Mitochondrial flashes: new insights into mitochondrial ROS signalling and beyond

Abstract

Respiratory mitochondria undergo stochastic, intermittent bursts of superoxide production accompanied by transient depolarization of the mitochondrial membrane potential and reversible opening of the membrane permeability transition pore. These discrete events were named ‘superoxide flashes’ for the reactive oxygen species (ROS) signal involved, and ‘mitochondrial flashes’ (mitoflashes) for the entirety of the multifaceted and intertwined mitochondrial processes. In contrast to the flashless basal ROS production of ‘homeostatic ROS’ for redox regulation, bursting ROS production during mitoflashes may provide ‘signalling ROS’ at the organelle level, fulfilling distinctly different cell functions. Mounting evidence indicates that mitoflash frequency is richly regulated over a broad range, and represents a novel, universal, and ‘digital’ readout of mitochondrial functional status and of the mitochondrial stress response. An emerging view is that mitoflashes participate in vital processes including metabolism, cell differentiation, the stress response and ageing. These recent advances shed new light on the role of mitochondrial functional dynamics in health and disease.

Tingting Hou is a PhD candidate (expected to graduate in 2015) in biophysics at the Institute of Molecular Medicine, Peking University. Instructed by professor Heping (Peace) Cheng and Professor Xianhua Wang, she focuses on the regulatory mechanisms and molecular basis of mitochondrial flashes as well as on novel functions of mitochondrial proteins. Heping (Peace) Cheng has a PhD in physiology and biophysics and is a founder of the Institute of Molecular Medicine at Peking University. His early achievements include the co-discovery of ‘calcium sparks’. His current research focus is on the mechanism, regulation and biology of mitochondrial flashes as well as on methodology for flash study.

Introduction

The mitochondrion is arguably the most complex and intriguing cellular organelle in eukaryotic cells – not only for its unique symbiotic origin, retention of its own genome, maternal inheritance and double-layered membranous structure, but also for its unparalleled functional diversity. More than a powerhouse central to cellular bioenergetics, it plays pivotal roles in calcium signalling, redox homeostasis and cell fate regulation (Green, 1998; Duchen, 2000; Droge, 2002; Nunnari & Suomalainen, 2012). Since Jensen reported that the respiratory chain produced reactive oxygen species (ROS) in 1966 (Jensen, 1966) followed by the pioneering work of Chance and colleagues on the mitochondrial production of hydrogen peroxide (H₂O₂) (Loschen et al. 1971; Boveris & Chance, 1973; Chance et al. 1979), a huge literature has been developed on the sources and consequences of mitochondrial ROS production (see Sies, 2014 for a review). Normally, the flux of electrons from substrates flows through various redox centres in the electron transport chain (ETC), and is ultimately terminated to water in a four-electron reduction of molecular oxygen, catalysed by cytochrome c oxidase. However, a small fraction of ‘leak’ electrons participate in a single-electron, incomplete reduction of O₂ to produce the free radical superoxide anion (O₂^•−). Superoxide is subsequently converted to H₂O₂ through spontaneous or superoxide dismutase (SOD)-catalysed dismutation and then to oxygen and water by antioxidant enzymes such as catalase, thioredoxin and glutathione peroxidases (Droge, 2002).

The production of ROS by mitochondria can lead to oxidative damage that underlies many pathologies including malignant diseases, diabetes mellitus, atherosclerosis, ischaemia–reperfusion injury, chronic inflammatory processes and neurodegenerative diseases (Halliwell, 2001; Droge, 2002; Newsholme et al. 2007). A paradigm-shifting concept in recent years, however, is that mitochondrial ROS also contribute to retrograde redox signalling from the organelle to the cytosol and nucleus (Droge, 2002; Balaban et al. 2005; Wallace, 2012). In so doing, mitochondrial ROS play signalling roles in a variety of pathways in differentiation and organogenesis (Owusu-Ansah & Banerjee, 2009), cell fate regulation (Maryanovich & Gross, 2013) and the stress response (Adler et al. 1999). In this brief review, we focus on the surprising recent discovery of a novel mode of ROS generation or ‘superoxide flash’ (Wang et al. 2008), also known as ‘mitochondrial flash’ (mitoflash) (Shen et al. 2014), in respiratory mitochondria. In particular, we present evidence that mitoflashes are universal and multifaceted, and highlight features that distinguish them from basal mitochondrial ROS production and regulation. We also synthesize the data on possible signalling roles of mitoflashes in the context of metabolism, cell differentiation, stress response, disease and ageing.

