to be not required for metabolic hyperemia, and other unknown mediators act together with H₂O₂ to open K_ATP channels and are thus responsible for metabolic hyperemia in isolated mouse hearts.

It should be pointed out that there are some limitations regarding the translation of our findings based on our present experimental model into the clinical applications regarding the pathophysiology of CF regulation. First, although the effects of neurohomonal and blood components were excluded to scrutinize the metabolic control mechanism in CF regulation, buffer-perfused isolated hearts did not allow us to clearly separate the contribution of shear- and/or pressure-induced CF changes from local metabolic effects. Second, the continuously oxygenated (95% O₂) Krebs buffer used in isolated hearts has much less O₂ carrying capacity than blood, which might cause hypoxia and relatively lower cardiac contractile function compared with blood-perfused hearts (40). However, in buffer-perfused hearts paced from 400 to 600 beats/min, we observed a correlated increase in mvo₂ from 160 to 230 μl·min⁻¹·g⁻¹ as well as CF (increased by ∼40%). More importantly, CF increased more than twofold over baseline after 15-s flow occlusion (42, 57), indicating that buffer-perfused hearts have a minimum hypoxia and coronary vessels have the capacity to further dilate. However, due to the relatively lower myoglobin O₂ saturation in buffer-perfused hearts (39), the cardiac interstitial adenosine level might be higher compared with blood-perfused hearts, which may overestimate the role of adenosine in modulating resting coronary blood flow in our model. Future studies are needed to compare adenosine levels in buffer- versus blood-perfused hearts. Finally, to extend our study to an atherosclerotic model, where blunted metabolic hyperemia has been reported, it should be noted that other factors, such as phenotypic changes of coronary vessels (more stiff and fibrotic) and hemodynamic alterations due to collateral circulation and/or coronary steal (less flow to the ischemic region where the vascular network is already maximally dilated), need to be considered for studies.

In conclusion, blockade of ARs, in particular, A_2AARs and A_2B ARs using combined pharmacological compounds with targeted gene deletion, decreased CF at rest but failed to reduce the pacing-induced increase in CF. Inhibition of NO synthase decreased resting CF without an effect on pacing-induced hyperemia, which was not altered by combined blockade of ARs, suggesting both NO and adenosine are not required for coronary metabolic hyperemia. The mechanisms responsible for pacing-induced coronary hyperemia involve mostly K_ATP channels and, to a lesser extent, H₂O₂, since blockade of K_ATP channels by glibenclamide almost abolished pacing-induced coronary vasodilation, whereas scavenging of H₂O₂ by catalase only partially inhibited the response. Thus, although adenosine-mediated H₂O₂ production and the subsequent opening of K_ATP channels in the coronary vasculature play an important role in coronary ischemic vasodilation, other mechanisms, but not adenosine, are responsible for H₂O₂ production and opening of K_ATP channels, leading to coronary metabolic vasodilation in isolated mouse hearts.

A compromised coronary vasodilation may not necessarily cause ischemia under resting condition. However, during increased myocardial demand, e.g., exercise, the coronary vascular dysfunction will lead to a mismatch between O₂ demand and supply, manifesting the clinical symptom known as angina. Although mice differ significantly from humans in many aspects, which may limit the extrapolation of present study to clinical situations, our model, using paced mouse hearts to gradually increase rates that are within the range of HR change during exercise in humans, will provide an important basis for further mechanistic studies in humans. The different mechanisms regarding the roles of adenosine, H₂O₂, and K_ATP channels in CF regulation during ischemic versus metabolic hyperemia may add new knowledge to the understanding of pathophysiology of coronary artery diseases.

GRANTS

This work was supported by National Institutes of Health Grants HL-027339, HL-09444, HL-071802, and U54-GM-104942.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: X.Z. and S.J.M. conception and design of research; X.Z. performed experiments; X.Z. analyzed data; X.Z. interpreted results of experiments; X.Z. prepared figures; X.Z. drafted manuscript; X.Z. and S.J.M. edited and revised manuscript; X.Z., B.T., S.L.T., C.L., and S.J.M. approved final version of manuscript.