Isoflurane, but not Halothane, Induces Protection of Human Myocardium via  Adenosine A¹Receptors and Adenosine Triphosphate–sensitive Potassium Channels

Human right atrial appendages were acquired from patients undergoing coronary artery bypass grafting surgery before bypass in accordance with guidelines of the University Human Investigation Committee. Patients taking oral hypoglycemics were excluded from the study. Appendages were transported from the operating room to the laboratory in oxygenated solution containing 130 mm NaCl, 5 mm Na²SO⁴, 6 mm KCl, 5 mm MgCl², 2 mm NaH²PO⁴, 20 mm glucose, and 5 mm HEPES, and were oxygenated for a further 30 min. Trabeculae were carefully dissected from the appendages and, using 2-mm circular spring clips, were vertically suspended from a force transducer (Model 400; Cambridge Technology, Watertown, MA) in an organ bath maintained at 32°C. This lower temperature is used to prevent automaticity and facilitate pacing. The bath was filled with a Krebs-Henseleit buffer consisting of 118.5 mm NaCl, 4.8 mm KCl, 24.8 mm NaHCO³, 1.2 mm KH²PO⁴, 1.44 mm MgSO⁴, 2.5 mm CaCl², 10 mm glucose, and 10 mm pyruvate, pH 7.4, and bubbled constantly with 95% O²–5% CO². Muscles were equilibrated for 10 min before being field stimulated to contract at 0.1 Hz (Grass Stimulator, model S88; Grass Instruments, Watertown, MA) for another 20 min. Muscles were then stretched to the minimum length to produce maximum isometric contractions (L^max) and stimulated at 1 Hz until a steady force of contraction was maintained for at least 20 min. The resting tension on muscle and stimulation voltage remained constant for the duration of the experiment. The buffer was changed every 15 min before and during the experiments. Muscles that failed to generate 1 mN of force were not used in the experiments.

After stabilization of contractile performance, muscles were subjected to a 5-min treatment interval (fig. 1). During this treatment interval, muscles were alternatively pretreated with 100 nm N⁶-cyclohexyladenosine (CHA;fig. 1C), an adenosine A¹agonist, or with anoxic preconditioning (APc) by bubbling the organ bath with 95% N²–5% CO²instead of 95% O²–5% CO²(fig. 1B) to simulate ischemia. These pretreatments were also repeated in the presence of 0.3 μm glibenclamide, a K^ATPchannel antagonist, or 10 nm (with APc) or 100 nm (with CHA) 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), an A¹antagonist, to determine the role of K^ATPchannels and A¹receptors, respectively. These concentrations of drugs were similar to those used in previous in vitro  studies. 14,21,22DPCPX and glibenclamide treatments were also administered alone as controls.

In anesthetic treatment groups, 1% halothane or 1.2% isoflurane was delivered to the organ chamber for 5 min either in 95% O²–5% CO²(fig. 1C) to determine the effect of prior exposure to anesthetic alone, or in 95% N²–5% CO²(fig. 1B) to determine interaction of anesthetics with preconditioning anoxia. The specific carrier gas was passed through a calibrated anesthetic vaporizer to produce the stated vapor concentration. The vapor concentrations used resulted in aqueous concentrations of 0.23 mm halothane and 0.26 mm isoflurane, which approximates the aqueous concentration at 37°C achieved by 1 minimum alveolar concentration of halothane (0.75%) or isoflurane (1.3%). Anesthetic treatments were also performed in the presence of 0.3 μm glibenclamide and 10 nm DPCPX (fig. 1C). Muscles that served as time control or those subjected only to the subsequent anoxic insult alone (nonpreconditioned) received no treatment during this initial 5-min interval (fig. 1A). At the end of the treatment (time 5 min) and after a washout, a 15-min interval followed during which no treatments were administered. Next, muscles in all groups (except control) were made anoxic by replacing 95% O²–5% CO²with 95% N²–5% CO²in the buffer for 30 min, followed by a 60-min oxygenated recovery period.