Universality of mitoflashes

By serendipity, we found that the fluorescent moiety of the Ca²⁺ indicator pericam (Nagai et al. 2001), a circularly permuted yellow fluorescent protein (cpYFP), is a novel biosensor of O₂^•−, the primal ROS generated by the mitochondrial ETC (Wang et al. 2008). By targeting cpYFP to the mitochondrial matrix using the signal peptide of cytochrome c oxidase subunit IV (COX IV), we discovered superoxide flashes – sudden, quantal, 10 s bursts of superoxide production – in single mitochondria in intact cells (Wang et al. 2008). More than a free radical-producing event, a superoxide flash is always accompanied by transient depolarization of the mitochondrial membrane potential (ΔΨm), but not vice versa (Wang et al. 2008; Li et al. 2012). There is also a transient, reversible increase of mitochondrial permeability at the onset of a superoxide flash, evidenced by a concomitant, irreversible loss of matrix-loaded indicators such as Rhod-2 (mol. mass 752 Da) (Wang et al. 2008) or its Ca²⁺-insensitive analogue (mol. mass 980 Da) (Wang et al. 2012) (Fig. 1). In the following discussion, we use ‘mitoflash’ when referring to this complex mitochondrial phenomenon in toto, and ‘superoxide flash’ when specifically referring to its free radical-producing component.

Figure 1. Mitoflash is a complex phenomenon comprising multifaceted and intertwined processes.

A cpYFP flash is associated with a transient loss of mitochondrial membrane potential (indexed by tetramethyl rhodamine methyl ester (TMRM)) (A), mitoSOX signal (B), DCF signal (modified from Zhang et al. 2013) (C), transient depletion of NADH (D) and FADH₂ (reflected by FAD⁺ autofluorescence) (E), and an MPT (evidenced by irreversible loss of Rhod-2 analogue) (modified from Wang et al. 2012) (F). NADH autofluorescence was measured by 720 nm two-photon excitation. FAD⁺ autofluorescence was measured between 500 and 650 nm at 488 nm excitation. Scale bar: 10 s for x-axis and 0.2ΔF/F₀ for y-axis in A, D, E and F; 20 s for x-axis and 1 arbitrary unit for y-axis in B and C.

Since their discovery, mitoflashes have been detected in many types of cells and tissues, including cardiomyocytes, skeletal muscle myotubes and fibres, neurons, glial cells, fibroblasts, chondrocytes, and many types of cancer cells (Wang et al. 2008; Pouvreau, 2010; Fang et al. 2011; Ma et al. 2011; Wei et al. 2011; Hou et al. 2012; Cao et al. 2013; Hou et al. 2013; Wei-LaPierre et al. 2013) (Fig. 2). The experimental systems used range from isolated single mitochondria (Wei-LaPierre et al. 2013; Zhang et al. 2013) and intact cells to ex vivo beating hearts (Wang et al. 2008) and even live animals (Fang et al. 2011; Shen et al. 2014) (Fig. 2). That mitoflashes occur in isolated single mitochondria attests that the mechanism of their formation is intrinsic to this organelle; on the other hand, the manifestation of mitoflashes in vivo underscores the physiological relevance of this dynamic mitochondrial activity. Remarkably, mitoflashes are evolutionarily conserved – events of nearly identical properties have been found in different cell types and experimental systems across species from Caenorhabditis elegans (Shen et al. 2014), to zebrafish (M Zhang & J Xiong, unpublished observations), to humans (Wang et al. 2008; Ma et al. 2011; Hou et al. 2013). In worms and mammals alike, the generation of mitoflashes requires the integrity of the ETC and mitochondrial respiratory function (Wang et al. 2008; Shen et al. 2014). Moreover, the intersection of mitoflashes with the key mitochondrial function – respiration – strongly suggests that they represent a conserved and fundamental activity of this ancient organelle, and are quintessential to mitochondrial ROS signalling and other functions (see below).

Figure 2. Mitoflashes represent a universal and conserved mitochondrial activity.

Mitoflashes are found across species in different tissues, cells and experimental systems. Their unitary properties are highly comparable. MTS: mitochondrial targeting signal peptide of cytochrome c oxidase subunit IV (COX IV) (for transgenic mouse and zebrafish) or succinate dehydrogenase (ubiquinone) iron-sulfur subunit (SDHB-1) (for transgenic C. elegans).

