Why don’t mice lacking the mitochondrial Ca²⁺ uniporter experience an energy crisis?

Abstract

Current dogma holds that the heart balances energy demand and supply effectively and sustainably by sequestering enough Ca²⁺ into mitochondria during heartbeats to stimulate metabolic enzymes in the tricarboxylic acid (TCA) cycle and electron transport chain (ETC). This process is called excitation-contraction-bioenergetics (ECB) coupling. Recent breakthroughs in identifying the mitochondrial Ca²⁺ uniporter (MCU) and its associated proteins have opened up new windows for interrogating the molecular mechanisms of mitochondrial Ca²⁺ homeostasis regulation and its role in ECB coupling. Despite remarkable progress made in the past 7 years, it has been surprising, almost disappointing, that germline MCU deficiency in mice with certain genetic background yields viable pups, and knockout of the MCU in adult heart does not cause lethality. Moreover, MCU deficiency results in few adverse phenotypes, normal performance, and preserved bioenergetics in the heart at baseline. In this review, we briefly assess the existing literature on mitochondrial Ca²⁺ homeostasis regulation and then we consider possible explanations for why MCU-deficient mice are spared from energy crises under physiological conditions. We propose that MCU and/or mitochondrial Ca²⁺ may have limited ability to set ECB coupling, that other mitochondrial Ca²⁺ handling mechanisms may play a role, and that extra-mitochondrial Ca²⁺ may regulate ECB coupling. Since the heart needs to regenerate a significant amount of ATP to assure the perpetuation of heartbeats, multiple mechanisms are likely to work in concert to match energy supply with demand.

Graphical Abstract

We propose that MCU and/or mitochondrial Ca²⁺ may have limited ability to set ECB coupling and that other mitochondrial Ca²⁺ handling mechanisms and/or extra-mitochondrial Ca²⁺ may regulate ECB coupling. The diagram shows the major mitochondrial Ca²⁺ handling processes (influx and efflux) and the role of intra- and extra-mitochondrial Ca²⁺ in regulating cardiac bioenergetics. AP, action potential; C, cytochrome C; Q, co-enzyme Q; I-V, the five complexes of electron transport chain; LCC, L-type Ca²⁺ channel; P_i, phosphate; Δψ_m, mitochondrial membrane potential. Please refer to the text for the other abbreviations.

Introduction

The normal heart in an average sized person pumps 350 l of blood per hour. The energy required to do this work is 5 W of power (250 g of ATP) (Neubauer, 2007). Without perpetual ATP generation, the heart will use up its reservoir of energy within 1 min of contractions and life will cease. To ensure its continued pumping function, the heart possesses the most active energy metabolism in the body, which is centred on mitochondria. Mitochondria are double-membrane-bound organelles in eukaryotic cells that carry out multiple functions, from bioenergetics to cell signalling and cell fate determination (Kroemer, 1999; Ohgami et al. 2005). Their most prominent function is generating ATP downstream of serial metabolic cascades from intermediate catabolism of glucose, fatty acids and amino acids to the tricarboxylic acid (TCA) cycle and to oxidative phosphorylation (OXPHOS) via the electron transport chain (ETC) (Balaban, 2002). This process provides 95% of the ATP in the adult heart (Henze & Martin, 2003). The heart must keep its metabolic rate constantly high at baseline and be able to adjust the rate promptly and proportionately in the face of a fluctuating workload. How mitochondria control this high rate of energy flow (i.e. 250 g of ATP per hour, which far exceeds the capacity of substrate storage in the heart) is a fascinating question.

Mitochondrial energy metabolism is a multi-step and rate-limiting process, in which activation of one step does not impact overall ATP synthesis. A single signal that accelerates all of the partially rate-limiting steps simultaneously would be ideal to regulate energy metabolism, and mitochondrial Ca²⁺ appears to be the perfect candidate to act as such a master regulator of bioenergetics. Indeed, it has been known for decades that mitochondria can take up Ca²⁺ (Slater & Cleland, 1953; DeLuca et al. 1962; Vasington & Murphy, 1962). During cardiac excitation-contraction (EC) coupling, transiently increased cytosolic Ca²⁺ can lead to elevation in intra-mitochondrial free Ca²⁺, which stimulates enzymes in the TCA cycle and ETC, leading to accelerated ATP synthesis (Balaban, 2009; Denton, 2009; Carafoli, 2010; Ivannikov & Macleod, 2013). When researchers identified the mitochondrial Ca²⁺ uniporter (MCU), a long-sought mitochondrial Ca²⁺ uptake channel, it seemed that the molecular link between mitochondrial Ca²⁺ uptake and excitation-contraction-bioenergetics (ECB) coupling had finally been found (Baughman et al. 2011; De Stefani et al. 2011). Yet, unexpectedly, genetic ablation of the MCU in different strains of mice by various means resulted in viable mice with little or no energy crisis in the heart (Pan et al. 2013; Luongo et al. 2015; Rasmussen et al. 2015). These puzzling results challenge the theory that the MCU and/or mitochondrial Ca²⁺ play a key role in ECB coupling and open up new opportunities for research.

Mitochondrial Ca²⁺ uptake and efflux

Given the important role of mitochondrial Ca²⁺ in bioenergetics, researchers have been actively studying how mitochondria handle Ca²⁺ for decades. It is now believed that a number of channels or carriers are in place to control Ca²⁺ transportation across the outer and inner mitochondrial membranes. The outer mitochondrial membrane is permeable to ions, and Ca²⁺ can cross it freely through the voltage-dependent anion channel (VDAC) (Gincel et al. 2001; Báthori et al. 2006; Wu et al. 2015). Therefore, the intermembrane space has similar Ca²⁺ concentrations as the cytosolic side of the outer membrane. However, it has been suggested that, under certain conditions, the VDAC may provide a barrier to the free diffusion of Ca²⁺ (Rapizzi et al. 2002). For example, the VDAC provides a higher Ca²⁺ permeability in the closed states (Tan & Colombini, 2007), indicating that Ca²⁺ transport across the outer membrane could be regulated. The inner mitochondrial membrane is impermeable to nearly all ions, and Ca²⁺ crosses it through three influx mechanisms or modes and two efflux mechanisms (Rizzuto et al. 2000; Bianchi et al. 2004; Gunter & Sheu, 2009; Rizzuto et al. 2009). In the steady state, mitochondrial Ca²⁺ uptake must equal efflux so that a stable mitochondrial Ca²⁺ concentration is maintained.

The most studied mitochondrial Ca²⁺ uptake mechanism is the MCU, a channel transporting Ca²⁺ into the matrix driven by an electrochemical gradient across the inner membrane. Functional and pharmacological studies revealed long ago that the MCU is a Ca²⁺-selective channel whose activity is blocked by Ruthenium Red and lanthanides (DeLuca & Engstrom, 1961; Vasington & Murphy, 1962). Electrophysiological channel recordings of Ca²⁺-selective currents in isolated mitoplasts from human hearts have uncovered distinct modes of mitochondrial Ca²⁺ uptake, and one of them is attributed to the MCU (Michels et al. 2009). But despite its being proposed as the main mechanism for mitochondrial Ca²⁺ uptake, the molecular identity of the MCU was only characterized recently, when two groups using different approaches determined that a mitochondrial protein with unknown function was the MCU (Baughman et al. 2011; De Stefani et al. 2011). These findings opened up the exciting possibility of investigating mitochondrial Ca²⁺ handling via genetic approaches. The other identified mechanism of Ca²⁺ uptake, namely rapid uptake mode (RaM), has received less attention (Sparagna et al. 1994). RaM was first observed by exposing isolated mitochondria to rapid Ca²⁺ pulses in a cylindrical cuvette similar to those observed in vivo in many types of tissue (Sparagna et al. 1994). At the same time, this approach permits the authors to measure mitochondrial Ca²⁺ uptake accurately using a dual label isotope technique. The ability of RaM to rapidly switch mitochondrial Ca²⁺ sequestration on and off led to the suggestion that it is associated more closely with metabolic signalling, while the MCU is associated both with metabolic signalling and with induction of the permeability transition or with apoptotic signalling (Gunter et al. 1998). The third mitochondrial Ca²⁺ uptake mechanism, specifically in cardiomyocytes, is the skeletal muscle type I ryanodine receptor (RyR1) (Munch et al. 2000; Beutner et al. 2001; Jeyakumar et al. 2002; Beutner et al. 2005; Ryu et al. 2011).

