Sarcoplasmic reticulum-mitochondria communication; implications for cardiac arrhythmia

Abstract

Sudden cardiac death due to ventricular tachyarrhythmias remains the major cause of mortality in the world. Heart failure, diabetic cardiomyopathy, old age-related cardiac dysfunction and inherited disorders are associated with enhanced propensity to malignant cardiac arrhythmias. Both defective mitochondrial function and abnormal intracellular Ca²⁺ homeostasis have been established as the key contributing factors in the pathophysiology and arrhythmogenesis in these conditions. This article reviews current advances in understanding of bidirectional control of ryanodine receptor-mediated sarcoplasmic reticulum Ca²⁺ release and mitochondrial function, and how defects in crosstalk between these two organelles increase arrhythmic risk in cardiac disease.

Graphical Abstract

Figure created with BioRender.com.

1.1. Introduction

The function of cardiomyocyte (CM) contractile machinery depends upon Ca²⁺ released from the sarcoplasmic reticulum (SR) and energy provided by mitochondria in the form of ATP. These two systems must be finely tuned to each other to ensure appropriate cardiac output during the constantly changing metabolic demands of the body. Over the last two decades, significant progress has been made in delineating mechanisms of bidirectional control of mitochondrial function and SR Ca²⁺ release. More recently, disease-related changes in mitochondria-SR crosstalk have become rapidly growing area of research. Evidence suggest that initial defect in either causes aberrant function of the other organelle, creating vicious cycle that aggravates contractile deficiency and increases arrhythmic risk. We will review how aberrant Ca²⁺ release contributes to mitochondrial injury, how defective mitochondria disturb intracellular Ca²⁺ cycling, and potential strategies to interrupt this vicious circle and reduce pro-arrhythmic behavior in cardiac disease.

2.1. Sarcoplasmic reticulum Ca²⁺ homeostasis

The sarcoplasmic reticulum (SR), the major Ca²⁺ storage organelle, occupies only 2~4 % of CM volume. However, given its enormous Ca²⁺ buffering capacity, the SR provides 70–90% of Ca²⁺ to the cytosol during Ca²⁺ release to activate contractile machinery [1]. The SR can accumulate up to 10–40 mM Ca²⁺, with the majority associated with the low affinity high capacity Ca²⁺ buffering molecule, calsequestrin type 2 (CASQ2). The main SR Ca²⁺ release channel in CMs is the ryanodine receptor type 2 (RyR2). The tetrameric 2.2 KDa RyR2 channel is associated with a number of auxiliary regulatory proteins including junctin, triadin, CASQ2, calmodulin, sorcin and a whole array of enzymes that regulate its phosphorylation, nitrosylation and oxidative states [2,3]. The opening of the RyR2 complex is initiated by cytosolic Ca²⁺ in the micromolar range, and can be terminated by a decrease in intraluminal Ca²⁺ either directly, or indirectly due to reduced Ca²⁺ efflux during SR depletion. Activity of RyR2 is also positively regulated by ATP and negatively regulated by intracellular Mg²⁺. The majority of RyR2 complexes are organized in clusters (8–100 channels per cluster) in the junctional SR (jSR) opposite to T-tubular sarcolemmal L-type Ca²⁺ channels (LTCCs). Sarcolemmal depolarization during systole leads to LTCC activation, permitting a small amount of Ca²⁺ to enter the cell. Given very tight organization of T-tubule and jSR contacts in dyads, structures where jSR envelopes T-tubules in a pancake-like fashion [4], the short 10 nm distance between LTCCs and RyR2 clusters allows Ca²⁺ to reach sufficient concentrations to activate RyR2s [5]. During systolic RyR2-mediated SR Ca²⁺ release, [Ca²⁺] in the dyadic space increases up to 100–200 μM [6] and reaches 20–40 μM in the submembrane space, while Ca²⁺ in the cytosol increases from 100 nM at baseline to ~500 nM at its peak [2,7,8]. Under β-adrenergic stimulation, in conditions mimicking stress, systolic Ca²⁺ in cytosol can reach ~4 μM [9,10]. During diastole, the majority of Ca²⁺ is returned to the SR by SR Ca²⁺ ATPase type 2 (SERCa2a) with most of it situated in free SR outside release clusters. Activity of SERCa2a is regulated indirectly by reversible serine/threonine phosphorylation of auxiliary inhibitory polypeptide phospholamban (PLB), and directly via posttranslational modifications such as sumoylation and oxidation [11,12]. After SR Ca²⁺ release, the remainder of Ca²⁺ which was not resequestered by SERCa2a is returned to extracellular milieu by sarcolemmal Na⁺/Ca²⁺ exchanger type 1 (NCX1) to offset LTCC-mediated Ca²⁺ influx. Given that NCX1 brings in 3 Na⁺ per 1 Ca²⁺ ion removed, its activation depolarizes the sarcolemma, which can contribute to triggered activity in the form of early and delayed afterdepolarizations (EADs and DADs respectively). Notably, CMs express another class of SR Ca²⁺ release channels, IP₃ receptors (predominantly type 2, [13]), which can contribute to intracellular Ca²⁺ homeostasis in ventricular myocytes [14].