Nature of mitoflashes

It has been increasingly appreciated that the mitoflash is a complex phenomenon comprising multifaceted and intertwined processes: superoxide flashes are not only coupled with transient depolarization of ΔΨm, but are also concurrent with a rapid depletion of the electron donors NADH and FADH₂ (Fig. 1). In some cases, mitoflashes are associated with conspicuous, reversible mitochondrial swelling that may masquerade as mitochondrial ‘contraction’ (Ma et al. 2011; Breckwoldt et al. 2014) and, in filamentous mitochondria, give rise to the transient beads-on-thread shape during a mitoflash (Ma et al. 2011). Among all these dynamic activities, bursting ROS signals constitute one but not the sole message that mitoflashes could convey.

Given that cpYFP is also a pH sensor with a pK_a of 8.5 (Wang et al. 2008; Schwarzländer et al. 2011), Schwarzländer et al. proposed that a similar mitochondrial phenomenon found in Arabidopsis reflects a transient matrix alkalization rather than a superoxide burst (Schwarzländer et al. 2011). Specifically, they suggested that spontaneous membrane depolarization accelerates proton pumping by the ETC and results in an increase in matrix pH, or ‘pH flash’ as termed by Schwarzländer et al. This controversy on the origin of mitoflashes has been examined in recent reviews by ourselves and others (Quatresous et al. 2012; Schwarzländer et al. 2012b; Wang et al. 2014); however, it is not yet fully resolved. Because mitochondrial superoxide metabolism and pH regulation are inseparably interlinked, this fact renders most evidence ambivalent and inconclusive for either hypothesis. For instance, both hypotheses would be compatible with the requirement of an intact ETC (Wang et al. 2008; Schwarzländer et al. 2011; Shen et al. 2014), dependence on oxygen (Wang et al. 2008; Huang et al. 2011), activation by ROS (Huang et al. 2011; Hou et al. 2013; Shen et al. 2014), inhibition by antioxidants (Wang et al. 2008; Schwarzländer et al. 2011), and the coupling with transient depolarization of ΔΨm (Wang et al. 2008; Schwarzländer et al. 2012a).

The evidence in favour of the pH hypothesis includes the finding that SypHer, a mutant of Hyper, with its two cysteine residues critical to H₂O₂ sensing changed to serine, which is insensitive to H₂O₂ but sensitive to pH with a pK_a of 8.7 (Poburko et al. 2011), detects similar flashes in astrocyte (Azarias & Chatton, 2011) and HeLa mitochondria (Santo-Domingo et al. 2013). However, caution has been voiced for this interpretation because the possibility that SypHer might also sense superoxide has not yet been excluded (Quatresous et al. 2012). While the pH hypothesis predicts a mirror relationship between ΔΨm and the cpYFP signal, which is supported by data from Arabidopsis, experimental data from mammalian cells have shown that this is not always the case: oscillatory ΔΨm depolarizations are often uncoupled from or, when coupled, outlast the cpYFP flashes (Wang et al. 2008; Li et al. 2012; Wei-LaPierre et al. 2013). Furthermore, nigericin, a H⁺/K⁺ antiporter, used in the micromolar range to clamp matrix pH without altering ΔΨm, inhibits mitoflashes, and this observation was interpreted as strong evidence to bolster the pH hypothesis of mitoflash origin (Schwarzländer et al. 2012a). However, a counter-argument is that nigericin at high doses also perturbs mitochondrial proton gradients that might be essential to superoxide production, and nigericin at nanomolar concentrations markedly stimulates, rather than inhibits, mitoflashes (Wei-LaPierre et al. 2013).

The observation that mitoflash events comprise a burst of superoxide production (superoxide flash) has been independently corroborated by several groups using multiple approaches. The synthetic ROS indicators mitoSOX (Pouvreau, 2010; Wei-LaPierre et al. 2013; Zhang et al. 2013) and 2,7-dichlorodihydrofluorescein diacetate (DCF) (Zhang et al. 2013), both pH insensitive in the milieu of intact mammalian cells at extracellular pH ranging from 6.0 to 9.5, faithfully report mitoflashes in both single-cell and single-mitochondrion systems. When used individually to avoid possible fluorescence resonance energy transfer (FRET) effects and spectral cross-contamination, these pH-insensitive ROS sensors confirmed flash events of nearly identical frequencies (Zhang et al. 2013). Azarias and Chatton demonstrated concurrent mitoSOX-reported bursts of superoxide during the SypHer-reported pH flashes (Azarias & Chatton, 2011). More recently, Breckwoldt et al. reported reversible redox changes lasting ∼200 s arising from brief oxidative bursts and coincident with spontaneous ‘contractions’ of axonal mitochondria in transgenic mice with neuron-specific expression of the redox sensor Grx1-roGFP (Breckwoldt et al. 2014). However, mitoSOX and DCF are not universally accepted as reliable ROS indicators and, so far, the structural information and biochemistry of how cpYFP senses superoxide also remain a mystery.

