Midazolam Selectively Potentiates the A^2A- but not A¹-receptor– mediated Effects of Adenosine: Role of Nucleoside Transport Inhibition and Clinical Implications

The benzodiazepines alprazolam, chlorodiazepoxide, diazepam, and flurazepam were purchased from Research Biochemicals (RBI, Natick, MD) and prepared as stock solutions in ethanol. Midazolam was obtained from Roche Laboratories (Nutley, NJ). S -(4-nitrobenzyl)-6-thioinosine (NBMPR, RBI), its tritiated derivative [³H]NBMPR (New England Nuclear, Dupont, Boston, MA), the selective A^2A-adenosine receptor antagonist ZM 241385, (4-(2-[7-amino-2-(2-furyl)[1,2,4]triazolo[2,3-a][1,3,5]triazin-5-yl amino]ethyl)phenol), a gift from Dr. Simon Poucher (Zeneca Pharmaceuticals, London, UK), and its tritiated derivative [³H]ZM 241385 (New England Nuclear) were prepared as stock solutions in dimethylsulfoxide. Stock solutions were further diluted in normal saline. Neither dimethylsulfoxide nor ethanol had a significant nonspecific effect in binding experiments. The concentration of dimethylsulfoxide did not exceed 0.01% (vol/vol) in the perfusion medium for the isolated heart experiments and did not affect coronary flow or the stimulus-to-ventricle (S-V) interval, a measure of atrioventricular conduction time (AVCT). 13,14 

Membranes for binding studies were prepared from porcine brain striatum and guinea pig ventricle in ice-cold Tris-HCl buffer (50 mM, p  H 7.4). The protein content of membrane suspensions was determined using the Bradford method (Bio-Rad Laboratories, Hercules, CA) with bovine serum albumin as the standard. The membranes were stored at −80°C until used for binding assays.

Equilibrium saturation binding studies were performed to determine the dissociation constant (K^d) of NBMPR. Membrane suspensions were incubated for 120 min at 25°C with varying concentrations of [³H]NBMPR (0.17–10 nM). Nonspecific binding was defined as binding not displaced by an excess (10 μM) of unlabeled NBMPR and was subtracted from total binding to calculate specific binding. K^dwas 1.9 and 1.0 nM for guinea pig ventricular and porcine striatal membranes, respectively. Competition assays to determine the affinities of alprazolam, chlorodiazepoxide, diazepam, flurazepam, and midazolam for the [³H]NBMPR binding site were performed by incubating membrane suspensions for 90 min at 25°C with 4 nM [³H]NBMPR and progressively higher concentrations of the respective benzodiazepines in a final volume of 300 μl. Similarly, competition binding assays were performed with porcine striatal membranes to determine the affinity of midazolam for A^2Areceptors labeled with 4 nM [³H]ZM 241385. Protein content, final volume, incubation time, and temperature were 10–50 μg, 300 μl, 120 min, and 25°C, respectively. Nonspecific binding was defined as binding not displaced by 30 μM 5′-N -ethylcarboxamidoadenosine (NECA). In all assays, separation of free from bound radioligand was conducted by vacuum filtration with glass fiber filters (type 32 glass, Schleicher & Schuell, Keene, NH) using a Brandel cell harvester (Biomedical Research & Development Laboratories, Inc., Gaithersburg, MD). The radioactivity trapped on the filter was counted using a liquid scintillation counter. All determinations were conducted in triplicate. Binding was linear over the range of protein concentrations used (protein content 10–50 μg for pig striatum and 100–300 μg for guinea pig ventricle). Assay times and temperatures were chosen to assure equilibrium conditions.

All protocols involving animals were reviewed and approved by the Animal Use Committee of the University of Florida Health Science Center. Hartley guinea pig hearts were isolated and perfused according to the Langendorff method as previously described. 15The perfusate was supplied through a nonrecirculating system fed by a peristaltic pump at constant flow, set to exceed maximal coronary flow, with the outflow variably split between coronary flow and overflow. With this arrangement, administration of drugs at the inflow allowed delivery of a fixed drug concentration regardless of changes in coronary flow. The oxygen tension, temperature, and p  H of the perfusate were maintained at 500–600 mmHg, 36 ± 0.5°C, and 7.3–7.4, respectively, unless stated otherwise. Because of the limited oxygen-carrying capacity of salt solutions, a partial pressure of oxygen > 500 mmHg in the perfusate represents normoxia for a nonworking heart preparation. 16The left atrium was excised to facilitate placement of a small plastic catheter into the left ventricle to prevent increases in wall tension through fluid accumulation. Hearts were paced at a fixed cycle length of 270 ms (222 beats/min). A unipolar electrode was placed on the interatrial septum to continuously acquire an electrogram. The S-V interval was used for the analysis of AVCT because neither the stimulus-to-atrium nor the His bundle-to-ventricle intervals are affected by adenosine or adenosine-regulating agents. 14Coronary flow was measured as total perfusate flow using an ultrasonic transit time flow probe (Model 2N, Transonic Systems, Ithaca, NY) placed in the perfusate line. Data were continuously recorded to the hard disk of an IBM-compatible PC (PII-300 MHz, Gateway 2000, North Sioux City, SD) using an analog-to-digital data board (TL-1-125, Axon Instruments, Foster City, CA).

