OUP user menu

Calcium release microdomains and mitochondria

Michael Kohlhaas, Christoph Maack
DOI: http://dx.doi.org/10.1093/cvr/cvt032 259-268 First published online: 15 February 2013


The processes of excitation–contraction (EC) coupling consume large amounts of energy that need to be replenished by oxidative phosphorylation in the mitochondria. Since Ca2+ activates key enzymes of the Krebs cycle in the mitochondrial matrix, it is important to understand the mechanisms and kinetics of mitochondrial Ca2+ uptake to delineate how in cardiac myocytes, energy supply is efficiently matched to demand. In recent years, the identification of various proteins involved in mitochondrial Ca2+ signalling and the tethering of mitochondria to the sarcoplasmic reticulum (SR) has considerably advanced the field and supported the concept of a mitochondrial Ca2+ microdomain, in which Ca2+ concentrations are high enough to overcome the low Ca2+ affinity of the principal mitochondrial Ca2+ uptake mechanism, the Ca2+ uniporter. Furthermore, defects in EC coupling that occur in heart failure disrupt SR-mitochondrial Ca2+ crosstalk and may cause energetic deficit and oxidative stress, both factors that are thought to be causally involved in the initiation and progression of the disease.

  • Calcium
  • Mitochondria
  • Microdomain
  • Redox
  • Heart failure

1. Introduction

Excitation–contraction (EC) coupling in cardiac myocytes consumes vast amounts of energy in the form of ATP that need to be efficiently replenished by oxidative phosphorylation in the mitochondria, which occupy roughly a third of cell volume. During a single heartbeat, ∼2% of the cellular ATP is consumed, and the whole ATP pool of cardiac myocytes is turned over within less than a minute.13 To orchestrate oxidative phosphorylation in response to constantly changing workloads of the heart, various factors tightly control the mitochondrial redox state and electron flux along the respiratory chain, securing constant availability of ATP. Two of the most important regulatory factors in this regard are Ca2+ and ADP,47 whose communication between different organelles and/or compartments within cardiac myocytes is organized by ‘microdomains’.8,9 Spatial regulation by microdomains is a particular requirement for signalling efficacy considering the highly organized architecture of cardiac myocytes, in which mitochondria comprise a third of the cell volume and are aligned regularly along the ATP-consuming myofilaments, tightly connected to the Ca2+ stores (i.e. the sarcoplasmic reticulum, SR) and strategically positioned in vicinity to the ‘dyads’, where activation of L-type Ca2+ channels (LTCC) triggers Ca2+ release from the junctional SR in a process coined ‘Ca2+-induced Ca2+ release’.

In the last couple of years, considerable progress has been made regarding the identification of the molecular nature and functional roles of various components involved in the transmission of Ca2+ between the SR and mitochondria. In particular, the identification of the molecular identities and functional roles of various proteins in the inner mitochondrial membrane (IMM) and outer mitochondrial membrane (OMM), but also the SR, has paved new avenues and spurred new enthusiasm in research on the mechanisms of mitochondrial Ca2+ uptake. Here, after giving a brief overview on EC coupling and mitochondrial energetics (for a more comprehensive review, see refs.5,10), we focus on these more recent findings with a special emphasis on the aspects relevant to cardiac myocytes and the pathophysiological changes that occur in chronic heart failure.

2. Excitation–contraction coupling

2.1 Physiology

During a cardiac action potential (AP), Ca2+ enters cardiac myocytes via LTCC, triggering an even greater release of Ca2+ from the SR via ryanodine receptors (RyR2), and this Ca2+ is available at the myofilaments to induce contraction. The amount of Ca2+ that entered the cell via LTCC is exported primarily via the sarcolemmal Na+/Ca2+ exchanger (NCX) during diastole, whereas the Ca2+ that was released from the SR is taken back up by the SR Ca2+ ATPase (SERCA). In human, rabbit and guinea-pig cardiac myocytes, the distribution between SR-derived Ca2+ and LTCC-triggered Ca2+ influx is ∼70/30, whereas in smaller animals (such as mice and rats), this is shifted towards an even higher contribution of the SR (∼90/10).10 The interplay between LTCC and RyR2 is facilitated by the close vicinity of both channels in the dyads between t-tubular sarcolemma and the junctional SR. The concept of such a Ca2+ microdomain is that for a brief period, the Ca2+ concentrations at the site of release (i.e. near the RyR2 or the LTCC) exceed the concentrations in the bulk cytosol of the cell by several orders of magnitude.8,11 These Ca2+ peaks are limited spatially and temporarily by diffusion of Ca2+ away from these ‘hot spots’ to the rest of the cytosol, where Ca2+ binds to myofilaments to induce contraction. A similar microdomain exists between the SR and mitochondria (Figure 1 with inset), facilitating an efficient Ca2+ transfer between both organelles to match the energy produced in the mitochondria to the demand generated by Ca2+-dependent processes of EC coupling. We will allude to this mitochondrial Ca2+ microdomain—the focus of this review—in more detail further below.

Figure 1

Mitochondrial Ca2+ signalling and its impact on mitochondrial energetics and redox state in health and disease. The Krebs cycle regenerates NADH required for oxidative phosphorylation at the electron transport chain, but also NADPH through isocitrate dehydrogenase (IDPm), malic enzyme (MEP) and nicotinamide nucleotide transhydrogenase (Nnt). Ca2+ stimulates three key enzymes of the Krebs cycle and thus, regulates NADH and NADPH regeneration. In heart failure, defects in EC coupling and mitochondrial ion homoeostasis (indicated by red arrows) contribute to energetic mismatch and oxidative stress. ΔΨm, mitochondrial membrane potential; Mn-SOD, Mn2+-dependent superoxide dismutase; PRX, peroxiredoxin; GPX, glutathione peroxidase; TRXr/o, reduced/oxidized thioredoxin; GSH/GSSG, reduced/oxidized glutathione; TR, thioredoxin reductase; GR, glutathione reductase; α-KG, α-ketoglutarate; MCU, mitochondrial Ca2+ uniporter; NCLX, mitochondrial Na+/Ca2+ (and Li+) exchanger; po, open probability; RyR, ryanodine receptor; SERCA, SR Ca2+ ATPase; Mfn, mitofusin. Inset: detailed aspects of a mitochondrial Ca2+ microdomain. MICU1, mitochondrial Ca2+ uptake 1; mNHE, mitochondrial Na+/H+ exchanger; IMM and OMM, inner and outer mitochondrial membranes; VDAC, voltage-dependent anion channel.

