OUP user menu

Slow calcium waves and redox changes precede mitochondrial permeability transition pore opening in the intact heart during hypoxia and reoxygenation

Sean M. Davidson, Derek M. Yellon, Michael P. Murphy, Michael R. Duchen
DOI: http://dx.doi.org/10.1093/cvr/cvr349 445-453 First published online: 23 December 2011


Aims Opening of the mitochondrial permeability transition pore (mPTP) is an important step on the pathway towards cardiomyocyte death, defining the extent of injury following cardiac ischaemia and reperfusion. In isolated mitochondria, mPTP opening is triggered by calcium overload facilitated by oxidative stress. In isolated cells, however, it has been suggested that mPTP opening occurs before calcium overload and is stimulated by oxidative stress. Our objective was to establish the events that cause mPTP opening in the intact heart.

Methods and results We performed multiphoton imaging of Langendorff-perfused mouse hearts expressing an inducible, Ca2+-sensitive reporter (circularly Permuted GFP and calmodulin (CaM), version 2), to examine the spatiotemporal relationship between [Ca2+]c, redox state, and mPTP opening in the intact heart during hypoxia and reoxygenation at sub-myocyte resolution. We found that during reperfusion, calcium waves propagated across multiple cells at 3.3 µm/s. mPTP opening caused an abrupt loss of mitochondrial membrane potential, measured using a potentiometric dye, which was invariably preceded by a rise in [Ca2+]c. The probability that localized [Ca2+]c waves led to mPTP opening was greater early during reoxygenation. During reoxygenation, coordinated redox changes also occurred across large regions and preceded mPTP opening on average by 122 ± 38 s. Fewer [Ca2+] waves led to mPTP opening in the presence of mPTP inhibitor cyclosporin A or mitochondrial-targeted scavenger of reactive oxygen species, MitoQ.

Conclusion These experiments define the spatiotemporal relationship between changes in [Ca2+]c, redox state and mPTP opening during reoxygenation in the intact heart. Tissue oxidation coincident with localized calcium waves together conspire to cause mPTP opening and subsequent cell death.

  • Calcium
  • Redox state
  • Mitochondria
  • Hypoxia
  • Reoxygenation

1. Introduction

Occlusion of a coronary artery causes cardiac ischaemia and myocardial infarction. Reperfusion is necessary to salvage the myocardium, but can paradoxically cause further damage itself, generating ischaemic-reperfusion injury.1 Opening of the mitochondrial permeability transition pore (mPTP) is an irreversible step on the pathway towards cardiomyocyte death during reperfusion.2,3 Radio-tracer experiments have demonstrated that the majority of mPTP opening occurs early during reperfusion,2,4 although by direct imaging, on-going mPTP opening was seen throughout 60 min of reperfusion.5 The factors leading to mPTP opening have been extensively studied in isolated mitochondria, and to a lesser degree in intact cardiomyocytes, but have not been extensively studied in the intact heart. In isolated mitochondria, two critical factors leading to mPTP opening are mitochondrial calcium overload coupled with oxidative stress. However, in intact cells, the temporal relationship between [Ca2+] and mPTP is not as clear-cut. For example, in primary adult cardiomyocytes, [Ca2+] in both the cytosol and in mitochondria returns rapidly to baseline levels early during reperfusion,68 and the subsequent [Ca2+]c and [Ca2+]m overload which occurs has been suggested to be a consequence, rather than a cause, of mPTP opening.6 Emerging evidence suggests that after cellular injury, mPTP opening in cardiomyocytes is largely caused by reactive oxygen species (ROS) produced by mitochondria, rather than an increase in [Ca2+].6,9 Ischaemic contracture has been suggested to contribute to reperfusion injury by generating damaging forces, but this can only be studied in the intact heart.6,10 Only by understanding the precise sequence of events that converge to cause mPTP opening in the intact heart can we hope to design appropriate protective strategies or understand how cardioprotective modalities such as ischaemic preconditioning and postconditioning work.

