Monophosphoryl Lipid A Attenuates Myocardial Stunning in Dogs: Role of ATP-Sensitive Potassium Channels

Abstract

Results of previous studies indicate that monophosphoryl lipid A (MLA) reduces myocardial infarct size when administered 24 but not 1 h before a prolonged period of regional ischemia in dogs and rabbits. This cardioprotective effect of MLA could be reversed by the administration of the adenosine triphosphate (ATP)-sensitive potassium channel (K_ATP) blockers, glibenclamide, or 5-hydroxydecanoate. MLA also was shown to attenuate myocardial stunning in dogs; however, its mechanism in this model remains unknown. Therefore the major aim of our study was to determine the dose-related effect of MLA to enhance contractile function in stunned myocardium and to determine the role of the K_ATP channel in mediating its cardioprotective effect. To produce myocardial stunning, barbital-anesthetized dogs were subjected to five cycles of 5 min of left anterior descending (LAD) coronary artery occlusion interspersed with 10 min of reperfusion and finally followed by 2 h of reperfusion. Regional segment shortening (%SS) was determined by sonomicrometers implanted in the subendocardium of the ischemic region. Single intravenous doses of MLA in the range of 10-35 μg/kg given 24 h before ischemia resulted in an improvement in %SS over a 2-h reperfusion period. Similar to results obtained in the canine and rabbit infarct models, cardioprotection against stunning with MLA appears to require activation of K_ATP channels during ischemia, because glibenclamide (50 μg/kg, 15 min before ischemia) completely blocked the effect of MLA to improve regional %SS during reperfusion. Cardioprotective doses of MLA were without effect on systemic hemodynamics, blood gases, and pH throughout the experiment. No treatment-related effects on regional myocardial blood flow were observed during ischemia or reperfusion. These results suggest that MLA improves %SS at doses of 10-35 μg/kg by an ATP-sensitive potassium channel-dependent process, and that MLA may mimic the antistunning effects observed during the second window of ischemic preconditioning.

Ischemia and subsequent reperfusion of myocardial tissue occur clinically during thrombolysis, unstable angina, cardiopulmonary bypass, or after coronary artery balloon angioplasty. This ischemia/reperfusion injury can manifest itself in the form of irreversible injury (necrotic and apoptotic myocyte death or infarction) and reversible injury (contractile dysfunction or stunning and disorders of automaticity manifest by atrial and ventricular arrhythmias; ^1,2).

A great deal of recent attention has focused on under-standing ischemic preconditioning, an endogenous cardioprotective mechanism, first described in 1986, in which brief periods of antecedent ischemia result in a rapidly evolving adaptive response whereby the myocardium is resistant to infarction in a subsequent prolonged ischemic challenge ⁽³⁾. Ischemic preconditioning has been observed in many species, including rats, rabbits, pigs, and dogs, and develops within minutes of reperfusion after a single brief preconditioning period and persisted from 30 to 120 min thereafter in most studies ^(3-6). Efforts to document that ischemic preconditioning (IP) additionally protects against myocardial stunning have failed ^(7,8). Paradoxically, episodes of antecedent ischemia not only do not reduce stunning to a more prolonged ischemic insult but may actually produce stunning, although it appears that stunning is not required for IP to reduce infarct size.

Recently a second window of preconditioning (SWOP) was observed whereby a brief period of ischemia produces a reduction in myocardial infarct size 24-48 h after antecedent ischemia ⁽⁹⁾. Interestingly, this delayed preconditioned state in response to ischemia is not associated with myocardial stunning as an associated consequence; presumably the stunning observed in the short term has dissipated by the time the SWOP occurs. More recently Sun et al. ^(8,10) showed a SWOP against myocardial stunning on exposure of the preconditioned swine heart to a secondary ischemic stress.