After equilibration and acquisition of baseline coronary flow and AVCT measurements, hearts were exposed to 5 and 10 μM midazolam for 25 min at each concentration. Subsequently, 1 μM ZM 241385 was administered in the continued presence of 10 μM midazolam for 10 min. The coronary flow and AVCT were recorded at the end of each intervention. To control for possible time-related deterioration of the heart preparation or agonist-induced desensitization, return to baseline values was confirmed after a 30-min washout period, and the coronary flow response after a 2.5-μg bolus dose of adenosine was measured. Bolus doses of adenosine were administered as 100-μl aliquots over 2 s by syringe pump into the perfusate line proximal to the flow probe. As the maximal A^2A-adenosine receptor–mediated response was a twofold to threefold increase in coronary flow (fig. 1) in hearts not treated with midazolam, data were retained for hearts in which at least a doubling of coronary flow was observed. 16,17 

In a first set of experiments, the changes in coronary flow and AVCT caused by a bolus dose of 0.5 μg adenosine were measured in the absence and presence of 5 μM midazolam followed by a third bolus dose after a 30-min washout period. Coronary flow and the AVCT were recorded before each bolus dose and at peak coronary flow. To assure the validity of the experimental design, the reproducibility of the response to three consecutive bolus doses of adenosine was assessed in hearts not exposed to midazolam.

In a second set of experiments, the effects of midazolam were assessed in hearts exposed to a continuous infusion of 30 nM adenosine, a concentration just sufficient to increase coronary flow (fig. 1). In the continued presence of adenosine, midazolam was administered at progressively higher concentrations of 5, 10, and 20 μM for 15 min at each concentration followed by the administration of 1 μM ZM 241385 for 20 min concurrently with 20 μM midazolam and adenosine. The coronary flow and AVCT were recorded before each intervention. Postintervention measurements were performed after a 30-min washout, followed by testing the coronary flow response to a bolus dose of 2.5 μg adenosine. A doubling of coronary flow was considered indicative of a viable preparation.

Hearts were exposed to three consecutive 3-min episodes of hypoxic perfusion by changing the perfusate to Krebs-Henseleit solution gassed with 20% O², 5% CO², and 75% N²(temperature 36°C, p  H 7.3–7.4, partial pressure of oxygen 200 ± 20 mmHg) to stimulate the release of endogenous adenosine. Each 3-min period of hypoxic perfusion was followed by a 20-min normoxic recovery. After control measurements, 5 μM midazolam was administered for 15 min before a second episode of hypoxia. ZM 241385 (1 μM) was then infused for 5 min in the continued presence of midazolam before the last episode of hypoxia. Coronary flow and AVCT were recorded before and after each intervention and at peak coronary flow. To validate the experimental design, the reproducibility of the response to three consecutive exposures to hypoxic perfusion was assessed in hearts not exposed to midazolam or ZM 241385.

Binding parameters describing the results of competitive displacement experiments (affinity constant [Kⁱ] and Hill coefficient) were determined using the radioligand binding analysis program GraphPad v 2.0 (GraphPad Software Inc., San Diego, CA). Kⁱvalues were calculated using the Cheng-Prusoff transformation. 18Because values of Kⁱare not normally distributed, mean values were calculated from the logarithmically transformed data (p  Kⁱ) for each experiment. 19The S-V interval and the coronary flow were measured at the times indicated. Changes in coronary flow were determined by integrating the recorded flow–time curves to generate the area under the curve using TableCurve 2D (Jandel Scientific, San Rafael, CA).