2.2 Pathophysiology

In chronic heart failure, defects of EC coupling underlie systolic and diastolic dysfunction.5,12,13 A central defect in failing cardiac myocytes is the decreased Ca2+ load of the SR, which is related to decreased SERCA activity and a leak of the RyR2. Furthermore, the three-dimensional t-tubular structure that provides efficient coupling of the membrane potential to triggering SR Ca2+ release is disturbed in various models of heart failure, which induces spatio-temporal dyssynchrony of cytosolic Ca2+ transients.1416 Initially considered as an adaptive mechanism, the cytosolic Na+-concentration ([Na+]i) is elevated in failing cardiac myocytes, presumably related to an activation of the ‘late Na+ current’ (late INa) and/or increased activity of the Na+/H+ exchanger.1721 This elevation of [Na+]i, in particular in combination with up-regulation of the NCX protein, favours the reverse-mode of the NCX to import Ca2+ during the AP and to decrease Ca2+-efflux during diastole.17,19,2225 Thus, it is generally perceived that this elevation in [Na+]i partly compensates for decreased SR Ca2+ load and improves contractility in heart failure.12,2227 In this review, we will also discuss how the elevation of [Na+]i has a negative impact on mitochondrial Ca2+ uptake and redox state.

A principal mechanism to regulate EC coupling is β-adrenergic stimulation, which signals through phosphorylation of Ca2+ handling proteins mediated by protein kinase A (PKA) and Ca2+/Calmodulin-dependent protein kinase II (CaMKII).28,29 In heart failure, chronic β-adrenergic stimulation with subsequent dysregulation of CaMKII- and PKA-regulated phosphorylation of target proteins substantially contributes to the maladaptive changes of EC coupling (e.g. the leak of RyR2 and activation of late INa).2830 Furthermore, an energetic deficit31,32 and mitochondrial oxidative stress33,34 are thought to play causative roles in the progression of heart failure. Over the past couple of years, we developed the concept that based on the tight interplay between EC coupling, mitochondrial redox state, and the regulation of reactive oxygen species (ROS) production, defects in EC coupling play a causal role for the development of energetic deficit and oxidative stress in heart failure (Figure 1).35

3. Regulation of mitochondrial energetics and redox state during EC coupling

3.1 Matching energy supply and demand

In the mitochondrial matrix, the Krebs cycle produces NADH and FADH2, which fuel electrons into the electron transport chain (ETC) through complexes I and II (Figure 1). This induces sequential redox reactions along the complexes of the ETC that promote the translocation of protons (H+) across the IMM, creating a proton gradient (ΔpH) and an electrical gradient (DYm) which constitute the proton motive force (ΔµH) which is utilized by the F1Fo-ATP synthase to regenerate ATP from ADP. In vivo, cardiac workload is increased especially by β-adrenergic stimulation, increasing the rate and amplitude of cytosolic Ca2+ transients, which in turn accelerates ATP consumption by Ca2+-cycling proteins and the myofilaments. This pronounced ATP hydrolysis elevates the formation of ADP, stimulating ATP regeneration at the F1Fo-ATP synthase. Since the F1Fo-ATP synthase reaction is coupled to ΔµH, ATP production principally dissipates ΔµH. To maintain ΔµH, NADH donates more electrons to the ETC, which oxidizes NADH to NAD+ (and FADH2 to FAD). Since β-adrenergic stimulation elevates the rate and frequency of cytosolic Ca2+ transients, mitochondria accumulate Ca2+ through several uptake mechanisms,36,37 but in particular, the mitochondrial Ca2+ uniporter (MCU; Figure 1).38,39 In the mitochondrial matrix, Ca2+ stimulates key enzymes of the Krebs cycle to accelerate the regeneration of oxidized NAD+ and FAD to reduced NADH and FADH2.4,7,4046 Furthermore, Ca2+-induced stimulation of the ETC, the F1Fo-ATP synthase and the aspartate–glutamate shuttle further contribute to an acceleration of oxidative phosphorylation (for a recent review, see Glancy and Balaban47). Thus, Ca2+ and ADP regulate oxidative phosphorylation in a complimentary manner and secure relatively constant ratios of ATP/ADP,48,49 and constant or even increasing ratios of NADH/NAD+50 during physiological variations of cardiac workload.

This complex interplay between ADP- and Ca2+-induced regulation of oxidative phosphorylation and the mitochondrial redox state has been revealed by experiments in which abrupt changes of workload were provoked in isolated cardiac trabeculae,4,4042 cardiac myocytes43,44 and/or computational modelling.7,45,46 During more gradual escalations of workload, as they occur during β-adrenergic stimulation in vivo, these transitions of energetic intermediates are presumably less accentuated, reflecting the high efficiency of regulatory control mechanisms (i.e. Ca2+ and ADP) that quickly adapt the supply to the increased energy demand. Thus, given the central role of Ca2+ for regulating mitochondrial energetics, the mechanisms and kinetics of mitochondrial Ca2+ uptake are of utmost importance for our understanding of energy supply-and-demand matching in the heart.

