The mitochondrial permeability transition pore: an evolving concept critical for cell life and death

ABSTRACT

In this review, we summarize current knowledge of perhaps one of the most intriguing phenomena in cell biology: the mitochondrial permeability transition pore (mPTP). This phenomenon, which was initially observed as a sudden loss of inner mitochondrial membrane impermeability caused by excessive calcium, has been studied for almost 50 years, and still nodeinitive answer has been provided regarding its mechanisms. From its initial consideration as an in vitro artifact to the current notion that the mPTP isa phenomenon with physiological and pathological implications, along road has been travelled. We here summarize the role of mitochondria in cytosolic calcium control and the evolving concepts regarding the mitochondrial permeability transition (mPT) and the mPTP. We show how the evolving mPTP models and mechanisms, which involve many proposed mitochondrial protein components, have arisen from methodological advances and more complex biological models. We describe how scientiic progress and methodological advances have allowed milestone discoveries on mPTP regulation and composition and its recognition as a valid target for drug development and a critical component of mitochondrial biology.

Keywords: mitochondria, mitochondrial permeability transition pore, calcium, cell death

I. INTRODUCTION

The mitochondrial permeability transition (mPT) is a pathophysiological state of the inner mitochondrial membrane (IMM). Under favourable conditions, including calcium (Ca2+) overload, oxidative stress, increased phosphate concentration, and decreased adenine nucleotide availability, the IMM becomes highly permeable to solutes with a molecular weight of up to 1500 Da (Halestrap, 2009; Morciano et al; 2015). Since biological and evolutionary features generally make the IMM relatively impermeable, the mPT leads initially to a considerable and unselective inﬂux of solutes and to an abrupt loss of mitochondrial metabolites, causing mitochondrial homeostasis perturbation that can result in cell death (Izzo etal; 2016). ThemPT is evolutionarily highly conserved, with critical features present in yeast, mammals, and plants; only the crustacean Artemia franciscana seems to lack a regulated mPT, which may contribute to its longlasting hypoxia tolerance (Menze et al 2005). The mPT is caused by the opening of proteinaceous channels at the juxtaposition of the IMM and outer mitochondrial membrane (OMM) rather than changes to the phospholipid bilayer composition (Crompton, Costi & Hayat, 1987). The opening of mPT pores (mPTPs) opening can be monitored using ﬂuorescent dyes or absorbance assays in living cells, isolated mitochondria and tissues. Pore opening results from relatively severe perturbations of the mitochondrial matrix and the consequences of mPT are dictated by several factors such as the pore open time, the number of channels present, and the number of mitochondria affected by this pathophysiological event. Along-lasting mPT can lead to irreversible consequences, including matrix swelling, dissipation of mitochondrial potential, ATP hydrolysis into ADP and free phosphate ions, and uncoupling of oxidative phosphorylation, causing cell death. The modalities by which mPT leads to cell death are still debated; based on recent reports, apoptosis or necroptosis may occur depending on ATP concentration and availability (Brenner & Moulin, 2012).

mPTP opening, considered by some authors as a cellular catastrophe (Briston et al; 2019), has been considered a critical mechanism in the development of several pathologies and organ damage caused by toxicity. For nearly two decades, the mPTPhas been associated with cardiac dysfunction observed after ischemia/reperfusion of the heart (Borutaite et al; 2003). In early work, Steenbergen et al. (1987) demonstrated an increase in cytosolic Ca2+ concentration after an ischemic episode. Cytosolic Ca2+ overload induces the opening of the mPTP and, consequently the release of cytochrome c from mitochondria to the cytosol and activation of the mitochondria-dependent apoptotic pathway (Whittington et al; 2018). During ischemia, the mPTP remains closed because of cytosolic acidiication since protons are mPTP inhibitors (Ong et al; 2015). By contrast, opening of the mPTP is strongly potentiated during reperfusion after an ischemic episode. During reperfusion, thereintroduction of oxygen has a deleterious effect, resulting in a burst of reactive oxygen species (ROS) production (Di Lisa et al; 2001; Hausenloy, Duchen & Yellon, 2003; GonzalezMontero et al; 2018). ROS potentiate mPTP opening, and this can result in cardiac cell death and cardiac dysfunction. mPTP opening after episodes of ischemia/reperfusion was also observed in the brain (Halestrap, 2006). Inhibition of mPTP opening has been suggested as a neuroprotective strategy to prevent cerebral ischemia–reperfusion injuries (Leger et al; 2011; Rekuviene et al; 2017; Matsumoto et al; 2018). The mPTP is also associated with metabolic diseases, including insulin-resistance or diabetes: Taddeo et al. (2014) demonstrated that mPTP opening is required for induction of insulin resistance in skeletal muscle. Work performed in animal models demonstrated how the mPT is induced in diabetic conditions, which may contribute to complications including diabetic cardiomyopathy or hyperglycemia (Oliveira et al; 2003; Lumini-Oliveira et al; 2011; Taddeo et al; 2014; Riojas-Hernandez et al; 2015). As another example, accumulation of hydrophobic bile acids in hepatic cells during cholestasis induces apoptosis of hepatic cells through mPTP opening (Yerushalmi et al; 2001; Rolo, Palmeira & Wallace, 2003).

The opening of the mPTP may also be observed during drug-induced toxicity. For example, high doses of caffeine enhance the Ca2+-dependent cardiac mPT in isolated mitochondrial fractions (Sardao, Oliveira & Moreno, 2002),causing concern over the excessive consumption of high-energy drinks and dietary supplements with high doses of caffeine. Doxorubicin,a potent chemotherapeutic drug but with associated cardiotoxicity, decreases the threshold for mPTP opening in cardiac cells, impairing the contraction performance of the heart (Montaigne et al; 2011),as demonstrated in rodent models (Oliveira et al; 2004, 2005; Pereira et al; 2011). The role of the mPTP in cell toxicity induced by xenobiotic compounds that cause oxidative stress and in different diseases has been extensively explored (Bernardo et al; 2013; Carvalho et al; 2018; Teixeira et al; 2018). Due to the importance of the mPTP for mitochondrial function and, consequently, cellular performance, the dynamics and regulation of the mPT have been the subject of intensive research.

Although mPTP opening is often associated with disease conditions, there is evidence for a critical physiological role in its ﬂickering or transient opening mode, especially in the heart (during cardiac development and in damage protection) (Hausenloy et al; 2004; Korge et al; 2011), brain (putative or indirect involvement in synaptic eficacy and plasticity) (Mnatsakanyan et al; 2017), and in metabolic functions (Hom et al; 2011).

II. THE mPTP AND MITOCHONDRIAL Ca2+ UPTAKE OR OVERLOAD

The cellular concentration of free Ca2+ regulates an array of biochemical reactions and is crucial for signal transduction (Rizzuto, Duchen & Pozzan, 2004; Elustondo et al; 2017; Herst et al; 2017; Krebs, 2017; Santulli, 2017; Giorgi et al; 2018a; Giorgi, Marchi & Pinton, 2018b; Del Re et al; 2019; Glaser et al; 2019; Hausenloy et al; 2020). Mitochondria are fundamental to cellular energy metabolism, supplying energy in the form of ATP and affecting cell physiology through the regulation of Ca2+ homeostasis (Picard, Wallace & Burelle, 2016; Krebs, 2017). Mitochondria have a large capacity to accumulate Ca2+ and can transiently store it, thus contributing to cell calcium homeostasis. Their ability to accumulate calcium for later release makes mitochondria essential cytosolic stores or buffers for Ca2+ in the context of cell physiology and pathophysiology (Dedkova & Blatter, 2008; Elustondo etal; 2017; Ludtmann & Abramov, 2018; Delierneux et al; 2020). Intramitochondrial free calcium plays a signiicant part in Ca2+ homeostasis in cells, also being important in cell survival and death (Bhosale et al; 2015; Picard et al; 2016; Santulli, 2017; Ludtmann & Abramov, 2018; Del Re et al; 2019). It has been demonstrated that a basal level of Са2+ in the mitochondrial matrix is needed for correct mitochondrial functioning, while the pathophysiological role of Ca2+ overload, which occurs in a wide range of pathologies, still remains to be clariied (Burgoyne et al; 2012; Bertero & Maack, 2018; Ludtmann & Abramov, 2018).

(1) Mitochondrial calcium inﬂux

Calcium accumulation in mitochondria was irst described in the 1960s (Deluca & Engstrom, 1961). Since then, the role of Ca2+ in the regulation of mitochondrial bioenergetics and diverse cellular functions has been well established. Calcium homeostasis in mitochondria is regulated by a complex system of mitochondrial Ca2+ inﬂux and efﬂux mechanisms. This Ca2+ transport system in mitochondria comprises speciic transporters in the IMM and OMM. To access the mitochondrial matrix, Ca2+ must irst pass through the OMM. This membrane is permeable to ions, in particular to Ca2+, and to small proteins, due to the presence of a large conductance channel – the voltage-dependent anion channel (VDAC) – which allows the exchange of molecules of molecular weight up to 1500 Da (Schein, Colombini & Finkelstein, 1976; Colombini & Mannella, 2012; Krebs, 2017; Becker & Wagner, 2018; Magri, Reina & de Pinto, 2018). The VDAC is responsible for Ca2+ transport from the cytoplasm into mitochondria, with its permeability controlled by ATP and other regulatory factors. Three different VDAC isoforms have been identiied: VDAC1, VDAC2 and VDAC3 (Mertins, Psakis & Essen, 2014; Krebs, 2017; Ponnalagu & Singh, 2017). Although these isoforms share some structural and functional properties, they appear to perform different physiological roles (Cheng et al; 2003; De Pinto et al; 2010; Magri et al; 2018; Rostovtseva et al; 2020). While limited information is available regarding the functions of VDAC2 and VDAC3 in the Ca2+-inﬂux mechanism (De Pinto et al; 2010; Lemasters et al; 2012; Magri et al; 2018), VDAC1 has been the subject of detailed research. VDAC1 is highly expressed in most cells (Shoshan-Barmatz & Golan, 2012) and seems to be the most prevalent isoform involved in Ca2+ transport into the intermembrane mitochondrial space (Krebs, 2017).

Ca2+ transport through the IMM is regulated via several transporters. At present, three main mechanisms of Ca2+ inﬂux through the IMM are proposed (Fig. 1): (i) an electrogenic mitochondrial Ca2+ uniporter (MCU), (ii) the rapid mode of Ca2+ uptake (RaM), and (iii) the mitochondrial ryanodine receptor (mRyR). Three main mechanisms of Ca2+ efﬂux through the IMM are also known: (i) a Na+/Ca2+ exchanger (NCXm), (ii) a H+/Ca2+ exchanger (HCXm), and (iii) the mPTP. The leucine zipperEF-hand containing transmembrane protein (LETM1) was also proposed as a Ca2+-transport system. However, its role in mitochondrial Ca2+ inﬂux and efﬂux through the IMM is still under discussion. Ca2+ transport across the IMM was initially thought to involve a single mechanism that was demonstrated to be highly sensitive to ruthenium red and lanthanides (Gunter & Pfeiffer, 1990). The molecular identity of this rutheniumand lanthanide-sensitive Ca2+ transport was unclear for several decades until this transporter was identiied as the MCU complex, following identiication of the gene encoding the pore-forming subunit of the MCU (Baughman et al; 2011; De Stefani et al; 2011).

Currently, Ca2+ inﬂux through the MCU multi-protein complex is the best-characterized pathway for mitochondrial Ca2+ uptake. It is driven by the large electrochemical gradient (mitochondrial membrane potential 曾 180 mV) for Ca2+ across the IMM (Elustondo et al; 2017; Mishra et al; 2017; Mammucari et al; 2018; Belosludtsev et al; 2019). It is now considered that this multi-protein MCU complex adapts to multiple states and is composed of several subunits, including transmembrane core components and membrane-associated regulatory subunits in the intermembrane space. Three proteins have been identiiedascore components of theMCU complex: the mitochondrial Ca2+ uniporter (MCU), theMCU dominant negative beta subunit (MCUb), and the essential MCU regulator (EMRE) (Raffaello et al; 2013; Sancak et al; 2013; Mishra et al;2017;Mammucarietal;2018;Cui et al;2019;Wang, Baradaran & Long, 2020a). The MCU gene (previously known as CCDC109a; 40 kDa protein) was identiied through bioinformatics screening of the MitoCarta database, a compendium of mitochondrial proteinsidentiied by mass spectrometry analyses on mitochondrial preparations from different mouse tissues (Baughman et al; 2011; De Stefani etal; 2011; Mammucari et al; 2018). TheMCUb gene (CCDC109b;33kDa protein) was identiied through an MCU sequence homology screening (Raffaello et al; 2013), and the incorporation of MCUb into the MCU complex has also been demonstrated by proteomic experiments (Sancak et al; 2013). EMRE (C22ORF32; 10 kDa protein) was the last identiied component of the MCU pore complex (Sancak et al; 2013) and is essential for MCU activity, as demonstrated by experiments in EMRE knockout cells (Patron et al; 2014). EMRE has been proposed to play a fundamental role in interactions between the pore core subunits and the regulatory subunits (Sancak et al; 2013; Mammucari et al; 2018).

