Excitation-contraction coupling and mitochondrial energetics

Abstract

Cardiac excitation-contraction (EC) coupling consumes vast amounts of cellular energy, most of which is produced in mitochondria by oxidative phosphorylation. In order to adapt the constantly varying workload of the heart to energy supply, tight coupling mechanisms are essential to maintain cellular pools of ATP, phosphocreatine and NADH. To our current knowledge, the most important regulators of oxidative phosphorylation are ADP, P_i, and Ca²⁺. However, the kinetics of mitochondrial Ca²⁺-uptake during EC coupling are currently a matter of intense debate. Recent experimental findings suggest the existence of a mitochondrial Ca²⁺ microdomain in cardiac myocytes, justified by the close proximity of mitochondria to the sites of cellular Ca²⁺ release, i. e., the ryanodine receptors of the sarcoplasmic reticulum. Such a Ca²⁺ microdomain could explain seemingly controversial results on mitochondrial Ca²⁺ uptake kinetics in isolated mitochondria versus whole cardiac myocytes. Another important consideration is that rapid mitochondrial Ca²⁺ uptake facilitated by microdomains may shape cytosolic Ca²⁺ signals in cardiac myocytes and have an impact on energy supply and demand matching. Defects in EC coupling in chronic heart failure may adversely affect mitochondrial Ca²⁺ uptake and energetics, initiating a vicious cycle of contractile dysfunction and energy depletion. Future therapeutic approaches in the treatment of heart failure could be aimed at interrupting this vicious cycle.

Physiological aspects

The most important function of the heart is to pump blood to supply the body with oxygen and substrates. In order to adapt cardiac output to the constantly changing demand of blood supply, the heart utilizes three major mechanisms to increase cardiac output: the force-frequency relationship (the Bowditch effect or Treppe; [22]), the Frank-Starling mechanism (sarcomere length-dependent activation of contractile force) [66, 149, 164], and sympathetic activation [28]. All of these mechanisms increase and accelerate myocardial force generation, and at the same time increase cellular energy demand. By these mechanisms, cardiac output during exertion can be increased more than five-fold compared to resting conditions.

Excitation-contraction coupling

Of fundamental importance for cardiac contraction and relaxation are the processes of excitation-contraction (EC) coupling (Fig. 1). During the action potential (AP), voltage-gated Na⁺-channels are activated, and the inward Na⁺-current (I_Na) induces a rapid depolarization of the cell membrane, facilitating voltage-dependent opening of L-type Ca²⁺-channels (I_Ca,L). The Ca²⁺ influx triggers the opening of the ryanodine receptor (RyR2 subtype), inducing the release of even greater amounts of Ca²⁺ from the sarcoplasmic reticulum (SR), a process termed Ca²⁺-induced Ca²⁺-release (CICR) [14, 15, 63, 64]. The Ca²⁺ released from RyRs floods the space between the SR and the cell membrane (i. e., the dyadic or junctional cleft), a gap of typically ∼10 nm and covering a volume with a radius of ∼200 nm [102, 151]. Computational modeling predicted a temporarily and spatially limited increase of the Ca²⁺-concentration ([Ca²⁺]) in the range from ∼10 µM to ∼7 mM within the first couple of ms after opening of the RyR2 [14, 15, 102, 133, 151, 152, 158, 174, 217]. This high junctional [Ca²⁺] decays rapidly by diffusion into the submembrane space and then further into the bulk cytosol, where it peaks later and at much lower concentrations than in the cleft (at ∼1.5 µM in the submembrane space and ∼0.5 µM in the bulk cytosol [174, 209]). Due to the spatial and temporal heterogeneity of [Ca²⁺] throughout different compartments of the cytosol, the term of “Ca²⁺ microdomains” was coined [160], with the junctional cleft being the best characterized Ca²⁺ microdomain in cardiac myocytes.

Fig. 1.

Overview on processes of excitation-contraction coupling and mitochondrial energetics. SR sarcoplasmic reticulum; SERCA SR Ca²⁺ ATPase; Mito mitochondria; TCA tricarboxylic acid cycle; RC respiratory chain; Δψ_m mitochondrial membrane potential; NCE mitochondrial Na⁺/Ca²⁺-exchanger; NHE mitochondrial Na⁺/H⁺ -exchanger; NKA sarcolemmal Na⁺/K⁺-ATPase; NCX sarcolemmal Na⁺/Ca²⁺-exchanger; RyR2 ryanodine receptor type 2; mCU mitochondrial Ca²⁺-uniporter; I_Na and I_Ca, currents of voltage-gated Na⁺- or Ca²⁺-channels, respectively

Experimentally, single Ca²⁺ release events can be observed by confocal microscopy as “Ca²⁺ sparks” that occur spontaneously in quiescent cardiac myocytes [42]. These Ca²⁺ sparks arise from the concerted activation of a whole cluster of RyRs [43, 177, 181]. Ventricular myocytes have a highly organized three-dimensional structure, with invaginations of the plasma membrane reaching into the interior of the cell (Fig. 1). This transverse tubular system (t-tubules, [182]) facilitates the spatial and temporal coordination of CICR during an AP, resulting in the summation of a large number of Ca²⁺ sparks, whose activation is a stochastic process [31]. In contrast, atrial myocytes lack t-tubules and thus, Ca²⁺-transients initiate in the periphery of the cell and then spread centripetally [20, 110, 119]. The increasing cytosolic Ca²⁺ binds to the myofilaments (i. e., troponin C) and induces the contraction of the cell. The contraction of the left and right ventricle leads to the ejection of blood through the aortic and pulmonary valves, respectively, initiating systole. During diastole, relaxation of the myocardium allows passive refilling of the ventricles with blood. Relaxation is initiated by the diffusion of Ca²⁺ from the myofilaments back to the cytosol. The main mechanisms removing Ca²⁺ from the cytosol are the SR Ca²⁺-ATPase, the sarcolemmal Na⁺/Ca²⁺-exchanger (NCX) and the plasmalemmal Ca²⁺-ATPase (PMCA) [14, 15, 179].

Mitochondrial energetics

When the body is at rest, i. e., at normal heart rates of ∼60 beats per minute in human, EC coupling already consumes vast amounts of energy. The main cellular energy consumers are the myosin ATPase of the contractile filaments, the plasmalemmal Na⁺/K⁺-ATPase, and the SR Ca²⁺-ATPase [15]. It is assumed that ∼2% of the cellular ATP pool is consumed in each heart beat, and during maximal workload, the whole ATP pool is turned over within a couple of seconds [7, 80, 139]. The main sites of energy production are the mitochondria, which take up ∼30% of cellular volume [15, 222] and are located in close vicinity to the main sites of energy consumption, i. e., the myofilaments, the SR and t-tubules (Figs. 1 and 2; [194, 220]). Under physiological conditions, glucose is transformed to pyruvate, which enters mitochondria and is transformed to acetyl-coenzyme A (CoA) by pyruvate dehydrogenase (PDH). Fatty acids are activated to fatty acyl-CoA in the cytosol and transported into mitochondria via the carnitine-acyltranslocase. Acyl-CoA enters β-oxidation, with acetyl-CoA, NADH and FADH₂ as the main products. Acetyl-CoA from both glycolysis and β-oxidation enters the tricarboxylic acid (TCA) cycle, resulting in the formation of NADH and FADH₂ (Fig. 3A).

Fig. 2.

Longitudinal (A) or scanning (B) electron micrographs of cardiac myocytes. M mitochondria; SR sarcoplasmic reticulum. With permission from Territo et al. [194] and Yoshikane et al. [220], respectively

Fig. 3.

(A) Enzymatic reactions of the tricarboxylic acid (TCA) cycle, and their regulation by Ca²⁺. DH dehydrogenase. (B) Stimulation of pyruvate-dehydrogenase phosphatase (PDH-PPase), NAD-isocitrate dehydrogeNase (NAD-ICDH) and α-ketoglutarate dehydrogenase (α-KGDH) enzymatic activity by Ca²⁺ in extracts of mitochondria. Activity in the presence of EGTA was ∼ 10–20% of maximal velocity (V_max), respectively. Reproduced from Denton & McCormack [54] with permission

NADH feeds electrons into the respiratory chain at complex I (Fig. 4), and succinate enters at complex II. By sequential redox reactions at complexes I–IV, protons are translocated across the inner mitochondrial membrane, establishing a proton gradient (ΔpH) and an electrical gradient (Δψ_m) which constitute the proton motive force (Δµ_H). The electrons that travel down the respiratory chain are eventually transferred to O₂, forming H₂O. O₂ consumption is respiration. At complex V of the respiratory chain (the F₁/F₀-ATPase), Δµ_H provides the free energy for the generation of ATP from ADP (oxidative phosphorylation) [45, 187].

Fig. 4.