All data are presented as mean ± SD. Before parametric testing, the assumption of normality was validated using the Kolmogorov-Smirnov test with Lilliefors correction (SSPS v9.0, SPSS, Inc., Chicago, IL). Differences between mean values were analyzed by one- or two-way repeated measures analysis of variance of raw data, followed by Student–Newman-Keuls testing. A P  value < 0.05 was considered to indicate statistically significant differences.

Specific binding of [³H]NBMPR to membranes prepared from guinea pig ventricles and porcine striatum was displaced in the presence benzodiazepines (alprazolam, chlorodiazepoxide, diazepam, flurazepam, and midazolam;fig. 2). Although the displacement of [³H]NBMPR was concentration-dependent and complete (fig. 2), the ligand potencies were significantly different (table 1). In both tissues, midazolam was the most potent inhibitor of the nucleoside transporter. The rank orders of potency for the ligands to displace the specific binding of [³H]NBMPR from guinea pig ventricle and from porcine striatum were midazolam > alprazolam = flurazepam = diazepam = chlorodiazepoxide and midazolam > alprazolam = chlorodiazepoxide > flurazepam = diazepam, respectively. When mean values of p  Kⁱfor all drugs were analyzed across tissues, it was found that drug affinities for nucleoside transporter were dependent on the tissue source of the nucleoside transporter to a significant degree (P = 0.034). Both diazepam and flurazepam had significantly higher affinities for nucleoside transporter in guinea pig ventricle than for nucleoside transporter in porcine striatum (table 1and fig. 2).

Exposure of normoxic hearts (n = 4) to midazolam caused a significant, concentration-dependent, reversible increase in coronary flow (table 2). ZM 241385, a selective A^2A-adenosine receptor antagonist, eliminated approximately 50% of this effect (P = 0.003). The S-V interval tended to shorten in parallel to the increase in coronary flow. In contrast to the shortening of the S-V interval during the administration of an adenosine bolus dose (see below), this change did not achieve statistical significance (P = 0.08). These effects were fully reversible as washout values for coronary flow, and the S-V interval did not differ from control measurements. Likewise, all hearts demonstrated at least a doubling of coronary flow after a 2.5-μg bolus dose of adenosine (data not shown).

A bolus dose of 0.5 μg adenosine caused a transient, reproducible increase in coronary flow from 5.1 ± 0.6 to 7.9 ± 1.3 ml/min (P = 0.034;figs. 3A and 3B) and tended to shorten the S-V interval by approximately 2 ms (P = 0.07;figs. 3C and 3D) in a control group of three hearts. In the presence of 5 μM midazolam, the increase in coronary flow after an adenosine bolus—expressed as the area under the flow curve—was significantly greater than at baseline or after washout of midazolam (P = 0.002; n = 6;figs. 4A and 4B). The presence of 5 μM midazolam caused similar increases in baseline coronary flow (from 6.5 ± 1.2 to 7.4 ± 1.0 ml/min;P < 0.05) and in peak coronary flow (from 11.1 ± 1.8 to 12.7 ± 1.7 ml/min;P < 0.01). The S-V interval shortened in parallel with the coronary flow increase for all three adenosine administrations. At the peak increase of coronary flow, AVCT decreased from 52.3 ± 2.6 ms to 50.7 ± 2.4 ms (P = 0.006). However, this change in AVCT was not affected by midazolam (figs. 4C and 4D).

Figure 5demonstrates the interaction between a constant concentration of adenosine (30 nM) and successively higher concentrations of midazolam (5, 10, and 20 μM). In the absence of midazolam, adenosine caused minimal to no change in coronary flow as shown by the concentration–response relationship depicted in figure 1. In the presence of 30 nM adenosine, coronary flow increased 66 ± 14%, 123 ± 31%, and 165 ± 44% above control measurements during administration of 5, 10, and 20 μM midazolam, respectively (P < 0.001; n = 4). Approximately 70% of the maximal increase in coronary flow caused by 20 μM midazolam in the presence of adenosine was antagonized by ZM 241385. Adenosine and midazolam did not significantly affect the S-V interval (values ranged from 49.0 ± 2.4 to 52.5 ± 2.6 ms), whereas the addition of ZM 241385 significantly prolonged AVCT to 56.5 ± 5.5 ms (P = 0.007; n = 4). The changes in coronary flow and S-V interval were reversible on washout of all drugs (data not shown).