The family of mitochondrial Ca2+ uptake proteins (MICU 1– 3), mitochondrial Ca2+ uniporter regulator 1 (MCUR1), and SLC25A23 are now considered the main membraneassociated regulatory subunits of the MCU multi-protein complex (Hoffman et al; 2014; Patron et al; 2014; Wang et al; 2014; Mishra et al; 2017; Marchi et al; 2019; Vais et al; 2020). MICU1 (CBARA1/EFHA3; 54 kDa protein) is a soluble (or membrane-associated) protein in the intermembrane space. It is proposed to be pivotal in both gatekeeping and cooperative activation of MCU; keeping the channel closed under resting conditions (Csordas et al; 2013; Patron et al; 2014; Wang et al; 2014; Foskett, 2020). Other isoforms of theMICU family of proteins – MICU2 (known as EFHA1) and MICU3 (known as EFHA2) – are also identiied as regulatory subunits of the MCU multi-protein complex, and display EF-hand domains in their protein structure, but share only 25% sequence identity with MICU1 (Plovanich et al; 2013). However, the location of MICU proteins and the nature of MICU-dependent regulation are still controversial (Vais et al; 2020; Wu et al; 2020; Wang et al; 2020b). The diverse functions of MICU family proteins maintain normal mitochondrial Ca2+ levels under resting conditions and enable prompt activation of the MCU to mediate rapid mitochondrial Ca2+ uptake (Cui et al; 2019).

MCUR1 (CCDC90A; 40 kDa protein), which consists of two transmembrane domains and one coiled-coil region, was also demonstrated to be a regulatory component of the MCU complex (Mallilankaraman et al; 2012). More recently, it was shown that MCUR1 binds to the MCU pore and EMRE through their coiled-coil domains that stabilize the MCU complex (Tomar et al; 2016).

SLC25A23, which belongs to a family of Mg-ATP/Pi solute carriers, was also proposed as an essential component of the MCU complex (Bassi et al; 2005; Hoffman et al; 2014; Krebs, 2017). A mutation of the EF-hand domain of SLC25A23 reduces mitochondrial Ca2+ accumulation, but whether this depends on direct MCU activity regulation or whether it affects mitochondrial bioenergetics or mitochondrial Ca2+ buffering capacity is still debated (Bassi et al; 2005; Rueda et al; 2015; Mammucari et al; 2018). In addition to the MCU, there are other mitochondrial Ca2+ inﬂux mechanisms, including the mRyR, RaM, and LETM1, which all have unique biophysical properties that differ from those of theMCU (Gunter & Gunter, 2001; Beutner et al; 2005; Jiang, Zhao & Clapham, 2009; Elustondo et al; 2017; Krebs, 2017; Mammucari et al; 2018).

mRyR is the largest known ion channel (about >2 MDa), localized in the IMM, and can function as an alternative mechanism for mitochondrial Ca2+ uptake, especially in the mitochondria of cardiac and neuronal cells (Beutner et al; 2001, 2005;Jakob et al; 2014). Three different subtypes of RyR isoforms (RyR1, RyR2, and RyR3) with different pharmacological properties and tissue-speciic expression have been described. RyR1, the primary isoform in skeletal muscle, was identiied in the IMM of isolated heart mitochondria through [3H]ryanodine binding, immunogold labelling and Western blot techniques. It is thought to mediate ryanodine-sensitive, rapid mitochondrial Ca2+ transport and is believed to play a central role in mitochondrialCa2+ uptake (Beutner etal; 2001, 2005). RyR2 is mostly present in cardiac muscle cells (Bhat et al; 1999), while RyR3 is widely expressed in the endoplasmic reticulum (ER) of different vertebrate tissues (Giannini et al; 1995) and maybe coexpressed with RyR1 and RyR2. There are suggestions that under certain situations (e.g. mitochondrial Ca2+ overload), mRyR channels may also mediate Ca2+ efﬂux (Ryu et al; 2010).

In isolated heart mitochondria, RaM has been described as an additional mechanism for Ca2+ transport, capable of sequestering signiicant amounts of Ca2+ hundreds of times faster than the MCU (Gunter & Gunter, 2001). RaM is activated only transiently, facilitates rapid sequestration of Ca2+ by mitochondria at the beginning of each cytosolic Ca2+ pulse, and rapidly recovers between pulses, allowing mitochondria to respond to repeated Ca2+ transients (Sparagna et al; 1995). Compared with the MCU, this transporter is activated at much lower Ca2+ concentrations (−50 to 100 nM versus >500 nM) (Sparagna et al; 1995). However, a protein responsible for this rapid mode of Ca2+ uptake has not been identiied, although it has been speculated that the RaM comprises a substate of MCU operation since both are inhibited by ruthenium red and RaM activity does not appear to be present in MCU-knockout mitochondria (Baughman et al; 2011; De Stefani et al; 2011).

LETM1 was initially identiied as a K+/H+ exchanger; however, it was later reported as a Ca2+/H+ antiporter, localized to the IMM (Jiang et al; 2009). LETM1 transports Ca2+ bidirectionally across the inner membrane, depending on the pH gradient, and is inhibited by ruthenium red (Jiang et al; 2009). However, further studies are needed to characterize its role in Ca2+ transport, as well as its sensitivity to ruthenium red, given recent demonstrations in which LETM1 protein was reconstituted in liposomes and was demonstrated to be a ruthenium red-insensitive electroneutral Ca2+/2H+ antiporter (Tsai et al; 2014). Other studies have reinforced the role of LETM1 in K+ homeostasis and suggested that it functions as an electroneutral H+/K+ exchanger (Nowikovsky & Bernardi, 2014). This hypothesis was supported by results showing that LETM1 was not responsible for efﬂux of Ca2+ from the mitochondria (De Marchi et al; 2014b). There are some suggestions that the role of LETM1 could change according to speciic conditions (O-Uchi, Pan & Sheu, 2012; Austin & Nowikovsky, 2019).

(2) Mitochondrial calcium efﬂux

While the biochemical characteristics and physiological functions of the mitochondrial systems for Ca2+ inﬂux have been widely studied, understanding the molecular nature and properties of mitochondrial Ca2+ efﬂux systems has just begun, although functional characterization of this system began in the 1970s (Carafoli et al; 1974). Currently, two separate mechanisms have been proposed to account for Ca2+ extrusion from the mitochondrial matrix: Na+-dependent (NCXm) and Na+-independent (HCXm).

In most cells, the main mechanism of Ca2+ extrusion from mitochondria is the NCXm. Although Na+-dependent Ca2+ efﬂux from mitochondria was irst discovered in isolated rat heart mitochondria and was described as the mitochondrial Na+/Ca2+ exchanger years ago (Carafoli et al; 1974), its molecular identity was also only recently resolved (Palty et al; 2010). It appears to extrude Ca2+ from themitochondrial matrix to the intermembrane space and, more speciically, to constitute a Na+/Ca2+/Li+ exchanger in the IMM (Palty et al; 2010; Palty, Hershinkel & Sekler, 2012). This Na+-dependent Ca2+ exchange activity is benzodiazepine and CGP-37157 sensitive and is found in a wide variety of tissues. Although it is dominant in the heart, brain, skeletal muscle, parotid gland, adrenal cortex, and brown fat (Gunter et al; 2004; Takeuchi, Kim & Matsuoka, 2015), it is also present in liver, kidney, and lung mitochondria, although its activity there is lower (Haworth, Hunter & Berkoff, 1980). The NCXm is active primarily in excitable cells and, in contrast to the plasma membrane Na+/Ca2+ exchanger, it uniquely has the ability additionally to transport Li+ ions (Carafoli etal; 1974; Palty et al; 2004). The stoichiometry (ion-exchange ratio) and the electrogenicity of the NCXm were controversial, but it was believed to be electroneutral (Affolter & Carafoli, 1980; Wingrove & Gunter, 1986b). The use of permeabilized rat ventricular myocytes demonstrated that the NCXm is voltagedependent and electrogenic, suggesting a stoichiometry higher than 3Na+ for one Ca2+ (Kim & Matsuoka, 2008). This stoichiometry and the electrogenic nature of the NCXm were proved recently using the whole-mitoplast patch-clamp technique (Islam, Takeuchi & Matsuoka, 2020). Detailed mechanisms of the regulation of NCXm activity and sensitivity to effectors have not yet been clariied, but some studies demonstrate its regulation by a stomatin-like protein 2 (Da Cruz et al; 2010) and the mitochondrial phosphatase and tensin homolog deleted on chromosome 10 (PTEN)induced kinase 1 (PINK1) (Gandhi et al; 2009).

In the tissues with low NCXm activity (e.g. the liver, kidney, lung, and smooth muscle; Takeuchi et al; 2015), the HCXm has a dominant effect on the release of Ca2+ from mitochondria (Gunter & Pfeiffer, 1990). The molecular identity of the HCXm is still debated, but most likely its activity is electroneutral with a stoichiometry 2H+ for one Ca2+ (Gunter, Zuscik & Gunter, 1991). Studies of HCXm activity in rat isolated liver mitochondria demonstrated that the rate of efﬂux via the HCXm decreases with increasing pH gradient and suggest that this mechanism is an active rather than passive Ca2+ for 2H+ exchanger (Gunter et al; 1991).

LETM1 is considered an additional and/or alternative mechanism of Ca2+ efﬂux from mitochondria (Nowikovsky et al; 2012; Takeuchi et al; 2015; Krebs, 2017; Austin & Nowikovsky, 2019). Ca2+ transport through the IMM could be mediated by LETM1, since this protein functions as a Ca2+/H+ antiporter under certain conditions (Mailloux & Harper, 2011). Studies using digitonin-permeabilized S2 or 293 cells expressing the mitochondrial Ca2+-sensor protein pericam, and puriied protein reconstituted in liposomes, showed that LETM1 mediates H+/Ca2+ exchange (Jiang et al; 2009). Later work using reconstituted proteoliposomes reported that LETM1 mediates electroneutral 2H+/Ca2+ antiport, which is insensitive to ruthenium red (Tsai et al; 2014). Further studies focused on intracellular-storedependent Ca2+ dynamics provided evidence that LETM1 acts as a 2H+/Ca2+ exchanger (Huang et al; 2017). Combined with a theoretical analysis (Nowikovsky et al; 2012), this lends credibility to the role of LETM1 as an important Ca2+ efﬂux mechanism.

Another important mechanism for Ca2+ release from mitochondria is suggested as the transiently open form of the mPTP. Under pathophysiological conditions, in which this high-conductance non-speciic pore opens, it may function as a Ca2+-efﬂux system (Takeuchi et al; 2015; Biasutto et al; 2016; Hurst, Hoek & Sheu, 2017; Britti et al; 2018; Briston et al; 2019). Opening of the mPTP is directly regulated by the concentration of free calcium, is triggered by Ca2+ overload, and enables rapid Ca2+ efﬂux (Hurst et al; 2017). Detailed information regarding the function of the mPTP in Ca2+ transport is provided in Section VII.

It is important to note that the localization of mitochondria within the cell is a crucial factor for mitochondrial Ca2+ uptake. It is now well accepted that the location of mitochondria in proximity to the plasma membrane or the ER/sarcoplasmic reticulum (SR) is essential in modulating a variety of cellular functions, as well as in Ca2+ transport to subcellular structures, in particular to mitochondria (Rowland & Voeltz, 2012; van Vliet, Verfaillie & Agostinis, 2014; Stefan, 2018; Silva et al; 2020). These interactions between mitochondria and the ER/SR have been described as physiologically and pathophysiologically signiicant for Ca2+ crosstalk between mitochondria and other cellular and subcellular structures (Giacomello et al; 2010; Rowland & Voeltz, 2012; van Vliet et al; 2014; Takeuchi et al; 2015; Rieusset, 2018; Yousuf et al; 2020). During Ca2+ movement through the plasma membrane or its release from the ER/SR, these interactions promote Ca2+ inﬂux to neighbouring mitochondria, determining the particular properties of Ca2+ transport into these organelles (Lawrie, Zolle & Simpson, 1997; Park et al; 2001; Giacomello et al; 2010). There is some evidence of Ca2+ channelling through plasma membrane Ca2+ channels to nearby mitochondria, as well as in the opposite direction (Korzeniowski et al; 2009). Mitochondria–ER/SR communication is also reported as an essential regulatory factor in a variety of cellular processes, including Ca2+ signalling, lipid biosynthesis, and mitochondrial division (Friedman et al; 2011; Rieusset, 2018; Granatiero et al; 2019; Namgaladze, Khodzhaeva & Brune, 2019). It is now clear that bidirectional Ca2+ crosstalk between both mitochondria and the plasma membrane and mitochondria and the ER/SR is crucial for regulating a wide range of cellular functions.

(3) Ca2+ induces mitochondrial swelling

Hunter, Haworth & Southard (1976) irst introduced the concept of Ca2+-induced mitochondrial swelling. In a series of seminal experiments performed on isolated beefheart mitochondria, the authors observed that when low levels of Ca2+ were added, mitochondria underwent a conigurational transition from ashrunken matrix with large intracristal spaces to a swollen matrix with decreased intracristal spaces (Fig. 2). Mitochondriawere described to transition from anaggregated to an orthodox coniguration in this process. Using electron microscopy, it was concluded that for each time point postCa2+ addition, the mitochondrial population was heterogeneous with only the two different conigurations present. The authors argued that the transition sequence for each mitochondrion consisted of a lag phase in the aggregated form followed by asudden transformation to the orthodox form.

To assess the correlation between this conigurational change and membrane permeability, Hunter et al. (1976) measured the permeability of mitochondrial membranes of isolated beefheart mitochondria to [14C]sucrose in the presence of Ca2+. In samples taken at the same time points as those used for electron microscopy imaging, it was observed that permeability to [14C]sucrose increased simultaneously with the decrease in the number of aggregated mitochondria. This important result made it clear that Ca2+increased the permeability of mitochondria to [14C]sucrose. Importantly, the swelling that follows the addition of Ca2+ was caused by mitochondrial osmotic water inﬂux that accompanies sucrose entry into the matrix space (Hunter et al; 1976).