Processes of oxidative phosphorylation at the mitochondrial respiratory chain. I–V complexes I–V of the respiratory chain. Complex V F₁/F₀-ATPase; c cytochrome c. Electrons (e⁻) donated by NADH to complex I (and by succinate to complex II; not shown) elicit sequential redox reactions at complexes I–IV, promoting H⁺-translocation from the matrix to the intermembrane space across the inner mitochondrial membrane (IMM). This creates a proton gradient (ΔpH), which together with the electrochemical gradient (Δψ_m) constitutes the driving force for protons to flow back into the matrix space via F₁/F₀-ATPase, promoting ATP-production from ADP. At complex III (and I; not shown), the superoxide anion radical is produced already under physiological conditions. It is dismutated to hydrogen peroxide (H₂O₂) by Mn²⁺-dependent superoxide-dismutase (Mn-SOD), which is eliminated to H₂O by glutathione-peroxidase (GPX). While Ca²⁺ activates (+) TCA-cycle dehydrogenases (but also complex V of the respiratory chain; not shown), ADP regulates respiration by increasing complex V activity (+)

Energy supply and demand matching

To tightly orchestrate the enormous energetic flux during the permanent changes of cardiac workload, a highly efficient matching process of energy supply and demand is essential. The two most important regulatory factors identified to date are ADP and Ca²⁺ [7, 25, 26, 37, 45, 46, 79, 129, 164, 165, 194]. The classical respiratory controlhypothesis of Chance and Williams [37] implies that the rate of respiration is regulated by the availability of ADP to the F₁/F₀-ATPase. This is based on the observation that when adding ADP to a suspension of isolated mitochondria, O₂ consumption increases (“state 3 respiration”), and when ADP and/or P_i in the assay are consumed, the rate of respiration decreases again (“state 4 respiration”). However, at the end of the 1980s, a series of experiments by Katz and Balaban et al. [106–108] called into question the regulatory role of ADP for respiration in vivo. In isolated rat hearts [106, 107] or instrumented dogs [108] they observed that an increase of cardiac work (by pacing or hormonal stimulation) increased O₂ consumption, however, without any changes of ADP, ATP, P_i or phosphocreatine (PCr). They concluded that respiration is regulated by the (upstream) availability of electrons to the respiratory chain, i. e., the redox state of NADH/NAD⁺, rather than ADP [107, 108]. Since NADH is produced by the TCA cycle, and three key enzymes of the TCA cycle (PDH, isocitrate- and α-ketoglutarate dehydrogenases) are activated by Ca²⁺ [53, 54, 56, 57, 127, 128] (Fig. 3A and B), it was proposed that Ca²⁺ rather than ADP may regulate respiration in vivo [106]. In more recent studies, Territo and Balaban et al. extended this concept by showing that besides dehydrogenases, Ca²⁺ also activates the F₁/F₀-ATPase [193, 195], and confirmed earlier studies [36] demonstrating that Ca²⁺ increases respiration in less than 100 ms, rapidly enough to support cardiac work transitions in vivo.

Of course, Ca²⁺ stimulation of oxidative phosphorylation alone is insufficient to account for the adaptation of mitochondrial energetics to variations in workload induced by changes in muscle length, since these occur largely without variations in the Ca²⁺ transient (metabolic basis of the Frank-Starling law [164]). Furthermore, while Ca²⁺ is able to increase respiration by a factor of ∼2, changes of left ventricular filling in isolated perfused rat hearts increases respiration by a factor of more than 10-fold [45, 164, 194, 216]. This has led to other proposed messengers, including P_i [21], although the correlations between this metabolite and cardiac work are also not robust. A difficulty of drawing conclusions regarding the relationship between levels of energy metabolites and ATP production stems from the assumption that bulk measurements of intracellular ATP and P_i (e.g., from NMR measurements) reflect the actual local intramitochondrial concentrations responsible for controlling respiratory flux. This assumption is as weak as assuming that gain of CICR is determined by the bulk cytoplasmic Ca²⁺ concentration. During CICR, bulk cytosolic Ca²⁺ can be completely buffered with no effect of local EC coupling, as demonstrated in studies of “Ca²⁺ spikes” [173, 183]. Similarly, cytoplasmic high energy phosphate buffer systems (e.g., creatine kinase (CK) isoenzymes and the rapidly diffusable phosphocreatine, PCr) are present that can limit changes in bulk ATP while shuttling ADP to the mitochondrial adenine nucleotide transporter (ANT), effectively transferring the energetic signal from the site of ATP hydrolysis to the mitochondrion (reviewed in [164]). Hence, in agreement with recent experimental and computational evidence, Ca²⁺, ADP, and P_i are likely to regulate respiration in a complementary way at sites both upstream and downstream of the respiratory chain, thus providing a balanced availability of ATP and reduced NADH [25, 26, 45, 46, 101, 117, 164] (see also further below in Bioenergetic consequences of mitochondrial Ca²⁺ uptake).

Mitochondrial Ca²⁺ uptake

A prerequisite for Ca²⁺-induced activation of respiration is that the amount of Ca²⁺ that enters mitochondria is sufficient to activate the dehydrogenases of the TCA cycle. The K_0.5 for Ca²⁺ activation of the TCA cycle dehydrogenases is in the range of 0.7–1 µM (Fig. 3B; [54, 56, 57, 127, 128]). As reviewed previously [76, 77, 79, 129], Ca²⁺ is taken up by mitochondria via a Ca²⁺ uniporter (mCU; Fig. 1), driven by the large electrochemical gradient for Ca²⁺ across the inner mitochondrial membrane (Δψ_m∼ −180 mV). From experiments on suspensions of isolated mitochondria, a half-maximal activation constant (K_0.5) of ∼10–20 µM Ca²⁺ was calculated for mCU-mediated Ca²⁺ transport, yielding a maximum Ca²⁺ flux (V_max) of approximately 2·10⁴ Ca²⁺ s⁻¹ for a single mCU molecule [76, 77]. The mCU is inhibited by ruthenium red [124, 138, 157] or Ru360 [109, 117, 124], with the latter being more mCU-selective than ruthenium red [124], which also inhibits SR Ca²⁺ release via RyR2 [35]. Furthermore, dissipation of Δψ_m by agents that uncouple respiration (i.e., a combination of FCCP and oligomycin) also inhibits mitochondrial Ca²⁺ uptake by reducing the driving force for Ca²⁺ entry (Δψ_m) rather than interacting with the mCU itself.

Recently, Kirichok et al. [109] characterized the mCU by patch-clamping the inner mitochondrial membrane of isolated mitoplasts. Its inwardly rectifying Ca²⁺ current (maximum at Δψ_m = −160 mV) had a half-activation constant (K_0.5) of ∼20 mM with a Ca²⁺ carrying capacity that was ∼250-fold higher than calculated from experiments on isolated mitochondria (V_max = 5·10⁶ Ca²⁺ s⁻¹ per single mCU molecule). The current lacked Ca²⁺-dependent inactivation, but was highly sensitive to ruthenium red and Ru360. The channel density of the mCU at the inner mitochondrial membrane was estimated to amount ∼10–40 channels per µm², which is only slightly less than the density of voltage-gated Ca²⁺ channels at sarcolemmal membranes of excitable cells (∼100 per µm²; [90, 109]). In the absence of Ca²⁺, the mCU was able to transmit Na⁺ (or K⁺ to a much smaller degree) instead of Ca²⁺, but not Mg²⁺. However, already at low [Ca²⁺]_c (K_d for Ca²⁺-binding <2 nM), Na⁺ and K⁺ conductances were completely inhibited. Thus, under physiological conditions in cardiac myocytes ([Ca²⁺]_c ≈100 nM), the mCU is a highly selective Ca²⁺ channel, allowing Ca²⁺ to accumulate in mitochondria with the least dissipation of Δψ_m [109].

An important difference between the study by Kirichok et al. [109] and previous studies [76, 77] concerning ionic flux estimates is that in mitochondrial suspensions [76, 77] or whole cardiac myocytes [222], dissipation of Δψ_m due to Ca²⁺ entry at very high extramitochondrial [Ca²⁺] may reduce the driving force for further Ca²⁺ uptake, leading to (pseudo-) lower transport rates that saturate at lower extramitochondrial [Ca²⁺]. In contrast, in the study of Kirichok et al. [109], the potential across the inner mitochondrial membrane could be maintained by voltage-clamp, allowing Ca²⁺ flux at much higher extra- and intramitochondrial [Ca²⁺] than in mitochondrial suspensions or whole cells. This may explain the differences of K_0.5 and V_max for mitochondrial Ca²⁺ transport between those studies [76, 77, 109].

The primary mitochondrial Ca²⁺ efflux mechanism in cardiac cells is the mitochondrial Na⁺/Ca²⁺ exchanger (mNCE) [54, 77], which is presumably electrogenic [10, 103] by exchanging 1 Ca²⁺ for 3 Na⁺, with a K_m for [Na⁺]_i of ∼8 mM [150]. Earlier studies, however, reported an electroneutral transport mode [48, 49].

A mitochondrial Ca²⁺ microdomain

At this point, the question arises as to why the mitochondrial inner membrane expresses a high density of selective Ca²⁺ channels that transport Ca²⁺ primarily at concentrations exceeding cytosolic bulk [Ca²⁺] (maximally ∼1–3 µmol/L). A possible explanation may be the concept of a mitochondrial Ca²⁺-microdomain, which is already well established in non-cardiac cell types (reviewed by Rizzuto et al. [159, 160]). According to this concept, the spatially- and temporally-limited occurrence of very high concentrations of Ca²⁺ (“hot spots”) at cellular Ca²⁺-release sites (e.g., inositol 1,4,5-triphosphate- (IP₃-) receptors of the endoplasmatic reticulum, ER) allows rapid mitochondrial Ca²⁺ uptake despite a relatively low affinity of the mCU for Ca²⁺ [159, 160]. On the one hand, this privileged Ca²⁺ signal transmission efficiently adapts cellular energy demand to supply, and on the other hand, it shapes the dynamics of [Ca²⁺]_c [159, 160]. Furthermore, under pathological conditions involving cytosolic Ca²⁺ overload, the low affinity of the mCU may prevent premature mitochondrial Ca²⁺ overload, since the route and rate of cellular Ca²⁺ influx is considerably slower than a triggered release of Ca²⁺ from an intracellular store, i. e., the ER or the SR. Mitochondrial Ca²⁺ overload, together with increased production of reactive oxygen species (ROS), can activate the permeability transition pore (PTP), dissipating Δψ_m and allowing the release of cytochrome c into the cytosol. This can initiate necrotic cell death or, by cytochrome c-dependent activation of cytosolic caspases, apoptosis [29].