Ca²⁺ leaves the mitochondria via two mechanisms: a Na⁺-dependent mechanism and a Na⁺-independent mechanism, both of which were initially discovered in the 1970s (Puskin et al. 1976). It has been known for some time that Ca²⁺ efflux from heart mitochondria is dependent on the Na⁺ gradient, that Li⁺ can also stimulate Ca²⁺ efflux, and that Ruthenium Red stimulates rather than inhibits Ca²⁺ efflux (Carafoli et al. 1974; Nicholls & Åkerman, 1982). The mitochondrial matrix Na⁺ concentration is thought to be lower than cytosolic Na⁺ due to the function of the mitochondrial Na⁺-H⁺ exchanger, which transports Na⁺ out of the matrix driven by the pH gradient (Murphy & Eisner, 2009). The Na⁺ gradient, in turn, drives Ca²⁺ efflux from the matrix via the Na⁺-Ca²⁺-Li⁺ exchanger (NCLX) (Palty et al. 2010). Some evidence suggests that NCLX is electrogenic, exchanging 3 Na⁺ ions for 1 Ca²⁺ ion (Mannella et al. 2013). Therefore, NCLX could potentially function in reverse mode and transport Ca²⁺ into the matrix when the mitochondrial membrane potential is dissipated. NCLX can be inhibited pharmacologically by CGP37157. The second mechanism for Ca²⁺ efflux from mitochondria is through transient opening of the mitochondrial permeability transition pore (mPTP) (Pfeiffer & Tchen, 1975; Hunter & Haworth, 1979). This large and non-selective pore was initially linked to pathological consequences such as mitochondrial dysfunction and apoptosis, but its brief opening has since been viewed as a physiologically important mechanism for mitochondrial and cell function (Nicklas et al. 1987; Halestrap et al. 2004; Przedborski et al. 2004; Halestrap, 2006). The mPTP may also allow Ca²⁺ to enter mitochondria under conditions of membrane potential dissipation (Saotome et al. 2005).

The physiological and pathological roles of mitochondrial Ca²⁺

Mitochondrial Ca²⁺ uptake was initially viewed as a way to regulate cytosolic Ca²⁺ by serving as intracellular Ca²⁺ storage and by releasing Ca²⁺ upon stimulation (Nicholls, 1978). But this hypothesis was largely refuted when the endoplasmic reticulum (ER) and sarcoplasmic reticulum (SR) were identified as the primary sites for Ca²⁺ storage. Although manipulating mitochondrial Ca²⁺ uptake has been shown to impact cytosolic Ca²⁺ levels at baseline and after stimulation, whether mitochondria can buffer physiological Ca²⁺ transients in the cytosol, such as during cardiac EC coupling, is still under debate (David et al. 1998; Parekh, 2003; Nicholls, 2005; Rizzuto & Pozzan, 2006). The major physiological role of mitochondrial Ca²⁺ is attributed to energy metabolism. Ca²⁺ regulates the activity of three mitochondrial dehydrogenases: pyruvate dehydrogenase, isocitrate dehydrogenase and α-ketoglutarate dehydrogenase (Denton & McCormack, 1980). Ca²⁺ can also activate F_oF₁-ATP synthase, the Complex V in the ETC that utilizes proton motive force for ATP synthesis (Territo et al. 2000). This effect has been observed in large species with slow heart rates (e.g. pigs and dogs) and shown to be retained in isolated mitochondria (Phillips et al. 2012). Taken together, the consensus has been that the key physiological role of mitochondrial Ca²⁺ uptake is to elevate matrix Ca²⁺ to support energy metabolism via activating enzymes in the TCA and ETC. The idea that mitochondrial Ca²⁺ is a master regulator of bioenergetics is particularly plausible for the excitable cells, such as neurons and muscle cells that have dynamic intracellular Ca²⁺ cycling and huge energy demands. For instance, in adult cardiomyocytes, fluctuations in workload lead to immediate increases in cytosolic Ca²⁺, which could then be taken up by mitochondria to stimulate metabolism and ATP generation. Thus, mitochondrial Ca²⁺ could be the key mediator of ECB coupling by which workload is directly coupled with energy supply (Glancy et al. 2013).

Under pathological conditions, an increase or decrease in mitochondrial Ca²⁺ levels can result in mitochondrial and cellular dysfunction. In most cases, a dramatically increased mitochondrial Ca²⁺ level, termed Ca²⁺ overload, is a key contributor to disease. For instance, mitochondrial Ca²⁺ overload has been causally linked to necrotic cell death (Murphy & Steenbergen, 2007). One of the major mediators in Ca²⁺-induced cell death is the mPTP (Halestrap et al. 2000; Seidlmayer et al. 2015). Indeed, the experiments that originally identified the mPTP used large amounts of Ca²⁺ to trigger mitochondrial swelling, a hallmark of the opening of this large pore (Hunter et al. 1976). Subsequent studies have shown that mitochondrial matrix Ca²⁺ targets a key mPTP regulator and component, cyclophilin D, to trigger mPTP opening and subsequent cell death under various pathological conditions (Crompton et al. 1988; Griffiths & Halestrap, 1991; Chalmers & Nicholls, 2003). One of the most important diseases linked to mitochondrial Ca²⁺ overload and mPTP opening is ischaemia reperfusion (I/R) injury (Steenbergen et al. 1987; Marban et al. 1989; Murphy et al. 1991; Hausenloy & Yellon, 2013). Cyclophilin D knockout or inhibition by cyclosporine A has been shown to protect the heart from I/R injury (Griffiths & Halestrap, 1993; Baines et al. 2005). Under chronic stresses, such as heart failure, mitochondrial Ca²⁺ is either increased or decreased and causally linked to the pathogenesis of disease. It has been reported that the leaky RyR2 in failing heart leads to mitochondrial Ca²⁺ overload, which induces oxidative stress, mitochondrial dysfunction and declining heart function (Palty et al. 2010; Fauconnier et al. 2011; Santulli et al. 2015). Others have reported that mitochondrial Ca²⁺ is decreased in the failing heart due to elevated cytosolic Na⁺, which accelerates Ca²⁺ efflux via NCLX, and that inhibiting NCLX by CGP37157 benefits the failing heart (Liu et al. 2014). A decreased matrix Ca²⁺ level compromises TCA activity and the generation of NAD(P)H, which are important antioxidants inside mitochondria (Bertero & Maack, 2018). Thus, either decreased or increased mitochondrial Ca²⁺ may lead to oxidative stress and cell damage in heart failure.

