Comparison of Three Different A₁ Adenosine Receptor Antagonists on Infarct Size and Multiple Cycle Ischemic Preconditioning in Anesthetized Dogs

A₁ adenosine receptor (AR) antagonists are currently being developed for use in humans to treat fluid retention disorders including congestive heart failure (Gottlieb, 2001). A₁AR antagonists are effective diuretic agents due to their ability to inhibit the actions of adenosine to constrict afferent arterioles and increase sodium reabsorption at the proximal and distal tubules (Holz and Steinhausen, 1987; Takeda et al., 1993; Balakrishnan et al., 1996; Wilcox et al., 1999). An additional mechanism for the renal actions of A₁AR antagonists is the interruption of tubuloglomerular feedback linked to sodium delivery to macula densa cells (Wilcox et al., 1999; Brown et al., 2001; Sun et al., 2001; Schnermann, 2002). Unlike thiazide and loop diuretics, A₁AR antagonists have the unique ability to promote natriuresis without reducing glomerular filtration rate (GFR) (Gellai et al., 1998; Wilcox et al., 1999).

BG 9719 (1,3-dipropyl-8-[2-(5,6-epoxy-s-norbornyl)]xanthine; previously named CVT-124) is one of the most potent and selective A₁AR antagonists developed to date (Belardinelli et al., 1995; Pfister et al., 1997). It is a xanthine derivative containing a norbornyl ring at the C-8 position (Fig. 1) that effectively increases A₁AR affinity while decreasing potency at A_2AARs (Belardinelli et al., 1995; Pfister et al., 1997). The affinities of BG 9719 for rat and human A₁ARs are 0.67 and 0.45 nM (Pfister et al., 1997), respectively, with selectivity versus A_2AARs of 1800-fold (rat) and 2400-fold (human). In human patients with congestive heart failure, acute administration of BG 9719 increased urine output without decreasing GFR (Gottlieb et al., 2000). In a follow-up study in congestive heart failure patients in which furosemide induced diuresis at the expense of decreased GFR, combining BG 9719 with furosemide increased urine volume additionally while preventing the deterioration in GFR (Gottlieb et al., 2002). These studies demonstrate the efficacy of BG 9719 as a renal modulating agent in vivo and suggest that the combined use of A₁AR antagonists with conventional diuretics may be an effective approach for the treatment of congestive heart failure (Gottlieb et al., 2000, 2002; Gottlieb, 2001).

Although the clinical usefulness of A₁AR-selective antagonists is promising, one concern with the use of these agents in patients with cardiovascular disease is their potential to counteract the beneficial actions of adenosine in nonrenal tissues. This is especially a concern in the heart. Adenosine is produced in response to ischemic stress, which is believed to serve a protective role to limit tissue injury by multiple mechanisms (Ely and Berne, 1992; Vinten-Johansen et al., 1999). Adenosine interacting with A₁ARs also appears to be one of several endogenous mediators of ischemic preconditioning (IPC), the phenomenon in which brief periods of ischemia activate defense mechanisms within the myocardium that increase resistance to subsequent ischemic episodes. Blockade of these actions of adenosine in the heart may be undesirable in patients with ischemic heart disease. Interestingly, however, many studies in experimental animal models have found that AR antagonists have no effect on the extent of tissue injury induced by acute ischemia and reperfusion (Auchampach and Gross, 1993; Thornton et al., 1993; Zhao et al., 1994; Haessler et al., 1996; Todd et al., 1996; Auchampach et al., 1997b; Kitakaze et al., 1997; Domenech et al., 1998). On the contrary, it has recently been proposed that selective blockade of A₁ARs during reperfusion may actually be an effective means to reduce myocardial infarct size (Neely et al., 1996; Forman et al., 2000).

The goal of the present investigation was to examine the effect of BG 9719 on infarct size in an in vivo dog model of infarction. In addition, we examined the effects of BG 9719 on the development of IPC in a clinically relevant model of multiple ischemia/reperfusion cycles (four 5-min occlusion/5-min reperfusion cycles). We compared the effects of BG 9719 with that of the traditional A₁AR antagonist CPX (1,3-dipropyl-8-cyclopentylxanthine) and a newer xanthine antagonist, BG 9928 (1,3-dipropyl-8-[1-(4-propionate)-bicyclo-[2,2,2]octyl)]xanthine), being developed as a renal modulating agent (Ticho et al., 2003). BG 9928, a C-8 substituted bicyclo-[2,2,2]octylxanthine, binds with high affinity to the A₁AR and possesses improved physiochemical properties (solubility and stability) compared with BG 9719 (Fig. 1) (Ticho et al., 2003). Since the affinities of the different antagonists used in the present study for dog ARs have not been assessed previously, and since the relative selectivity of the antagonists for A₁ARs versus A_2B and A₃ARs is unknown, radioligand binding studies were performed with recombinant dog ARs expressed in HEK 293 cells.

