Mitochondrial Flash: Integrative Reactive Oxygen Species and pH Signals in Cell and Organelle Biology

The serendipitous discovery of mitochondrial flash

The word “serendipity” accurately describes how our groups observed flash activity with mitochondria-targeted circularly permuted yellow fluorescent protein (mt-cpYFP). Our initial intention was to investigate mitochondria Ca²⁺ handling during skeletal muscle and cardiac excitation–contraction (EC) coupling. At the time, while the role of mitochondrial Ca²⁺ uptake in promoting energy metabolism and mitochondrial Ca²⁺ overload in cell death was well known, whether mitochondria sequester Ca²⁺ during EC coupling was under debate. A study reported transient Ca²⁺ uptake events in individual mitochondria, referred to as “Ca²⁺ marks” since they reflected Ca²⁺ spark-like behavior in single mitochondria (59). This intriguing finding prompted us to use a mitochondrial-targeted genetic fluorescent Ca²⁺ indicator, named ratiometric mt-pericam (56), to probe mitochondrial Ca²⁺ handling during EC coupling. However, initial testing was disappointing, since transient millisecond duration mitochondrial events were not detected. This could be due to the relatively slow Ca²⁺ binding kinetics of mt-pericam, in comparison to small molecule Ca²⁺ indicators (such as rhod-2) used previously to measure Ca²⁺ marks. Based on this speculation, we increased the scanning time and focused on 2D time-lapse scanning. Very unexpectedly, we found stochastic and transient increases in mt-pericam fluorescence from individual mitochondria under resting conditions in intact cells (76). These events exhibited much longer time-to-peak (∼3 s) and mean duration (10–20 s) compared to that of Ca²⁺ sparks (∼30 ms duration) and Ca²⁺ marks (∼170 ms duration) and were only observed at one of the two Ca²⁺-sensitive excitation wavelengths of mt-pericam (488 nm but not 405 nm). In addition, flash events in cardiac myocytes occurred at a very low frequency, more than 3 orders of magnitude lower compared with Ca²⁺ sparks in the same cells. Further characterization indicated that acute manipulation of extracellular Ca²⁺ concentrations or application caffeine/ryanodine to alter Ca²⁺ spark frequency did not significantly impact mitochondrial flash frequency. These initial observations suggested that the events detected with mt-pericam were unlikely due to mitochondrial Ca²⁺ signals similar to “Ca²⁺ marks” and, thus, were referred to as “flashes” to differentiate them from Ca²⁺ sparks/marks. Indeed, subsequent structure–function studies of the probe revealed that identical flash events were also observed even after removal of the Ca²⁺ sensing elements (calmodulin and M13) of mt-pericam (resulting in mt-cpYFP) (76).

Mitochondrial flash renamed to superoxide flash

We puzzled over the nature and mechanism of mitochondrial flash activity for nearly 2 years following its initial observation. First, we found that flash activity was observed in all cell types screened, including rodent cardiomyocytes, skeletal myotubes, primary neurons, and human cancer cell lines (76, 82). Although the frequency of flash events differed dramatically across the different cell types, likely due to the differences in mitochondrial number, morphology, and bioenergetics, the properties of individual events, including amplitude and kinetics, were essentially invariant. These findings strongly suggest that flash activity reflects a fundamental, well-conserved mitochondrial biological process or phenomenon. Since probe structure–function studies revealed that identical events were observed using a Ca²⁺-insensitive variant of the probe (mt-cpYFP), subsequent studies focused on the role of changes in other potential mitochondrial signals (e.g., redox, superoxide, pH, membrane potential, ATP, ADP, and NADH), during flash events (76).

After confirming that flash events were not the result of laser-induced phototoxicity, we monitored changes in mitochondrial membrane potential during flash activity. Indeed, the stochastic nature of flash activity in quiescent cells is reminiscent of previously described spontaneous transient depolarization in the mitochondrial membrane potential (termed “flickers”) that exhibits similar spatial–temporal properties (14, 22, 38, 58). Consistent with this, we found that every flash is accompanied with an abrupt reduction in mitochondrial membrane potential (76). However, since not all mitochondrial membrane potential flickers are coincident with a flash event (76, 79), mitochondrial flashes likely represent only a subset of these events. Since ETC activity is stimulated by mitochondrial depolarization, studies manipulating mitochondrial respiration revealed that flash activity requires an intact mitochondrial ETC. Besides using known ETC inhibitors, one piece of key evidence comes from the study on PC12 cells after long-term treatment of ethidium bromide, which gradually depletes mitochondrial DNA. To our surprise, flash frequency decreased proportionally with a decrease in expression of mitochondrial DNA encoded proteins. In a ρ⁰ cell line, in which mitochondrial DNA is totally depleted, no flash events were observed. Together, these observations provide strong evidence that flash events are coupled to ETC activity or mitochondrial respiration (76). Since mitochondrial respiration is the major process for the generation of mitochondrial ROS, the relationship between ROS and mitochondrial flash activity was tested. Indeed, flash frequency and amplitude were found to be decreased and increased by antioxidants and oxidants, respectively. Following in vitro studies of purified cpYFP (see The Superoxide/ROS Component in Mitochondrial Flash section) showing that the probe fluorescence is reversibly increased by superoxide ions, mitochondrial flashes were named “superoxide flashes” (76).