3.1. Mitochondrial Ca²⁺ homeostasis

Mitochondria are the major energy source in CMs and they occupy up to 35 % of CM volume. Flux of Ca²⁺ in and out of mitochondria is essential for the activity of enzymes involved in the Krebs cycle and ATP generation [15,16]. Over the last decade, significant progress has been made in identifying components of mitochondrial Ca²⁺ handling machinery. The current consensus is that the outer mitochondrial membrane (OMM) is relatively permeable to Ca²⁺, which crosses it through the low selectivity high conductance voltage-dependent anion channel VDAC [17], and to a lesser degree through other non-selective pores. To allow Ca²⁺ to enter the matrix, the inner mitochondrial membrane (IMM) is equipped with multimolecular Ca²⁺ transport complexes, comprised of the mitochondrial Ca²⁺ uniporter (MCU, [18]) and auxiliary regulatory proteins MICU1, MICU2, EMRE, MCUb and MCUR1 (For recent review see [15]). The MCU complex is the major but not the only Ca²⁺ transport pathway into the mitochondria. The leucine zipper EF hand-containing transmembrane protein 1 (LETM1, [19]), uncoupling proteins 2 and 3 (UCP2/3, [20]), the transient receptor potential canonical 3 (TRPC3, [21]), the mitochondrial ryanodine receptor type 1 (RyR1, [22]), and possibly more Ca²⁺ transporters are present in the IMM [23,24]. To remove Ca²⁺ from the matrix, mitochondria express a specific form of the Na⁺/Ca²⁺ exchanger, called the Na⁺/Ca²⁺/Li⁺ exchanger (NCLX, [25]), with intracellular Na⁺ Km of ~8 mM and mito-Ca²⁺ Km of 13 μM [26,27]. In addition, under certain conditions Ca²⁺ can leave via the mitochondrial permeability transition pore (mPTP). Ca²⁺ regulates the threshold of permeability transition activation via the mitochondrial calcium uniporter regulator 1 (MCUR1) [28], thereby self-limiting accumulation of Ca²⁺ in the matrix.

Experiments with isolated CMs showed that the MCU activity/cytosolic Ca²⁺ relationship is steep, and uptake can be initiated when cytosolic [Ca²⁺] reaches 300–500 nM [29]: the typical values for cytosolic Ca²⁺ transient peak amplitudes. Patch clamp studies by Kirichok et al. [17] using isolated cardiac mitochondria from COS7 cells revealed MCU Km for Ca²⁺ is ~20 mM, with extremely high affinity (Kd less or equal ~2 nM) to Ca²⁺, which makes this channel highly capable to bring large amounts of Ca²⁺ into the matrix, even at extremely low cytosolic Ca²⁺ concentrations. In cardiac myocytes with large frequent cytosolic Ca²⁺ elevations, this will invariably result in mito-Ca²⁺ overload (up to 10–100 mM) in the absence of robust mechanisms to limit channel activity, given the large driving force for Ca²⁺ uptake through the highly charged inner mitochondria membrane. The follow-up study by Kirichok’s group revealed that cardiac mito-Ca²⁺_MCU current is 25–30 times smaller than current in the other tissues, including closely related skeletal muscle [30]. The authors suggested that low MCU activity/channel number may prevent mito-Ca²⁺ overload in cells with frequent Ca²⁺ elevations in cytosol. However, this does not explain “threshold”-like behavior when mito-Ca²⁺ accumulation only becomes apparent when Ca²⁺ transients amplitude reach ~300–500 nM, as shown by Zhou et al. [29] and more recently by Wescott et al. [31]. Mallilankaraman et al. [32] suggested that MICU1, a component of the mitochondrial Ca²⁺ uptake complex, may play a “gatekeeper” role, preventing MCU opening at low cytosolic [Ca²⁺] (<3 μM). The works of several laboratories suggested that MCU sensitivity to cytosolic Ca²⁺ may be regulated by MICU1 not only from the cytosolic side but from within the matrix as well, shutting off the channel at cytosolic [Ca²⁺] lower than 500 nM [33].