Exposure of hearts to three consecutive episodes of endogenous adenosine release caused by hypoxic perfusion reproducibly increased coronary flow from a baseline of 5.7 ± 1.5 to a peak flow of 11.2 ± 2.2 ml/min (P = 0.01; n = 3). The hypoxia-induced increases in coronary flow, expressed as the areas under the flow–time curve, for the three episodes were 9.8 ± 1.3, 10.1 ± 2.3, and 11.5 ± 0.5 (P = 0.45; n = 3). During hypoxia, coronary flow increased linearly with a stable rate of increase of 2.8 ± 0.5 ml/min²(P < 0.67 for the three consecutive episodes). The S-V interval did not shorten consistently during the period of hypoxic perfusion.

Administration of 5 μM midazolam during hypoxic perfusion increased the coronary flow response by > 50% (P = 0.005; n = 6;figs. 6A and 6B). Similar to previous experiments, coronary flow increased reversibly from a baseline of 5.6 ± 0.6 ml/min to a new baseline of 7.4 ± 0.9 ml/min in the presence of midazolam. This new baseline was used to determine the total flow increase in response to hypoxic perfusion in the continued presence of midazolam (figs. 6A and 6B). ZM 241385 attenuated the midazolam-induced increase in coronary flow. The presence of ZM 241385 also attenuated the rate of increase in coronary flow to 2.0 ± 0.8 ml/min², compared with 2.9 ± 1.0 ml/min²during control and 2.8 ± 0.9 ml/min²in the presence of 5 μM midazolam (P = 0.04). However, the total increase in coronary flow caused by global hypoxia was similar during control (hypoxia 1) and in the presence of both midazolam and ZM 241385 (hypoxia 3). The S-V interval did not change significantly or consistently during the consecutive hypoxic episodes (figs. 6C and 6D).

To determine whether the selective augmentation of adenosine-induced coronary vasodilation is caused by direct activation by midazolam of A^2A-adenosine receptors, in competition binding assays we assessed the affinity of midazolam to porcine striatal A^2A-adenosine receptors labeled with [³H]ZM 241385. At concentrations used in the isolated heart experiments, midazolam did not displace specific binding of [³H]ZM 241385 (fig. 7). In contrast, the adenosine receptor agonist NECA caused concentration-dependent, complete displacement with a Hill coefficient not different from unity and a value of Kⁱ(p  Kⁱ± SD) of 78 nM (7.13 ± 0.16), indicating the presence of A^2A-adenosine receptors in the membrane preparation.

Our results demonstrate that (1) among five commonly used benzodiazepines, midazolam is the most potent inhibitor of nucleoside transporter in heart (guinea pig ventricle) and brain (porcine striatum), and (2) midazolam can be used as an “adenosine-regulating agent” to selectively enhance the cardiac actions of adenosine in an event- and site-specific manner in the guinea pig isolated heart.

A novel finding of the current study is that among five commonly used benzodiazepines, midazolam is the most potent inhibitor of nucleoside transporter in two distinct tissues from different species. In agreement with our results, other investigators also found significant differences among benzodiazepines in their ability to displace [³H]NBMPR binding to the nucleoside transporter in various tissues. 7,9,20Midazolam, the benzodiazepine most widely used in the perioperative setting, was only studied in human erythrocytes. 9No prior study assessed midazolam binding to transporter in cardiac and neural tissue. All benzodiazepines tested completely displaced [³H]NBMPR in a concentration-dependent and competitive manner from heart and brain (table 1and fig. 2). The rank order of potency of benzodiazepines to compete with [³H]NBMPR for binding to guinea pig ventricle differed slightly from the rank order of potency of benzodiazepines to compete for binding to porcine striatum. Although this difference in rank order of potency of benzodiazepines to compete with [³H]NBMPR for binding sites in these two tissues may reflect the existence of distinct subtypes of nucleoside transporter, these slight differences in the rank order of ligand potency are not unusual and have been described for other binding studies. 9,20,21In addition, species-dependent factors may contribute to this difference.

