Preconditioning with Sevoflurane Reduces Changes in Nicotinamide Adenine Dinucleotide during Ischemia–Reperfusion in Isolated Hearts: Reversal by 5-Hydroxydecanoic Acid

The investigation conformed to the Guide for the Care and Use of Laboratory Animals (US National Institutes of Health No. 85–23, revised 1996) and was approved by the institutional animal studies committee. Our methods have been described previously. 3,4,11,12Thirty milligrams of ketamine and 1,000 units of heparin were injected intraperitoneally into 60 albino English short-haired guinea pigs (weight, 250–300 g). Ketamine given in this manner does not inhibit preconditioning, as reported when given before IPC in the isolated rat heart. 13Animals were decapitated 15 min later when unresponsive to noxious stimulation. After thoracotomy, the aorta was cannulated distal to the aortic valve, and the heart was immediately perfused retrograde with 4°C cold oxygenated Krebs-Ringer solution. The inferior and superior venae cavae were ligated, and the heart was rapidly excised. After cannulation of the pulmonary artery to collect the coronary effluent, the heart was placed in the support system and perfused at 55 mmHg at 37°C. The Krebs-Ringer perfusate was equilibrated with ∼97% O²and ∼3% CO²to maintain a constant perfusion pH of 7.4 (pH, 7.40 ± 0.01; carbon dioxide partial pressure, 25 ± 2 mmHg; oxygen partial pressure, 570 ± 10 mmHg). The perfusate was filtered (5-μm pore size) in-line and had the following calculated composition (nonionized): 138 mm Na⁺, 4.5 mm K⁺, 1.2 mm Mg²⁺, 2.5 mm Ca²⁺, 134 mm Cl⁻, 14.5 mm HCO³⁻, 1.2 mm H²PO⁴⁻, 11.5 mm glucose, 2 mm pyruvate, 16 mm mannitol, 0.1 mm probenecid, 0.05 mm EDTA, and 5 U/l insulin.

Left ventricular pressure (LVP) was measured isovolumetrically with a saline-filled latex balloon inserted into the left ventricle through a cut in the left atrium. At the beginning of the experiment the balloon volume was adjusted to achieve a diastolic LVP of 0 mmHg, so that any subsequent increase in diastolic LVP reflected an increase in left ventricular wall stiffness, or diastolic contracture. Characteristic data from LVP were as follows: systolic, diastolic, systolic–diastolic LVP, and the maximal and minimal first derivatives of LVP (dLVP/dt^maxand dLVP/dt^min) as indices of contractility and relaxation, respectively. Spontaneous heart rate was monitored with bipolar electrodes placed in the right atrial and ventricular walls. Coronary flow (CF) was measured at constant temperature and perfusion pressure by an ultrasonic flowmeter (Transonic T106X, Ithaca, NY) placed directly into the aortic inflow line.

Coronary inflow (a) and coronary venous (v) Na⁺, K⁺, Ca²⁺, oxygen partial pressure (Po²), pH, and carbon dioxide partial pressure (Pco²) were measured off-line with an intermittently self-calibrating analyzer system (Radiometer Copenhagen ABL 505, Copenhagen, Denmark). Po²v tension was also measured continuously on-line with an O²Clark type electrode (model 203B; Instech, Plymouth Meeting, PA). Myocardial oxygen consumption (MVo²) was calculated as CF · heart wet weight⁻¹· (Po²a − Po²v) · 24 μl O²/ml at 760 mmHg, and the cardiac efficiency index was calculated as [sys − diaLVP]· HR · MVo²⁻¹.

Sevoflurane (Abbott Laboratories, Chicago, IL) was bubbled into the perfusate using an agent specific vaporizer (Vapor 2000; Dräger Medizintechnik GmbH, Lübeck, Germany) placed in the O²–CO²gas mixture line. Samples of coronary perfusate were collected from a port in the aortic (inflow) and pulmonary cannulas (outflow) to measure sevoflurane concentrations by gas chromatography. Inflow and outflow sevoflurane concentrations were 1.3 ± 0.1 and 1.2 ± 0.1 mm, equivalent to 8.9 ± 0.7 and 8.3 ± 0.7% atmospheres at 37°C, respectively. These concentrations, which are too high to maintain general anesthesia but may be used temporarily during mask induction, 14,15were chosen to unmask any possible effects on NADH concentrations during sevoflurane exposure or during IR. 12 