Materials and Methods

Materials

All reagents were obtained from Sigma-Aldrich (St. Louis, MO) except for the following: HEK 293 cells from American Type Culture Collection (Manassas, VA); [³H]CPX and radioactive microspheres from PerkinElmer Life and Analytical Sciences (Boston, MA); adenosine deaminase from Roche Diagnostics (Indianapolis, IN); LipofectAMINE, G418, fetal bovine serum, cell culture media, and pcDNA3.1 from Invitrogen (Carlsbad, CA); BG 9719 and BG 9928, provided by Biogen, Inc. (Cambridge, MA); and Whatman GF/C glass fiber filters from Brandel Inc. (Gaithersburg, MD). [¹²⁵I]ZM 241385 (4-(2-[7-amino-2-(2-furyl)[1,2,4]-triazolo[2,3-a]-[1,3,5]triazin-5-yl amino]ethyl)-3-[¹²⁵I]iodophenol) and [¹²⁵I]AB-MECA (N⁶-(4-amino-3-[¹²⁵I]iodobenzyl)adenosine-5′-N-methylcarboxamide) were synthesized by the chloramine-T method and purified by reverse-phase high-performance liquid chromatography. [³H]MRS 1754 (1,3-dipropyl-8-[4-[((4-cyano-[2,6-³H]-phenyl)carbamoylmethyl)oxy]-phenyl]xanthine) was custom synthesized according to the procedure of Ji et al. (2001).

Radioligand Binding Assay

The affinities of the AR antagonists for recombinant canine ARs expressed in HEK 293 cells were determined by radioligand binding assay using the antagonist radioligands [³H]CPX, [¹²⁵I]ZM 241385, and [³H]MRS 1754 for the A₁,A_2A, and A_2BARs, respectively, and the agonist radioligand [¹²⁵I]AB-MECA for the A₃AR. The full coding region of the receptor cDNAs was subcloned into the mammalian expression vector pcDNA3.1, transfected into HEK 293 cells using LipofectAMINE, and then selected with 2 mg/ml G418. After antibiotic selection, the cells were maintained in Dulbecco's modified Eagle's medium cell culture media containing 10% fetal bovine serum with 0.6 mg/ml G418. Dr. Guy Vassart (Universite Libre de Bruxelles, Brussels, Belgium) provided the full-length canine A₁ and A_2AAR cDNAs (Libert et al., 1989), and we previously cloned the full-length canine A₃AR cDNA from a mast cell cDNA library (Auchampach et al., 1997a). The canine A_2BAR cDNA was obtained by reverse transcription-polymerase chain reaction from RNA isolated from large intestine (GenBank accession number AY313204) using primers based on the sequences of the A_2BAR cloned from human, rabbit, rat, and mouse. Crude membranes were prepared from transfected cells, as described previously (Auchampach et al., 1997a), and stored in aliquots at –20°C.

Binding assays were performed in triplicate with 50 μg of membrane protein in a total volume of 0.1 ml of HE buffer (10 mM Na-HEPES, pH 7.4, 1 mM EDTA, and 0.1 mM benzamidine) with 1 unit/ml adenosine deaminase and 5 mM MgCl₂. Membranes were incubated with radioligands at room temperature for 3 h, and the reactions were terminated by rapid filtration over Whatman GF/C glass fiber filters using a 48-well Brandel cell harvester, followed by four 3-ml washes with ice-cold 10 mM Tris-HCl (pH 7.4) containing 10 mM MgCl₂. Nonspecific binding was determined in the presence of 100 μM adenosine-5′-N-ethylcarboxamide (NECA). Saturation binding assays were conducted first to calculate K_d and B_max values by incubating the membranes with six to eight concentrations of radioligand (the specific activity of [¹²⁵I]AB-MECA was reduced 10- to 20-fold with cold ligand to achieve saturation of the A₃AR). For competition experiments, 5 to 10 nM [³H]CPX, 0.5 to 1.0 nM [¹²⁵I]ZM 241385, 5 to 10 nM [³H]MRS 1754, or 0.25 to 0.50 nM [¹²⁵I]AB-MECA were incubated with cell membranes in the presence of inhibitors.

