Cyclooxygenase-2 Mediates Ischemic, Anesthetic, and Pharmacologic Preconditioning In Vivo 

All experimental procedures and protocols used in this investigation were reviewed and approved by the Institutional Animal Care and Use Committee of the Medical College of Wisconsin (Milwaukee, Wisconsin). Furthermore, all conformed to the Guiding Principles in the Care and Use of Animals  5of the American Physiologic Society and were in accordance with the Guide for the Care and Use of Laboratory Animals . 6 

Instrumentation of dogs has been previously described in detail. 7Briefly, mongrel dogs (weight, 20–24 kg) of either sex were anesthetized with sodium barbital (200 mg/kg) and sodium pentobarbital (15 mg/kg) and ventilated using positive pressure with an air-and-oxygen mixture after tracheal intubation. A dual micromanometer-tipped catheter was inserted into the aorta and left ventricle for measurement of aortic and left ventricular (LV) pressures and the maximum rate of increase of LV pressure (+dP/dt^max). Heparin-filled catheters were inserted into the left atrial appendage, the right femoral artery, and the right femoral vein for administration of radioactive microspheres, withdrawal of reference blood flow samples, and fluid or drug administration, respectively. The left anterior descending coronary artery (LAD) was isolated, and a silk ligature was placed around this vessel immediately distal to the first diagonal branch for production of coronary artery occlusion and reperfusion. Hemodynamic data were continuously monitored on a polygraph and digitized using a computer interfaced with an analog-to-digital converter.

Baseline systemic hemodynamic data were recorded 90 min after instrumentation was completed (fig. 1). All dogs were subjected to a 60-min LAD occlusion followed by 3 h of reperfusion. Dogs were randomly assigned to receive oral celecoxib (200 mg), aspirin (650 mg), or acetaminophen (500 mg) three times (24, 12, and 2 h) before surgical instrumentation in the presence or absence of IPC, APC, or PPC in 12 separate experimental groups. IPC was produced with four 5-min periods of LAD occlusion interspersed with 5-min periods of reperfusion. APC was produced by administration of isoflurane (1.0 minimum alveolar concentration) for 30 min and discontinued 30 min before LAD occlusion. End-tidal concentrations of isoflurane were measured at the tip of the endotracheal tube by an infrared anesthetic analyzer that was calibrated with known standards before and during experimentation. Last, PPC was produced by administration of intravenous diazoxide (2.5 mg/kg over 15 min) immediately before coronary occlusion. Transmural myocardial blood flow was measured under baseline conditions, during LAD occlusion, and after 1 h of reperfusion. Dogs that developed intractable ventricular fibrillation or those with subendocardial blood flow greater than 0.15 ml · min⁻¹· g⁻¹were excluded from subsequent analysis.

Myocardial infarct size was measured at the end of each experiment as previously described. 8Briefly, the LV area at risk (AAR) for infarction was separated from the normal zone, and the two regions were incubated at 37°C for 20–30 min in 1% 2,3,5-triphenyltetrazolium chloride in 0.1 m phosphate buffer adjusted to a pH of 7.4. After overnight storage in 10% formaldehyde, infarcted and noninfarcted myocardial areas within the AAR were carefully separated and weighed. Infarct size was expressed as a percentage of the AAR.

Carbonized plastic microspheres (15 ± 2 μm [SD] in diameter) labeled with ¹⁴¹Ce, ¹⁰³Ru, or ⁹⁵Nb were used to measure regional myocardial perfusion as previously described. 9Transmural tissue samples were selected from the ischemic region (distal to the LAD occlusion) and subdivided into subepicardial, midmyocardial, and subendocardial layers of approximately equal thickness. Samples were weighed and placed in scintillation vials for isotope activity determination. Similarly, the activity of each isotope in the reference blood flow sample was assessed. Tissue blood flow was calculated as Q^r· C^m· C^r−1, where Q^ris the rate of withdrawal of the reference blood flow sample (ml/min), C^mis the activity (counts · min ⁻¹· g⁻¹) of the myocardial tissue sample, and C^ris the activity (counts/min) of the reference blood flow sample. Transmural blood flow was considered to be the average of subepicardial, midmyocardial, and subendocardial blood flows.

In separate experiments (n = 8), dogs were anesthetized and instrumented as described above. LV myocardium was biopsied at baseline and after 30 min of isoflurane in the absence or presence of pretreatment with celecoxib (200 mg orally at 24, 12, and 2 h before instrumentation). Myocardial 6-keto-prostaglandin F^1α(6-keto-PGF^1α) was measured with enzyme immunoassay (Cayman Chemical, Ann Arbor, MI).