The trabeculae that were used in these experiments were 4–5 mm in length with a 1-mm²cross-sectional area. The average weight of trabeculae (n = 132) used was 3.11 ± 2.8 mg. Pretreatment force generated in muscles ranged from 1 to 17 mN and averaged 7.35 ± 8.8 mN (table 1). Actual baseline values for developed force did not differ significantly among treatment groups. Some experiments were excluded from the study because of insufficient recovery after treatment. Nine studies were excluded from the APc group, two from the isoflurane in the presence of DPCPX group, and one from the glibenclamide control group.

Atrial myocytes were isolated from right atrial appendages using a procedure based on that of Nánási et al.  22Trabeculae minced into 1-mm³pieces were allowed to equilibrate for 40 min in a low Ca²⁺buffer containing 130 mm NaCl, 0.3 mm CaCl², 5 mm Na²SO⁴, 6 mm KCl, 5 mm MgCl², 2 mm NaH²PO⁴, 5 mm HEPES, and 20 mm glucose adjusted to pH 7.3 with NaOH. The tissue was then digested for 25–30 min with 1 mg/ml collagenase (Type I; Sigma, St. Louis, MO) and 0.15 mg/ml pronase (Type XIV; Sigma). Initial digestion was followed by a 1–2-min wash in calcium-free solution with 0.25 mg/ml aprotinin (Sigma) adjusted to pH 7.4 with NaOH. A second enzyme solution containing 1 mg/ml albumin, 5 mg/ml Na²ATP, 0.5 mg/ml creatine, 0.5 mg/ml pyruvate, and 0.5 mg/ml β-hydroxybutyrate with 2 mg/ml collagenase and 0.1 mg/ml elastase (Type III; Sigma) was superfused for 3–12 min. The tissue fragments were gently triturated and filtered through a 250-μm nylon mesh. The myocytes were stored at 4°C in medium (20 mm KCl, 10 mm glucose, 70 mm glutamate, 10 mm β-hydroxybutyrate, 10 mm taurine, 10 mm EGTA, and 1% albumin) for 1 h before recordings were made. Throughout the isolation procedure, all solutions were gassed with 100% O²and maintained at 37°C.

The myocyte suspension was placed in a chamber mounted on an inverted microscope and allowed to settle to the bottom. Quiescent rod-shaped myocytes with clear striations were whole cell voltage clamped at room temperature (22–25°C) in external solution consisting of 140 mm NaCl, 5 mm KCl, 1 mm MgCl², 1 mm CaCl², and 10 mm HEPES, pH 7.4. Microelectrodes were pulled (Model PB-7; Narishige Instruments, Tokyo, Japan) from glass pipettes and polished with a microforge (CPM2; ALA Scientific, New York, NY). Pipette resistances were typically 3–4 MΩ. Pipette internal solution consisted of 140 mm KCl, 5 mm EGTA, 5 mm HEPES, 10 mm KOH, 1 mm MgCl², 0.1 mm Na²ATP, and 0.1 mm NaADP. Currents were evoked by 10-ms voltage steps from −100 to 30 mV from a holding potential of −80 mV. Halothane or isoflurane was bubbled into the extracellular solution via  air and superfused over the cells in the absence or presence of 50 μm 2,4-dinitrophenol (DNP) to enhance K^ATPcurrents. Halothane administered at vapor concentrations of 0.5%, 1.0%, 1.5%, 2.0%, and 2.5% resulted in aqueous concentrations in the cell chamber of 0.23, 0.51, 0.68, 1.0, and 1.1 mm, respectively. Isoflurane administered at 1.0%, 2.0%, and 3.0% resulted in aqueous concentrations of 0.47, 0.96, and 1.25 mm, respectively. Currents were recorded with pCLAMP 5.0 (Axon Instruments, Foster City, CA) and analyzed using PCS analysis software developed in our laboratory.