A third possibility remains open: the mitoflash could be sphinx-like; the superoxide flash and the pH flash are just two facets of this complex phenomenon (Wang et al. 2014). In this regard, quantitative appraisal of the respective contributions of pH (measured with SNARF-1) and superoxide have revealed superoxide as the predominant signal with a minor pH (0.08 units) contribution for mitoflashes in the heart and skeletal muscles (Wei-LaPierre et al. 2013). It should be informative to implement similar quantitative analyses in Arabidopsis, because it is conceivable that the relative contributions of superoxide and pH to mitoflash events vary in a species-, cell type-, and context-dependent manner.

Flash and flashless ROS production: similarities and differences

As the primary source of intracellular ROS, mitochondria contain a sophisticated, multilayered system for ROS metabolism and signalling. On the one hand, mitochondria contain at least 10 known sites that are capable of generating ROS, with 9 for superoxide anions (Andreyev et al. 2005; Starkov, 2008). The basal ROS production is believed to be a continuous and flashless process. On the other hand, mitochondria harbour numerous antioxidants and ROS-defence enzymes including MnSOD, catalase, glutathione, glutathione-S-transferase and phospholipid hydroperoxide glutathione peroxidase (for the removal of lipid peroxides) (Andreyev et al. 2005). The concerted actions of these molecular players help to set the steady-state ROS level and maintain redox homeostasis. In contrast, superoxide flashes are bursting superoxide-producing events that occur only intermittently and are spatially well confined. As a hallmark of ‘digital’ signals, the properties of individual superoxide flashes appear to be stereotypical. Depending on the tissue and cell type, metabolic rate, developmental stage, age and degree of oxidative stress, the frequency of superoxide flashes can vary by orders of magnitude, but there are only mild to moderate changes in the amplitude and duration of individual events. Mitoflash regulation and signalling occur predominantly in a frequency-modulated manner (Wang et al. 2014).

Apart from the distinctive digital feature, superoxide flashes differ from basal ROS in the high levels of ROS attained, as evidenced by the steep rise of cpYFP, mitoSOX or DCF (Zhang et al. 2013), and Grx1-roGFP signals (Breckwoldt et al. 2014). Thus, whereas sustained elevation of global ROS is undoubtedly detrimental, brief ROS pulses may activate high-threshold ROS pathways locally, fulfilling signalling roles while limiting cellular and mitochondrial damage. Indeed, while the rate of superoxide flashes waxes and wanes, the global ROS level and redox balance are exquisitely safeguarded by sophisticated multi-layered systems. Recently, we have proposed that basal ROS production may fulfil housekeeping roles, such as maintaining redox balance (i.e. homeostatic ROS); against this homeostatic background, superoxide flashes as dynamic events are well poised to serve as physiological signalling units (i.e. signalling ROS) (Wang et al. 2012). In analogy to ‘calcium sparks’ for calcium signalling (Cheng & Lederer, 2008), mitoflashes with their spatiotemporally controlled high ROS signals may also build up hierarchical multi-scaled intracellular ROS dynamics, conferring efficiency, specificity and diversity on ROS signalling (see below).