Spectrofluorometry has been used to measure global [Ca2+]c in the intact, perfused heart.1114 As expected, [Ca2+]c increases transiently during each action potential. During severe ischaemia, these Ca2+ transients disappear as cardiac contraction ceases. In the absence of oxygen, ATP supplies are exhausted within about 20 min, and diastolic [Ca2+]c increases progressively reaching values as high as 3–4 µM resulting in ischaemic contracture.15 During reperfusion of perfused hearts, as in isolated cardiomyocytes, the recovery of diastolic [Ca2+]c is actually relatively rapid,7,14 returning to baseline as the heart resumes beating, within 5 min of the flow resuming.13 [Ca2+]m appears to be tightly controlled, since it regulates mitochondrial metabolism.16

Thus, the exact spatiotemporal relationship between changes in [Ca2+]c, redox changes, and mPTP opening during reperfusion has not been determined in the intact heart. A limitation of spectrofluorimetric measurements is that the measurement is restricted to spatially averaged measurements of [Ca2+]c that obscure the responses of individual cells. In the present study, we therefore performed multiphoton imaging of Langendorff-perfused mouse hearts expressing circularly Permuted GFP and calmodulin (CaM), version 2 (GCaMP2), a genetically encoded, inducible, Ca2+-sensitive reporter. Simultaneously, we used the mitochondrial potential-sensitive dye, tetramethyl rhodamine methyl ester (TMRM), to identify mPTP opening. This method has been previously verified5 and was confirmed here by the use of the mPTP inhibitor cyclosporine A (CsA). In this way, we could examine the spatial and temporal relationship between [Ca2+]c and mPTP opening. These studies were undertaken in the intact heart during hypoxia and reoxygenation at sub-myocyte resolution. Here, multiphoton microscopy has several advantages over standard confocal microscopy including better resolution at greater tissue depths and reduced photo-damage.17 The GCaMP2 Ca2+ reporter is sensitive, fast enough to detect cellular calcium transients and is stable at 37°C.18 Furthermore, using the fluorescent signal from NAD(P)H as a readout of cellular redox status, we were able to investigate the timing of redox changes, and determine directly whether localized [Ca2+]c increases precede (and therefore trigger) mPTP opening, or are consequences of mitochondrial Ca2+ release after mPTP opening.

2. Methods

2.1 Multiphoton imaging

Animal experiments were in accordance with the United Kingdom Home Office Guide on the Operation of Animal (Scientific Procedures) Act of 1986. The investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health. Mice with cardiac expression of the cytosolic, calcium reporter GCaMP218 were terminally anaesthetized by ip injection of 160 mg/kg Na.pentobarbitone. Once pedal pinch reflexes were completely inhibited, hearts were removed and Langendorff-perfused with a solution containing 100 nmol/L TMRM which labels respiring mitochondria and is lost upon mPTP opening5 (see Supplementary material online). Hearts were perfused for 45 min with a hypoxic, substrate-depleted buffer followed by 45 min perfusion with normoxic buffer. Twenty micromole per litre blebbistatin (an inhibitor of actin–myosin cross-bridge cycling)19 was added to prevent contraction and movement of muscle (Supplementary material online). Multiphoton images were obtained on a Zeiss 510 NLO microscope, with a laser tuned to 990 nm, allowing simultaneous detection of GCaMP2 (emission at 500–550 nm) and TMRM (emission 575–640 nm), or 720 nm allowing simultaneous imaging of reduced NAD(P)H autofluorescence (emission at 435–485 nm) and TMRM (emission > 515 nm) (Supplementary material online). During reoxygenation, 0.2 µmol/L CsA or 100 nmol/L MitoQ was added to the perfusate.