3.2 Role of redox state for mitochondrial reactive oxygen species formation

In addition to its role for energy production, mitochondrial Ca2+ uptake also plays a key role in governing mitochondrial formation of ROS. It has been initially proposed that ∼2% of O2 consumption is aberrantly funnelled to superoxide (O2−) production, however, these initial estimates were corrected to levels of ∼0.2% more recently (for review, see Balaban et al.51). The major site of O2− formation are complexes I and III of the ETC (Figure 1),51,52 and O2−formation is highest when substrates (glucose, fatty acids) are available, but energy (and thus, O2) consumption are low, i.e. a condition that is experimentally defined as ‘state 4 respiration’. In this situation, which typically does not occur in the heart where energy is constantly consumed, the ETC is highly reduced, and electrons are more likely to aberrantly ‘slip’ to O2 to produce O2−.51,53 When workload increases, ADP-induced acceleration of respiration oxidizes the ETC by favouring the enzymatic reduction of O2 to H2O at complex IV (cytochrome oxidase), during which O2− formation is efficiently avoided. At the same time, aberrant O2− formation at complexes I and III is diminished.51

Besides the ETC, other mechanisms of mitochondrial ROS production have been proposed. Dedkova and Blatter54 revealed that an uncoupled nNOS located in the mitochondria is a source for O2− which can induce mPTP opening and cell death under pathological conditions of mitochondrial Ca2+ overload. The existence of a mitochondrial NOS, however, is still controversial (for review, see Brookes55). Furthermore, translocation of MAO56 and NAPDH oxidase 4 (Nox4)57,58 to mitochondria may contribute to mitochondrial oxidative stress. The respective contribution of these various potential sources to overall ROS production, however, requires further evaluation.

To protect the matrix enzymes and mitochondrial DNA from oxidative damage, antioxidative mechanisms are in place to eliminate ROS.51,59 O2− is rapidly transformed to H2O2 by superoxide dismutase (SOD), and H2O2 is detoxified by glutathione peroxidase (GPX) and peroxiredoxin (PRX; Figure 1). These H2O2-eliminating enzymes need to be regenerated by NADPH-dependent enzymes, such as glutathione reductase (GR) and thioredoxin reductase (TR). This assigns NADPH a central role in the terminal elimination of mitochondrial ROS.59 NADPH is regenerated by three enzymes that all derive their substrates from products of the Krebs cycle, i.e. isocitrate dehydrogenase (IDPm), malic enzyme and the nicotinamide nucleotide transhydrogenase (Nnt; Figure 1).59 Since Ca2+ activates three rate-limiting enzymes of the Krebs cycle, we recently discovered that mitochondrial Ca2+ uptake is not only important to match energy supply to demand, but also to keep the redox state of NADPH in a reduced state to prevent overflow of H2O2.44 During β-adrenergic stimulation, transient oxidation of NAD(P)H in response to ADP-induced acceleration of respiration was associated with a transient increase in mitochondrial H2O2 formation.44 When blocking mitochondrial Ca2+ uptake with the MCU-inhibitor Ru360, the Ca2+-induced (and Krebs cycle mediated) regeneration of NAD(P)H was blunted and the formation of H2O2 potentiated.44

Since RyRs are regulated by the cellular redox state,60 a recent study by Liu and co-workers61 revealed that variations of the mitochondrial NAD(P)H redox state and H2O2 production correlate with RyR2 activity, indexed by the frequency of elementary Ca2+ release events, i.e. Ca2+ sparks. These data were in agreement with a previous study that observed a bidirectional regulation of spark frequency by mitochondrial ROS production.62 In this latter study, low ROS levels increased Ca2+ spark frequency, while higher levels suppressed them. Along similar lines, a recent study by Prosser et al.63 linked stretch-induced increases in ROS production from an NADPH oxidase (Nox2) to increased spark frequency. Collectively, these studies suggest that moderate increases in ROS production activate RyRs and suggest a close interplay of mitochondrial and cytosolic redox state/ROS production and elementary Ca2+ release events, implying a redox feedback on EC coupling (Figure 1).

4. Mechanisms and kinetics of mitochondrial Ca2+ uptake and release

Although in recent years, considerable progress was made with regard to the molecular identity of proteins involved in mitochondrial Ca2+ uptake and release,3639,6469 several questions still remain unanswered and some controversies persist.5,7072 As already addressed by Hüser et al.73 in 2000, a central matter of debate was and is whether mitochondria take up Ca2+ rapidly, on a beat-to-beat basis, or rather slowly, integrating changes of amplitudes and frequency of cytosolic Ca2+ transients.5,37 This controversy is based partly on different findings related to different techniques that can be applied to measure mitochondrial Ca2+ uptake in cardiac myocytes during EC coupling (for more comprehensive reviews, see refs.5,71,74). While most groups agree that beat-to-beat mitochondrial Ca2+ uptake can occur, the absolute quantities of mitochondrial Ca2+ fluxes during cytosolic Ca2+ transients remain incompletely resolved,70,72 ranging from estimates of 10 nM/beat75,76 to ∼26% of the whole SR Ca2+ content during a cytosolic Ca2+ transient.77