To potentially block mK^ATPchannels, 5-HD (Sigma Chemical, St. Louis, MO) was given at a concentration of 200 μm. 8If ventricular fibrillation occurred, a bolus of 250 μg lidocaine was immediately injected in the aortic cannula. All data were collected from hearts naturally in, or converted, to sinus rhythm. At the end of 120 min of reperfusion, hearts were removed and ventricles cut into six to seven transverse sections of 3-mm thickness. These were immediately stained with 1% 2,3,5-triphenyltetrazolium chloride in 0.1 m KH²PO⁴buffer (pH 7.4, 38°C) for 5 min. 2,3,5-Triphenyltetrazolium chloride stains viable tissue red, indicating the presence of a formazan precipitate that results from the reduction of 2,3,5-triphenyltetrazolium chloride by dehydrogenase enzymes present in viable tissue. 16All slices were digitally imaged by a photoscanner, and the infarcted areas of each slice were measured in a blinded fashion by planimetry using software (Image 1.62) from the National Institutes of Health (Bethesda, MD). Infarcted areas of individual slices were averaged on the basis of their weight to calculate the total %IS of both ventricles (in percent). Reproducibility of this method is approximately ± 5% based on studies of fresh ex vivo , but nonperfused Langendorff prepared hearts (M. L. Riess, laboratory observations performed at Dr. David Stowe's laboratory at Medical College of Wisconsin).

Tissue autofluorescence is a widely used and accepted method to measure NADH in isolated hearts and myocardial tissue. 10–12,17,18To assess NADH fluorescence, each experiment was conducted in a light-blocking Faraday cage. The distal end of a bifurcated fiberoptic cable (6.8 mm²per bundle) was placed gently against the left ventricular anterior wall. A net was applied around the heart for optimal contact with the fiberoptic tip. This maneuver did not affect LVP. The two proximal ends of the fiberoptic cable were connected to a modified spectrophotometer (Photon Technology International, London, Canada). Fluorescence was excited with light from a xenon arc lamp at 75 W. The light was filtered through a 350-nm monochromator (Delta RAM; Photon Technology International), and the beam was focused onto the in-going fibers of the optic bundle. The arc lamp shutter was opened only for 2.5-s recording intervals to prevent photobleaching; there were 26 recordings for a total exposure of 65 s over the course of each experiment. Fluorescence emissions were collected by fibers of the second limb of the cable. This light was separated by a dichroic beamsplitter at 430 nm and filtered by interference filters (Chroma Technology Corp., Brattleboro, VT) at 405 ± 15 and 460 ± 10 nm. Intensities were measured by photomultipliers (Photomultiplier Detection System 814; Photon Technology International, London, Canada).

Although autofluorescence at 460 nm could also arise from unknown intracellular constituents or cytosolic NADH, the majority is derived from mitochondrial NADH. 17,19Motion artifacts in the NADH fluorescence at 460 nm are diminished by using 405 nm as a second reference wavelength that is less sensitive to changes in NADH; thus, the ratio of the intensities at 460 and at 405 nm is interpreted as a measure of NADH. 18The use of these two tissue light isobestic wavelengths also accounts for possible alterations in myoglobin light absorption, e.g. , by hypoxia. 18Since we did not calibrate NADH fluorescence, we obtained more conservative estimates of changes in NADH. 20NADH is given in arbitrary fluorescence units.

All analog signals were digitized (PowerLab/16 SP; ADInstruments, Castle Hills, Australia) and recorded at 200 Hz (Chart & Scope v3.6.3; ADInstruments) on Power Macintosh® Computer G4 (Apple, Cupertino, CA) for later analysis using MATLAB® (The MathWorks, Natick, MA) and Microsoft Excel® (Microsoft Corporation, Redmond, WA) software. All variables were averaged over the sampling period of 2.5 s.