Several groups were able to resolve individual “mito-[Ca²⁺] transients” in parallel with cytosolic Ca²⁺ transients using mitochondria-loaded fluorescent indicators or matrix-targeted Ca²⁺ biosensors [16]. Single cytosolic Ca²⁺ transient can result in ~10–30 nM Ca²⁺ increase in the mitochondrial matrix [34]. Importantly, the decay of mitochondrial Ca²⁺ transients is considerably slower than those of cytosolic Ca²⁺ transients [34,35], which allows mitochondria to gradually accumulate Ca²⁺ to substantially higher levels during continuous periodic pacing (Figure 1A and 1B). Under baseline conditions at rest, free mitochondrial matrix [Ca²⁺] is estimated within the 100–200 nM range nM in cytosol [36,37]. During periodic pacing of CMs, free mito-[Ca²⁺] increases in accordance to cytosolic Ca²⁺ transients amplitudes and frequency of stimulation (Figure 1C), and can reach up to 4–5 μM when CMs are exposed to β-adrenergic stimulation [9,31,34].

Figure 1. Free mitochondrial matrix [Ca²⁺] dynamics in intact ventricular myocytes.

Free mitochondrial matrix [Ca²⁺] was monitored using a Leica SP8 Lightning laser scanning confocal system at 8 ms per frame at room temperature in a cultured rat ventricular myocyte adenovirally expressing red mtRCamp1h biosensor (Kd = 1.3 μM) [85,114]. Cytosolic Ca²⁺ transients were recorded using Fluo-3 AM as described in [9], and presented as F/F₀. Mito-[Ca²⁺] was estimated using following equation: Kd*(F-Fmin)/(Fmax-F). Fmin was fluorescence in the presence of 5 mM EGTA, and Fmax was fluorescence at 200 μM [Ca²⁺] in cell permeabilized with saponin. A. Superimposed representative traces of cytosolic Ca²⁺ transients and mito-[Ca²⁺] in a field-stimulated rat ventricular myocyte in the presence of 50 nM isoproterenol. Increase in stimulation frequency from 1 Hz to 2 Hz leads to an increase in mito-[Ca²⁺] due to slow removal kinetics. B. Enlarged superimposed traces from A demonstrate that mito-Ca²⁺ time to-peak is 5–10 fold slower than cytosolic Ca²⁺ transient time-to peak. C. The dependence of mito-[Ca²⁺] level upon stimulation frequency.

4.1. The role of SR Ca²⁺ release in regulation of mitochondrial function

The Ca²⁺ K_m of MCU in physiological conditions is estimated to be 20–30 μM and [Ca²⁺] must be ~2–5 μM in the vicinity of transporter for appreciable uptake to occur [38]. Based on the data from non-muscle cells, the concept of restricted “high [Ca²⁺] microdomains” has emerged to explain how MCU still can bring Ca²⁺ into the matrix [39]. Using a Ca²⁺ biosensor specifically targeted to the OMM in neonatal rat CMs, Drago et al. [40] showed that there are hotspots on OMM where Ca²⁺rises up to 30 μM during SR Ca²⁺ release, which is sufficient for MCU activation. Although computer modeling predicts [Ca²⁺] near mitochondria ends, i.e. nanodomains where MCUs are concentrated [41], is within a similar 10–20 μM range during RyR2-mediated SR Ca²⁺ release [42], the concept of a “high [Ca²⁺] microdomain” is yet to be experimentally tested in live adult CMs. Computer modeling also predicts that during a Ca²⁺ transient, the exposure time of to 10 μM [Ca²⁺] is very short (~10 ms), which is hard to reconcile with slow 100–300 ms rise times reported by mitochondrial Ca²⁺ probes [34] (See Figure 1B). As mentioned in the previous section, the dependence of mito-Ca²⁺ uptake upon cytosolic [Ca²⁺] in CMs is steep, with uptake starting at ~300–500 nM and becoming prominent at 2–5 μM. The latter values are achievable during the peak of Ca²⁺ transients when CMs are challenged with β-adrenergic agonists such as isoproterenol [9]. Interestingly, mouse models with cardiac specific MCU knockout or overexpression of dominant-negative MCU paralogs (MCU-DN or MCUb) showed very little impact on cardiac function and overall mito-[Ca²⁺] under basal conditions [37,43,44]. However, hearts from these mice exhibited blunted response to catecholaminergic stimulation. Measurements of mito-[Ca²⁺] and respiration rates in MCU knockout CMs showed that challenge with isoproterenol failed to increase both. However, mito-[Ca²⁺] and respiration rates were increasing, although slowly, and within 30–60 minutes both parameters reach the values similar to values in normal myocytes. These data suggest that MCU-mediated mito-Ca²⁺ influx during RyR2-mediated SR Ca²⁺ release plays an important and very specific role in CMs, accelerating metabolic output to match increased demand during the fight-or-flight response. Under baseline conditions, the physiological function of this pathway in CMs is rather limited.