Because of the prolonged duration (24-48 h in length) of the protective adaptation of delayed preconditioning (PC) and also because of the antistunning effect of SWOP, many investigators have been interested in identifying pharmacologic means of inducing delayed PC. One such agent that appears to induce delayed myocardial PC is monophosphoryl lipid A (MLA). Administered as a single 10- to 35-μg/kg dose 9-36 h before ischemia, MLA reduces infarct size in dogs and rabbits ^(11-16) and improves ventricular functional recovery in the isolated rat heart ⁽¹⁷⁾ and in the in situ canine cardiopulmonary-bypass model ⁽¹⁸⁾. Cardioprotection in the dog and rabbit infarct model appears to require activation of myocardial adenosine triphosphate (ATP)-sensitive potassium (K_ATP) channels, as investigations showed that intravenous or intracoronary glibenclamide (GLB) or 5-hydroxydecanoate administered 30-60 min before ischemia blocked the cardioprotective effect of MLA ^(14,19,20). Recently it was observed that pretreatment with 35 μg/kg of MLA 24 h but not 1 h before six cycles of 5 min of ischemia and 10 min of reperfusion in dogs reduces regional myocardial stunning ⁽²¹⁾; however, the mechanism by which MLA produced this effect is unknown.

Our study was undertaken to evaluate the effect of several doses of MLA delayed preconditioning against regional myocardial stunning in the dog in an effort to determine the optimal cardioprotective dose. The potential role of K_ATP channels in mediating the antistunning effect of MLA also was evaluated by using in vivo pharmacologic blockade with the K_ATP-channel antagonist, GLB.

MATERIALS AND METHODS

Animal model

Animal experiments were conducted according to guidelines in "Position of the American Heart Association on Research and Animal Use" of 1984 by the American Heart Association and in compliance with the American Association of Laboratory Animal Care (AALAC) and approved regulations of the Animal Care Committee of the Medical College of Wisconsin.

Mongrel dogs of either sex (19.0-25.0 kg) were fasted overnight and were administered sodium barbital (200 mg/kg) and sodium pentobarbital (15 mg/kg) and ventilated with room air plus 100% O₂ to a PO₂ of ∼120 mm Hg at an end-expiratory pressure of 5-7 cm H₂O (model 607; Harvard Apparatus, S. Natick, MA, U.S.A.). Aortic and left ventricular pressures were measured via a double pressure transducer-tipped catheter (PC 771; Millar Instruments, Houston, TX, U.S.A.), and ±dP/dt values were calculated as the first derivative of the left ventricular pulse pressure.

After performance of a left thoracotomy at the fifth intercostal space, the lung was retracted, a partial pericardiectomy was performed, and the heart suspended in a cradle. The left anterior descending (LAD) coronary artery was dissected free of the heart wall for a 1.5-cm length distal to the first diagonal branch, and an electromagnetic flow probe (Statham SP 7515, Oxnard, CA, U.S.A.) was placed around the vessel for measurement of coronary blood flow via a flowmeter (Statham 2202). Vessel occlusion was accomplished by using a mechanical micrometer-driven occluder placed distal to the flow probe.

A minimal heart rate of 150 beats/min was maintained by pacing with a rectangular pulse width of 4-ms and voltage twice threshold by using bipolar electrodes sutured to the left atrial appendage.

Heart rate, intravascular pressures, and LAD blood flow were monitored continuously throughout the experiment by a polygraph (model 7; Grass Instrument Co., Quincy, MA, U.S.A.). Radioactive microspheres were injected into the left atrial appendage via an indwelling catheter, and reference blood samples removed from the right femoral artery for determination of myocardial blood flow. Drug was administered via the right femoral vein.

Arterial blood gases (PO₂, PCO₂) and pH were measured via automated analysis (AVL 995; AVL Scientific Corp., Roswell, GA, U.S.A.) and maintained within a normal physiologic range (pH 7.35-7.45; PO₂, 80-120 mm Hg; PCO₂, 25-40 mm Hg) by manipulation of ventilation rate, O₂ mix, and via i.v. administration of 1.5% bicarbonate as needed. Body temperature was maintained with a heating pad at 38 ± 1°C.

Determination of area at risk and regional myocardial blood flow

Ischemic area at risk was determined by intracoronary injection of India ink at the site of occlusion at the end of the experiment. Myocardial blood flow, measured in the normal and ischemic zone, was evaluated by using published methods by a radiomicrosphere technique ⁽²²⁾. In brief, microspheres (15 μm; New England Nuclear, Boston, MA, U.S.A.), labeled with ¹⁴¹Ce or ⁹⁵Nb, suspended in isotonic saline and 0.01% Tween 80 were sonicated for 5 min, vortexed for 5 min, and injected (1 ml; 2-4 × 10⁶ microspheres) via the left atrial catheter followed by a 5-ml saline flush. A reference blood-flow sample was withdrawn from the femoral artery at a rate of 7.1 ml/min beginning immediately before microsphere injection. Regional myocardial blood flow was determined at 2 min into the first occlusion and at the end of 2 h of reperfusion in the ischemic and nonischemic regions.