Cardiac myocytes, in particular, display a highly organized juxtaposition of sarcolemma, SR and mitochondria (Fig. 2). Mitochondria are ensnared in the SR network and are bounded at each end by the junctional Ca²⁺ release sites (Figs. 1 and 2). Whether this spatial association of mitochondria with the SR results in rapid mitochondrial Ca²⁺ uptake during EC coupling, however, has been highly controversial for over a decade now (for a previous review see Hüser et al. [95]). Over the past 10–15 years, a number of studies have been aimed at elucidating whether during EC coupling, mitochondria take up Ca²⁺ rather slowly, integrating cytosolic Ca²⁺ transients, or rapidly, on a beat-to-beat basis [95]. The clarification of this issue may be of paramount importance to understanding energy supply and demand matching and the interplay between mitochondrial energetics, EC coupling and cellular signaling processes mediated by, e. g., ROS or Ca²⁺ [50]. Thus, in the following discussion, we will review the evidence for and against rapid mitochondrial Ca²⁺ uptake during EC coupling.

Evidence against fast mitochondrial Ca²⁺ uptake

One of the first quantitative estimations of mitochondrial Ca²⁺ uptake during EC coupling comes from a study by Bassani et al. [9]. The aim of that study was to calculate the relative contribution of SR Ca²⁺-ATPase, NCX, PMCA and mitochondria for cytosolic Ca²⁺ removal after a rapid release of Ca²⁺ from the SR. In rabbit ventricular myocytes loaded with the Ca²⁺-indicator indo-1/AM, Ca²⁺ was released from the SR by caffeine. Since in the persistent presence of caffeine, the SR Ca²⁺-ATPase cannot refill the SR, the decay of [Ca²⁺]_c was achieved primarily by forward-mode I_NCX. Consequently, with 0 Na⁺ and 0 Ca²⁺ in the extracellular solution (to inhibit I_NCX), the decay of [Ca²⁺]_c was substantially slowed [9]. The remaining decay was attributed to Ca²⁺ removal by mitochondria and the PMCA, since it was blocked by mitochondrial uncoupling (by application of FCCP and oligomycin 5 s prior to the caffeine pulse) or inhibition of PMCA, each contributing about equally to the Ca²⁺ removal rate [9]. Thus, the relative contributions of SR Ca²⁺-ATPase, NCX, PMCA and mitochondria to diastolic Ca²⁺-removal in rabbit ventricular myocytes were 70%,28%,1% and 1% [9, 14, 15].This implies that under steady-state conditions, 70% of Ca²⁺ cycle into and out of the SR, whereas ∼29% enter the cell via L-type Ca²⁺ channels and leave the cell via the NCX and PMCA. Consequently, only 1% of Ca²⁺ would enter and leave mitochondria on every beat. This was in agreement with estimates by Sipido and Wier [179] and the conception that the affinity of the mCU for Ca²⁺ transport was too low to allow beat-to-beat oscillations of mitochondrial [Ca²⁺] ([Ca²⁺]_m) during cytosolic Ca²⁺ transients [14, 15].

In the following years, a number of studies attempted to measure mitochondrial Ca²⁺ uptake during EC coupling more directly, with quite controversial results [12, 26, 33, 59–61, 68, 72–74, 100, 117, 119, 136, 137, 140, 141, 144, 147, 148, 161, 170, 171, 176, 191, 200, 201, 213]. One reason for these controversial results may be that different techniques were employed to investigate the kinetics of mitochondrial Ca²⁺-uptake. It is a technical challenge to reliably and selectively monitor changes in [Ca²⁺]_m without contamination by cytosolic Ca²⁺ signals. In most approaches, cell-permeable fluorescent Ca²⁺ indicators (e.g., indo-1/AM [59, 60, 72–74, 221, 222], rhod-2/AM [26, 117, 119, 148, 176, 191, 200, 201] or fluo-3 [33, 144, 170]) penetrate sarcolemmal and mitochondrial membranes and accumulate (to different degrees) in the mitochondrial matrix. To avoid contamination of the fluorescence signal by cytosolic Ca²⁺, the groups of Hansford, Silverman [59, 60, 74, 136, 137] and Bers [26, 222] applied MnCl₂ in the extracellular solution. Mn²⁺ enters the cell, presumably via L-type Ca²⁺-channels, binds to cytosolic indo-1 with ∼20-fold higher affinity than Ca²⁺ and quenches its fluorescence [123, 137, 222]. Although Mn²⁺ competes with Ca²⁺ for the mCU [69, 123], and thus can be taken up into mitochondria as well [26, 222], it was proposed that at the concentrations used, cytosolic free Mn²⁺ was too low to effectively compete with Ca²⁺ for the mCU, since most cytosolic Mn²⁺ should be bound to indo-1 [222]. Indeed, MnCl₂ did not affect cell shortening, indicating that despite a potential inhibitory effect of Mn²⁺ on I_Ca,L [123], EC coupling was not hampered [137]. Depending on the loading protocol of indo-1/AM, MnCl₂ reduced its overall fluorescence by ∼30–55%, indicating that ∼45–70% of fluorescence stemmed from non-cytosolic compartments, i. e., primarily from mitochondria [137, 222].

In all studies using the Mn²⁺-quench method [59, 60, 74, 136, 137], no beat-to-beat oscillations of [Ca²⁺]_m in response to cytosolic Ca²⁺-transients could be detected (Fig. 5A and B). Instead, a slow, cumulative increase of [Ca²⁺]_m from resting levels (∼90 nmol/L) to ∼400–600 nmol/L was observed in response to β-adrenergic stimulation and an increase of stimulation frequency to 4 Hz (Fig. 5C; [60, 137]). When cells were paced at 0.2 Hz, lowering of extracellular [Na⁺] induced reverse-mode I_NCX-dependent cytosolic Ca²⁺-influx with subsequent slow mitochondrial Ca²⁺-accumulation [137]. Plotting [Ca²⁺]_m against the respective [Ca²⁺]_c (determined in a different set of cells) revealed a nonlinear relationship between [Ca²⁺]_m and [Ca²⁺]_c, with a threshold at [C²⁺]_c ≈500 nmol/L and an exponential increase of [Ca²⁺]_m above this threshold (Fig. 5D; [137]). This exponential relationship indicates positive cooperativity of mitochondrial Ca²⁺ uptake, a phenomenon observed also in whole cardiac myocytes [67, 222], isolated mitochondria [126, 129] and other cell types [44]. The exponential slope was heavily influenced by extramitochondrial [Na⁺] [45, 126, 129], reflecting the critical influence of mNCE activity on steady-state levels of [Ca²⁺]_m. Mitochondrial Ca²⁺-uptake was sensitive to the mCU-inhibitors ruthenium red or Ru360, or dissipation of Δψ_m [60, 137], whereas mitochondrial Ca²⁺-decay was reduced by the mNCE-inhibitors clonazepam or CGP-37157 [74].

Fig. 5.

[Ca²⁺]_c and [Ca²⁺]_m during a single contraction in the presence and absence of norepinephrine (NE): myocyte loaded with indo-1/salt (A); myocyte loaded with indo-1/AM and quenched with Mn²⁺ (B). A and B each show results from a single cell, stimulated at steady-state 0.2 Hz in presence and absence of 1 µM NE. It is seen that there is no response of [Ca²⁺]_m to a single electrical stimulation. (C) Kinetics of rise and fall of [Ca²⁺]_m in response to changed pacing rate. Top: [Ca²⁺]_m for a single myocyte, superfused with medium containing 1 µM NE; bottom: mechanical response of cell to electrical stimulation. At 1^st arrow, stimulation frequency was increased from 0 to 4 Hz; at 2^nd arrow, cell begins to contract in synchrony with 4 Hz stimulation; at 3^rd arrow, pacing was discontinued. (D) Relation between [Ca²⁺]_m and mean [Ca²⁺]_c during Na⁺-replacement experiments with slow cytosolic and mitochondrial Ca²⁺ uptake. Reproduced with permission from Miyata et al. [137]

In an elaborate study on indo-1/AM loaded myocytes, Zhou et al. [222] used different approaches involving Mn²⁺-quenching, patch-clamping with whole cell dialysis and measurement of cell shortening as a biosensor for [Ca²⁺]_c to determine the kinetics of mitochondrial Ca²⁺ uptake at physiological [Ca²⁺]_c in ferret and cat cardiac myocytes. [Ca²⁺]_c was increased by voltage steps from −40 to +110 mV, inducing a relatively slow I_NCX-induced Ca²⁺ influx that lasted several seconds. In response to cytosolic Ca²⁺ transients, there was no detectable increase of [Ca²⁺]_m as long as resting [Ca²⁺]_c was below a threshold of ∼300–500 nM [222]. However, above 500 nmol/L, a similar absolute change of [Ca²⁺]_c did increase [Ca²⁺]_m, confirming the threshold phenomenon and positive cooperativity of the mCU described before [44, 45, 126, 129, 137]. Since [Ca²⁺]_m followed [Ca²⁺]_c with a lag of ∼1 s and a maximum flux across the mCU of ∼10 µM/s, it was concluded that in agreement with the studies by Miyata et al. [137] and Di Lisa et al. [60], this flux would be too small to allow mitochondrial beat-to-beat Ca²⁺-transients during EC coupling.