2.2 Isolated cardiomyocytes

Primary adult cardiomyocytes (used in Supplementary material online, Results) were isolated from rats that had been terminally anaesthetized by single ip injection of 130 mg/kg Na.pentobarbitone. Once pedal pinch reflexes were completely inhibited, the chest was opened, the heart rapidly removed, and perfused on Langendorff apparatus. Cells were isolated by standard collagenase perfusion (Supplementary material online).

2.3 Statistics

Average results are presented with standard error of the mean, and compared using t-test or Z-test for comparing two proportions, where appropriate.

3. Results

Mouse hearts expressing the previously described calcium reporter GCaMP2,18 were isolated and perfused with a normoxic buffer containing TMRM, which distributes across the membrane of healthy mitochondria according to their membrane potential (ΔΨm). In normoxic-perfused hearts, fluorescence from the calcium reporter, TMRM, and NAD(P)H, remained constant throughout the experiment (Figure 1E and G). Repetitive [Ca2+]c transients were observed synchronously throughout the heart as expected (Figure 1A and B), but no other spontaneous [Ca2+]c activity was observed. We subjected hearts to 45 min oxygen–glucose depletion followed by 45 min oxygenated perfusion and visualized infarct formation in real time using multiphoton microscopy. As expected, cardiomyocytes remained viable for some time after mPTP opening during reoxygenation, as demonstrated by perfusion with fluorescent calcein, which diffuses throughout the heart, but not across the intact plasma membrane of depolarized cardiomyocytes (Figure 1C). However, mPTP opening was inevitably followed by necrotic cell death, as demonstrated by propidium iodide staining of nuclei in cells with compromised plasma membrane integrity (Figure 1D).

Figure 1

(A) Low magnification image of an isolated, perfused heart, perfused with TMRM to indicate mitochondrial membrane potential (ΔΨm), in addition to green GCaMP2 fluorescence reporting [Ca2+]c, and NAD(P)H autofluorescence reporting redox state. The white dots indicate the border of one cardiomyocyte. Bar 100 µm. (B) Higher power image as per panel A, demonstrating localization of TMRM (red) and NAD(P)H (blue) within individual mitochondria of cardiomyocytes. Bar 10 µm. (C) A heart was perfused with TMRM (red) and free calcein (green fluorescence) after hypoxia and reoxygenation. Calcein enters the interstitium and blood vessels, but not cardiomyocytes lacking TMRM signal, demonstrating that the plasma membrane remains intact in the minutes after mPTP opening. Bar 100 µm. (D) Perfusion with the vital dye, propidium iodide late during reoxygenation labels the nuclei of dead cardiomyocytes (red, arrows) that are lacking TMRM (also red). Bar 100 µm. (E) Total NAD(P)H fluorescence over the entire field did not during perfusion of a normoxic control heart. (F) The total fluorescent intensity of TMRM and GCaMP2 did not change markedly during perfusion of a normoxic control heart. (G) Total NAD(P)H fluorescence increased during simulated ischaemia and returned to baseline on reoxygenation. (H) The total fluorescent intensity of TMRM and GCaMP2 over the entire field did not change markedly during reoxygenation after 45 min of simulated ischaemia.

During reoxygenation, global [Ca2+]c did not change markedly or consistently apart from a slight increase in average fluorescence intensity when Ca2+ transients resumed (Figure 1H). Occasionally, however, dramatic increases in [Ca2+]c were detected that propagated as waves extending up to several hundred micrometres from the origin (Figure 2A and B; see Supplementary material online, Movie S1) (average distance 108 ± 1.6 µm, n= 26 Ca2+ waves), equivalent to the length of several cardiomyocytes (which have typical dimensions ∼100 µm × 20 µm aligned horizontal to the epicardial plane). The average wave-front velocity was surprisingly slow, at only 3.3 ± 0.3 µm/s, taking ∼30 s to traverse the length of a cell (Figure 2C, see also Figure 3F and H). The total duration of raised [Ca2+]c in the region lasted up to 220 s (average 76 ± 2 s, n= 26 Ca2+ waves), and occasionally initial waves were followed by further waves in the same region and with the same direction, suggesting that their initiation was due to localized disturbance in calcium handling. Based on an estimate of approximately 1000 cells per field, we calculated the frequency of waves was 9.6 ± 3.7/min per 105 cells (Figure 3C), though the probability of their occurrence was greater during the initial 5–20 min of reperfusion (Figure 3D).