After excision, hearts were sectioned in a transverse fashion and further sectioned into subepicardium, midmyocardium, and subendocardium in nonischemic and ischemic regions and weighed. All biopsies used were ≥1 cm from the perfusion boundaries, as indicated by Patent blue dye. Samples were evaluated for radioisotopic activity by using a gamma counter (Tracer Analytic 1195; TM Analytic, Elk Grove, IL, U.S.A.). Activity of each isotope also was determined in reference blood samples. Calculations of myocardial blood flow were made via a preprogrammed computer (Apple IIe) to obtain the activity of each isotope in individual tissue samples with tissue blood flow determined from the equation Equation 1 where Q_m is myocardial blood flow (ml/min/g), Q_r is the rate of withdrawal of reference blood samples (7.1 ml/min), C_r is the activity of the reference blood-flow sample (counts/min), and C_m is the activity of the tissue sample (counts/min/g).

Determination of myocardial segment shortening

Piezoelectric crystals were inserted 7-9 mm into the subendocardium of the ischemic and nonischemic regions of the heart. Sound pulses were transformed to electrical signals via an ultrasonic amplifier. Electrical signals, which displayed a frequency proportional to the distance between the crystals, were monitored by an oscilloscope (Soltec model 520; Sun Valley, CA, U.S.A.) and measured as changes in transmission time. Segment length during diastole (DL) and systole (SL) was determined at the onset of isovolumetric contraction (start of positive dP/dt) and at peak negative dP/dt, respectively. Percentage segment shortening (%SS) was calculated as %SS = (DL − SL)/DL × 100. Segment-length data were adjusted to a value of 10.0 assigned for baseline DL.

Test material

Monophosphoryl lipid A, formulated at 300 μg/ml in 10% ethanol, 40% propylene glycol, 60% Water for Injection (Ribi ImmunoChem Research, Inc., Hamilton, MT, U.S.A.) and vehicle control were diluted 1:10 in 5% dextrose in Water for Injection and administered by i.v. bolus over a 3-min period. Glibenclamide was dissolved in a 1:1:1:2 mixture of polyethylene glycol (PEG)/95% ethanol/0.1N sodium hydroxide and 0.9% saline and administered at 50 μg/kg 15 min before inducing repetitive cycles of transient ischemia and reperfusion. Our laboratory previously showed that this dose of GLB produces a reduction of blood glucose concentrations in anesthetized dogs, which indicates effective blockade of pancreatic K_ATP channels ⁽²³⁾.

Experimental protocol

Twenty-four hours before regional ischemia, MLA was administered to randomly assigned dogs at doses of 3, 10, or 35 μg/kg. Vehicle was used to equalize the test material injection volume of the lower doses to that of the high dose level (0.117 ml/kg). Control dogs received an equal volume of vehicle 24 h before ischemia. Dogs were subjected to five 5-min cycles of regional ischemia interspersed with 10-min periods of reperfusion, followed thereafter by 120 min of sustained reperfusion. Regional segment shortening (%SS) and hemodynamics were recorded throughout the experiment. Regional myocardial blood flow was measured during the first occlusion and after 2 h of reperfusion.

To evaluate the potential role of K_ATP-channel activation on the cardioprotective activity of MLA, two additional groups of dogs were administered the K_ATP-channel antagonist GLB ⁽²²⁾. One treatment group received MLA at a dose (10 μg/kg) previously determined in this study to produce a maximal attenuation of stunning in this model. MLA was administered i.v. 24 h before ischemia, and GLB (50 μg/kg) was administered i.v. 15 min before ischemia. A second group of dogs received only i.v. GLB (50 μg/kg) 15 min before ischemia. The dose of GLB used was determined from pilot studies, in which it was shown that this dose did not aggravate stunning (%SS) by using the protocol chosen for these experiments.

Criteria for animal exclusions

Subendocardial collateral flow during ischemia >0.15 ml/min/g, ventricular fibrillation, heart rate >160 beats/min, or heart worms were the preestablished criteria for excluding a dog from the database.