The superoxide/ROS component in mitochondrial flash

Substantial evidence supports the conclusion that superoxide or ROS represent the major signal detected by mt-cpYFP during each flash event. First, we conducted in vitro calibrations on purified cpYFP protein using a spectrofluorometer while strictly ensuring inert environmental conditions throughout the whole process (76). When purified, the cpYFP protein is fully oxidized and exhibits a maximal fluorescence at 488 nm excitation because the purification process is conducted under ambient oxygenated conditions. Following the detailed protocol described previously (30), we successfully reduced purified cpYFP using dithiothreitol (DTT) and, after removing the reducing agent, were able to reoxidize cpYFP back to the fully oxidized condition. Surprisingly, however, fully reduced cpYFP could only be oxidized by superoxide (from xanthine + xanthine oxidase) and O₂, but not various other ROS (including hydrogen peroxide [H₂O₂], hydroxyl radical, peroxynitrite, nitric oxide) or changes in redox potential. More importantly, addition of superoxide dismutase (SOD) before or after the xanthine + xanthine oxidase reaction attenuated or blocked the increase in fluorescence (76). This inhibition by SOD provides strong evidence that superoxide underlies the increased cpYFP fluorescence. During these in vitro experiments, the solution pH was set at 8.0 to better mimic the physiological pH of the mitochondrial matrix. The superoxide sensitivity of cpYFP was further tested in live cells and animals, which showed acute increases in cpYFP fluorescence upon addition of multiple superoxide generators or oxidants, including menadione, aldrithiol, and paraquat (37, 76, 83).

The superoxide component of mitochondrial flash events in cells was supported by the inhibitory effect on flash frequency and amplitude of various general and mitochondria-targeted antioxidants (35, 37, 60, 76). Acute treatment of oxidants such as menadione and paraquat (which generate superoxide inside cells) or deletion of SOD2 (which scavenges superoxide in the mitochondrial matrix) dramatically increased flash activity (35, 37, 69, 76). The most physiologically relevant evidence for the contribution of superoxide to mitochondrial flash events comes from studies of ischemia–reperfusion, which is well known to first decrease (during ischemia) and then transiently increase (during early reperfusion) mitochondrial superoxide production. Indeed, ischemia drastically suppressed and anoxia abolished, while reperfusion caused a transient increase in flash frequency in intact cardiomyocytes (37, 76). Superoxide/ROS signal in flash events detected using mt-cpFYP was confirmed by several groups using various cell types, oxidants, and antioxidants (32, 35, 37, 60, 65, 86). Finally, other superoxide/ROS indicators, including MitoSOX, 2′,7′-dichlorodihydrofluorescein diacetate (DCF) and redox green fluorescent protein (roGFP), have been used to detect increases in superoxide during single mitochondrial flash events (Table 1) (5, 11, 60, 79, 87).

The pH component in mitochondrial flash events

During the in vitro calibration of cpYFP, we confirmed that cpYFP is pH sensitive as its fluorescence emission increases with pH as is documented for other GFP-based biosensors (76). Given the high HEPES concentration (20 mM) used in the calibration studies, pH changes were not a concern for the in vitro determination of the superoxide response of purified cpYFP. However, since cpYFP has a pK_a of ∼8.5, which approximates the pH of the mitochondrial matrix, changes in mitochondrial pH during a flash could contribute to the signal (64, 66) (see Transient pH Flash Events Detected with mt-cpYFP and mt-SypHer section for details). This possibility was initially ruled out, since a known pH-sensitive GFP derivative, enhanced yellow fluorescent protein (EYFP), when targeted to mitochondria, failed to detect transient flash-like signals (76). To further investigate whether pH contributes to mt-cpYFP-detected mitochondrial flashes, we conducted experiments to simultaneously monitor changes in fluorescence of cpYFP and a red-shifted commercial pH indicator, SNARF-1, in the same mitochondria. SNARF-1 is a small molecule with a pK_a of 7.5 determined in vitro. By using an optimized protocol, we preferentially loaded SNARF-1 into the mitochondrial matrix. Intracellular calibration of SNARF-1 using two imaging protocols revealed a pK_a of ∼8.3, which is close to that of cpYFP (∼8.5). The pH fluorescence response curves of cpYFP and SNARF-1 in cells were overlapping, suggesting the two indicators respond equally to changes in intramitochondrial pH. This approach enabled the use of SNARF-1 as an internal control to determine the relative contribution of changes in pH to mitochondrial flash events. Indeed, we found that SNARF-1 detected a small alkalization signal during each flash event, which was equal to ∼30% of the peak cpYFP signal increase during the flash (79). Based on in-cell calibrations, the pH increase during a flash was estimated to be ∼0.1 pH unit. Thus, each flash event represents a mixed signal with ∼30% due to a modest alkalization of the matrix and the rest from superoxide/ROS. In light of this fact, the events were renamed as “mitochondrial flashes” (or “mitoflashes” for short, which have been used in several reports) to acknowledge the composite nature of signal (21, 27, 31, 69, 84 –86).

Mitochondrial flash detected by other ROS/redox indicators

Shortly after the first report of mitochondrial flashes with mt-cpYFP, Pouvreau detected similar flash events in skeletal muscle cells using mt-pericam (60). The mt-pericam-detected flashes in skeletal muscles share many of the same features as the mitochondrial flashes detected by cpYFP in cardiomyocytes. In the skeletal muscle, flashes were transient with a time course of 10–20 s, were reduced by antioxidants, depended on ETC activity, and were accompanied by a drop in mitochondrial membrane potential (60). However, the skeletal muscle flash did not depend on transient opening of the mitochondrial permeability transition pore (mPTP). In addition, the author also determined the spectra of mt-pericam inside skeletal muscle cells and showed that when it binds to Ca²⁺, its fluorescence at 405 nm excitation dropped 90%, while its fluorescence at 488 nm excitation only increased by 25% (60). This is consistent with previous in vitro calibrations (56), but drastically differs from the fluorescence changes during mitochondrial flashes. During mitochondrial flashes, the increased fluorescence of mt-pericam and mt-cpYFP is solely at 488 nm excitation with no change detected at 405 nm excitation. This largely refutes the possibility of Ca²⁺-dependent signals in mt-pericam-detected mitochondrial flashes. Pouvreau was also the first to simultaneously monitor MitoSOX red with mt-pericam to independently confirm an increase in ROS production during each flash event (60). These findings critically support the superoxide contribution of mitochondrial flashes, since MitoSOX is an accepted and widely used mitochondrial superoxide indicator that is not sensitive to physiological pH changes (87).

