Mitochondrial depolarization and asystole in the globally ischemic rabbit heart: coordinated response to interventions affecting energy balance

Transcription

1 Am J Physiol Heart Circ Physiol 308: H485 H499, First published December 31, 2014; doi: /ajpheart Mitochondrial depolarization and asystole in the globally ischemic rabbit heart: coordinated response to interventions affecting energy balance Paul W. Venable,* Katie J. Sciuto,* Mark Warren, Tyson G. Taylor, Vivek Garg, Junko Shibayama, and Alexey V. Zaitsev Nora Eccles Harrison Cardiovascular Research and Training Institute, University of Utah, Salt Lake City, Utah Submitted 21 April 2014; accepted in final form 18 December 2014 Venable PW, Sciuto KJ, Warren M, Taylor TG, Garg V, Shibayama J, Zaitsev AV. Mitochondrial depolarization and asystole in the globally ischemic rabbit heart: coordinated response to interventions affecting energy balance. Am J Physiol Heart Circ Physiol 308: H485 H499, First published December 31, 2014; doi: /ajpheart Mitochondrial membrane potential ( m ) depolarization has been implicated in the loss of excitability (asystole) during global ischemia, which is relevant for the success of defibrillation and resuscitation after cardiac arrest. However, the relationship between m depolarization and asystole during no-flow ischemia remains unknown. We applied spatial Fourier analysis to confocally recorded fluorescence emitted by m -sensitive dye tetramethylrhodamine methyl ester. The time of ischemic m depolarization (t mito_depol) was defined as the time of 50% decrease in the magnitude of spectral peaks reflecting m. The time of asystole (t asys ) was determined as the time when spontaneous and induced ventricular activity ceased to exist. Interventions included tachypacing (150 ms), myosin II ATPase inhibitor blebbistatin (heart immobilizer), and the combination of blebbistatin and the inhibitor of glycolysis iodoacetate. In the absence of blebbistatin, confocal images were obtained during brief perfusion with hyperkalemic solution and after the contraction failed between 7 and 15 min of ischemia. In control, t mito_depol and t asys were and min, respectively. Tachypacing did not significantly affect either parameter. Blebbistatin dramatically delayed t mito_depol and t asys ( and min, respectively; both P vs. control). Iodoacetate combined with blebbistatin accelerated both events (t mito_depol, min; and t asys, min; both P 0.03 vs. control). In all groups pooled together, t asys was strongly correlated with t mito_depol (R ; P ). These data may indicate a causal relationship between m depolarization and asystole or a similar dependence of the two events on energy depletion during ischemia. Our results urge caution against the use of blebbistatin in studies addressing pathophysiology of myocardial ischemia. myocardial ischemia; mitochondrial depolarization; ATP-sensitive potassium channel; asystole; blebbistatin COLLAPSE OF MITOCHONDRIAL inner membrane potential ( m )is a major adverse event in the course of global myocardial ischemia and/or reperfusion (I/R). It has been implicated in loss of excitability during ischemia and subsequent postreperfusion ventricular fibrillation (VF) (2), and in the development of proarrhythmic action potential alternans during ischemia (25). Despite the obvious importance of m loss for the outcomes of an ischemic insult, the timing and the determinants of this event are still poorly understood. One reason for * P. W. Venable and K. J. Sciuto contributed equally to this study. Address for reprint requests and other correspondence: A. V. Zaitsev, Nora Eccles Harrison Cardiovascular Research and Training Inst., Univ. of Utah, 95 South 2000 E, Salt Lake City, UT ( zaitsev@cvrti. utah.edu). that is the difficulty of monitoring m in realistic, whole heart models of ischemia. We addressed these issues in our previous publication (29) and developed a new method for detection of m collapse in globally ischemic hearts based on spectral analysis of confocally derived fluorescence emitted by mitochondrial probe tetramethylrhodamine perchlorate methyl ester (29). In that study we provided evidence that a subcellular resolution afforded by confocal or multiphoton microscopy is necessary to detect m collapse during no-flow ischemia. This, however, necessitates complete immobilization of the heart, because otherwise the scanned microscopic images are distorted, preventing recognition of subcellular features. The recently discovered myosin-ii inhibitor blebbistatin (8) has been routinely used in both wide-field and confocal heart imaging studies addressing the pathophysiology of myocardial I/R (6, 24). However, these studies typically lacked appropriate controls for the possible modulating effects of blebbistatin with regard to the studied phenomena. The original purpose of this study was to provide experimental validation of the previously postulated cause-effect relationship between m collapse and electrical failure during ischemia (2). We reasoned that a strong correlation between the timing of m depolarization (t mito_depol ) and the timing of asystole (t asys ) during global ischemia in whole hearts would provide strong support, even though not complete proof, of the postulated causal relationship. The lack of such correlation would clearly refute it. We also hypothesized that high excitation rate, a condition relevant to VF-induced cardiac arrest, might affect t asys, t mito_depol, or both during ischemia. The difficult, but crucially important, part of the experimental design was to estimate the confounding effects of blebbistatin, which is required to perform confocal experiments in contracting hearts. For that purpose we transiently immobilized hearts with high-potassium Tyrode solution for obtaining baseline (preischemic) images and performed confocal imaging during ischemia only after the hearts experienced ischemic contraction failure. The experiments in blebbistatin-free hearts revealed, quite unexpectedly, that blebbistatin dramatically (by 100%) and proportionally increased both t mito_depol and t asys, whereas rapid pacing had minimal effects on these parameters. The effects of blebbistatin were fully abolished by the inhibitor of glycolysis iodoacetate, suggesting energy preservation as the underlying mechanism. Importantly, even in the absence of blebbistatin, a significant m depolarization occurred relatively late (at 25 min of ischemia), which is much later than it was previously thought (14) and suggests that m has little or no role in early ischemic events, such as contraction failure, action potential shortening (2), and action potential alternans (25). However, under all tested conditions, there was a direct correlation /15 Copyright 2015 the American Physiological Society H485