The main role of Ca²⁺ in mitochondrial matrix has been ascribed to activation of Ca²⁺-dependent dehydrogenases in Krebs cycle, resulting in increased production of NADH, facilitation of the electron transport and activation of F₁/F_o-ATP synthase to produce ATP [45]. Notably, later work of Wescott et al. [31], measuring both matrix Ca²⁺ and ATP came to a conclusion that stimulation of ATP synthesis by Ca²⁺ does not involve direct effect on ATP synthase. Furthermore, they showed that ATP production occurs through micromolar matrix Ca²⁺-dependent modulation of pyruvate and glutamate dehydrogenase activity. This is line with classical biochemistry works revealing that these dehydrogenases activities are tightly linked to changes in [Ca²⁺] within 0.1–10 μM range [46,47]. Furthermore, changes in [Ca²⁺] can modulate electron transport efficiency via regulation of mitochondrial volume, by activating mitochondrial Ca²⁺-dependent K⁺ channels [48]. Large conductance Ca²⁺-activated K⁺ channels (BK channels) sense matrix Ca²⁺ and their protective role against ischemia-reperfusion injury associated with mitochondrial Ca²⁺ overload is well established. However, given their relatively low [Ca²⁺] Kd of ~10 μM and strong voltage dependence, the role of BK channels under normal physiological conditions can be limited. On the other hand, small conductance Ca²⁺-activated K⁺ channels present in the IMM are voltage independent and gated exclusively by Ca²⁺ via constitutively bound calmodulin, with affinity for Ca²⁺ within the physiological range ~0.3–1 μM. Experiments to determine SK channel orientation in the IMM in neurons showed that the SK Ca²⁺ sensor is facing the intermembrane space [49], suggestive that the activity of mitochondrial SK channels is linked to cytosolic Ca²⁺ fluctuations.

5.1. Mitochondrial subpopulations and sources of Ca²⁺

Based on specific location, CM mitochondria are often divided into three distinct subpopulations that may perform different functions; namely: 1) interfibrillar, 2) subsarcolemmal and, 3) perinuclear mitochondria (see Figure 2). The primary role of interfibrillar mitochondria role is to support contractile function of CMs. Subsarcolemmal mitochondria feed sarcolemmal energy-consuming transporters, while perinuclear mitochondria contribute to the synthetic activity, i.e. transcription and translation, of CMs. Although differences in metabolic profiles between mitochondrial subpopulations were previously established [50], information with regard to potential differences in Ca²⁺ handling remains limited [51,52].

Figure 2. Intracellular Ca²⁺ sources and mitochondrial subpopulations in adult ventricular myocytes.

Upon RyR2-mediated SR Ca²⁺ release, free [Ca²⁺] in the dyadic space increases from 200 nM to ≥200 μM, which leads to an increase in submembrane [Ca²⁺] to 20–40 μM. SR free-[Ca²⁺] reciprocally decreases from ~1 mM to 200–400 μM. The cytosolic [Ca²⁺] during SR Ca²⁺ release increases from basal 100 nM to 300 nM, and can reach 3–4 μM when myocytes are exposed to β-adrenergic stimulation. Under periodic stimulation, mito-[Ca²⁺] increases from 100–200 nM to 1 μM and up to 5 μM in the presence of β-agonists. In interfibrillar mitochondria, MCU complexes are situated at the end that faces the back of jSR cisternae. The theoretical estimate for jSR-mitochondria nanodomain [Ca²⁺] is 10–20 μM. Supporting experimental data is missing. NCLX is situated further from the jSR cisternae, which may underlie mito-[Ca²⁺] gradient. Subsarcolemmal mitochondria may be exposed to higher Ca²⁺ during Ca²⁺ transient due to higher submembrane [Ca²⁺] vs cytosolic [Ca²⁺], and more direct communication of RyR2 in corbular SR and mitochondrial VDAC. The exact free mito-[Ca²⁺] for this subpopulation remains unknown. During periodic pacing, perinuclear mitochondria accumulates [Ca²⁺] slower probably due to longer distances to RyR2 clusters. Stimulation of IP₃ receptors increases perinuclear mito-[Ca²⁺]. Whether interfibrillar and perinuclear mito-[Ca²⁺]s differ under certain conditions remains unknown. Figure created with BioRender.com.