Further experiments addressed the effect of IMM permeability on mitochondrial respiratory activity. The uncoupler respiratory control index (i.e. the magnitude of the respiration increase) was compared with conigurational changes at several time points after the addition of Ca2+. This analysis showed that coupling in an individual mitochondrion follows an all-or-nothing law, precisely as for the conigurational transition. However, experiments using hypotonically swollen mitochondria led to the conclusion that the changes in coupling are caused by changes in the permeability of the IMM, rather than changes in conigurational transition (Hunter et al; 1976). This clariied previous misconceptions (Harman & Feigelson, 1952; Lehninger, 1962) that mitochondrial swelling by itself did not determine mitochondrial functioning. Hunter et al. (1976) also observed that the transition was reversible and that arsenate, phosphate, and fatty acids induced the phenomenon.

In later studies, light scattering was used to measure mitochondrial swelling, based on changes in pseudo-absorbance of a suspension of beef heart mitochondria (Hunter & Haworth, 1979a). This methodology, still used today for isolated mitochondrial fractions, enabled the authors to show that mitochondrial metabolites and small molecules and ions can counteract the Ca2+-induced membrane transition: endogenous NADH, ADP, and Mg2+ all prevented the mPT. The same effect could be achieved with mitochondrial energization (Hunter et al; 1976). Following these pioneer discoveries, Haworth & Hunter (1979) then explored the mechanisms by which Ca2+ leads to mPT, and how its inhibitors could function.

One important observation was that mitochondria which had been previously subjected to Ca2+-induced mPT lost their endogenous protection. Also, the concentration of Ca2+ in the incubation medium was able to regulate the permeability of the IMM ina dose-dependent manner, as seen by measuring light scattering in suspensionstitrated with different concentrations of Ca2+. Together with the fact that the membrane transition needed no energy source (Hunter & Haworth, 1979a), this led to the conclusion that the physical binding of Ca2+ to units in the IMM could cause the mPT. An important observation was that each preventive agent showed a speciic mode of action by which it counteracted the membrane transition. For example, Mg2+ was observed to inhibit the transition competitively. Hunter et al. (1976) had already proposed that Mg2+ could bind to ‘trigger sites ’, but Ca2+, with an apparently higher afinity for those sites, could exclude Mg2+and induce the transition. The effects of ADP and NADH were investigated in more detail later (Haworth et al; 1980). ADP and NADH have synergistic effects on inhibition of the Ca2+-induced transition in isolated mitochondria. ADP was thought to have an inner membrane binding site which, when bound to ADP, could signiicantly inhibit binding of Ca2+ to the trigger site and therefore inhibit the transition. The inhibition by NADH was observed to be reduced by NADPH (Haworth et al; 1980). Several cations were identiied as competitive inhibitors of the modulating effects of Ca2+ (Hunter & Haworth, 1979a), including divalent cations, such as Sr2+ and Mn2+, and the monovalent cations, K+, Na+, and tetramethylammonium (TMA). The trivalent metal inhibitor La3+ ion exhibited competitive inhibition more potent than these mono and divalent cations. Protons (H+) were also a competive inhibitor, an observation that formed the basis for studies that eventually led to inhibition of the mPT by ischemia-induced cytosolic acidiication (Qian et al; 1997). These observations suggested that the trigger site could be involved in high-afinity Ca2+ uptake. However, Ca2+induced transition progression was found to be ruthenium red insensitive. Because high-afinity Ca2+ uptake is sensitive to ruthenium red (Moore, 1971), Hunter & Haworth (1979a) assumed that another mitochondrial site, independent from those involved in high-afinity Ca2+ uptake, must be involved in the transition.

Hunter & Haworth (1979b) further investigated pathways for the release of Ca2+ that occurred during the transition. They concluded that Ca2+ release was due to the Ca2+induced transition because: (i) agents that blocked the transition, such as Mg2+, ADP and bovine serum albumin (BSA), inhibited Ca2+ release under steady-state respiration; (ii) mitochondria that accumulated Sr2+, did not release it subsequently,indicating that the spontaneous release mechanism was selective for Ca2+, similar to the transition itself; (iii) the use of light-scattering and electron microscopy to assess the conigurational change during Ca2+ release showed that the transition to the orthodox state took place at the same time as Ca2+ release. Additional observations showed that the kinetics of Ca2+ release was similar to that of the transition: measurement of 45Ca2+ ﬂuxes indicated that mitochondria release their entire pool of Ca2+ at the same time, following a lag phase without any release. The authors also investigated the release of Ca2+ induced by the addition of an uncoupler molecule that dissipated the transmembrane electric potential (ΔΨ) (Hunter & Haworth, 1979b), observing that as Ca2+ accumulated in mitochondria, so did the amount of Ca2+ released by addition of the uncoupler triﬂuoromethoxy carbonylcyanide phenylhydrazone (FCCP). Addition of FCCP also provoked a conigurational change from the aggregated to the orthodox form,i.e. a permeability transition was observed. The percentage of mitochondria that underwent the transition was higher in populations with higher levels of Ca2+ accumulation. Again the pattern of uncoupler-induced Ca2+-dependent Ca2+ release was sudden and total and followed a lag phase of no release. In conditions where the NADH protective mechanism was lost, mitochondria were observed to transition to an orthodox conformation without the requirement for additional Ca2+, and to release Ca2+ much faster than mitochondria that still possessed this protective mechanism, thus implying that Ca2+release depended on the permeability transition.

In addition, Sr2+ release was dependent on Ca2+ accumulation and was ruthenium red insensitive, both characteristics of the transition. Supported by this evidence, the uncouplerinduced release of Ca2+ was concluded to be another consequence of the Ca2+-induced transition, but dependent on the level of Ca2+ accumulated.

Hunter & Haworth (1979b) also explored whether the release of Ca2+ caused by Na+ was mediated by the Ca2+induced transition. Because no mitochondrial conigurational changes upon addition of Na+ and consequent Ca2+ release were observed, it was concluded that not all release of Ca2+ from mitochondria depends on the mPT. Na+induced Ca2+ release was later shown to originate from a protein membrane exchanger (Wacquier et al; 2017).

III. FROM THE INITIAL OBSERVATIONS TO THE COMPOSITION AND REGULATION OF THE PERMEABILITY TRANSITION: AN EVOLVING MODEL

Over the past three decades, our understanding of the structure of the mPTP has evolved; while we now know some of the modulators involved, precise details of the pore-forming parts, including their dynamics, remain elusive. Advances in technology have enabled a better understanding of the mPTP pore complex composition and assembly, and its regulation in intact cells and tissues. Early work (Raaﬂaub, 1953; Lehninger & Remmert, 1959; Lehninger, 1962) used gravimetric and optical methods to show that isolated liver mitochondria can take up water and thereby increase in volume. Later studies were based on measuring either light transmittance (transparency) or light scattering (opacity, turbidity) of mitochondrial suspensions. Interestingly, Lehninger (1959) also observed the reverse process, which he termed ‘contraction ’, consisting of the extrusion of water from mitochondria, resulting Coronaviruses infection in a decrease in mitochondrial volume manifested by increased turbidity (light scattering) of mitochondrial suspensions. The standard method used for isolated mitochondria is still the mitochondrial swelling assay, which records a decrease in absorbance at 520/540 nm that accompanies mitochondrial swelling, with Ca2+ administered to preparations at pH 7.4.
Thismitochondrialswelling, subsequently studiedbyAzzi& Azzone (1965a), was termed ‘large-amplitude swelling’ as it eventually led to complete rupture of the OMM and the formation of IMM ‘ghosts’ of low light absorbance (Fig. 2). By contrast, ‘low-amplitude swelling’, irst observed by Chance & Packer (1958) and studied by Azzi & Azzone (1965b),was fully reversible and exhibited variation in the energy and metabolic state of mitochondria. These low-amplitude changes in mitochondrial volume and structure could also be observed inside living cells (Hackenbrock et al; 1971). The term ‘permeability transition’ of the IMM was probably irst used by Wingrove & Gunter (1986a). Large-amplitude and (most likely) lowamplitude mitochondrial swelling could be interpreted as resulting from an unspeciic increase in permeability of the IMM to low molecular weight solutes. This increased permeability allowed the concentrations of low molecular weight compounds to equilibrate inside and outside the mitochondria, whereas concentrations of high molecular weight compounds, in particular, soluble intramitochondrial proteins, remained unchanged. This resulted in a higher osmotic pressure (so-called colloidal or oncotic pressure) inside mitochondria, resulting in the inﬂux of water.

In contrast to mitochondrial swelling, mitochondrial contraction was assumed to be an active process (Lehninger, 1959). While the releaseofnon-esteriied fatty acids accompanied mitochondrial swelling, their contraction involved the reincorporation of these fatty acids into mitochondrial phospholipids as demonstrated using isotopically labelled [14C]oleate and glycerol 3-[32P]phosphate (Wojtczak, Wlodawer & Zborowski, 1963). These studies suggested that contraction was enabled by the removal of accumulated free fatty acids and the restoration of some lipid compounds indispensable for thenormallow permeability of the inner membrane.

A systematic study using electron microscopy revealed that the swelling-related increase in mitochondrial matrix volume is accompanied by the rupture of the OMM (Wlodawer et al; 1966) (Fig. 2). By contrast ATP-induced contraction was reﬂected by matrix condensation, although this never led to the restoration of the original structure of the intact mitochondrion. Impermeability of the OMM to cytochrome c (Wojtczak & Zaluska, 1969; Wojtczak & Sottocasa, 1972) led to the development of an assay for its intactness in preparations of isolated mitochondria based on oxidation of externally added reduced cytochrome c (Wojtczak et al; 1972). The same protocol can be used to detect the release of cytochrome cfrom mitochondria that might occur during mitochondrial swelling. In living cells, release of cytochrome c to the cytosol accompanied by the rupture of the E7766 outer membrane (Vander Heiden et al; 1997; Petit et al; 1998) can be investigated with the use of antibodies against cytochrome c; however, this requires isolation of cytosolic fractions. It is also important to note that there are other possible ways in which cytochrome c can be released from mitochondria into the cytosol without swelling and rupture of the OMM (Wieckowski et al; 2001).

Experiments on isolated mitochondria were carried out in the late 1970s (Haworth & Hunter, 1979; Hunter & Haworth, 1979a,b) (see Section II.3) in research into the molecular mechanisms underlying the mPTP, its regulation, and the biological function of this transition that employed the ‘controlled ’ environment offered by isolated and deenergized organelles. Despite being artiicially devoid of endogenous substrates, de-energized mitochondria preserve a permeability transition, allowing its analysis in a relatively variable-free environment. Methods applied to isolated mitochondria partly replaced previous methodologies using enzymatic reactions performed, for instance, on ashed preparations from rat heart that were often inaccurate and technically dificult (Slater & Cleland, 1953). Although the earlier methods allowed Ca2+-dependent mitochondrial swelling to be observed (Slater & Cleland, 1953), studies on isolated mitochondria enabled the demonstration of the sudden opening of a reversible permeability state of the IMM, termed the ‘Ca2+-induced transition ’ (Hunter et al; 1976), and enhancement of that effect by increased concentrations of phosphates, arsenate and fatty acids (Hunter & Haworth, 1979a). The ability to add different substrates to mitochondrial preparations to identify their regulatory properties allowed exploration of the protective mechanisms against mPTP opening, such as Mg2+ which competes with Ca2+ for a shared binding site inside mitochondria. Thus, Ca2+ was established as the central positive modulator of the mPT (Azzi & Azzone, 1966; Hunter et al; 1976). Other protective mechanisms included the reduction status of mitochondrial NAD+, bongkrekic acid, and the energetic status of mitochondria (Hunter et al; 1976).
This evidence was conirmed more recently when it was demonstrated that Ca2+-dependent swelling was eficiently inhibited by bongkrekic acid administration, an ADP-based medium and by the immunosuppressant Cyclosporin A (Halestrap & Davidson, 1990). This work highlighted the usefulness of mitochondrial swelling to study this unselective pore’s opening and highlighted a regulatory role for the adenine nucleotide translocator (ANT) and cyclophilin D (CypD),the molecular target of Cyclosporin A.

Critical to modern studies was the description of two ways by which anon-speciic pore opens: one caused by low physiological Ca2+ concentrations involving a peptidylprolyl isomerase (PPi)-dependent mechanism that is insensitive to Cyclosporin A, and a second in response to higher Ca2+ concentrations and that is Cyclosporin A-sensitive (Davidson & Halestrap, 1987). These observations led to the proposal of a model of pore opening, based on the CypD—ANT protein interaction in the presence of Ca2+ overload (Halestrap & Davidson, 1990). In this model, by binding CypD, Cyclosporin A would promote its dissociation from the translocator and block the pore. The ability of ADP, ATP, and bongkrekic acid strongly to inhibit mPTP opening suggested that they also may bind the ANT carrier, reversing its conformation caused by Ca2+ addition. A combination of isolated mitochondrial preparations and patch-clamp techniques allowed monitoring of mPTP properties in response to ANT oligomers in artiicial membranes (Brustovetsky & Klingenberg, 1996).

It was also suggested that the mPTP might result from structural and functional ‘cooperation ’ between the IMM and OMM (Kottke et al; 1988; McEnery et al; 1992; Kinnally et al; 1993). This view identiied the contact sites between the two membranes as possible loci regulating not only the permeability but also the metabolic and energetic functions of mitochondria (Nicolay et al; 1990; Bucheler, Adams & Brdiczka, 1991; Wieckowski, Brdiczka & Wojtczak, 2000). Arguments that the mPTP is located at the junction between the two membranes were based on observations that proteins supposedly participating in this pore came from both the OMM and IMM.