Thanks, in part, to an ongoing debate over the relative pH and superoxide sensitivity of mt-cpYFP, we and others used other commercially available genetic and organic small molecule indicators (it should be noted that each indicator has its own limitations such as lack of specificity for single ROS and dependence on mitochondrial membrane potential) to detect superoxide/ROS/redox flash-like events in cells and live animals (Table 1). Several groups confirmed that MitoSOX red detects sudden steep and irreversible increases in superoxide production during mitochondrial flash events (5, 60, 79, 87). Since the MitoSOX signal is not reversible, a burst in superoxide results in an enhanced rate of fluorescence increase (i.e., an increase in slope) during the time-to-peak of a flash event. Thus, plotting the rate of change in MitoSOX fluorescence over time results in a trace that shares a similar time course as the changes in mt-cpYFP fluorescence during a flash (87). Similarly, the cytosolic ROS indicator, DCF, preferentially loaded into mitochondria also exhibits an increased rate of change in fluorescence during mitochondrial flash events (87). Finally, a recent study using mitochondrial targeted mt-roGFP, a redox indicator, observed “redox transients” that were coincident with changes in mitochondrial morphology (“contraction”) in neurons in vivo (11). Moreover, mt-roGFP redox transients were accompanied by an abrupt loss in membrane potential, confined within single mitochondria, and subjected to antioxidant regulation. These features of “redox transients” are remarkably similar to mitochondrial flash activity detected with mt-cpYFP and mt-pericam. However, the mt-roGFP redox events occurred at a much lower frequency and were longer in duration compared to that reported for mitochondrial flash events (Table 1). The longer time course of redox transient could be explained by the shorter superoxide lifetime compared to a general redox potential change, which requires the re-equilibration of the concentrations of electron donors and electron acceptors, and/or slower redox reaction between the indicator and underlying redox event. In summary, both genetic (mt-cpYFP, mt-pericam, and mt-roGFP) or small molecule (MitoSOX and DCF) indicators confirm that ROS production is an important component of single mitochondrial flash events in mammalian cells across multiple species and tissues (Table 1 and Fig. 1). Similarities in the basic characteristics of these flash events, such as frequency, amplitude, kinetics (Fig. 1), and sensitivity to antioxidants, indicate that the events likely reflect a common fundamental quantal event of mitochondrial function. To further elucidate their role and origin, more studies are warranted to more thoroughly characterize mitochondrial flash activity in single mitochondria using the different indicators in the same cell type and under the same conditions.

OPEN IN VIEWER

FIG. 1.

Single mitochondrial flash events detected with various indicators. Representative traces are from the indicated publications from different groups and aligned in a way to better show the coordinated change of the various signals. Note in (E), the mt-roGFP-detected redox transient is much longer than the other events and the scale bar is different from panels (A–D). roGFP, redox green fluorescent protein. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Transient pH flash events detected with mt-cpYFP and mt-SypHer

As mentioned earlier, the pH sensitivity of cpYFP was demonstrated in our first report on mitochondrial flash activity. Following introduction of mt-cpYFP into root epidermis cells of Arabidopsis, Schwarzlander et al. found single mitochondrial flashes in isolated mitochondria from these cells (64). Next, they manipulated mitochondrial respiration, ROS, and pH of the isolated mitochondria. However, instead of quantifying flash events from individual mitochondria, they used a plate reader to monitor changes in steady-state (basal) fluorescence of mt-cpYFP from a large population of isolated mitochondria. The results showed that energizing mitochondria increased basal cpYFP fluorescence, but further stimulation or inhibition of respiration did not change the mt-cpYFP fluorescence in the mitochondrial population (64). Antioxidants also failed to decrease mt-cpYFP fluorescence. Therefore, the authors concluded that mt-cpYFP is not sensitive to superoxide or ROS. In addition, the authors also monitored the pH response of mt-cpYFP and found that nigericin, a K⁺/H⁺ antiporter that dissipates the pH gradient across mitochondrial membrane, totally abolished flash events in isolated mitochondria. Taken together, the authors concluded that flash activity detected by mt-cpYFP was not related to superoxide production, but rather was due to changes in mitochondrial pH.

It should be noted, however, that monitoring basal mt-cpYFP fluorescence in a large population of mitochondria using a plate reader may not be sensitive enough to detect small transient stochastic changes in mt-cpYFP signal. This is because superoxide is short lived and mt-cpYFP fluorescence is fully reversible. Indeed, we recently monitored basal mt-cpYFP fluorescence by confocal imaging in the intact perfused heart. The results confirmed a significant change in basal mt-cpYFP fluorescence occurred during manipulation of mitochondrial substrate availability (Fig. 2A–C), the magnitude of which positively correlated with flash frequency in the same cells (Fig. 2D). In intact adult rat cardiomyocytes, stimulating mitochondrial respiration induced a mild, but significant, increase in basal mt-cpYFP fluorescence, which was sensitive to antioxidants (Fig. 2E, F). However, parallel imaging of SNARF-1 revealed no significant change in mitochondrial pH during manipulation of mitochondrial substrate availability (Fig. 2G).

OPEN IN VIEWER

FIG. 2.