Given the close proximity, interfibrillar mitochondria are thought to specialize in providing energy to the contractile apparatus in CMs. They are tethered at one end to the jSR by several structural proteins including mitofusins 1 and 2, PDZD8 and GRP75 [53,54]. According to De La Fuente et al. [41], functional MCU complexes are concentrated at the mitochondrial-jSR interface, which can bring them within 40 nm distance of RyR2 clusters. However, given structural restrictions of Ca²⁺ diffusion from release sites located on jSR sheets that wrap around T-tubules, it seems unlikely that [Ca²⁺] in jSR-interfibrillar mitochondria interface can rise orders higher than [Ca²⁺] in the cytosol. Nevertheless, the Bers group confirmed the existence of interfibrillar mitochondrial matrix Ca²⁺ gradient in rabbit CMs, with higher values near the ends of mitochondria tethered to jSR [34]. De La Fuente et al. [41] showed that NCLX is distributed outside of uptake sites further away from the Ca²⁺ source, which may contribute to the Ca²⁺ gradient in the mitochondrial matrix.

The Bers group recently showed that Ca²⁺ accumulates slower in perinuclear mitochondria during periodic pacing of intact CMs than in interfibrillar mitochondria, and this was attributed to lower local [Ca²⁺] exposure [55]. On the other hand, subsarcolemmal mitochondria may be exposed to a substantially higher local [Ca²⁺] given that submembrane [Ca²⁺] is estimated at ~20–40 μM during a Ca²⁺ transient vs. maximum cytosolic [Ca²⁺] of ~2–5 μM under β-adrenergic stimulation [7,8,9,56]. Furthemore, RyR2 clusters in corbular SR under the CM surface membrane are not as tightly restricted to face the intermembrane space as RyR2 clusters in jSR within dyads that face T-tubular membrane [57]. Moreover, parallel co-immunoprecipitation and localization studies revealed that RyR2 interacts with VDAC2 from subsarcolemmal mitochondria [58], which suggests more direct SR-mitochondrial Ca²⁺ transfer for this subpopulation. Another study using superresolution imaging technique established the close proximity of subsarcolemmal mitochondria with sarcolemmal Na⁺ channels [59]. Increased local Na⁺ is expected to facilitate NCLX-mediated mito-Ca²⁺ extrusion, promoting matrix unloading [35]. Together, these data strongly suggest that mito-Ca²⁺ dynamics can differ greatly between mitochondrial subpopulations. However, functional consequences of these differences and their relations with regard to different metabolic profiles were not unraveled yet.

The Dedkova group recently showed that activation of IP₃Rs in CMs increases intramitochondrial [Ca²⁺] in perinuclear and interfibrillar mitochondrial populations, but not in subsarcolemmal mitochondria [60]. This effect was preventable by knocking down mitofusin 2 suggesting mitochondria tethering to free SR as well far from jSR RyR2 Ca²⁺ release clusters. It also point out that certain mitochondria subpopulations are integral part of hypertrophic signaling pathways. A recent report from the Bers group demonstrated that perinuclear mitochondrial fission and fusion rates are several orders higher than those for interfibrillar mitochondria [55]. Fission and fusion along with mitophagy are important components of mitochondrial quality control. Murphy et al. [61] recently showed that facilitation of mitophagy rescues aberrant RyR2-mediated SR Ca²⁺ release in ventricular myocytes from aging hearts. The authors suggest such intervention can reduce arrhythmic risk associated with aging.

Taken together, these data establish that CM mitochondria are fully equipped to translate changes in intracellular Ca²⁺ homeostasis into metabolic response, and three mitochondrial subpopulations exhibit different alignments with regard to specific Ca²⁺ sources according to their particular functions.

6.1. The role of mitochondria in regulation of SR Ca²⁺ release

6.1.1. ATP and Mg²⁺

As the major source of ATP, mitochondria regulate activity of SERCa2a. Diminished ATP levels in diseased hearts, such as in late stages of heart failure [62], contribute to the reduced capacity of SERCa2a to re-sequester Ca²⁺, leading to smaller systolic Ca²⁺ transients and increased diastolic Ca²⁺ that promote pro-arrhythmic spontaneous RyR2-mediated Ca²⁺ release. ATP is the major buffer for intracellular Mg²⁺. Given that RyR2 activity is positively regulated by ATP and negatively by Mg²⁺ [63], increased free Mg²⁺ in heart failure reduces RyR2 hyperactivity, at least partially offsetting SR Ca²⁺ loss during diastole [64]. However, this may negatively affect systolic Ca²⁺ release and exacerbate contractile dysfunction.