Statistical analysis

A two-way analysis of variance (ANOVA) followed by Fisher's least significant difference (LSD) was used to evaluate treatment and time effects, respectively. A one-way ANOVA followed by Dunnett's t test was used to compare differences within a treatment group. A p value of <0.05 was considered significant. Data are presented as the mean ± SEM.

RESULTS

A total of 48 dogs was entered into the protocol. Three were eliminated because of a high baseline heart rate (one each in MLA, 3- and 10-μg/kg, and GLB, 50-μg/kg, groups), four because of ventricular fibrillation (one each in MLA, 3-, 10-, and 35-μg/kg, and MLA, 35-μg/kg, + GLB, 50-μg/kg, groups) and two because of high collateral blood flow (one each in MLA, 35-μg/kg, and MLA, 35-μg/kg, + GLB, 50-μg/kg, groups). Therefore, a total of 39 dogs successfully completed the protocols; MLA, 3 μg/kg (n = 6); 10 μg/kg (n = 6); 35 μg/kg (n = 6); MLA vehicle (n = 8); GLB, 50 μg/kg (n = 6); MLA, 10 μg/kg, + GLB, 50 μg/kg (n = 7).

Pretreatment with three doses of MLA resulted in no significant changes in systemic hemodynamics (Table 1), blood gases, or blood pH (Table 2) when compared with the MLA vehicle-treated control group, either at baseline or at various points during the brief occlusion periods or after the prolonged reperfusion period. MLA, at various doses, also had no effect on rate-pressure product throughout the experiment (data not shown). Similarly, the addition of GLB had no effect on systemic hemodynamics, blood gases, or blood pH throughout the experiment, with the exception that mean blood pressure tended to be lower during reperfusion in the MLA (10 μg/kg) + GLB group compared with the vehicle control group, and positive LV dP/dt tended to be lower during reperfusion in the GLB group than in the vehicle-treated control group (Tables 1 and 2). Additionally, there were no differences in regional myocardial blood flows in the nonischemic or ischemic regions during the occlusion period or at 2 h of reperfusion between the various treatment groups (Table 3). Thus all groups were subjected to similar intensities of ischemia during the first occlusion period and had similar flows at 2 h of reperfusion (Table 3).

TABLE 1

TABLE 2: Arterial blood gas and pH values

TABLE 3: Regional myocardial blood flow

Pretreatment with MLA 24 h before stunning resulted in a significant improvement in the recovery of %SS at 30, 60, and 120 min of the 2-h reperfusion period in animals given a 10- or 35-μg/kg dose (Fig. 1). MLA at 3 μg/kg also tended to produce an improvement in %SS in the ischemic region throughout reperfusion; however, the cardioprotective effect was statistically significant only at 15 min of reperfusion (Fig. 1). This drug-associated improvement in the recovery of regional segment shortening was rapid, with significant differences as compared with the controls apparent by 15-30 min of reperfusion at the two higher doses. The maximal improvement in %SS was observed at the 10- rather than the 35-μg/kg dose, although the recovery in %SS was not statistically different between the two groups. There were no statistically significant differences in %SS in the nonischemic region between groups throughout the experimental protocol [range, 83 ± 11% to 102 ± 13% (data not shown)].

FIG. 1: Pretreatment (24 h) with monophosphoryl lipid A (MLA) at 3, 10, or 35 μg/kg, i.v., produced an improvement in functional recovery (%SS) in stunned myocardium after five 5-min occlusions of the left anterior descending (LAD) coronary artery interspersed with 10 min of reperfusion and finally followed by 2 h of reperfusion. All values expressed as the mean ± SEM (n = 6−8 per group). *p < 0.05 MLA₁₀ vs. Vehicle, ^†p < 0.05 MLA₃₅ vs, Vehicle, ^‡p < 0.05 MLA₃ vs. Vehicle. Absolute values for %SS at baseline for each group were 18.7 ± 0.6% (control), 20.5 ± 1.2% (MLA, 3), 18.2 ± 1.5% (MLA, 10) and 22.0 ± 0.8% (MLA, 35).

Administration of GLB (50 μg/kg) to control animals had no adverse effect on the recovery of %SS during reperfusion; however, this relatively small dose of the K_ATP antagonist administered 15 min before ischemia completely abolished the protective effect of 10 μg/kg MLA administered 24 h before ischemia (Fig. 2).