2 H486 between t asys and t mito_depol, which is compatible with a causal relationship between the two events, although it may also indicate a mutual dependence on a critical level of energy depletion. The observed, life supporting effects of blebbistatin urge great caution in the interpretation of ischemic studies conducted in the presence of this drug but also underscore the outstanding cardioprotective properties of blebbistatin, which might be useful in the setting of clinical I/R and organ preservation. METHODS Ethical approval. All procedures involving animals were approved by the Animal Care and Use Committee of the University of Utah (protocol number ) and complied with the National Institutes of Health s Guide for the Care and Use of Laboratory Animals (8th ed., 2011). Langendorff-purfused rabbit hearts. Adult New Zealand White rabbits of either sex ( kg) were euthanized by pentobarbital sodium (130 mg/kg iv). Hearts were quickly removed via midline sternotomy and attached to the Langendorff-perfusion apparatus in 3 min. Hearts were perfused with Tyrode solution consisting of 130 mm NaCl, 24 mm NaHCO 3, 1.2 NaH 2PO 4, 1.0 mm MgCl 2, 5.6 mm glucose, 4.0 mm KCl, 1.8 mm CaCl 2, and 0.1 g/l albumin (ph 7.4, and C) at a fixed rate of 30 ml/min. After an equilibrium period of 35 min, hearts were perfused with Tyrode solution containing the cationic flurophore tetramethylrhodamine perchlorate methyl ester (TMRM; 450 nm) for 30 min, followed by washout with normal Tyrode solution. Subsequent perfusion protocols depended on the experimental group as described below in detail; however, in all groups, the time interval between the heart cannulation and the onset of global ischemia was between 90 and 110 min, depending on the group. All hearts were placed in a custom confocal imaging chamber with a coverslip built in the bottom of the chamber (29). Temperature was monitored in the right ventricular cavity and near the imaged area in the posterior left ventricular (LV) epicardial surface and maintained at both sites at C throughout the protocol. A global bipolar electrogram (an analog of ECG) was continuously recorded from silver electrodes placed on the bottom of the imaging chamber on the two sides of the ventricles. Additionally, a small flexible bipolar electrode was placed between the imaging coverslip and the heart in the vicinity of the area imaged in the confocal microscope. Two bipolar pacing electrodes made of a pair of Teflon-coated silver wires were inserted into the interventricular septum and the LV free wall to probe the excitation threshold and to apply rapid pacing. Outflow of the perfusate from the heart was allowed to fill the imaging chamber to a level of 4 mm from the bottom, which helped to maintain moisture and temperature at the imaged area and provided a volume conductor for the global ECG. Global ischemia was initiated by the cessation of aortic perfusion and maintained for at least 60 min, followed by reperfusion. During ischemia, temperature was maintained at C by resistor heaters located in the perfusion bath walls and superfusion with heated Tyrode solution gassed with 95% N 2-5% CO 2. In four experiments, a miniature oxygen probe was inserted between the imaged area and the coverslip during ischemia to ensure hypoxic conditions at the imaged surface. Measured PO 2 was 20 mmhg during ischemia. Experimental groups. A total of seven experimental groups were used in this study. Hearts in all groups were subjected to at least 60 min of global no-flow ischemia. In the control group (n 7), the hearts were not treated with any drug and were not paced until the spontaneous ventricular activity was lost later in ischemia. In the Tachy group (n 8), the hearts were rapidly paced during ischemia (details below). In the BBS group (n 8), the hearts were treated with the myosin II ATPase inhibitor blebbistatin, which is routinely used for heart immobilization in cardiac imaging studies (8). In the BBS-IA group (n 5), the hearts were treated with the combination of blebbistatin and the inhibitor of anaerobic glycolysis sodium iodoacetate. In the Glyb group (n 6), the hearts were treated with the ATPsensitive potassium channel (K ATP) blocker glybenclamide before ischemia. In the Oligo10 and Oligo30 groups (n 3 each), the hearts were treated with F 0F 1 ATPase blocker oligomycin at the concentration of 10 or 30 M, respectively, for min before the onset of ischemia. In control, Tachy, Glyb, and Oligo groups, TMRM was washed out with normal Tyrode solution for 10 min. To arrest contraction and record baseline TMRM images before ischemia, hearts in these four groups were perfused for 15 min with a modified high-potassium Tyrode solution in which [KCl] was raised to 20 mm and [NaCl] was reduced to 114 mm; otherwise, the composition of the solution was the same as described above. The high-potassium solution was replaced with normal Tyrode solution, and normal sinus rhythm was observed for at least 10 min before the initiation of global ischemia. In the Tachy group, pacing at a cycle length (CL) of 150 ms and stimulus current amplitude of 1.5 times excitation threshold was initiated immediately preceding ischemia. During ischemia, as the excitation threshold increased and one-to-one capture was lost, the stimulus amplitude was increased to 3 times and then to 10 times preischemic excitation threshold, and the CL was increased to 200, 300, and 700 ms to maintain the highest ventricular activation rate possible. In all other groups, pacing was not initiated until atrioventricular block or asystole was observed