Although myocyte shortening and thus, the processes of EC coupling were minimally affected by Mn²⁺-treatment [60, 137], it cannot be ruled out completely that mitochondrial Ca²⁺-uptake was affected by Mn²⁺ to some extent. Mn²⁺ competes with Ca²⁺ for the mCU [76, 77] (although patch-clamp studies suggest that the selectivity of the mCU for Ca²⁺ vs Mn²⁺ is > 15-fold [109]) and can be taken up into mitochondria as well [26, 33, 69, 76, 77, 222], where it may inhibit oxidative phosphorylation [70]. Thus, Mn²⁺ could potentially reduce the rate of the mCU, on the one hand by competing with Ca²⁺, on the other hand by reducing Δψ_m. Furthermore, Mn²⁺ might quench mitochondrial indo-1 [222]. Finally, the mNCE is inhibited by Mn²⁺ as well [69, 76, 77, 137], which could affect Ca²⁺ efflux kinetics. To avoid these undesired effects of MnCl₂, Griffiths et al. [73], who had previously used the Mn²⁺-quench method [74], developed the “heat treatment” protocol, in which incubation of indo-1/AM loaded cardiac myocytes at 25 °C for 2.5 h and 37 °C for 1.5 h promoted the loss of cytosolic indo-1 through probenecid-sensitive plasma membrane anion pumps [73]. With this technique, a similar behavior of [Ca²⁺]_m as after MnCl₂-treatment was observed in rat myocytes, i. e., no beat-to-beat transients, but a cumulative increase of [Ca²⁺]_m [73]. Interestingly, in a follow-up study using the same heat treatment protocol on indo-1 loaded myocytes, Griffiths [72] reported that in guinea-pig, but not rat cardiac myocytes, beat-to-beat [Ca²⁺]_m transients could be observed, suggesting that species-dependent differences of mitochondrial Ca²⁺ uptake may exist.

More support for slow mitochondrial Ca²⁺ uptake comes from a very recent study by Sedova et al. [170], who determined [Ca²⁺]_m by incubating cells with fluo-3, which localized primarily to mitochondria. To eliminate cytosolic traces of fluo-3, the cell membrane was permeabilized with digitonin, and cells dialyzed with a Ca²⁺-free “intracellular” solution. To simulate cytosolic Ca²⁺ transients, cells were exposed to rapid ejections of a Ca²⁺ holding solution (up to 100 µM) from a micropipette into the bath solution, yielding a peak [Ca²⁺] of ∼5 µM in the cytosol. To prevent cells from contracting, 2, 3-butanedione monoxime (BDM) was applied. Under these conditions, no beat-to-beat transients of [Ca²⁺]_m could be observed. Nevertheless, mitochondria accumulated Ca²⁺ in a ruthenium red- and FCCP-sensitive manner as extramitochondrial Ca²⁺ ([Ca²⁺]_em) was raised above 500 nM, with a K_0.5 of 4.4 µM [Ca²⁺]_em, again confirming the phenomenon of a threshold for mitochondrial Ca²⁺ uptake [44, 45, 126, 129, 137, 222]. Interestingly, elevation of [Na⁺] from 0 to 40 mM decreased the velocity and amplitude of (slow) mitochondrial Ca²⁺ uptake by about half, indicating that net [Ca²⁺]_m was critically governed by the equilibrium between mCU and mNCE kinetics, even if the mNCE is manyfold slower than the mCU.

Although highly elegant in design, a potential disadvantage of the study by Sedova et al. [170] could be that cytosolic Ca²⁺ transients were simulated by extracellular application of high [Ca²⁺]. This may neglect to some extent the spatio-temporal aspects of EC coupling, including the potential existence of a mitochondrial Ca²⁺ microdomain. Furthermore, BDM inhibits respiration at complex I [167] and could thus affect the electrochemical driving force for mitochondrial Ca²⁺ uptake.

Evidence for fast mitochondrial Ca²⁺ uptake

The first experimental evidence for rapid mitochondrial Ca²⁺ uptake in intact cardiac myocytes comes from the group of Wendt-Gallitelli and Isenberg [100, 213]. By performing electron-probe microanalysis (EPMA) after shock freezing guinea-pig cardiac myocytes at different time points of a voltage-clamp-induced cytosolic Ca²⁺ transient, they observed a mitochondrial Ca²⁺ transient that peaked ∼15–20 ms after the cytosolic transient [213]. This technique, however, allowed determination of total, but not free [Ca²⁺] in the respective compartments. According to their calculations, total [Ca²⁺]_m rose from 0.5 to 1.3 mmol/kg dry weight within ∼45 ms, yielding a flux of ∼100 nmol/s/mg protein across the mCU. Thereafter, [Ca²⁺]_m decayed back to ∼1.3 mmol/kg dry weight within ∼400ms with a rate of ∼36 nmol/s/mg protein [100]. These values were ∼2 orders of magnitude greater than results from isolated mitochondria [76, 77, 79, 129]. Interestingly, a mitochondrial Na⁺ transient was observed with a delay of ∼55 ms after the mitochondrial Ca²⁺ transient, indicating that matrix Ca²⁺ was rapidly replaced by Na⁺ via the mNCE [100]. The decay of matrix [Na⁺] is explained by exchange of Na⁺ against H⁺ via the Na⁺/H⁺-exchanger (mNHE), which dissipates the proton motive force (Δµ_H) and thus, Δψ_m (Fig. 1) [77, 79, 129].

In a different set of experiments, Isenberg et al. [100] performed patch-clamp experiments in guinea-pig myocytes in which [Ca²⁺]_c was determined by indo-1 salt that was introduced via the patch pipette. Cytosolic Ca²⁺ transients were induced by voltage-clamping in the absence and presence of mitochondrial uncouplers to inhibit the mCU. In the presence of the uncouplers, an ∼50% increase of the amplitude of the cytosolic Ca²⁺ transient was observed for ∼1–3 min. By subtracting the cytosolic Ca²⁺ transient in the absence from the one in the presence of uncouplers, they obtained an indirect estimate of a mitochondrial Ca²⁺ transient. It has to be taken into account though that when elevating [Ca²⁺]_c by blocking mitochondrial Ca²⁺ uptake, allosteric Ca²⁺-activation of I_NCX [118, 208] or Calmodulin (CaM) kinase II dependent phosphorylation of phospholamban (PLB) and subsequent increase of SR Ca²⁺-ATPase activity [15, 120] could have accelerated cytosolic Ca²⁺ decay. This may, in turn, overestimate the rate of decay of [Ca²⁺]_m derived from the indirect mitochondrial Ca²⁺ transient [100]. Furthermore, it should be considered that uncoupling of mitochondrial respiration is problematic since a number of important factors involved in EC coupling (ATP, sarcolemmal K_ATP current, Δψ_m, cellular redox state etc.) change in response to uncouplers, which may hamper the interpretation of data. Nevertheless, the data for the first time suggested that cardiac mitochondria take up Ca²⁺ rapidly enough to shape cytosolic Ca²⁺ transients.

In contrast to the results of Isenberg et al. [100, 213], Moravec and Bond [140, 141] were unable to detect mitochondrial Ca²⁺ transients using a similar technique (EPMA). However, instead of single guinea-pig myocytes, they used hamster papillary muscles. In this context it may be important to consider that also Di Lisa et al. [59, 60] were unable to observe beat-to-beat [Ca²⁺]_m transients in hamster myocytes, again raising the question of potential species-differences, as suggested by Griffiths [72]. Another possible explanation for differential results comes from a follow-up study of the Isenberg group [68] on guinea pig myocytes, where Ca²⁺ and Na⁺ rapidly increased in response to cytosolic Ca²⁺ transients in peripheral, but not in central mitochondria.

A series of studies from the Lemasters group [33, 144, 200, 201] investigated mitochondrial Ca²⁺ uptake by confocal microscopy in fluo-3/AM [33, 144]-, rhod-2/AM- [200, 201] and indo-1/AM [144] loaded rabbit cardiac myocytes. In their first study [33], cytosol and mitochondria were equally loaded with fluo-3, and it was assumed that by high spatial resolution, cytosolic and mitochondrial signals were collected specifically without spill-over from the respective other compartment. In the other studies [144, 200, 201], a cold/warm loading protocol was performed to enhance mitochondrial accumulation of the respective dye. Mitochondrial localization was verified by colocalization of fluo-3 signals with fluorescence from tetramethylrhodamine methyl ester (TMRM), a potentiometric dye that localizes to mitochondria and is an indicator of Δψ_m, or rhodamine 123 [33, 144]. Using this technique, rapid mitochondrial Ca²⁺ transients were observed, with more or less similar kinetics as cytosolic Ca²⁺ transients. β-Adrenergic stimulation increased amplitudes of both cytosolic and mitochondrial Ca²⁺ transients [33, 144]. In one study [144], quantification of mitochondrial Ca²⁺ transients with indo-1/AM revealed an increase of [Ca²⁺]_m from ∼200 nmol/L during diastole to 500 or 700 nmol/L during systole in the absence and presence of β-adrenergic stimulation, respectively [144]. In the most recent study [200], [Ca²⁺]_c and [Ca²⁺]_m were measured by simultaneous recordings of (cytosolic) fluo-3 and (mitochondrial) rhod-2 fluorescence. The specific localization was verified by inhibiting mitochondrial, but not cytosolic Ca²⁺ transients with the mCU-inhibitor ruthenium red [200].

