Detection of mitochondrial depolarization/recovery during ischaemia–reperfusion using spectral properties of confocally recorded TMRM fluorescence

Abstract

Timing and pattern of mitochondrial potential (ΔΨ_m) depolarization during no-flow ischaemia–reperfusion (I–R) remain controversial, at least in part due to difficulties in interpreting the changes in the fluorescence of ΔΨ_m-sensitive dyes such as TMRM. The objective of this study was to develop a new approach for interpreting confocal TMRM signals during I–R based on spatial periodicity of mitochondrial packaging in ventricular cardiomyocytes. TMRM fluorescence (F_TMRM) was recorded from Langendorff-perfused rabbit hearts immobilized with blebbistatin using either a confocal microscope or an optical mapping system. The hearts were studied under normal conditions, during mitochondrial uncoupling using the protonophore FCCP, and during I–R. Confocal images of F_TMRM were subjected to spatial Fourier transform which revealed distinct peaks at a spatial frequency of ∼2 μm⁻¹. The area under the peak (MPA) progressively decreased upon application of increasing concentrations of FCCP (0.3–20 μm), becoming undetectable at 5–20 μm FCCP. During ischaemia, a dramatic decrease in MPA, reaching the low/undetectable level comparable to that induced by 5–20 μm FCCP, was observed between 27 and 69 min of ischaemia. Upon reperfusion, a heterogeneous MPA recovery was observed, but not a de novo MPA decrease. Both confocal and wide-field imaging registered a consistent decrease in spatially averaged F_TMRM in the presence of 5 μm FCCP, but no consistent change in this parameter during I–R. We conclude that MPA derived from confocal images provides a sensitive and specific indicator of significant mitochondrial depolarization or recovery during I–R. In contrast, spatially averaged F_TMRM is not a reliable indicator of ΔΨ_m changes during I–R.

Key points

• Mitochondrial inner membrane potential (ΔΨ_m) collapse during myocardial ischaemia is one of the key events determining the physiological consequences of ischaemic attack in terms of post-ischaemic arrhythmias and cell survival.

• Timing and pattern of ΔΨ_m collapse during ischaemia remain controversial, in part due to difficulties in interpreting the fluorescence of potentiometric cationic probes commonly used for assessment of ΔΨ_m in cellular and multicellular experimental models.

• This manuscript presents a new method for monitoring ΔΨ_m in whole hearts based on the regular arrangement of mitochondria in cardiac myocytes, thus permitting detection of ΔΨ_m collapse using spectral analysis of fluorescence.

• The proposed method will help to ascertain the role of mitochondrial function in acute cardiovascular conditions, such as acute myocardial infarction or sudden cardiac arrest.

Introduction

The dynamics of ΔΨ_m loss and recovery during I–R have been implicated in the immediate (loss of excitability, post-ischaemic arrhythmias; Akar et al. 2005) and long-term (cell survival or death; Petronilli et al. 2001) outcomes of a regional ischaemic event due to coronary occlusion, or global ischaemia due to cardiac arrest or trauma. To date, however, most of these concepts are based on knowledge obtained in isolated mitochondria or isolated cardiomyocytes, model systems which cannot fully recapitulate all the essential features of I–R in whole hearts. In recent years, several groups started studying the dynamics of ΔΨ_m during I–R or its surrogates in whole hearts by monitoring fluorescence emitted by the rhodamine-based cationic dyes TMRM and TMRE using either confocal microcopy (Matsumoto-Ida et al. 2006; Slodzinski et al. 2008; Kato et al. 2009; Davidson et al. 2012) or wide-field optical mapping (Jin et al. 2010; Lyon et al. 2010). Reviewing these studies, one cannot fail to notice a diversity of outcomes with regard to the timing (early (Lyon et al. 2010) or late: (Matsumoto-Ida et al. 2006; Slodzinski et al. 2008; Kato et al. 2009) in ischaemia, or upon reperfusion (Matsumoto-Ida et al. 2006; Slodzinski et al. 2008; Davidson et al. 2012)) and spatial pattern (mosaic (Matsumoto-Ida et al. 2006; Slodzinski et al. 2008; Davidson et al. 2012), diffuse (Kato et al. 2009), or propagating as a wave (Lyon et al. 2010)) of delta ΔΨ_m depolarization during I–R. Without better understanding of the actual dynamics of ΔΨ_m during realistic I–R, it will be impossible to translate the vast amount of prior knowledge obtained in reductionist approaches to the level of integrative physiology and eventually to the level of human cardiac disease.

At least in part, the variability in results may be due to the difficulty of interpreting changes in fluorescence emitted by the dyes used to monitor ΔΨ_m (superbly reviewed by Duchen et al. (2003)). In the first approximation, the cationic dyes TMRM and TMRE distribute across the extracellular space, cytosol and mitochondrial matrix based on a 3-chamber Nernstian system which is governed by both ΔΨ_m and the sarcolemmal transmembrane potential (ΔΨ_s). This leads to the highest concentration of the dye in the most negatively charged compartment, the mitochondrial matrix. Mitochondrial depolarization causes a redistribution of the dye between the mitochondrial matrix and the cytosol, with subsequent redistribution to extracellular space. The difficulty stems from the fact that individual mitochondria in cardiac myocytes cannot be reliably resolved as separate objects, even using confocal microscopy, and therefore the event and the degree of the dye redistribution has to be assessed from spatially averaged measurements that unavoidably include contributions from compartments other than the mitochondrial matrix. In the worst-case scenario, if the total pool of dye molecules in the compartments contributing to the recorded fluorescence does not change, then ΔΨ_m loss may not be detectable at all. Notably, during no-flow ischaemia the total pool of the fluorophores in all myocardial compartments does remain the same, thus the ability to detect ΔΨ_m depolarization is critically dependent on separating the movement of the dye between mitochondria and cytosol from other redistributions, including those which might be unrelated to ΔΨ_m dissipation.