Figure 2

(A) Some localized calcium waves (green) precede mPTP opening (detected as a wave of loss of TMRM signal, red; arrows) while other calcium waves do not (arrowheads). The same field is shown over the first 300 s reperfusion. Bar 100 µm. (B) An image from one heart in which all the reoxygenation calcium waves were superimposed demonstrates that they tend to occur in regions. Bar 100 µm. (C) A line scan along a region with a Ca2+ wave (green) demonstrating its rate of progression in the direction of the arrow.

Figure 3

(A) A line scan illustrating the coincidence between [Ca2+]c waves and mPTP opening (loss of red TMRM signal, arrows), while other regions experience elevated [Ca2+]c without mPTP opening (arrowheads). (B) A line scan of [Ca2+]c signal (green) illustrates progression of the wave-front in the direction of the arrow. (C) Frequency of Calcium waves leading to mPTP opening under different conditions. (D) mPTP opening is observed more frequently following [Ca2+]c flashes that are more intense and that occur early during reperfusion. (E) A region of myocardium imaged at low resolution. Line scans were generated along the regions indicated and are shown in panels FI. (FI) Line scans demonstrate calcium waves (green) that preceded mPTP opening (arrows), or resolve spontaneously (arrowheads).

mPTP opening is an irreversible step towards myocyte death and causes an abrupt loss of mitochondrial potential, which can be detected as a rapid loss of TMRM fluorescence.5 In primary cardiomyocytes as in the perfused heart, mPTP opening occurs in a distinctive wave along the length of the cardiomyocyte.5,20,21 After 44% of [Ca2+]c waves, [Ca2+]c returned to basal levels and the mPTP remained closed (ΔΨm remained stable). However, the remaining 56% of [Ca2+]c waves were followed after a mean delay of 100 ± 27 s (n= 10 hearts) by mPTP opening in the cell at the [Ca2+]c wave origin (Figure 3D). This relationship can be demonstrated by viewing the fluorescence intensity along a line scan across the heart, spread out over time (Figure 3A, B, and E–I). A line scan along a region with particularly extensive mPTP opening revealed that mPTP opening in one cell appeared to increase the likelihood of [Ca2+]c waves and mPTP opening in adjacent cells (Figure 3B and F). Interestingly, TMRM fluorescence increases in cells adjacent to the cell with mPTP opening, suggesting that they either undergo slight hyperpolarization or take up dye released from the adjacent cell (Figure 3B).

Some hearts were imaged at a lower resolution, allowing a 0.5 Hz scan rate—rapid enough to observe the progression of the waves of [Ca2+]c and subsequent mPTP opening in line scans along the cardiomyocyte long axes (Figure 3E–I, see Supplementary material online, Movie S2). These suggested that the [Ca2+]c wave could spread to and prompt mPTP opening in adjacent cardiomyocytes (Figure 3F). In cells where control of intracellular calcium was regained, the elevated [Ca2+]c gradually dissipated (Figure 3G and H). In cells where irretrievable damage occured, however, subsequent massive [Ca2+]c overload was followed by mPTP opening (Figure 3I).