Due to the relatively low temporal resolution in these studies, no discrimination of cytosolic and mitochondrial Ca²⁺ kinetics could be obtained. Another potential disadvantage is that the spatial resolution of a confocal microscope (∼0.9–1.2 µm) may not be high enough to detect fluorescence exclusively from a single mitochondrion and thus, contamination of the mitochondrial signal by cytosolic dye may occur [95], especially when considering that myocytes move during a twitch, and no attempts were made to eliminate signal contamination by cytosolic dye. On the other hand, the fact that mitochondrial, but not cytosolic Ca²⁺ transients were inhibited by ruthenium red [200] indicates that rhod-2 and fluo-3 monitored Ca²⁺ quite selectively from the respective compartments.

To avoid contamination of mitochondrial Ca²⁺-signals with cytosolic traces of a Ca²⁺-indicator, we recently developed a technique (modified from the approach of Zhou et al. [222]) in which we first loaded adult guinea-pig myocytes with cell-permeable rhod-2/AM, and then patch-clamped these cells in order to wash out cytosolic traces of rhod-2 by whole-cell dialysis [117]. At the same time, the cell-impermeable Ca²⁺-indicator indo-1 (penta-K⁺ salt) was introduced via the patch pipette. This allowed us to specifically monitor [Ca²⁺]_c and [Ca²⁺]_m within the same cell. Using a voltage-clamp protocol, we induced cytosolic Ca²⁺ transients by step depolarizations from −80 to + 10 mV at different frequencies. The main results were that with every cytosolic Ca²⁺ transient, a rapid mitochondrial Ca²⁺ transient occurred, its amplitude (Δ) linearly correlating with the amplitude of Δ[Ca²⁺]_c (Fig. 6A–D). In contrast to most previous studies [33, 68, 72, 100, 144, 161, 200, 201, 222], the upstroke of [Ca²⁺]_m preceded the upstroke of [Ca²⁺]_c by several ms, whereas the decay of [Ca²⁺]_m was ∼2.5-fold slower than of [Ca²⁺]_c. In response to β-adrenergic stimulation or an increase of stimulation frequency, this led to a stepwise accumulation of diastolic [Ca²⁺]_m (Fig. 6A–D). Mitochondrial Ca²⁺-uptake could be inhibited by intracellular application of Ru360 (Fig. 6F), and the Ca²⁺-decay was slowed by CGP-37157 [117]. A drawback of our study is that since [Ca²⁺]_m was determined by rhod-2, which is not a ratiometric dye as indo-1, no direct quantification of [Ca²⁺]_m was obtained.

Fig. 6.

(A-D) Representative experiment determining [Ca²⁺]_c and [Ca²⁺]_m in the same cell. [Ca²⁺]_c transients were elicited at 1 Hz in the absence and presence of isoproterenol (Iso, 1–100 nmol/L; (A), (C)). The [Ca²⁺] transients in (B) and (D) were recorded at the indicated time points of the protocol (arrows in (A), (C)). (E) and (F) Cytosolic (E) and mitochondrial Ca²⁺-transients (F) in the absence and presence of Ru360 (10 nmol/L)

A whole different approach to circumvent cytosolic contamination with Ca²⁺-indicators was chosen by Robert et al. [161], who transfected neonatal cardiac myocytes with the Ca²⁺-sensitive photoprotein aequorin, specifically targeted to the cytosol or mitochondria, respectively. With this technique they also observed beat-to-beat oscillations of [Ca²⁺]_c and [Ca²⁺]_m in spontaneously contracting neonatal myocytes (∼1 Hz), with [Ca²⁺]_c peaking after 50–100 ms, and [Ca²⁺]_m peaking with ∼50 ms delay. The Ca²⁺ decay was ∼100–150 ms slower in mitochondria than in the cytosol. In response to increasing extracellular [Ca²⁺] or β-adrenergic stimulation, an increase of amplitude and/or frequency of Ca²⁺-oscillations was observed in both compartments. Similar results were obtained with the green fluorescent protein- (GFP-) based fluorescent Ca²⁺-indicator “ratiometric-pericam” [161]. Bell et al. [12], recently used this technique in adult rat cardiac myocytes, which were adenovirally infected with targeted aequorin to selectively monitor [Ca²⁺]_c and [Ca²⁺]_m. Agreeing with the results on neonatal cells [161], they observed beat-to-beat oscillations of [Ca²⁺]_c and [Ca²⁺]_m during electrical field stimulation. [Ca²⁺]_m increased simultaneously with [Ca²⁺]_c, but decreased more slowly [12]. The strength of these studies [12, 161] is the compartmental specificity of the Ca²⁺-signals, which was illustrated by permeabilization experiments. However, a disadvantage is the relatively low signal-to-noise ratio of the fluorescence signals that necessitates the recording of a whole optical field with multiple cardiac myocytes.

Evidence for a mitochondrial Ca²⁺ microdomain

To functionally address whether rapid mitochondrial Ca²⁺ uptake is achieved due to a mitochondrial Ca²⁺ microdomain, Sharma et al. [176] and Szalai et al. [191] loaded rat cardiac myocytes or cardiac myotubes (H9c2 cells) with rhod-2/AM and eliminated cytosolic dye by chemical membrane permeabilization. [Ca²⁺]_c or perimembrane [Ca²⁺] were monitored by dialyzing the per-meabilized cells with fura-2 salt or fura-2 C18, respectively. Application of caffeine induced rapidly increasing Ca²⁺ transients in both the cytosol and mitochondria, with the decay of [Ca²⁺]_m being slower than of [Ca²⁺]_c. Interestingly, when buffering [Ca²⁺]_c with BAPTA or EGTA [191], cytosolic, but not mitochondrial Ca²⁺ transients were inhibited, suggesting that either the size of the Ca²⁺-complexing molecules or their Ca²⁺-complexing kinetics prevented buffering of highly localized and short-lived increases of [Ca²⁺] in the space between RyR2 and mCU (hot spots; Fig. 1 and 2).

Furthermore, a more rapid increase of [Ca²⁺]_c facilitated a greater increase of [Ca²⁺]_m compared to a slower increase of [Ca²⁺]_c with similar amplitude, indicating that the rate of SR Ca²⁺ release has an important impact on mitochondrial Ca²⁺ uptake [176]. To match the velocity of mitochondrial Ca²⁺ uptake after caffeine-induced SR Ca²⁺ release, a [Ca²⁺] of ∼30 µM had to be established in the cytosol of permeabilized cells, suggesting that the local [Ca²⁺]_c around the mCU may be several-fold higher than in the bulk cytosol, where it peaks at ∼1–2 µmol/L (Fig. 7). The steep part of the concentration-response curve between 10 and 30 µmol/L (Fig. 7B) is in agreement with the K_0.5 of the mCU for Ca²⁺ transport determined in isolated mitochondria (10–20 µM; [54, 76, 77]). In a parallel study from the lab of Hajnoczky, Pacher et al. [147] calculated that ∼26% of the Ca²⁺ released from the SR after caffeine-exposure was taken up by mitochondria. This contrasts to the 1% of Ca²⁺ removed by mitochondria after SR Ca²⁺ release as calculated by Bassani et al. [9].

Fig. 7.

Full activation of mitochondrial Ca²⁺ uptake during RyR-mediated Ca²⁺ release. (A) Ca²⁺- and caffeine induced [Ca²⁺]_m and Δψ_m responses were measured in permeabilized myotubes using rhod-2 and TMRM, respectively. Uncoupler (1 µM CCCP and 2.5 µg/ml oligomycin) was added as indicated. (B) The rate of mitochondrial Ca²⁺ uptake was measured at varying [Ca²⁺] obtained by the addition of CaCl₂ in adherent rhod-2 loaded permeabilized cells. The added CaCl₂ concentration values are indicated with arrows below the x axis, and the effective [Ca²⁺]_c were calculated. To prevent Ca²⁺-induced Ca²⁺-release via RyR2, the cells were preincubated with thapsigargin to inhibit the SR Ca²⁺ ATPase and hence to deplete the SR prior to Ca²⁺ addition. Reproduced from Szalai et al. [191] with permission

To test the existence of a mitochondrial Ca²⁺ microdomain in silico, we used a computer model that integrates the kinetics of mitochondrial Ca²⁺ handling (i. e., mCU and mNCE), the TCA cycle and oxidative phosphorylation [45]. Pulses of Ca²⁺ resembling those expected to be encountered in a mitochondrial Ca²⁺-microdomain [174, 176, 191] ([Ca²⁺]_MD) were simulated with an exogenous Ca²⁺ spike function with peak [Ca²⁺] of either 10 or 20 µM for a duration of 50 ms (Fig. 8D). These pulses of [Ca²⁺]_MD elicited rapid [Ca²⁺]_m transients with slow decay. An increase of either [Ca²⁺]_MD amplitude (Fig. 8A) or frequency (Fig. 8B) independently increased diastolic [Ca²⁺]_m. An increase in both amplitude and frequency of the pulses potentiated diastolic [Ca²⁺]_m accumulation (Fig. 8C). These simulations were in qualitative agreement with experimental results [12, 117, 144, 161] and lend support to the hypothesis that mitochondria sense [Ca²⁺] in a microdomain, rather than the bulk [Ca²⁺]_c.

