Recent studies have demonstrated that the adenosine A1 receptor agonist 2-chloro-N6-cyclopentyladenosine (CCPA) and the adenosine A3 receptor agonist N6-(3-iodobenzyl)adenosine-5'-N-methyluronamide (IB-MECA) produce a delayed phase of protection against infarction similar to the late phase of ischemic preconditioning (PC). However, the mechanism for adenosine A1 or A3 receptor-induced late PC remains unknown. The goal of this study was to determine whether the delayed cardioprotective effects of adenosine A1 or A3 receptors are mediated by cyclooxygenase-2 (COX-2), which is an obligatory mediator of ischemic PC. We found that COX-2 protein expression (Western blotting) did not increase 24 h after the administration of either CCPA (100 microg/kg iv) or IB-MECA (300 microg/kg iv) compared with controls. To probe the role of constitutive COX-2 expression, conscious rabbits were subjected to 30-min coronary occlusion followed by 72-h reperfusion. Twenty-four hours before the occlusion, the rabbits were pretreated with CCPA (100 microg/kg iv) or IB-MECA (300 microg/kg iv). Both CCPA and IB-MECA resulted in a marked (approximately 47%) reduction in infarct size vs. controls [36.2 +/- 4.0% of the risk region (n = 9), 31.2 +/- 4.7% (n = 9), and 59.5 +/- 3.8% (n = 9), respectively; P < 0.05], similar to that induced by the late phase of ischemic PC [31.8 +/- 3.2% (n = 9)]. The selective COX-2 inhibitor N-(2-[cyclohexyloxy]4-nitrophenyl)methanesulfonamide (NS-398, 5 mg/kg), which abolished the protective effect of ischemic late PC, failed to block the protection of either CCPA or IB-MECA, indicating that COX-2 does not mediate the delayed protection of either CCPA or IB-MECA [CCPA + NS-398, 29.1 +/- 3.4% (n = 7); IB-MECA + NS-398, 34.9 +/- 2.9% (n = 8)]. NS-398 in itself did not affect infarct size [54.9 +/- 3.7% (n = 9)]. Taken together, these results demonstrate that, in contrast to ischemia-induced late PC, the mechanisms of adenosine A1 or A3 receptor-induced late PC is independent of COX-2.