Glibenclamide (Sigma) was made as a stock solution in 1:1:1 ethanol, polyethyl glycol, and 1 N sodium hydroxide. DPCPX (RBI, Natick, MA) was made as a stock solution in 95% ethyl alcohol, and CHA (RBI) was made as a stock solution in distilled water.

All data are expressed as mean ± SD percent of the percent pretreatment (time 0 min) force, except as specified in table 1. Anesthetic and preconditioning treatment groups were compared between groups using a one-way analysis of variance and Student-Neuman-Keuls post hoc  analysis. P < 0.05 was considered statistically significant.

All data are expressed in the text as a percent of the force of contraction at the initiation (time 0 min) of the experimental protocol. The time course of the changes in force of contraction for APc and anesthetic experiments is shown in figure 2. Contractile force of untreated trabeculae showed a modest decrease during the 110 min of these experiments (fig. 2A) to 82.3 ± 5.0% of the time-0 value. Untreated muscles made anoxic for 30 min and reoxygenated for 60 min contracted with a force of only 30.4 ± 8.3% of pretreatment at the conclusion of the experiment. Pretreatment of the muscles with a single 5-min anoxic interval reduced contractile force to less than 25% of pretreatment control but resulted in postanoxic recovery of contractile force to 71.6 ± 5.5% of pretreatment (time 0 min) after 60 min of reperfusion (P < 0.05 vs.  recovery of nonpreconditioned muscles). This protocol was used throughout the remainder of the experiments to determine the relative effects of combining treatments with APc.

The effects of the volatile anesthetics on contractile force were evaluated at the end of the 5-min treatment interval (fig. 3) and again 60 min postanoxia (110 min;fig. 4) in both untreated and pretreated muscles. When administered in the absence of oxygen, neither anesthetic altered the anoxia-induced decrease in contractile force over the 5-min treatment interval. In the presence of oxygen, 1.2% isoflurane produced no significant change in contractile force. In contrast, delivery of 1% halothane via  oxygen decreased force of contraction to 48.3 ± 13.6% pretreatment after 5 min. The postanoxic recovery of muscles in these treatment groups also differed. Muscles pretreated with 1.2% isoflurane alone recovered 76.6 ± 7.5% pretreatment control force postanoxia (110 min;fig. 4), not different from APc muscles. When administered concurrently with APc, 1.2% isoflurane did not alter the degree of cardioprotection provided by APc as assessed after 60-min recovery (fig. 4). In muscles pretreated with 1% halothane alone, force of contraction at the conclusion of the experiments was only 39.6 ± 14.9% pretreatment control. Furthermore, recovery of muscles that were anoxically pretreated in the presence of halothane was significantly decreased compared with APc muscles (P < 0.05). Therefore, although 5-min pretreatment with halothane directly decreased force of contraction in isolated muscles (fig. 3), it inhibited the ability of anoxia to precondition myocardium and provide protection after 60-min recovery. In contrast, 1.2% isoflurane had cardioprotective activity similar to that of APc.

Pretreatment with the A¹receptor agonist CHA resulted in improved postanoxic recovery (110 min) similar to APc (74.6 ± 8.4%). Postanoxic recovery in APc and CHA-pretreated muscles was similarly inhibited by DPCPX to 39.5 ± 9.6% and 36.9 ± 8.6% of pretreatment control, respectively. These findings corroborate previous studies that suggested a role of A¹receptors in physiologic preconditioning. 22When given concurrently with isoflurane pretreatment, 10 nm DPCPX inhibited the postanoxic recovery to 52.5 ± 6.2% of pretreatment control, comparable to that seen in APc muscles. Pretreatment with 0.3 μm glibenclamide to block activation of K^ATPchannels also inhibited postanoxic recovery in CHA and APc-pretreated muscles (fig. 4), as has been previously demonstrated. This dose of glibenclamide also produced a decrease of postanoxic recovery of force in 1.2% isoflurane-pretreated muscles (P < 0.05;figs. 2 and 4). Recovery of halothane-treated muscles was not significantly altered by the inclusion of glibenclamide. Pretreatment with these antagonists alone (glibenclamide, DPCPX;figs. 2 and 4) did not alter the postischemic recovery of contractile force as compared with nonpreconditioned muscles.