Fig. 8.

Computational modeling of [Ca²⁺]_m, assuming a microdomain (MD) in which mitochondria are exposed to [Ca²⁺] pulses of 10–20 µmol/L for 50 ms (D; representative pulse taken from t = 215 s in A). Starting from steady-state conditions with [Ca²⁺]_MD oscillating from 0.1 to 10 µmol/L at 1 Hz, an increase of amplitude (A, from 10 to 20 µmol/L), frequency (B, from 1 to 2 Hz) or both (C) of [Ca²⁺]_MD transients were simulated. Reproduced from Maack et al. [117] with permission

Implications of rapid mitochondrial Ca²⁺ uptake for EC coupling

Several studies indicate that by buffering [Ca²⁺]_c, rapid mitochondrial Ca²⁺ uptake directly affects EC coupling [61, 100, 117, 119, 124, 147, 148, 171]. When blocking mitochondrial Ca²⁺ uptake, an increase of cytosolic Ca²⁺ transients was observed (Fig. 6E and F; [100, 117, 124, 171]). It also appears that the highly organized pattern of mitochondria localized close to the SR provides a “local control” of SR Ca²⁺ release events. In cardiac myotubes, spontaneous Ca²⁺ release events restricted to a small number of RyRs (Ca²⁺ sparks) elicited miniature mitochondrial matrix Ca²⁺ signals (Ca²⁺ marks) that lasted less than 500 ms [148] and roughly resembled the kinetics of mitochondrial Ca²⁺ uptake and release kinetics recorded under steady-state conditions during whole-cell cytosolic Ca²⁺ transients (Fig. 6D; [12, 117]). Accordingly, localized Ca²⁺ sparks induced transient focal depolarizations of Δψ_m by ∼10–15 mV [61]. The propagation of caffeine-induced Ca²⁺ sparks was suppressed by mitochondria, since inhibition of the mCU (with Ru360) resulted in regenerative Ca²⁺ waves across the whole cell [171]. Similarly, in atrial myocytes, where cytosolic Ca²⁺ transients occur primarily in the periphery due to the lack of t-tubules, uptake of Ca²⁺ into mitochondria and the SR prevented the propagation of the peripheral Ca²⁺ transients to more central parts of the cell [119], even though the density of mitochondria is lower in atrial than in ventricular cells [171]. Taken together, these results suggest that mitochondria may function as spatial buffers for Ca²⁺, regulating the propagation of Ca²⁺ release events and thus, contractility of the cell. However, regarding the controversy on the amount of Ca²⁺ taken up by mitochondria during EC coupling [9, 117, 147], the definite relevance of mitochondrial Ca²⁺ uptake for cytosolic Ca²⁺ homeostasis is still incompletely understood.

Molecular identity of the mCU

So far, the mCU has been characterized functionally, but not on a molecular level. Thus, most studies rely on the mCU’s sensitivity to Ru360 or ruthenium red. Sparagna et al. [186] observed a rapid mode of uptake, termed RaM, that allowed rapid mitochondrial Ca²⁺ uptake typically at the first of a series of Ca²⁺ pulses in liver mitochondria. The RaM was less sensitive to ruthenium red than the mCU. However, in cardiac mitochondria, the RaM required > 30 s to become fully reactivated at extramitochondrial [Ca²⁺] < 100 nmol/L [30] and thus, cannot be responsible for beat-to-beat mitochondrial Ca²⁺ uptake. Furthermore, Beutner et al. [18, 19] have proposed that mitochondrial RyR (RyR1 subtype) channels may mediate Ca²⁺ flux across the inner mitochondrial membrane. The binding of [³H]ryanodine to this mRyR1 has a bell-shaped Ca²⁺-dependence, with an optimum between 10–30 µM [19], the latter value comparable to the reported K_0.5 of the mCU for Ca²⁺ [54, 77]. Currently, there is no specific inhibitor that could differentiate between the mCU and mRyR1. Ruthenium red inhibits the mCU [124], but also the mRYR1 [19] and the RyR2 subtype [124]. In contrast, Ru360 inhibits the mCU, but not the RyR2 subtype [124]. However, whether the mRyR1 is inhibited by Ru360 has not been determined [19]. Thus, it may be possible that the mRyR1 contributes to mitochondrial Ca²⁺-uptake, or may even resemble the mCU.

Bioenergetic consequences of mitochondrial Ca²⁺ uptake

A series of studies by Brandes and Bers [23–27] provided important insights into the physiological role played by mitochondrial Ca²⁺ uptake with respect to the regulation of respiration. In isolated rat cardiac trabecule loaded with rhod-2/AM, [Ca²⁺]_m and NADH were determined in response to an abrupt change of workload (Fig. 9A; [26]). When increasing stimulation frequency from 0.5 to 2 Hz, a rapid initial oxidation of NADH occurred (undershoot). This oxidation was attributed to pronounced ATP hydrolysis resulting in an increase of ADP, which stimulated ATP-generation at the F₁/F₀-ATPase [25]. Since ATP synthesis is associated with dissipation of Δψ_m, more electrons need to be donated by NADH to the respiratory chain to maintain Δψ_m (Fig. 4). After the initial oxidation, a phase of slow NADH recovery was observed, which reached a new steady state after several seconds. The time course of NADH recovery was closely associated with the accumulation of [Ca²⁺]_m (Fig. 9A, lower panel), and thus related to Ca²⁺-induced stimulation of TCA cycle dehydrogenases [26]. After returning to 0.5 Hz stimulation frequency, an overshoot of NADH was observed, which is explained by maintained elevation of TCA cycle dehydrogenase activities despite decreased ADP-stimulated respiration [26]. NADH again slowly recovered towards baseline levels in parallel to the decay of [Ca²⁺]_m.

Fig. 9.

Time-dependent behavior of [Ca²⁺]_m and NADH after changes in workload. (A) Experimental data showing the responses of NADH (upper panel) and [Ca²⁺]_m (lower panel) to an increase in stimulation frequency from 0.5 to 2 Hz in a rat cardiac trabecula. (B) NADH (upper panel) and [Ca²⁺]_m (lower panel) in response to an increase of stimulation frequency from 0 to 4 Hz in the presence of isoproterenol (100 nM). (C) Computational modeling of NADH and [Ca²⁺]_m by integrating the kinetics of EC coupling, mitochondrial Ca²⁺ transporters and energetics in response to a change in stimulation frequency from 0.25 Hz to 2 Hz (for parameters, see [46]). I, undershoot; II, recovery, and III, overshoot of NADH. Reproduced from Brandes and Bers (A; [26]), Maack et al. (B; [117]) and Cortassa et al. (C; [46]) with permission

Similar results were obtained in isolated cardiac myocytes (Fig. 9B; [101, 117]), albeit with different relative amounts of the initial oxidation- and the secondary recovery phases, respectively. These differences may be related to the fact that in prestretched cardiac trabeculae [26], work during contraction may be severalfold higher than in unstretched single myocytes. Thus, ADP-related oxidation of NADH was less pronounced in single cells [101, 117], whereas the Ca²⁺-dependent NADH recovery was comparable between trabeculae [26] and single cells [101, 117], resulting in net reduction of the NADH/NAD⁺ pool after increasing stimulation frequency in single cells (Fig. 9B; [101, 117]). Jo et al. [101] corroborated the concept that NADH oxidation at the onset of work is related to ADP-induced stimulation of respiration, since addition of ADP or P_i oxidized NADH in permeabilized myocytes. Furthermore, the NADH recovery phase was inhibited by blocking mitochondrial Ca²⁺ uptake with Ru360 [101] during electrical stimulation, confirming the Ca²⁺-dependence of this process. In an earlier set of studies, White and Wittenberg [214, 215] had observed that in response to 5 Hz stimulation, NADH was oxidized at 95% O₂, whereas it was reduced at 20% O₂. They concluded that if O₂ availability limits the rate of respiration (i. e., at 20% O₂), then the equilibrium of the NADH/NAD⁺ redox potential is shifted towards NADH reduction by Ca²⁺-stimulated TCA cycle dehydrogenases, and away from respiratory chain-induced oxidation [214, 215].

The transient behavior of NADH determined experimentally [23, 25, 26, 101, 117, 214, 215] could be reproduced by computational modeling in different studies [45, 46, 101]. In our recent model that integrates the processes of EC coupling and mitochondrial bioenergetic responses of the cardiac myocyte [46], the transient changes of NADH and [Ca²⁺]_m in response to a change of workload (Fig. 9C) closely resembled the experimental results (Fig. 9A and B). Moreover, the model predicted the changes in mitochondrial ADP (ADP_m), cytoplasmic ATP (ATP_i), phosphocreatine (PCr) and average oxygen consumption (Av V_O2) during the protocol (Figs. 5 and 8 in Cortassa et al. [46]). After the step-increase of work, both ADP_m and Av V_O2 abruptly increased in parallel to the initial oxidation of NADH. Since high energy phosphates were buffered by cytoplasmic PCr, the equilibrium of the reaction ATP ↔ ADP + P_i was shifted ∼400-fold to the left and thus, the relative decrease of ATP after the step-increase of workload was comparably small (by 0.4% from 7.99 to 7.96 mM). This is consistent with the observation by Balaban et al. [8] who did not detect considerable changes in cellular ATP in response to up to 5-fold changes in workload in the dog heart. Due to increased ATP consumption, however, the PCr pool was continuously reduced by ∼15% [46].