The mitochondrial Ca²⁺ uniporter complex (MCU)

The MCU is the pore forming subunit of the MCU complex

The identity of the MCU was revealed in recent groundbreaking studies that show it is encoded by the coiled-coil domain-containing protein 109A (CCDC109A) (Baughman et al. 2011; De Stefani et al. 2011). The MCU gene encodes for a ~35 kDa protein composed of two coiled-coil domains and two transmembrane domains linked by a short loop enriched in acidic residues and is well conserved in all eukaryotes except yeasts (De Stefani et al. 2011). The MCU reconstituted on lipid bilayers exhibits similar currents with a lower open probability to those previously recorded by patch clamp of isolated mitoplasts (Kirichok et al. 2004; De Stefani et al. 2011). Thus, the MCU alone is necessary and sufficient to transport Ca²⁺ across the membrane. Silencing the MCU inhibits mitochondrial Ca²⁺ transients while overexpressing the MCU significantly increases them (De Stefani et al. 2011). Follow-up studies have tested the essential role of the MCU in mitochondrial Ca²⁺ uptake in different systems. For example, in vivo delivery of MCU siRNA abolishes mitochondrial Ca²⁺ uptake in the liver, cardiomyocytes, heart, pancreatic β cells, neurons and cancer cells (Baughman et al. 2011; Drago et al. 2012; Joiner et al. 2012; Tarasov et al. 2012; Qiuetal.2013).Finally, mitochondria isolated from tissues of the MCU knockout mice do not take up any Ca²⁺ when treated with increasing extra-mitochondrial Ca²⁺, further supporting the prevailing idea that the MCU is the one and only mitochondrial Ca²⁺ uptake mechanism (Pan et al. 2013). Interestingly, the MCU cannot regulate its own activity due to a lack of any Ca²⁺ binding domains. Indeed, the MCU is identified by screening the binding proteins of its regulator, mitochondrial Ca²⁺ uptake protein 1 (MICU1), which contains canonical EF-hand Ca²⁺ binding domains (Perocchi et al. 2010). Regulation of MCU channel activity is mainly achieved by its associated proteins (see below), although transcriptional (Marchi et al. 2013; Diaz-Juarez et al. 2016), translational (Zaglia et al. 2017), and post-translational regulation (Joiner et al. 2012; O-Uchi et al. 2014) of the MCU have also been reported. The channel activity can be inhibited by Ruthenium Red, Gd³⁺, or site-specific mutagenesis on two residues (D261, E264) in the loop region (De Stefani et al. 2011). An earlier structural study indicates that C. elegans MCU forms pentamers for Ca²⁺ transportation (Oxenoid et al. 2016). Most recently, cryo-EM and X-ray structures of fungi MCU indicate a tetrameric architecture (Fan et al. 2018; Nguyen et al. 2018; Yoo et al. 2018).

MCUb is an inhibitory homologue of the MCU

MCUb is encoded by CCDC109B and was originally thought to be a subtype of the MCU. It shares ~50% homology in sequence and has high structural similarity with the MCU (De Stefani et al. 2015). MCUb is conserved in most vertebrates and in many plants but is absent in other organisms where the MCU is present (De Stefani et al. 2015). MCUb has a crucial amino acid substitution in its loop region (E256V), which has a negative effect on the channelling properties as shown by molecular dynamics simulations and confirmed by experiments in HeLa cells and lipid bilayers (Raffaello et al. 2013). Overexpression of MCUb decreases agonist-induced mitochondrial Ca²⁺ transients, while silencing MCUb significantly increases mitochondrial Ca²⁺ uptake. Although recombinant MCUb in artificial membranes shows no Ca²⁺ current, it significantly lowers the MCU channel open probability when co-expressed at a very low level with the MCU. These results support the idea that MCUb plays a dominant negative role in MCU-mediated Ca²⁺ transportation. Indeed, the ratio between MCUb and MCU is different among tissues and correlates with tissue-specific MCU activity (Fieni et al. 2012). However, how the inhibitory effect of MCUb is regulated and why this endogenous inhibitory homologue is needed remain elusive. Given their structural similarity, it is believed that MCUb can form hetero-oligomers with the MCU during channel formation and its incorporation decreases the channel open probability (Raffaello et al. 2013).

The MICUs are gatekeepers of the MCU complex

MICU1 was the first member of the MCU complex to be discovered (Perocchi et al. 2010). Subsequently, another two isoforms of MICU were found: MICU2 and MICU3 (Plovanich et al. 2013; Kovács-Bogdán et al. 2014). Recent proteomics and functional studies have indicated that MICUs are soluble proteins located in the intermembrane space (Mallilankaraman et al. 2012b; Hung et al. 2014; Patron et al. 2014; Lam et al. 2015). The gatekeeper role of MICU1 was revealed by a study showing that it keeps the MCU closed when extra mitochondrial Ca²⁺ concentration is low (Mallilankaraman et al. 2012b). Evidence suggests that silencing MICU1 engenders mitochondrial Ca²⁺ overload. Later, it was shown that mitochondrial Ca²⁺ uptake is less efficient without MICU1 (Csordas et al. 2013). Based on evidence to date, MICU1 not only keeps the MCU closed in the resting condition but also facilitates its opening when stimulated. The gatekeeper role of MICU fits well with previous findings that mitochondrial Ca²⁺ uptake follows a sigmoidal relationship with extra-mitochondrial Ca²⁺ concentrations. Mitochondrial Ca²⁺ uptake rate is low at resting cytosolic Ca²⁺ levels and increases significantly at higher cytosolic Ca²⁺ levels (Csordás et al. 2010; Giacomello et al. 2010). This unique feature is important for mitochondrial Ca²⁺ regulation because it prevents Ca²⁺ overload and ensures a prompt response to agonists. Although MICU1 critically regulates mitochondrial Ca²⁺ uptake, it lacks a transmembrane domain and cannot form the channel. MICUs may fine-tune MCU channel activity through their two conserved EF-hand Ca²⁺ binding domains (Plovanich et al. 2013). MICU2 seems to share a gatekeeper function with MICU1, forming a heterodimer with MICU1, and interacting with the loop region of the MCU, which is located in the intermembrane space (Patron et al. 2014). Interestingly, the stability of MICU2 is dependent on MICU1 (Plovanich et al. 2013). MICU3 is predominantly expressed in the central nervous system (Plovanich et al. 2013), and its exact function in the MCU complex regulation needs further investigation.

Essential MCU regulator (EMRE) is an essential regulator for MCU activity

The essential MCU regulator (EMRE) is a 10 kDa inner membrane protein broadly expressed in the body (Sancak et al. 2013). EMRE knockdown or knockout totally abolishes mitochondrial Ca²⁺ uptake even when the MCU is overexpressed. Thus, it is deemed essential for the Ca²⁺ channelling activity of the MCU complex in live cells. How the EMRE controls MCU function is not clear. It has been suggested that the EMRE may facilitate the interaction between the MICU1/2 dimer and the MCU (Sancak et al. 2013). However, this is not supported by results showing that MICU1 still regulated MCU activity in lipid bilayers in the absence of the EMRE (Patron et al. 2014). Nevertheless, lack of the EMRE leads to a smaller sized MCU complex, as revealed on blue-native gels (Sancak et al. 2013), and therefore the EMRE probably helps with the assembly of the MCU complex. On the other hand, EMRE protein stability is dependent on the MCU, because MCU knockout lowers EMRE levels. Recent studies showed that the EMRE may also serve as the gatekeeper of the MCU complex (Vais et al. 2016), and its protein level is controlled by mitochondrial proteases (Tsai et al. 2017). Interestingly, in lower organisms where the MCU and MICU1 are highly conserved, such as plants, fungi and protozoa, EMRE homologues have not been found and thus are dispensable for functional MCUs (Kovács-Bogdán et al. 2014).

MCUR1 is a putative regulator of the MCU complex

In a study that screened 45 mitochondrial membrane proteins in HEK293T cells, MCU regulator 1 (MCUR1, encoded by CCDC90A) was found to be a strong regulator of mitochondrial Ca²⁺ uptake (Mallilankaraman et al. 2012a). MCUR1 is a 40 kDa protein of the inner mitochondrial membrane with the major part located in the matrix. It has a predicted transmembrane domain, one coiled-coil region, and an N-terminus located in the intermembrane space. MCUR1 silencing decreases basal matrix Ca²⁺ and inhibits agonist-induced mitochondrial Ca²⁺ uptake. MCUR1 can beco-immunoprecipitated with the MCU but not with MICUs, indicating different forms or combinations of the MCU complex (Mallilankaraman et al. 2012a). A recent report has questioned the role of MCUR1 in MCU regulation by showing that silencing MCUR1 dissipates mitochondrial membrane potential and decreases Complex IV assembly and activity (Paupe et al. 2015). Thus, whether MCUR1 can directly regulate MCU activity or indirectly regulate mitochondrial Ca²⁺ through the membrane potential needs to be further studied. The same study that identified MCUR1 found that SLC25A23 moderately impacts mitochondrial Ca²⁺ uptake (Mallilankaraman et al. 2012a). Later, it was found to interact with the MCU and promote mitochondrial Ca²⁺ uptake (Hoffman et al. 2014).