Regulation of basal mt-cpYFP fluorescence by mitochondrial respiration substrates and antioxidants. (A) Experimental protocol for manipulating physiological substrates in the intact perfused mouse heart. (B) Representative images showing mt-cpYFP fluorescence in intact myocardium of mt-cpYFP transgenic mice during substrate manipulation. (C) Changes in whole cell mt-cpYFP fluorescence (ratio of 488 nm and 405 nm excitation, F488/F405) during substrate manipulation. Data are mean ± SEM. n = 22–34 images from three hearts for each group. ^† p < 0.001 versus Control (with substrate). ^# p < 0.01 versus No substrate. (D) Correlation between whole cell mt-cpYFP fluorescence (F488/F405) and mitochondrial flash frequency during substrate manipulation in the perfused heart. The data points are from Figure 2C and Figure 1D of Ref. (27). (E) Whole cell mt-cpYFP fluorescence (F488/F405) in freshly isolated adult cardiac myocytes from the mt-cpYFP mouse with no substrate (No sub), with glucose (10 mM), palmitate (0.5 mM), or with mitoTEMPO (1 μM) added after glucose (10 mM). Data are mean ± SEM. n = 43–104 cells from five mice for each group. ^# p < 0.01 versus No sub. ^† p < 0.001 versus Glucose. (F) Whole cell mt-cpYFP fluorescence in adult rat cardiac myocytes in the presence of glucose (10 mM) with or without SOD2 overexpression (+Ad-SOD2). Data are mean ± SEM. n = 27–35 cells from three rats for each group. ^† p < 0.001 versus glucose. (G) Mitochondrial pH monitored by SNARF-1 (SNARF-1 ratio) during the same treatments as in F. Data are mean ± SEM. n = 27–45 cells from three rats. Detailed methods for these experiments are described in Refs. (27, 79). mt-cpYFP, mitochondria-targeted circularly permuted yellow fluorescent protein; SOD, superoxide dismutase. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

In a follow-up study, the same group attributed mt-cpYFP flash events solely to matrix alkalization and showed that they were coupled to transient reductions in mitochondrial membrane potential (pulsing of membrane potential) in cells and isolated mitochondria from the root of Arabidopsis (Table 1) (65). Although oxidants increased the incidence of membrane potential pulsing events, the study did not assess whether oxidants also increased the frequency of mt-cpYFP flash events. A simultaneous increase in the matrix Ca²⁺ levels was detected during membrane potential pulsing, which was proposed to serve as the trigger for the membrane potential pulsing. However, blockers of the mitochondrial calcium uniporter (MCU) failed to inhibit the pulsing events. Nevertheless, the authors proposed a model in which Ca²⁺ influx into the matrix, through an unknown non-MCU channel, stimulates membrane potential pulsing to uncouple mitochondria, accelerate electron flux, and trigger transient matrix alkalization (detected by mt-cpYFP) (65). In addition, the authors proposed that opening of a nonspecific large pore (e.g., mPTP) was not responsible for the pulsing and the transient alkalization events, since the pulsing events were not blocked by cyclosporine A, a known mPTP inhibitor, and were accompanied by transient alkalization and increased matrix Ca²⁺ levels, two events that should decrease when the mPTP opens (65). Therefore, besides differences with regard to interpretations of the signal detected by mt-cpYFP (i.e., pH vs. superoxide), the trigger of membrane depolarization during flash events in plants was fundamentally different from that previously reported for mammalian cells (Ca²⁺ influx vs. opening of the mPTP). Despite these specific differences, the proposed model is in general agreement with previous hypotheses on flash activity (5, 66, 76), since it also suggests that individual mitochondria switch between quiescent and activated states, which are closely coupled to mitochondrial respiration.

Using a newly developed pH indicator, Santo-Domingo et al. reported pH flash events in Hela cells (Fig. 1C) (63). The pH indicator, SypHer, was derived from a previous H₂O₂ indicator named HyPer (8) by mutating a cysteine residue in the H₂O₂ sensing domain. It should be noted that HyPer or SypHer shares a very similar cpYFP domain as the one originally used to detect mitochondrial flashes. They differ only by a few amino acids (61, 62). mt-SypHer-detected pH flashes were dependent on mitochondrial respiration and an intact ETC, induced by oxidants, but not inhibited by mPTP blockers or antioxidants. The authors calibrated mt-SypHer in vitro and found that it senses pH, but not superoxide or other ROS. The pH flashes exhibited high amplitude, a peak increase of ∼0.4 U pH (63), which was significantly higher than ∼0.1 U determined using mt-cpYFP and SNARF-1 (79). Another feature of the pH flashes in this study was that they occurred together with a simultaneous loss of membrane potential that always mirrored the flash event time course. The authors further determined pH flash events in cells with altered mitochondrial morphology by genetic manipulations of fission and fusion proteins. Intriguingly, they found that pH flash events were dependent on optic atrophy 1 (OPA1), an inner membrane fusion protein (63). Since OPA1 regulates inner membrane cristae and mitochondrial respiration, this is consistent with the mitochondrial respiration dependence of pH flash activity. Azarias and Chatton also reported pH flash events using the same SypHer indicator. However, these studies also used MitoSOX to show that the pH flash events were accompanied by a burst in superoxide (5) and that the alkalization and ROS events coexisted in individual mitochondria (5). In the recent report that used roGFP to detect “redox transients,” the authors also monitored mitochondrial pH using mt-SypHer and found that brief matrix alkalization preceded the onset of a redox transient (Fig. 1E) (11). Interestingly, their results showed that during the redox transients, matrix pH actually decreased (became more acidic) rather than increased (11). Taken together, in these reports, pH flashes in single mitochondria share many similarities, but also exhibit important differences compared to mitochondrial flash activity detected using ROS indicators (e.g., mt-cpYFP, MitoSOX, DCF).