Amid these inherent difficulties and the diversity of previous observations, we sought to develop a new approach, which would be a more reliable and sensitive index of ΔΨ_m loss in intact ventricular myocardium during no-flow ischaemia. Here we present a method taking advantage of spectral properties of confocally recorded TMRM fluorescence (F_TMRM), resulting from the spatially periodic arrangement of mitochondria within ventricular myocytes. This method permits detection of redistribution of a cationic fluorophore which is highly specific to mitochondrial depolarization. We provide evidence that, in confocal imaging, the spectral method is superior to tracking the mean F_TMRM level or its spatial dispersion (Brennan et al. 2006) and, unlike the other two measures, allows detection of the complete ΔΨ_m loss event with a high degree of certainty.

Methods

Ethical approval

All procedures involving animals were approved by the Animal Care and Use Committee of the University of Utah and complied with the National Institutes of Health Guide for the Care and Use of Laboratory Animals 1996.

Isolated adult rabbit ventricular myocytes

Adult ventricular myocytes were isolated from rabbit hearts (n= 3) by enzymatic digestion as previously described (Skolnick et al. 1998; Ajiro et al. 2011). Animals were killed by i.v. injection of 130 mg kg⁻¹ sodium pentobarbital, then the excised heart was attached to an aortic cannula and perfused with solutions gassed with 100% O₂ and held at 37°C, pH 7.3. The heart was first perfused for 5 min with a 0 mm Ca²⁺ solution containing (in mm): 92.0 NaCl, 4.4 KCl, 11.0 dextrose, 5.0 MgCl₂, 20.0 taurine, 5.0 creatine, 5.0 sodium pyruvate, 1.0 NaH₂PO₄, 24.0 Hepes, and 12.5 NaOH. This was followed by 13–15 min of perfusion with the same solution containing 0.075 mg ml⁻¹ collagenase P (Roche Diagnostic, Manheim, Germany), 0.05 mg ml⁻¹ protease (type XIV, Sigma Chemical, St Louis, MO, USA), and 0.05 mm CaCl₂. The heart was then perfused for 5 min with the same solution containing no enzymes. The left ventricle was minced and shaken for 10 min, and then filtered through a nylon mesh. Cells were stored at room temperature in normal Hepes-buffered solution containing 1.1 mm CaCl₂. All myocytes used in this study were rod shaped, had well-defined striations, and did not spontaneously contract. Experiments were performed within 10 h of isolation.

Cells were incubated in 400 nm tetramethyl rhodamine perchlorate methyl ester (TMRM) for 3–10 min at ambient temperature, placed in a perfusion chamber mounted on the stage of an inverted confocal microscope (Zeiss LSM 510, Carl Zeiss, Germany) and perfused with normal Hepes solution (126 mm NaCl, 1.0 mm MgCl₂, 4.4 mm KCl, 1.1 mm CaCl₂, 11 mm dextrose, 24 mm Hepes) at 37°C. Cells were confocally imaged using an oil-immersion 40× objective lens. F_TMRM was excited using a 543 nm laser and collected using a 560 long pass filter. To uncouple mitochondria and induce ΔΨ_m depolarization, perfusion was rapidly switched to Hepes solution containing various concentrations of protonophore carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP).

Langendorff-perfused rabbit heart

Adult rabbits of either sex were killed by i.v. injection of 130 mg kg⁻¹ sodium pentobarbital. Hearts were removed and attached to the Langendorff apparatus in less than 3 min. Hearts were perfused with Tyrode's solution (130 mm NaCl, 24 mm NaHCO₃, 1.2 mm NaH₂PO₄, 1.0 mm MgCl₂, 5.6 mm glucose, 4.0 mm KCl, 1.8 mm CaCl₂ and 0.1 g l⁻¹ albumin) at a fixed rate of 30 ml min⁻¹. After an equilibration period of 35 min the heart was perfused with Tyrode's solution containing TMRM (450 nm) and the electromechanical uncoupler blebbistatin (5.7 μm, Sigma-Aldrich, St Louis, USA; Fedorov et al. 2007) for 30 min, unless stated otherwise. For the remainder of the experiment, hearts were perfused with Tyrode's solution containing blebbistatin. In some experiments the mitochondrial uncoupler FCCP (5 μm) and an inhibitor of glycolysis, sodium iodoacetate (15 mm), were added to the perfusate. Global ischaemia was initiated by cessation of aortic perfusion and maintained for at least 60 min, followed by reperfusion. Volume-conducted ECG was monitored continuously throughout the protocol. A bipolar pacing electrode was inserted into the depth of the left ventricular (LV) anterio-basal free wall and the pacing threshold was periodically measured throughout the experiment.

Confocal imaging in the whole heart

Langendorff-perfused hearts (n= 12) were placed in a custom confocal imaging chamber (Supplemental Fig. 1). Temperature was monitored in the right ventricular (RV) chamber and near the imaged epicardial area and was maintained in both sites at 37.7 ± 0.5°C throughout the experimental protocol. During ischaemia the temperature was maintained by the combined action of superfusion with Tyrode's solution bypassed from the aortic cannula and inflow of warm moisturized air into the imaging chamber (see Supplemental Fig. 1). 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 of 450 μm × 450 μm, with resolution of 0.439 μm per pixel. The slice thickness was 2.7 μm. When z-stacks were obtained, 9 to 14 slices were taken at the distance of 2.7 μm between the centres of each slice. TMRM was excited using a 543 nm laser; emission was collected using a 560 nm long pass filter. The laser power and detector gain was adjusted to ensure the TMRM signal was not saturated. The mean TMRM fluorescence (F_TMRM) was calculated after applying manual masking to exclude grossly areas not containing solid sheets of myocardial cells. The identification and quantification of mitochondrial spectral peaks was performed using the Fourier transform function available in ImageJ and custom software routines as described in the Results.