In untreated hearts, an average of 9.6 ± 3.7/min per 105 cells, waves were observed during reperfusion (n= 10), 56% of which were followed by mPTP opening (Figure 3C). In contrast, in hearts that were perfused with 0.2 µM CsA, an inhibitor of mPTP opening, although the total frequency of [Ca2+]c waves was similar (8.9 ± 4.5/min per 105 cells, n= 3), only 19% of these were followed by mPTP opening (Figure 3C). Importantly, although not all [Ca2+]c waves led to mPTP opening (Figure 3C), it was always preceded by a [Ca2+]c increase. In addition, plotting the intensity of [Ca2+]c waves against the time from reoxygenation shows that a wave was more often associated with mPTP opening when it occurred at earlier times after ischaemia (Figure 3D). This prompted us to investigate whether an additional factor such as redox state was involved in mPTP opening during this time-period.

Since only the reduced forms of the NAD(P)+/NAD(P)H redox couples are fluorescent, cellular redox state can be interrogated by measuring the blue autofluorescence from NAD(P)H (Figure 1A, B, and F).22,23 We imaged NAD(P)H and TMRM simultaneously in the perfused heart (see the ‘Methods section’). As expected, global NAD(P)H fluorescence was elevated during hypoxia (confirming the effective inhibition of mitochondrial respiration), and recovered to baseline levels within the first few minutes of reperfusion (Figures 1G and 4A (single asterisk, double asterisks). During subsequent reperfusion, NAD(P)H fluorescence was relatively stable, but occasionally in large regions of the myocardium NAD(P)H became transiently oxidized. A representative line scan (double asterisks inFigure 4A) demonstrates one of these regions, which coincides with and overlaps a region where mPTP opening occurred. Interestingly, as can be seen on a graph of intensity along two different regions over time, a precipitous drop in NAD(P)H fluorescence accompanied mPTP opening (double asterisks in Figure 4A), while in regions of the myocardium where the NAD(P)H signal recovered, the mPTP remained closed (single asterisk in Figure 4A). There was a brief (<1 min) transient mitochondrial hyperpolarization immediately after NAD(P)H became reduced again, visible as a small spike in the TMRM signal (Figure 4A, arrow). A magnified image of the NAD(P)H signal shows that the region of oxidation spread out and receded as a wave originating from the damaged cell (Figure 4A, see Supplementary material online, Movie S3).

Figure 4

(A) A line scan along a region with mPTP opening (red panel, arrowhead) demonstrating a large, transient region of NAD(P)H oxidation (blue panel, arrow). The TMRM signal increases adjacent to the cell with mPTP opening (double arrowhead). A graph of intensity over time in the marked regions shows that the mPTP remains closed only where NAD(P)H recovers. A magnified region of NAD(P)H shows that the oxidized region spreads and recedes in a wave. (B) In this line scan, oxidation of NAD(P)H can be seen to precede loss of TMRM (red) by several minutes. The movement (contracture) of the myocyte (yellow dashed line) occurs several minutes before loss of TMRM, as is highlighted in a 3D plot of TMRM intensity over time.

Interestingly, in most cells, NAD(P)H oxidation occurred well before mPTP opening (average 122 ± 38 s, n= 9 events, e.g. Figure 4B) although occasionally they were detected contemporaneously (e.g. Figure 4A). This timing coincides closely to the [Ca2+]c wave described above (100 ± 27 s before mPTP opening), suggesting that they may be related. Unfortunately, it is not possible to directly determine the relationship between NAD(P)H and [Ca2+]c, since the 2-photon excitation spectra of NAD(P)H and GCaMP2 do not overlap, and the infra-red laser cannot be re-tuned rapidly enough to image both simultaneously. Occasionally, however, with fortuitous positioning of the heart in the imaging chamber, it was possible to obtain stable images without the addition of blebbistatin. In one such heart, contracture of a cardiomyocyte, which is Ca2+-dependent,24 was observed. This preceded both NAD(P)H oxidation and mPTP opening, as is evident in the 3D intensity map of TMRM over time (Figure 4B). Furthermore, in in vitro studies with isolated cardiomyocytes to examine the relationship between NAD(P)H and mPTP opening, we found that NAD(P)H fluorescence decreased progressively to a minimum at which time the mPTP opened, consistent with the inhibition of mPTP opening by NADH (see Supplementary material online, Figure S1).