Mitochondrial flash events reflect an integration of ROS and pH signals

As discussed earlier, although being named differently due to the different indicators used (e.g., mt-cpYFP, mt-SypHer, mt-roGFP) and signals measured (e.g., ROS, pH, or redox), the underlying events exhibit remarkably similar features, including transient fluorescence changes, stochastic signals confined within single mitochondria, comparable quantal properties, including frequency, amplitude, and kinetics, coincidence with membrane potential depolarization, and dependence on mitochondrial respiration/metabolism (Table 1). Despite disagreement and debate over the pH and superoxide sensitivity of mt-cpYFP, the universal existence of “flash-like” events detected by mitochondrial pH and ROS indicators in intact cells and live animals represents a major breakthrough as characterization of their activity advances our understanding of real-time mitochondrial function/signals at the individual mitochondrion level (23, 28, 70, 77, 81). Moreover, the similar unitary features and regulatory mechanisms of “flash-like” events across species and cell types suggest that the events reflect different components/signals of the same underlying mitochondrial phenomenon (31, 79). Previous work has suggested that many mitochondrial processes/functions are integrated or interconnected and changes in one component directly or indirectly affect the others (2, 10, 12, 39, 55). Since flash events are coupled to mitochondrial respiration, it is very likely that the coincidence of increased ROS generation and matrix alkalization during a flash is due to transient bursts of electron flow, which are accompanied by parallel transient increases in electron slippage to oxygen and a burst of H⁺ efflux. Perhaps, it is fair to say that just like the story of “the blind men and an elephant” (where a group of blind men touch different parts of an elephant and conclude that they are in complete disagreement), the different indicators preferentially detect different components of the same phenomenon. In other words, mitochondrial flash activity reflects an integration of ROS and pH signals within the matrix of single mitochondria, which are difficult to separate. In addition, these integrated events also involve changes in other mitochondrial signals, including Na⁺, Mg²⁺, Ca²⁺, membrane potential, and more (5, 11). The relationship and dynamics between these different mitochondrial parameters (including energetics, pH, membrane potential, redox, and ROS) have been nicely integrated with computational models (43, 80).

Mitochondrial flashes are coupled to mitochondrial respiration through ETC electron flow

As discussed earlier, mitochondrial flashes detected by different groups, using different indicators and in different species/cell types, share the common feature of a strong dependence on mitochondrial respiration through ETC electron flow (5, 11, 27, 63, 65, 76, 79). This finding is in agreement with the notion that mitochondrial ETC electron flow underlies both ROS production and proton pumping. Therefore, mitochondrial flash activity can be used as a biomarker for mitochondrial respiration (Fig. 3).

OPEN IN VIEWER

FIG. 3.

Mitochondrial flash events integrate the inseparable signals of ROS, pH, and Ca²⁺ in individual mitochondria. The diagram was generated based on the models presented in Refs. (5, 27, 65, 66, 76, 82). ROS, reactive oxygen species. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Several studies have reported that mitochondrial flashes are stimulated by respiration substrates, including oxygen, glucose, fatty acids, and specific substrates for ETC complexes (23, 27, 60, 63, 69, 76, 82). We demonstrated that stimulation of respiration with physiological substrates increased the frequency of mt-cpYFP-detected mitochondrial flash events in living cells, tissues, and animals (23, 27, 76, 82). In human cell lines, both mt-cpYFP- and mt-SypHer-detected mitochondrial flashes were absent when mitochondrial DNA was depleted (in ρ⁰ human osteosarcoma cells) (63, 76). Mitochondrial flash activity was also enhanced in permeabilized cells and isolated mitochondria under conditions promoting State 3 respiration using Complex I, II, or IV substrates (27). In addition, mitochondria under State 4 respiration also exhibit increased flash frequency, consistent with increased superoxide generation when electron flow through the ETC slows down (19, 63, 79). For mt-cpYFP-detected mitochondrial flashes in plant cells/mitochondria, indirect evidence also supports the idea that respiration is coupled to mitochondrial flash activity (65). For instance, mt-cpYFP flash activity in plant cells is tightly coupled with membrane potential pulsing events, which are increased in State 3 respiration and further increased in State 4 respiration (65). In addition, the pulsing events are coupled to transient increase in matrix Ca²⁺, a known activator of mitochondrial respiration.

The causal role of mitochondrial respiration through ETC electron flow in mitochondrial flash genesis is supported by the fact that all pharmacological inhibitors of the ETC reduce or abolish mitochondrial flash activity. Specifically, mt-cpYFP- and mt-SypHer-detected mitochondrial flash activity in mammalian cells is abolished by mitochondrial inhibitors, including rotenone (Complex I), antimycin A (Complex III), NaCN or azide (Complex IV), oligomycin A (Complex V), and carbonyl cyanide m-chlorophenylhydrazone (CCCP) or carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) (uncoupler) (63, 76). Overall, these results suggest that electron flow along the entire spectrum of ETC is required for mitochondrial flash generation (2, 43, 80). However, the effect of uncouplers is an exception. Although uncouplers stimulate maximal electron flow, they also dissipate the proton gradient across the inner membrane, which is a key component of the membrane potential and proton motive force. Therefore, uncouplers decrease electron leak, prevent ROS production, and cause acidosis of the matrix. In addition, previous reports have shown that antimycin A induces reverse electron flow and increases mitochondria ROS production (2, 53). However, antimycin A significantly reduced mitochondrial flash activity detected with either mt-cpYFP or mt-SypHer. This observation could be explained by the fact that antimycin A (like FCCP) collapses the proton motive force needed for ROS production by Complex I (47, 48), thus inducing ROS production from Complex III toward the intermembrane space rather than within the mitochondrial matrix where the probe is located (63, 76). The decreased pH gradient may further decrease mt-cpYFP fluorescence. In addition, the fluorescence of mt-cpYFP is reversible and depends on the balance between superoxide production and scavenging. In contrast, other ROS indicators, such as Amplex red, DCF, and MitoSOX red, detect the cumulative signal from stable ROS such as H₂O₂ likely from both inside and outside of mitochondrial matrix. Results obtained with oligomycin A are also interesting. Oligomycin A acutely induces mitochondrial flash production followed by an inhibition after 20–30 min of treatment (27). This is likely due to the fact that oligomycin A slows down rather than totally blocks electron flow, which, similar to State 4 respiration, promotes electron leak and temporarily induces ROS production (73).