Early studies on isolated mitochondria identiied putative proteinaceous channels between the IMM and OMM with ANT and VDAC as the core components of the mPTP (Fig. 3A), and a plethora of regulators, including the OMM 18-kDa peripheral benzodiazepine receptor (TSPO), glycogen synthase kinase 3β (GSK3β), hexokinase II (HKII) and creatine kinase (CK) (Ong et al; 2014; Morciano et al; 2015; Tanaka et al; 2018). Although the VDAC enables the transport of most solutes across the IMM, the newera of genetic studies challenged this proposed pore structure (Baines et al; 2007), demonstrating that mitochondria isolated from VDAC-knockout (KO) mice still exhibited a Ca2+-dependent mPT very similar to that found in wild-type mitochondria. Similarly, cell death was unaltered in VDACKO cells. Gene inactivation studies using KO animal models conirmed the presence of a functional mPTP also in ANTKO mice, thus excluding it as a pore-forming component of the mPTP (Kokoszka et al; 2004) and conferring to CypD a more important modulatory role (Schinzel et al; 2005; Hurst et al; 2020). The recent innovative generation of atriple KO for ANT1, ANT2, and ANT4 has revisited the contribution of this translocator to the mPTP cascade; the progressive deletion of these isoforms led to equal decreases in sensitivity of the pore to opening (Karch et al; 2019), implying that ANT, under favourable biochemical conditions and in a given tissue, may constitute an alternative to ATP synthase during mPTP formation.

CypD then remained the only protein whose involvement in the mPTP was uncontested: experiments performed with transgenic mice lacking the peptidylprolyl isomerasef (Ppif) gene conirmed that this protein is a key element of the mPTP that is responsible for its sensitivity to Cyclosporin A (Basso et al; 2005; Valasani et al; 2014, 2016; Lindblom et al; 2020; Torpey et al; 2020; Panel et al; 2021). However, it should be stressed that CypD plays a regulatory role and is not involved in the pore itself (Fayaz, Raj & Krishnamurthy, 2015). The elucidation of the regulatory role of CypD in mPTP opening was facilitated by discovering that Cycloporin A (the gold-standard inhibitor of the mPTP) targeted this protein. To date, multiple investigations have shown that CypD inactivation (using genetic or pharmacological approaches) inhibits mPTP induction and cell death in several in vitro and in vivo models. Signalling events also can target CypD to regulate mPTP opening.

McEnery etal. (1992) investigated the interaction of TSPO with the ANT and VDAC proteins. At that time, studies on IMM— OMM contact sites were the focus of interest, and ligands of TSPO with nanomolar afinity and joined to VDAC/ANT proteins were shown to have mPTP-like channel activities, as recorded using the patch-clamp technique (Kinnally et al; 1993). However, biochemical attempts to identify the mPTP core complex by conditional deletion assays led researchers to conclude that TSPO was dispensable in terms of mPTP regulation and hence does not participate in mPTP-dependent cell death (Sileikyte et al; 2014).

It was proposed independently by two research teams (Alcala et al; 2008; Leung, Varanyuwatana & Halestrap, 2008) that the phosphate carrier (PiC) could be a good candidate to form the core of the mPTP (Fig. 3B). The observed concentration-dependent inhibitory effect of n-ethylmaleimide (NEM), ubiquinone 0 (UQ0), and (spiro [cyclopentane-1,50 -[5H]dibenzo[a,d]cyclohep-ten]-2-one,100 ,110 -dihydro-3-methylene (Ro 68-3400) on both the mPTP and PiC suggested that the PiC could act as a pore-forming component. This was reinforced by the fact that phosphates greatly enhanced mPTP opening, so it was believed that an additional protein of the IMM (such as the PiC) could represent the putative pore-forming part. Varanyuwatana & Halestrap (2012) subsequently questioned this concept by showing that a 70% or more decrease in expression of the PiC in HeLa cells did not affect mPTP opening. This was conirmed by other authors using KO models (Gutierrez-Aguilar et al; 2014). Interestingly, the involvement of the PiC in the PTP could not be fully excluded because complete genetic deletion of this carrier in mouse cardiac mitochondria desensitized the mPTP (Kwong et al; 2014). The PiC was thus instead suggested to be a regulatory rather than a core component of the mPTP (Kwong et al; 2014).

In another perspective, Karch et al. (2013) provided evidence for B-cell lymphoma-2 (Bcl-2)-associated X (Bax) and Bcl-2-antagonist/killer 1 (Bak) members of the Bcl-2 family as mPTP components; using transgenic animal models, they demonstrated that the absence of both proteins promoted resistance to mitochondrial swelling and resulted in the lack of channel activity, as evidenced by patch-clamp studies on mitoplasts (Karch et al; 2013). Thus, even though Bax and Bak play a structural role in the OMM part of the mPTP complex, an increase in solutes in the mitochondrial matrix also requires permeability of the IMM. This pore model suggesteda subdivision between IMMand OMM-pore formation, returning to previous concepts of IMM–OMM contact sites in mPTP assembly (Beutner et al; 1998). Despite solid data supporting the contact site model of the mPTP, another model of mPTP formation has been proposed (Kowaltowski, Castilho & Vercesi, 2001; He & Lemasters, 2002). This model suggests that the pore could be formed by misfolded mitochondrial proteins modiied, among other agents, by oxidative damage and unconnected with the presence of a pre-existing inner membrane pore. It has been proposed that opening of such unregulated pores occurs when the number of amphipathic protein clusters exceeds the number of chaperones available to block their conductance (Fig. 3C).

Although Ca2+ overload is widely recognized as the primary inducer of the mPT, other positive regulators have been reported. Halestrap (1991) demonstrated that pH can act as an additional and precise regulator of the mPT in rat heart and liver mitochondria. mPTP opening is thought to be inhibited in an acidic matrix pH because of the displacement of Ca2+ at the trigger site by H+ ions (Halestrap, 1991). Atractyloside, a natural toxic glycoside, increases mPTP opening by modifying the conformational state of ANT, locking it in a cytoplasmic-side open conformation. ROS production and low mitochondrial membrane potential (MMP) are two other known inducers. Mitochondria are an important source of ROS (Giorgi et al; 2018c), and conditions that lead to increased oxidative stress in cells, such as reoxygenation following hypoxia, strongly sensitize mPTP opening and consequent cell demise (Assaly et al; 2012). Thus, it is a classical notion that the mPT is related to the redox state of mitochondria, including that of coenzyme Q(Kowaltowski, Castilho & Vercesi, 1995), and NADPH (Bernardes et al; 1994), with mitochondrial oxidative stress playing an important role even in de-energized mitochondria (Kowaltowski, Castilho & Vercesi, 1996; Vercesi et al; 2018).

A unique roleisalso playedby thiol groups ofproteins localized in the IMM. Connern & Halestrap (1994) showed that exogenous administration of thiol oxidizing reagents to puriied mitochondria preparations promoted key modiications in Cys56, Cys159 and Cys256 residues of the ADP/ATP translocator. In particular, Cys56 oxidation potentiated the binding of mitochondrial CypD with the IMM leading to pore opening (Connern & Halestrap, 1994). Cys56 and Cys159 are thought to be involved in antagonizing the inhibitory properties of ADP artificial bio synapses via the nucleotide-binding site on the ANT protein, as shown by adding phenylarsine oxide (PhAsO) to deenergized mitochondria, with its effect independent of CypD binding (Halestrap, Woodield & Connern, 1997). All these effects were reversed in the presence of antioxidants (Kowaltowski et al; 2001; Vercesi et al; 2018). Further, evidence for mPTP inhibition in the presence of rotenone and the sensitivity of the mPT to inorganic phosphate (Pi) prompted the suggestion that NADH:ubiquinone oxidoreductase (Complex I) could also be implicated in the negative modulation of the mPTP, establishing a novel level of regulation. This was proposed to occur via conformational changes of Complex I affecting its interaction with the mPTP, in a manner that depends on the availability of Pi and CypD expression (Li et al; 2012).

The open conformations of the mPTP may assume a low or a high conductance. The ability to switch from a low to a high conductance was investigated by Ichas & Mazat (1998) who also reported that this transition became irreversible upon reaching the high-conductance state. Historically, physiological roles have been attributed to the lowconductance open conformation; indeed, mitochondrial functions are preserved while in the so-called ‘ﬂickering mode ’ that regulates cellular Ca2+ homeostasis (Gunter & Pfeiffer, 1990; Altschuld et al; 1992), with very limited diffusion of solutes through the IMM (cutoff <300 Da) and precise regulation by matrix pH changes and mitochondrial Ca2+ uptake (Ichas, Jouaville & Mazat, 1997). Conversely, when the mPTP switches to the high-conductance state, the consequences that are not compatible with cell life, including increased permeability of the IMM to solutes of 1500 Da, rupture of the OMM, and activation of the apoptotic cascade.

Our understanding of the processes leading to the association of ATP synthase (Complex V) with the mPTP (Fig. 3D, E) began with the discovery of regulatory analogies between these two complexes. Two mPTP inhibitors (ADP and Mg2+) interact to block the hydrolytic activity of Complex V (Feniouk, Suzuki & Yoshida, 2006), while phosphates, known to be positive regulators of mPTP opening, reversed this state. In addition, the mPTP subunits ANT and PiC form a so-called ATP synthasome by interacting with ATP synthase (Ko et al; 2003). Similarly, mitochondrial CypD, the target of the mPTP inhibitor Cyclosporin A, has been demonstrated to interact with the peripheral stalk of ATP synthase, the oligomycin-sensitivity-conferring protein (OSCP) (Giorgio et al; 2009). This protein is able to modulate ATP synthase activity, causing it to decrease when mitochondrial CypD is anchored and to regulate mPTP opening (and cell death) via its C141 and H112 residues. These residues respectively affect the sensitivity of the pore to oxidation and its dependence on H+ binding (Antoniel et al; 2018; Carraro et al; 2020).

In recent years, much effort has been made to identify putative component(s) of the IMM with channel properties. The c-subunit of Fo-ATP synthase, an IMM-resident protein, was identiied as a key mPTP component with voltage-sensitive channel properties in its monomeric form and in the presence of both CypD binding and high Ca2+ concentrations (Alavian et al; 2014). Studies from three independent research groups conirm the knowledge that this csubunit plays a pivotal role in mPTP opening and link physio-pathological effects to its intracellular expression (Bonora et al; 2013) and phosphorylation status (Azarashvili et al; 2014). How the mPTP complex takes shape and which other proteins contribute to the structure of the pore-forming region remains elusive. Biochemical studies (Pavlov et al; 2005; Elustondo et al; 2016) reported the presence of higher numbers of c-subunits in mitochondria following induction of the mPT. This suggested the possibility that the c-subunit modiies interaction within F1/FO-ATP synthase dimers. They described an essential role of complexes formed by the c-subunit–inorganic polyphosphate (polyP)-hydroxybutyrates (PHB) axis in the formation of a non-speciic pore channel during Ca2+-induced stress. These results suggest that the c-subunit may form water-illed pores with polyP possibly serving as a hydrophilic coating of the pore, despite the F 1/FO-ATP being a hydrophobic protein. Further conirmation of the importance of the c-subunit in mPTP activity has been published recently (Neginskaya et al; 2019).

Two separate hypotheses justify a contribution of ATP synthase to mPTP assembly and activity (Fig. 3D, E): irst, dimerization of ATP synthase is essential for the generation of the pore opening (Giorgio et al; 2013) (Fig. 3E); second, ATP synthase monomers with a proper c-ring (the arrangement of c-subunits in the IMM) conformation constitutes the pore-forming part (Fig. 3D) (Bonora et al; 2017). However, currently both models present certain critical issues that remain to be addressed (Bauer & Murphy, 2020). Concerning the irst model (Giorgio et al; 2013), there is irrefutable proof of the involvement of ATP synthase dimers (and oligomers) in beneicial bioenergetic functions of mitochondria, partially due to their role in ensuring the correct curvature of the IMM (Daum et al; 2013), thus dimers of ATP synthase are unlikely also to be a mediator of Ca2+-dependent cell death. For the second hypothesis, a structural model by which a non-speciic current occurs in the c-ring is still needed. The controversial nature of the mPTP was highlighted by two recent publications in which KO experiments showed that the c-subunit and the peripheralstalk subunit of ATP synthase are not required for mPT opening (He et al; 2017a,b). However, electrophysiological analysis of isolated mitoplast KOs for the c-subunit demonstrated that, in their absence, the current recorded differed from that expected for mPTP. Indeed, the c-subunit-KO cells did exhibit a Ca2+-inducible current, which could be inhibited by Cyclosporin A,but this was much lower than in wild-type cells (0.3 nS versus 1.3 nS). Moreover, the data suggest that, in c-subunit-KO cells,a second current could be generated by ANT (Neginskaya et al; 2019). These indings thus conirm the key role of the c-subunit in mPTP activity. In another study, fully reconstituted active ATP synthase in liposomes was responsive to Ca2+, converting dimers, but not monomers, into achannel. Interestingly, the activity was sensitive to adenine nucleotides, but not to ligands of ANT or VDAC (Urbani et al; 2019). Recent decades of research have led to the discovery of a large number of inhibitors (Morciano et al; 2018) and inducers of the mPT (see Fig. 4).

IV. UNDERSTANDING THE mPTP WITH THE USE OF ELECTROPHYSIOLOGICAL STUDIES

Important information about the mPTP structure and functioning has been obtained using the patch-clamp technique. To record channel activity of the IMM using this technique, it is necessary to remove the OMM and to create singlemembrane mitoplasts. Two methods can be used to achieve this: the French press (Decker & Greenawalt, 1977) or osmotic swelling (Gupte et al; 1984).