Whole cell currents ranged from −198.5 ± 44.5 to 487.5 ± 81.1 pA at voltages from −100 to 30 mV. To determine if either volatile anesthetic was directly activating or inhibiting K^ATPchannel currents, anesthetics were applied to myocytes in the absence of channel activators or in the presence of DNP. In the absence of DNP-enhanced current, the baseline K⁺conductance was not enhanced by either halothane or isoflurane (n = 6, data not shown). Application of 50 μm DNP augmented whole cell outward currents by 65.4 ± 23.6% (n = 11), with negligible effect on inward current. DNP-induced currents recorded at voltages from −100 to 30 mV ranged from −400 to 1,500 pA. Halothane (1%) decreased inward and outward currents stimulated by DNP at −100 and 30 mV by 29.8 ± 4.5% and 31.9 ± 6.3%, respectively. The results from a representative experiment are shown in figure 5. Increasing concentrations of halothane produced slightly greater decreases in outward current but no further reduction of inward current (table 2). Halothane maximally reduced inward current at a concentration of 1%, whereas 2.5% halothane produced the greatest reduction of outward current observed in these experiments. Isoflurane failed to produce any significant changes in either inward or outward K^ATPcurrent (I^KATP) when administered in concentrations as high as 3% (table 2). These currents were susceptible to 50–65% inhibition by glibenclamide (data not shown). The high degree of variability observed in the isoflurane-treated groups is insignificant and probably resultant of variations in the whole cell current. This variability was not apparent in halothane-treated cells because of the substantial and consistent effect of halothane treatment.

The objectives of this study were to verify studies suggesting protective roles of adenosine receptors and cardiac K^ATPchannels using this human model of preconditioning, and to determine if volatile anesthetics could pharmacologically imitate or modulate the physiologic effects of APc. These findings support previous reports suggesting that preconditioning in human myocardium involves adenosine receptor activation and K^ATPchannel activation 14,23as well as providing evidence toward a mechanism for anesthetic effects in ischemic myocardium.

The anesthetics studied were isoflurane, which represents the most widely used volatile anesthetic despite prior concerns regarding its potential for causing ischemia, 3–5and halothane, a historically widely applied anesthetic with somewhat greater ability to depress myocardial contractility. 24,25At clinically relevant concentrations, isoflurane protected the human myocardium in this model against a protracted 30-min anoxia to a similar extent as did hypoxic preconditioning. This effect of isoflurane seems to be mediated through A¹receptors and the K^ATPchannel, as the A¹antagonist DPCPX and the K^ATPchannel antagonist glibenclamide, respectively, inhibit it. Halothane, at clinically relevant concentrations, provided no benefit to anoxic myocardium and seemed to inhibit the cardioprotection provided by hypoxic preconditioning. In isolated myocytes, halothane inhibited induced K^ATPcurrents, but isoflurane produced no effect.