It is also instructive to compare and contrast the similarities and differences in the regulation of basal ROS and superoxide flashes. Both depend on ΔΨm (Starkov, 2008; Wang et al. 2008), and are sensitive to matrix pH (Murphy, 2009; Wei-LaPierre et al. 2013) and oxygen tension (Huang et al. 2011), as well as the concentration of substrate supplied (Starkov, 2008). In addition, no flashes have been detected in ρ⁰ 143B cells, human osteosarcoma cells devoid of all mitochondrial (mt) DNA-encoded ETC components (Wang et al. 2008). In ρ⁻ PC12 cells, rat PC12 phechromocytoma cells treated with ethidium bromide to partially deplete mitochondrial DNA, both the superoxide flash incidence and mtDNA decrease proportionally (Wang et al. 2008). More recently, it has been shown that C. elegans mutants with defective respiratory chain complexes, including the complex I mutants gas-1, nuo-6, the complex II mutant mev-1, the complex III mutants cyc-1, isp-1, and the complex V mutant atp-3, all exhibit extremely low mitoflash activity (Shen et al. 2014). Interestingly, the ETC inhibitors differentially regulate mitoflash and basal ROS production. For basal ROS, they can be either inhibitory or stimulatory depending on the sites of the ETC complex affected, the direction of electron flow, and the ‘side-ness’ of ROS emission (Starkov, 2008). In contrast, all the ETC inhibitors tested, including antimycin A, rotenone, myxothiazole and oligomycin, suppress or even abolish mitoflash production (Wang et al. 2008). As to the pH sensitivity, alkalization of the mitochondrial matrix accelerates basal ROS production and acidification suppresses it (Murphy, 2009), but the mitoflash incidence in cardiac cells shows little change over the physiological pH range and is even depressed at extreme pH values (X. Wang & H. Cheng, unpublished observations). Future investigation is warranted to establish the superoxide-generating sites and the mechanism of their synchronous activation as well as their prompt termination after a 10 s burst.

Mitochondrial permeability transition during mitoflashes

An important feature of mitoflashes is that their ignition is tightly coupled with another dynamic mitochondrial activity, the mitochondrial permeability transition (MPT). The MPT is defined as a sudden increase of inner mitochondrial membrane permeability to ions and small solutes (Haworth & Hunter, 1979; Hunter & Haworth, 1979a,b1979b), and is thought to be mediated by the opening of a high-conductance channel, the MPT pore (mPTP) which is also known as the mitochondrial megachannel (Kinnally et al. 1989; Petronilli et al. 1989; Szabo & Zoratti, 1991; Szabo et al. 1992). Extensive functional studies have revealed two gating modes of the mPTP: irreversible, full-conductance opening for a permanent MPT (pMPT) leading to cell death, and reversible opening with smaller and variable conductance for a transient MPT (tMPT) (Haworth & Hunter, 1979; Ichas et al. 1997). The involvement of the MPT in a mitoflash is strongly supported by the precipitous dissipation of ΔΨm (Wang et al. 2008; Pouvreau, 2010; Fang et al. 2011; Ma et al. 2011; Wei et al. 2011), irreversible loss of small solutes (molecular mass up to 980 Da) from the matrix (Wang et al. 2008; Wang et al. 2012), and transient mitochondrial swelling that sometimes masquerades as mitochondrial ‘contraction’ during a mitoflash, presumably arising from water rushing into the hypertonic matrix (Wang et al. 2008; Ma et al. 2011). Because of the ability to generate repeated mitoflashes in the same mitochondrion (Wang et al. 2008; Hou et al. 2012), it is clear that the mitoflash is coupled with reversible MPT. In the following discussion, we refer to the putative reversible MPT underlying the mitoflash as an ‘fMPT’. (i.e. we use mitoflashes as the optical readout of fMPTs operating in situ). Functionally, an fMPT can either be the mitoflash-igniter, by inducing ΔΨm loss and swelling that somehow trigger bursting superoxide production (Wang et al. 2008; Wei & Dirksen, 2012), or the mitoflash-terminator, by dissipating the electrical and proton gradients required for supporting an ongoing mitoflash (Wei & Dirksen, 2012).

In Table 1 we compare the unitary properties, modulators, and possible molecular players in the pMPT, tMPT and fMPT. Briefly, all three types of MPTs are associated with abrupt dissipation of ΔΨm and non-selective passage of small solutes (at different molecular mass cutoffs). In addition, they are all cyclosporin A (CsA) sensitive and regulated by cyclophilin D (CypD), the fMPT in skeletal muscles being an exception. Elevating the mitochondrial global ROS using menadione or paraquat (Hou et al. 2013), photoreaction with Killer Red (Shen et al. 2014), or direct application of hydrogen peroxide (Shen et al. 2014) activates mitoflashes in a dose-dependent manner; likewise, oxidative stress stimulates both the pMPT and tMPT (Crompton et al. 1987, 1988; Zorov et al. 2000). Strikingly, sustained elevation of mitochondrial calcium to the submicromolar range (Rhod-2 ΔF/F₀ = 0.6–1.7) induces robust mitoflash activity via a slow onset process (Jian et al. 2013). In the presence of subthreshold global ROS elevation, even smaller mitochondrial Ca²⁺ increases (Rhod-2 ΔF/F₀ = 0.3) can greatly augment the mitoflash activity (Hou et al. 2013). In contrast, mitochondrial Ca²⁺ overload activates a pMPT only at supraphysiological levels (10 μm; Al-Nasser & Crompton, 1986; Crompton et al. 1988) while also displaying a strong synergism with ROS. Thus, the Ca²⁺ regulation of mitoflashes is qualitatively similar to that of pMPTs and fMPTs, yet, the former occur at physiological Ca²⁺ levels whereas the latter require excessive Ca²⁺ overload attainable only during extreme stress or damage.