Wide-field optical mapping in the whole heart

Hearts (n= 11) were fully immersed in a superfusion solution inside the imaging chamber to maintain the temperature at 37°C both during normoxic perfusion and no-flow ischaemia. During no-flow ischaemia the superfusion was switched to a hypoxic solution gassed with 95% N₂–5% CO₂. TMRM was excited by a 532 nm solid-state laser (Coherent, Santa Clara, CA, USA). Emitted fluorescence was collected by an EMCCD camera (iXon DU-860D, Andor Technology, Belfast, UK) using a 600 ± 40 nm bandpass filter. The imaging area was 3 cm × 3 cm encompassing the entire posterior view of the right and left ventricles (see Supplemental Fig. 8). Wide-field images were analysed using Scroll (Sergey Mironov, Center for Arrhythmia Research, University of Michigan, USA) after a careful manual selection of the area exclusive to RV and LV in each heart. The details of different measurements applied to the wide-field TMRM images are described in Results.

Results

Spectral analysis and MPA computation

The idea of using spectral analysis to track mitochondrial polarization was suggested by a highly regular, ‘crystal-like’ pattern of mitochondrial arrangement in ventricular myocytes, such that individual mitochondria usually span a sarcomere (Vendelin et al. 2005). Thus, fluorescence emitted by dyes preferentially accumulating in mitochondria might reflect the highly periodic spacing of the z-disks. We first validated this idea in isolated cells. Figure 1 shows TMRM fluorescence and transmitted light images of an isolated ventricular cardiomyocyte taken in the absence (panels Aa and Ac) and in the presence (panels Ab and Ad) of the mitochondrial uncoupler FCCP (20 μm). The high concentration of the uncoupler was chosen to ensure complete dissipation of ΔΨ_m. Indeed, one can see that the brightness of F_TMRM is greatly reduced in the presence of FCCP (Fig. 1Ab), as is expected due to redistribution between mitochondrial matrix, cytosol and the (virtually infinite) perfusion bath. In addition, the spatial pattern of F_TMRM image is dramatically altered. Specifically, the ‘granularity’ apparent in the control image (Fig. 1Aa) is completely lost in the presence of FCCP (Fig. 1Ab). A zoomed-in fluorescent image of the cardiomyocyte (Fig. 1B) shows an example of regular mitochondrial arrangement (with a period of ∼2 μm, see red arrows in Fig. 1Ba) along the longitudinal axis of the cell, which is abolished by application of FCCP (Fig. 1Bb). Such periodic features, even those not obvious by eye, can be revealed by spectral analysis.

Figure 1. Spatial Fourier transform (SFT) reveals ΔΨ_m-dependent spatial periodicity in confocal images of F_TMRM.

Aa and b, F_TMRM image of a single ventricular cardiomyocyte before and after, respectively, administration of 20 μm FCCP. Ac and d, transmitted light images corresponding to those shown in Aa and b, respectively. Ba and b, zoomed-in region of interests indicated by the white square in Aa and b, respectively. Note the presence of a highly periodic array of mitochondria at baseline spaced at ∼2 μm (red arrows in Ba) and lack thereof after application of FCCP (Bb). Ca–d, SFT spectra of images shown in Aa–d, respectively. The spectral peaks (red arrows) in Ca, Cc, and Cd correspond to the spatial period of ∼2 μm. The fact that after mitochondrial uncoupling by FCCP these peaks are absent in the SFT of the F_TMRM image (Cb), but are preserved in the SFT of the transmitted light image (Cd) suggests that these peaks are specific to TMRM accumulation in charged mitochondria. Yellow dashed line, the longitudinal axis of the myocyte.

Figure 1Ca–d shows spatial Fourier transformations (SFTs) of the images shown in Fig. 1Aa–d, respectively. By convention, the SFT spectra are displayed in polar coordinates, where the distance from the centre (radius) represents the size of periodic features and the angle represents the orientation of the features (see Supplemental Fig. 2). Note that in the absence of FCCP the SFT spectra of both the F_TMRM and transmitted light images (Fig. 1Ca and c) show distinct bands (see red arrows) at the radius corresponding to a spacing interval of ∼2 μm and at the angle corresponding to the longitudinal axis of the cell (see yellow dashed line). In the presence of FCCP, however, the bands corresponding to spacing interval of ∼2 μm are visible only in the SFT spectrum of the transmitted light image (Fig. 1Cd), but are absent in the SFT spectrum of F_TMRM image (Fig. 1Cb). These observations suggest that the periodic features imposed by structure (presumably z-disks) are not affected by FCCP, whereas the periodic fluctuations of TMRM concentration along the longitudinal axis of the cell have dissipated.

Figure 2 shows that the essential spectral features observed in single myocytes are fully present in the spectrum of a F_TMRM image obtained from a whole heart (Fig. 2A). In particular, the bands corresponding to spatial intervals of ∼2 μm and aligned with the longitudinal axis of the myocytes are readily recognizable (Fig. 2B). We will refer to these bands/peaks as ‘mitochondrial peaks’. In order to quantify the magnitude of these peaks we developed an algorithm illustrated in Fig. 2. Angles from 0 to π rad were scanned at small angular increments. At each angle Θ, the spectral power within the sector centred at Θ and having a user-defined angular half-width (ΔΘ) was computed as the function of radius (r) and normalized to the arc length. Due to two-fold rotational symmetry of the SFT spectrum, the same calculation was performed for the sector centred at the angle of Θ+π rad (see the two red lines in Fig. 2B) and the results were added at each value of r. At the angle Θ approximately corresponding to the longitudinal axis of the myocytes (Θ_axis), the resulting one-dimensional spectral profile showed a well-defined mitochondrial peak and sometimes its first harmonic (see Fig. 2C). In order to assess, at least semi-quantitatively, the degree of mitochondrial polarization, we estimated the area under the peak. The algorithm first searched for the absolute maximum in the user-defined search range. This range should be based on the expected range of variation of the sarcomere length, which should be quite narrow for any particular species and experimental model. We found, however, that the position of the mitochondrial peak is different between isolated cardiomyocytes (∼1.95 μm cycle⁻¹), normoxemic ventricular tissue (∼2.4 μm cycle⁻¹) and ischaemic ventricular tissue (∼2.0 μm cycle⁻¹) (see Supplemental Figs 3 and 5). When the maximum was found, the background-subtracted area within the user-defined half-width on each side of the peak (red arrowheads in Fig. 2C) was computed yielding MPA (green area in Fig. 2C). The algorithm selected the highest MPA values among all scanned angles Θ. Whenever mitochondrial peaks were detectable, the highest MPA value was always found at the angle approximately corresponding to Θ_axis. However, when mitochondrial peaks were not present (due to ΔΨ_m dissipation), the algorithm could yield a low MPA value picked at a random angle and a random spatial frequency. Such cases are easily identifiable upon close inspection and can be classified as ‘non-detectable’. Alternatively, the false-detected peaks can be tolerated, or MPA can be re-measured at a fixed, user-defined Θ and spatial frequency. In our experience, the choice among these options did not affect the principal conclusion of a significant ΔΨ_m loss. The values of input parameters used in this study are shown in Supplemental Table 1.