The cause of the localized changes in redox status in the heart is not known, but a series of higher magnification images demonstrates a clear border of NAD(P)H oxidation (decrease in blue fluorescence) that spreads progressively from a vessel during reoxygenation (Figure 5A), indicating that regional perfusion differences may exist.

Figure 5

(A) High-resolution image of a redox wave in NAD(P)H (blue) spreading from the region of a vessel during the return to normoxia. TMRM (red) labels mitochondria in cardiomyocytes and in the cells of the vessel. Bar 100 µm. (B) Perfusion of a GCaMP2-expressing heart with tert-butyl peroxide eventually causes [Ca2+]c overload throughout the field, followed rapidly by loss of TMRM signal.

One reason for the occurrence of [Ca2+]c waves during reperfusion may be increased ROS production. We tested this by perfusion with tert-butyl hydroperoxide, however, this resulted in [Ca2+]c overload and mPTP opening that was qualitatively different, i.e.: after an extended delay, [Ca2+]c overload occurred in all cells without any evidence of [Ca2+]c waves, followed by a huge wave of mPTP opening (Figure 5B, see Supplementary material online, Movie S4). We also perfused hearts during reoxygenation with 100 nM MitoQ, a mitochondrially targeted ROS scavenger known to be cardioprotective.25 This did not change the frequency of [Ca2+]c waves significantly (4.4 ± 3.0/min per 105 cells vs. 9.6 ± 3.7/min per 105 cells n= 3, n.s.; Figure 3C); however, the probability of mPTP opening was much reduced (15% vs. 56% in control hearts, P< 0.05).

4. Discussion

At the onset of reperfusion, a wave of reoxygenation spreads from arterioles. We have shown that, during reoxygenation, highly localized, transient waves of elevated [Ca2+]c diffuse across the myocardium at an average speed of 3.3 µm/s. In 44% of cases, [Ca2+]c waves of a sufficient intensity preceded mPTP opening, particularly during the first 20 min of reoxygenation. Interestingly, cytosolic [Ca2+]c often returned to basal levels 100 ± 27 s before mPTP opening, though it is possible [Ca2+]m remained elevated. Perfusion with CsA significantly reduced the appearance of an abrupt loss of TMRM fluorescence, confirming our interpretation of these events as mPTP opening. However, CsA did not significantly affect the frequency of [Ca2+]c waves, indicating they are not a consequence of mPTP opening. Large, well-defined regions of NAD(P)H oxidation occurred on average 122 ± 38 s before mPTP opening, and appear to be an important contributing factor since scavenging of mitochondrial ROS decreased the likelihood of mPTP opening. Ca2+-dependent cellular contracture occurred before mPTP opening.

As far as we are aware, such slow [Ca2+]c waves have not been observed previously. Spontaneous waves have been observed in severely calcium-overloaded myocytes within the intact heart and in isolated cells, but these travel much more rapidly (>80 µm/s), have a much shorter duration (∼160 ms), and rarely propagate between cells.26,27 The [Ca2+]c waves observed here remain elevated for up to 3.7 min, but since the KD of the Ca2+ reporter is 146 nM,18 fluctuations of [Ca2+]c above ∼1 µM are likely to be undetectable. In faster line scans along cells with [Ca2+] overload, some [Ca2+]c oscillations were observed after the initial wave (Figure 3F–I). These either spontaneously resolved, or were followed by large [Ca2+]c increases and mPTP opening, supporting suggestions that [Ca2+]c oscillations are also an important characteristic of compromised cardiomyocytes in vivo.