In saturation assays, specific binding data with [³H]CPX, [¹²⁵I]ZM 241385, and [³H]MRS 1754 fit optimally to a single-binding site model using Marquardt's nonlinear least-squares interpolation (Marquardt, 1963), whereas [¹²⁵I]AB-MECA fit optimally to a two-site binding model. The two binding sites reflect binding of [¹²⁵I]AB-MECA to the high-affinity, G protein-coupled form of the A₃AR and the low-affinity, uncoupled form of the receptor, since the addition of guanosine 5′-O-(3-thio)triphosphate converts all of the binding to the low-affinity state (Auchampach et al., 1997a). The K_d and B_max values of the radioligands for their respective receptors were as follows: [³H]CPX/A₁AR, K_d = 18.1 ± 4.4 nM, B_max = 23,870 ± 1480 fmol/mg membrane protein; [¹²⁵I]ZM 241385/A_2AAR, K_d = 0.76 ± 0.02 nM, B_max = 1501 ± 269 fmol/mg membrane protein; [³H]MRS 1754/A_2BAR, K_d = 12.8 ± 1.7 nM, B_max = 8573 ± 784 fmol/mg membrane protein; and [¹²⁵I]AB-MECA/A₃AR, K_d1 = 0.84 ± 0.04 nM, K_d2 = 21.1 ± 1.2 nM, B_max1 = 781 ± 42 fmol/mg membrane protein, B_max2 = 2146 ± 469 fmol/mg membrane protein. For analysis of competition data, IC₅₀ values of antagonist ligands were fit to: where i is the number of binding sites, SB is specific binding, and NS is nonspecific binding. K_i values were calculated from IC₅₀, B_max, the concentration of radioligand, and the radioligand K_d value with correction for radioligand and competing compound depletion, as described previously by Linden (1982). For A₃AR binding using [¹²⁵I]AB-MECA, we used nonlinear least-squares fitting to obtain K_i values of antagonists in competition for two binding sites by solving the following four equations simultaneously: where LB is radioligand bound, CB is the competitor bound, L is free radioligand, C is free competitor, LT is the total ligand, CT is total competitor, and f is the fraction of L or C nonspecifically bound (Auchampach et al., 1997a). K_d1 and K_d2 values and B_max1 and B_max2 values are known from independent binding isotherms performed in the absence of competitor. Since antagonists are assumed to bind with similar affinity to G protein-coupled and -uncoupled receptors, a single K_i value for each competitor was obtained by setting K_i1 and K_i2 values within the equations to be equal.

Anesthetized Dog Model

Surgical Preparation. The open-chest, barbital-anesthetized dog model of infarction used in this investigation has been described previously in detail (Auchampach and Gross, 1993). Adult mongrel dogs (average weight 23.3; range = 19.0–28.0 kg) of either sex were anesthetized [mixture of sodium pentobarbital (15 mg/kg i.v.) and barbital sodium (200 mg/kg i.v.)], ventilated (tidal volume, 15 ml/kg; 10–15 breaths/min), and instrumented to measure left ventricular pressure and aortic pressure by inserting a double pressure transducer-tipped catheter into the aorta and left ventricle via the left carotid artery. Left ventricular dP/dt was recorded by electronic differentiation of the left ventricular pulse pressure, and heart rate was determined by a tachometer. A left thoracotomy was performed at the fifth intercostal space, and a 1.0- to 1.5-cm section of the left anterior descending (LAD) coronary artery was dissected free from surrounding tissue just distal to the first diagonal branch. A calibrated electromagnetic flow probe was placed around the vessel to measure coronary flow continuously with a flowmeter. The heart was paced at 150 beats/min (2.5 Hz) with rectangular pulses of 4-ms duration and a voltage of twice the threshold via bipolar leads clipped to the atrium. All measurements were recorded throughout the experiment on a Grass model 7 polygraph. All of the dogs received humane care in accordance with the guidelines established by the Medical College of Wisconsin, which conform to the Guide for the Care and Use of Laboratory Animals published by the U.S. National Institutes of Health (NIH Publication 85-23, revised 1996).