FIG. 2: Pretreatment with the adenosine triphosphate-sensitive (K_ATP)-channel antagonist, glibenclamide (GLIB; 50 μg/kg, i.v.) 15 min before stunning completely blocked the cardioprotective effect of monophosphoryl lipid A (MLA; 10 μg/kg). This same dose of gliben-clamide had no effect in vehicle-treated controls. All values expressed as the mean ± SEM (n = 6−8 per group). *p < 0.05 MLA₁₀ vs. Vehicle. Absolute values for %SS at baseline for GLIB alone and MLA + GLIB were 21.8 ± 1.2% and 20.0 ± 1.0%, respectively.

DISCUSSION

It was previously reported that pretreatment with MLA at doses of 10-35 μg/kg ∼24 but not 1 h before regional in situ myocardial ischemia reduces infarct size in the dog ⁽¹²⁾. In rabbits, the cardioprotective effect of MLA against infarction is apparent 6 h after dosing at 35 μg/kg and is optimal between 9 and 36 h after drug administration ⁽²⁴⁾. Clearly, pharmacologic cardioprotection against cardiac ischemia/reperfusion injury elicited by MLA mimics in a temporal fashion the SWOP induced by ischemia ^(8-10,25).

Recently it was shown that SWOP reduces myocardial stunning in a swine model of regional in situ ischemia, which is a point of distinction from the first window of ischemic preconditioning, in which an antistunning effect cannot be observed ^(8,10). Delayed pharmacologic preconditioning with 35 μg/kg of MLA also was previously shown to reduce contractile dysfunction after five cycles of transient regional ischemia in the dog ⁽²¹⁾.

Dose-response studies of the infarct-limiting activity of MLA pretreatment were conducted in the rabbit and dog, and the results suggest that doses of 10-35 μg/kg offer significant cardioprotection when administered 24 h before ischemia ^(14,19). Investigation of the dose-response relation for MLA in the dog-stunning model was conducted so a comparison could be made with the results obtained in the infarct models. The results of the study suggest that 10 μg/kg of MLA is an optimally cardioprotective dose against myocardial stunning in the dog. MLA at 3 μg/kg was clearly less efficacious than at 10 μg/kg. Interestingly, 35 μg/kg of MLA, although clearly cardioprotective, was not superior to 10 μg/kg. Collateral flow during the first 5-min occlusion period tended to be higher in the 10- versus the 35-μg/kg group, although not significantly. It is unclear why contractility appeared to be better in animals pretreated with 10 versus 35 μg/kg MLA, although the trend toward higher collateral flow in the 10-μg/kg group, which was likely a chance occurrence, may have played some role in the result. In any event, these authors are not convinced that MLA truly has a bimodal dose-response curve of protection in this model. The functional recovery of the 10 and 35 μg/kg dose levels, in fact, were not significantly different from each other.

It was previously demonstrated that direct K_ATP-channel agonists such as nicorandil and aprikalim reduce myocardial stunning in the canine model ⁽²⁶⁾. In both dog and rabbit infarct models, the cardioprotective activity of a 24 h MLA pretreatment could be blocked by administration of the K_ATP-channel antagonists, GLB, or 5-hydroxydecanoate ^(14,19,20), as was previously reported for the first window of ischemic preconditioning ⁽²²⁾. These observations, in combination with epicardial action-potential measurements made in the dog during early periods of ischemia ⁽¹⁹⁾, suggested that MLA pretreatment may be enhancing early K_ATP-channel activation on initiation of ischemia, thereby leading to enhanced ischemic tolerance. Consequently, it was obvious to ask the question as to whether activation of the K_ATP channel could play a role in the previously reported antistunning activity of MLA. Administration of GLB just before ischemia, resulting in complete abrogation of the antistunning activity of MLA administered 24 h before ischemia, is reminiscent of observations previously made in infarct models with this drug ^(14,19,20). Therefore activation of K_ATP channels during ischemia may be important in the cardioprotective activity of MLA in both myocardial infarct and stunning models. On the other hand, GLB was shown to have effects independent of its ability to block K_ATP channels ^(27-29); thus it is possible that it may block the effect of MLA by an alternative mechanism not related to K_ATP channels. Glibenclamide has been shown to block Na-K ATPase ⁽²⁷⁾, cardiac chloride channels ⁽²⁸⁾, and the expression of inducible nitric oxide synthase (iNOS) in vivo and in vitro ⁽²⁹⁾. This latter effect may be of particular importance because the iNOS inhibitor, aminoguanidine, was recently shown to block the effect of MLA to reduce infarct size in rabbits ⁽²⁶⁾.