Recent reports using genetic animal models have provided explicit evidence to demonstrate the ETC dependence of mitochondrial flash activity. In these studies, knock out (KO) of ETC components or related mitochondrial proteins resulted in decreased or abolished mitochondrial flash activity in live cells and animals. For example, using a Complex I-deficient mouse model, where the Ndufs4 subunit of Complex I is knocked out in the heart, mitochondrial flash activity was decreased proportionally with decreased Complex I activity. Importantly, Complex I substrates cannot stimulate flash activity in myocytes from Ndufs4 KO mice to the same level as that observed for myocytes from wild-type mice (27). The parallel reduction in respiration and mitochondrial flash activity is consistent with decreased ROS production and a more reduced state in cardiomyocytes/mitochondria from the Ndufs4 KO hearts under resting and stress conditions (42). In Caenorhabditis elegans, a number of strains with defective respiratory chain were generated, all of which exhibited dramatically decreased mitochondrial flash activity (69). In the Hela cells, mt-SypHer-detected mitochondrial flashes are abolished by depletion of OPA1 (63). However, the effects of OPA1 ablation of flash activity were not due to altered mitochondrial morphology and fusion since flash activity was unaltered by inhibiting fusion through KO of the outer membrane fusion protein mitofusin 1. Since OPA1 also plays an important role in inner membrane cristae organization and formation of ETC supercomplexes that augment respiration efficiency (17, 74), it is possible that OPA1 ablation abolished mitochondrial flash activity through suppression of respiration.

Mitochondrial flash and transient mPTP opening

An abrupt reduction in membrane potential accompanies each mitochondrial flash event detected by the different flash indicators (mt-cpYFP, mt-SypHer, and mt-roGFP) (5, 8, 60, 63, 65, 76, 79). This is another common feature of these events and suggests a drastic change in inner membrane permeability during mitochondrial flash activity. In addition, the loss of membrane potential is temporary and recovers during or after each event, and two or more consecutive events can be detected within the same mitochondrion. These observations suggest that changes in mitochondrial inner membrane permeability are brief and reversible. Thus, transient openings of a large pore such as mPTP have been proposed to underlie this temporary loss of membrane potential.

Three lines of evidence suggest that mt-cpYFP-detected mitochondrial flashes in mammalian cells are accompanied by transient mPTP openings. First, the mPTP pore is a nonspecific large pore that can conduct small molecules up to 1.5 kDa. Mitochondria loaded with small fluorescence molecules, such as rhod-2 and mag-rhod-2 (molecular weight less than 1 kDa), were monitored during mt-cpYFP flash activity in the same mitochondria in skeletal muscle and adult cardiomyocytes. The results demonstrate a clear temporal coupling between flash events and the loss of fluorescence from these small molecules (76, 78, 79). These findings are consistent with the opening of a large pore during flash activity that permits release of these small molecules from the matrix. Second, mitochondrial flash activity is altered by pharmacological and genetic manipulations of mPTP opening. While the precise molecular identity of mPTP is not fully resolved, cyclophilin D (CypD) regulates the Ca²⁺ sensitivity of mPTP and is also a regulator of the ATP synthase (Complex V), in which the c-ring of the Fo or the dimers of the ATP synthase are proposed to form the mPTP pore (1, 6, 9, 24). The most widely used pharmacological inhibitor of mPTP, cyclosporine A, which targets CypD, blocks cpYFP-detected mitochondrial flash activity in cardiomyocytes and neurons. Conversely, flash activity is increased by atractyloside, an mPTP opener that binds to adenine nucleotide translocator and locks it into opening conformation (33 –35, 50, 76, 79). Knocking down or knocking out CypD, which suppresses but does not abolish mPTP openings, decreases flash frequency, while overexpressing CypD increases flash frequency (68, 76, 84). Similarly, mt-SypHer-detected mitochondrial flash activity is also induced by atractyloside, and the authors suggested that the pH flash events are triggered by the opening of a large pore during fusion (63). Finally, mPTP opening is regulated by a number of signals, such as ROS and Ca²⁺. The roles of oxidants and antioxidants on mitochondrial flash regulation are consistent with mPTP as a trigger, and mitochondrial Ca²⁺ uptake indeed enhances mitochondrial flash activity (26, 32, 41, 86). In plant cells, atractyloside and H₂O₂ induce mitochondrial flashes. However, based on the interpretation that mt-cpYFP flash activity reflects transient matrix alkalization, the authors excluded the possibility that opening of a large pore triggers the flash events and accompanying loss of membrane potential (pulsing events) (65). Mitochondrial redox events detected with mt-roGFP coincident with mitochondrial “contraction” are also accompanied by a loss of membrane potential and are regulated by oxidants/antioxidants. However, the role of mPTP activity in the contraction events was excluded due to the lack of an effect by cyclosporine A. Since mitochondrial flash frequency in skeletal muscle is not altered by either cyclosporine A or CypD KO, but is coincident with the opening of a large pore, it is likely that CypD-independent mPTP openings (possibly via ROS activation) are coupled to flash activity in skeletal muscle (60, 68, 79, 82). Taken together, the data suggest that transient openings of the mPTP likely accompany mitochondrial flash activity (Fig. 3).

While transient mPTP opening accompanies each mitochondrial flash, whether they are causally linked is a topic of current investigation. In cardiomyocytes and neurons, transient openings of mPTP play a causal role in triggering mitochondrial flashes. This conclusion is derived from the inhibitory effect of cyclosporine A and CypD KO or knockdown (27, 33 –35, 68, 76). In skeletal muscles, even though a large pore opening also accompanies each flash event, mPTP inhibition does not alter flash frequency (60, 68, 82). Because CypD controls the Ca²⁺ sensitivity of mPTP, but not other triggers of the mPTP, including ROS (7), flash activity in skeletal muscle may be triggered by CypD-independent mPTP openings (81). This possibility is supported by the very low abundance of CypD in skeletal muscle compared to the heart (68). Furthermore, a “pedestal precursor” signal was found before some flashes in both cardiomyocytes and skeletal muscle cells, which could reflect small increases in ROS that trigger mPTP opening (49, 60). A detailed discussion of the relationship between mPTP openings and flash activity is provided in our previous review article (81). Finally, whether genetic manipulation of mPTP constituents modulates mt-cpYFP-detected flashes in plant cells or mt-roGFP-detected redox flashes in neurons awaits future studies.