MCU deficiency in mice does not cause an energy crisis

Before the MCU and its associated proteins were discovered, mitochondrial Ca²⁺ uptake was linked to fundamental cellular processes, including energy metabolism, cytosolic Ca²⁺ buffering, cell secretion, survival, proliferation, migration and death. Genetic manipulations of MCU complex proteins in cells have largely confirmed these effects in vitro. For instance, the MCU mediates substrate-induced ATP generation (Tarasov et al. 2012) and regulates cell secretion (Alam et al. 2012; Quan et al. 2015) and migration (Tang et al. 2015). The precise molecular mechanism by which the gradually increased mitochondrial Ca²⁺ via the MCU activates metabolic enzymes, stimulates ETC activity, and promotes ATP synthesis remains unclear. Whether MCU activity is needed for basal mitochondrial ATP synthesis is also not known. Manipulating MCU expression modulates the magnitude of cytosolic Ca²⁺ elevation (De Stefani et al. 2011; O-Uchi et al. 2014; Lee et al. 2015) or store-operated Ca²⁺ entry (Deak et al. 2014; Samanta et al. 2014; Tang et al. 2015) in cell lines. In neonatal cardiomyocytes, MCU overexpression or siRNA silencing either reduces or enhances cytosolic Ca²⁺ transients during spontaneous contractions, respectively (Drago et al. 2012). Whether MCU-mediated mitochondrial Ca²⁺ uptake buffers systolic Ca²⁺ in adult cardiomyocytes is currently debated (Drago et al. 2012; Williams et al. 2013). Finally, manipulating MCU complex proteins in cells impacts reactive oxygen species (ROS) production, mPTP opening and apoptosis (Mallilankaraman et al. 2012b; Hoffman et al. 2013; Hou et al. 2013; Hoffman et al. 2014; O-Uchi et al. 2014).

Despite the above in vitro evidence supporting critical roles of the MCU in cellular energetics and other functions, surprisingly, several in vivo MCU-deficient models show little evidence of an energy crisis and adverse phenotypes at baseline when MCU-mediated mitochondrial Ca²⁺ uptake is abolished (Table 1). The first genetic model of the MCU is a pan-tissue germline MCU knockout in an outbred mouse strain (CD1 and C57BL/6), which yields live MCU knockout (MCU-KO) mice with a lower birth rate (Pan et al. 2013; Murphy et al. 2014). The same group also tried to use an inbred strain (C57BL/6). But germline MCU deletion in the inbred strain is embryonically lethal (Murphy et al. 2014). The results suggest that although MCU may play a critical role in embryonic development, its role in the development of some strains of mice is dispensable. The first heart-specific MCU-deficient model was generated in mice from the same mixed background by overexpressing a dominant-negative MCU isoform (MCU D260Q, E263Q, DN-MCU) driven by a cardiomyocyte-specific α-MHC promoter (Hofmann etal. 1989). Unlike the MCU-KO mice, the DN-MCU mice were born in Mendelian ratios (Rasmussen et al. 2015). The third model was generated in mice with a C57BL/6 background by floxing exons 5 and 6 of the MCU gene and expressing the tamoxifen-inducible Cre recombinase driven by the α-MHC promoter (Kwong et al. 2015; Luongo et al. 2015). Upon induction, the MCU protein level decreased 80% in the adult mouse heart (Luongo et al. 2015).

Table 1.

Major characteristics of the 3 MCU-deficient mouse models

	MCU-KO	DN-MCU	Heart MCU-KO
Basic characteristics
• Strain/background	CD1/BL6	CD1/BL6	C57BL/6
• Gene manipulation	Germline KO	Overexpression	Induced (8–10 weeks)
• Birth rate	↓	Normal	N/A
• Lifespan	Normal	Normal	Viable at 1 year
• Mitochondrial morphology	Normal	Normal	Normal (1 year)
• General phenotypes	Smaller size	Normal	Normal (1 year)
Ca2+ handling proteins
• MCU	Abolished	↑ (protein)	↓ 80% in heart
• MCUb	N/A	N/A	− (after 1 week)
• MICU1	N/A	↓ (mRNA)	− (after 1 week)
• EMRE	↓ (protein)	N/A	N/A
• NCLX	N/A	N/A	↓ (after 6 weeks)
Mitochondrial Ca2+
• Basal	↓ (a)	N/A	− (a,b)
• Uptake	Abolished (a,b,c)	Abolished (a,b)	Abolished (a,b)
Cytosolic Ca2+
• Basal	N/A	↑	–
• Ca2+ transient	–	↑	− (basal), ↑ (ISO)
SR Ca2+	N/A	↓	–
Bioenergetics
• PDH activity	↓	↓	↓
• NADH level (basal)	N/A	N/A	− (c)
• NADH level (stimulated)	N/A	↓ to ISO (c)	↓ to ISO (c)
• ATP level (basal)	− (d)	N/A	N/A
• ATP synthesis	N/A	↓ (b)	− (basal), ↓ (ISO)
• OCR (state 4/basal)	− (c,d)	− (a,b), ↑ (d)	− (c)
• OCR (state 3)	− (c)	− (a), ↓ (b)	− (a), ↓ (b)
• OCR (stimulated)	↓ to Ca2+ (a)	↑ to pacing (d)	↓ to Ca2+/ISO (a,c)
Heart Function
• Heart weight	Normal	Normal	Normal
• Heart rate	Normal	No ↑ to ISO	Normal
• Contractility	Normal	Normal	Normal (1 year)
• ↑ Pacing frequency	N/A	Lack of response	N/A
Stress response
• Acute β-AR stress	No effect	Impaired	Impaired
• Pressure overload	No effect	N/A	No effect
• I/R injury	Not protected	Not protected	Protected (in vivo)
Other findings	Skeletal muscle function ↓	Transcriptional reprogramming	Acute exercise performance ↓

−, ↓, or ↑: no change, decreased, or increased compared to wild type. N/A, not available; a, isolated mitochondria; b, permeabilized cardiomyocytes; c, intact cardiomyocytes; d, whole heart or whole body. ISO, isoproterenol.

Regardless of the approach, all three genetic models of MCU deficiency yielded live animals with normal or mild phenotypes (Table 1). Specifically, germline MCU-KO or DN-MCU mice lived to adulthood with a normal lifespan. The heart-specific conditional MCU-KO mice were normal at 1 year of age (induced at 8–12 weeks) (Luongo et al. 2015). In all of these mouse models, the function of the adult heart (e.g. heart weight, heart rate, ejection fraction, fractional shortening, stroke volume, chamber size) did not differ compared to the wild type controls under resting conditions. The normal phenotype in these mice contrasted sharply to the profound impact of MCU deficiency on mitochondrial Ca²⁺ uptake. In isolated mitochondria or permeabilized cardiomyocytes from the three MCU-deficient mice, increasing extra-mitochondrial Ca²⁺ to sub-millimolar levels failed to induce observable mitochondrial Ca²⁺ uptake. The basal matrix Ca²⁺ level was lower in MCU-KO mitochondria and unchanged in heart-specific MCU-KO mitochondria. Cytosolic Ca²⁺ handling was not changed in heart-specific MCU-KO cardiomyocytes (Luongo et al. 2015). In DN-MCU cardiomyocytes, the diastolic Ca²⁺ in the cytosol was higher, suggesting a loss of mitochondrial buffering which leads to cytosolic Ca²⁺ accumulation in cardiomyocytes (Rasmussen et al. 2015). The impact of MCU deficiency on mitochondrial or cytosolic Ca²⁺ handling in vitro cannot explain why the heart function of the genetic models was normal and also raises the question of the exact role of MCU-mediated Ca²⁺ uptake in the heart.