4.1 Calcium waves and mPTP opening

Calcium is believed to be a crucial factor in mPTP opening and is routinely added to isolated mitochondria in vitro to assay mPTP opening, although it can accumulate to millimolar levels in cells before causing mPTP opening.28 Direct confirmation that it contributes to mPTP opening in the intact heart has not however been previously obtained. Measurements of global [Ca2+]i in isolated perfused hearts suggest that [Ca2+]i increases progressively during ischaemia after the onset of contracture, but returns rapidly to baseline within a few minutes of reperfusion.1115,29 In line with this, we did not observe any major increase in global [Ca2+]c during the reoxygenation period, apart from the reappearance of calcium transients during the first few minutes. Global [Ca2+]m has also been shown to decrease quite rapidly during reperfusion, remaining only approximately two-fold elevated throughout reperfusion.13 Our observations explain these observations by demonstrating that [Ca2+]c overload does, in fact, occur during on-going reperfusion, but it manifests in the form of slow, regional waves which are undetectable using spatially averaged measurements. Although we are unable to directly measure [Ca2+]m, it is likely that [Ca2+]m is also elevated within the region of the wave. The [Ca2+]c waves correspond precisely with those cells in which mPTP opening occurs, and the phenomena are most likely connected. The substantial delay of 100 ± 27 s between [Ca2+]c overload and mPTP opening suggests that additional factors are required.

We considered the possibility that oxidative stress stimulated [Ca2+]c wave generation, since it can increase diastolic [Ca2+]c in isolated cardiomyocytes.24 Perfusion of isolated hearts with peroxide, however, was not sufficient to stimulate [Ca2+]c waves. In isolated cardiomyocytes subject to ischaemia and reperfusion, Ca2+ overload is the consequence of bioenergetic failure after mPTP opening rather than its cause.30 However, the fact that CsA had no effect on [Ca2+]c wave generation in our experiments, argues against the involvement of the mPTP in [Ca2+]c wave generation.

In isolated cardiomyocytes, mPTP opening results in ATP depletion in Ca2+-overloaded cardiomyocytes, resulting in hypercontracture.31,32 In the intact heart, however, we found that contracture occurred before mPTP opening. This observation suggests that mitochondrial deenergization or cellular ATP depletion is a causative factor contributing to mPTP opening (Figure 6). Indeed, ATP inhibits mPTP opening in isolated mitochondria.33

Figure 6

A hypothetical model suggesting the events which lead to mPTP in the myocardium. Within the Ca2+ wave, futile Ca2+ cycling may deplete ATP leading to cardiomyocyte contracture. Independently, oxidative stress depletes NAD(P)H, further exacerbating the oxidative stress. The combination of factors, including Ca2+ overload, oxidative stress, and ATP depletion increase the likelihood of mPTP opening.

Infarcted regions in myocardium are usually contiguous rather than dispersed. The occurence of [Ca2+]c waves in neighbouring regions of mPTP opening (Figure 3B and F) suggests a propagating mechanism that may contribute to the ‘wave-front’ phenomenon of infarct expansion.34

4.2 Redox state and mPTP opening

A number of factors can dramatically influence the sensitivity of the mPTP to [Ca2+], including intracellular pH, [phosphate], [ADP], and redox state. In particular, critical protein components of the mPTP are redox regulated.35 We therefore investigated the relationship between redox state, [Ca2+]c and mPTP opening.

Myocardial ROS production peaks during the first few minutes of reperfusion,36,37 although for the most part only global measurements have been obtained in the isolated heart. The main source of ROS appears to be the mitochondria,38 and the mitochondrially targeted ROS scavenger MitoQ is cardioprotective.25 As reviewed by Lemasters et al.30, in some settings mitochondrial Ca2+ accumulation has been determined to increase oxidation leading to mPTP opening. We could not observe this link directly due to inherent limitations in multiphoton imaging, but in our experiments, MitoQ decreased the proportion of [Ca2+]c waves resulting in mPTP opening supporting the important role of mitochondrial ROS in mPTP opening. It remains possible however that Ca2+ overload also contributes to oxidative stress in the heart.