After stabilization, each experiment lasted 200 min (fig. 1). Hearts were randomly assigned to one of five groups: (1) Untreated ischemic control hearts (ISC, n = 12) were not subjected to preconditioning or given the drug 5-HD. (2) Preconditioned hearts (APC, n = 12) were exposed to sevoflurane for 15 min, followed by a 30-min washout period before ischemia. (3) Other hearts were exposed to sevoflurane and 5-HD (APC+5-HD, n = 12); in this group sevoflurane exposure was bracketed by perfusion with 200 μm 5-HD from 5 min before to 15 min after sevoflurane exposure, followed by 15-min washout before ischemia. (4) Additional hearts were exposed to 5-HD only (5-HD, n = 12) for 35 min, followed by 15-min washout before ischemia. (5) Nonischemia time control hearts (CON) were perfused for 200 min. There were no differences in sevoflurane concentrations between APC and APC+5-HD hearts. Sevoflurane was essentially undetectable (0.05 ± 0.01 mm) in the effluent 30 min after its washout before ischemia. All ischemia hearts then underwent 30 min of global ischemia and 120 min of reperfusion.

All data are expressed as mean ± SEM. For among-group data, ISC, APC, APC+5-HD, 5-HD, and CON were compared by analysis of variance to determine significance (Super ANOVA 1.11® software for Macintosh®; Abacus Concepts, Berkeley, CA) at the following selected time points: before (at 0 and 5 min), during (at 20 min), and after (at 35 and 50 min) sevoflurane exposure; during ischemia (at 55 and 80 min); and during reperfusion (at 85, 140, and 200 min). If F  values (P < 0.05) were significant, post hoc  comparisons of means tests (Student-Newman-Keuls) were used to compare the five groups. Differences among means were considered statistically significant when P < 0.05 (two-tailed). Statistical symbols used were:^aISC, ^bAPC, ^cAPC+5-HD, and ^d5-HD versus  CON; ^eAPC, ^fAPC+5-HD, and ^g5-HD versus  ISC; ^hAPC+5-HD and ⁱ5-HD versus  APC; and ^j5-HD versus  APC+5HD. Regression–correlation analyses were used to determine relations between (1) %IS and the rate of the NADH decline during ischemia (dNADH^I/dt), and (2) %IS and the NADH deviation from baseline (ΔNADH^RP) at 120 min of reperfusion.

Figure 2shows the changes in NADH for each of the four ischemia groups before, during, and after ischemia and for the nonischemia time control group. NADH remained stable for 140 min in the nonischemia CON group and then slowly decreased by about 5% at 200 min. Exposure to 1.3 mm sevoflurane caused a reversible increase in NADH. This was fully blocked when bracketed by 5-HD. For all ischemia groups, NADH increased during initial ischemia; this was followed by a gradual decline toward preischemic concentrations. APC hearts showed a significantly lower initial increase in NADH and a slower decline (P < 0.05) during prolonged ischemia than did CON hearts. Bracketing 5-HD during sevoflurane exposure before ischemia abolished these differences during ischemia. Throughout reperfusion, NADH fluorescence deviated the least from preischemic concentrations in the APC group and was not different from the CON group after 120 min of reperfusion. 5-HD given before ischemia had a similar effect on NADH fluorescence before, during, and after ischemia as the CON group.

Figure 3shows changes in LVP for each of the four ischemia groups before, during, and after ischemia and for the nonischemia time control group. Systolic LVP decreased slightly over the 200-min time course of the experiment, as evidenced in the CON group; diastolic LVP remained constant. Sevoflurane exposure caused a large but completely reversible decrease in systolic LVP. 5-HD before, during, and after sevoflurane exposure did not alter this decrease; 5-HD alone did not alter systolic LVP. Diastolic LVP increased in each ischemic group during the final 5–10 min of ischemia and was lower at 30 min of ischemia in the APC group than in the ISC and APC+5-HD groups. Throughout reperfusion, diastolic LVP was lower in the APC group than in the ISC, APC+5-HD, and 5-HD groups. From 60 min of reperfusion on, systolic LVP was higher in APC hearts compared with ISC, APC+5-HD, and 5-HD hearts. Developed (systolic − diastolic) LVP (data not displayed) was also improved throughout reperfusion in the APC group compared with all other ischemic groups (P < 0.05).