To answer this question, cardiac bioenergetics was evaluated in the three MCU-deficient models. Ca²⁺ is known to activate metabolic enzymes in mitochondria. Indeed, in all three models, the phosphorylation level of pyruvate dehydrogenase (PDH) was increased and PDH activity decreased (Table 1), which is consistent with the notion that mitochondrial Ca²⁺ is a key regulator of mitochondrial metabolism. Thus, it is logical to assume that lack of mitochondrial Ca²⁺ uptake should compromise the TCA cycle, downstream ETC activity, and ATP production. However, the results did not support significant impairment in mitochondrial energy metabolism in these three MCU-deficient models. First, the direct product of the TCA cycle, NADH was unchanged in the pan-tissue or heart-specific MCU-KO mice, suggesting that decreased PDH activity does not lead to an impaired TCA cycle. Second, in the MCU-KO mouse heart, oxygen consumption rate (OCR) and ATP levels were unchanged at baseline, demonstrating preserved ETC activity and energy production (Pan et al. 2013). The same results were reported in the heart-specific MCU-KO model (Luongo et al. 2015). These results are consistent with and may underlie the normal cardiac function found in these two models. As for the DN-MCU mice, OCR measured in isolated perfused heart was increased, while in permeabilized cells or isolated mitochondria it slightly decreased (State 3) or remained unchanged (State 4) (Rasmussen et al. 2015). Despite the unchanged or even increased OCR, ATP levels measured in heart issues were significantly lower in DN-MCU mice (Rasmussen et al. 2015; Wu et al. 2015). Taken together, results from the two MCU-KO models both showed that the TCA cycle, ETC activity and ATP synthesis were all preserved despite a decrease in PDH activity (Table 1). The decrease in ATP level seen in the DN-MCU model was not observed in the two MCU-KO models and did not align with the conserved OCR and contractility found in the DN-MCU mouse heart.

The lack of significant functional and bioenergetics phenotypes in these MCU-deficient models under basal conditions was an unexpected result. However, when acute and chronic stresses were applied to the hearts of these mice, different results were obtained (Table 1). First, acute β-adrenergic receptor (β-AR) stimulation was done on all three models to simulate the activation of ‘fight-or-flight’ response. While the MCU-KO mouse heart had a normal response to isoproterenol (ISO), as indicated by an increase in cardiac output (Holmstrom et al. 2015), the DN-MCU heart response to ISO was impaired, showing a smaller increase in heart rate and in the kinetics of contraction and relaxation (Wu et al. 2015). The heart-specific MCU-KO mice also responded to ISO or dobutamine treatment with a lower increase in contractile kinetics and a normal increase in heart rate (Kwong et al. 2015; Luongo et al. 2015). Intriguingly, the fall in cardiac function in MCU-KO mice after ISO treatment gradually reverted over approximately one hour to the levels of the control mice. The authors suggested that the MCU is specialized for acute matching of mitochondrial energy output to cardiac metabolic demands but that long-term Ca²⁺ homeostasis can be achieved through other influx pathways. Some bioenergetics parameters have been reported in the DN-MCU and heart-specific MCU-KO mice after acute β-AR stimulation, which showed that ISO-induced OCR increase, NADH accumulation, or ATP increase was missing in the MCU deficient cardiomyocytes (Kwong et al. 2015; Luongo et al. 2015; Rasmussen et al. 2015). In the heart or pacemaker cells of the DN-MCU mice, ATP dialysis rescued the heart rate response to ISO as well as increasing cytosolic Ca²⁺ (Wu et al. 2015). Finally, transverse aortic constriction (TAC) surgery was performed on the MCU-KO and heart-specific MCU-KO mice to generate chronic pressure overload, which is a widely used heart failure model. The same cardiac function decline was observed in these mice compared to their wild type controls after the TAC (Holmstrom et al. 2015; Luongo et al. 2015). In summary, these data suggest that mitochondrial Ca²⁺ uptake through MCU may play a role in responding to acutely increased workload but not to chronic stress of the heart.

Finally, mitochondrial Ca²⁺ overload has long been regarded as a key factor for cardiac I/R injury (Crow et al. 2004). In vivo or ex vivo I/R injury studies were carried out on the three MCU-deficient mouse models (Table 1). In an ex vivo Langendorff perfusion set-up, germline MCU-KO hearts showed no sign of protection compared to wild type controls, based on recordings of similar infarct size and similar impairment in contractile function (Pan et al. 2013). Interestingly, cyclosporine A (CsA), which inhibits Ca²⁺-dependent mPTP opening and cell death, protected the wild type hearts but not the MCU-KO hearts. The authors speculated that mPTP-independent death pathways may be activated in the absence of the MCU (Pan et al. 2013). Similar to the germline MCU-KO model, the DN-MCU hearts were not protected against I/R injury (Rasmussen et al. 2015). However, the heart-specific MCU-KO mice were protected against I/R injury, as shown by markedly improved contractile function, less myocardium damage, and smaller infarct size (Kwong et al. 2015; Luongo et al. 2015). The protective effects of heart-specific MCU-KO on I/R injury can be attributed to the blockade of the mPTP during reperfusion. The reason for the discrepancy between the germline MCU-KO and heart-specific MCU-KO mice responses to I/R injury remains unknown.

Potential explanations on the lack of energy crisis in MCU-deficient mice

As outlined above, the three MCU-deficient mouse models showed normal cardiac energetics and function at baseline. These findings are incomprehensible and contrast sharply with current knowledge of the role of the MCU or mitochondrial Ca²⁺ in ECB coupling. If the MCU is the sole or major Ca²⁺ influx mechanism for mitochondria, then MCU knockout is expected to be embryonically lethal or at least to cause severe dysfunction in the heart irrespective of the genetic background of the mice or the developmental stage. The unexpected normal phenotypes of MCU-deficient models teach us a lesson – that biological systems are so complex that a single molecule or factor is not able to determine the fate of the organism or even influence a functional outcome. Especially in the heart, which experiences huge energy fluxes in a beat-to-beat manner, it is reasonable to speculate that multiple mechanisms or regulators are in place to shape the overall landscape of cardiac ECB coupling and a single factor is either dispensable or only plays a fine-tuning role. An analogue of the MCU-deficient model is the heart-specific Complex I-deficient mouse model, in which a 70% decline in Complex I activity does not lead to a defective pumping function or an energy crisis in the heart under resting conditions (Karamanlidis et al. 2013). In the following sections, we attempt to provide some explanations of why viable mice lacking MCU activity do not suffer an energy crisis (Fig. 1).

Figure 1. Diagram showing the major mitochondrial Ca²⁺ handling processes (influx and efflux) and the role of intra- and extra-mitochondrial Ca²⁺ in regulating cardiac bioenergetics.

C, cytochrome C; CypD, cyclophilin D; I–V, the five complexes of electron transport chain; LCC, L-type Ca²⁺ channel; IMM, inner mitochondrial membrane; IMS, intermitochondrial membrane space; OMM, outer mitochondrial membrane; P_i, phosphate; Q, co-enzyme Q. Please refer to the text for the other abbreviations.