Experimental Protocols. Three experimental protocols were performed in the study (Fig. 2). In all three protocols, the dogs were subjected to 60 min of LAD occlusion and 3 h of reperfusion. Coronary occlusion and reperfusion were performed using a micrometer-driven occluder placed around the LAD artery. At the end of reperfusion, infarct size was assessed ex vivo by incubating the hearts with triphenyltetrazolium chloride and expressed as a percentage of the area at risk or as a percentage of the entire left ventricle, as described previously in detail (Auchampach and Gross, 1993). Hemodynamic variables (heart rate, arterial blood pressure, left ventricular pressure, left ventricular dP/dt, and LAD coronary blood flow) were measured continuously throughout the experiments. Regional myocardial blood flow was measured by use of radioactive microspheres (¹⁴¹Ce and ⁹⁵Nb)(Auchampach and Gross, 1993) at 30 min into the prolonged 60-min occlusion period and after 3 h of reperfusion.

Protocol I (antagonist pretreatment). This protocol was designed to determine whether pretreatment with CPX, BG 9719, or BG 9928 influences the infarct size induced by acute regional myocardial ischemia and reperfusion. The dogs were subjected to 60 min of LAD occlusion and 3 h of reperfusion. Four groups of dogs (n = 8–12/group) were randomly assigned to receive vehicle, CPX, BG 9719, or BG 9928 beginning 10 min before the coronary occlusion. All of the antagonists were administered at a dose of 1 mg/kg as an i.v. bolus followed by an infusion of 10 μg/kg/min, which was continued until immediately before reperfusion (70 min total, total dose = 1.7 mg/kg).

Protocol II (ischemic preconditioning). The goal of this protocol was to determine whether pretreating with the AR antagonist influences the development of IPC induced by multiple ischemia/reperfusion cycles. Four groups of dogs (n = 6–8/group) were subjected to 60 min of coronary artery occlusion followed by 3 h of reperfusion. Preconditioning was elicited by four 5-min occlusion/5-min reperfusion cycles produced 10 min before the 60-min occlusion. The dogs were randomly assigned to receive vehicle, CPX, BG 9719, or BG 9928 beginning 10 min before the first preconditioning occlusion. The antagonists were administered at a dose of 1 mg/kg i.v. bolus followed by an infusion of 10 μg/kg/min, which was continued until the release of the prolonged occlusion (115 min total; total dose = 2.15 mg/kg).

Protocol III (antagonist at reperfusion). Since we observed that CPX and BG 9928 reduced infarct size when administered prior to and throughout the ischemic period in protocol I, we included a third protocol in which we determined whether administration of the AR antagonists reduced infarct size if administered just before and during the first hour of reperfusion. Three groups of dogs (n = 8/group) were subjected to 60 min of coronary artery occlusion followed by 3 h of reperfusion. The dogs were randomly assigned to receive CPX, BG 9719, or BG 9928 beginning 10 min before the release of the occlusion. The antagonists were administered at a dose of 1 mg/kg i.v. bolus followed by an infusion of 10 μg/kg/min for 1 h (70 min total; total dose = 1.7 mg/kg).

Exclusion Criteria. Strict criteria were used to ensure that the animals included in data analysis were healthy and exposed to similar degrees of ischemia. Dogs were excluded if heart worms were found after the animals were killed, subendocardial blood flow exceeded 0.15 ml/min/g, or more than three consecutive attempts were required to convert ventricular fibrillation with low-energy DC pulses applied directly to the hearts.

Statistical Analysis

All values are expressed as the mean ± S.E. Hemodynamic variables were analyzed by a two-way repeated-measures analysis of variance (time and drug treatment) to determine whether there was a main effect of time, a main effect of treatment, or a time-treatment interaction. If global tests showed a main effect or interaction, post hoc contrasts between time points or treatments were performed with Student's t test for unpaired or paired data, as appropriate, with the Bonferroni correction. Infarct sizes and risk region sizes were compared using a one-way analysis of variance followed by Student's t test with the Bonferroni correction.