4.1 Kinetics and molecular identity of the mitochondrial Ca2+ uniporter

Mitochondria take up Ca2+ primarily via the MCU, a recently identified 40-kDa channel protein in the IMM that is highly selective for Ca2+ (Figure 1).38,39,78 In COS7 cells, the MCU has a high Ca2+ carrying capacity (5 × 106 Ca2+ molecules per second per single MCU molecule), and MCU density in the IMM (∼10–40 per µm2) is only slightly lower than that of voltage-gated Ca2+ channels in the sarcolemmal membrane (∼100 per µm2).78,79 More recent evidence, however, showed that mouse heart mitochondria have dramatically lower MCU current densities than skeletal muscle mitochondria.80 It is currently unclear whether this low current density is limited to the mouse with heart rates of ∼600/min to avoid mitochondrial Ca2+ overload, or conserved across many species. The driving force for mitochondrial Ca2+ uptake is the large electrochemical gradient across the IMM (ΔΨm, ∼ −180 mV), and early experiments on suspensions of isolated mitochondria revealed Ca2+ concentrations for half-maximal activation (K0.5) of the MCU in the upper micromolar range.81,82 In more recent patch-clamp-based studies, the K0.5 was even higher in the millimolar range.78,83 These Ca2+ concentrations are clearly beyond levels expected in the bulk cytoplasm during Ca2+ transients in myocytes (∼1–2 µmol/L). In a study on human cardiac mitoplasts, two distinct mitochondrial Ca2+ channels were identified, coined mCa1 and mCa2, with mCa1 resembling the properties of the MCU and mCa2 with slightly different properties, possibly responsible for part of the non-Ru360-sensitive component of mitochondrial Ca2+ uptake.83 Similarly, Wei et al.84 identified two different modes of mitochondrial Ca2+ uptake with different sensitivities to Ru360, termed MCUmode1 and MCUmode2. This will be discussed in more detail below.

The MCU is part of a macromolecular complex that also contains MICU1 (Figure 1, inset), the first protein involved in mitochondrial Ca2+ uptake whose molecular nature could be identified in HeLa cells by the Mootha group.66 The identity of MICU1 was revealed by an integrative strategy that predicted human genes involved in mitochondrial Ca2+ uptake based on clues from comparative physiology, evolutionary genomics, and organelle proteomics.66 Systematic RNA interference experiments against 13 top candidates of this screen revealed MICU1 to be a protein that is associated with the IMM and that has two canonical EF hands that are required for its activity, indicating a role in Ca2+ sensing (Figure 1, inset).66 Silencing MICU1 abolished mitochondrial Ca2+ uptake and attenuated metabolic coupling between cytosolic Ca2+ transients and activation of Krebs cycle dehydrogenases.66 The observation that MICU1 overexpression did not enhance mitochondrial Ca2+ uptake and the fact that MICU1 has only one membrane-spanning domain, however, made it an unlikely candidate for the Ca2+ uniporter per se, but rather a regulatory protein that would act as a Ca2+ sensor based on the requirement of its two EF hands for proper function (see below).

In late 2011, two independent seminal papers from the Mootha and the Rizzuto groups unequivocally revealed the molecular identity of the Ca2+ uniporter.38,39 Using comparable systematic computational screening approaches, a protein with two membrane-spanning domains that locates to the IMM was identified, now termed the MCU (Figure 1, inset). In contrast to MICU1, overexpression of MCU doubled mitochondrial Ca2+ uptake in HeLa cells,39 while silencing MCU substantially reduced mitochondrial Ca2+ uptake in HeLa cells in vitro,38,39 but also in liver mitochondria in vivo.38 Reconstitution of the MCU in lipid bilayers yielded channel activity that resembled the electrophysiological properties and inhibitor sensitivity of the uniporter.39 The MCU forms oligomers, physically interacts with MICU1 and is part of a multi-protein complex. According to both studies, a highly conserved linker between the two transmembrane domains that faces the intermembrane space is required for the inhibitory effect of Ru360 (Figure 1, inset), the most specific inhibitor of the MCU so far.38,39 A recent study revealed that also in neonatal cardiac myocytes, silencing of the MCU reduced-, and overexpression potentiated mitochondrial Ca2+ uptake.85

Following up on MICU1,66 whose exact modulatory role for mitochondrial Ca2+ uptake was initially unresolved, a recent study revealed an important physiological role of this protein.65 Under resting conditions, mitochondrial Ca2+ levels are kept about five to six orders of magnitude lower than would be predicted by the electrochemical gradient across the IMM driven primarily by ΔΨm. In fact, earlier studies on isolated rat86 and guinea-pig cardiac myocytes43 had already observed a ‘threshold’ for mitochondrial Ca2+ uptake at cytosolic Ca2+ concentrations ([Ca2+]c) of ∼400–500 nmol/L; below this [Ca2+]c, [Ca2+]m was <[Ca2+]c, while at higher values of [Ca2+]c, [Ca2+]m rapidly increased.86 The low Ca2+ affinity of the MCU is one explanation for this phenomenon,78 restricting mitochondrial Ca2+ uptake to situations when [Ca2+]c rises substantially. In this regard, MICU1 interacts with the MCU and sets a Ca2+ threshold for mitochondrial Ca2+ uptake without affecting the kinetic properties of MCU-mediated Ca2+ uptake per se.65 Thus, MICU1 is considered a ‘gatekeeper’ of mitochondrial Ca2+ uptake that prevents mitochondrial Ca2+ overload,65 which can lead to activation of the permeability transition pore (PTP) and subsequently, programmed cell death by necrosis (or potentially apoptosis).8789

Another regulatory protein that binds to the MCU is MCUR1, an integral membrane protein required for MCU-dependent mitochondrial Ca2+ uptake.69 MCUR1 knockdown abrogated mitochondrial Ca2+ uptake in intact and permeabilized cells and disrupted oxidative phosphorylation, which lowered cellular ATP and activated AMP kinase-dependent pro-survival autophagy. The roles of neither MCUR1 nor MICU1 have been addressed in cardiac myocytes yet.

Recently, the group of Anderson proposed an important regulatory role for CaMKII in regulating MCU activity.90 They observed that CaMKII promoted opening of the mitochondrial PTP and myocardial death by enhancing the MCU current. A mitochondrial-targeted CaMKII-inhibitory protein reduced the MCU current, prevented mPTP opening (presumably by limiting mitochondrial Ca2+ overload after ischaemia/reperfusion), ΔΨm dissipation, programmed cell death and infarct size to a similar extent as cyclosporine A, an inhibitor of the PTP.90 Based on patch-clamp measurements, it was proposed that CaMKII regulates the MCU from the matrix side of the IMM.90 Since CaMKII activity is increased by GPCR activation and oxidative stress, CaMKII-mediated regulation of MCU may resemble an attractive control mechanism of mitochondrial Ca2+ uptake. This concept, and in particular its relevance under physiological conditions, will require further investigation.