Figure 4shows changes in dLVP/dt^max(contractility) and dLVP/dt^min(relaxation) for each of the four ischemia groups before, during, and after ischemia and for the CON group. dLVP/dt^maxand dLVP/dt^minremained stable in CON hearts over the 200-min experiment. dLVP/dt^maxand dLVP/dt^minwere reversibly depressed during sevoflurane exposure. 5-HD before, during, and after sevoflurane exposure did not alter this decrease; 5-HD alone did not alter dLVP/dt^maxand dLVP/dt^min. During reperfusion, the contraction and relaxation indices were improved only in APC hearts and were not different among the ISC, APC+5-HD, and 5-HD hearts.

Sevoflurane exposure altered the heart rate reversibly from 247 ± 5 beats/min in CON hearts and 241 ± 7 beats/min in ISC hearts to 173 ± 6 in APC and 182 ± 15 beats/min in APC+5-HD hearts, respectively (^bcefijP < 0.05). Heart rate recovered fully on washout of sevoflurane before ischemia. 5-HD alone did not affect the heart rate (250 ± 7 beats/min). At all other perfusion periods, there was no significant difference among the groups (data not displayed). Ventricular fibrillation occurred in each ischemia group. In each incident, ventricular fibrillation was converted immediately to sinus rhythm with lidocaine. There were no differences in the incidence of ventricular fibrillation among the ischemia groups.

Figure 5shows changes in coronary flow for each of the five groups. Coronary flow increased slightly over the 200-min experiment as evidenced in the CON group. Coronary flow reversibly increased during sevoflurane exposure. 5-HD before, during, and after sevoflurane exposure did not alter this increase; 5-HD alone did not affect coronary flow. All ischemia groups exhibited a similar postischemic flow response on early reperfusion. During later reperfusion, coronary flow recovered better in the APC group than in other ischemia groups but did not reach CON values by the end of the experiment.

Myocardial oxygen consumption is displayed in figure 6. It was also reversibly decreased during sevoflurane exposure; 5-HD before, during, and after sevoflurane exposure did not alter this decrease. 5-HD alone had no effect on Mvo². During reperfusion, Mvo²steadily improved in APC hearts and reached CON values by the end of the experiment. In all other ischemia groups, Mvo²remained lower.

The cardiac efficiency index (heart rate ·[systolic − diastolic LVP]· Mvo²⁻¹) (fig. 7) was also reversibly decreased during sevoflurane exposure. 5-HD before, during, and after sevoflurane exposure did not alter this decrease. 5-HD alone had no significant effect on this index. The cardiac efficiency index steadily improved in all ischemic groups during the first 30 min of reperfusion but remained below CON values. The index was not different in CON and APC groups by the end of the experiment.

Figure 8shows that APC resulted in decreased cardiac %IS. This decrease was abolished by bracketing sevoflurane exposure with 5-HD before ischemia. 5-HD alone had no effect on %IS. As shown in figure 9A, there was a significant correlation between %IS and the rate of NADH decline (−dNADH^I/dt in arbitrary fluorescence units per minute) from the peak increase in NADH at 5 min ischemia to the end of ischemia in the 48 ischemic hearts. We also found a significant correlation between %IS and the deviation of NADH (ΔNADH^RPin arbitrary fluorescence units) from baseline values at 120 min of reperfusion in all 60 hearts (fig. 9B). Both of these data sets were better fit by a linear than by a nonlinear relation (form of y = a + bx). See details in figure legend.

Anesthetic preconditioning is well known to attenuate cardiac IR injury, but its underlying mechanisms have not been fully elucidated. In particular, it is unknown how anesthetics trigger preconditioning. This study demonstrates that administration of 5-HD, presumed to be a selective mK^ATPchannel blocker, not only prevented functional, metabolic, and tissue cardioprotection by APC on reperfusion in guinea pig isolated hearts, but also reversed APC-induced changes in mitochondrial NADH before and during IR. The capability to assess changes in NADH continuously as a marker of mitochondrial bioenergetics in the intact heart furnishes additional insight into the mechanism of APC-induced cardioprotection.