Results

Radioligand Binding Data.Table 1 reports the affinity of the AR antagonists used in the present investigation for the four canine ARs. We have also listed affinity values of 1,3-dipropyl-8-sulfophenylxanthine (DPSPX) for canine AR receptors, since this xanthine antagonist has also been shown to reduce myocardial infarct size in an experimental dog model by Forman et al. (2000). All four of the antagonists bound with highest affinity for the A₁AR, with a rank order of potency of A₁ > A_2B > A_2A > A₃. The absolute affinity of the antagonists for the canine A₁AR was lower than that reported previously for rat and human A₁ARs (Pfister et al., 1997). This finding corresponds with previous work (Tucker et al., 1994) demonstrating that a single amino acid difference in the canine A₁AR at position 270 within the seventh transmembrane region (isoleucine to methionine) reduces binding of xanthine antagonists ∼10- to 20-fold. One important finding of these studies is that all of the antagonists exhibited relatively high affinity for the canine A_2BAR. BG 9719 and BG 9928 showed the greatest overall selectivity for the A₁AR, being 79-, 21-, and 556-, and 163-, 24-, 1457-fold selective versus the A_2A, A_2B, and A₃AR, respectively. CPX was only 9-, 6-, and 119-fold selective for the A₁AR versus A_2A, A_2B, and A₃ARs. DPSPX was essentially equipotent at binding to all four AR subtypes.

Effect of the AR Antagonists on A₁ and A_2AAR-Mediated Responses in Anesthetized Dogs. All of the antagonists were used in the present study at a dose of 1 mg/kg, followed by an infusion of 10 μg/kg/min. This dose of BG 9719 and BG 9928 has previously been shown to produce a maximal natriuretic/diuretic effect in rats, nonhuman primates, and humans (Gellai et al., 1998; Wilcox et al., 1999; Gottlieb et al., 2000, 2002; Ticho et al., 2003). In preliminary studies in un-paced barbital-anesthetized dogs, we determined whether this dose of each of the AR antagonists effectively blocked changes in heart rate and blood pressure induced by bolus injection (100 μg/kg) of the A₁AR-selective agonist N⁶-cyclopentyladenosine (CCPA). We also examined in preliminary studies the effect of the same dose of the antagonists (1 mg/kg followed by an infusion of 10 μg/kg/min) on changes in LAD coronary conductance (LAD blood flow/mean arterial blood pressure) induced by intracoronary administration of the A_2AAR-selective agonist CGS 21680. In these studies, peak changes in coronary conductance were measured after bolus injections of 100-μl aliquots of 3, 10, or 100 μM solutions of CGS 21680 given in a needle catheter inserted directly into the LAD coronary artery immediately distal to the flow probe.

In the absence of the antagonists, bolus administration of CCPA (100 μg/kg) decreased heart rate 24 ± 7% and decreased blood pressure 10 ± 2% (n = 3). Pretreatment with each of the antagonists blocked the hemodynamic actions of CCPA completely. The effects of CPX, BG 9719, and BG 9928 on changes in coronary conductance are depicted in Fig. 3. In the absence of inhibitors, bolus injection of CGS 21680 produced a dose-dependent increase in coronary conductance. The dose-response relationship of CGS 21680 on coronary conductance was shifted significantly to the right by both CPX (estimated EC₅₀ values for CGS 21680: control = 37.3 ± 3.1 μM; in the presence of CPX = 125.8 ± 1.8 μM) and BG 9928 (control = 33.1 ± 1.9 μM; BG 9928 = 74.1 ± 2.3 μM), but not by BG 9719 (control = 32.1 ± 1.8 μM; BG 9719 = 41.0 ± 2.2 μM). These data demonstrate that at a dose of 1 mg/kg followed by an infusion of 10 μg/kg/min, all three of the antagonists efficiently antagonized A₁ARs. At this dose, CPX and BG 9928 (but not BG 9719) also antagonized A_2AARs in the coronary circulation. Differences in the pharmacodynamic/pharmacokinetic properties of BG 9719 likely explain its lower in vivo potency.