In addition to the MCU, several other mechanisms and/or channels for mitochondrial Ca2+ uptake have been proposed. Among them are the rapid mode of uptake (RaM), coenzyme Q10, RyR1, uncoupling proteins 2 and 3 and leucine-zipper-EF-hand-containing transmembrane protein 1 (Letm1). For a more comprehensive review on these Ca2+ uptake mechanisms, we refer the reader to two recent comprehensive reviews by O-Uchi et al.36 and Dedkova and Blatter.37

4.2 Mitochondrial Ca2+ efflux

The primary mitochondrial Ca2+ efflux mechanism in cardiac myocytes is the mitochondrial NCX,81,91 which is presumably electrogenic92,93 by exchanging 1 Ca2+ for 3 Na+, with a Km for [Na+]i of ∼8 mM94 and thus, within the physiological range of [Na+]i in cardiac myocytes.21,95 Recently, the molecular nature of the mitochondrial NCX was identified as well.64 The exchanger forms dimers, locates to the mitochondrial cristae, transports either Na+ or Li+ in exchange for Ca2+ and thus, is termed NCLX (Figure 1).64 Ca2+-dependent mitochondrial Na+ import via the NCLX is counterbalanced by mitochondrial Na+/H+ exchange (mNHE). This makes mitochondrial Ca2+ uptake an energetically expensive process, since ΔµH is dissipated by the Na+-dependent H+ import to the matrix and needs to be regenerated by the ETC (Figure 1).

4.3 Integrative physiology of mitochondrial Ca2+ uptake and release in cardiac myocytes

While these studies are important to understand the molecular composition of the MCU with its regulatory proteins in general, they have not (yet) resolved the kinetics and quantities of mitochondrial Ca2+ uptake in cardiac myocytes during EC coupling. In the last couple of years, we have focused on this controversial issue by establishing a method by which cytosolic ([Ca2+]c) and mitochondrial Ca2+ concentrations ([Ca2+]m) are monitored in the same cell, respectively,43 by loading myocytes with a cell-permeable Ca2+ indicator that—due to its positive charge—accumulates primarily in the mitochondrial matrix (the ester form of rhod-2, i.e. rhod-2/AM). To eliminate cytosolic traces of the dye, myocytes are patch-clamped and dialysed with a rhod-2-free pipette solution that contains indo-1 salt that locates only to the cytosol.43 Eliciting cytosolic Ca2+ transients by voltage-clamping, we observed that during β-adrenergic stimulation (a situation when the heart requires increased ATP production), mitochondria take up Ca2+ rapidly, with [Ca2+]m transients peaking slightly earlier than cytosolic Ca2+ transients.43,44 This slightly earlier [Ca2+]m peak could not be explained by different dye kinetics and would be in agreement with the idea that mitochondria are located close to the SR where they would sense the RyR-released Ca2+ earlier (and at higher concentrations) than Ca2+ that is monitored by indo-1 in the bulk cytosol. Furthermore, [Ca2+]m transients had 2.5-fold slower decay kinetics than [Ca2+]c.43,44 This leads to accumulation of diastolic [Ca2+]m during an increase in the amplitude and/or frequency of cytosolic Ca2+ transients. Blocking mitochondrial Ca2+ uptake with Ru360 reduced the mitochondrial Ca2+ transients by ∼66%, whereas cytosolic Ca2+ transients slightly increased.43,44 Blocking the primary mitochondrial Ca2+ export mechanism (i.e. the NCLX) increased diastolic [Ca2+]m and decreased [Ca2+]c transients,43 while elevating [Na+]i to activate the NCLX had the opposite effect, accelerating mitochondrial Ca2+ efflux, reducing steady-state [Ca2+]m while slightly elevating [Ca2+]c.43

Since in our studies, rhod-2 was not calibrated to quantify [Ca2+]m, it remained unclear so far how much Ca2+ indeed enters mitochondria during a cytosolic Ca2+ transient. This question had spurred some controversy, since from indirect estimations (see below under ‘Evidence for a mitochondrial microdomain’) it seemed that mitochondria took up more Ca2+ than just 1% of overall cytosolic Ca2+ as originally proposed by the Bers group.96,97 Maack et al.43 and others85,98 observed that when blocking mitochondrial Ca2+ uptake, the amplitude of cytosolic Ca2+ transients increased, although in our more recent studies,44 this effect was smaller than in our initial study and no longer significant.43

Recently, the Bers group further addressed this issue. After their previous study using rhod-2 in permeabilized rat cardiac myocytes, in which they estimated that 2–10 nM Ca2+ enter mitochondria per beat,75 they now used the mitochondrially targeted inverse pericam indicator Mitycam, which was previously established in cardiac myocytes by the group of Smith,99 to measure mitochondrial Ca2+ uptake in intact rabbit cardiac myocytes.76 With high spatial resolution microscopy, they observed that areas of mitochondria that are located within 500 nm from the dyads (i.e. the Z-line) have an amplitude of [Ca2+]m of 37 nM per beat, while areas of mitochondria that are further (>500 nm) from the dyads have increases in [Ca2+]m of 26 nM per cytosolic Ca2+ transient. Furthermore, the time-to-peak of [Ca2+]m was shorter close to the Z-lines than in more remote areas of mitochondria, and since the MCU was evenly distributed over mitochondria, these data support the idea that spatial aspects (i.e. the existence of a mitochondrial Ca2+ microdomain; see also below) play an important role for the efficiency of mitochondrial Ca2+ uptake. Overall, these measurements support the concept that the quantities of mitochondrial Ca2+ uptake contribute only ∼1% to cytosolic Ca2+ handling, which argues against the idea that mitochondrial Ca2+ uptake ‘shapes’ cytosolic Ca2+ transients in cardiac myocytes.