The unanswered question is how MLA induces a state of delayed cardioprotection developing over a number of hours after dosing, which could ultimately result in modulation of K_ATP activation on subsequent ischemic challenge. Data generated recently in the rabbit suggest that induction of calcium-iNOS may play a role in the mechanism of action of this agent ⁽³⁰⁾. Myocardial ischemia (15 or 30 min) appears to lead to enhanced cardiac iNOS activity in MLA-pretreated rabbits, although enzyme levels were similar in nonischemic myocardium of treated and untreated animals ⁽³⁰⁾. Administration of the selective iNOS inhibitor, aminoguanidine, completely blocked the effect of MLA to reduce infarct size in the rabbit ⁽³⁰⁾. Nitric oxide has been suggested to be a potential mediator of delayed preconditioning ^(25,31) through its ability to activate kinases or scavenge free radicals ^(25,32,33). Most recently it was reported that NO may increase open-state probability of cardiac K_ATP channels, as measured by patch-clamp technique ⁽³⁴⁾.

At present, it is hypothesized that MLA leads to activation of iNOS gene transcription in myocytes possibly via a tyrosine kinase-dependent pathway ⁽³⁰⁾, as is reported to occur with the pharmacologically related xenobiotic, LPS ^(35-37). Induction of myocardial iNOS messenger RNA (mRNA), peaking at 6 h after administration of cardioprotective doses of MLA, was observed in the rat ⁽³⁸⁾. This induced iNOS pool may exist in a latent form, which is subsequently activated as an immediate consequence of a second phosphorylation cascade provoked by ischemic challenge ⁽³⁹⁾. Nitric oxide signaling in ischemic myocardium may result, among other things, in activation of K_ATP channels, as originally proposed by Cameron et al. ⁽³⁴⁾, and consequently to a heightened tolerance to ischemia/reperfusion injury.

Incubation of primary rabbit cardiomyocytes with MLA (200 ng/ml) for 4 h, followed by a 20-h incubation in drug-free media, was recently demonstrated to result in enhanced tolerance of myocytes to lethal cell injury on exposure to 2 h of simulated ischemia in media containing deoxy-d-glucose at pH 6.5 ⁽³⁶⁾. Glibenclamide and 5-hydroxydecanoate are able to block this in vitro pre-conditioning of cardiomyocytes by MLA against simulated ischemic injury. One of the conclusions of this study was that the cardiomyocyte has an inherent capacity to be preconditioned by MLA.

At this time, it is unknown whether MLA can protect vascular endothelium during ischemia and reperfusion, although studies in this regard are presently under way. It was recently reported that another compound that shares the lipid A structure with MLA, endotoxin, induces iNOS mRNA in a variety of cell types in the rat after in vivo dosing, including cardiomyocytes, vascular smooth-muscle cells, and to a lesser extent, in vascular endothelium ^(40,41). Therefore it is not unreasonable to consider that MLA may protect vascular endothelium as well as cardiomyocytes.

To summarize, it is proposed that induction of myocardial iNOS (in cardiomyocytes and possibly endothelial cells) occurs over a number of hours after administration of MLA. The iNOS protein induced by MLA may exist in a latent form, which is subsequently activated on ischemic challenge, possibly through rapid phosphorylation, leading to enhanced NO signaling in ischemic myocardium of MLA-pretreated animals. As was previously reported, NO may then rapidly activate K_ATP channels, leading to the enhanced ischemic tolerance of MLA-pretreated animals to infarction and myocardial stunning as manifestations of ischemia-reperfusion injury.

The ability of MLA to reduce infarction, myocardial stunning, and ventricular arrhythmias associated with ischemia/reperfusion injury through a pharmacologic means of inducing delayed myocardial preconditioning may represent an opportunity for the development of a cardioprotective drug for human use. To that end, MLA is undergoing clinical testing for evaluation of its ability to reduce reversible and irreversible myocardial injury after cardiopulmonary bypass in association with coronary artery bypass grafts or aortic valve replacement.

Acknowledgment: We thank Jeannine Moore for excellent technical assistance and Kathy Wallinder for preparation of this manuscript. This work was supported by NIH grant HL 08311 and a grant from Ribi ImmunoChem, Inc.