MCU-mediated mitochondrial Ca²⁺ may have limited ability to set ECB coupling

To understand how important mitochondrial Ca²⁺ is for cardiac ECB coupling, the first question we should ask is whether mitochondria take up Ca²⁺ during each heartbeat. So far, there is no clear answer to this question (Dedkova & Blatter, 2013). As summarized in recent reviews, early studies reported either beat-to-beat oscillations of mitochondrial Ca²⁺ or the lack of them (Hüser et al. 2000; Dedkova & Blatter, 2013; Williams et al. 2013). One limitation of these studies, particularly those carried out in adult cardiomyocytes, was the use of small fluorescent Ca²⁺ indicators that may not be exclusively located in the mitochondrial matrix. More recently, mitochondria-targeted fluorescent protein-based genetic Ca²⁺ indicators have been developed. Using one of these indicators, a study showed beat-to-beat mitochondrial Ca²⁺ transients at very low pacing frequencies (0.1–0.2 Hz) in adult rabbit cardiomyocytes. However, at a higher pacing frequency (0.5 Hz, which equates to 30 beats per minute) the transients were replaced by tonic increases in mitochondrial Ca²⁺ (Lu et al. 2013). In another recent study using the Forster resonance energy transfer (FRET)-based calcium indicator, MitoCam, the authors reported beat-to-beat fluctuations of mitochondrial Ca²⁺ at a pacing frequency of up to 4 Hz in adult rat cardiomyocytes (Wust et al. 2017). Interestingly, both studies showed that diastolic Ca²⁺ levels in mitochondria increased gradually during EC coupling. These findings raised other questions: whether the accumulated matrix Ca²⁺ during EC coupling is eventually released, and if so, how? Ca²⁺ fluxes through mitochondrial membranes need to be balanced so that Ca²⁺ cycling can be sustainable. This is the same rationale as has been applied to cytosolic Ca²⁺ cycling in the beating heart, in which Ca²⁺ influx through L-type Ca²⁺ channels and release via RyR2 must be matched by Ca²⁺ removal via the Na⁺-Ca²⁺ exchanger (NCX) on the cell membrane and the SR/ER Ca²⁺-ATPase (SERCA) (Eisner et al. 2017). Balanced Ca²⁺ fluxes are the key to maintaining a stable diastolic Ca²⁺ level and allow the next cycle of Ca²⁺ transients to happen. Perturbations of any of the influx/efflux mechanisms will lead to imbalanced flux, which, if not corrected, will lead to dysfunction (Eisner et al. 2017). In this regard, we have reported that mitochondria-accumulated Ca²⁺ during EC coupling can trigger transient mPTP openings, which is one of the mechanisms besides NCLX for matrix Ca²⁺ efflux and resetting of mitochondrial energetics (Gong et al. 2014). Future studies are needed to determine whether this mechanism applies to in vivo conditions and whether other efflux mechanisms are involved in balancing mitochondrial Ca²⁺ cycling.

If mitochondria do take up or accumulate Ca²⁺ during EC coupling, is the amount of Ca²⁺ influx significantly high? Accurate measurement of mitochondrial Ca²⁺ concentration is challenging due to the lack of a quantitative approach and the fact that most of the Ca²⁺ is bound and thus not detectable. The studies that have monitored mitochondrial Ca²⁺ during EC coupling mostly show slow, small amplitude and tonic increases in matrix Ca²⁺ during electrical pacing (Hüser et al. 2000; Lu et al. 2013; Williams et al. 2013; Gong et al. 2014). Alternatively, monitoring the consequence of blocking mitochondrial Ca²⁺ uptake on cytosolic Ca²⁺ levels is frequently used to indirectly evaluate the amount and significance of mitochondrial Ca²⁺ uptake during EC coupling. Previous studies have shown that mitochondria play little role in cytosolic Ca²⁺ removal (Bers et al. 1993; Negretti et al. 1993). The MCU inhibitor Ru360 has no effect on the amplitude of the cytosolic Ca²⁺ transient, suggesting that net flux of Ca²⁺ into mitochondria is small. As mentioned above, the heart-specific MCU-KO cardiomyocytes exhibit no change (increase or decrease) in diastolic and systolic Ca²⁺ levels, Ca²⁺ sparks and SR Ca²⁺ load, confirming that mitochondrial Ca²⁺ uptake is small and does not significantly impact cytosolic Ca²⁺ homeostasis (Kwong et al. 2015). It has been estimated that only ~1% of the Ca²⁺ released during EC coupling may be taken up by mitochondria (Williams et al. 2013). Since the estimated mitochondrial free Ca²⁺ is close to resting cytosolic Ca²⁺ levels (Boyman et al. 2014), this amount of Ca²⁺ influx would be insignificant or have only a very mild effect on mitochondria. On the other hand, some reports suggest that mitochondria could take up a significant amount of Ca²⁺ during EC coupling. Knockdown of the MCU in neonatal cardiomyocytes has been shown to increase cytosolic Ca²⁺ transients by 50–60% (Drago et al. 2012) and cardiomyocytes from DN-MCU mice have significantly increased diastolic and systolic Ca²⁺ (Rasmussen et al. 2015). However, these studies found no compensatory increase in SR Ca²⁺ uptake or deranged ECB coupling and contractility. These results did not agree with the concept of balanced intracellular Ca²⁺ fluxes. This is because, if huge amounts of Ca²⁺ are not buffered by mitochondria, they should be quickly removed by counterbalancing mechanisms (e.g. Ca²⁺ enters the SR or moves out of the cell), otherwise detrimental outcomes will be inevitable. Taken together, current experimental evidence suggests that mitochondria may or may not take up a significant amount of Ca²⁺ during EC coupling. Another point worth mentioning is potential species differences in mitochondrial Ca²⁺ uptake during EC coupling. In rabbit and guinea-pig cardiomyocytes, beat-to-beat oscillations of mitochondrial Ca²⁺ concentrations have been recorded (Chacon et al. 1996; Maack et al. 2006). However, these studies used rhod-2 and Fluo-3 to monitor the mitochondrial Ca²⁺ transients and pharmacological agents to inhibit the MCU, which may result in non-specific effects. Future application of mitochondria-targeted Ca²⁺ probes and generation of MCU knockouts in these species will be useful to resolve this issue.

Finally, MCU-mediated mitochondrial Ca²⁺ may play a more important role in the acute stress response rather than regulating baseline ECB coupling under physiological conditions. It has been proposed that prolonged and higher levels of cytosolic Ca²⁺ may lead to much higher (e.g. 10- to 1000-fold) mitochondrial Ca²⁺ uptake, which may reshape cytosolic Ca²⁺ dynamics, mitochondrial ECB coupling, and cardiac function (Williams et al. 2013). This notion is supported by the results from the three MCU-deficient mouse models. First, OCR at baseline is not changed despite the lower PDH activity in all three MCU-deficient models. Together with the unchanged baseline cardiac contractility, these findings imply that manipulating PDH activity may not be sufficient to impact baseline ECB coupling in the heart. Since the normal heart mostly relies on fatty acids for energy, it is possible that perturbations in glucose metabolism would be well tolerated. It would therefore be interesting to determine whether substrate metabolism in the MCU deficient mice models is altered. In contrast, when energy demand is increased by acute β-AR stimulation, OCR increases in wild type but not MCU deficient cardiomyocytes (Luongo et al. 2015), suggesting that MCU-mediated Ca²⁺ uptake is needed to acutely boost metabolism and energy production. Indeed, β-AR stimulation is known to enhance ECB coupling and contractility, which is compromised by MCU-KO in the heart (Luongo et al. 2015). However, under chronic stress conditions, such as pressure overload, MCU-mediated mitochondrial Ca²⁺ seems not to play a key role. Chronic stresses are known to increase diastolic Ca²⁺ in the cytosol, which may lead to mitochondrial Ca²⁺ overload (Santulli et al. 2015). An explanation of the lack of effect of MCU deficiency in chronic heart failure could be that chronic stress elevates cytosolic Ca²⁺ slowly or mildly, so that it does not reach the threshold for activating MCU-mediated Ca²⁺ uptake.

Other mitochondrial Ca²⁺ handling mechanisms may regulate ECB coupling

As mentioned earlier, multiple mitochondrial Ca²⁺ uptake mechanisms were proposed to synergistically regulate mitochondrial Ca²⁺ homeostasis before the discovery of the molecular identity of the MCU. The MCU is thought to be the major Ca²⁺ channel but has a low Ca²⁺ sensitivity, meaning that it only functions at high cytosolic Ca²⁺ levels. Currently two findings support the function of the MCU under physiological conditions. First, the Ca²⁺ level at the SR-mitochondria junctions is high enough to trigger MCU-mediated Ca²⁺ influx (Rizzuto et al. 1998; Tubbs & Rieusset, 2017). Second, the MCU is located at these junctions and in close proximity to the SR Ca²⁺ release channel, RyR2, in cardiomyocytes (De La Fuente et al. 2016). However, the MCU-mediated mitochondrial Ca²⁺ uptake at these microdomains has not been measured during EC coupling. Furthermore, baseline mitochondrial Ca²⁺ is decreased but not abolished in the germline MCU-KO mouse or is unchanged in the heart-specific MCU-KO mouse. These results clearly point to the existence of other mitochondrial Ca²⁺ uptake pathways that may play an important role in keeping the physiological relevant mitochondrial Ca²⁺ homeostasis.