Adenosine triphosphate–sensitive potassium channel opening is thought to be an important mechanism in APC as well as IPC. This is based on observations that several putative sarcolemmal K^ATPand mK^ATPchannel antagonists attenuate preconditioning and cardioprotection induced by volatile anesthetics. 2,4–6Because 5-HD abrogated the changes in NADH induced by APC, our results suggest in addition that mK^ATPchannel opening is involved in providing the protective mitochondrial effect. Additional evidence for involvement of the mK^ATPchannel is provided by the putative mK^ATPchannel opener diazoxide, which appears to mimic IPC on %IS 8and to afford protection during ischemia. 21–23However, there is a debate over the role of sarcolemmal K^ATPchannels and if mK^ATPchannels have a greater role in triggering or effecting preconditioning. 9,24 

Although opening mK^ATPchannels may be an effective way to attenuate mitochondrial damage, it remains unclear how mK^ATPchannel opening might protect mitochondrial function. mK^ATPchannel opening and subsequent mitochondrial K⁺uptake has been reported to cause a significant matrix swelling and a small decrease in the negative mitochondrial membrane potential. 25Other data suggest that K⁺influx partially dissipates the mitochondrial membrane potential and, if compensated by an acceleration of electron transfer, leads to a net oxidation in the mitochondria. This was demonstrated by an increase in flavoprotein fluorescence by diazoxide in rabbit ventricular myocytes. 26 

Mitochondrial ATP synthesis at Complex V (F¹F^o-ATP synthetic complex) is driven by the electrochemical proton gradient in the inner mitochondrial membrane that is generated by the transfer of electrons from NADH (and FADH²) to oxygen. 27Complex I (NADH CoQ oxidoreductase) is the first stage of proton and electron donation for NADH. Complex III (ubiquinol:ferricytochrome c oxidoreductase) spans the central part of the respiratory chain, catalyzing the electron transfer from ubiquinol to oxidized cytochrome c, which is also coupled to proton translocation. With adequate substrate and oxygen, this system is in equilibrium, and oxidation of NADH to NAD⁺and phosphorylation of adenosine diphosphate to ATP are matched. NADH and FADH²are products of the tricarboxylic acid cycle and the β-oxidation of fatty acids. 28Dehydrogenases generate NADH from NAD⁺during glycolysis. Steady state NADH concentrations are therefore determined by the balance between NADH generation and oxidation. During ischemia, e.g. , the supply of oxygen to accept electrons diminishes, electron flux through the electron transport chain falters, and NADH accumulates. 11,12,29 

We reported recently 12that either a temporary exposure to a high concentration of sevoflurane or brief ischemia (IPC) reversibly increased NADH in normothermic guinea pig intact hearts but offered no mechanism. The current study offers the first evidence that 5-HD given before, during, and after anesthetic exposure not only prevents the preconditioning stimulus and subsequent cardioprotective effects, but also completely abolishes the anesthetic-induced increase in NADH. If 5-HD is purely a mK^ATPchannel blocker, this would suggest that direct or indirect opening of mK^ATPchannels is involved in the triggering mechanism of APC and that the anesthetic-induced increase in NADH is a consequence of the mK^ATPchannel opening.

Another possibility is that volatile anesthetics directly inhibit complex I 30,31to cause an increase in NADH. A concentration-dependent increase in NADH fluorescence was reported earlier with exposure to other volatile anesthetics in an isolated rat heart model. 32Increased NADH caused by complex I inhibition may cause the formation of reactive oxygen species. Reactive oxygen species in turn have been shown to activate K^ATPchannels. 33,34This could create a feed-forward interaction between NADH accumulation (decreased oxidation) and K^ATPchannel opening that might be inhibited by blocking the mK^ATPchannel.

Direct blockade of complex I activity by volatile anesthetics would also be expected to lead to a compensatory increase in electron transfer via  flavoprotein oxidation at complex II. Indeed, a recent study 35showed that a comparable sevoflurane concentration mildly increased flavoprotein fluorescence (increased oxidation) in isolated myocytes bathed in a substrate-free solution. This would imply then that 5-HD, like diazoxide, 36may not exclusively target the mK^ATPchannel. Indeed, 5-HD may functionally antagonize the inhibition of the respiratory chain by volatile anesthetics at complex I and diazoxide at complex II, as recently suggested by Hanley et al.  375-HD, as a substrate for acyl-CoA synthetase, could provide a bypass for electrons around complex I or II to ubiquinone at complex III. Their study raises a serious concern about the assumed selectivity of diazoxide and 5-HD for the mK^ATPchannel. Our data are also compatible with their theory because 5-HD reversed the sevoflurane-induced increase in NADH. Therefore, any definite conclusions on the role of the mK^ATPchannel using these drugs must be limited at this point. Clearly, a rigorous examination of the coupling of electron transport chain and oxidative phosphorylation during anesthetic exposure will be required from experiments in isolated mitochondria and isolated myocytes, as well as in intact hearts, to more specifically address the triggering mechanisms of APC.