Although this study is presently the best quantitative estimation of mitochondrial Ca2+ uptake during EC coupling in cardiac myocytes, there are still some methodological issues that need to be taken into account when interpreting the data. To express Mitycam in cardiac myocytes, cells had to be cultured for 36 h before the functional experiments,76 which may have an effect on the t-tubular organization of myocytes. Furthermore, the Kd of Mitycam for Ca2+ is ∼200 nmol/L,76,99 which may lead to saturation of fluorescence should [Ca2+]m approach micromolar concentrations. A potentially important difference between this study76 and our previous studies with rhod-2 AM43,44,100 is that our experiments were carried out at 37°C in freshly isolated cardiac myocytes, and the times to peak (TTP) of cytosolic and mitochondrial Ca2+ transients were <60 ms, respectively43,44,100. In contrast, the study by Lu et al.76 was carried out at room temperature, and TTP of [Ca2+]c and [Ca2+]m were between 200 and 300 ms. We observed that the rate of increase in [Ca2+]c (which correlates with SR Ca2+ release flux)101,102 had a strong impact on the efficiency of mitochondrial Ca2+ uptake. In conditions under which the TTP of [Ca2+]c was 200–300 ms (i.e. in response to NCX-mediated Ca2+ influx triggering SR Ca2+ release), the efficiency of mitochondrial Ca2+ uptake was only 50% compared with conditions with rapid ICa,L-induced SR Ca2+ release.100 Taken together, the study of Lu et al.76 makes a strong and conclusive point that overall mitochondrial Ca2+ fluxes are rather low, but due to the mentioned technical limitations the actual mitochondrial Ca2+ uptake under ‘real life’ conditions may still be slightly underestimated. Nevertheless, the concept is further supported by a recent study that observed that the current of the MCU is dramatically lower in cardiac compared with skeletal muscle mitochondria of the mouse.80

Another level of complexity to the quantitative aspects of mitochondrial Ca2+ uptake and free matrix concentrations is added by the largely unknown Ca2+ buffering capacity of mitochondria. By measuring [Ca2+]c and [Ca2+]m in isolated mitochondria, Wei et al.84 identified two modes of mitochondrial Ca2+ uptake, termed MCUmode1 and MCUmode2. MCUmode1 mediated a large and rapid change of free [Ca2+]m in response to small additions of extramitochondrial Ca2+. In contrast, MCUmode2 mediated a slow and lower affinity Ca2+ uptake which was capable of taking up large amounts of Ca2+ that led to only relatively small changes of free [Ca2+]m. The different modes can be explained by access of these two MCU modes to different Ca2+ buffer systems, with MCUmode1 allowing pronounced changes in free [Ca2+]m without major intramitochondrial Ca2+ buffering, potentially controlling Krebs cycle dehydrogenase activation for energy supply-and-demand matching.84 In contrast, MCUmode2 has access to a dynamic phosphate-dependent buffer system that serves as a Ca2+ sink in pathological situations of cytosolic Ca2+ overload (e.g. during ischaemia/reperfusion).84 In this respect, it was an interesting observation that the activation of the mPTP was uncorrelated with free [Ca2+]m and may rather be mediated by a ‘downstream by-product of [Ca2+]m, perhaps a Ca2+-phosphate species’.84

5. Evidence for a mitochondrial Ca2+ microdomain

An important question is how mitochondria can take up Ca2+ at rapid kinetics considering that the K0.5 of the MCU for Ca2+ is in the upper micromolar range, while the bulk cytosolic Ca2+ concentrations during systole are 1–2 orders of magnitude lower even during β-adrenergic stimulation. This seeming paradox would be resolved by the concept of a mitochondrial Ca2+ microdomain, in which mitochondria are in close vicinity to the RyRs of the SR (Figure 1 and inset), a concept that is already well accepted in non-cardiac cells with regard to the interaction between the endoplasmic reticulum (ER) and mitochondria.8 In fact, early electron micrographic studies revealed that in cardiac myocytes, the average distance between RyR2s and the mitochondrial surface ranges between 37 and 270 nm,103 and in our more recent studies analysing transmission electron micrographs, the distance between the junctional SR and mitochondria averaged 15–20 nm.68 Computational modelling predicts that the [Ca2+] at this distance from RyRs is at least 30 µmol/L during Ca2+ release.11 When simulating such ‘microdomain Ca2+ transients’ by exposing mitochondria to brief Ca2+ pulses of 10–30 µmol/L in a computer model that integrates known kinetics of mitochondrial Ca2+ uptake and release with Krebs cycle metabolism, our experimental findings in cardiac myocytes could be reconciled.43,45 Also functional data support the existence of a mitochondrial Ca2+ microdomain in cardiac myocytes: in cardiac H9c2 cells, mitochondrial Ca2+ uptake in response to caffeine-induced rapid SR Ca2+ release was comparable with exposing mitochondria to an extramitochondrial [Ca2+] of 30 µmol/L.104 When adding Ca2+ buffers (EGTA or BAPTA) to the cytosol, the cytosolic, but not the mitochondrial Ca2+ transient could be suppressed after rapid caffeine-induced SR Ca2+ release in cardiac cells, indicating privileged Ca2+ communication between both organelles.103,104 In mitochondria that were isolated by standard procedures from adult hearts, the application of caffeine elicited a Ca2+ transient in the mitochondrial matrix, but not in the extramitochondrial solution, which could be explained by Ca2+ transfer from the SR still being intimately linked to the OMM.105 Two recent studies that used Ca2+-sensitive fluorescent proteins genetically targeted to the OMM in non-cardiac cells or cardiac myotubes identified Ca2+ concentrations at the interface between the ER and mitochondria 5–10 times higher than in the bulk cytosol,106,107 further supporting the idea that mitochondria are exposed to Ca2+ concentrations sufficient to overcome the low Ca2+ affinity of the MCU.