The source of NAD(P)H autofluorescence is in the heart is primarily mitochondrial NADH.39 Mitochondrial redox state affects the degree of NADH oxidation, and can therefore be measured using the fluorescence from NAD(P)H.22,23 As expected, NAD(P)H fluorescence was elevated at the end of ischaemia due to the reduced hypoxic state. This rapidly returned to baseline levels of fluorescence after perfusion of oxygenated buffer. Reoxygenation spread out across myocardium from arterioles, which is supported by previous studies suggesting that perfusion is regulated at the level of arterioles rather than capillaries.40

Previous confocal microscope studies of isolated guinea pig hearts have observed cell-wide oscillations in ΔΨm, NADH and ROS after ischaemia and reperfusion or in response to oxidative stress, although these did not involve the mPTP or Ca2+ overload.4143 We did not observe such oscillations in our model, possibly because of slightly different perfusion conditions or buffers.

Examining the relationship between NAD(P)H and mPTP opening at the level of individual cardiomyocytes within the myocardium, we found that NAD(P)H oxidation preceded the loss of TMRM signal by 122 ± 38 s. This might be expected, as NADH has a specific inhibitory effect on the mPTP,44 although differs from studies in isolated mitochondria in which NAD(+) depletion followed Ca2+ overload induced mPTP opening.45

Previous experiments using isolated cardiomyocytes have obtained contradictory results suggesting alternatively that mitochondrial ROS during reperfusion triggers mPTP opening,46 or is a consequence of mPTP opening.27 Our results suggest that oxidation precedes mPTP opening in vivo, but do not exclude that further oxidative stress occurs following mPTP opening. Notably, ROS-induced ROS release has been proposed as the mechanism underlying the wave of mPTP opening which occurs in cardiomyocytes.21

4.3 Limitations

These results must be considered in the light of methodological limitations particularly the use of contractile inhibitor blebbistatin, and perfusion with ischaemic buffer, both necessary to reduce movement artefacts. This model of hypoxia and reoxygenation may differ subtly from the standard model of global ischaemia and reperfusion in the Langendorff mouse heart. Changes in TMRM fluorescence can occur for reasons other than mPTP opening and TMRM accumulation depends on both plasma membrane and mitochondrial membrane potential, although in our experiments the ability of CsA to decrease loss of TMRM signal suggests that it was due to mPTP opening. NAD(P)H fluorescence does not distinguish NAD from NADP pools and can be influenced by pH.

4.4 Conclusion

In the light of previous contradictory results in studies using isolated cardiomyocytes, our results define the events that lead to mPTP opening in cardiomyocytes within the intact heart (Figure 6). They support a model in which global [Ca2+]c remains relatively constant throughout reoxygenation, but increases in slow-moving, highly localized waves. Through mechanisms such as those described above, [Ca2+]c overload leads to ATP depletion and consequently inevitable contracture. In parallel, oxidative stress causes NAD(P)H to become oxidized and depleted in maintaining ROS defences. The combination of Ca2+ overload, ATP depletion, and NAD(P)H oxidation increase the likelihood of mPTP opening. It seems likely that these factors also interact with each other. For example, since many calcium channels are redox regulated, oxidative stress can further perturb calcium homeostasis, promoting catastrophic Ca2+ overload.47

In summary, in the isolated perfused mouse heart during reoxygenation, localized [Ca2+]c waves occur, and in combination with NAD(P)H oxidation and ATP depletion leads to mPTP opening. If the main initiating cause of the Ca2+ waves can be identified this may present a new target for cardioprotection upstream of the mPTP.


This work was supported by Medical Research Council grant EAA/17568 and was undertaken at UCLH/UCL who received a proportion of funding from the Department of Health's NIHR Biomedical Research Centres funding scheme.


We are extremely grateful to Prof. Michael Kotlikoff for supplying the GCaMP2 mice, and to Dr Gyorgy Szabadkai for critical reading of the manuscript.

Conflict of interest: none declared.


View Abstract