Figure 2. Spectral analysis of F_TMRM in confocal images obtained from subepicardial ventricular myocytes in intact heart.

A, F_TMRM image. Θ_axis, the angle of the longitudinal axis of the myocytes. B, SFT of the image shown in A. ΔΘ, the half-angular width of the sector (red lines) centred at Θ_axis and used for computing the linear spectral profile shown in C. C, the spectral profile obtained from the SFT spectrum shown in B. The x axis is the size of spatial features, from the largest (450 μm) to the smallest (0.9 μm). The profile reveals the mitochondrial peak and its first harmonic. The position of the mitochondrial peak is at 2.4 μm cycle⁻¹ with corresponds to the sarcomere length. The algorithm searched for the maximum in the user-defined search range and then computed the area under the peak within a user-defined half-width, subtracting the background level. MPA, mitochondrial peak area.

Sensitivity of MPA, mean F_TMRM, and SD/mean of F_TMRM to mitochondrial uncoupling by FCCP

In the past, either mean F_TMRM or standard deviation of F_TMRM divided by the mean (SD/mean of F_TMRM, a measure of spatial heterogeneity; Duchen et al. 2003; Brennan et al. 2006) were used for tracking changes in ΔΨ_m. In order to compare MPA to these other metrics, we first had to ensure stability of these parameters at baseline conditions. Supplemental Fig. 3 shows that even during long perfusion with TMRM (of the order of 1 h), mean F_TMRM did not reach steady state and kept increasing. In contrast, after TMRM washout all three parameters were reasonably stable (see Supplemental Fig. 4), even though MPA and SD/mean of F_TMRM typically fluctuated more than mean F_TMRM.

Figure 3 shows results from an experiment in which the response of MPA to mitochondrial uncoupling by FCCP was compared to that of mean F_TMRM and SD/mean of F_TMRM. The three parameters were measured as functions of FCCP concentration, laser power, and the size of the region of interest (ROI). The top row in Fig. 3A shows the full image under control conditions and indicates how the large ROI and the small ROI were selected from the full image. The small ROI roughly encompassed just two myocytes. Note that for the purposes of computing MPA for ROIs, the black peripheral areas were included in the image in order to preserve the spatial frequency resolution. The bottom row in Fig. 3A shows the full image recorded at different FCCP concentrations. All images shown in Fig. 3A were obtained at a laser power of 0.48 mW (40% of maximum power of 1.2 mW). One can see that the brightness of the images changes very little up to an FCCP concentration of 5 μm, and then decreases dramatically at 20 μm FCCP. Figure B–D shows mean F_TMRM, SD/mean of F_TMRM and MPA, respectively, for different FCCP concentrations and ROIs. There were slight differences between measurements of the same parameter in different ROIs, reflecting local heterogeneities in the image features. Note, however, the principally different response to increasing levels of mitochondrial uncoupling between mean F_TMRM, SD/mean F_TMRM, and MPA. In agreement with visual perception of images shown in Fig. 3A, mean F_TMRM changed very little, at most by 10%, at FCCP concentrations up to 5 μm and then dropped abruptly to an almost background level at 20 μm FCCP (see Fig. 3B). SD/mean of F_TMRM exhibited a slightly higher sensitivity, with 15–20% decrease between 0 and 5 μm FCCP followed by a dramatic decrease between 5 and 20 μm FCCP (Fig. 3C). In contrast, the largest decrease in MPA value (40–50% depending on the ROI) occurred in response to 1 μm FCCP (Fig. 3D). Subsequent increases in FCCP concentrations to 5 and 20 μm caused further progressive decreases in MPA, by an additional 20–30%, approaching zero at 20 μm.

Figure 3. The dependence of MPA, mean F_TMRM and SD/mean F_TMRM on FCCP concentration, laser power, and the size of the region of interest.

A, confocal images of F_TMRM. Top row, left to right: three different areas (full image, large ROI and small ROI) in which the parameters were calculated. Bottom row, left to right: full image at 1, 5 and 20 μm FCCP. At each FCCP concentration, the parameters of TMRM fluorescence were calculated for the full image and the two ROIs. B–D, mean F_TMRM, SD/mean F_TMRM and MPA, respectively, computed for the full image and the two ROIs in the absence and the presence of FCCP (1, 5 and 20 μm). E–G, mean F_TMRM, SD/mean F_TMRM and MPA, respectively, computed for the full image in the absence and in the presence of FCCP (5 μm) as the function of laser power.