The irst patch-clamp recordings of mitochondrial channel activity from the IMM (Sorgato, Keller & Stuhmer, 1987) led to the description of a single, slightly anion-selective channel of 107 pS conductance in 150 mM KCl. A 350 pS channel attributed to the OMM was also observed (Sorgato et al; 1987). Subsequently, a study on mouse liver mitochondria (Kinnally, Campo & Tedeschi, 1989) reported a variety of conductances of 10– 20, 45, 80, 120– 150, 200, 350, and 1000 pS,with the latter uncharacterized due to its rarity. A parallel study on rat liver mitochondria (Petronilli, Szabo & Zoratti, 1989) observed a similarly large-conductance activity of about 1.3 nS. This activity exhibited a ﬂickering nature, and its main subconductance state was around 0.63 nS at +20 mV. However, a range of other subconductance levels was observed at different voltages, for instance, at +30 mV subconductances were 450, 350, 860 pS while at −40 mV they were 650, 450, 1000 pS. This study was carried out in 150 mM KCl with 0.1 mM CaCl2, while Kinnally et al. (1989) employed no addition of CaCl2. A follow-up study (Szabo & Zoratti, 1991) found that this large-conductance activity characterized by multiple subconductance states, for which they coined the term ‘mitochondrial megachannel ’ (MMC), was inhibited by 100–200 nM Cyclosporin A, a known inhibitor of the mPTP (Broekemeier, Dempsey & Pfeiffer, 1989). The kinetic features of the single-channel events supported the idea that the MMC is composed of cooperating subunits. Further properties of the MMC were characterized in a subsequent study, showing that the MMC was non-selective, activated by Ca2+, and inhibited by Mg2+, Cyclosporin A, and ADP, probably acting at matrix-facing sites (Szabo & Zoratti, 1992).

It then was shown that Mg2+, Mn2+, Ba2+, and Sr2+ and Cyclosporin A act as competitive inhibitors for Ca2+ (Szabo, Bernardi & Zoratti, 1992), and that the MMC is regulated by pH in the physiological range. Lower pH values caused MMC closure in a Ca2+-reversible manner. For example, MMC activity that was blocked by pH 6.5 in the presence of 0.5 mM Ca2+ could be reinstated by 1.2 mM Ca2+. Patch-clamp experiments were carried out using a so-called ‘inside-out ’ coniguration in which the matrix side of the membrane patch is exposed to the medium. Together, these patchclamp results led to the conclusion that the modulating sites involved in these effects are located on the matrix side of the IMM. Bernardi et al. (1992) provided evidence that the Ca2+-induced mPT is affected by the above agents, supporting the identiication of the MMC as responsible for mPT. This pharmacological characterization of the MMC/PTP was set aside in subsequent studies in which the voltage dependence of the megachannel was re-investigated (Szabo & Zoratti, 1993). It was noted that the closed state(s) of the MMC was favoured at negative (physiological) transmembrane potentials. MMC conductance was 4.35 nS in symmetrical 0.5 M KCl with gating events involving a ﬂickering half-size conductance (2.2 nS), which corresponded to that of the fully open VDAC in these conditions (Szabo, de Pinto & Zoratti, 1993). This led to the proposal that the MMC consisted of two cooperating porin (VDAC) molecules. Beutner et al. (1996) then characterized high molecular weight complexes isolated by Triton X-100 extraction from rat brain homogenate. These complexes were tested with speciic antibodies and contained hexokinase, creatine kinase, VDAC, and ANT. After incorporating artiicialbilayers, channel activity of 6 nS in 1 M KCl with an asymmetrical voltage dependence was recorded.

A claim that ANT was responsible for the mPTP was made a couple of years earlier (Tikhonova et al; 1994). In their studies, puriied ANT was incorporated into liposomes and fused into black lipid membranes (BLM) in the presence of 800 mM urea. Channel activity was induced by mersalyl. They observed a range of conductances from 200 pS to 2.2 nS in 180 Na2SO4, 20 mM 2-(N-morpholino)ethanesulfonic acid (MES)-NaOH, pH 7.0, in the presence of 3 mM Mg2+. Involvement of ANT in MMC activity was also proposed in other studies (Brustovetsky & Klingenberg, 1996; Brustovetsky et al; 2002). Brustovetsky & Klingenberg (1996) used bovine heart ANT, puriied and reconstituted into giant membrane vesicles. Large conductance channels of 300– 600 pS [in buffer containing 100 mM KCl, 2 mM MgCl2, 1 mM CaCl2, 4 mM potassium gluconate, 5 mM MES, and 5 mM trisaminomethane (Tris) at pH 7.2] were observed. These channels exhibited low cation selectivity (PK+/PCl− = 4.3 ± 0.6), were activated by Ca2+ (1 mM), inhibited by protons (pH 5.2), and by a combination of bongkrekate and ADP. Channel closing was induced at extreme voltages. In a follow-up study, recombinant ADP/ATP carrier (rAAC) from Neurospora crassa was expressed in Escherichia coli (Brustovetsky et al; 2002). Puriied rAAC was reconstituted and its activity recorded by patchclamp. Its behaviour was similar to that observed for ANT from bovine heart. In addition, it was shown that cyclophilin isolated from Neurospora crassa suppressed channel gating, thus increasing channel open probability, while Cyclosporin A abolished the cyclophilin effect. When ADP was applied to cyclophilin-activated channels it induced ﬂickering of the channel, effectively decreasing channel open probability. By contrast, channel gating was diminished by the prooxidant tert-butyl hydroperoxide (Brustovetsky et al; 2002).

Although attempts to characterize the mPTP by means of various assays, such as mitochondrial swelling or mitochondrial Ca2+ accumulation were carried out at this time, only a few electrophysiological studies on the MMC were published. In one such study it was observed in patch-clamp experiments with rat liver mitochondria that ubiquinone 0 and decylubiquinone inhibited the activity of the MMC, in line with earlier observations for the mPTP (Fontaine, Ichas & Bernardi, 1998). Inhibition by these compounds was reversed by increasing [Ca2+], a similar behaviour to that observed for several other MMC inhibitors. Classical MMC activity was observed in human hepatoma HepG2 cells. This channel had a high conductance of 1.23 nS (in 150 KCl), and frequently occupied a 640 pS subconductance level; it was active at high (1 mM) and closed at low (1 μM) Ca2+ and was inhibited by 10 μM Cyclosporin A (Loupatatzis et al; 2002). Biochemical studies suggested that Bax cooperates with ANT in apoptotic events and thus was likely to be a component of the mPTP (Marzo et al; 1998). However, other work (Campello et al; 2005) showed that the activity of the MMC does not require Bax. In this study, using the human HCT116 cancer cell line, the MMC was found in 10–20% of patches from both Bax+ and Bax− cells, indicating that the mPTP was independent of the presence of Bax. MMC activity was recorded in 150 mM KCl, 0.5 mM CaCl2, 1 mM Pi, 20 mM 4-(2-hydroxyethyl)1-piperazineethanesulfonic acid (Hepes), pH 7.35, and under these conditions had a conductance of about 1 nS (range 0.9– 1.3 nS) with a hallmark ﬂickering substate of about half this size. Weakly anionic or no selectivity (PCl−/PK+ = 1.8) for the high conductance state, and slight cationic selectivity (PK+/PCl− = 2.8) for the low conductance state was observed.

It was known that Cyclosporin A binds CypD, as described above. However, it was not clear whether CypD was a component of the pore itself. To solve this question, MMC channel properties from liver mitochondria from wild-type and CypD-deleted mice were compared in detail (De Marchi et al; 2006). The pores observed in the two cases were indistinguishable, with the only clear difference being their sensitivity to Cyclosporin A. It was therefore concluded that CypD is a modulatory component of the PTP but is not part of the MMC pore (De Marchi et al; 2006).

The work of Giorgio et al. (2013) proved that electrical mPTP activity was related to ATP synthase dimers. First, they identiied the activity of monomers and dimers of ATP synthase after separation of mitochondrial proteins using blue native electrophoresis. Second, these gel-puriied ATP synthase monomers or dimers, devoid of ANT,VDAC, and CypD, were incorporated into planar lipid bilayers made of puriied soybean azolectin and channel activity was recorded. When the experimental medium contained 50 mM KCl, 1 mM Pi, and 0.3 mM Ca2+ no channel activity was observed when either monomer or dimer of ATP synthase were added. However, the addition of Bz-423, a proapoptotic agent (Boitano et al; 2003) that was previously shown to target ATP synthase (Johnson et al; 2005), elicited channel activity only when dimers but not monomers of ATP synthase were added to the bilayer. Bz-423 elicited similar activity in the presence of PhAsO, a sensitizer of the mPTP to Ca2+ (Krauskopfet al; 2006). This channel activity could be inhibited by γ-imino ATP (AMP-PNP), a nonhydrolyzable ATP analog, and by ADP in the presence of Mg2+ ions; ADP also exhibited a partial inhibitory effect when present alone. Channel activity was not inhibited by Cyclosporin A, in agreement with the absence of CypD in the preparation and the fact that ATP synthase dimer was extracted in the presence of 10 mM Pi, which sensitizes the mPTP even in the absence of CypD. Channel opening was still observed in the presence of bongkrekic acid and could not be elicited by atractyloside, a selective inhibitor of ANT. These results have been further conirmed by recording the MMC activity using a highly puriied preparation of bovine ATP synthase dimers (Urbani et al; 2019).

Further evidence for the role of ATP synthase dimers in the formation of the pore came from the planar bilayer recordings of blue native polyacrylamide gel electrophoresis (BN-PAGE) puriied yeast F1/FO-ATP synthase (Carraro et al; 2014). The ATP synthase dimers did not elicit currents unless Ca2+, PhAsO, and Cu(OP)2 were added. Moreover, the channel activity was inhibited by Mg2++ ADP as found for the mammalian ATP synthase dimer. The unitary conductance of the channels formed by ATP synthase dimers anilide; JM-20, 3-ethoxycarbonyl-2-methyl-4-(2-nitrophenyl)-4,11-dihydro-1H-pyrido[2,3-b][1,5]benzodiazepine; MitoQ, 10-(60 ubiquinonyl)decyltriphenylphosphonium bromide; O2.–, superoxide anion; Pi, inorganic phosphate; PK11195, 1-(2-chlorophenyl)N-methyl-N-(1-methylpropyl)-3-isoquinolinecarboxamide; SfA, Sanglifehrin A; TMD#7538, N-phenethyl-6-phenyl-2, 3, 4, 9-tetrahydro-1H-carbazol-1-amine; TRO40303, 3,5-seco-4-nor-cholestan-5-one oxime-3-ol; ΔΨ, mitochondrial transmembrane potential. ranged from 250 to 300 pS, and thus was lower than the conductance observed for the mammalian MMC. It should be noted that the dimer preparation did not contain translocase of outer mitochondrial membrane 20 (Tom20) or translocase of inner membrane 54 (Tim54) and therefore, that channel activity could not be due to the twin-pore translocase (Rehling et al; 2003). The same approach showed that dimers of Drosophila Fo-ATPase form channels opened by Ca2+, Bz-423, PhAsO, and Cu(OP)2 with a single-channel conductance of only 53 pS (in 100 mM KCl) (von Stockum et al; 2015).

This sequence of papers thus implicated ATP synthase dimers as the molecular entity responsible for mPTP activity. A subsequent study (Alavian et al; 2014) showed that the puriied human Fo subunit of ATP synthase alone, which is formed by an octameric ring of c-subunits (see Fig. 3D, E), when reconstituted into liposomes exhibited channel activity. This channel had a high conductance of around 100–300 pS, 500– 750 pS, 1500 pS, and 1800–2000 pS and resembled the activity of the MMC. It was unselective with a permeability ratio Na+/K+ = 1.5. AMP, ADP, and ATP inhibited this c-subunit activity, and for ATP the half maximal effective concentration (EC50) was found to be 660 μM. The same study also determined the EC50 for ATP on MMC activity recorded in submitochondrial vesicles (SMVs), which was found to be lower by at least an order of magnitude (50 μM). An anti-pan-c-subunit antibody that inhibits the activity of the c-subunit ring also inhibited MMC activity induced by Ca2+ in SMVs. In contrast to puriied ATP synthase monomers, the activity of the c-subunit was unaffected by Ca2+. To demonstrate further that the c-subunit ring could create a pore, they substituted up to four highly conserved glycines within the irst alpha-helical region of the c-subunit with valines, under the assumption that this would interfere with the tight packing of the c-subunit molecules within the ring structure. When reconstituted into liposomes, these mutants all demonstrated increased singlechannel conductance compared to that of the wild-type, with the conductance of the quadruple mutant being the largest (Alavian et al; 2014). This channel was also shown to be insensitive to blocking by ATP. They then hypothesized that F 1 binding to the c-subunit ring would inhibit the channel’s activity, and applied puriied individual F1 proteins to reconstituted active c-subunit channels.

Curiously, only the β subunit but not γ, δ, or ε subunits of ATP synthase had an inhibitory effect. Alavian et al. (2014) tested further the channel activity and regulation in the following preparations:
(i) puriied recombinant c-subunit lacking CypD and OSCP reconstituted intoproteoliposomes (neither Ca2+ nor Cyclosporin A had an effect on channel activity);
(ii) puriied ATP synthase monomers containing OSCP but lacking CypD reconstituted intoproteoliposomes (infrequent channel activity was observed, strongly enhanced by the addition of recombinant CypD protein either in the presence or absence of Ca2+; this activity was inhibited by Cyclosporin A);
(iii) mitochondria and SMV containing endogenous CypD and OSCP, and SMV exposed to urea to denature and remove extramembrane proteins, including F 1 components such as the OSCP, β subunit, and CypD (the activity of mitochondria and SMVs was regulated by Ca2+ and Cyclosporin A, but was completely absent from the urea-exposed SMVs; 1 mM ATP was still able to inhibit the activity of the channel).
Mnatsakanyan et al. (2019) conirmed these initial conclusions by recording megachannel activity of puriied porcine monomeric ATP synthase. Altogether, these patchclamp experiments in which channel activity was recorded from reconstituted highly puriied protein complexes indicated in a compelling way that ATP synthase, either as a dimer or as a monomer, is responsible for MMC activity.