To test the relative contribution of ADP and [Ca²⁺]_m to the changes in respiration and NADH, distinct modifications of the model parameters were performed [46]. When halfing the rate of ATP hydrolysis by the myofibrillar ATPase (Fig. 10 B), the undershoot and overshoot of NADH were substantially reduced, while the steady-state NADH after equilibration at 2 Hz was elevated compared to control conditions (Fig. 10A). As mentioned above, this may explain why in (unstretched) cells, steady-state NADH after an increase in stimulation frequency was at a more reduced level than in (prestretched) isolated trabeculae (Fig. 9A and B; [26, 101, 117]). Conversely, when reducing the maximal velocity of the mCU by 90%, mitochondrial Ca²⁺ uptake was substantially reduced (Fig. 10D), oxidation of NADH pronounced and the recovery of NADH abolished (Fig. 10C).

Fig. 10.

Effect of changing the actomyosin- (AM-) ATPase and mCU on the computational model (described by Cortassa et al. [46]; compare Fig. 9C) on energetic behavior after changes in workload. (A) The behavior of NADH with the maximal rate of ATP hydrolysis by the AM-ATPase being decreased to onehalf (solid trace; V_max/AM = 3.6·10⁻³ mM ms⁻¹) of the control value (shaded trace). (B) The profile of average ATP_i for changes in workload under control conditions (shaded trace) or with decreased AM-ATPase activity (solid trace). (C) The NADH profile with the maximal rate of the mCU decreased to 1/10^th (solid trace; V_max/mCU = 2.75·10⁻³ mM ms⁻¹) of the control value (shaded trace). (D) The profile of [Ca²⁺]_m accumulation for the low mCU condition described in (C) (solid trace) as compared to the control (shaded trace). Reproduced from Cortassa et al. [46] with permission

Taken together, both experimental results and computational modeling indicate that in response to an increase of workload, the immediate increase in respiration is mediated by ADP, while the resulting oxidation of NADH is recovered by Ca²⁺-induced stimulation of TCA cycle enzymes. Thus, both ADP and Ca²⁺ contribute to the energetic adaptation that provides constant ATP/ADP and NADH/NAD⁺ ratios in the cell.

Pathophysiological aspects

Defects in EC coupling in chronic heart failure

In chronic heart failure, contractile dysfunction is related to structural remodeling of the left ventricle and significant defects in EC coupling [13, 84, 94]. A major deficit in failing myocytes is the reduced Ca²⁺ content of the SR, which is related to decreased expression and activity of the SR Ca²⁺-ATPase [84, 92, 146, 153] and an increased Ca²⁺ leak of the RyR2 due to hyperphosphorylation [84, 94, 112, 122]. These defects, potentially aggravated by L-type Ca²⁺ channel dysfunction [41, 71, 81, 86, 113, 131, 146, 184] or t-tubular derangement [32, 86, 115, 145, 184], lead to a smaller and more dyssynchronous SR Ca²⁺ release during an AP, resulting in slower rates of increase and decay of [Ca²⁺]_c, but higher diastolic [Ca²⁺]_c compared to normal cardiac myocytes [17, 81, 113, 115, 146, 184]. Decreased SR Ca²⁺-ATPase activity is partly compensated by increased expression and activity of the NCX [16, 65, 93, 146, 178, 190], since pronounced forward mode I_NCX may maintain diastolic function [85] by removing Ca²⁺ to the extracellular space. On the other hand, this pronounced forward mode I_NCX may further aggravate SR Ca²⁺ depletion, since inhibition of I_NCX with an inhibitor peptide (XIP [118]) restored SR Ca²⁺ load in failing myocytes [91].

In heart failure, but also cardiac hypertrophy, [Na⁺]_i is elevated by ∼3–6 mM (for review, see [156]). The physiological range of [Na⁺]_i in resting cardiac myocytes is ∼5–15 mM and is species-dependent: While in small animals (i.e., mice and rats), [Na⁺]_i is between 10 and 15 mM, larger animals and humans have [Na⁺]_i between 5 and 10 mM [154]. [Na⁺]_i rises by ∼3–5 mM in response to an increase of stimulation frequency [58, 154, 155]. In humans, [Na⁺]_i is 8 mM in normal resting myocytes and elevated to 12 mM in cells from patients with heart failure [155]. Similar absolute increases of [Na⁺]_i were observed in various animal models of cardiac failure [5, 58, 154] and hypertrophy [154, 156].

The underlying mechanisms for elevated [Na⁺]_i are incompletely understood, but may involve a decrease in Na⁺/K⁺-ATPase activity [58, 156, 169, 172, 175, 206], enhanced Na⁺/H⁺-exchanger (NHE) activity [2, 6, 40, 142], or an increase in a tetrodotoxin-sensitive persistent (late) I_Na [58, 121, 203–205]. During the AP, increased [Na⁺]_i facilitates repolarization and pronounced cytosolic Ca²⁺-influx via reverse-mode I_NCX, which partly compensates the impaired SR Ca²⁺-release and contractility in failing myocytes [3, 58, 153, 210, 212]. In this context, the elevation of [Na⁺]_i in cardiac failure and hypertrophy may be regarded as a beneficial and compensatory mechanism [94]. On the other hand, cariporide, an inhibitor of the NHE, reduced [Na⁺]_i and improved cytosolic Ca²⁺ handling [5] and LV remodeling [6] in a rabbit heart failure model. Similarly, cariporide completely prevented the development of hypertrophy, fibrosis and heart failure in the β₁-AR overexpressing mouse [62]. These data indicate that elevated [Na⁺]_i, besides its compensatory effects on SR Ca²⁺ load, may actually play a maladaptive role in the development of cardiac hypertrophy and failure. The precise mechanisms for such a maladaptive role, however, are currently unclear.

Are defects in EC coupling linked to energy starvation in heart failure?

Besides defects in EC coupling, the failing heart is energy-starved [99, 187, 211]. The total cellular levels of PCr, but also NAD and adenine nucleotides, are reduced in patients with heart failure [11, 99, 188]. Such changes already occur at the stage of cardiac hypertrophy [99]. In patients with heart failure, the decreased ratio of PCr/ATP predicts an adverse outcome [143]. As a more dynamic parameter of energy turnover, in vivo ATP flux through creatine kinase was reduced by 50% in patients with mild-to-moderate heart failure [211].

The mechanisms for the energetic deficits are manifold. As has been reviewed previously [187], mitochondria from failing hearts are characterized by more frequent membrane disruption and matrix depletion, a lower capacity for respiration with a variety of substrates, defects in complexes of the electron transport chain (ETC), and a decreased capacity for oxidative phosphorylation. Furthermore, defects in substrate metabolism such as downregulation of fatty acid oxidation and increased glycolysis and glucose oxidation may contribute to the energetic deficit [187]. Finally, mitochondria are the source for an increased production of ROS in failing hearts [97, 98]. It has been proposed that increased mitochondrial ROS-production may lead to defects in mitochondrial DNA, which could contribute to reduced expression of ETC complexes in heart failure [96].

In a recent study from our lab [117], we investigated the hypothesis that an elevation of [Na⁺]_i, as observed in heart failure, may have an adverse effect on mitochondrial energetics. Since mitochondrial Ca²⁺ efflux is governed by the mNCE with a K_0.5 for [Na⁺]_i in the range of ∼8 mM, an increase of [Na⁺]_i as observed in heart failure may accelerate the rate of mitochondrial Ca²⁺ efflux and thus, decrease steady-state [Ca²⁺]_m. Indeed, in isolated mitochondria, increasing extramitochondrial [Na⁺] decreased matrix Ca²⁺ accumulation, Ca²⁺-stimulated activity of TCA cycle enzymes (Fig. 11), NADH production and rate of oxidative phosphorylation, while inhibition of mNCE eliminated the Na⁺ sensitivity of these parameters [47, 55]. Accordingly, in voltage-clamped guinea-pig cardiac myocytes, an increase of [Na⁺]_i from 5 to 15 mM reduced the amplitude of rapid mitochondrial Ca²⁺ transients during EC coupling by accelerating mitochondrial Ca²⁺ decay [117]. As a result, steady-state accumulation of [Ca²⁺]_m was substantially reduced (Fig. 12A and B; [117]). Physiologically, [Na⁺]_i is elevated in species with higher heart rates (e. g., mice or rats [154]). Furthermore, [Na⁺]_i generally increases at an increase of heart rate in many different species [58, 154, 155]. In this context, the elevation of [Na⁺]_i could potentially represent a physiological, protective mechanism by inhibiting mitochondrial Ca²⁺-overload and fatal opening of the PTP [29]. On the other hand, pathophysiologically, when [Na⁺]_i is chronically elevated in species with normally low [Na⁺]_i (and slower heart rates), this may have adverse effects on mitochondrial energetics.

Fig. 11.

Stimulation of α-ketoglutarate oxidation by Ca²⁺ in the absence and presence of Na⁺ (15 mM) and the mCU-inhibitor ruthenium red (RR). Reproduced from Denton et al. [55] with permission

Fig. 12.