We also assessed how the sensitivity of mean F_TMRM, SD/mean F_TMRM and MPA to mitochondrial depolarization caused by 5 μm FCCP depends on the variation in laser power (Fig. 3E–G, respectively). The data are from the ‘full image’ (see Fig. 3A, top row, left image). As expected, mean F_TMRM linearly decreases with attenuation of laser power, approaching the threshold of detection at 10% of maximum power (see Fig. 3E). Note that the difference in the mean F_TMRM in the absence and in the presence of 5 μm FCCP is minor, and is further diminished at reduced levels of laser power. SD/mean F_TMRM (Fig. 3F) showed a slightly better sensitivity, again dependent on the laser power. However, MPA (shown in Fig. 3G) showed a significant improvement in sensitivity over SD/mean F_TMRM. Even though MPA also tended to decrease with decreasing laser power (which might be due to increasing contribution of background noise to the Fourier spectrum), the mitochondrial depolarization caused by 5 μm FCCP was still clearly detectable, even under the low-light conditions observed at the lowest laser power, due to a decrease of ∼63% in the value of MPA.

Supplemental Figs 5–8 provide additional examples of FCCP effects on mean F_TMRM and MPA under various experimental conditions. Note that the invariable decrease of mean F_TMRM upon application of FCCP seen in Supplemental Figs 5, 7 and 8 rules out the presence of TMRM quenching (Duchen et al. 2003) in our experiments. Regarding the dose dependence of the FCCP effect, the smallest concentration of FCCP having detectable effect was between 300 nM and 1 μm, which is similar to published data (Brennan et al. 2006). However, even at the concentration of 5 μm we could still see residual mitochondrial peaks (see Supplemental Figs 6 and 8), suggesting the presence of residual mitochondrial polarization. The peaks were no longer visually discernible from noise when FCCP concentration was raised to 20 μm (see Supplemental Fig. 6) or when the glycolytic inhibitor iodoacetate was added after 5 μm FCCP (see Supplemental Fig. 8). Thus, it appears that a relatively high concentration of FCCP (>5 μm) is necessary to fully dissipate ΔΨ_m and that anaerobic glycolysis might contribute to ΔΨ_m maintenance in partially uncoupled mitochondria. Note also that, even though in isolated cells both MPA and mean F_TMRM responded relatively quickly to ΔΨ_m depolarization by FCCP, in whole hearts MPA still responded fast, but mean F_TMRM responded very slowly (see Supplemental Fig. 8). The slower speed of TMRM loss from cells within the tissue might be due to the presence of additional diffusional barriers for TMRM in the intact heart. In summary, our data suggest that in whole hearts MPA is a more sensitive marker of mitochondrial polarization than the mean F_TMRM and SD/mean F_TMRM, and it is conceivable that the absence of mitochondrial peak (MPA ≍ 0) signifies a complete or near-complete ΔΨ_m loss.

The dynamics of MPA, mean F_TMRM and SD/mean of F_TMRM during no-flow ischaemia

Confocal imaging during ischaemia–reperfusion was seriously confounded by massive changes in the heart shape, which invariably occurred upon the onsets of both ischaemia and reperfusion (see Fig. 5 below and Supplemental Fig. 9). Therefore, it was virtually impossible to keep the same field of view throughout the entire ischaemia–reperfusion protocol. We lost the pre-ischaemic field of view after the onset of ischaemia in seven out eight cases, and we lost the ischaemic field of view after the onset of reperfusion in eight out of eight cases. In none of the experiments were we able to maintain the same field of view throughout the entire ischaemia–reperfusion protocol. The average duration of blind periods were 8.1 ± 1.9 min and 4.6 ± 3.2 min upon the onset of ischaemia and resumption of flow, respectively.

Figure 5. The dynamics of mean F_TMRM recorded using an optical mapping setup during global ischaemia and reperfusion.

A–C, recordings obtained from 3 different hearts. The measured parameters included the projected size of the ventricles (pixel count of the mapped area), mean F_TMRM, total F_TMRM, and examples of recordings obtained from a fixed pixel (i.e. the fixed position in the CCD sensor regardless of the changes in the heart shape) and a fixed location (i.e. the same physical point on the heart surface which is projected on different pixels as the heart shape changes). Note that the total F_TMRM decreased in all hearts at least partially due to the decrease in the size of the ventricles. However, the dynamics of other types of F_TMRM measurements were in general inconsistent.

Despite these limitations, the MPA analysis enabled us to extract definitive information about the dynamics of ΔΨ_m depolarization and recovery during ischaemia–reperfusion. Figure 4 shows an example of confocal imaging of F_TMRM before, during, and after a 70 min episode of no-flow ischaemia. The plots in Fig. 4A shows the time course of MPA (green), mean F_TMRM (blue), and SD/mean F_TMRM (light blue). The gaps in the curves indicate changes in the field of view upon both the start of ischaemia and the start of reperfusion, and the ‘blind periods’ during which no stable images could be obtained. Figure 4Ba–f shows the spectral profiles and mitochondrial peaks at the respective time points a–f indicated in panel A. Before ischaemia, a robust mitochondrial peak can be seen in the SFT spectrum (see Fig. 4Ba). The field of view was re-established after 10 min of ischaemia. Between 10 and 25 min of ischaemia the value of MPA was close to the pre-ischaemic value, despite the fact that the field of view was different. Note, however, that the position of the peak changed from 2.39 μm cycle⁻¹ to 1.98 μm cycle⁻¹ (compare panels Ba and Bb), which is consistent with a decrease in sarcomere length and shrinking of the heart (see Supplemental Fig. 9). After 25 min of ischaemia the MPA began to drop, a process that took about 36 min to reach the level below detectability at approximately 60 min of ischaemia (see Fig. 4Bc). Although mean F_TMRM and SD/mean F_TMRM also decreased during ischaemia, MPA exhibited the largest fractional decrease and, unlike mean F_TMRM and SD/mean F_TMRM, MPA dynamics enabled a definitive statement of complete or near-complete ΔΨ_m loss at around 60 min of ischaemia.

Figure 4. The dynamics of MPA, mean F_TMRM and SD/mean F_TMRM F_TMRM during global ischaemia and reperfusion.