In addition to the MCU, which is activated by high levels of extra-mitochondrial Ca²⁺ and accounts for large amount of Ca²⁺ influx, RaM is another mechanism that is activated by lower extra-mitochondrial Ca²⁺ and is responsible for rapid but transient Ca²⁺ influx (Sparagna et al. 1995; Buntinas et al. 2001). Direct electrophysiological channel recordings have shown two Ca²⁺-selective currents in isolated mitoplasts that match the features of the MCU and RaM (Michels et al. 2009). The MCU and RaM modes of mitochondrial Ca²⁺ uptake have also been detected in isolated mitochondria through measurement of free matrix Ca²⁺. Findings have indicated that these two Ca²⁺ uptake mechanisms activate different intra-mitochondrial Ca²⁺-buffering systems and lead to significant differences in free matrix Ca²⁺ (Wei et al. 2012). More recently, three apparently different Ca²⁺ currents were recorded, of which two types appeared to result from RaM (e.g. low sensitivity to Ruthenium Red) (Bondarenko et al. 2013). However, since the reported studies on RaM were all carried out in isolated mitochondria, its involvement in regulating ECB coupling in intact cardiomyocytes remains elusive. It is not known whether RaM is generated by a new and unknown channel or represents a certain mode or regulation of the MCU complex. Future studies are needed to determine whether different regulation of the MCU complex can alter channel conductance and lead to diverse Ca²⁺ influx currents.

Another mitochondrial Ca²⁺ uptake mechanism in the heart is the skeletal muscle RyR1, which is located on the inner membrane of cardiac mitochondria and which transports Ca²⁺ into the matrix (Beutner et al. 2001, 2005). Three different RyR isoforms (RyR1–3) have been cloned, and their different physiological and pharmacological properties have been reported (Lanner et al. 2010). RyR2 is predominately located in the SR of cardiomyocytes and RyR1 is predominately located in skeletal muscle SR (Lanner et al. 2010). Researchers detected the mRNA and protein of RyR1 in cardiac tissue at low levels years ago (Munch et al. 2000; Jeyakumar et al. 2002), but its exact functional and physiological significance in the heart is still not fully understood. The Sheu lab, using multiple approaches, first reported that RyR1 is expressed at a low level on the inner membrane of cardiac mitochondria and plays a role in fast Ca²⁺ uptake (Beutner et al. 2001, 2005). The RyR1 in cardiac mitochondria was found to have similar biochemical, pharmacological and functional properties to skeletal muscleRyR1, butis differentfrom the RyR2 in cardiac SR. Therefore, cardiac RyR1 was termed mRyR1 (mitochondrial RyR1). The molecular identity of mRyR1 was carefully analysed and confirmed by a variety of functional and biochemical experiments using not only native heart but also knockout heart (RyR1 knockout mice) (Beutner et al. 2005). The electrophysiological, biophysical and pharmacological properties of mRyR1 were determined in lipid bilayers of mRyRs purified from the inner membrane of cardiac mitochondria (Altschafl et al. 2007) and in single channel recordings of cardiac mitoplasts using patch-clamp techniques (Ryu et al. 2011). Ryanodine was found to dose-dependently modulate mRyR1, stabilizing it at a subconductance state at low doses and blocking its opening at high doses (Ryu et al. 2011). Collectively, these results supported the conclusion that mRyR1 is another mitochondrial Ca²⁺ uptake channel in the heart. Unlike the MCU, the Ca²⁺ selectivity of mRyR1 is low and its conductance is high (Ryu et al. 2011). Thus, mRyR1 opening may dissipate mitochondrial membrane potential and halt energy metabolism. Given its low expression level, bell-shaped Ca²⁺ dependency, and rapid activation and inactivation profiles, it is believed that the opening of mRyR1 will only cause short-term or localized depolarization within a mitochondrion and will not adversely impact mitochondrial energetics. How mRyR1-mediated mitochondrial Ca²⁺ uptake contributes to physiological and pathological mitochondrial Ca²⁺ homeostasis and ECB coupling remains elusive. Because the genetic and molecular identity of RyR1 are well known, future studies with cardiac-specific inducible RyR1 knockout mice will help to define the physiological and pathological roles of mRyR1 in ECB coupling.

Mitochondrial Ca²⁺ is regulated not only by influx but also by efflux (Sekler, 2015). In the three MCU-deficient mouse models, some of the Ca²⁺ influx and efflux proteins were assessed. The other components of the MCU complex, such as MCUb, MICU1/2 and MCUR were mostly unchanged in the DN-MCU (mRNA) and heart-specific MCU-KO (protein) hearts. The EMRE was missing in germline MCU-KO hearts, consistent with the fact that EMRE binds to the MCU and regulates its function (Holmstrom et al. 2015). The protein level and activity of the mitochondrial Ca²⁺ efflux protein NCLX was measured only in the heart-specific MCU-KO mice. Interestingly, short-term MCU-KO (1–2 weeks) that mildly decreased the MCU protein level did not alter the NCLX protein level in the heart (Kwong et al. 2015). After a longer term (10 weeks) MCU-KO, both MCU and NCLX protein levels decreased significantly, and NCLX activity also decreased dramatically (Luongo et al. 2015). The decrease in NCLX protein levels and activity could be a compensatory mechanism to maintain balanced mitochondrial Ca²⁺ fluxes and may underlie the unchanged basal mitochondrial Ca²⁺ level. More recently, severely compromised phenotypes at baseline were reported in an inducible cardiac-specific NCLX-KO mouse model (Luongo et al. 2017). This phenotype contrasted sharply to the functionally and energetically uncompromised phenotypes of three MCU-deficient models. The NCLX-KO hearts developed acute heart failure accompanied by hypertrophy, fibrosis, ventricular dilatation and compromised left ventricular function (Luongo et al. 2017). Although these detrimental phenotypes in the NCLX-KO hearts can be attributed to decreased mitochondrial Ca²⁺ efflux, no significant change in baseline mitochondrial Ca²⁺ level, PDH phosphorylation and NADH/NAD⁺ ratio were found, indicating normal Ca²⁺ homeostasis and ECB coupling. The EMRE level was decreased in NCLX-KO hearts, but other MCU complex components were the same and mitochondrial Ca²⁺ uptake was not compromised. Therefore, it seems that deranged ECB coupling is not a key mechanism underlying the severe heart failure phenotype of the NCLX-KO mice. Perhaps delayed Ca²⁺ efflux triggers the mPTP and induces cell death, leading to heart dysfunction. Meanwhile, NCLX overexpression in the heart protects it from I/R injury by preventing mPTP opening rather than changing basal mitochondrial Ca²⁺ homeostasis (Luongo et al. 2017). Taken together, these findings indicate that NCLX may compensate for the lack of MCU, and NCLX-mediated mitochondrial Ca²⁺ efflux may be equally if not more important for maintaining mitochondrial Ca²⁺ homeostasis and heart function than MCU-mediated Ca²⁺ uptake.