Pretreatment with CPX or BG 9928 Reduces Myocardial Infarct Size (Protocol I). Pretreatment with CPX or BG 9928 resulted in a significant reduction in myocardial infarct size induced by 60 min of LAD coronary artery occlusion and 3 h of reperfusion in barbital-anesthetized dogs (Fig. 4). Infarct size, expressed as a percentage of the area at risk, was reduced from 22.6 ± 1.3% in vehicle-treated dogs to 11.1 ± 2.2% in CPX-treated dogs (∼51% reduction) and to 11.5 ± 4.6% in BG 9928-treated dogs (∼49% reduction). Pretreatment with an equivalent dose of BG 9719, however, did not produce a significant reduction in infarct size (18.8 ± 1.0% of the area at risk). The protection against infarction provided by CPX or BG 9928 was not the result of differences in the area at risk size (control, 32.5 ± 1.5%; CPX, 31.9 ± 1.8%; BG 9719, 30.9 ± 1.6%, BG 9928, 30.4 ± 1.7%), changes in hemodynamic parameters (Table 2), or increases in regional myocardial blood flow (Tables 3 and 4). When infarct size, expressed as a percentage of the area at risk, was plotted versus transmural collateral blood flow, an inverse relationship was evident in all of the treatment groups such that infarct size was progressively smaller with higher levels of collateral flow (Fig. 4B). This relationship between collateral blood flow was shifted downward in dogs treated with CPX or BG 9928 compared with vehicle-treated control dogs, indicating that the smaller infarcts observed in these two groups of dogs were independent of changes in collateral blood flow during the ischemic period.

Pretreatment with CPX, BG 9719, or BG 9928 Does Not Block IPC (Protocol II). To determine whether any of the AR antagonists blocked the development of IPC, we subjected dogs to four 5-min coronary occlusion/5-min reperfusion cycles 10 min before the 60-min occlusion period. This protocol has been shown previously to induce a preconditioning effect that resulted in a robust reduction in myocardial infarct size (Gross and Auchampach, 1992). As shown in Fig. 5, we also observed a marked reduction in myocardial infarct size in dogs subjected to multiple-cycle IPC. Infarct size as a percentage of the area at risk was reduced by IPC to 8.1 ± 2.3% (versus 22.6 ± 1.3% in control dogs from protocol I) or 2.6 ± 0.7% of the left ventricle (control = 7.3 ± 0.5%). In the three groups of dogs pretreated with CPX, BG 9719, or BG 9928, infarct size continued to be reduced by IPC (3.2 ± 1.6%, 7.4 ± 1.6%, and 2.7 ± 2.6% of the area at risk, respectively). Similar to our observations in protocol I, we also observed that none of the antagonists influenced the area at risk size (control, 32.5 ± 1.5%; Precond, 32.0 ± 1.3%; Precond + CPX, 33.4 ± 1.6%; Precond + BG 9719, 33.7 ± 3.0; Precond + BG 9928, 33.7 ± 3.2%), systemic hemodynamics (Table 5), or regional myocardial blood flow (Tables 3 and 4). Plots of infarct size versus transmural collateral blood flow (Fig. 5B) demonstrated that infarct size was reduced by IPC in a manner that was independent of changes in collateral blood flow and that none of the antagonists interfered with this relationship.

Treatment with CPX or BG 9928 Reduces Infarct Size when Administered Just Before Reperfusion (Protocol III). This protocol was designed to determine whether the AR antagonists reduce infarct size if administered just before the reperfusion period. In this protocol, CPX, BG 9719, or BG 9928 was administered at the same dose and duration as protocol I, except that the drugs were given 10 min before the release of the 60-min occlusion period and continued for the first hour of reperfusion. Interestingly, the two groups of dogs that were treated with CPX or BG 9928 during reperfusion continued to exhibit smaller infarcts compared with the vehicle-treated control group from protocol I (Fig. 6; infarct sizes as a percentage of the area at risk were: CPX, 12.9 ± 2.6%; BG 9719, 20.8 ± 4.5%; BG 9928, 12.6 ± 3.0%). The relationship between transmural collateral blood flow and infarct size was similarly shifted downward in the two groups of dogs given CPX or BG 9928 (Fig. 6B). None of the three antagonists influenced the area at risk size (control, 32.5 ± 1.5%; CPX, 31.4 ± 1.2%; BG 9719, 33.6 ± 2.4%; BG 9928, 36.9 ± 1.9%), systemic hemodynamics (Table 6), or regional myocardial blood flow (Tables 3 and 4).

Discussion

This study demonstrates that three different AR antagonists, with preferential affinity for the A₁AR, did not exacerbate the development of irreversible tissue injury induced by a prolonged coronary occlusion and reperfusion in an anesthetized dog model at a dose that produces maximal natriuresis/diuresis. Indeed, we found that two of the three AR antagonists that we studied, CPX and BG 9928, provided a cardioprotective effect and resulted in reductions in infarct size of ∼40 to 50%. The protective effects of CPX and BG 9928 were evident if the drugs were administered prior to and during the ischemic period or if they were administered just before and during the reperfusion period, suggesting that they may act at least partially by reducing reperfusion-mediated injury. In a clinically relevant model of IPC produced by multiple ischemia/reperfusion cycles, the present study also demonstrates that blockade of ARs does not interfere the infarct size-sparing actions of IPC.