6.1.2. Mitochondria as buffers of intracellular Ca²⁺

Given the significant capacity of mitochondria to accumulate Ca²⁺ and shape cytosolic Ca²⁺ signals in many cell types, the potential of CM mitochondria to serve as an effective Ca²⁺ buffer received significant attention [16]. Pacher et al. [65] reported that mitochondria are able to intercept up to 26% of Ca²⁺ released from the SR in permeabilized H9C2 myotubes. Assessments using neonatal rat CMs demonstrated that pharmacological of genetic manipulation of MCU activity significantly affects SR Ca²⁺ release amplitude [40]. This was further corroborated in guinea pig adult ventricular myocytes [35]. Maack et al. [35] estimated that treatment with 10 nM Ru360, a specific MCU inhibitor, can increase cytosolic Ca²⁺ elevation during systole up to 50 nmol/L. However, higher concentrations of Ru360 decreased Ca²⁺ transient amplitude. Likewise, unchanged cytosolic Ca²⁺ transients were reported in adult rabbit ventricular myocytes treated with 1 μM Ru360 [34]. In this work, Lu et al. estimated that the mitochondrial contribution to removal of cytosolic Ca²⁺is only ~1%, in support of previous work suggesting that the role of mitochondria in this process is dispensable when compared to those of SERCa2a and NCX1 [66]. Experimental and computational works of the Lederer group provided further support of little if any role of mitochondria as cytosolic Ca²⁺ buffers in adult cardiac ventricular myocytes [36,42]. Furthermore, genetic inhibition of MCU [66] or ablation of mitofusin 2 [68] to reduce local communication between RyR2 and mitochondria did not alter parameters of Ca²⁺ transients in mouse ventricular myocytes.

6.1.3. Mitochondria as sources of reactive oxygen species

While the debate regarding mitochondrial Ca²⁺ buffering continues, the role of mitochondria as major sources of reactive oxygen species (ROS) is well established [69,70,71]. At least 11 distinct mitochondrial sites involved in oxidative phosphorylation or substrate oxidation were shown to produce superoxide anion or H₂O₂ [72]. Both reduced and enhanced electron transport chain activity increases ROS production, which can overwhelm antioxidant defenses and affect numerous intracellular targets and processes. Increased mito-ROS has been associated with abnormally high RyR2 activity in many cardiac diseases including HF, myocardial infarct (MI), cardiac hypertrophy, diabetes- and aging-related cardiac dysfunction, and more recently hereditary catecholaminergic polymorphic ventricular tachycardia (CPVT) [9,73,74,75,76,77,78,79]. Mito-ROS can modulate RyR2 function directly [80,81,82], given that the tetrameric channel has 84 cysteines prone to oxidative modifications [80]. It can also affect RyR2 activity indirectly. Oxidative stress leads to disruption of protein-protein interactions within the SR Ca²⁺ release channel, such as reducing the amount of RyR2-tethered calmodulin, thereby reducing its stabilizing influence [83,84]. Another pathway involves oxidation of CaMKII, which increases enzyme activity and results in phosphorylation and hyperactivation of RyR2 [85]. H₂0₂-mediated modification of Cysteine-674 of SERCa2a decreases ATPase activity, contributing to impaired SR Ca²⁺ reuptake and thereby relaxation [12]. Together, mito-ROS-mediated RyR2 gain-of-function and SERCa2a loss-of-function induces a reduction of SR Ca²⁺ content and enhancement of spontaneous SR Ca²⁺ release, underlying pro-arrhythmic EADs and DADs especially prominent during challenge with catecholamines to mimic stress. Scavenging of mito-ROS using mitochondrial-specific agent mitoTEMPO, or mitochondrial overexpression of enzymes superoxide dismutase or catalase leads to significant improvement of intracellular Ca²⁺ homeostasis in various cardiac disease states [9,73,74,75,78,86,87,88].