The results of competition binding studies indicate that benzodiazepines are weak inhibitors of transporter at concentrations that facilitate GABAergic transmission in vivo . For example, the Kⁱvalue of midazolam to displace [³H]NBMPR from guinea pig ventricular membranes was approximately 5 μM (table 1). According to the law of mass action, in the absence of transporter substrate (i.e. , adenosine), 5 μM midazolam will only inhibit 50% of nucleoside transporter. The affinity of benzodiazepines for the nucleoside transporter is thus more than 1,000-fold lower than that of prototypical high-affinity nucleoside transporter ligands such as NBMPR (K^d= 1.6 nM in guinea pig ventricle 22,23) or dipyridamole (Kⁱ= 1.9 nM in human erythrocytes 9). To achieve relatively selective inhibition by midazolam of nucleoside transporter and thus minimize nonspecific effects provided the rationale of using a concentration of midazolam at its Kⁱvalue in most of the experimental protocols. Although this concentration of midazolam is approximately an order of magnitude lower than those used in previous isolated heart studies, 7,8,12,17it is nonetheless at the upper end of the clinically relevant dose spectrum (6 μM [2 μg/ml] during bolus administration and 1.7 μM for deep hypnosis 24,25). Furthermore, the high degree of protein binding of benzodiazepines (> 90%, at least under physiologic conditions) further reduces the free concentration of drug in vivo . Therefore, it is unlikely that under physiologic conditions a significant midazolam-induced potentiation of the cardiac effects of adenosine would occur. Regardless, the major goal of this study was to demonstrate the feasibility and rationale of using the benzodiazepine with the greatest binding affinity to nucleoside transporter, namely, midazolam, as an “adenosine-regulating agent” to selectively enhance the cardiac actions of adenosine in an event- and site-specific manner.

Midazolam selectively augmented coronary vasodilation (A^2A-receptor–mediated) but not atrioventricular conduction delay (A¹-receptor–mediated) caused by adenosine, regardless of whether the nucleoside was exogenously administered or endogenously released during global hypoxia. Midazolam alone caused a concentration-dependent increase in coronary flow without prolonging atrioventricular conduction (table 2). This finding is in agreement with a previous study in constant pressure perfused guinea pig isolated hearts that also showed a concentration-dependent increase in coronary flow caused by midazolam, despite a concomitant decrease in myocardial oxygen consumption. 17Based on the actions of ZM 241385, a potent adenosine receptor antagonist that is > 100-fold more selective for A^2Areceptors than for other adenosine receptor subtypes (i.e. , A¹, A^2B, and A³receptors), 26only approximately half of this coronary vasodilator response can be ascribed to increased adenosine levels. Although the etiology of the ZM 241385–insensitive component cannot be determined with the design of the current study, it may be secondary to enhanced GABA receptor activity. 27In any case, the focus of our study was neither to compare the potency of benzodiazepines as coronary vasodilators nor to define further the underlying mechanism(s) mediating the effects of midazolam on coronary flow, but rather to demonstrate that benzodiazepines can potentiate the cardiac effects of adenosine.

In agreement with our finding that midazolam consistently augmented the effects of exogenous adenosine on coronary flow, pretreatment of rats with diazepam has been shown to enhance the A^2A-adenosine receptor–mediated relaxation of isolated pulmonary artery rings. 28Likewise, diazepam enhanced coronary flow increases caused by exogenous adenosine in anesthetized dogs. 29Our observation of an augmentation of the cumulative coronary flow during hypoxic perfusion extends these findings to pathophysiologic situations wherein adenosine release is increased. The time course of the coronary flow increase in the presence of midazolam suggests that the increase in flow is predominantly caused by a slowed return to baseline coronary flow (figs. 4A and 6A), lending further support to the hypothesis that midazolam delays the elimination of extracellular adenosine.

Although benzodiazepines were previously shown to sensitize guinea pig atria to the negative inotropic effect of exogenously administered adenosine (A¹-receptor–mediated), 11,12we did not observe a prolongation of AVCT in the presence of midazolam and adenosine. On the contrary, exogenous adenosine, administered as a bolus dose, caused a transient but significant shortening of AVCT. This counterintuitive finding may be explained by the increase in shear stress caused by increased flow, which was shown to alter Ca²⁺transients and cause an inverse relationship between coronary flow and AVCT. 30 