Recently, we tested the existence and physiological implications of a mitochondrial Ca2+ microdomain in adult guinea-pig cardiac myocytes by comparing the efficiency of mitochondrial Ca2+ uptake (indexed by the ratio of [Ca2+]m over [Ca2+]c increases in the same cells, respectively) after either a coordinated SR Ca2+ release or a trans-sarcolemmal Ca2+ influx via the reverse-mode of the NCX with Ca2+-depleted SR.100 The efficiency of mitochondrial Ca2+ uptake was twice as high after SR Ca2+ release compared with NCX-mediated Ca2+ influx, with a tight correlation between the rate of rise of the cytosolic Ca2+ transient and the efficiency of mitochondrial Ca2+ uptake.100 This was of bioenergetic consequence, since the ratio of NAD(P)H/FAD, a sensitive index of Krebs cycle dehydrogenase activity, was increased more efficiently after SR Ca2+ release than after NCX-driven Ca2+ influx of similar magnitude. Taken together, these data strongly support the concept of privileged Ca2+ communication between mitochondria and the SR over other sources for cytosolic Ca2+ influx.

A very recent study by the Rizzuto group85 used genetically encoded fluorescent Ca2+ indicators targeted to the cytosol, the OMM or the mitochondrial matrix in combination with genetic silencing or up-regulation of the MCU in neonatal cardiac myocytes and observed that (i) Ca2+ concentrations at the OMM clearly exceeded bulk cytosolic Ca2+ concentrations by several orders of magnitude, (ii) down-regulation of the MCU decreased and up-regulation of the MCU increased rapid [Ca2+]m transients while (iii) the opposite occurred in the cytosol, where MCU down-regulation increased the amplitude of [Ca2+]c transients and up-regulation of MCU decreased [Ca2+]c transients. These data and some previous results43 lend further support to the concept of a mitochondrial Ca2+ microdomain and—in contrast to other studies75,76,108—suggest that in cardiac myocytes, rapid mitochondrial Ca2+ uptake shapes cytosolic Ca2+ transients.

In this context, it is interesting to note that in atrial cardiac myocytes, the lack (or paucity) of T-tubules limits cytosolic Ca2+ transients to the subsarcolemmal space. Mackenzie et al.109 observed that the propagation of subsarcolemmal Ca2+ transients to the central regions of the myocytes was limited by the SR and mitochondria, since inhibition of either SERCA or mitochondrial Ca2+ uptake increased the propagation of cytosolic Ca2+ to the central regions of the cell. These data support the idea that mitochondria could serve as a spatial buffer and thus, affect the magnitude and propagation of cytosolic Ca2+ transients.

6. Proteins that tether mitochondria to the SR

6.1 Role of mitofusin

Electron tomography studies on adult rat cardiac myocytes identified linker proteins that directly attach mitochondria to the SR, but whose molecular identity remained unresolved.105,110 In a seminal study on mouse embryonic fibroblasts, de Brito and Scorrano67 identified mitofusin (Mfn) 2 to physically link the ER to mitochondria. Mfns are proteins in the OMM with a well-characterized role in mitochondrial fusion (for a recent review, see Dorn and Maack111). Mfn1 and Mfn2 fuse mitochondria with each other by forming homo- and heterotypic complexes and thus, serve redundant roles for this process, so that deletion of neither Mfn1 nor Mfn2 alone affected mitochondrial fusion in vivo.68,112 Since Mfn2 is also located to the cardiac SR,113 it might serve as a linker between the SR and mitochondria, as revealed by de Brito and Scorrano67 for ER-mitochondrial tethering in non-cardiac cells. To test this, we analysed mice with cardiomyocyte-specific post-natal deletion of either Mfn1 or Mfn2.68 Up to an age of 6 weeks, these mice developed normal with no cardiac phenotype. However, transmission electron microscopy revealed that in Mfn2-, but not Mfn1-deficient mice, the contact length between the junctional SR and mitochondria was decreased, with a trend towards a widening of the gap between the two organelles. Likewise, RyR-containing mitochondrial associated membranes as a biochemical measure of SR-mitochondrial contact were strikingly decreased in Mfn2-KO mice.68

These data contrast somewhat with data from the Walsh group by Papanicolaou et al.,114 who reported that cardiac-specific ablation of Mfn2 in mice did not alter the close associations between SR and mitochondria. There were, however, important methodological differences between our studies and those of Papanicolaou et al..114 First, we used a mouse model with post-natal knock-out of Mfn2,68 while in the other study,114 Mfn2 was knocked out already in the embryo, potentially facilitating mechanisms to compensate for the loss of Mfn2. Furthermore, Papanicolau et al.114 analysed only the distance between mitochondria and the centre of T-tubules, which was in the range of 150 nm and not different between Mfn2-KO and control animals.114 In contrast, we analysed the distance between the junctional SR and mitochondria, which is in the range of only 15 nm and a more direct parameter for SR-mitochondrial tethering.

To elucidate the functional relevance of these ultrastructural alterations, we performed experiments on isolated cardiac myocytes which were exposed to a physiological increase in workload (increased pacing frequency and β-adrenergic stimulation for 3 min). While in Mfn1 KO myocytes, mitochondrial Ca2+ uptake and the redox states of NAD(P)H and FAD were unchanged compared with WT littermates, in Mfn2-KO myocytes, mitochondrial Ca2+ uptake was hampered during β-adrenergic stimulation at low stimulation frequencies.68 At the same time, a transient oxidation of NAD(P)H and FADH2 occurred, indicating that reduced mitochondrial Ca2+ uptake was associated with decreased Krebs cycle dehydrogenase activation. In agreement with our previous observation that the Krebs cycle also controls NADPH and ROS elimination,43 a transient increase in ROS formation was observed in Mfn2-KO myocytes. Taken together, these results uncover critical roles for cardiomyocyte Mfn2 in tethering mitochondria to the junctional SR and creating a mitochondrial Ca2+ microdomain. Disruption of SR-mitochondrial cross-talk hampers energy supply-and-demand matching, and possibly provokes oxidative stress. At the same time, these data are in line with previous reports43,85,98 that mitochondrial Ca2+ uptake shapes cytosolic Ca2+ transients.