[Ca²⁺]_c (A), [Ca²⁺]_m (B) and reduced NADH (C) in guinea-pig cardiac myocytes paced at 3 Hz and exposure to 10 and 100 nM isoproterenol, with either 5 or 15 mM [Na⁺] in the pipet. Reproduced from Maack et al. [117] with permission

The transient changes of NADH in response to an abrupt increase of work (as initially characterized by Brandes and Bers (Fig. 9; [23–27]) were affected by acutely elevated [Na⁺]_i in a way that the initial oxidation of NADH due to ADP-induced stimulation of respiration was pronounced, whereas the (counterbalancing) recovery of NADH (related to dehydrogenation by the TCA cycle) was substantially impaired [117]. As a result, the steady-state redox potential of NADH/NAD⁺ was oxidized at 15 compared to 5 mM [Na⁺]_i (Fig. 12C; [117]). Since the rate of respiration and thus, ATP production, depends on the availability of electrons to the respiratory chain in the form of NADH (Fig. 4; [45]), these results suggest that elevated [Na⁺]_i may critically hamper the matching process of energy supply and demand in cardiac cells.

Accordingly, in a study by Di Lisa et al. [59] on field-stimulated myocytes of the cardiomyopathic hamster, the amplitude of cytosolic Ca²⁺ transients and subsequently, mitochondrial Ca²⁺ accumulation was reduced compared to control myocytes (488 nM versus 830 nM, respectively). This was associated with reduced pyruvate dehydrogenase (PDH) activity, whose K_m for Ca²⁺ activation is in the range of ∼650 nM to 1 µM (Fig. 3B; [54, 56, 57, 59, 127, 128]). Thus, both the reduced amplitude of cytosolic Ca²⁺ transients [59] and elevated [Na⁺]_i [117] may hamper Ca²⁺-activation of TCA cycle enzymes in heart failure. Furthermore, glycolysis and, thus, the availability of pyruvate as a substrate for the TCA cycle was reduced in myopathic compared to control hamsters, resulting in more oxidized NADH and decreased phosphorylation potential and developed LV pressure [4]. Application of pyruvate in this model improved both energetic and hemodynamic parameters [4]. Accordingly, application of pyruvate to human heart failure patients in vivo increased LV systolic and diastolic function [87, 88], which was related to an increase of SR Ca²⁺ load and thus, cytosolic Ca²⁺ transient amplitude in vitro [82, 89].

Clinical implications

Taken together, these results suggest that in chronic heart failure, defects in EC coupling and glycolysis may trigger decreased supply of the respiratory chain with NADH. However, energy demand in the failing heart, despite reduced systolic Ca²⁺ transients and contractile force may be even higher than in the normal heart. First of all, due to sympathetic activation, heart rate is elevated in patients with heart failure. Furthermore, β-adrenergic activation decreases the energetic economy of contraction in human failing myocardium [83]. An elevation of stimulation frequency increases diastolic tension (and reduces systolic force) in failing myocardium with decreased SR Ca²⁺-ATPase expression and a lack of upregulation of the NCX [85]. Diastolic force development, however, is occurring at the same energy expenditure as systolic force generation in human failing myocardium [132]. If indeed, higher energy consumption and, thus, ATP hydrolysis occurs in failing myocardium, but with an insufficient compensatory increase of NADH production by the TCA cycle, this may explain the pronounced oxidation of the NADH-pool in failing hearts [4], which may eventually result in a mismatch between ATP supply and demand. It has been suggested that such a mismatch precedes the reduction in the total cellular levels of PCr in patients with heart failure [11, 99, 188].

Besides the PCr system, which buffers high energy phosphates in the cytosol and in close spatial association with ATPases [39], direct ATP/ADP channeling also exists between mitochondria and the main sites of energy consumption, i. e., SR Ca²⁺ ATPase and myofilaments [104]. Furthermore, ATP generation from SR-associated glycolytic enzymes supports SR Ca²⁺-ATPase activity [219]. The SR Ca²⁺-ATPase, in fact, seems to be the most sensitive ATPase in response to a drop in the free energy released from ATP hydrolysis (ΔG_∼p), which may directly affect cardiac contractility [105, 130, 196, 197, 199]. This implies that decreased ΔG_∼p may directly translate into defects in EC coupling and contractile function [166, 196–199].Although it is difficult to determine cause and effect as the pathological conditions evolve, a vicious cycle may develop that leads to progressive cardiac dysfunction and energy starvation (Fig. 13). Interrupting this vicious cycle of defective EC coupling and energy starvation may be a novel future strategy to improve LV function, and potentially the prognosis of patients with heart failure.

Fig. 13.

Hypothetical vicious cycle (and potential targets for therapeutic interventions) of defects in EC coupling and energy depletion in chronic heart failure. PCr phosphocreatine; PLB phospholamban; SERCA SR Ca²⁺-ATPase

In this context, approaches aimed directly at improving defective EC coupling or mitochondrial energetics (Fig. 13) without activating receptor-mediated pathways (e. g., β-adrenergic receptors) could be promising. In experimental heart failure models, transgenic inhibition or deletion of phospholamban [51, 116, 134] or adenoviral increase of SR Ca²⁺-ATPase expression [52] frequently, but not always [185], improved LV function and survival. Interestingly, the increase of SR Ca²⁺-ATPase expression was associated with an improvement of the PCr/ATP ratio in rats with heart failure after aortic banding [52], supporting the hypothesis that reconstitution of EC coupling may improve energetics. However, a specific pharmacological drug for such an intervention is currently not available.

Another concept is to inhibit forward mode I_NCX to increase SR Ca²⁺ load [91], however, this has not yet been tested in heart failure models in vivo. A traditional pharmacological approach to increase cytosolic Ca²⁺ transients is to inhibit Na⁺/K⁺-ATPase with digitalis, which increases [Na⁺]_i and thus, I_NCX-mediated Ca²⁺ influx. However, digitalis does not improve survival despite beneficial effects on morbidity in patients with heart failure [75]. One could speculate that adverse energetic effects by the concomitant increase of [Na⁺]_i (and consequent effects on mitochondrial Ca²⁺ handling) could explain this lack of prognostic benefit [6, 117]. Furthermore, besides inhibiting Na⁺/K⁺-ATPase, digitalis may have other yet unknown effects improving contractility [207, 218].

Thus, in the light of a potential maladaptive role of elevated [Na⁺]_i in heart failure, it may be a promising approach to reduce cellular Na⁺-influx. This could be achieved by inhibiting I_NHE or late I_Na. Both approaches were cardioprotective in experimental models of ischemia/reperfusion, since in this situation, an excessive increase of [Na⁺]_i contributes to progressive cytosolic and mitochondrial Ca²⁺-overload and thus, the induction of apoptosis through activation of the mitochondrial PTP [78, 135, 168, 189]. Also in animal models of heart failure, both the inhibition of I_NHE by cariporide and of late I_Na by ranolazine reduced [Na⁺]_i and improved EC coupling and LV remodeling [6, 34, 38, 62, 163, 202]. In conditions of cardiac ischemia/reperfusion, inhibition of I_NHE preserved mitochondrial energetics (i.e., ATP and PCr content) [111, 162]. However, these effects were not related to an improvement of EC coupling, but rather to a delay of mitochondrial matrix acidification by a mechanism unrelated to the sarcolemmal NHE [162]. Furthermore, beneficial effects of ranolazine on mitochondrial energetics have so far been related to its ability to decrease fatty-acid oxidation, promote glucose oxidation, and by increasing pyruvate dehydrogenase complex activity [1, 125, 192] rather than by improving EC coupling. Thus, at this point, it is unknown whether reducing [Na⁺]_i per se improves mitochondrial energetics, especially in heart failure.

A novel target to improve mitochondrial energetics and cardiac function in heart failure or ischemia/reperfusion could be to inhibit the mNCE. While elevated [Na⁺]_i reduced steady-state [Ca²⁺]_m, TCA cycle activity and rate of oxidative phosphorylation, inhibiting the mNCE with CGP-37157 improved mitochondrial Ca²⁺-accumulation and energetics in isolated mitochondria and cardiac myocytes [47, 114, 117]. Furthermore, during ischemia and reperfusion with cytosolic Ca²⁺ overload and a collapse of Δψ_m, the mNCE in reverse mode has been shown to contribute to mitochondrial Ca²⁺ overload [180], which eventually may lead to PTP opening and cell death [29]. In this setting, CGP-37157 inhibited mitochondrial Ca²⁺ overload [180]. Thus, inhibiting the mNCE may improve energetics and potentially protect from apoptosis in various pathologic situations. However, data on the effects of mNCE-inhibition on cardiac energetics in vivo are presently not available. Furthermore, considering that the physiological increase of [Na⁺]_i during an increase of heart rate [58, 154, 155] could potentially prevent mitochondrial Ca²⁺ overload (as discussed in the previous chapter), it remains unclear whether mNCE-inhibition may have adverse effects on energetics under physiological conditions. Clearly, more in vitro and in vivo research is needed to further evaluate the mNCE as a therapeutic target.

In conclusion, the processes of EC coupling and mitochondrial energetics are highly interrelated, and defects in EC coupling may directly translate into defects in mitochondrial energetics in pathological situations. The understanding of the relation between EC coupling and energetics may help to develop new strategies to improve LV function and potentially survival in patients with chronic heart failure.