A human model of simulated ischemia was used to examine the interactions between anesthetics and preconditioning, which have been somewhat characterized in several other species, 15,16but it remained to be determined whether the effects persisted in the human model. Although experiments performed in a human model have the benefit of more relevant clinical extrapolations, there are a few limitations. For example, the human model possesses a high degree of variability because of possible drug interactions and disease states. Therefore, appendages were not used if obtained from patients currently on an oral hypoglycemic regimen in an attempt to preclude from the study samples with K^ATPchannels altered by prior pharmacologic interventions. Another consideration was the use of volatile anesthetic in the surgical procedure during which the atrial specimen is removed, because this study attempted to demonstrate anesthetic-specific effects on myocardial protection. Because of its widespread use as a supplementary anesthetic during cardiac surgery at our institution, patients were frequently exposed to isoflurane (0.5–1.0%), which could theoretically precondition the sample. However, studies were initiated at least 90 min (and anoxic insult performed 110–140 min) after removal of the atrium, times that extend beyond the early window of preconditioning. 8Furthermore, comparison is made with control experiments that were performed in parallel. Nevertheless, our experimental results may be superimposed on some small residual effects. The contribution of endothelial function to a model of protection must also be examined, because myocardial protection mediated by IPc has recently been shown to include not only the myocytes themselves, but also to extend to endothelium. 26Obviously, protection of the endothelium may be critical to preventing subsequent alterations in perfusion due to disruptions in endothelial regulation of vascular tone. However, anesthetic effects in this model of superfused muscle segments is unlikely involve a major contribution from endothelium. Another issue that must be considered is the temperature at which each experiment was performed. Temperatures below physiologic 37°C have been found to decrease channel responsiveness to potassium channel openers, 27,28which suggests an alteration in channel activity. This effect could theoretically alter the anesthetic responses in isolated muscles contracting at 32°C and patch-clamped myocytes at 22–25°C. No studies to date specifically address temperature effects on K^ATPchannel function. Conducting these experiments at the stated temperatures allows comparison with other studies that have typically been performed at these somewhat subphysiologic temperatures.

Finally, inherent to models of preconditioning is a variability in the methods used to simulate ischemia. In our experiments, ischemia was simulated by anoxia, which was achieved by substitution of nitrogen for oxygen. Although many studies use a duration of anoxia up to 90 min, 30 min has been used in other human models. 29This anoxic period conceivably may result in a reversible cardiodepression without cell death (“myocardial stunning”), which has been found by some to be resistant to protection from IPc. 30However, in this study, the cardiodepression by 30-min anoxia was lessened by prior anoxic conditioning. The reason for this discrepancy could be explained by the fact that the current study uses restoration of contractile function, not cell viability, as the measure of postischemic recovery. Nonetheless, because this study focuses on contractile performance, a hypoxic insult sufficient to prevent functioning is adequate for the determination of cardioprotection.

The possible involvement of adenosine A¹receptors and K^ATPchannels as mediators of physiologic preconditioning has been suggested by studies in several animal models, including human ones. 14,21,31It has been suggested that adenosine receptors may be activated by preconditioning and, in turn, activate K^ATPchannels, possibly through an intermediate such as PKC. Evidence in support of this theory include: (1) adenosine-stimulated preconditioning of human atrial trabeculae can be partially blocked by glibenclamide 23; (2) IPc of rabbit myocardium enhances translocation of PKC isoforms η and ε7; (3) preconditioning of human atrial trabeculae with a K^ATPchannel opener is inhibited by a PKC antagonist 14; and (4) K^ATPcurrents evoked in rabbit myocytes by metabolic inhibition are increased in the combined presence of PKC activation and adenosine. 32The current study confirms that adenosine receptors and K^ATPchannels may sequentially mediate physiologic preconditioning as well as pharmacologic protection provided by anesthetics. Several possible mechanisms by which K^ATPchannel activation results in cardioprotection against an ischemic insult have been proposed, including a shortening of action potential duration. However, this has been refuted by findings demonstrating protection by potassium channel openers in the absence of action potential effects, 33suggesting that activation of sarcolemmal K^ATPchannels may not mediate cardioprotection. Recent studies have described a K^ATPchannel on the inner membrane of the mitochondria, which is selectively activated by diazoxide and inhibited by 5-hydroxydecanoate. 34Studies using these agents in isolated rat and rabbit hearts suggest mitochondrial K^ATPchannels can mediate physiologic preconditioning. 35,36Our results do not provide any information that would distinguish between a sarcolemmal and mitochondrial effect.