Extra-mitochondrial Ca²⁺-mediated mechanisms in ECB coupling

The current paradigm for explaining how ECB coupling is regulated by Ca²⁺ in the heart focuses on the role of mitochondrial Ca²⁺ in activating the TCA cycle and ETC enzymes (Balaban, 2002). The MCU-deficient models, however, challenge this dogma by showing that a significant inactivation of mitochondrial Ca²⁺ uptake has little impact on mitochondrial energy metabolism in the beating heart under physiological conditions (Pan et al. 2013; Luongo et al. 2015). Therefore, additional pathways other than the regulation of metabolic enzymes by mitochondrial Ca²⁺ must exist to account for the lack of defective ECB coupling in MCU-deficient models. In this regard, extra-mitochondrial Ca²⁺ signals may be a key player in ECB coupling and mitochondrial metabolism. One of the potential mechanisms is mitochondrial dynamics, which refers to the changes in mitochondrial morphology including fission, fusion and movement (Chan, 2012; Friedman & Nunnari, 2014). Mitochondrial dynamics play a critical role in dictating mitochondrial function, and disturbances in mitochondrial dynamics lead to human disease, including cardiovascular disease (Friedman & Nunnari, 2014; Disatnik et al. 2015; Dorn, 2015; Roy et al. 2015). Mitochondrial dynamics are regulated by a group of proteins that is highly expressed in the heart and linked to mitochondrial energetics and intracellular Ca²⁺ regulation (Imoto et al. 1998; Santel et al. 2003). However, mitochondria in adult cardiomyocytes have been found to exhibit static and fragmented morphology with very low levels of dynamic changes. Moreover, genetic ablation of the fusion protein mitofusin 1/2 (Mfn1/2) or the fission protein dynamin related protein 1 (Drp1) was found to provoke only modest changes in mitochondrial size in contrast to significant energetics and functional outcomes in adult ventricular myocytes (Ikeda et al. 2015; Song et al. 2015; Zhang et al. 2017). These results strongly support non-canonical roles for the dynamics proteins in adult cardiomyocytes (Song & Dorn, 2015; Wang et al. 2018), which may regulate ECB coupling in the heart. For instance, Mfn2 can serve as a tether to connect the SR and mitochondria (de Brito & Scorrano, 2008; Dorn et al. 2015) for Ca²⁺ cycling or as a signalling protein for mitophagy (Song & Dorn, 2015). Drp1 is involved in mitophagy in the heart and maintains efficient ATP synthesis in adult neural stem cells (Kim et al. 2015; Song & Dorn, 2015). The inner membrane fusion protein optic atrophy 1 (Opa1) can directly modulate mitochondrial respiration through controlling the cristae structure of the inner mitochondrial membrane (Frezza et al. 2006; Cogliati et al. 2013).

We recently demonstrated that Drp1, which is activated by cytosolic Ca²⁺ and translocates to the outer mitochondrial membrane, modulates mitochondrial morphology and ECB coupling in adult cardiomyocytes (Zhang et al. 2017; Wang et al. 2018). Previously, it has been shown that thapsigargin, which inhibits SERCA, or KCl, which depolarizes membrane potential and induces Ca²⁺ influx through L-type Ca²⁺ channels, can increase cytosolic Ca²⁺ and promote Drp1 translocation to mitochondria in adult rat cardiomyocytes (Hom et al. 2010). More recently, we showed that Drp1 modulated the mitochondrial flash activity (Zhang et al. 2017), which is a biomarker for a variety of mitochondrial energetics (Gong et al. 2015), morphology (Breckwoldt et al. 2014), and other functions (Wang et al. 2008, 2016a; Shang et al. 2016). Furthermore, genetic (overexpression of dominant negative Drp1 K38A mutation or knockdown Drp1 with shRNA) or pharmacological (mitochondrial division inhibitor Mdivi-1; Cassidy-Stone et al. 2008) inhibition of endogenous Drp1 significantly decreases mitochondrial OCR in isolated mitochondria and permeabilized or intact adult cardiomyocytes. Meanwhile, mitochondrial ROS generation and ATP levels, measured using a new mitochondria-targeted ATP indicator, ATeam (Imamura et al. 2009), were lower when Drp1 activity was inhibited. Collectively, these results point to a key role for Drp1 in bridging EC coupling with mitochondrial bioenergetics under physiological conditions (Zhang et al. 2017; Wang et al. 2018). Consistently, studies have shown that mitochondrial respiration and energetics are also inhibited in the heart-specific Drp1-KO mice (Ikeda et al. 2015). How Drp1 regulates mitochondrial metabolism is not clear. A potential mechanism is through the physiologically relevant transient openings of the mPTP, which has been shown to stimulate mitochondrial respiration in adult cardiomyocytes (Zhang et al. 2017). Supporting evidence for this hypothesis includes the ability of Drp1 inhibition to suppress mPTP opening and the absence of an effect of Mdivi-1 on respiration in adult cardiomyocytes from cyclophilin D (an mPTP regulator) null mice (Zhang et al. 2017). The physiological role of Drp1 in regulating mitochondrial metabolism also applies in disease conditions. We have reported that chronic β-AR stimulation activates Drp1 through phosphorylation and that hyper-activation of Drp1 is responsible for excessive mPTP opening, cardiomyocyte death and heart hypertrophy (Xu et al. 2016). Several reports have shown that Drp1 activation during I/R leads to cardiomyocyte death and functional decline of the heart, and inhibiting Drp1 activity is protective (Ong et al. 2010; Ishikita et al. 2016). This effect is also attributed to the blockade of excessive mPTP openings during I/R. Future experiments to elucidate how Drp1 activity can be modulated by Ca²⁺ transients, caused by Ca²⁺ release from SR during the heart beats, may provide new insights regarding the involvement of Drp1 in ECB coupling.

Summary and perspectives

The current wave of mitochondrial research has led to exciting findings in mitochondria biology and physiology that may guide the development of new therapies for mitochondria-and metabolism-related diseases (Wang et al. 2016b). The discovery of the MCU and its associated proteins is one such breakthrough, opening a new frontier for studying the molecular mechanism of mitochondrial Ca²⁺ regulation. However, the MCU complex is unlikely to be the only mitochondrial Ca²⁺ uptake channel. It is also dispensable for the physiological regulation of mitochondrial bioenergetics in the heart and other energy-craving organs. The fact that deranged MCU-mediated mitochondrial Ca²⁺ uptake does not compromise mitochondrial metabolism and heart function indicates that extra-mitochondrial pathways play a role in cardiac ECB coupling. Nevertheless, MCU-mediated mitochondrial Ca²⁺ uptake serves as a rapid stress response to boost energy metabolism.

Future studies are warranted to identify other channels or proteins that control mitochondrial Ca²⁺ homeostasis at baseline and after stimulation. Particularly, research on the molecular identity of RaM and its potential role in regulating ECB coupling may help to explain the lack of an energy crisis in MCU-deficiency mice under resting conditions. Similarly, research is needed to determine whether mRyR1 plays a key role in baseline mitochondrial Ca²⁺ influx regulation and physiological ECB coupling. In addition, extra-mitochondrial Ca²⁺-dependent signalling pathways may work synergistically with intra-mitochondrial Ca²⁺-dependent signalling pathways in setting and adjusting ECB coupling in the heart. In this regard, cytosolic Ca²⁺ hotspots adjacent to mitochondria may be important in activating pathways that either boost intermediate metabolism (e.g. glycolysis or fatty acid metabolism) or trigger mitochondrial translocation of cytosolic proteins (e.g. Drp1) to indirectly stimulate mitochondrial metabolism. Finally, other metabolic agents (e.g. ADP, ATP, ROS, redox, TCA cycle enzymes and substrates) could stimulate or synergize with mitochondrial Ca²⁺ in regulating ECB coupling under normal or stress conditions.

Biographies

Pei Wang is a postdoctoral researcher in Wang Wang’s Lab. His research focuses on the role of mitochondria Ca²⁺ signalling in the heart disease.

Wang Wang is an Associate Professor at the Mitochondria and Metabolism Center, Department of Anesthesiology and Pain Medicine, University of Washington in Seattle. Dr Wang’s research is focused on identifying new molecular mechanisms regulating mitochondrial function and energy metabolism in the normal and diseased heart.

Shey-Shing Sheu is William Wikoff Smith Professor of Cardiovascular Research and Associate Director at Center for Translational Medicine at Thomas Jefferson University in Philadelphia, Pennsylvania. Dr Sheu’s research has concentrated on understanding the mechanisms underlying mitochondrial ATP, Ca²⁺, ROS and fission/fusion dynamics in the heart.