A considerable controversy then arose as a result of studies showing that HAP1-A12 cells incapable of producing the ATP synthase c-subunit still show mPTP activity as measured by Ca2+ retention capacity of mitochondria (He etal; 2017b). Recently, mitochondria derived from this cell line were further investigated by the patch-clamp technique (Neginskaya et al; 2019). In contrast to the mitochondria of the wild-type HAP1 cells in which classic MMC activity of 1.3 ± 0.2 nS with a subconductance state of 0.4 ± 0.04 nS was detected, the mitochondria of HAP1-A12 cells contained a channel of a much smaller conductance of 0.3 ± 0.07 nS with a subconductance state of 0.13 ± 0.03 nS. Curiously, this channel was blocked by Cyclosporin A but also partially by ADP and bongkrekic acid. Similar features, including conductance and sensitivity to ADP and bongkrekic acid, were previously described for a puriiedANT converted to a channel by Ca2+ treatment (Brustovetsky & Klingenberg, 1996) indicating that in the absence of a mPTP its function could be substituted by ANT.

Doubts about the involvement of the c-subunit in mPTP activity were replicated for other subunits, excluding from this phenomenon contributions of both subunits of the peripheral stalk of ATP synthase and of the whole enzyme once assembled. To reach these unexpected conclusions, clones were generated from HAP1-A12 cells via the disruption of ATP5F1 and ATP5O genes encoding respectively for subunits b and OSCP (He et al; 2017a). In both cases, the properties of the mPTP appeared unaltered, still opening following stress stimulation and being inhibited by Cyclosporin A, refuting a role of OSCP as an interactor between the mPTP and CypD. Using the same approach and in the same cell line, subunits e, f, g, 6.8PL and DAPIT were removed (Carroll et al; 2019), leading to similar conclusions. Disruption of these proteins inevitably causes defects in ATP synthase assembly, thus arguing against the involvement of ATP synthase dimers in mPTP activity.

Nevertheless, a stronger case for the involvement of ATP synthase in MMC activity came from a study in which the activity of a single amino acid mutant of ATP synthase was studied by patch-clamp (Antoniel et al; 2018). It was known from earlier work that MMC activity was blocked by protons (Szabo etal; 1992). Antoniel et al. (2018) investigated whether a unique histidine in the OSCP subunit of the ATP synthase is important in this blockage by acidic pH. MMC activity from wild-type and from the OSCP H112Q mutant cells was recorded in a standard symmetrical solution of 150 mM KCl, 0.1 or 0.2 mM CaCl2, 10 mM Hepes, pH 7.3. Both channels exhibited a maximal conductance of 1 nS and several subconductance states with a prevalent substate of around 500 pS in agreement with previous observations for MMC activity. As observed previously for wild-type ATP synthase, acidiication of the bath to pH 6.5 resulted in almost complete MMC activity inhibition. However, in mitoplasts from OSCP H112Q cells, a decrease in pH to 6.5 did not cause considerable changes to open probability but the MMC was still sensitive to the classical mPTP inhibitor Ba2+ (Szabo et al; 1992).

Detailed information about the electrophysiological properties of the mPTP have recently been reviewed (Neginskaya, Pavlov & Sheu, 2021). This seminal paper discusses the evidence claiming that patch-clamp investigations can (i) discern among different mPTP pathways occurring in a given genetic or biochemical condition, and (ii) be used to understand the contribution of proteins or drugs in a highly controlled biophysical system.

V. THE mPTP – FROM MITOCHONDRIAL FRACTIONS TO INTACT CELLS

Since the 1990s, several methods have been developed to study the opening of the mPTP in intact cells, including cell imaging, the use of ﬂuorescent dyes, and pharmacological inhibition of the mPTP (Petronilli et al; 1998; Bonora et al; 2016) (see Fig. 5). Depolarization of the mitochondrial transmembrane potential is a recognized consequence of the mPT (Petronilli et al; 1993; Zamzami et al; 1996), and thus several studies have measured variations in the mitochondrial transmembrane potential as an indicator of the opening of the mPTP (Fig. 5A) (Huser & Blatter, 1999; Rama Rao, Jayakumar & Norenberg, 2003; Briston et al; 2017). However, since many other events besides the mPT can induce alterations in the mitochondrial transmembrane potential (including mitochondrial metabolism per se, lipid peroxidation, ion cycles, or activity of uncoupling proteins), it is unlikely that this is a useful method to investigate mPTP opening. Peroxidation of lipids within mitochondrial membranes can induce conigurational changes that will alter membrane properties, including a progressive increase in membrane permeability, and consequently, depolarization of the membrane potential (Stark, 1991; Wong-Ekkabut et al; 2007). When inducing the mPT with oxidant agents, one thus must eliminate artifacts caused by non-speciic alterations in mitochondrial membrane permeability. Furthermore, transient opening of the mPTP is very dificult to follow by measuring the MMP using ﬂuorescent dyes. The collapse of the MMP is not systematic and to detect measurable changes in the distribution of ﬂuorescent dyes, extended periods of PTP opening are usually required (Petronilli et al; 2001; Dumas et al; 2009). Also, the use of Cyclosporin A as an inhibitor of mPT-induced mitochondrial depolarization can be unreliable, as Cyclosporin A also inhibits calcineurin in cells, which can cause artifacts. FK506 (tacrolimus), which inhibits calcineurin but not the mPTP (Rodrigues-Diez et al; 2016), or mitochondrialdirected Cyclosporin A (Malouitre et al; 2009) can be used to account for the potential lack of speciicity of Cyclosporin A in intact cells. For this reason, other approaches to measure mPTP opening in situ should be employed to complement the information obtained by measurement of MMP.

The 2-deoxyglucose method was developed to follow mPTP opening in intact cells (Fig. 5B). While there was indirect evidence that pore opening occurred during thereperfusion of hearts, there was a need for a method that would provide direct evidence of mPTP opening at critical time points during the reperfusion phase. Grifiths & Halestrap (1995) developed an elegant methodology to follow the opening of the mPTP that used 2-deoxy[3H]glucose, which enters the cell through the glucose transporter and is phosphorylated to 2-deoxy[3H]glucose-6-phosphate. This metabolite cannot be further metabolized and thus is trapped in the cell. However, it can not cross the mitochondrial inner membrane unless the mPTP is in an open state. Once the mPTP opens, 2-deoxy[3H]glucose-6-phosphate enters the matrix of mitochondria. Subsequent treatment of isolated mitochondria with ethylene glycol-bis(β-aminoethyl ether)-N,N,N0 ,N0 -tetraacetic acid (EGTA) will chelate Ca2+, allowing the mPTP to close, trapping 2-deoxy[3H]glucose-6-phosphate in the matrix. The measurement of disintegrations per minute (d.p.m.) in isolated mitochondrial fractions allows calculation of the uptake of 2-deoxy[3H]glucose-6-phosphate by mitochondria and, therefore, demonstrates that the pore opened during that period. Since this method was developed, it has been used in several settings (Kerr, Suleiman & Halestrap, 1999; Rama Rao et al; 2003; Ayoub, Radhakrishnan & Gazmuri, 2017).

Nieminen et al. (1995) developed a different ﬂuorescent method using the ﬂuorescent dyes calcein-acetoxymethyl (AM) and tetramethylrhodamine methyl ester (TMRM) to monitor the mPT in intact cells (Fig. 5C). Calcein is a hydrophilic ﬂuorescein derivative, that when esteriied with anAM group, acquires the ability to cross the cellular membrane and become entrapped in a non-ﬂuorescent form. Once in the cytosol, intracellular esterases hydrolyse the AM group and the trapped calcein becomes ﬂuorescent. Cleavage of AM moieties is a widely used strategy to entrap ﬂuorescent probes inside cells. TMRM is a positively charged cellpermeant ﬂuorescent compound that once inside the cell is sequestered by mitochondria, depending on the MMP. Nieminen et al. (1995) demonstrated that when cells are loaded with calcein at 37。C, this ﬂuorescent dye accumulates predominantly in the cytosol, while mitochondria appear as dark spots when using confocal microscopy to image calcein ﬂuorescence. However, due to its positive charge, TMRM accumulates inactive mitochondria; hence the calcein-unlabelled dark spots now appear labelled with TMRM ﬂuorescence. Once the mPTP opens, mitochondrialose TMRM ﬂuorescence and the dark spots become illed with ﬂuorescent calcein. By using laser-scanning confocal microscopy, this method allowed the authors to monitor the mPTP in intact hepatocytes after exposure to t-butylhydroperoxide (Nieminen et al; 1995). Their method was, however, challenged by Petronilli et al. (1998) who pointed out caveats and advised caution in interpreting the results. Since the esteriied form of calcein (calcein-AM) can cross intracellular membranes, diffusion among the different organelles can also occur and it may label other cell spaces beside the cytosol (Petronilli et al; 1999). The esteriied form of calcein can also be cleaved by mitochondrial esterases and the ﬂuorescent calcein may become trapped inside the mitochondrial matrix. Regarding the dark spots observed by Nieminen et al. (1995) in confocal microscopy images of calcein ﬂuorescence, Petronilli et al. (1998) suggested that TMRM could quench calcein ﬂuorescence, or that the concentration of calcein inside the mitochondrial matrix could reach values high enough to cause calcein self-quenching. To overcome the ﬂaws of this method, Petronilli et al. (1998, 1999) suggested loading cells with calcein-AM in the presence of 1 mM of CoCl2 (Fig. 5D). In the presence of Co2+, calcein ﬂuorescence in the cytosol and nucleus is quenched, and because Co2+ does not cross the mitochondrial inner membrane, mitochondria appear as green-ﬂuorescent bodies against a dark background. Under a condition promoting mPTP opening, calcein can exit the mitochondria and Co2+ can ﬂow to the mitochondrial matrix. A decrease in calcein ﬂuorescence intensity in the mitochondrial matrix can be measured by ﬂuorescence microscopy or in a regular multiplate-based ﬂuorescence assay. This method proved to be a useful tool for the in situ study of mPTP modulation and it available in the form of commercial kits.

As described in Section III, mitochondrial swelling is a valuable method to study mPTP opening in isolated mitochondria. However, the observation of mitochondrial swelling in intact cells after mPTP opening is still somewhat controversial. While some authors argue that morphological alterations observed in mitochondria in intact cells result from mitochondrial swelling (Minamikawa et al; 1999), others believe that, in intact cells, mitochondrial swelling does not occur immediately after mPTP opening because of the presence of proteins in the cytosol that may inhibit osmotic water entry into mitochondria (Dumas et al; 2009). Thus mitochondrial swelling is not a uniformly accepted end-point for measuring mPTP opening in intact cells.

The evaluation of the mPT in intact cells thus is not a straightforward process and remains subject to confounding factors and artifacts. There is still no single reliable method to measure in situ the dynamic of themPT. Further dificulties arise when attempting to measure the low-conductance state of the mPTP, because most of the available methods are designed to evaluate its high-conductance form. Thus, for maximum reliability, a combination of different methods is required to evaluate mPTP dynamics in intact cells.

VI. THE mPTP IN CELL DEATH AND PATHOLOGY

As described in Section III, mPTP opening can lead to mechanical stress in the OMM due to swelling, leading to its rupture. It is of interest to consider whether this is a common phenomenon that takes place due to solute distribution between mitochondria and the cytosol in intact cells and tissues, or whether it is affected by the number of calciuminduced pores in a single mitochondrion. Neginskaya et al. (2020) calculated that the number of mPTP structures per mitochondrion was between one and nine, although they noted that this may be a possible underestimate. In any case, the mPTP is a central mechanism involved in cell death. Mitochondrial membrane depolarization resulting from mPTP opening leads to a deicit in mitochondrial ATP, a condition generally associated with necroptosis (Bauer & Murphy, 2020) or classical necrosis (Lemasters, 1999). Induction of the mPTP is also involved in Bax recruitment to mitochondria, an event that represents one of theirst steps in the mitochondrial apoptosis pathway (Narita et al; 1998; Precht et al; 2005). The binding of Bax to the ANT increases IMM permeability and triggers cell death (Marzo et al; 1998). It was also shown that the mPTP directs Bax translocation and multimerization in the OMM (De Giorgi et al; 2002). This phenomenon was observed to be inhibited by binding of the anti-apoptotic protein Bcl-2 (Brenner et al; 2000).

Further research demonstrated that mPTP induction is not always required for the apoptosis-inducing effects of Bax (Eskes et al; 1998). Apoptosis caused by Bcl-2 homology 3 (BH3)-interacting domain death agonist Bid,a pro-apoptosis ‘BH3-only’ member of the Bcl-2 family, which is cleaved by caspase-8 and translocated to the OMM, was initially shown not to depend on mPTP opening or Bax translocation (Kim et al; 2000), although later studies contradicted this (Zamzami et al; 2000). One mechanism to explain these differences may involve the concentration of calcium, which in low concentrations may induce cytochrome c release without mPTP induction, whereas higher concentrations can activate the mPTP and trigger cytochrome c release (Gogvadze et al; 2001).