7.1. Altered mito-Ca²⁺ homeostasis in cardiac disease

Studies of mechanisms underlying ischemia-reperfusion injury, which causes simultaneous cytosolic and mitochondria Ca²⁺ overload, established that abnormalities in function of SR Ca²⁺ handling machinery and mitochondria amplify each other, exacerbating damage-evoking cell death pathways and promoting pro-arrhythmic behavior [89,90]. Interventions to specifically protect mitochondria or restore SR Ca²⁺ handling were shown to improve both [61,78]. This verified the importance of bidirectional control of SR and mitochondrial function, and set the stage for in-depth studies in chronic, less extreme conditions such as HF, age-related cardiac dysfunction and diabetic cardiomyopathy. Diminished SR Ca²⁺ transient amplitude due to increased RyR2-mediated diastolic SR Ca²⁺ leak and reduced SERCa2a-mediated uptake are hallmarks of HF and diabetic cardiomyopathy [91]. It is well established that in both conditions, mitochondrial function is compromised [62,92] and mito-ROS production is accelerated [93]. In diabetes, it is accompanied by a reduction in mitochondrial Ca²⁺ content [94]. Suarez et al. [95] ascribed this change to a reduction in MCU expression levels. However, other diabetes models showed no changes [94]. Studies using a mouse post-MI HF model suggested that increased RyR2-mediated SR Ca²⁺ leak is accompanied by an increase in mito-[Ca²⁺], assessed using isolated mitochondria and Rhod-2AM loaded isolated ventricular myocytes [78]. A similar increase in Rhod-2 fluorescence was obtained in ventricular myocytes from a hypertension-induced, non-ischemic mouse HF model [96]. In this work, Xie et al. demonstrated increased Ca²⁺_MCU current in patch-clamped mitoplasts isolated from failing mouse hearts. Isolated mitochondria from aging human hearts also showed an increase in mito-[Ca²⁺] [97]. This may be explained by posttranslational modifications of MCU causing its gain-of function in conditions accompanied by oxidative stress and increased catecholaminergic tone such as HF. Dong et al. [98] recently established that MCU is a redox sensor and its oxidation increases activity. O-Uchi et al. [99] demonstrated that α-adrenergic stimulation evokes translocation of the ROS/Ca²⁺ dependent proline-rich tyrosine kinase Pyk2 from the cytosol to mitochondria, which promotes tetrameric MCU pore formation and accelerates mito-Ca²⁺ uptake. However, patch clamp studies in mitoplasts isolated from control and failing human hearts showed reduced activity of mitochondrial Ca²⁺ uptake channels in HF [100]. Furthermore, thorough assessment by the O’Rourke group showed a significant decrease of free mito-[Ca²⁺] in intact, periodically paced ventricular myocytes from a guinea pig HF model [101]. We obtained similar results in field stimulated ventricular myocytes from rat hearts with pressure-overload induced hypertrophy, when challenged with β-adrenergic agonist isoproterenol [86]. Maack et al. [35] associated this decrease with cellular Na⁺ overload characteristic of HF, which facilitates mito-Ca²⁺ removal by mitochondrial NCLX. Furthermore, the sharp decrease in systolic Ca²⁺ transient amplitude in disease states reduces [Ca²⁺] within the 2–5 μM range, suitable for MCU-mediated uptake. It appears that increased diastolic [Ca²⁺] due to increased RyR2-mediated SR Ca²⁺ leak and reduced SERCa2a-mediated uptake cannot compensate systolic Ca²⁺ reduction, resulting in net decrease of free mito-[Ca²⁺] [9].

To summarize, decreased Ca²⁺ transient amplitude and increased cytosolic [Na⁺] play major roles in diminished mito-[Ca²⁺] in cardiac disease. The increase in mito-[Ca²⁺] seen using isolated mitochondria preparations can be potentially explained by increased MCU activity due to posttranslational modifications, in conditions where disease-related changes in cytosolic [Ca²⁺] and [Na⁺] are eliminated.

8.1. Mito-Ca²⁺ handling machinery as an anti-arrhythmic target

Calcium-dependent arrhythmias have been attributed to the generation of diastolic intracellular Ca²⁺ waves that promotes sarcolemma depolarization and activation of extrasystolic action potentials [102,103,104]. General approaches to affect cellular arrhythmogenesis include; 1) suppression of Ca²⁺ wave generation by reduction of cellular Ca²⁺ accumulation (Ca²⁺ influx, AP duration); 2) stabilization of RyR2 activity; and 3) uncoupling elevation of cytosolic Ca²⁺ from sarcolemma depolarization by inhibiting NCX1.

Three strategies targeting mitochondrial homeostasis have recently emerged to reduce cardiac Ca²⁺-dependent arrhythmogenesis and SCD. These strategies were directed to affect different mechanisms of cellular arrhythmogenesis: to suppress Ca²⁺ wave initiation and propagation by 1) increasing mito-Ca²⁺ uptake; 2) prevention of sarcolemma depolarization induced by augmented mito-Ca²⁺ efflux; or 3) reducing mitochondrial-mediated posttranslational modification of RyR2.