Statistical analysis of data within and between groups was performed with analysis of variance for repeated measures followed by the Student-Newman-Keuls test. The Mann–Whitney rank sum test was used to compare myocardial prostaglandin concentrations. Changes within and between groups were considered significant when the probability (P ) value was less than 0.05. Data are expressed as mean ± SEM.

Ninety-one dogs were instrumented to obtain 86 successful experiments. Two dogs were excluded because of intractable ventricular fibrillation (1 control; 1 diazoxide alone). Three dogs were excluded because subendocardial blood flow exceeded 0.15 ml · min⁻¹· g⁻¹(1 isoflurane alone, 1 aspirin plus isoflurane, and 1 celecoxib alone).

There were no significant differences in systemic hemodynamics between experimental groups under baseline conditions, during LAD occlusion, or during reperfusion (table 1). Celecoxib, aspirin, and acetaminophen did not alter hemodynamics. Isoflurane caused significant (P < 0.05) decreases in heart rate, mean arterial pressure, LV systolic pressure, and LV +dP/dt^maxin the presence or absence of celecoxib, aspirin, or acetaminophen pretreatment. There were no differences in coronary collateral blood flow among groups (table 2).

The LV AARs were similar among groups (control, 37 ± 2%; celecoxib, 33 ± 4%; APC, 37 ± 2%; IPC, 39 ± 2%; celecoxib plus APC, 36 ± 2%; celecoxib plus IPC, 31 ± 3%; aspirin, 33 ± 2%; aspirin plus APC, 41 ± 3%; acetaminophen, 37 ± 3%; acetaminophen plus APC, 41 ± 4%; PPC, 44 ± 2%; celecoxib plus PPC, 36 ± 4% of left ventricle). IPC, APC, and PPC decreased myocardial infarct size (9 ± 1, 12 ± 2, and 11 ± 1% of the LV AAR, respectively) as compared with control (30 ± 1%;fig. 2). Celecoxib alone had no effect on infarct size (26 ± 3%) but abolished IPC (30 ± 3%), APC (30 ± 3%), and PPC (26 ± 1%). Aspirin and acetaminophen alone did not alter the extent of myocardial infarction (24 ± 3 and 29 ± 2%, respectively), nor did these drugs abolish the protection produced by APC (18 ± 1 and 19 ± 1%, respectively).

Isoflurane increased myocardial 6-keto-PGF^1αto 463 ± 267% of baseline (n = 4) in untreated dogs. In contrast, 6-keto-PGF^1αwas unchanged by isoflurane in dogs (n = 4) pretreated with celecoxib (94 ± 13% of baseline).

A variety of preconditioning stimuli are capable of eliciting protection in ischemic myocardium by activating common signal transduction pathways. Reactive oxygen species, membrane-bound receptors (e.g. , adenosine subtype 1, δ opioid), intracellular kinases (e.g. , protein kinase C, protein tyrosine kinase), and ion channels (e.g. , K^ATPchannels) have been identified that play critical roles in triggering or mediating IPC 10and APC. 11Two proteins, including inducible nitric oxide synthase and COX-2, have been the focus of recent studies characterizing the more prolonged transition of myocardium from a vulnerable to a protected state. 12This transition has been termed the delayed  or late phase  of IPC. Protection caused by acute or early IPC typically lasts for approximately 2 h after a brief ischemic episode; however, myocardium again becomes resistant to infarction 24 h later. This late phase of preconditioning is dependent on inducible nitric oxide synthase–derived nitric oxide and increased COX-2 expression and activity. 13Many drugs, including volatile anesthetics, 14may mimic the salutary action of both acute 3and delayed IPC by activating similar signaling pathways. 11The mitochondrial K^ATPchannel has been strongly implicated during IPC 15and APC 3and has also been shown to mediate delayed preconditioning phenomena. 16Direct activation of mitochondrial K^ATPchannels with diazoxide produces both early and delayed protection, and these beneficial actions are blocked by a nitric oxide synthase inhibitor. 16Prostaglandins also influence the activity of K^ATPchannels. Intravenous administration of prostaglandins E¹or E⁰decreased myocardial infarct size, and this beneficial effect was blocked by the selective mitochondrial K^ATPchannel antagonist 5-hydroxydecanoate. 17These provocative findings strongly suggested that interventions that alter prostanoid concentrations may favorably influence K^ATPchannel activity and reduce the degree of myocardial injury in response to a subsequent ischemic event. Therefore, it has become increasingly apparent that endogenous molecules including nitric oxide 18and COX-generated metabolites of arachidonic acid may potentiate K^ATPchannel activity 17and modulate the response of myocardium to ischemia.