Table 1.

Properties of pMPT, tMPT and fMPT

	pMPT	tMPT	fMPT
Molecular cutoff (permeability)	<1500 Da, permeable to large molecules	<300 Da, permeable to small solutes and ions such as H+, Ca2+ and K+	Up to 980 Da, permeable to Rhod-2 (Wang et al. 2008) and Rhod-2 analogue (Wang et al. 2012)
Mitochondrial swelling	Yes	No (Ichas & Mazat, 1998)	Yes (reversible)
ATP synthesis	Irreversible disruption	Transient disruption	Transient disruption
ETC dependence	No	Yes	Yes
Release of pro-apoptotic factors	Yes	No	No
CsA sensitivity	Yes	Yes	Yes (except in skeletal muscle)
Regulation by CypD	Yes	Yes	Yes (except in skeletal muscle)
Modulators
Ca2+	↑ (Haworth & Hunter, 1979; Hunter & Haworth, 1979a,b1979b), at high Ca2+ levels (>10 μm)	↑	Sensitive to submicromolar Ca2+ elevation, but with slow onset (Jian et al. 2013)
ROS	↑ (Crompton et al. 1987, 1988)	↑ (Zorov et al. 2000)	↑ (Hou et al. 2013; Shen et al. 2014)
Synergy between Ca2+ and ROS	Yes	ND	Yes (Hou et al. 2013)
Adenine nucleotide	↓ (Bogucka et al. 1995)	↓ (Evtodienko et al. 1994; Bogucka et al. 1995)	ND
Pi	↑ (Szabo et al. 1992; Ichas & Mazat, 1998)	↓ (Ichas & Mazat, 1998)	ND
	↓ (Basso et al. 2008; Di Lisa & Bernardi, 2009)
Acidification	↓ (Haworth & Hunter, 1979; Bernardi, 1992; Szabo et al. 1992)	↓ (Ichas & Mazat, 1998)	↓ at extreme acidification (Wei-LaPierre et al. 2013)
Alkalization	No effect (Ichas & Mazat, 1998)	↑ (Ichas & Mazat, 1998)	Mild ↑ (Wei-LaPierre et al. 2013)

ND, not determined; P_i, inorganic phosphate.

It has been shown that the mPTP is of highly variable conductance that may transit from low to high in a time-dependent and context-sensitive manner (Ichas & Mazat, 1998). We have reported that mitoflash ignition displays three different patterns: an abrupt rise from quiescence (44%), a rise with an exponential foot (27%), or a rise occurring after a pedestal precursor (29%) (Li et al. 2012). In principle, the polymorphism of mitoflash ignition could be ascribed to the multiplicity and variability in mPTP conductance. Alternatively, it might reflect a regenerative process dominated by stochastic, autonomous recruitment of a limited number of mPTPs in single mitochondria (Li et al. 2012). Future investigation is needed to distinguish among models of mPTP recruitment versus conductance transition.

Taken together, current data have shown close resemblances between fMPTs and tMPTs and striking quantitative distinctions between fMPTs and pMPTs. As to their molecular identities, it is possible that all three types of MPTs may stem from the same molecular complexes acting in entirely different regimes. Alternatively, they may originate from distinctive sets of molecular complexes with overlapping regulatory mechanisms. In either case, investigating mitoflashes should open a new avenue for elucidating the molecular basis of the MPT.