Increased mPTP induction is associated with several pathological conditions. This is not surprising given its roles in cellular bioenergetic dysfunction and cell death and its association with redox and ionic disruption. We provide below a brief description of selected research in which the mPTP has been recognized as part of a pathophysiological mechanism.
The mPTP is a recognized mediator of neuronal death in several neurodegenerative diseases including amyotrophic lateral sclerosis (ALS) (Sun et al; 2019), Parkinson’s disease (Iravanpour et al; 2021), hypoxic—ischemic encephalopathy (Chen et al; 2021), and Alzheimer’s disease (Du et al; 2008) in different animal models. Interestingly, increased activation of the mPTP in Alzheimer’s disease was recorded even in non-neuronal tissues such as skin ibroblasts from patients (Perez et al; 2018).

Cardiac injury has been frequently associated with an increased activity of the mPTP. Anthracycline-induced cardiomyopathy, a pressing concern in anti-cancer therapy, has an important mitochondrial component with increased mPTP activation mediated by oxidative stress and CypD signalling (Oliveira et al; 2004; Dhingra et al; 2020; Wallace, Sardao & Oliveira, 2020). CypD hyperacetylation appears to be involved in increased mPTP opening during heart failure (Castillo et al; 2019).

It is known that CypD gene silencing is protective against ischemic renal injury in a mouse model (Devalaraja-Narashimha, Diener & Padanilam, 2009). The protective effect of silencing CyPD and that of Cyclosporin A both evidence the role of the mPTP in kidney injury. Indirect evidence for kidney cell injury caused by mPTP opening is provided by Bendavia, a mPTP inhibitor, which inhibits damage caused by revascularization in experimental renal artery stenosis (Eirin et al; 2012).

A recent study showed that hepatic steatosis in a mouse model was associated with mitochondrial swelling and depolarization, two mPTP-linked outcomes that were inhibited by Cyclosporin A (Li etal; 2021). This result conirmed previous indings showing that CypD inhibition improved fatty liver in animal models (Wang et al; 2018).

Aging is associated with progressive alterations in mitochondrial function, including altered mPTP regulation. Zhou et al. (2019) showed that reduced mPTP opening increases the lifespan of Caenorhabditis elegans. Their work also suggested an essential link between mPTP opening, IMM permeability, and autophagy in the regulation of yeast and mammal healthy aging (Zhou et al; 2019). An earlier study showed that mitochondria from older (36-month-old) animals had a higher response to Ca2+ in terms of mPTP induction compared to younger (1-month-old) animals (Goodell & Cortopassi, 1998).

Questions still remain on how over-activation of the mPTP occurs in these and other pathologies. Increasedoxidative stress is an obvious explanation, a condition that could be exacerbated by prolonged mPTP opening. Prolonged mPTP opening leads to mitochondrial depolarization, increased ROS production, and triggers cell death signalling, leading to tissue injury. Nevertheless, the situation is likely to be more complex. Some of the examples above include a role of CypD in potentiating mPTP opening, including post-translational modiications such as acetylation,resulting from the metabolic stress associated with different pathologies.

VII. NOT ALWAYS A BAD GUY: THE PHYSIOLOGICAL ROLES OF THE mPTP

In addition to the known pathological results of mPTP opening, it has been linked to diverse normal physiological processes, such as cell fate and differentiation (Hom et al; 2011; Folmes et al; 2012; Vega-Naredo et al; 2014). Several studies have addressed the importance of mPTP regulatory molecules under physiological conditions that prevent the cell from deleterious changes to mitochondrial functions that could lead to cell death.

In the context of a physiological role of the mPTP, two possible forms of its open state were proposed: a fullconductance irreversible opening for permanent permeability, or an alternative transient and ﬂickering short-term opening of the mPTP (Wang et al; 2008; Hou et al; 2014; Li et al; 2020) (Fig. 6). The full-conductance open state leads predominantly to apoptosis and cell death, while the transient short-term open state with smaller and more variable conductance is likely to function during physiological conditions (Perez & Quintanilla, 2017).

Evidence for such physiological opening of the mPTP has been described, including a ﬂickering opening activity (Crompton, 1999; Hausenloy et al; 2004, 2010; Korge et al; 2011); furthermore, an association between transient mPTP opening and ‘superoxide ﬂashes ’ was observed in striated muscle mitochondria (Wang et al; 2008). Based on the suggestion that this transient opening of the mPTP may release mitochondrial matrix Ca2+ to maintain mitochondrial homeostasis,a model was proposed for a physiological function of the mPTP, in addition to its well-known role in cell death (Elrod etal; 2010). The most prominent physiological role for the mPTP in mitochondrial and cellular Ca2+ homeostasis thus is apparently its capacity to act as a Ca2+ efﬂux mechanism (Altschuld et al; 1992; Takeuchi et al; 2015; Biasutto et al; 2016; Hurst et al; 2017; Krebs, 2017; Li et al; 2020; Xu et al; 2020).

Evidence for the transient opening of the mPTP asaphysiological Ca2+ efﬂux pathway includes early demonstrations of the inhibition of Ca2+ release in the presence of mPTP inhibitor Cyclosporin A in isolated adult rat ventricular cardiomyocytes (Altschuld et al; 1992). Transient or lowconductance opening of the mPTP was proposed to serve as an additional mode of Ca2+ efﬂux that mitigates sustained matrix Ca2+ overload (Ichas & Mazat, 1998; Bernardi & von Stockum, 2012). Numerous studies have supported this hypothesized physiological role (Elrod et al; 2010; Korge et al; 2011; Bernardi & von Stockum, 2012; Elrod & Molkentin, 2013; Gainutdinov et al; 2015). However, in a recent study, mitochondrial Ca2+ efﬂux rates measured in intact HeLa cells were completely unaffected by mPTP inhibition either by Cyclosporin A or by small interfering RNA (siRNA)-mediated reduction of the ATP synthase c-subunit (De Marchi et al; 2014a), suggesting that the mPTP may not always play a role in Ca2+ efﬂux under physiological conditions. Most experiments addressing the role of themPTPin Ca2+ homeostasis use Cyclosporin A, which introduces inherent problems regarding its speciicity. Notably, the inhibitory effect of Cyclosporin A depends on the expression level of CypD, which is rarely assessed (Bernardi et al; 2015). Cyclosporin A inhibitory ability can vary signiicantly, with known differences in sensitivity to Cyclosporin A in different tissues; for example, NIH3T3 ibroblasts and HL60 cells show no sensitivity to this inhibitor (Li et al; 2012). The relative expression of CypD and Fo-ATP synthase may also be crucial; cross-linking experiments in beefheart mitochondria indicate that there is much less CypD present than subunits b, d, and OSCP and that many Fo-ATP synthase channels are insensitive to Cyclosporin A even if CypD is expressed (Bernardi et al; 2015). Thus the role of Cyclosporin A-sensitive/insensitive mPTP functions, particularly as a mechanism of mitochondrial Ca2+ efﬂux, remains unproven. Despite this continuing debate, most research does suggest that the mPTP appears to function as a Ca2+-release mechanism required for proper metabolic regulation (Bernardi et al; 2015).

Ca2+ remains the most important regulator and inducer of pore opening, given its numerous indirect roles regulating and modulating the mPTP (Biasutto et al; 2016; Hurst et al; 2017). At physiological levels, Ca2+ could stimulate transient opening of the pore, while Ca2+ overload changes the balance from physiology to pathology, leading to sustained mPTP opening with subsequent mitochondrial and cellular dysfunction (Hurst et al; 2017; Mnatsakanyan et al; 2017; Perez & Quintanilla, 2017; Nesci et al; 2018; Lamb, 2020). Recent studies also highlighted a crucial role of the mPTP in cardiac, neurodegenerative and other pathologies and its involvement in cardiac and brain development (Folmes et al; 2012; Perez & Quintanilla, 2017), neutrophil activation and ROS release (Vorobjeva et al; 2020). ThemPTPhasbeenhighlightedasa gating mechanism underlying differentiation in the developing heart and brain, potentially involving cross-talk between genetic and metabolic signalling (Folmes et al; 2012). Transient mPTP opening directly regulates cellular energy metabolism as it uncouples oxidative metabolism from ATP synthesis, amechanism that operates in concert with ROS ﬂashes to promote cardiomyocyte differentiation (Folmes et al; 2012). Knockout of the mPTP component cyclophilin D results in elevated mitochondrial matrix Ca2+, enhancing the activation of Ca2+-dependent dehydrogenases and thusreducing metabolic ﬂexibility (Elrod et al; 2010). It was demonstrated that mPTP opening in the early heart is a physiological event required for organ development. In the fetal heart, myocytes exhibit low MMP, high levels of ROS production, and opening of the mPTP. Inhibition of mPTP opening with Cyclosporin A led to maturation of mitochondrial structure and function, decreased intracellular ROS levels and increased MMP, which accelerated myocyte differentiation (Hom et al; 2011).

Concerning the participation of the mPTP in neural development, it was reported that culturedembryonic mousecortical neural progenitor cells demonstrate intermittent spontaneous bursts of mitochondrial superoxide generation that require a transient opening of the mPTP (Hou et al; 2014). It was shown that both mitochondrial ROS scavengers and mPTP inhibitors such as Cyclosporin A reduced the frequency of mitoﬂashes and enhanced the proliferation of cortical neural progenitor cells, whereas prolonged mPTP opening and superoxide generation increased the incidence of mitoﬂashes and promoted differentiationoftheseneuronalcells(Hou et al;2012). Theevidencethat the mPTP participates in cardiac and brain development is thus convincing, but there remains much to understand regarding its rolesinothercell types[althoughforhaematopoietic progenitor cells and vascular progenitor cells see Arnold et al. (2000) and Davies et al. (2005)]. The physiological participation of the mPTP in cell development and differentiation appears to depend on the speciic cell type, the subcellular localization of mitochondria and on the developmental stage (Mnatsakanyan et al; 2017; Perez & Quintanilla, 2017). Transient pore opening has been shown to be associated with a transient depolarization of the MMP (Jou, 2011; Hurst et al; 2017). While the exact mechanism responsible is still unclear, the physiological roles are beginning to be elucidated. The frequency of these transient openings has been associated with metabolism, aging, wound healing, and an essential role in cell differentiation (Shen et al; 2014; Vega-Naredo et al; 2014; Ding et al; 2015).

VIII. THE ROAD TO THE CLINIC: FUTURE PERSPECTIVES IN mPTP REGULATION

The potentially dual physiological and pathological role of the mPTP has attracted attention in terms of drug discovery, with several pre-clinical trials already published. The mPTP has been extensively studied as a target for therapies aimed at blocking ATP synthesis, especially that of pathogens causing infective pathologies in humans (Andries et al; 2005). By binding the FO-ATP synthase csubunit and mycobacterial subunit ε, some antimicrobials can selectively and effectively eliminate strains of microorganisms that are resistant to conventional drugs. Another example of the use ofF1FO-ATP synthase asapharmacological target is 1,4-benzodiazepine (Bz-423), which induces a selective, ROSand mPTP-dependent apoptotic cell death in B lymphocytes via its interaction with the OSCP subunit (Johnson et al; 2005), the same site used by CypD to modulate mPTP. The discovery of Bz-423 has opened the way for potential treatments for autoimmune disorders such as lupus erythematosus.

Subunit c, together with CypD, have been deined as critical components of the mPTP. However, there are conceptual and technical dificulties in the use of both of these as targets for screening for inhibitors of mPTP opening: CypD is considered a regulator and not a pore component, thus, its targeting may only desensitize pore activity. Subunit c is part of the membrane-bound FO portion (see Fig. 3D, E) and involved in proton translocation for ATP generation. Therefore, its inhibition may lead to unwanted side effects, as is the case for oligomycin, an antibiotic that targets the FO component, and N,N-dicyclohexylcarbodiimide (DCCD), for example. Oligomycin is a natural macrolide that acts as a potent inhibitorofboththesynthesis andhydrolysis ofmitochondrial ATP. It is produced from Streptomyces and exists in six different isoforms, A toF, based onthe R group attached to the macrolide. Oligomycin binds side chains of amino acids located on two consecutive c subunits and can inhibit 50% of mPTP opening in living cells at a concentration of 10 μM (Bonora etal; 2017). DCCD is a lipid-soluble carbodiimide with strong inhibitory properties of both portions ofF1FO-ATP synthase, depending on its use. At low concentrations (about 50 μM), it interacts covalently with the c-ring through an essential carboxyl amino acid (Asp61) of subunit c and inhibits mPTP opening by 45% if used in the range 7.5– 15 μM; at higher concentrations, it also interacts with the F1 portion through a glutamine residue in the ß subunit. However, the toxic nature of these drugs makes themunacceptablefor use: despitesigniicant inhibition of mPTP activity in vitro, they also deplete mitochondrial ATP, causing toxic side effects in more complex disease models.

Considering the importance of the c-ring in mPTP modulation, many efforts are ongoing to ind less toxic c-subunit inhibitors. This goal could be achieved by identifying the essential core of a drug required to recognize subunit c, and adapting the surrounding chemical structure to minimize unwanted side effects while maintaining its inhibitory potential. A small-molecule library of c-subunit inhibitors has been obtained by modifying the oligomycin functional core, leading to new compounds that strongly reduce reperfusion damage in animal models of global ischemia without interfering with ATP production (Morciano et al; 2018). For example, compound 10 inhibited mPTP opening in vitro by 40–50% at very low concentrations (0.5– 1 μM) and reduced cardiac apoptotic cell death by 40% when administered during reperfusion for 10 min. Compound 10, together with compounds 5c and 6g, showed low toxicity, probably due to their exclusive localization in mitochondria and their reversible binding.