AR Antagonists during Ischemia/Reperfusion Injury. Adenosine accumulation in the ischemic myocardium is generally considered to serve a protective role by delaying the development of ischemic injury via multiple receptor and nonreceptor mechanisms (Ely and Berne, 1992). Receptor-mediated mechanisms are manifest by coronary dilation, which increases oxygen supply (A_2AARs), and by multiple effects, which decrease oxygen demand (A₁AR), including negative inotropism, chronotropism, and dromotropism (Ely and Berne, 1992). Via actions on A₁ARs and, potentially, A₃ARs expressed in cardiomyocytes, adenosine also provides direct cardioprotection likely to be mediated by ATP-sensitive potassium (K_ATP) channels. Finally, adenosine also reduces inflammation and reperfusion-mediated injury via actions on A_2AARs expressed in neutrophils and other inflammatory cells (Vinten-Johansen et al., 1999). Nonreceptor-mediated actions of adenosine involve serving as a substrate for purine salvage to restore energy supply during reperfusion (Ely and Berne, 1992).

Based on these important actions of adenosine, it has been postulated that AR antagonism would have negative effects during acute ischemia/reperfusion injury. However, previous studies are conflicting, and have reported that AR antagonists increase, decrease, or have no effect on infarct size (Auchampach and Gross, 1993; Thornton et al., 1993; Zhao et al., 1994; Haessler et al., 1996; Neely et al., 1996; Todd et al., 1996; Auchampach et al., 1997b; Kitakaze et al., 1997; Domenech et al., 1998; Forman et al., 2000). How can these discrepant data be reconciled? Forman et al. (2000) presented the theory that the differences may be related to the selectivity of the antagonist used in the various studies. These investigators (Forman et al., 2000) suggested that antagonists with preferential affinity for the A_2AAR may increase infarct size by reducing coronary blood flow and/or by inhibiting the numerous anti-inflammatory actions of adenosine mediated by A_2AARs. On the other hand, since A₁ARs on neutrophils promote chemotaxis, it was suggested that antagonists with preferential affinity for the A₁AR may reduce neutrophil-mediated reperfusion injury resulting in a reduction in infarct size. Thus, even though adenosine may mediate several salutary actions during ischemia, it was suggested that it may also exert deleterious actions during reperfusion such that selective blockade of specific AR subtypes could result in effective reduction in infarct size.

The results of the present investigation are in general agreement with this theory proposed by Forman et al. (2000). However, our results suggest that AR subtypes other than the A₁AR may be involved. This conclusion is based on our observation that only two of the three AR antagonists that we tested (CPX and BG 9928, but not BG 9719) effectively reduced infarct size, even though all three of the drugs were administered at a dose that efficiently blocked the A₁AR. We speculate, therefore, that CPX and BG 9928 reduced infarct size by blocking the A_2BAR rather than the A₁AR. This hypothesis is based on two pieces of evidence. First, we found in our radioligand binding studies that CPX, BG 9719, and BG 9928, as well as DPSPX, have relatively high affinity for the canine A_2BAR (Table 1). Second, we demonstrated that the dose of CPX and BG 9928 used in our in vivo studies was likely sufficient to antagonize A_2BARs. This latter conclusion is based on our observation that both CPX and BG 9928 (but not BG 9719) antagonized CGS 21680-mediated coronary vasodilation, implying that the drugs were administered at a dose sufficient to block A_2AARs and, by corollary, A_2BARs (since the antagonists have higher affinity for A_2BARs versus A_2AARs). The lack of effect of BG 9719 to block A_2BARs at the dose utilized in our investigation may explain its ineffectiveness at reducing infarct size (although our theory suggests that higher doses of BG 9719 would be effective). Additional mechanistic studies with selective A_2BAR antagonists, once developed and made readily available, are necessary to test this hypothesis. Although we predict that CPX and BG 9928 reduced infarct size by blocking the A_2BAR, we must also consider other potential mechanisms. For instance, CPX and BG 9928 may have been effective by blocking A₃ARs. It is also possible that CPX and BG 9928 were effective by mechanisms unrelated to AR blockade, such as inhibition of intracellular phosphodiesterases responsible for degrading cAMP. Finally, we also cannot discount the possibility that the dose of BG 9719 was too low and that CPX and BG 9928 reduced infarct size via efficient blockade of A₁ARs.