The current study confirms findings in animal models demonstrating a cardioprotective effect of isoflurane anesthesia that is also mediated through adenosine receptors and K^ATPchannels. 16,19In a porcine model, coronary vasodilation induced by isoflurane can be inhibited by K^ATPchannel inhibition with glibenclamide. 37Previous investigators have suggested that this vasodilation by isoflurane may induce coronary “steal” and thereby induce ischemic changes. 3–5However, subsequent studies suggest if coronary perfusion pressure is adequately maintained there is no evidence of such steal. 38This study now documents that in human myocardium, isoflurane can play a direct protective role, as has also been demonstrated in animal models. 16,39,40It is also noteworthy that the protective effect of isoflurane persists beyond its withdrawal for at least 15 min, similar to the “memory” effects previously reported experimentally 41and clinically. 17 

To determine whether isoflurane exerts a direct effect on K^ATPchannels, we applied isoflurane to voltage-clamped human atrial myocytes. In doses as high as 3% (roughly 3 minimum alveolar concentration), glibenclamide-sensitive K^ATPcurrents were not affected, either positively or negatively, by isoflurane. These findings suggest that isoflurane does not have a direct effect on channel activation. Because the activator used here, DNP, activates K^ATPcurrents by decreasing intracellular ATP as well as through a direct channel effect, 42it is possible that protective activity of isoflurane is exerted at a regulatory site on the channel that is circumvented by the direct action of DNP. For example, in rabbit ventricular myocytes, isoflurane has been shown to decrease channel sensitivity to ATP. 18Another possibility is that the site of protective action of isoflurane is an upstream intermediate, such as an adenosine receptor. The latter theory is supported by the demonstrated inhibition of isoflurane effects by 10 nm DPCPX. A similar study by Kersten et al.  40found that 0.8 mg/kg of DPCPX in dogs partially inhibits the cardioprotective activity of isoflurane. If this dose produced the maximal effect of A¹antagonism, as suggested by the investigators, A¹receptors may be only one mediator of the isoflurane effect on ischemic recovery, and partial mediation by A3 receptors remains a possibility.

The finding that halothane pretreatment diminishes the protective effects of APc contradicts previous studies that have demonstrated a cardioprotection by halothane. 43The discrepancy may be a result of the difference in experimental species and illustrates the rationale for conducting these experiments in a human model. This study also supports the hypothesis of a possible interaction of halothane with K^ATPchannels by two observations: (1) halothane blunts the cardioprotection provided by APc in isolated muscles, which is strongly evidenced to be mediated through K^ATPchannels; and (2) K^ATPcurrents in voltage-clamped isolated human atrial myocytes were blocked by approximately 30% in the presence of halothane. These observations strongly suggest that halothane affects APc through the inhibition of I^KATP, although it would require that its effects extend beyond its immediate period of application. Because inhibition of K^ATPcurrent by halothane was not as extensive as that seen with glibenclamide, this less-than-complete inhibition may explain why 1% halothane caused only a partial interference with APc.

In conclusion, our findings demonstrate that the adenosine receptor activation of the K^ATPchannel, which mediates IPc in human atrium, contributes to the cardioprotective effects of the volatile anesthetic isoflurane. Such a protective effect, which can be elicited even when isoflurane is briefly applied clinically in cardioplegic solution, 17could contribute to the relatively low incidence of myocardial ischemia observed intraoperatively. Disappearance of protection in the immediate postoperative period might be anticipated as the preconditioning effect declines. 42Furthermore, we show this property is not present among all volatile anesthetics because halothane did not protect and even seemed to attenuate protection by APc.

The authors thank Drs. Scott Arnold, Michael Butler, Arthur Coffey, Richard Gitter, John Kern, Irving Kron, David Robiciewski, William Spotnitz, and Curtis Tribble of the Division of Cardiothoracic Surgery for their help in obtaining the specimens used in this investigation.