Physiological functions of mitochondrial flash activity

Mitochondrial flash activity is an evolutionarily conserved process that crosses a broad range of species and tissues. Thus, mitochondrial flashes likely reflect a fundamental physiological mitochondrial phenomenon that has important implications in mitochondrial biology and cell biology. However, debate over the elusive nature of the different overlapping signals that occur during these events has hampered broader investigations into the biological importance of the phenomenon. Moreover, the relative low frequency and discrete/stochastic occurrence of flash activity have raised concerns as to whether these events exert a significant functional impact on whole cell energy metabolism, redox regulation, and/or ion homeostasis. Nevertheless, several studies have clearly shown that mitochondrial flash activity contributes to both local physiological signaling and pathological conditions in the heart, skeletal muscle, neurons, and stem cells (Fig. 4).

OPEN IN VIEWER

FIG. 4.

Mitochondrial flash activity as a biomarker for health (physiology) and disease (pathology). Various physiological and pathological triggers modulate mitochondrial flash frequency. Pathological stresses usually increase flash frequency, thus contributing to stress-induced mitochondrial and cell dysfunction, ultimately leading to the progression of disease. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

We have studied the role of mitochondrial flash activity in cardiac EC coupling, which provides the driving force for the continuous pumping function of the heart. The results revealed that a positive feedback loop exists in the cardiomyocytes, whereby mitochondria respond to an increased cytosolic energy demand by mitochondrial Ca²⁺ uptake boosting respiration. Furthermore, the increased mitochondrial respiration is coupled to increased generation of mitochondrial flashes, which provide local ROS signaling to further promote Ca²⁺ release from the sarcoplasmic reticulum. This feedback loop provides an important mechanism for the heart to continually match energy production with increased energy demand on a beat-to-beat basis (26).

The role of mitochondrial flash activity in skeletal muscle development was investigated in mt-cpYFP-expressing transgenic zebrafish. The results demonstrate striking multiphasic changes in mitochondrial flash frequency during muscle development, which are accompanied by mitochondrial morphological remodeling and functional maturation (85). However, whether alterations in mitochondrial flash activity play a causal role in mitochondrial maturation and muscle development is an unanswered question ripe for future study. The roles of mitochondrial flash activity in somatic cell reprogramming and neural progenitor differentiation have also been studied. Mitochondrial flash activity is increased during the early phase of somatic cell reprogramming, which induces DNA demethylation in the nucleus and transcriptional upregulation of Nanoga, a key gene in self-renewal and maintaining stemness of cells (84). In neural progenitor cells, mitochondrial flash activity negatively regulates cell proliferation and promotes cell differentiation (34, 35). These data indicate that mitochondrial flash activity regulates context-dependent stem cell differentiation and somatic cell dedifferentiation. With the demonstration that mitochondrial flash activity is coupled to both ETC-dependent respiration and mPTP opening, future investigations will undoubtedly uncover additional important cell type-specific physiological functions linked to mitochondrial flash activity.

Mitochondrial flash activity and disease

Given the importance of mitochondrial ROS production and mPTP in human disease, several studies have explored the role of mitochondrial flash activity under pathological conditions. Selenite-induced Hela cell death is mediated partially by increased mitochondrial ROS production through increased mitochondrial flash activity, and the antiapoptotic protein, Bcl-2, ameliorates this effect by reducing mitochondrial flash frequency (51). Proinflammatory cytokines stimulate both mitochondrial flash activity and mPTP openings in articular chondrocytes, which may contribute to cartilage inflammatory diseases (15). During heart failure, mitochondrial flash activity is decreased, which likely reflects compromised mitochondrial metabolism (26). During reperfusion following cardiac ischemia, mitochondrial flash activity mirrors the burst in mitochondrial ROS production and mPTP openings (54, 76). In the skeletal muscle, uncontrolled mitochondrial flash activity contributes to enhanced muscle oxidative stress in a murine model of malignant hyperthermia (82), a muscular dystrophy model in zebrafish (85), and in the ob/ob mouse model of obesity and insulin resistance (21).

Several groups have reported a role for altered mitochondrial flash activity in neuronal disease. Increased mitochondrial flash activity mediates the suppressive effect of amyloid beta 1–42 on neural progenitor cell proliferation (33), mitochondrial Ca²⁺-induced mitochondrial DNA damage in fibroblasts from Huntington disease patients (75), and glutamate toxicity in motor neurons (50). roGFP-detected redox flashes and accompanying mitochondrial “contraction” events are induced by stresses that promote neuron damage (11). Recent studies conducted on C. elegans revealed an important role of mitochondrial flash activity in aging and wound healing. In one study, mitochondrial flash frequency peaked 3 days after maturation, which was negatively correlated to the life span of the worm. In general, the higher the level of day 3 flash activity, the shorter the worm's life span, such that mitochondrial flash activity could be used to predict how long the organism will live (69). In a second study, physical damage of the worm skin (by laser or needle stab) induced a wave of mitochondrial ROS production featured by a dramatic increase in mitochondrial flash activity around the injured site. This ROS production was induced by a wave of intracellular Ca²⁺ right after the injury and was responsible for actin-based wound closure (83). In most of these studies, increased mitochondrial flash activity serves as a universal stress response signal downstream of changes in mitochondrial and/or cytosolic Ca²⁺ and upstream of either detrimental outcomes, including DNA damage, or compensatory outcomes, such as wound healing (Fig. 4).