6.2 Other tethering proteins

Previous studies in non-cardiac cells have revealed several proteins involved in tethering the ER to mitochondria. Already in 2002, Rizzuto’s group identified voltage-dependent anion channel (VDAC), a well-characterized Ca2+ permitting protein in the OMM to govern Ca2+ transmission from the ER to mitochondria in HeLa cells.115 Subsequently, the same group found that VDAC is physically linked to the ER Ca2+ release channel, the inositol 1,4,5-triphosphate receptor (IP3R) by glucose-regulated protein 75 (grp75).116 In a more recent study, Hajnoczky’s group observed that during cardiomyocyte development, the ER-mitochondrial Ca2+ transfer that is governed by IP3Rs in early developmental stages of cardiomyocytes transforms to a RyR2-dependent SR-mitochondrial Ca2+ transfer.117 In agreement with this concept, a recent biochemical study revealed physical interaction between RyR2 and VDAC2 with co-localization in SR-mitochondrial junctions in HL-1 cells (Figure 1).118 It is currently unclear whether this interaction is direct or maybe mediated by another tethering protein-like grp75 (which tethers IP3Rs to VDAC).116 Other studies in non-cardiac cells identified the Mmm1/Mdm10/Mdm12/Mdm34 complex119 and PACS-2120 as linker proteins between the ER and mitochondria. Their role in cardiac myocytes, however, is currently unknown.

The role of tethering proteins is obviously thought to create Ca2+ microdomains between the SR and the OMM, and recent evidence suggests that the areas of the OMM which are tethered closely to the SR are also closer to the IMM than areas of the OMM that do not have a close SR contact113 (Figure 1, inset). In fact, these contact points between IMM and OMM are enriched in VDAC,113 which associates with RyR2,118 and a lighter version of the Mfn2 protein (50 kDa), while those areas of the OMM that are not engaged in contact points harbour the full-length Mfn2 protein (∼80 kDa).113 While the functional implications of the lighter Mfn2 protein are presently unclear, these data nevertheless suggest that the contact points function as anchorage sites for the SR-mitochondrial physical coupling, and that close coupling of the SR, OMM, and IMM is likely to ‘provide a favourable spatial arrangement for local RyR2-mitochondrial Ca2+ signalling’.113

7. Pathophysiological changes in heart failure

In chronic heart failure, maladaptive remodelling of EC coupling induces contractile dysfunction. The most prominent changes are (i) decreased SR Ca2+ load, (ii) an elevation of [Na+]i that reduces cytosolic Ca2+ efflux (via the forward mode of the NCX) and increases Ca2+ influx (via the reverse mode; Figure 1), and (iii) t-tubular remodelling that induces a dyssynchronization of cytosolic Ca2+ transients.5,1216 Irrespective of the absolute quantities of mitochondrial Ca2+ uptake during EC coupling, all these processes have a negative impact on mitochondrial Ca2+ uptake. First, the amplitudes of mitochondrial Ca2+ transients correlate with the amplitudes of cytosolic Ca2+ transients,43,44,100 which are decreased in failing cardiac myocytes.25 Furthermore, we observed that reverse-mode NCX-driven Ca2+ influx is less efficient for mitochondrial Ca2+ uptake than coordinated SR Ca2+ release (due to its slower kinetics and less privileged localization to mitochondria; Figure 1).100 Similarly, t-tubular dysorganization leads to dyssynchronized and thus, slower cytosolic Ca2+ transients, which based on our experimental results100 should also negatively affect mitochondrial Ca2+ uptake. A study on human cardiac mitoplasts revealed that the opening probability of the mCa1 (equalling MCU) and mCa2 were reduced in the mitochondria from patients with heart failure.83 Finally, since Ca2+ is exported from the mitochondria by the NCLX, the elevation of [Na+]i observed in failing myocytes reduces steady-state [Ca2+]m and leads to an oxidation of NADH and NADPH, which on the one hand impairs energy supply-and-demand matching and on the other hand leads to oxidative stress43,44,121 (Figure 1). Increased ROS production, in turn, activates CaMKII,122 which elevates the late INa30 and thus, [Na+]i, inducing a positive feedback-loop that sustains defects in EC coupling, energetics deficit and oxidative stress.35 Furthermore, mitochondrial ROS activate RyRs and thus, have a direct feedback on SR Ca2+ release events, which may further contribute to SR Ca2+ leak and arrhythmias (Figure 1).6062

8. Conclusions

Considerable progress has been made on identifying the mechanisms and molecular determinants of mitochondrial Ca2+ uptake and release in cardiac myocytes. In particular, the identification of the MCU, its regulatory factors MICU1 and MCUR1 as well as of the NCLX, but also tethering proteins between the SR and mitochondria, such as Mfn2, RyR2, and VDAC will allow further characterization of the processes of mitochondrial Ca2+ uptake and release in cardiac myocytes. These studies together with the further improvement of mitochondrially located Ca2+ indicators will hopefully resolve some of the remaining open issues such as the exact quantities and kinetics of mitochondrial Ca2+ uptake during EC coupling and its role for cytosolic Ca2+.

Conflict of interest: none declared.


The authors are supported by the Deutsche Forschungsgemeinschaft (DFG; Heisenberg Programm, SFB 894 and Klinische Forschergruppe 196).


  • The article is part of the Spotlight Issue on: T-tubules and ryanodine receptor microdomain signalling in cardiac hypertrophy and failure.


View Abstract