A, the plots of MPA (green), mean F_TMRM (blue) and SD/mean F_TMRM (light blue) before, during and after 70 min episode of ischaemia. Continuous curves indicate recordings from the same field of view. The field of view had to be changed during early ischaemia and early reperfusion because of macroscopic changes in the heart shape. Note that MPA rapidly decreases to an undetectable level between 50 and 60 min of ischaemia and recovers to the level comparable to that before ischaemia within 5 min of reperfusion. Changes in mean F_TMRM and SD/mean F_TMRM during ischaemia–reperfusion are much less prominent and inconclusive. Panels Ba–f show the spectral profiles and mitochondrial peaks at the respective time points indicated in A. Note that the mitochondrial peak shifts to a shorter spatial period (1.98 μm cycle⁻¹) during ischaemia and returns to the pre-ischaemic position (2.38 μm cycle⁻¹) upon reperfusion.

Upon reperfusion, the field of view had to be re-established four times due to the continuous expansion of the heart. The earliest reperfusion image was taken at 2 min of reperfusion (see point d in Fig. 4A and the respective spectrum in Fig. 4Bd) and already showed a robust recovery to near pre-ischaemic levels of all measured parameters. The second and third post-ischaemic fields of view yielded somewhat discordant fluctuations of mean F_TMRM and SD/mean F_TMRM reflecting a high dependence of these parameters on the specific features of individual fields of view. The fourth field of view was retained for another 50 min (between points e and f in Fig. 4A) without showing any major changes either in MPA, or mean F_TMRM, or SD/mean of F_TMRM. Note that during this time frame MPA and SD/mean of F_TMRM maintained highly stable levels suggestive of a maintained recovery whereas mean F_TMRM showed stable low level, which per se could be interpreted as the lack of recovery of ΔΨ_m upon reperfusion.

In six out of seven experiments MPA exhibited a dramatic decrease during ischaemia approaching values comparable to those induced by 5–20 μm of FCCP, but only after a delay ranging between 30 and 70 min (see Supplemental Fig. 10). Thus, MPA analysis suggests a much later timing of ΔΨ_m collapse during myocardial ischaemia than reported previously (Lyon et al. 2010). Sampling of MPA in multiple fields of view during reperfusion suggested spatially heterogeneous and incomplete recovery of ΔΨ_m, not entirely unexpected after these long episodes of global ischaemia (see Supplemental Fig. 10).

Cell-to-cell heterogeneities in parameters of TMRM fluorescence

Although the MPA method was mainly tested on multicellular fields of view, differences in the TMRM parameters between individual myocytes were frequently observed and could be potentially quantified in terms of mean F_TMRM, SD/mean F_TMRM and MPA. In qualitative terms, two principally different alterations of TMRM fluorescence could be observed during I–R at the level of individual cells. The first pattern, and by far the most widespread during ischaemia, was a radical decrease in MPA amid a small (<20%) decrease in mean F_TMRM (see Fig. 4 above and Supplemental Fig. 11 and 12). Visually such pattern was perceived as a transition from a ‘granular’ to a ‘smooth’ cell appearance, which could initially occur in individual cells and propagate as a wave (see Supplemental Video 1). MPA computed for the area exhibiting wave-like mitochondrial depolarization showed a monotonic decrease of MPA (see Supplemental Fig. 11). Thus, in essence, MPA decreased whenever any portion of the analysed region exhibited ΔΨ_m loss.

The second pattern was an abrupt appearance of individual myocytes or small groups of myocytes exhibiting a radical decrease in both mean F_TMRM and MPA (see Supplemental Fig. 13 and 14). Such ‘dropouts’ were observed in five out of eight hearts at late stages of ischaemia or during reperfusion and are probably similar to those reported previously.(Matsumoto-Ida et al. 2006; Slodzinski et al. 2008; Davidson et al. 2012) The abrupt and catastrophic loss of TMRM from ‘dropouts’ appears to be a distinct and independent phenomenon from the slow and more global decrease in MPA not accompanied by a major dye loss (see more in Discussion).

The dynamics of F_TMRM derived from optical mapping during mitochondrial uncoupling and I–R

Previous reports using wide-field optical mapping indicated large changes in F_TMRM within the first few minutes of no-flow ischaemia (Lyon et al. 2010), which is in apparent contradiction with the confocal data presented above. Optical mapping provides data averaged over many cells, and at a larger spatial scale (millimetres). To test the possibility that the signs of early mitochondrial depolarization can be detected at this larger scale, we investigated the dynamics of spatially averaged F_TMRM using a standard optical mapping technique in the same species (rabbit) and under exactly the same experimental conditions as we used in confocal experiments described above.

Whereas application of FCCP (5 μm) in normally perfused hearts induced an expected (albeit slow) decrease of mean F_TMRM in optical maps (see Supplemental Fig. 15), no unequivocal conclusion could be made with regard to changes in F_TMRM in hearts subjected to global I–R. The interpretation of these data was seriously confounded by shrinking of the heart upon the onset of ischaemia and recovery of the shape (sometimes incomplete) upon reperfusion (see Supplemental Fig. 9). Figure 5A–C shows results from three different hearts subjected to I–R. In each heart we tracked changes in the size of the ventricles (as the number of pixels covering the projected image) in parallel to changes in F_TMRM. Four different F_TMRM measurements are shown for each heart: total F_TMRM (fluorescence integrated over the mapped area); mean F_TMRM (total F_TMRM divided by the pixel count); fixed pixel F_TMRM (fluorescence recorded from a single pixel with fixed coordinates) and fixed location F_TMRM (our best effort to track fluorescence coming from the same physical point on the surface of the heart, based on anatomical landmarks). One can see that the total F_TMRM decreased during ischaemia, closely following the decrease in the projected surface area. At the same time, the mean F_TMRM did not show any consistent change. Likewise, measurements from a ‘fixed pixel’ and even ‘fixed location’ could show either an increase or a decrease during ischaemia. Amid these inconsistent outcomes, we concluded that, at least in our experimental model, F_TMRM derived from wide-field optical mapping could not faithfully represent ΔΨ_m changes during ischaemia.