AR Antagonists and IPC. The elucidation of the mechanisms of IPC has uncovered a new physiological role of adenosine. IPC is the phenomenon whereby brief periods of ischemia induce adaptive responses that increase the tolerance of the heart to subsequent ischemic episodes. There are two phases of IPC: an early phase that provides immediate protection (classical or “early” IPC) and lasts approximately 1 to 2 h, and a second phase (second window of protection or “late” IPC) that develops 12 to 24 h later and lasts for days. The early phase involves acute changes in metabolism due to post-translational modification of protective proteins (Nakano et al., 2000). The second phase involves increased synthesis of cardioprotective proteins (Bolli, 2000). Both the early and late phases of IPC can be triggered by adenosine produced during the preconditioning ischemia, since pretreatment with AR antagonists blocks the induction of both early and late IPC (Vinten-Johansen et al., 1999). Thus, studies of IPC have demonstrated that adenosine not only provides acute cardioprotection, but that it also induces a sustained ischemia-tolerant phenotype.

In addition to adenosine, subsequent work has shown that other mediators generated during ischemia are also capable of inducing the early phase of IPC, including bradykinin and opioid peptides (Nakano et al., 2000). Since all three of these receptor systems couple to similar intracellular signaling pathways (i.e., protein kinases, ATP-sensitive potassium channels), it has been proposed (Nakano et al., 2000) that a threshold level of stimulation is necessary to precondition the myocardium. That is, the combined actions of adenosine, bradykinin, opioid peptides, and potentially other mediators released during ischemia stimulate kinase signaling pathways to a threshold level that results in the cardioprotective phenotype of IPC. This additive theory of IPC (Nakano et al., 2000) predicts that IPC induced by a mild stimulus (i.e., a single occlusion/reperfusion cycle) can be inhibited by blocking a single endogenous mediator of IPC, presumably by preventing threshold activation of cardioprotective signaling mechanisms, whereas IPC induced by a more robust stimulus (i.e., multiple ischemia/reperfusion cycles) cannot be blocked efficiently since other mediators are generated in sufficient amounts to attain threshold. This hypothesis has recently been tested directly in an in vivo rabbit model of infarction (Goto et al., 1995; Miki et al., 1998). In this model (Goto et al., 1995; Miki et al., 1998), blockade of opioid receptors or bradykinin receptors efficiently blocked IPC induced by a single 5-min occlusion period. However, pharmacological blockade of either type of receptor was not capable of blocking IPC induced by multiple ischemia/reperfusion cycles (Goto et al., 1995; Miki et al., 1998).

The present study provides additional support for this theory. We observed that pretreating dogs with three potent A₁AR antagonists did not block the reduction in infarct size provided by IPC induced by multiple ischemia/reperfusion cycles. In fact, CPX and BG 9928 appeared to provide additional cardioprotection in preconditioned dogs (Fig. 5). In contrast, in a previous study (Auchampach and Gross, 1993) we observed that pretreating dogs with a single dose of CPX completely blocked the reduction in infarct size provided by IPC induced by a single 5-min coronary occlusion. In this study, it is important to note that CPX was administered as a single bolus dose prior to the IPC stimulus (Auchampach and Gross, 1993) and was not administered during reperfusion. Collectively, the results of the present study as well as those from previous work in our laboratory (Auchampach and Gross, 1993) demonstrate that the early phase of IPC in the dog is not mediated solely by adenosine. Rather, it appears likely that multiple mediators are involved that act in concert with adenosine.

Conclusions

The present investigation unveiled several new aspects of adenosine biology during ischemia/reperfusion injury. Our results demonstrate that A₁AR blockade is not detrimental in this experimental model and does not block IPC induced by multiple ischemia/reperfusion cycles. Although caution is certainly necessary to extrapolate our findings in a dog model to humans, especially since it has been suggested that A₁AR-mediated responses in the dog may differ from that of humans (Belloni et al., 1989; Martin, 1992), our data provide evidence that the use of A₁AR antagonists as diuretics inpatients with ischemic heart disease may not pose a problem. Dose-relationship studies in additional models will be necessary to address this issue further.