Danshensu (DSS), the main constituent of Salvia miltiorrhiza (Danshen), a traditional Chinese herb, provided substantial cardioprotection against myocardial ischemia/reperfusion injury as measured by cell viability loss, and creatine kinase-isozyme MB, cardiac troponin and lactate dehydrogenase (LDH) release. This cardioprotection was dependent of the modulation of subunit c protein (Yin et al; 2013). DSS acted by downregulating subunit c messenger RNA (mRNA) and protein levels in reperfused rat hearts, and inhibited mPTP opening and the consequent cardiac ischemia/reperfusion injury, improving heart parameters and cardiomyocyte survival (Gao et al; 2017).

Genetic studies in animals have highlighted the importance of CypD in mPTP-mediated cardioprotection by modifying itsPpif gene expression. CypD can be manipulated by a long list of drugs, almost all derived from Cyclosporin A. Cyclosporin Awas irst isolated from a fungus and entered clinical practice some years later; its binding to CypD is due to a tryptophan (Trp-121) within a short α-helical region of the protein (Davis et al; 2010). Its excellent activity in inhibiting mPTP opening at low concentrations (0.2– 1.2 μM) and its unquestionable results in in vitro and preclinical models, have made Cyclosporin A one of the most promising positive controls for assessing mPTP function. However, Cyclosporin A also has non-mPTP-related effects. Probably due to its diffuse localization pattern inside cells, in both cytosol (Youn et al; 2002; Abikhair et al; 2016) and nucleus (Le Hir et al; 1995), Cyclosporin A can act as an immunosuppressant. Its incorporation into poly-lactic/glycolic acid (PLGA) nanoparticles (CsA-NP) allows better mitochondrial localization; treatment with CsA-NP at the time of reperfusion increased cardioprotection with a signiicant reduction in infarct size using lower concentrations compared to Cyclosporin A alone (Ikeda et al; 2016). Cyclosporin A derivatives with a signiicantly decreased (several thousand-fold less) immunosuppressant effect have also been developed. Examples include N-methyl-isoleucine-4-cyclo-sporin (NIM-811) and N-methylD-alanine-3-N-ethyl-valine-4-cyclosporin (Debio025). NIM811 and Debio025 are both semisynthetic analogs of CsA in which cytosolic side effects have been partially abolished by eliminating the calcineurin-binding motif (Waldmeier et al; 2002; Hansson et al; 2004). In assays targeting the mPTP induced with Ca2+ overload and Pi in both living cells and isolated mitochondria, NIM-811 showed the same potency as Cyclosporin A. The reduction in its immunosuppressant effects offers an additional advantage since Cyclosporin A has a limited window of action in terms of concentrations (0.2 to 1.2 μM). NIM-811 does not show the same toxicity. Debio025, a NIM-811 derivative, in a comparative study performed in isolated brain and heart mitochondria, showed a 10-fold more potent activity than Cyclosporin A. Multiple studies have reported beneicial effects of these two drugs in reduced cell death, recovery of left ventricle contractile function, and improved survival (Gomez et al; 2007, 2009; Cour et al; 2011).

Despite the preclinical potential of its derivatives, only Cyclosporin A has so far entered clinical trials, which ended after 15 years with a phase III failure against cardiac reperfusion injury. Indeed, CIRCUS (Cung et al; 2015) and CYCLE (Ottani et al; 2016) trials, consisting of a single intravenous bolus of Cyclosporin A (2.5 mg/kg) before revascularization, showed no improvement in clinical outcome. These indings did not conirm the results of a pilot study (Piot et al; 2008), which initially raised hope for the use of Cyclosporin A to treat reperfusion injury.

Another non-selective inhibitor of CypD is Sanglifehrin A (SfA) (Clarke, McStay & Halestrap, 2002). SfA is as potent as Cyclosporin A regarding mPTP opening; recovery of cardiac performance upon ischemia/reperfusion and a signiicant reduction of LDH release was observed after SfA treatment (Hausenloy et al; 2003). Since SfA differs structurally from Cyclosporin A, the formation of calcineurin–SfA complexes is avoided, while the ability to bind CypD although at a different site,is maintained, giving this molecule discrete immunosuppressant activities. Similar to the effects of DCCD on ATP synthase subunit c, SfA strongly binds CypD, failing to detach upon washing. Together with its immunosuppressant activity, this may discourage its use in clinical practice, as prolonged residence inside cells may lead to side effects.

Small-size molecule inhibitors also exist for CypD, such as C-9 and C-19. C-9 was initially identiied as a therapeutic agent to delay Alzheimer’s disease symptoms by preventing the interaction between CypD and amyloid-beta (Aβ) to decrease mitochondria-dependent neuronal stress (Valasani et al; 2014). C-9 was also found in vitro to show potential for the treatment of many other diseases such as acute pancreatitis, ultraviolet radiation damage, and in other neurodegenerative diseases, due to its mPTP-inhibiting properties. The most potent inhibitor in this category is C31 (Panel et al; 2019), which was able to restore mitochondrial parameters following hepatic injury.

Suggested initially as mPTP regulator and most commonly known as the peripheral benzodiazepine receptor for its high afinityin binding benzodiazepines, TSPO has been proposed as a potent inducer of mPTP opening upon interaction with protoporphyrin IX (PPIX) (Pastorino et al; 1994). Targeting of TSPO by 3,5-seco-4-nor-cholestan-5-one oxime-3-ol (TRO40303) showed promising cardioprotective effects in a rat model of cardiac ischemia; its administration prior to reperfusion reduced infarct size (IS) and concomitant cell death by about 40%. However, desensitization of mPTP opening seems to be secondary to its remarkable antioxidant properties (Schaller et al; 2010). Further supporting an indirect mechanism in the modulation of mPTP, studies on TSPO KO mice have excluded the possibility that TSPO ligands, and TSPO itself, may regulate mPTP activity by showing that the presence of the protein in hearts subjected to ischemia and reperfusion was dispensable (Sileikyte et al; 2014). The safety and eficacy of TRO40303 were evaluated in myocardial infarction patients undergoing percutaneous coronary intervention. This multicenter, double-blinded, phase II study (MITOCARE), in which TRO40303 was administered just before revascularization failed to show eficacy in reducing or limiting reperfusion injury (Atar et al; 2015). In this study, infarct size, left ventricle (LV) ejection fraction evaluation, creatinekinase, and troponin I dosage did not differ between placeboand treated patients.

Investigations of the properties of the mPTP have also highlighted compounds derived from cinnamicanilide, such as GNX-4728 and GNX-4975. These molecules modulate themPTPin a CypDand subunit c-independent way, showing beneicial effects in an ALS transgenic mouse model (Martin et al; 2014) with delayed onset of symptoms, increased lifespan, and reduced inﬂammation. It has been hypothesized that GNX-4975 shares the same binding site as calcium in mPTP opening; once open, ANT and PiC would be subject to conformational changes to forman interface in the inner membrane to which the compound binds (Richardson & Halestrap, 2016).

Screening performed on isolated mitochondria with thousands of compounds in the National Institute of Health repository identiied many other small-molecule inhibitors, based on an isoxazole functional core. Of these compounds, ML-404 selectively inhibited mPTP opening without side effects in concentrations up to 100 μM; this compound was classiied with twice the potency of GNX-series compounds (Sileikyte et al; 2010). Having a synergistic effect with Cyclosporin A, ML-404 and other isoxazole-based compounds (such as compound 60) do not act through CypD binding (Roy et al; 2015).

Another strategy to inhibit mPTP opening via CypDindependent interactions involved compounds based on the N-phenylbenzamide scaffold; these have pharmacological relevance due to their complete inhibition of the pore at low concentrations without interfering with ATP synthase production (Roy et al; 2016).

Similar to TRO40303, many compounds that promote mPTP desensitization act primarily as potent oxidantscavengers: these include a gallic acid-derivative, themitochondriotropic antioxidant AntiOxBEN3 (Teixeira et al; 2018) and Agomelatine (AGO), a melatonin receptor agonist. These two molecules showed low toxicity, particularly for AGO which protected the rat ischemic myocardium by acting on targets upstream of the mPTP, including the enhanced phosphorylation of GSK3β. Melatonin itself, which has antioxidant activity, also modulates mPTP opening. New evidence and ongoing work has focused on additional and direct effects on the pore (Zhou et al; 2017; Tarocco et al; 2019). Currently, melatonin is considered one of the safest drugs available for use as a mPTP inhibitor. Melatonin is a chronobiotic indolamine mainly synthesized and secreted in the pineal gland with a marked circadian rhythm. Given its high level of lipophilicity, pineal melatonin can diffuse across cell membranes, allowing the distribution of this molecule throughout all cells of the body and inﬂuencing organ function (Tan et al; 1999). However, increasing evidence supports that rather than a passive diffusion process, there are active or facilitated mechanisms that favour melatonin uptake and its cellular internalization, probably via members of the solute carrier family 2 (SLC2)/glucose transporter (GLUT) and peptide transporter 1 (PEPT1) families (Hevia et al; 2008; Huo et al; 2017; Mayo et al; 2018). Indeed, some cellular organelles, especially mitochondria, were found to accumulate high melatonin concentrations (Leon et al; 2004). A recent study also showed that large amounts of melatonin are synthesized by mitochondria during oocyte maturation, being essential for the maintenance of energy metabolism, mitochondrial function, and the quality of oocytes (He etal; 2016). Enzymes involved in the synthesis of melatonin were found in mitochondria and isolated mitochondria retain the capacity to produce melatonin (Coelho et al; 2015; He et al; 2016). These indings underline the importance of melatonin for the mitochondria and raise questions regarding the physiological and protective effects of this neurohormone on these organelles. In addition to its antioxidant role, melatonin has a protective effect against neural disorders and other diseases via modulation of the activity of the mPTP and apoptosis responses (Petrosillo et al; 2009; Espino et al; 2010). Likewise, melatonin stimulates uncoupling proteins (UCPs), contributing to dissipation of the electrochemical proton gradient across the IMM and the consequent reduction of the MMP (Tan et al; 2016; Pan et al; 2018).

mPTP modulation by melatonin is a novel role within the broad spectrum of protective actions of melatonin in diverse diseases, especially in neurological disorders. Early observations in rat brain astrocytes showed that melatonin seems to suppress mitochondrial ROS formation and to target Ca2+mediated mPT to protect against cell death (Jou et al; 2010). A later study using ﬂuorescence laser scanning microscopy found that melatonin prevents mitochondrial depolarization and mPTP neurotoxicity under disturbed Ca+2 homeostasis, preserving the protective conformation of the mPTP (Jou, 2011). A recent study demonstrated that melatonin addition to isolated brain mitochondria inhibits the opening of the mPTP, probably hindering mPTPmediated mitochondrial dysfunction (Waseem, Tabassum & Parvez, 2016). In this study, isolated mitochondria were incubated with Ca2+ and 5-hydroxydecanoate to stimulate mitochondrial swelling and induce mPTP opening. Melatonin administration signiicantly reduced mitochondrial swelling and MMP and improved mitochondrial respiration. Beneits of melatonin are also known in the heart from aged mice and in in frozen–thawed sperm, where melatonin administration inhibited mPTP opening and improved cellular respiration and ATP production (Sahach et al; 2008; Fang et al; 2019). A recent mechanistic study revealed that a melatonin receptor 1 (MT1) antagonist eliminated the mPTP opening inhibition generated by melatonin, whereas aMT1 agonist had the opposite effect (Fang et al; 2020). However, the exact mechanism ofaction by which melatonin regulates the mPTP still remains unclear, although related molecular mechanisms have been proposed. Interestingly, promising results may provide evidence that melatonin can regulate the mPTP via inhibition of CypD (Zhou et al; 2018). This study transfected CypD mutants (mimicking permanent phosphorylation) into melatonin-treated endothelial cells and demonstrated that melatonin represses the receptor-interacting serine/threonine-protein kinase-3 (Ripk3)–phosphoglycerate mutase family member 5 (PGAM5)–CypD cascade, attenuating necroptosis and cardiac ischemia–reperfusion injury. Another study used recordings of IMM potentials with a patch-clamp approach in liver mitoplasts from rodents to evaluate the direct effects of melatonin on the mPTP (Andrabi et al; 2004). This work showed that melatonin inhibits mPTP opening in a dose-dependent manner, although there was no evidence regarding its target protein. Together, these publications show that melatonin can modulate mPTP activity and preserves the optimal MMP and mitochondrial integrity, contributing to the maintenance of cell functions and survival.

IX. CONCLUSIONS

(1) Advances in methodologies, including genetic manipulation, have accelerated our understanding of the mPTP as a constantly evolving entity present in many species, although with different forms of regulation.
(2) The mPTP plays crucial roles in cell physiology and pathology; excessive mPTP opening is involved in the pathophysiology of different human diseases, but regulated mPTP opening is critical for regulating cell and mitochondrial ionic and redox balance, and plays an apparently important role in fetal development.
(3) Opening of the mPTP increases the permeability of mitochondrial membranes to different solutes, and can switch from a lower to a higher conductance state according to different stress or physiological stimuli.
(4) The mPTP is notoriously activated by high levels of oxidative stress and increased Ca2+ concentrations, but it is also precisely regulated by several other metabolites.
(5) No single methodology can accurately measure mPTP opening rates.
(6) More recent models suggest that ATP synthase is a structural component of the mPTP, although doubt still exists on the exact subunit(s) responsible for channel activity.

Nur77 Signal

Reproductive isolation also tends to increase with genetic divergence.

The mitochondrial permeability transition pore: an evolving concept critical for cell life and death

Leave a Reply Cancel reply