As we described previously, a sudden shortage in cellular oxygen during the onset of ischemia leads to a rapid imbalance between NADH generation and oxidation. 12In the current study, NADH increased and peaked in all groups at 5 min of ischemia. The decline in NADH after 5 min of ischemia may be interpreted as a relatively lower rate of NADH generation than oxidation during prolonged ischemia. APC hearts showed a slower NADH decline; this could reflect a slower remaining rate of oxidative phosphorylation or a higher dehydrogenase activity or larger substrate stores compared with control hearts. Giving 5-HD to bracket exposure to the anesthetic abolished both the lower initial NADH increase as well as the slower NADH decline during prolonged ischemia. This suggests that the lesser changes in NADH during ischemia after sevoflurane exposure are caused by the preconditioning effect of sevoflurane, rather than being merely accompanied by the previous exposure to sevoflurane per se . Similarly, because the same degree of temporary and fully reversible cardiac depression (contractility, coronary flow, Mvo², cardiac efficiency) during sevoflurane exposure was observed in APC and APC+5-HD hearts, this temporary cardiac depression alone cannot be responsible for the subsequent APC. Temporary cardiac depression with other drugs such as pentobarbital or propofol does not necessarily lead to preconditioning. 38 

The overall NADH fluorescence signal measured in this study is a product of average NADH per cell and the number of viable cells contributing to the observed fluorescence. Infarction in our model is concentric and subepicardial. Thus, infarcted cells as well as viable cells underlie the fiberoptic probe on reperfusion. Because there was either incomplete or no recovery in NADH fluorescence over time on reperfusion, and because the magnitude of the deviation correlated with %IS, the observed decrease in tissue fluorescence could be a result of infarcted tissue not contributing to the overall signal rather than to a permanent decrease in NADH per cell. In contrast to the reduced NADH fluorescence in ISC, APC+5-HD, and 5-HD hearts, its relative normalization in APC hearts may be evidence of greater cell salvage.

We used an isolated heart model with a blood-free perfusate. In vivo , volatile anesthetics may modify neutrophil function related to IR injury. IR injury may also be modified by hormonal input or the autonomic nervous system. Many investigators have used other species, particularly the rat, as a model for APC, but the guinea pig heart is believed to be closer in design to the human heart. 3,39–41We used only one pulse of high concentration sevoflurane to induce APC in this study, but we have shown previously that sevoflurane has a concentration-dependent effect to induce APC 12and that APC can be induced with two pulses at lower concentrations of sevoflurane. 3,4Finally, it is important to acknowledge that any conclusion on the role of mK^ATPchannel opening depends on the selectivity of 5-HD for blocking the mK^ATPchannel.

In conclusion, 5-HD, a putative mitochondrial K^ATPchannel blocker, not only abolished protection induced by APC against mechanical dysfunction and tissue necrosis on reperfusion, but also reversed all changes in mitochondrial energetics induced by APC before, during, and after ischemia, as evidenced by the changes in NADH fluorescence. Our study also demonstrates that subsequent %IS can be predicted by changes in NADH during ischemia and by the deviation of NADH fluorescence during reperfusion. The reversible increase of NADH by sevoflurane exposure before ischemia suggests a blockade of mitochondrial electron transport is part of the triggering mechanism of APC and that its reversal by 5-HD is consistent with an mK^ATP-independent action of 5-HD. 37If we assume that the results of our research are applicable to patients at risk for cardiac ischemia during surgery, 42then even brief exposure to a volatile anesthetic may be beneficial in the perioperative period.

The authors thank Ming Tao Jiang, Ph.D., James Heisner, B.S., and Samhita Rhodes, M.S. (all Department of Anesthesiology, Medical College of Wisconsin, Milwaukee, Wisconsin), for their valuable contributions to this study.