The current results indicate that COX-2 is a critical mediator of IPC, APC, and PPC in vivo . The findings further suggest that COX-derived prostanoids may modulate K^ATPchannel activity elicited by IPC or pharmacologic agonists, including isoflurane and diazoxide. The role of COX-2 and its prostanoid products in myocardial protection has been extensively studied and validated in models of delayed preconditioning against infarction and stunning. 19,20In contrast, whether prostanoids are capable of favorably affecting acute ischemic injury remains controversial, as does the specific role of COX-2 in IPC and other forms of acute preconditioning. However, genetic and pharmacologic evidence suggests that cyclooxygenase or cyclooxygenase-derived metabolites may contribute to protection against ischemic injury. Recovery of LV pressure after global ischemia and reperfusion was significantly attenuated in mice with targeted deletions of the COX-1 or -2 gene. 21Findings that myocardial ischemic injury is increased in COX-2 null mice implies that endogenous production of COX-2–derived prostanoids produces cardioprotective effects even though constitutive expression of protein is low or undetectable. Furthermore, prostaglandin E²(PGE²) and 6-keto-PGF^1α, a stable metabolite of prostacyclin (PGI²), are present in myocardium of COX-1 null mice, indicating that constitutive COX-2 expression is sufficient to produce measurable quantities of cardioprotective eicosanoids. 2Pharmacologic inhibition of COX-2 with celecoxib caused dose-dependent decreases in functional recovery of myocardium after global ischemia and reperfusion in isolated rabbit hearts. 22The nonselective cyclooxygenase inhibitor indomethacin abolished reductions of infarct size after IPC in enflurane-anesthetized pigs. 23Another nonselective cyclooxygenase inhibitor (meclofenamate) also abrogated the beneficial effects of IPC on ischemia and reperfusion-induced arrhythmias in dogs. 24Conversely, administration of indomethacin 25or aspirin 26failed to block IPC in rat myocardium in vitro  or in vivo , respectively. IPC-induced protection against contractile dysfunction after ischemia and reperfusion also remained intact in COX-1 and -2 knockout mice. 21These findings suggested that important species differences (rodents vs.  larger mammals) may exist that complicate interpretation of studies designed to evaluate the role of cyclooxygenase in protection against ischemic injury. In addition, the previous data also suggested that differences in experimental design (isolated hearts vs. in vivo  preparations), variability in drug selectivity for cyclooxygenase isoforms, or shunting of arachidonic acid through alternate metabolic pathways (e.g. , leukotrienes, hydroxyeicosatetraenoic acids) may also be important in determining the outcome of studies in which cyclooxygenase activity is manipulated.

A wide variety of physical, chemical, inflammatory, and mitogenic stimuli liberate arachidonic acid from membrane phospholipids by activating phospholipase A², thereby providing the substrate for cytosolic cyclooxygenase. 27Two major forms of cyclooxygenase have been identified. COX-1 is often described as the constitutively expressed isoform, whereas COX-2 is classically identified as an inducible form of the enzyme whose production may be stimulated by cytokines, growth factors, phorbol esters, and bacterial lipopolysaccaride. 28Despite the relative simplicity of these definitions, COX-2 may also be constitutively expressed in renal parenchyma, 29neural tissue, 30and myocardium. 20Prostaglandins synthesized by COX-1 have been strongly implicated in the maintenance of gastrointestinal mucosal function and vascular and cellular homeostasis. 27,28COX-1 is also the only enzyme isoform expressed in platelets, and activation of this isoform is directly responsible for thromboxane A²(TXA²)–dependent platelet aggregation. 27,28Nonsteroidal antiinflammatory drugs (e.g. , ibuprofen) inhibit the activity of both cyclooxygenase isoforms at the cyclooxygenase active site 31that initiates the committed step in the biosynthesis of prostaglandins. Aspirin inhibits cyclooxygenase activity by irreversibly acetylating the enzyme and is 100 times more selective for COX-1 than -2. 32In contrast, celecoxib is 375 times more selective for COX-2 than -1. 28Both COX-1 and -2 are relatively unaffected by acetaminophen, 32but acetaminophen has recently been shown to inhibit a newly identified isoform variant of COX-1, identified as COX-3. 32COX-2 has been shown to be a major source of PGE²and PGI²biosynthesis in humans. 28Production of these eicosanoids is significantly decreased by celecoxib. Aspirin and other COX-1 antagonists significantly decrease TXA²concentrations and TXA²-dependent platelet aggregation. 28These drugs also reduce PGE²and PGI²concentrations, but decreases in these vasodilatory prostaglandins are less pronounced than concomitant declines in thromboxanes. 28,33Therefore, aspirin seems to favorably modify the ratio of PGI²to TXA², whereas COX-2 selective antagonists may theoretically produce adverse cardiovascular consequences by selectively reducing PGI²concentrations and allowing the relatively unopposed action of TXA². 1 