On another note, Aon, O'Rourke and colleagues have shown that mitochondrial energetic and redox variables oscillate autonomously under both physiological and pathological conditions (Aon et al. 2008). They coined the term ‘mitochondrial criticality’ (Aon et al. 2004) to describe the state in which the mitochondrial network of cardiomyocytes becomes very sensitive to small perturbations. At the point of criticality (e.g. under metabolic stress when the balance between ROS generation and ROS scavenging is perturbed), the mitochondrial network throughout the cardiac cell locks into one main low-frequency, high-amplitude oscillatory mode, and this alteration can scale up to the level of the whole organ, giving rise to fatal arrhythmias (Aon et al. 2004). Evidently, mitoflashes differ from mitochondrial oscillations because of their action potential-like features (Li et al. 2012). They are solitary, brief, high-amplitude events with a stereotyped time course and are confined to single mitochondria or local mitochondrial networks. The distribution of the intervals between consecutive events (seen on a confocal section) is consistent with the idea that mitoflash production is a Poisson process (Li et al. 2012). The involvement of fMPT is another distinctive feature of the mitoflash. Nevertheless, it is conceivable that the mitoflash mechanism may interact with those of mitochondrial oscillation, the latter setting the background and hence modulating the propensity for mitoflash production.

Signalling roles of mitoflashes

Several lines of evidence suggest that mitoflashes are active and dynamic signalling events, rather than by-products of mitochondrial respiration. First, they are actively controlled and generated by the fMPT (see above), allowing for regulation through means independent of mitochondrial respiration. Second, they occur at the expense of a significant sum of energy (e.g. collapse of ΔΨm and depletion of NADH and FADH₂ pools) (Wang et al. 2008; Wei-LaPierre et al. 2013). As such, their rate of occurrence can vary 10- to 100-fold, but not always in parallel with changes in respiratory activity. Third, they are universal and tightly coupled with vital mitochondrial signals (e.g. basal ROS and matrix Ca²⁺) (Hou et al. 2013; Jian et al. 2013; Wang et al. 2014). More importantly, mitoflash activity is a rapid, robust and highly sensitive responder to changes in metabolic state, developmental signal transduction, ageing and pathological stressors (Fig. 3).

Figure 3. Schematic view of possible signalling roles of mitoflashes.

Mitoflashes serve as an early predictor when lifespan is altered by diverse genetic, environmental and stochastic factors in C. elegans, and participate in early developmental (proliferation) and pathophysiological stress responses (hyperosmotic and apoptotic stress). Note that, except for a few cases, little is known about the specific pathways involved in mitoflash signalling. Continuous lines, activation (→) or inhibition (―|); dashed arrows, simplified signal pathway.

By generating mt-cpYFP transgenic mice and performing in vivo imaging of mitoflashes, we have demonstrated that their frequency surges in response to systemic glucose challenge or insulin stimulation (Fang et al. 2011). A similar mitoflash response has also been found in isolated flexor digitorum brevis muscle fibres with either glucose or pyruvate stimulation (Pouvreau, 2010; Wei et al. 2011). These results indicate that mitoflash frequency decodes the rate of cell metabolism. Mitoflashes also participate in early developmental signalling. In cultured embryonic mouse cerebral cortical neural progenitor cells (NPCs), decreasing mitoflash frequency by ROS scavengers or CsA treatment enhances NPC proliferation, whereas increasing it has the opposite effect (Hou et al. 2012). NPCs lacking mitochondrial MnSOD exhibit a significant increase in mitoflash frequency accompanied by a decrease in proliferation. Importantly, fewer proliferative NPCs and differentiated neurons at mid-gestation are found in the embryonic cerebral cortex of MnSOD null mice as compared with wild-type littermates (Hou et al. 2012). These results suggest that mitoflashes also participate in early developmental signalling. Remarkably, that mitoflashes mediate a negative regulation of NPC proliferation is in contrast to the role of global ROS which is generally thought to promote cell differentiation (Hou et al. 2012). Regarding the cellular signalling pathways involved, it has been shown that inhibition of mitoflashes promotes ERK1/2 activation, a mitogen-activated protein kinase (MAPK) that is typically activated by growth factor stimulation and promotes cell proliferation (Hou et al. 2012). In a most recent study, we have shown the surprising result that mitoflash frequency is an early predictor of lifespan in C. elegans. By tracking mitoflash activity in the pharyngeal muscles of live worms through the whole life-cycle, we found that it peaks on day 3 during active reproduction and on day 9 when animals start to die off. The mitoflash frequency on day 3, but not day 9, negatively correlates with the lifespan of the worm across diverse genetic, environmental and stochastic factors. Given these findings, we have speculated that the tempo of the mitoflash clock in young adults faithfully reflects the rate of the ageing clock (Shen et al. 2014).