Discussion

In this methodological study we developed and validated a novel spectral method (the MPA method) for tracking significant changes in ΔΨ_m during I–R in confocally imaged ventricular tissue. Below we will discuss the strengths and limitations of this method in comparison to other approaches, as well as general limits of our ability to image ΔΨ_m during real ischaemia–reperfusion.

The major advantage of the MPA method is that it is highly sensitive and selective to the accumulation of the dye in the mitochondrial matrix. The method takes advantage of the mitochondrial packaging specific to cardiac myocytes, such that many (although not all) mitochondria are constrained between the z-disks (Vendelin et al. 2005). When the mitochondria are well polarized, the spatially periodic increases in F_TMRM corresponding to the mitochondrial arrays (see Fig. 1Ba) create a distinct mitochondrial peak in the spatial Fourier spectrum (see Fig. 1Ca). The observation that this peak fully disappears upon mitochondrial uncoupling by a high concentration of FCCP (see Fig. 1Cb) provides strong evidence that the absence of the peak heralds a complete or almost complete loss of ΔΨ_m. While we cannot completely exclude the possibility that mitochondrial fusion/fission may affect the MPA value, available evidence suggests a quantitatively modest change in the mitochondrial size distribution during ischaemia in the whole heart (Ong et al. 2010), which can hardly explain the essentially all-or-none changes in MPA observed during I–R in our study.

To the best of our knowledge, none of the previous approaches afforded such a definitive conclusion with regard to the event of ΔΨ_m loss. Since the MPA method relies on the specific pattern of mitochondrial packaging in sarcomeres, it is limited to myocardium and presumably skeletal muscle where mitochondrial organization is somewhat similar (Vendelin et al. 2005). However, knowledge of specific shape/size constraints of mitochondria in other cell types can potentially help to develop image-processing approaches for a more mitochondria-specific interpretation of those fluorescent dyes which are present in multiple cellular compartments (Gerencsér & Adam-Vizi, 2001).

A serious obstacle for continuous assessment of MPA in the same field of view during I–R, inherent to microscopic imaging in ischaemic hearts, is the large change in the heart shape accompanying both ischaemia and reperfusion. This phenomenon effectively prevented us from tracking the same group of cells throughout the entire ischaemia–reperfusion protocol in the vast majority of hearts. It should also be noted that tissue drift (albeit at a smaller scale) usually occurred during the rapid phase of ischaemic mitochondrial depolarization (see for example Supplemental Video 1) and could also diminish our ability to make a definitive statement regarding the timing of ΔΨ_m loss in a specific group of cells. Note also that the tissue displacement, in general, occurred not only in the plane of view but also in depth, further complicating tracking the same group of cells during relevant events. Our approach to alleviate this problem was acquiring z-stacks (10 to 15 slices, at ∼3 μm steps) and then automatically selecting the slice with the largest value of MPA (other criteria of selection, including algorithms of image correlation and visual recognition of image traits, can be applied). Trade-offs of acquiring z-stacks include reduced maximal temporal resolution and increased laser exposure. A recently reported approach for automated real-time tracking/stabilization of confocal images in whole hearts (Schroeder et al. 2010), combined with the MPA method described here, may provide an ultimate tool for the analysis of ΔΨ_m dynamics during I–R in whole hearts. Also, multi-photon imaging may improve the robustness of the MPA technique due to inherently sharper spatial resolution and decrease in blur, especially in deeper myocardial layers.

Another significant limitation of MPA is the lack of a simple method to calibrate MPA values in terms of actual millivolts of ΔΨ_m. Again, this limitation is shared with all prior work in perfused myocardial preparations (or in isolated cardiomyocytes, for that matter). In that regard, we presented the data of MPA titration with different concentrations of FCCP (Fig. 3 and Supplemental Fig. 4), which may serve as a semi-quantitative approach for interpretation of MPA values. A dose–response function of ΔΨ_m versus[FCCP] can be obtained in a mitochondrial assay (Scaduto & Grotyohann, 1999). The validity of such calibration, however, will rely on the (rather strong) assumption that isolated mitochondria have the same sensitivity to FCCP as the mitochondria in intact myocardium. In any case, values of MPA obtained during ischaemia can be compared to those obtained in the same experimental model in the presence of FCCP. In previous studies FCCP concentrations in the micromolar range were considered to be sufficient to completely dissipate ΔΨ_m (Johnson et al. 1981; Petronilli et al. 2001; Duchen et al. 2003). In our study 1 μm FCCP decreased MPA by 50%, and 5 μm FCCP brought MPA values to the level of 136 ± 21 but still did not fully eliminated mitochondrial peaks. At the end of ischaemia, we observed MPA values below 136 in all experiments (see Supplemental Fig. 10), and in some cases the mitochondrial peak was visually undetectable (comparable to the effect of 20 μm FCCP). Thus, we can conclude that during ischaemia the degree of ΔΨ_m depolarization was comparable to, or exceeded that, of ΔΨ_m depolarization caused by 5 μm FCCP.

It is more difficult to form conclusions regarding intermediate levels of MPA. Even during normal perfusion, the variability of MPA was quite high, ranging from 550 to 210 (see Supplemental Fig. 10). Whereas we cannot exclude heterogeneities in ΔΨ_m at baseline, we have an impression that this variability was mostly due to variability in the quality and sharpness of the image. Note, however, that in those cases when we were able to maintain the same field of view during ischaemia, the observation of a continuous monotonic decrease in MPA provided valuable information about the timing and the speed of ΔΨ_m loss and recovery. For example, the ischaemic data obtained in this study suggest that a significant ΔΨ_m depolarization occurs relatively late in ischaemia (after ∼30 min) and is a relatively slow process (takes at least 10–15 min), but can be essentially complete by 60–70 min of ischaemia. However, even after such a prolonged time of ischaemia, a relatively fast recovery of ΔΨ_m could be observed upon reperfusion at least in some hearts, without evidence of massive and fatal ΔΨ_m loss up to 1 h following reperfusion. Also, the recovery upon reperfusion appeared to be heterogeneous (see Supplemental Fig. 10).