In the current investigation, celecoxib alone did not affect infarct size but abolished the protective effects of preconditioning by repetitive brief ischemia, exposure to isoflurane, and administration of diazoxide in a canine model of myocardial infarction independent of alterations in systemic hemodynamics or coronary collateral blood flow. In contrast, aspirin or acetaminophen did not block reductions of myocardial infarct size in response to preconditioning with isoflurane. In addition, the current results are the first to show that isoflurane increases myocardial concentrations of 6-keto-PGF^1α. Pretreatment with celecoxib abolished the formation of this cardioprotective prostanoid in response to isoflurane. In view of previous results obtained in large mammals, the current findings strongly suggest that COX-2–derived prostanoids, such as PGI², may enhance mitochondrial K^ATPchannel activation in response to ischemic and pharmacologic stimuli, including isoflurane. The current results further suggest that COX-1 selective antagonists seem to be less likely to impair signal transduction during APC by maintaining a favorable balance between PGI²and TXA², but further investigation will clearly be needed to verify this hypothesis. The actions of cyclooxygenase antagonists have been shown to be dose dependent. 27For example, small quantities of aspirin did not block late preconditioning against myocardial stunning, but higher doses abolished IPC-enhanced recovery of contractile function and prevented increases in myocardial 6-keto-PGF^1αand PGE²after late preconditioning. 33Taken together, the current and previous findings suggest that prostanoids derived from COX-2 are important endogenous mediators of preconditioning that seem to be recruited in response to ischemic or pharmacologic stimuli. Inhibition of COX-2 with selective antagonists or higher doses of nonselective blockers may interfere with this important adaptive response during ischemia–reperfusion injury.

The current investigation should be interpreted within the constraints of several potential limitations. The dose of cyclooxygenase inhibitors was based on a comparable dose in humans and on previous reports in the literature. Dose–response relations to cyclooxygenase inhibitors were not specifically examined in the current investigation. However, the dose of aspirin used in the current investigation that did not block APC was similar to the dose previously used in rabbits to block delayed preconditioning. Therefore, the relative dose-dependent selectivity of cyclooxygenase inhibitors may be species related, and similar results may not be observed in humans. The area of the left ventricle at risk for infarction and coronary collateral blood flow are important determinants of the extent of myocardial infarction. No differences in these variables were observed between experimental groups that would account for the current findings. Isoflurane caused similar alterations in the hemodynamic determinants of myocardial oxygen consumption in the presence or absence of cyclooxygenase inhibitors. Therefore, it seems highly unlikely that the current findings are related to differences in myocardial oxygen consumption among groups. However, coronary venous oxygen tension was not measured and myocardial oxygen consumption was not directly quantified in the current investigation.

In summary, the current results indicate that reductions in myocardial infarct size produced by repetitive brief ischemia, exposure to isoflurane, and administration of diazoxide were abolished by selective inhibition of COX-2 independent of alterations in systemic hemodynamics or coronary collateral blood flow in dogs. In contrast, aspirin and acetaminophen failed to block isoflurane-induced protection against ischemic injury. The current results are consistent with the hypothesis that inhibition of COX-2 decreases PGI²and PGE²concentrations and alters signaling responsible for myocardial protection, including activation of K^ATPchannels, during IPC, APC, and PPC. The findings confirm and extend recent evidence suggesting that COX-2 is an essential cardioprotective protein.

The authors thank David A. Schwabe, B.S.E.E. (Department of Anesthesiology, Medical College of Wisconsin, Milwaukee, Wisconsin), for technical assistance and Mary Lorence-Hanke, A.A. (Department of Anesthesiology, Medical College of Wisconsin), for assistance in preparation of the manuscript.