The participation of mitoflashes in pathophysiological signalling has been reported by several groups. Mitoflash activity is largely reduced and even abolished during sustained anoxic or hypoxic treatment, and overshoots upon reoxygenation (Wang et al. 2008; Huang et al. 2011), suggesting a potential role of mitoflashes in the oxidative stress of ischaemia–reperfusion. In HeLa cells, mitoflashes participate in the hyperosmotic stress response: a reversible 20-fold increase of mitoflash frequency occurs in response to hyperosmotic stress, and inhibition of the mitoflash response by either CsA or mitoTEMPO, a mitochondrial targeted antioxidant, largely prevents the activation of JNK and p38, two MAPKs essential for the induction of adaptive responses required for cell survival (Hou et al. 2013). Moreover, mitoflash frequency is linked to muscle activities: it is significantly increased following brief tetanic stimulation and markedly decreased following prolonged tetanic stimulation (40 tetani) in skeletal muscle fibres (Wei et al. 2011). Specifically, in the fibres from RYR1^Y522S/WT mice, a model of malignant hyperthermia, a significant temperature-dependent increase of mitoflash frequency has also been documented (Wei et al. 2011). Pro-inflammatory cytokines, interleukin-1β (IL-1β), or tumour necrosis factor-α (TNF-α) stimulates mitoflash activity by 2- to 5-fold in cultured chondrocytes and intact cartilage (Cao et al. 2013). In oxidative stress-induced apoptosis, we have demonstrated that mitoflashes constitute early mitochondrial signals that are not associated with detectable release of cytochrome c. Yet, they accelerate the progression of cell death and serve as a convergence point for pro- and anti-apoptotic regulation mediated by the CypD and Bcl-2 proteins (Ma et al. 2011). In addition, an increase of mitoflash incidence is an early intracellular event in the NPC response to amyloid beta-peptide-mediated inhibition of proliferation (Hou et al. 2014). Increased mitoflash activity has also been reported in both fibroblasts from patients with Huntington's disease (HD) and the HD mouse model, while the excessive mitoflashes are associated with an overall doubled level of mtDNA mutation, suggesting a role for mitoflash hyperactivity in the pathology of HD (Wang et al. 2013). Notably, the contribution of such signalling ROS in the form of mitoflashes to global homeostatic ROS is often marginal (Ma et al. 2011; Hou et al. 2012), further distinguishing mitoflashes from basal ROS.

Perspective

Recent advances have shown that mitoflashes represent a universal and fundamental mitochondrial activity that plays important signalling roles. The bursting superoxide production in a mitoflash is different from the flashless basal ROS production in terms of ignition, regulation, quantity and function. In particular, the mitoflash is tightly coupled to a transient and reversible MPT that is remarkably similar to the tMPT.

Investigation of the signalling roles of mitoflashes is capturing much attention and imagination because it unravels new insights into mitochondrial ROS signalling and functional dynamics as well as molecular constituents of fMPT. Acting in a frequency-encoded manner, mitoflashes intertwine with mitochondrial core functions and participate in pivotal processes, including metabolism, cell fate regulation, ageing and the stress response across species from worms to humans. Emerging data show that mitoflashes and global ROS may fulfil distinctive and even opposing cellular functions.

At present, much more endeavour is needed to elucidate the mitoflash mechanism and to discriminate true biological signals and unveil signalling pathways stemming from the mitoflash. For instance, what physiological signals directly trigger an fMPT? What are the exact molecular constituents of an fMPT? On which sites are superoxide anions produced and how could these sites operate synchronously in a burst? In appreciation of the mitoflash as a complex phenomenon, which subset of changes conveys the mitoflash signals and how are they further transduced at the organelle level and beyond? Answering these and many related challenging questions would lay the cornerstones for a better understanding of mitochondrial dynamic function and signalling through the interrogation of mitoflashes.

Glossary

CsA

cyclosporin A

CypD

cyclophilin D

cpYFP

circularly permuted yellow fluorescent protein

DCF

2,7-dichlorodihydrofluorescein diacetate

ETC

electron transport chain

fMPT

putative reversible mitochondrial permeability transition underlying the mitoflash

MAPK

mitogen-activated protein kinase

MPT

mitochondrial permeability transition

mPTP

MPT pore

NPCs

neural progenitor cells

pMPT

permanent MPT

ROS

reactive oxygen species

SOD

superoxide dismutase

tMPT

transient MPT

Additional information

Competing interests

None declared.

Funding

This work was supported by the National Key Basic Research Program of China (2013CB531200 and 2011CB809100) and the National Science Foundation of China (31221002, 31130067 and 31327901).