The comparative analysis of the dynamics of MPA and mean F_TMRM gave us a new perspective on previous observations in the setting of I–R in whole hearts. Several prior publications were focused on the ‘dropouts’– cells exhibiting ‘rapid, complete, and irreversible’ (Matsumoto-Ida et al. 2006) loss of mean F_TMRM during I–R (Matsumoto-Ida et al. 2006; Slodzinski et al. 2008; Davidson et al. 2012). We also observed accidental ‘dropouts’ during I–R in our experiments (see Supplemental Fig. 13 and 14). The abrupt and catastrophic decline in both MPA and mean F_TMRM observed in ‘dropouts’ appears to be a distinct and independent phenomenon from the slow and global decrease in MPA (which reflected mitochondrial dye loss) amid only a minor change in mean F_TMRM. Why the ‘dropouts’ lose TMRM so quickly from the cytoplasm is not clear, but might reflect a significant loss of sarcolemmal polarization and/or integrity in addition to the loss of the mitochondrial potential. Further studies are needed to resolve the timing and possible cause–effect relationships between critical mitochondrial and sarcolemmal events during I–R.

We should also comment on the previously published notion of ‘waves of mitochondrial membrane potential collapse’ occurring within the first 5 min of ischaemia (Lyon et al. 2010). The timing of ΔΨ_m loss during ischaemia is relevant to explaining electrical depression and arrhythmias during early ischaemia or cardiac arrest, which might be related to mitochondrial depolarization (Akar et al. 2005). We have to mention that, based on our study, any conclusions regarding fluorescence changes in this very acute phase of global ischaemia are severely confounded by the attendant changes in the heart shape (see Supplemental Fig. 9). In our confocal experiments we were unable to track myocytes during this stage, but even after 10–15 min of ischaemia all the measured TMRM parameters (mean F_TMRM, SD/mean of F_TMRM, and MPA) were typically not different from those before ischaemia. At the same time, a major decrease in MPA was typically observed only after 30–40 min of ischaemia (see Supplemental Fig. 10). Moreover, performing optical mapping experiments in the same species (rabbit) and using the same experimental conditions as in our confocal experiments, we realized that the complex dynamical changes in the projection of the heart image to the imaging sensor, in the presence of inherent heterogeneities in dye distribution and illumination, makes the local estimate of F_TMRM dynamics dependent on a large number of unknown geometric factors. As a result, changes in per-pixel F_TMRM signal cannot be disentangled from changes in the heart geometry. Thus, the timing and the dynamics of ΔΨ_m loss/recovery during I–R in whole hearts remain to be settled.

Summary and conclusions

Quantification of mitochondrial spectral peaks derived from confocal TMRM images may be the most sensitive method for tracking significant ΔΨ_m changes during no-flow ischaemia and reperfusion in intact hearts. Using this method, we conclude that a significant ΔΨ_m depolarization occurs no earlier than 30 min of ischaemia, and at least a partial recovery of ΔΨ_m occurs upon reperfusion following 60+ min of ischaemia. This phenomenon is different from a catastrophic and irreversible loss of F_TMRM in individual myocytes (‘dropouts’) amid neighboring cells maintaining high F_TMRM level. The relationship between the two ‘modes’ of presumed ΔΨ_m depolarization remains unclear. In the isolated Langendorff-perfused rabbit heart, wide-field imaging of F_TMRM can be used for detection of ΔΨ_m loss in the presence of coronary perfusion, but does not provide reliable information during no-flow ischaemia and subsequent reperfusion.

Glossary

ΔΨ_m

mitochondrial membrane potential

FCCP

carbonylcyanide-p-trifluoromethoxyphenyl hydrazone (a protonophore and potent uncoupler of oxidative phosphorylation)

F_TMRM

fluorescence of TMRM

mean F_TMRM

spatially averaged fluorescence of TMRM

IAA

sodium iodoacetate (an inhibitor of anaerobic glycolysis)

I–R

ischaemia–reperfusion

LV

left ventricle

MPA

mitochondrial peak area

ROI

region of interest

RV

right ventricle

SD/mean of F_TMRM

standard deviation of TMRM fluorescence over space divided by the mean

SFT

spatial Fourier transform

TMRM

tetramethylrhodamine methyl ester (a cationic fluorescent dye)

Author contributions

Experiments described in this article were performed in the laboratory of Dr Zaitsev at the Nora Eccles Harrison Cardiovascular Research and Training Institute (CVRTI), University of Utah, Salt Lake City, Utah. The Zeiss confocal system is a shared facility at the CVRTI. The authors’ contributions are as follows. Conception and design of the experiments: A.V.Z., P.W.V., K.W.S.; collection, analysis and interpretation of data: P.W.V., K.W.S., A.V.Z., M.W., J.S., K.J.S., J.Z., T.G.T.; drafting the article or revising it critically for important intellectual content: P.W.V., A.V.Z., K.W.S., T.G.T. All authors read and approved the final version of the manuscript.

Supplementary material

Supplemental Figure 1.

Supplemental Figure 2.

Supplemental Figure 3.

Supplemental Figure 4.

Supplemental Figure 5.

Supplemental Figure 6.

Supplemental Figure 7.

Supplemental Figure 8.

Supplemental Figure 9.

Supplemental Figure 10.

Supplemental Figure 11.

Supplemental Figure 12.

Supplemental Figure 13.

Supplemental Figure 14.

Supplemental Figure 15.

Supplemental Table 1.

Supplemental Video 1.