Three factors may explain the differential augmentation of A^2A-versus  A¹-adenosine receptor–mediated effects by midazolam. First, this finding may be a result of differential receptor reserve of A^2A- and A¹-adenosine receptor–mediated responses. Receptor reserve is a phenomenon whereby submaximal receptor occupancy elicits a maximal response. 13In tissues with a large receptor reserve, a small increase in receptor occupancy produces a large response. Significant differences for receptor reserve among distinct receptor subtypes mediating the effects of an endogenously released substance such as adenosine are an important determinant of organ and tissue responsiveness and form the basis of agonist selectivity in vivo . 13,15,31,32In guinea pig isolated hearts, the receptor reserve for A^2A-receptor–mediated coronary vasodilation (70%13) is much greater than that for A¹-receptor–mediated slowing of atrioventricular nodal conduction (2%32). Therefore, the concentration of adenosine required to cause half-maximal coronary vasodilation (EC⁵⁰= 108 nM;fig. 1) is > 40-fold lower than that required to produce half maximal slowing of atrioventricular conduction (EC⁵⁰= 4,200 nM 31). This phenomenon, at least in part, explains how concentrations of adenosine that are subthreshold for slowing atrioventricular conduction can markedly increase coronary flow. 13Midazolam may cause small increases in the interstitial concentration of adenosine, sufficient to elicit a significant A^2A-receptor–mediated response but subthreshold to concurrently affect A¹-receptor–mediated effects. Consistent with this explanation, adenosine levels in normoxic isolated hearts are just below the inflection point of the steep portion of the concentration–response relationship. 1,33However, in the setting of ischemia, where coronary flow is limited or absent, this hierarchy in adenosine-mediated responses may be overcome, and adenosine accumulation may be sufficient to convey a protective effect via  activation of both A¹and A^2Areceptors. 34,35In keeping with this interpretation, we found a small but significant increase in AVCT (A¹-receptor–mediated) in the presence of 20 μM midazolam and 30 nM adenosine only, when the concomitant increase in coronary flow was attenuated by the A^2A-adenosine receptor antagonist ZM 241385.

A second factor that may also contribute to an apparent differential responsiveness of A¹and A^2Areceptors to endogenous and exogenous adenosine is the phenomenon of “compartmentalization.” The concentration of adenosine in the interstitium differs from that in the coronary perfusate (arterial compartment). 36An analysis using a mathematical model of adenosine metabolism in guinea pig heart indicates that the concentration of adenosine in the interstitium is approximately threefold lower than that in the arteries during the exogenous administration of adenosine and is not a linear function of arterial adenosine concentration. 36How transient mild hypoxia, which inhibits adenosine kinase 37and thus increases adenosine release from cardiomyocytes, affects the relative adenosine concentration in these compartments is not known. However, the hypoxia-induced increase in coronary flow may augment oxygen delivery to the atrioventricular node and cause sufficient adenosine washout to keep the increase in interstitial adenosine levels subthreshold for activation of A¹-adenosine receptors. Alternatively, the atrioventricular node may have a higher density of nucleoside transporter sites, which more efficiently remove adenosine, and thus render the atrioventricular node less susceptible to the effects of the nucleoside.

Third, our finding that midazolam selectively amplified A^2A- but not A¹-adenosine receptor–mediated responses may simply be a result of direct activation of A^2Areceptors by midazolam. However, this conclusion is not supported by the results of our competition binding experiments that indicate no significant affinity of midazolam to A^2Areceptors.

The therapeutic benefits of adenosine in the heart are undisputed. The potential cardioprotective effects of exogenous and endogenous adenosine provide the rationale for adding adenosine to crystalloid cardioplegic solutions and explain the improvement in cardiac morbidity and mortality observed with acadesine administration during coronary artery bypass grafting. 6Exogenously administered adenosine is the drug of choice to treat supraventricular tachyarrhythmias involving the atrioventricular node as part of the reentrant circuit. In a similar manner, the clinical utility of using endogenously released adenosine as an effective antiarrhythmic agent has been demonstrated in humans. 38Conti et al.  used the prototypical blocker of nucleoside transporter dipyridamole as an adenosine-regulating agent to terminate supraventricular tachyarrhythmias by a mechanism involving interstitial accumulation of endogenous adenosine acting at the A¹receptor. 39In the current study, we extend the results of this prior work to the A^2Areceptor. Midazolam, by a mechanism involving inhibition of nucleoside transporter and subsequent accumulation of extracellular adenosine, selectively augmented the effects of the nucleoside on coronary flow without affecting AVCT (site-specific action) during global hypoxia (event-specific action). Therefore, benzodiazepines have the therapeutic potential to be used as “adenosine regulating agents.” Our results may provide the framework whereby a new series of benzodiazepines could be developed that have even greater nucleoside transporter blocking properties while still retaining their allosteric enhancing effects on GABA receptor–mediated anxiolysis and sedation. By promoting the known cardioprotective effects of adenosine, benzodiazepines of this type may potentially mitigate hypoxia- or ischemia-induced organ injury, particularly in the perioperative setting.

The authors thank Dr. Stephen P. Baker for his support and Jackie Ruble and Debbie Otero for expert technical assistance with the binding studies.