on the ECG, after which point the same pacing algorithm was used as described above but starting at acl 300 ms (approximate CL of the spontaneous sinus rhythm). Hearts were paced from the LV free wall location until threshold exceeded 10 times preischemic excitation threshold at which point pacing was switched to the second electrode located in the interventricular septum and the same pacing algorithm was applied. The duration of tachycardic rhythm during ischemia was defined as the time during which the excitation rate was maintained at 5 Hz (CL 200 ms) or above, which included paced rhythms and ventricular tachycardia/fibrillation induced in some hearts by rapid pacing. In BBS and BBS-IA groups, the TMRM staining solution and all remaining Tyrode solutions included 5.7 M blebbistatin( ) (Sigma- Aldrich, St. Louis, MO). No high-potassium solution was needed because of the effective abolishment of contraction by blebbistatin. The hearts in BBS-IA group were perfused with the inhibitor of glycolysis sodium iodoacetate (250 or 1,000 nm, Sigma-Aldrich), dissolved in normal Tyrode solution for 15 min before the onset of ischemia. Confocal imaging. The posterior LV epicardial surface was imaged in a line scan mode using a Zeiss LSM 510 confocal microscope with a 20 objective lens, giving a field of view (FOV) of 450 m 450 m, with a resolution of m per pixel. The slice thickness was 2.7 m. In the standard imaging mode, we obtained z-stacks with 9 to 14 slices taken at a distance of 2.7 m between the centers of each slice, at time intervals of 3 5 min. The acquisition of one z-stack took 60 s. Shorter time intervals between z-stacks were avoided as they could cause a significant loss of TMRM signal (presumably because of TMRM bleaching) if repetitive recordings were continued over 30 min (not shown). Z-stacks were helpful in 1) selecting the slice with the sharpest images of synsitial myocytes and 2) tracking the same image plane amid macroscopic changes in myocardial geometry that consistently occurred during ischemia (for more, see RESULTS). In some cases, when a fast change in the pattern of TMRM fluorescence suggesting m depolarization was evident, a continuous time series was recorded at a fixed z-depth giving the sharpest image. In this mode the XY scans were repeated as soon as the previous XY scan was finished (3.4 s per scan). TMRM was excited using a 543-nm laser; emission was collected using a 560-nm long pass filter. The laser power and detector gain were adjusted to ensure that the TMRM signal was not saturated. In some later experiments, we switched

3 quickly between 20, 10, and 2.5 lenses to keep recognizable features of the preischemic imaged area in the view, which drifted away during fast changes in heart geometry associated with the early phase of ischemia. Data analysis. To identify the time of a significant t mito_depol during ischemia, we used a spectral method introduced in our previous publication (29). This method relies on the periodic arrangement of the interfibrillar mitochondria, yielding distinct ( mitochondrial ) peaks in the spatial Fourier spectrum of the TMRM fluorescence image obtained with a confocal microscope. Loss of m leads to equalization of TMRM concentration between the mitochondrial matrix and the cytosol, causing disappearance of these peaks (29). Consequently, the area under the mitochondrial peak (MPA) provides an index of mitochondrial polarization in ventricular myocytes (29). The time point when MPA values reached 50% of the difference between the minimum and maximum levels was used as t mito_depol. This time point was linearly interpolated when necessary. The t asys was defined as the time point at which no spontaneous ventricular activity was observed in either global or local electrogram, and there was no response to stimulus at 10 times preischemic excitation threshold applied at both pacing locations. Statistical analysis. To compare t mito_depol and t asys between experimental groups, we used a one-way ANOVA with Newman-Keuls post hoc test for individual pairwise comparisons. Linear regression analysis was applied to test for correlation between t mito_depol and t asys in four groups (control, Tachy, BBS, and BBS-IA) pooled together. A one-sample, two-tailed t-test was used to determine if the difference between t mito_depol and t asys in each group was significantly different from zero. All tests were performed using XLSTAT v (Addinsoft USA, New York, NY). Statistical significance was declared at P 0.05, and all the values presented in text, tables, and figures are means SD. RESULTS H487 Figures 1, 2, 3, and 4 show representative examples of confocally recorded TMRM fluorescence and corresponding MPA dynamics, along with time-aligned global bipolar ECG from control, Tachy, BBS, and BBS-IA groups, respectively. All these figures have the same layout. Figures 1 4, A C, show, from left to right, the original confocal TMRM images, their Fourier power spectrum, and spectral profiles, respectively, obtained at the selected time points indicated in Figs. 1 4, D. The spectral profiles are computed along the approximate longitudinal axis of myocytes indicated with yellow lines in the Fourier spectra. This line crosses two mitochondrial peaks. Because the spectra have radial symmetry, the two halves of the spectral profile and the respective mitochondrial peaks are summed up to yield the curves shown on the right-hand side of Figs. 1 4, A C. MPA values are computed as the integral of the region indicated with green color in the spectral profiles [see Venable et al. (29) for more detail]. Figures 1 4, D, show the time course of MPA at 3 5 min time steps, time-aligned with a global bipolar electrogram (an analog of ECG). Red and green vertical, dashed lines indicate t asys and t mito_depol, respectively. Note that in hearts not treated with blebbistatin (control, Tachy, Glyb, and Oligo groups), the preischemic images were obtained during a brief period of perfusion with high-potassium solution (performed min before the ischemic episode) as a means to abolish motion (indicated in cyan). In these hearts, complete cessation of contraction due to the ischemic contractile failure usually set the earliest time point at which confocal imaging became possible. In addition to rhythmic contractions, macroscopic changes in the heart shape (shrinking/swelling) could render confocal imaging impossible, because the imaged area significantly shifted even during a single scan. In 6 of 7 hearts from the control group, the blind period caused by the two types of motion artifact ranged from 7 to 13 min. In one control heart, the blind period was extended to 19 min because of technical problems with the microscope. In the BBS group, the main source of motion artifact was the macroscopic change in heart shape, although in some cases blebbistatin appeared to lose efficacy within the first 5 10 min of ischemia, leading to microscopic rhythmic contractions. In 7 of 8 hearts from the BBS group, the blind period (predominantly caused by heart shape changes) ranged from 5 to 10 min. In one BBS heart, the blind period was close to 20 min due to technical difficulties. Despite these limitations, the earliest MPA values obtained during ischemia were not significantly different from the preischemic MPA values in both control and BBS groups [89 23 and % of the preischemic value, respectively; P not significant (NS)]. In contrast, the end-ischemic MPA values were very significantly lower than preischemic MPA values in both control and BBS groups ( and % of the preischemic value, respectively; P ). In summary, confocal imaging in whole hearts was not possible during the early phase of ischemia either in the absence or in the presence of blebbistatin, but it is unlikely that the critical m depolarization occurred during that time period. Figures 5, 6, and 7 show expanded segments of both the global and the local LV electrogram at selected time points (including the moment of asystole) during representative experiments from the control, BBS and BBS-IA groups, respectively. BBS and BBS-IA groups exhibited the largest deviation in t asys from the control group; in all other groups, the values of t asys fell in the range between the two extreme cases. It can be seen that the shape and amplitude of both global ECG and local LV EG varied greatly in the course of the experiment, which reflected changes in electrical activity due to evolving ischemia, the presence or absence of pacing or episodes of spontaneous VF, as well as changes in the heart shape occurring during ischemia. The timing of the last ventricular beat in each experiment required meticulous inspection of the ECG to distinguish between the R waves diminishing with the time of ischemia, stimulus artifacts, and occasional artifacts caused by movement of the microscope stage or rotation of the lens turret. The amplitude of the stimulus artifact varied a great deal between experiments and, in some cases, exceeded the amplitude of the signal at later stages of ischemia. To aid a detection of faint R waves close to t asys, in some cases we applied a median filter to remove stimulus artifacts. However, in each experiment, the last R wave was clearly identifiable. In some cases the local LV electrogram lost detectable activity a few minutes before the global ECG (as in the control case shown in Fig. 5), which could reflect local heterogeneities in the timing of electrical failure (30). On average, the loss of R waves in the local LV electrogram preceded global asystole by 1.6, 3.0, 1.5, and 0.7 min in control, Tachy, BBS, and BBS-IA groups, respectively. We never observed the opposite, i.e., detectable local LV activity in the absence of detectable R waves in the global ECG. Since the local LV electrode was not truly representative of the imaged area (because it was about 3 mm away, not to interfere with the imaging), we decided to limit

4 H488 A Fig. 1. The relationship between mitochondrial membrane potential ( m) loss and asystole during global ischemia in a control heart. A C, left to right: confocal images of TMRM fluorescence, the respective fast Fourier transform (FFT) power spectra, and the spectral profiles obtained along the direction of the longitudinal axis of myocytes (yellow dashed line). These data are obtained at the respective time points indicated in D. Light green in the spectral profiles denotes the mitochondrial peak area (MPA) that reflects the magnitude of m. Vertical dashed lines indicate the position of mitochondrial peaks at the spatial frequency axis. Note the virtual absence of a mitochondrial peak in C, which corresponds to the loss of granular pattern in the tetramethylrhodamine perchlorate methyl ester (TMRM) image and heralds a significant, if not complete, loss of m. D: time course of MPA time-aligned with the volume-conductor ECG (bottom). Gray area indicates the period of ischemia. Cyan color indicates the period of perfusion with a high (20 mm) concentration of potassium. Discontinuities in the MPA curve indicate changes in the field of view caused by the macroscopic shifts in the heart shape (shrinking/swelling) occurring during ischemia. Red arrow indicates the moment of ischemia-induced atrioventricular block, at which time ventricular pacing was initiated. Green dashed line indicates the time of 50% decrease in MPA (t mito_depol). Red dashed line, the time of asystole (t asys). Note that in this case asystole occurred 4 min after the apparent m loss. Red asterisks indicate artifacts on the ECG caused by the movement of the microscope stage and manipulations on the perfusion tubing. B C D further analysis to the global ECG, which probably was the best available indicator of total electrical failure and provided the closest analog of information used for diagnosis of asystole in the setting of sudden cardiac arrest. Ischemic mitochondrial depolarization and asystole in control. Figure 1 shows data from a representative control experiment. Figure 1A shows data obtained under normoxemic conditions 18 min before ischemia in the presence of 20 mm extracellular [K ]. During perfusion with the high-potassium solution (shown with cyan in Fig. 1D), cardiac excitation and thus contraction were fully abolished. It can be seen that the TMRM fluorescence had a granular appearance typical of wellpolarized mitochondria and yielded prominent mitochondrial bands in the Fourier spectrum. Accordingly, the spectral profile along the longitudinal axis of myocytes featured a large-amplitude mitochondrial peak at spatial period approximately corresponding to the interval between the Z disks (2.15 m), as well as its first harmonic. Figure 1B shows data obtained at 9 min of ischemia. Note that the actual imaged myocytes are not the same as in the preischemic image (see Fig. 1A), because the heart rapidly changed shape within the first 8 min of ischemia and the original FOV was lost. However, the TMRM image at 9 min of ischemia was similar in appearance to that before ischemia. Likewise, the Fourier spectrum and the corresponding spectral profile (Fig. 1B, middle and right) showed large-amplitude mitochondrial peaks, yielding MPA at about 98% of the preischemic value. These data suggest that within the first 10

5 H489 Fig. 2. The relationship between m loss and asystole during global ischemia in a heart from the group paced during ischemia at cycle length 150 ms (Tachy group). A C, left to right: layout is the same as in Fig. 1, A C; confocal images of TMRM fluorescence, the respective FFT power spectra, and the spectral profiles obtained at the respective time points indicated in D. Note the virtual absence of a mitochondrial peak in C, which corresponds to the loss of granular pattern in the TMRM image and heralds a significant, if not complete, loss of m. D: time course of MPA time-aligned with the volume-conductor ECG. The dashed line connecting points A and B indicates the fact that there were no confocal recordings between these two time points, but the imaged area in B could be recognized as the same or almost the same as the imaged area in A. Rapid pacing (cycle length 150 ms) was initiated 1 min before the onset of ischemia. Green dashed line indicates t mito_depol. Red dashed line indicates t asys. Note that in this case asystole occurred 5 min before the apparent m loss. Red asterisks indicates artifacts on the ECG caused by the movement of the microscope stage and manipulations on the perfusion tubing. min of ischemia, the m was largely preserved. Note, however, that the position of the mitochondrial peak shifted during ischemia to a shorter spatial period (from 2.15 to 1.95 m), reflecting shortening of the sarcomere. MPA moderately decreased between 10 and 20 min and then catastrophically decreased between 23 and 24 min of ischemia. The confocal TMRM image obtained after this transition (Fig. 1C, left, corresponding to point C in Fig. 1D) shows disappearance of the granular pattern, suggesting redistribution of TMRM between the mitochondrial matrix and the cytosol. Accordingly, the Fourier spectrum and the spectral profile (Fig. 1C, middle and right) reveal the virtual absence of the mitochondrial peaks. These changes strongly suggest that in this experiment, a significant m loss occurred between 23 and 24 min of global ischemia (t mito_depol 23.5 min). Due to additional changes in the heart shape (typically associated with rapid phases of the apparent m loss) and the ensuing drift of the imaged area and the focal plane, the imaged area continuously imaged between time points B and C was lost. However, the two adjacent regions imaged during the rest of the ischemic period showed a similar smooth appearance of the TMRM image and similarly low levels of MPA (not shown), suggesting that a significant m loss had also occurred in those regions. In one control experiment, there was evidence of compromised myocytes at baseline, and an extreme contracture lead-

6 H490 A Fig. 3. The relationship between m loss and asystole during global ischemia in a heart from the group treated with 5.7 M blebbistatin (BBS group). A C, left to right: layout is the same as in Fig. 1, A C; confocal images of TMRM fluorescence, the respective FFT power spectra, and the spectral profiles obtained at the respective time points indicated in D. Note the very low amplitude of the mitochondrial peak in C, which corresponds to the loss of granular pattern in the TMRM image and heralds a significant, if not complete, loss of m. D: time course of MPA time aligned with the volume-conductor ECG. Note that unlike in Figs. 1 and 2, the preischemic perfusion with high potassium was not necessary because blebbistatin afforded complete immobilization. The fluctuations in the MPA values before and during the first 40 min of ischemia most likely reflect a high sensitivity of the MPA to slight changes in the focal plane. Red arrow indicates the moment of ischemia-induced atrioventricular block, at which time ventricular pacing was initiated. Green dashed line indicates t mito_depol. Red dashed line indicates t asys. Note a slow MPA decline compared with the control and Tachy groups (see Figs. 1 and 2). Asystole occurred 8 min before the apparent 50% m loss, but after the (relatively slow) process of apparent m loss had begun. ing to highly contorted shapes of myocytes had occurred relatively early in ischemia. Since these phenomena were not observed in other control experiments, data from this experiment were excluded from statistical analysis. In the remaining seven control hearts, t mito_depol occurred at min of ischemia (range, min). In 5 out of 7 control experiments, there was an apparent rapid phase of m loss, such that the most of MPA decrease occurred within 10 min, whereas in the remaining two control experiments, the MPA decrease was more gradual during ischemia. The global bipolar electrogram (an analog of ECG) recorded simultaneously with confocal imaging is presented in the bottom of Fig. 1D. It shows transient cessation of electrical activity caused by perfusion with high-potassium solution, during which the baseline confocal image was acquired (point A). Complete atrioventricular block occurred at 7 min of ischemia (red arrow) and was followed after a short delay with B C D ventricular pacing. The red dashed line indicates the time at which pacing from two ventricular locations at stimulus amplitude 10 times preischemic excitation threshold failed (t asys ), which in this case followed t mito_depol by 4 min. In this and other control hearts, asystole occurred at min of ischemia (range, 17 to 33 min). The average time delay between t asys and t mito_depol (t asys t mito_depol ) was equal to min and was not significantly different from zero. Ischemic mitochondrial depolarization and asystole in hearts rapidly paced during ischemia (Tachy group). Tachycardic rhythm is present in a large fraction of cases of sudden cardiac arrest. Because of increased metabolic load and increased intracellular Na and Ca 2, it is possible that a high excitation frequency during global ischemia affects t mito_depol, t asys, or the relationship between the two. Thus, in the Tachy group (n 8), we applied rapid pacing (CL 150 ms) starting at 2 to 3 min before ischemia and continuing during ischemia

7 H491 A B C D Fig. 4. The relationship between m loss and asystole during global ischemia in a heart from the group treated with 5.7 M blebbistatin and 1 mm Na iodoacetate (BBS-IA). A C, left to right: layout is the same as in Fig. 1, A C; confocal images of TMRM fluorescence, the respective FFT power spectra, and the spectral profiles obtained at the respective time points indicated in D. Note the very low amplitude of the mitochondrial peak in C, which corresponds to the loss of granular pattern in the TMRM image and heralds a significant, if not complete, loss of m. D: time course of MPA time aligned with the volume-conductor ECG. Note that unlike in Figs. 1 and 2, the preischemic perfusion with high potassium was not necessary because blebbistatin afforded complete immobilization. Black vertical arrow indicates the time when the perfusion was switched to the solution containing Na iodoacetate. Green dashed line indicates t mito_depol. Red dashed line indicates t asys. Note that the phase of fast MPA decline occurred very early (the point of 50% MPA loss was at 11.5 min of ischemia) compared with Figs. 1 3 but was followed by a partial recovery. Asystole occurred even earlier (at 6 min of ischemia) and never recovered. Red asterisks indicates artifacts on the ECG caused by the movement of the microscope stage and manipulations on the perfusion tubing. for as long as the heart could follow this pacing rate in a one-to-one fashion. As the ability to capture the heart progressively deteriorated during ischemia, the CL and the stimulus current were progressively increased as described in METHODS. Rapid pacing precipitated transient episodes of ventricular tachycardia or fibrillation in 7 of 8 hearts with an excitation rate similar to the rate of pacing. Tachycardic rhythm (either paced rhythm or VT/VF at CL not exceeding 200 ms) was maintained for the first min of ischemia, after which time point increasing postrepolarization refractoriness reduced the maximal rate the heart could follow. Neither t mito_depol nor t asys was significantly affected by the presence of tachycardic rhythm during the above-mentioned time period of early ischemia (t mito_depol, min; and t asys, min; P NS vs. control). A representative example of confocal imaging and ECG data from the Tachy group are shown in Fig. 2 (the same layout as in Fig. 1). Blebbistatin greatly postpones ischemic mitochondrial depolarization and asystole. Blebbistatin, a recently discovered selective inhibitor of myosin II ATPase, has become a standard of care electromechanical uncoupler for use in experimental studies involving optical mapping or confocal imaging, including those addressing effects of I/R on electrophysiological parameters and m dynamics (6, 23, 24). It was recently

8 H min before ischemia 1 min before ischemia Global ECG Local EG Fig. 5. Expanded segments of global and local ECG recordings at selected time points (as indicated) from a representative control experiment. All recordings during ischemia show paced rhythm. Arrows show the last identified ventricular excitation in the local and the global ECG, respectively. Note that local electrical failure occurred 3 min before the global electrical failure (asystole). The dual ECG strips showing the moments of local and global electrical failure are supplemented with the tracings of stimuli (note a different time scale). Note also that P waves asynchronous with pacing and ventricular activity are present in all ischemic recordings and persist even after full loss of ventricular capture. Stim, stimulus. Stim Stim shown that blebbistatin postpones changes in cellular redox state associated with ischemia (32). So it was important to test whether blebbistatin also modulates the time of ischemic electrical failure and m loss. Results from a representative BBS experiment are shown in Fig. 3. One can see that both m depolarization and asystole were dramatically postponed compared with control (compare with data in Fig. 1). One can appreciate a relatively slow (as compared with Figs. 1 and also 2) dynamics of MPA decrease, which started at 35 min of ischemia and continued until 65 min of ischemia. Within the first 30 min of ischemia, MPA dynamics show minor fluctuations (at the level close to the preischemic level), which are most likely due to sensitivity of the spectral analysis to slight changes in the position of the focal plane, often unavoidable in experiments using whole hearts. Despite these fluctuations, the overall decrease of MPA by about 90% of the preischemic value is perhaps a robust indicator of true ischemic m loss. In this and other BBS experiments, the estimated t mito_depol was min of ischemia (range, min), an increase of 111% over the control value (P ). In the BBS group, the speed of the MPA loss appeared to be slower than in the control group, but it varied greatly among individual BBS experiments and it was difficult to quantify. In the case shown in Fig. 3, t asys preceded t mito_depol by 7 min. Note, however, that at t asys, a decrease in the MPA value was already apparent. In this and other BBS hearts, asystole occurred at min of ischemia (range, min). The average time delay between t asys and t mito_depol (t asys t mito_depol ) was equal to min and was not significantly different from zero, despite the fact that in 7 of 8 BBS experiments, asystole occurred before 50% MPA loss. 10 min of ischemia Local electrical failure (23.75 min ischemia) Global electrical failure (27.15 min ischemia) 20 min of ischemia 2sec 4sec Glycolysis inhibition greatly accelerates ischemic mitochondrial depolarization and asystole. To assess the role of anaerobic glycolysis in the prolonged maintenance of m and myocardial excitability during ischemia afforded by blebbistatin, four hearts were treated with an inhibitor of glycolysis sodium iodoacetate (250 1,000 M) in addition to blebbistatin (BBS-IA group). Figure 4 shows a representative example of such an experiment. Before the onset of ischemia, there were no major effects of iodoacetate on either m or electrical activity. However, electrical failure occurred extremely early in the course of ischemia (at 6 min of ischemia) and was shortly followed by a sharp MPA decrease, reaching the 50% level at 12.5 min of ischemia. Of interest, the phase of sharp MPA decrease was followed by a partial recovery in MPA (observed in 3 out of 5 BBS-IA experiments), but electrical activity never recovered. It should also be noted that upon reperfusion, BBS-IA hearts never recovered any electrical activity and apparently had a complete m loss within the first 10 min of reperfusion (not shown). In five BBS-IA experiments, t asys was min and t mito_depol was min. Both t asys and t mito_depol in BBS-IA group were significantly different from the values of the respective parameters in both BBS and control groups. Thus preservation of m and sarcolemmal excitability afforded by blebbistatin during ischemia is fully abolished by the blockade of anaerobic glycolysis. The coordinated response of t asys and t mito_depol to interventions affecting energy balance becomes more evident when the data from the control, Tachy, BBS, and BBS-IA groups are pooled together (Fig. 8). One can see that overall there is a strong direct correlation between the timing of the two critical events (R , P ), such that, on average, t asys is 1mV 1mV

9 H min before ischemia 1 min before ischemia Global ECG Local EG 10 min of ischemia 20 min of ischemia Stim 30 min of ischemia Global & local electrical failure (50.21 min of ischemia) slightly ahead of t mito_depol. Note that the points from the control and Tachy groups significantly overlap, whereas the points from BBS and BBS-IA groups form distinct clusters fully separated from the control/tachy cluster. Effects of blocking K ATP on ischemic mitochondrial depolarization and asystole. A possible reason for correlation between m collapse and electrical failure during ischemia is the activation of the sarcolemmal K ATP due to ATP consumption by depolarized mitochondria (2, 19). To further ascertain the role of K ATP in ischemic asystole we performed experiments using the K ATP blocker glybenclamide (10 M, Glyb group), the drug most commonly used for testing the role of the sarcolemmal K ATP channel during ischemia (2). In theory, glybenclamide may also block the mitochondrial K ATP (mk ATP ). The expected outcome would be prolonged maintenance of m during ischemia and possibly preservation of ATP. It is appropriate to note, however, that the very existence of mk ATP remains controversial (10). Whereas the role of mk ATP in acutely ischemic myocardium has never been directly demonstrated, it was shown that at the concentration used in this study (10 M), glybenclamide does not alter the dynamics of ATP depletion during ischemia (7). Therefore, regardless of possible effects of glybenclamide on the mk ATP, using this drug to test the role of the sarcolemmal K ATP in t asys was deemed to be appropriate. Data from a representative heart treated with glybenclamide are shown in Fig. 9. Glybenclamide postponed t asys (to vs min in control, P 0.05), supporting a role of K ATP in ischemic electrical failure (2, 27). t mito_depol in the Glyb group (measured as the time of 50% MPA loss) was not 40 min of ischemia 2sec 4 sec 1mV 0.2 mv Fig. 6. Expanded segments of global and local ECG recordings at selected time points (as indicated) from a representative BBS experiment (heart treated with 5.7 M blebbistatin). All recordings during ischemia show paced rhythm. Arrows show the last identified excitation in the local and the global ECG, respectively. In this experiment local and global electrical failure occurred simultaneously. Note a different time and voltage scale in the lowermost dual ECG strip. The recording at the very bottom shows the timing of stimulation. Note that the spikes seen in the local EG recording after the last detected excitation are due to stimulus artifacts. different from control ( vs min in control, P NS). However, unlike the control group, in the Glyb group the earliest possible ischemic MPA values (obtained at min of ischemia) were significantly lower than the preischemic values ( % of preischemic value, P 0.001). In one Glyb experiment, the earliest ischemic MPA value recorded at 10 min of ischemia fell below 50% of the preischemic value, which necessitated the exclusion of this experiment from t mito_depol analysis. Thus it is possible that glybenclamide caused partial m depolarization early in ischemia (see possible reasons in DISCUSSION). Effects of blocking F 0 F 1 ATP synthase on ischemic mitochondrial depolarization and asystole. To confirm that m during ischemia is maintained because of ATP hydrolysis by F 0 F 1 ATP synthase working in reverse, we used F 0 F 1 ATP synthase blocker oligomycin. At the concentration of 10 M (Oligo10 group, n 3), oligomycin did not have any significant effect (not shown). At the concentration of 30 M (Oligo30 group, n 3), oligomycin dramatically accelerated m depolarization. Specifically, the earliest ischemic MPA value (measured at min of ischemia) was already reduced to % of the preischemic level (P 0.023). In two hearts, t mito_depol (defined as the time point at which the MPA value fell below 50% of the preischemic) was 7.0 and 8.5 min of ischemia. In the remaining heart from this group, the earliest MPA value measured at 5.5 min of ischemia was already below 50% of the preischemic value. When 5.5 min is used as the conservative estimate of t mito_depol in this experiment, the average between three Oligo30 experiments was min, which was significantly shorter than in control (P 0.05).

10 H494 Fig. 7. Expanded segments of global and local ECG recordings at selected time points (as indicated) from a representative BBS-IA experiment (heart treated with 5.7 M blebbistatin and 1 mm iodoacetate). Arrows show the last identified excitation in the local and the global ECG, respectively. The dual ECG strips showing the moments of local and global electrical failure are supplemented with the tracings of stimuli. Note a different time and voltage scale in these two ECG strips. Noteworthy features: 1) partial atrioventricular blockade occurred already before the onset of ischemia; 2) a fine ventricular fibrillation is seen at 3 min of ischemia; and 3) before the full electrical failure, the amplitude of both global and local ECG is only about 5 10% of the preischemic amplitude. This supports the essential role of F 0 F 1 ATP synthase (AT- Pase) in m maintenance during ischemia. t asys in the Oligo30 group was min, which was not significantly different from control. Summary of t asys and t mito_depol in different experimental groups. Fig. 10, A and B, summarizes the statistical analysis of t asys and t mito_depol, respectively, in all experimental groups. All t asys (min) y = 0.85 x R 2 = p < Control Ta c h y BBS BBS-IA t mito_depol (min) Fig. 8. Direct correlation between t mito_depol and t asys in 4 groups (control, Tachy, BBS, and BB-IA) pooled together. Note that on average t asys was slightly ahead of t mito_depol, judging from the fact that the slope of the linear regression curve was 1. pairwise comparisons were performed, but only physiologically meaningful significant differences are indicated. When compared with control, t asys was significantly prolonged in the Glyb and the BBS group and significantly shortened in the BBS-IA group. Also, t asys was significantly shorter in the BBS-IA and Oligo30 groups than in the BBS group. When compared with control, t mito_depol was significantly prolonged in the BBS group and significantly shortened in the BBS-IA and Oligo30 groups. Also, t mito_depol was significantly shorter in the BBS-IA and Oligo30 groups than in the BBS group. Figure 11 shows the difference between t asys and t mito_depol in all experiments. In each group this difference was compared with zero (i.e., the 2 events are simultaneous). One can see that in the control and Tachy groups, there was no consistent order in the timing of the two critical events. In the BBS group, in all but one heart t asys was ahead of t mito_depol, but the presence of two outliers precluded statistically significant difference from zero. In all BBS-IA experiments, t asys was ahead of t mito_depol, and in all Glyb and Oligo30 experiments, t asys was behind t mito_depol. The difference between t asys and t mito_depol was significant (P 0.05) in BBS-IA and Glyb groups and was close to being significant (P 0.058) in the Oligo30 group. DISCUSSION Detection of m loss during no-flow ischemia. To our knowledge, this is the first study in which confocal imaging of m -senstitive fluorescence was performed in hearts subjected

11 H495 A B C D Fig. 9. Relationship between m loss and asystole during global ischemia in a heart from the group treated with 10 M glybenclamide (Glyb group). A C, left to right: confocal images of TMRM fluorescence, the respective FFT power spectra, and the spectral profiles obtained at the respective time points indicated in D. Note the virtual absence of the mitochondrial peak in C, which corresponds to the loss of granular pattern in the TMRM image and heralds a significant, if not complete, loss of m. D: time course of MPA time aligned with the volume-conductor ECG. Gray area indicates the period of ischemia. Cyan color indicates the period of perfusion with a high (20 mm) concentration of potassium. Discontinuity in the MPA curve between points A and B indicates a change in the field of view caused by macroscopic shifts in the heart shape (shrinking/swelling), which occurred during early ischemia. Green dashed line indicates t mito_depol. Red dashed line indicates t asys. Asystole occurred after the apparent 50% m loss, which was the case in all Glyb experiments (see Fig. 11). Note that the earliest ischemic value of MPA (point B) was much lower than the preischemic value (point A). This was a consistent finding in the Glyb group, contrasting it from other groups for reasons not completely understood (see text for more detail). Red asterisks indicate artifacts on the ECG caused by the movement of the microscope stage and manipulations on the perfusion tubing. to total (no flow) ischemia in the absence of electromechanical uncouplers. Lyon et al. (14) also investigated the dynamics of m in a globally ischemic heart without use of chemical immobilization, but these authors used wide-field video imaging instead of confocal microscopy. According to these authors, m depolarization to a level below 50% of the preischemic level occurred within 7 min of global no-flow ischemia, which is much earlier than t mito_depol determined in our study (on average, 24 min of ischemia). The discrepancy may be related to different species (rats vs. rabbits). In addition to that, the difference in the method may be important. Lyon et al. reported a decrease in TMRM fluorescence in some locations to almost zero level during the first few minutes of ischemia. We did not observe any consistent decrease in the average level of TMRM fluorescence during global ischemia, neither using video imaging nor using confocal microscopy, as we reported in our previous publication (29). Theoretically, the TMRM signal obtained using wide-field imaging (and thus not resolving individual mitochondria) is not expected to decrease during no-flow ischemia, since the system is closed and the total amount of TMRM molecules emitting fluorescence remains the same. To detect m depolarization under these conditions, an approach allowing distinction between mitochondrial and nonmitochondrial TMRM fluorescence is necessary. That is why we developed a method based on the spectral analysis of confocally recorded TMRM fluorescence described and validated in our previous publication (29). This method relies on the fact that the periodic packaging of interfibrillar mitochon-