Compatible with the significantly increased flash frequency under pathological conditions outlined earlier is the “ROS-induced ROS release” model. This model was developed based on the findings that laser-induced local ROS production triggers either mPTP or inner membrane anion channels (IMAC) to release ROS (2, 89). The released extramitochondrial ROS subsequently triggers neighboring mitochondria to undergo a similar process that leads to whole cell bursting of ROS oscillations. These whole cell mitochondrial ROS oscillations are coincident with loss of mitochondrial membrane potentials, which could lead to cellular dysfunctions such as cardiac arrhythmias (2 –4, 18, 71). Recently, computational models have integrated the ROS signals with other mitochondrial functions during ROS oscillations (43 –46, 80). Since synchronized flash activity is observed in a group of interconnected mitochondria in skeletal muscles (23, 25), individual mitochondrial flashes and whole cell ROS oscillations could be mechanistically linked and reflect the dynamic ROS regulation (and integrated mitochondrial functions) under physiological or pathological conditions, respectively. However, important characteristics of flash events that differentiate them from global ROS-induced ROS release events are that mitochondrial flashes occur spontaneously and reversibly in quiescent cells under physiological conditions, represent a stochastic rather than oscillatory process, and are confined to single mitochondria or a local interconnected mitochondrial network.

The autonomy and excitability of individual mitochondria

The discovery and characterization of mitochondrial flash activity provide new insights into mitochondrial biology. First, flash frequency provides an index of mitochondrial respiration at the individual mitochondrion level. Mitochondrial flash activity also represents a fundamental mechanism linking ATP utilization (energy demand) with ETC electron flow (energy production). Thus, mitochondrial flash activity may reflect a “switch” within each mitochondrion by which an increase in energy demand is sensed and translated into a transient burst or acceleration in aerobic respiration. In this regard, individual (or interconnected) mitochondria are regulated and work autonomously. Such regulatory signals could involve mitochondrial ROS and/or Ca²⁺, which both trigger transient mPTP opening and subsequent mitochondrial flash activity (Fig. 3). At the whole cell level, controlling the autonomy of individual mitochondria could enable precise and prompt manipulation of energy production to match fluctuations in energy demand. Under certain conditions, such as during whole cell ROS oscillations, individual mitochondrion could function in synchrony or form functional networks (2, 4, 25, 43, 45, 46, 80).

Moreover, mitochondria have been suggested to be excitable organelles in terms of Ca²⁺ signaling (39), and this concept now appears to extend to ROS and pH, which are concomitantly changed during mitochondrial respiration and thus are inseparable. We and others have proposed that mitochondrial excitability is reflected in integrated functions of mitochondria switching between quiescent and excited states (2, 5, 43, 49, 76, 80). In this regard, mitochondrial excitability could shift current paradigms that view mitochondria as passive energy producers to a more active participant and regulator of whole cell energy metabolism and signaling, which could broaden and deepen our understanding of the complex role of mitochondria in health and disease (Figs. 3 and 4).

Amplitude versus frequency mode of regulation

The concept of “switch” and “excitability” of mitochondria in bioenergetics discussed earlier brings forth a fundamental mechanism in encoding biological signals physiologically and effectively through stochastic rather than tonic signals. One example is how Ca²⁺ signals can be encoded by altering the frequency of Ca²⁺ oscillations. In an analogous manner, frequency-encoded ROS (and pH) signal via changes in mitochondrial flash activity represents a new mechanism for mitochondrial ROS signaling. As discussed earlier, mitochondrial flash activity is controlled by multiple physiological and pathological stresses, suggesting that the spatiotemporal regulation of mitochondrial flash frequency could represent a powerful mechanism of mitochondrial and cellular ROS regulation. We, therefore, refer to this “digitized” ROS production as a “frequency-modulatory (FM) mode” of ROS regulation. In comparison, sustained or steady-state mitochondrial ROS production (55) would represent an “amplitude-modulatory (AM) mode” of ROS regulation (Fig. 5). Similar distinct regulatory mechanisms have also been shown for controlling mitochondrial Ca²⁺, in which oscillatory and sustained Ca²⁺ changes play different roles in energy metabolism (29). Accumulating evidence suggests that the FM mode of ROS regulation is uniquely suitable to regulate local signals such as EC coupling and gene expression under physiological conditions (27, 84). Under extreme conditions or strong stresses, such as ischemia–reperfusion or cell damage, the FM mode activity becomes uncontrolled and significantly augments global or whole cell ROS signaling (2, 76, 84, 89). In contrast, the degree of AM mode activity may regulate the sensitivity of FM mode activity through triggering transient mPTP openings (Fig. 5) (32). Future studies are needed to provide more evidence for the existence, interaction, and significance of the two proposed modes of mitochondrial ROS regulation in cells. In addition, the composite nature of mitochondrial flashes indicates that other signals produced during these events (e.g., pH, Ca²⁺, membrane potential, and redox) could display similar dual-mode forms of regulation, one for local and compartmentalized regulation and another for global and steady-state regulation. In the face of dynamic nature of mitochondria that tend to preferentially localize in the hot spots where the energy demands are high as well as their critical role in the interorganelle (e.g., mitochondria–nucleus, mitochondria–endo/sarcoplasmic reticulum) communications, these oscillatory and localized mitochondrial signals will play a key role in cell regulation.

OPEN IN VIEWER

FIG. 5.

Proposed two modes of mitochondrial ETC-dependent superoxide production that coexist in vivo or under physiologically relevant conditions. The first mode, termed AM mode, reflects basal or constitutive ROS production. The second mode, termed FM mode, involves quantal events of mitochondrial flash activity that provide pulsatile transients of ROS production (plus other signals, including pH, Ca²⁺, redox, membrane potential). The two modes of ROS production may be linked through the effects of constitutive ROS production on transient mPTP activation. AM, amplitude-modulatory; ETC, electron transport chain; FM, frequency-modulatory; mPTP, mitochondrial permeability transition pore. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars