Prevention of Isoflurane-induced Preconditioning by 5-Hydroxydecanoate and Gadolinium: Possible Involvement of Mitochondrial Adenosine Triphosphate–sensitive Potassium and Stretch-activated Channels

All experiments performed in this study conformed to the Guiding Principles in the Care and Use of Animals approved by the American Physiologic Society. The Laboratoire de Physiologie Lyon Nord and the investigators received authorization from the French government to perform such animal studies.

A total of 68 New Zealand white rabbits of either sex (2.0–2.5 kg) were premedicated with xylazine (5 mg/kg intramuscularly) and anesthetized with ketamine HCl (50 mg/kg intramuscularly). Before surgical procedures were performed, adequate depth of anesthesia was ensured after determination of the absence of pedal and palpebral reflexes. After tracheostomy, animals were underwent mechanical ventilation (Servo ventilator 900B; Siemens-Elema, Solna, Sweden) with use of 100% oxygen. The tidal volume was set at 15 ml/kg and the respiratory rate at 35 strokes/min. Ventilation was adjusted to keep the end-tidal carbon dioxide in the physiologic range. End-tidal gas concentrations were measured continuously (gas analyzer; Capnomac Ultima, Datex, Helsinki, Finland). Body temperature, recorded by use of a thermistor inserted into the esophagus, was maintained between 39.0 and 40.5°C by means of a servocontrolled heating element incorporated into the operating table. Limb lead II of the electrocardiograph was continually monitored by means of subcutaneous needle electrodes. Anesthesia was maintained by infusion of a mixture of ketamine (3 mg · kg⁻¹· h⁻¹) and xylazine (1.5 mg · kg⁻¹· h⁻¹). Fentanyl 50 μg was injected systematically and intravenously in the animals before thoracotomy to provide adequate analgesia. Systemic blood pressure was monitored by use of a Gould pressure transducer connected to a 1-mm fluid-filled catheter inserted in the right carotid artery. The right internal jugular vein was catheterized with a 1-mm catheter to infuse fluids and drugs. Hetastarch (5 ml · kg⁻¹· h⁻¹) was continually infused via  this intravenous cannula. The heart was exposed via  a left thoracotomy and suspended in a pericardial cradle. The first large marginal branch of the circumflex artery was identified, and a 5/0 Dexon suture was passed around this artery approximately halfway between the apex and the base. The suture ends were threaded through a small vinyl tube to make a snare to perform further coronary occlusion and reperfusion. After the surgical procedure, a 20-min stabilization period was allowed.

In all groups, the coronary artery was occluded for 30 min. Myocardial ischemia was confirmed by the appearance of a regional cyanosis on the epicardium distal to the snare, akinesia or bulging in this area, and a progressive marked ST segment elevation in the electrocardiogram. After 30 min, the snare was released and reperfusion was allowed for a period of 3 h. Reperfusion was visually confirmed by the appearance of hyperaemia. The thread passed around the marginal artery was left in place.

At the end of the reperfusion period, the coronary artery was briefly reoccluded and diluted. Uniperse blue (Ciba-Geigy, Hawthorne, NY) was injected in the jugular vein to delineate the in vivo  area at risk. With this technique, the previously non–area at risk appears blue, whereas the area at risk remains unstained. Anesthetized rabbits were then killed by an intravenous injection of KCl (4 mEq). The heart was excised and cut into five or six 2-mm transverse slices. After removing right ventricular tissue, each slice was weighted and identified. The basal surface of each slice was photographed for further measurement of area at risk. Each slice was then incubated for 15 min in tetrazolium chloride to differentiate infarcted (pale) from viable (red) myocardial area. 9Each slice was then rephotographed. Enlarged projections of these slices were traced for determination of the boundaries of the different areas. Extent of left ventricle area, area at risk, and infarct size were quantified by computerized planimetry and corrected for the weight of tissue slices. Total weights of area at risk and area of necrosis were then calculated and expressed as weight (grams) or as percentages of total left ventricle (LV) weight. 1We decided prospectively that hearts with a risk region less than 10% of the LV weight would be excluded from the study. 5-Hydroxydecanoate (ICN Biomedicals, Aurora, OH) and gadolinium 3+ chloride hexahydrate (molecular weight = 371.7; Sigma Aldrich, Milwaukee, WI) were dissolved in NaCl 0.9% (wt/vol). Isoflurane was purchased from Abbott laboratories (Queensborough, United Kingdom).

The rabbits were randomly divided into six groups (fig. 1). All groups underwent a 30-min coronary artery occlusion and 3 h of reperfusion. Before this prolonged ischemia, they underwent a 40-min treatment period.

After 30 min of stabilization, the isoflurane group received 15 min of inhalational isoflurane (1% end tidal) followed by a 15-min washout period before the 30-min coronary artery occlusion.

To determine whether isoflurane-induced preconditioning is attenuated by the blocking MK^ATPchannel, 5 mg/kg 5-hydroxydecanoate was intravenously injected 10 min before isoflurane administration in the isoflurane–5-hydroxydecanoate group. Similarly, 5 mg/kg 5-hydroxydecanoate was injected intravenously 40 min before coronary occlusion in the 5-hydroxydecanoate group to check the lack of 5-hydroxydecanoate effect on myocardial protection.

To determine whether stretch-activated ion channels are involved in myocardial preconditioning afforded by isoflurane, gadolinium, a blocker of stretch-activated ion channels, was injected. Animals in the isoflurane–gadolinium group received 40 μmol/kg gadolinium intravenously injected just before isoflurane administration. The gadolinium group received 40 μmol/kg gadolinium 30 min before ischemia to determine the lack of effect of gadolinium on infarct size.

For each animal receiving isoflurane inhalation, end-tidal isoflurane concentration was less than 0.1% at the end of the washout period.

Statistical analyses of hemodynamics and end-tidal carbon dioxide were performed using two-way analysis of variance with repeated measures on one factor. LV weight and area at risk were analyzed by analysis of variance. Effect of pretreatment on percent of risk zone infarcted was analyzed by analysis of variance followed by post hoc  Tukey’s test. Differences of infarct zone among groups were evaluated by analysis of covariance and post hoc  Tukey’s test, with infarct zone as the dependent variable and area at risk as the covariant. Statistical calculations were performed using Statistica 5.0 computer software (Stasoft Inc., Tulsa, OK). All values are expressed as mean ± SD. A P  value < 0.05 was significant.

Among the 68 rabbits that have been included into this study, 16 were excluded for the following reasons: fatal ventricular fibrillation during coronary occlusion in one rabbit (isoflurane–5-hydroxydecanoate group), severe heart failure (obvious cardiac dilatation and systolic blood pressure < 40 mmHg) during coronary occlusion or reperfusion in nine rabbits (one control, one isoflurane, one isoflurane–5-hydroxydecanoate, two 5-hydroxydecanoate, three isoflurane–gadolinium, and one gadolinium). One rabbit was excluded because the risk region was smaller than 10% of the LV weight (gadolinium group). Five rabbits were excluded for technical problems during surgical preparation. Data are therefore presented for 52 rabbits: 9 in the control group, 9 in the isoflurane group, 10 in the isoflurane–5-hydroxydecanoate group, 9 in the 5-hydroxydecanoate group, 8 in the isoflurane–gadolinium group, and 7 in the gadolinium group.

Hemodynamic data, including heart rate and systolic blood pressure, are summarized in table 1. There were no significant differences among the six groups, although isoflurane tended to decrease systolic blood pressure during the treatment period (but this reduction did not reach statistical significance). End-tidal carbon dioxide and body temperature did not differ among groups.

The area at risk and infarct size data are presented in table 2and figure 2. LV weight and area at risk were comparable among the different groups, with mean value of area at risk ranging from 27 ± 6 to 35 ± 7% of LV weight. Infarct size in the control group averaged 60 ± 20% of the area at risk. Neither 5-hydroxydecanoate nor gadolinium affected infarct size, which averaged 54 ± 27 and 65 ± 15% of the area at risk in the 5-hydroxydecanoate and gadolinium groups, respectively.

As expected, isoflurane-treated animals developed significantly smaller infarcts than controls, with mean infarct size averaging 26 ± 11% of the area at risk in this group (P < 0.05 vs.  control). In addition, MK^ATPblockade by 5-hydroxydecanoate and stretch-activated channels blockade by gadolinium prevented this infarct size attenuation, because infarct size in these two groups was no longer different from that in the control group. Mean infarct size averaging 68 ± 23 and 56 ± 21% of the area at risk in the isoflurane–5-hydroxydecanoate and isoflurane–gadolinium groups, respectively (P = nonsignificant vs.  control;P < 0.05 vs.  isoflurane).

These data were confirmed when the weight of the infarct size was plotted versus  the weight of the area at risk, its major determinant in the rabbit model of myocardial infarction. As depicted in figure 3, data points for the isoflurane group lie below the control line, indicating that for any size of the risk region, isoflurane-preconditioned hearts developed significantly smaller infarcts. Analysis of covariance confirmed this and showed that for any area at risk, the isoflurane group displayed smaller infarcts. In contrast, data points for isoflurane–5-hydroxydecanoate (fig. 3A), and data points for isoflurane–gadolinium (fig. 3B), lie close to the control line, indicating that 5-hydroxydecanoate or gadolinium preadministration prevented the protecting effect afforded by isoflurane. The regression line of the 5-hydroxydecanoate (fig. 3C) and gadolinium (fig. 3D) groups did not differ from the control group, indicating a lack of effect of these agents per se .

In the current study, we demonstrated that pharmacologic preconditioning by isoflurane is prevented by 5-hydroxydecanoate and gadolinium. Recent studies have demonstrated that isoflurane can protect the ischemic heart. Kersten et al.  10reported that isoflurane can attenuate postischemic contractile dysfunction (i.e. , myocardial stunning) in the canine heart. Cope et al.  11showed in an in vitro  rabbit model that a 5-min preadministration of volatile anesthetic agents (enfurane, halothane, and isoflurane) decreased infarct size. In another in vivo  study, Kersten et al.  4showed that the canine heart can be preconditioned by isoflurane preadministration, a protection that was blocked by the K^ATPantagonist glyburide. Cason et al.  12showed that a 15-min isoflurane administration performed 15 min before a prolonged coronary occlusion significantly reduced infarct size in a rabbit model. Consistent findings have been reported by Ismaeil et al.  13in propofol-anesthetized rabbits. These investigators showed that the beneficial effect of isoflurane-induced preconditioning on infarct size was abolished by glyburide, a K^ATPchannel blocker, and by 8-(p-sulfophenyl)-theophylline, a nonspecific adenosine receptor antagonist. Our results are in close agreement with this: isoflurane was administrated 15 min before the sustained ischemic insult, and despite the fact that the end-tidal isoflurane concentration was almost null at the time of coronary occlusion, isoflurane-treated hearts developed significantly smaller infarcts. In other words, isoflurane can precondition the ischemic heart.

The underlying mechanism of this isoflurane-induced preconditioning remains unclear. Ischemic preconditioning has been initially described by Murry et al.  14and Reimer et al.  15on a canine model. These investigators showed that repetitive brief ischemic episodes before a longer ischemia delayed myocardial infarction and ATP depletion. Ischemic preconditioning involves several sarcolemmal receptors, including adenosine A1, opioids, and bradykinin receptors, intracellular messengers or enzymes such as calcium G protein and protein kinase C, ion channels such as stretch-activated channels, and K^ATPchannels. 16So far, much interest has focused on the sarcolemmal K^ATPchannel as a possible end-effector of preconditioning. Their activation would shorten action potential duration, limit calcium entry into the cell, and then reduce the detrimental energy-consuming process during ischemia–reperfusion. This is mainly supported by the fact that, in most experimental models, preconditioning is blocked by the sulfonylurea glibenclamide, a blocker of sarcolemmal K^ATPchannels. 4,10,17,18 

However, several investigators reported that preconditioning can be observed in the absence of action potential duration shortening. 19,20In addition, the K^ATPopener diazoxide can protect the ischemic heart without altering action potential duration. 21In the whole cell patch-clamp preparation, diazoxide was 50-fold less potent than cromakalim, a K^ATPchannel opener, to activate sarcolemmal K^ATPcurrents, yet both agents displayed a similar cardioprotective effect. 22Both glibenclamide and 5-hydroxydecanoate antagonized the diazoxide effect. Taken together, these findings suggested that activation of the sole sarcolemmal K^ATPchannel is not sufficient to explain the whole cardioprotective effect. In fact, MK^ATPchannels are thought to be involved in the cardioprotection afforded by ischemic preconditioning. 23,24MK^ATPchannels have been discovered by Inoue et al.  25on the inner face of rat liver mitochondrial membrane. Activation of these channels depolarizes the inner membrane and leads to an influx of potassium into the mitochondria, accelerates the mitochondrial respiratory chain, slows the ATP production, releases accumulated calcium, and facilitates mitochondrial swelling. 26Protein kinase C, which is known to be involved in the ischemic preconditioning pathway, can modulate MK^ATPactivation. 7The concomitant measure of sarcolemmal K^ATPcurrent by patch clamp and flavoprotein fluorescence, an indicator of mitochondrial redox state that correlates with mitochondrial depolarization, confirmed this findings: Liu et al.  6demonstrated that diazoxide can produce a reversible activation of flavoprotein without generating any surface K^ATPcurrent. In contrast, pinacidil, another K^ATPchannel opener, induced a mitochondrial oxidation associated with a sarcolemmal K^ATPcurrent. 5-Hydroxydecanoate specifically blocked flavoprotein oxidation triggered by pinacidil without effect on surface current, indicating a specific effect on MK^ATPchannels. 7In contrast, glibenclamide acts as a nonspecific blocker, with inhibitory activity on both types of channels. 5-Hydroxydecanoate, which has shown to be a selective blocker of MK^ATPin in vitro  experiments, blocks ischemic preconditioning in different species, such as the rabbit, 27dog, 28and rat. 29All of these studies suggest that MK^ATPchannels might be the end-effector of ischemic preconditioning. Other pharmacologic triggers of preconditioning, such as opioids, are sensitive to 5-hydroxydecanoate, and likely act through MK^ATPactivation. 30In the current study, 5-hydroxydecanoate clearly blocked isoflurane-induced infarct size reduction, suggesting that this protection is, in part, mediated by MK^ATP. Anesthetic-induced preconditioning seems to share common mechanisms with ischemic preconditioning, both depending on adenosine receptors 18and protein kinase C, 11,12and both being inhibited by 5-hydroxydecanoate, suggesting a specific role for MK^ATP.

It has been demonstrated that myocardial stretch, induced by acute volume overload, can precondition dog 1and rabbit hearts. 31These channels may also play a role in ischemic preconditioning (i.e. , in the absence of volume overload) because of their probable activation during possible distension of dysfunctional myocardium during preconditioning ischemia. 1This protective effect is triggered by sarcolemmal stretch-activated ion channels, and is also mediated by K^ATPchannels and protein kinase C. 31Two pore-domain-potassium channels have been recently described 32and individualized as TWIK-1, TREK-1, and TASK channels. Using a patch-clamp preparation, Patel et al.  demonstrated that the two-pore-domain potassium channels TREK-1 and TASK can be activated both by mechanical stimulation (or stretch) 33and by pharmacologic challenge by isoflurane and halothane. 5These channels have been described in different tissues, including the heart. 34,35In the current study, we found that the Lanthanide gadolinium, a potent blocker of mechanogated channels, 36can attenuate isoflurane-induced preconditioning. This strongly suggests that gadolinium-sensitive channels, such as mechanogated channels, are involved in isoflurane-induced protection. However, the link between MK^ATPand stretch-activated channels remains unclear and was not specifically studied in the current work.

Caution must be exercised before relating this experimental work to a clinical situation. All groups of animals received the same amount of ketamine and xylazine; however, we cannot exclude an interaction between intravenous anesthetics and isoflurane. The model we used is a classical one that has been validated in many studies. The rabbit model has two major advantages: (1) rabbits have a minimal collateral coronary flow degree, 37diminishing confounding factors affecting the extension of infarction; and (2) the rabbit model is widely used in experiments on preconditioning, validating the pathophysiology knowledge of this phenomenon. 1,11,12,13,27The end point of the current study is the percentage of infarct size expressed as percentage of risk area and not the ventricular functional improvement. It is justified because preconditioning delays the occurrence of infarction and likely has no effect on myocardial function in stunning models. 38Most of the previous studies used glibenclamide to assess the role of K^ATPchannels. 4,10,18,39However, glibenclamide is not only specific of sarcolemmal K^ATPand MK^ATPchannels, because it has been shown to inhibit other channels, such as Cl⁻channels. 40In the current study, we used 5 mg/kg 5-hydroxydecanoate and 40 μmol/kg gadolinium to block MK^ATPand stretch-activated channels, respectively. Theses doses are commonly used in similar myocardial protection models using 5-hydroxydecanoate 3,27,29or gadolinium. 1,31We cannot ascertain that the observed effects are fully related to a specific blockade of these channels. However, the doses we injected and the timing of injection are those usually used in similar in vivo  rabbit models to antagonize these channels. 27,31The action of 5-hydroxydecanoate on MK^ATPhas been shown in vitro  for a 100-μmol/l concentration; in our study we did not measure 5-hydroxydecanoate concentration. Gadolinium blocks in vitro  mechanogated channels in a concentration ranging from 1 to 100 μmol/l. However, it can block other channels (for review see reference 36), such as calcium L- and T-type voltagegated channels 41and sodium channels, and inhibit the delayed rectifier K⁺current IKs. 42Therefore, caution must be exercised before concluding there is a direct effect of gadolinium on stretch-activated channels.

In addition, we cannot rule out the possibility that gadolinium also blocked MK^ATPchannels. Sarcolemmal K^ATPcan be activated by membrane deformation. 43Although there is no report that MK^ATPchannels are mechanosensitive, or even blocked by gadolinium, there is some evidence that they are involved in the regulation of mitochondrial volume. 26 

Although there was no statistical difference in hemodynamic parameters, we cannot strictly rule out the effect of isoflurane-induced hypotension on the decrease of infarct size in the groups receiving isoflurane. Similar experiments 44using a different halogenated agent, such as sevoflurane, showed that preadministration of this agent significantly decreased blood pressure (mean blood pressure was 97 ± 3 mmHg in controls vs.  65 ± 4 mmHg in the sevoflurane-preconditioned group) without altering infarct size (23 ± 1% of the area at risk in controls vs.  21 ± 4% in the sevoflurane-preconditioned group). Kersten et al.  4suggested that isoflurane preconditioned the heart by a direct effect on K^ATPchannels. Using a similar in vivo  rabbit model, Cope et al.  11examined the effect of hypotension on halogenated cardioprotection. They showed that there was no correlation between infarction area and decrease in blood pressure. In an in vitro  Langendorff model, the investigators showed that halogenated agents exerted cardioprotection without alteration in perfusion pressure.

In conclusion, the current study demonstrated that preadministration of isoflurane preconditions the rabbit heart. 5-Hydroxydecanoate, a specific blocker of MK^ATPchannels, and gadolinium, a potent inhibitor of mechanogated channels, abolished this pharmacologic anesthetic preconditioning. These data indicate that MK^ATPchannels are probably involved in isoflurane ischemic-like preconditioning, and that mechanogated channels play possibly a role in this phenomenon. Further studies are needed to determine the link between MK^ATPand stretch-activated channels in this protective phenomenon.

The authors thank Florence Arnal, laboratory technician, for her contributive work and organization, and A. J. Patel and Y. Tourneur for their invaluable help.