8.1.1. Mito-Ca²⁺ uptake enhancement

Given the proximity of RyR2 clusters to mitochondria, it has been suggested that enhancement of mito-Ca²⁺ uptake can be used to interfere with Ca²⁺ diffusion from one RyR2 cluster to another, thereby reducing the probability of arrhythmogenic spontaneous Ca²⁺ wave initiation and propagation. Usage of pharmacological VDAC2 enhancer efsevin, and either VDAC2 or MCU overexpression, eliminated cardiac fibrillation in zebra fish mutants with NCX1 loss of function [105]. Experiments in the same work using human and mouse embryonic CMs in Ca²⁺ overload conditions demonstrated that efsevin accelerated Ca²⁺ transient and Ca²⁺ spark decay without changing amplitudes. Furthermore, efsevin reduced the frequency of pro-arrhythmic spontaneous Ca²⁺ waves in adult mouse myocytes evoked by increased 5 mM [Ca²⁺] in extracellular solution. The anti-arrhythmic potential of mito-Ca²⁺ uptake enhancement was further tested in mouse model of CPVT with gain of function RyR2 mutation, RyR2-R4496C^(+/−) [106]. Efsevin significantly attenuated diastolic Ca²⁺ waves and spontaneous action potentials in this model. The effect was abolished by MCU blockade and could be reproduced using the MCU activator kaempferol. Both efsevin and kaempferol reduced the incidence of stress-induced ventricular tachycardia (VT) in CPVT mice in vivo, and abolished spontaneous pro-arrhythmic Ca²⁺ events in human RyR2-S406L^(+/−) iPSC-derived CMs. Interestingly, efsevin significantly reduced the frequency, amplitude and spatiotemporal parameters of Ca²⁺ sparks in CPVT mouse cells challenged with β-adrenergic agonist isoproterenol. The authors attribute this to efsevin-mediated increase in the efficiency of mitochondria to intercept Ca²⁺ locally released by RyR2 clusters [106]. The main shortcoming of this approach is that augmentation of mito-Ca²⁺ uptake leads to increased mito-Ca²⁺ accumulation, driving excessive production of ROS. Indeed, we recently showed that MCU enhancement with kaempferol significantly reduces SR Ca²⁺ content in ISO-challenged CMs from rat hearts with pressure-overload induced hypertrophy [86]. This reduction was ascribed to enhanced RyR2s oxidation due to increased mito-ROS emission in the presence of kaempferol, thereby stimulating channel activity. Increased mito-ROS has been consistently associated with the pro-arrhythmic effects (see above). Accordingly, kaempferol did not reduce arrhythmia incidence and even increased frequency of ventricular fibrillation (VF) evoked by β-adrenergic stimulation in ex vivo optically-mapped hypertrophic hearts [86]. Intriguingly, in the recent manuscript of Liu et al. [77], authors showed that moderate (30%) MCU overexpression using adenovirus-mediated delivery in failing Guinea pig hearts significantly reduced premature ventricular contractions (PVCs), improved Ca²⁺ homeostasis, reversed ROS elevation measured using H₂O₂ fluorescent indicator, and restored RyR2 redox state. However, parallel work of our group suggests that two-fold MCU overexpression in hypertrophic rat ventricular myocytes does not attenuate increased propensity to pro-arrhythmic spontaneous Ca²⁺ waves in the presence of β-adrenergic agonist [107]. Furthermore MCU overexpression results in massive increase in mito-ROS levels measured with mito-targeted H₂O₂ biosensors beyond already high levels in diseased myocytes [107], in line with previous work using pharmacological MCU enhancer [86].

8.1.2. Mito-Ca²⁺ efflux inhibition

Given cytosolic Na⁺ overload in HF, Maack et al. [35] attributed HF-related mito-Ca²⁺ loss to an increase in NCLX-mediated exchange of mitochondrial Ca²⁺ for Na⁺. Accordingly, restoration of NCLX activity using pharmacological inhibition with NCLX inhibitor CGP-37157 restored mito-[Ca²⁺] in ventricular myocytes from failing hearts, and restored matching of metabolic demand with output [108]. Furthermore, CGP-37157 treatment markedly reduced the incidence of PVCs and prevented VF, substantially reducing SCD in a guinea pig model of HF [101]. Xie et al. [109] showed that NCLX inhibition with CGP-37157 in embryonic stem cell derived mouse ventricular myocytes reduces automaticity. This was confirmed in hiPSC-derived ventricular myocytes and ischemic adult mouse ventricular myocytes. The authors suggested that Ca²⁺ elevation resulted from enhanced activity of NCLX can activate electrogenic plasmalemmal NCX1, causing sufficient depolarization to evoke spontaneous action potential thereby contributing to triggered activity. Given that in interfibrillar mitochondria, NCLX is situated further from jSR and T-tubules [41], it is unlikely that Ca²⁺ released from mitochondria can reach NCX1 and is not intercepted by SERCa2a. On the contrary, the submembrane mitochondrial population, which in theory can be exposed to higher local [Na⁺] [110] may indeed release Ca²⁺ close enough to activate NCX1. Additionally, if large enough, Ca²⁺ efflux from mitochondria could reactivate adjacent RyR2 clusters contributing to spontaneous Ca²⁺ release. However, there is significant concern that CGP-37157 may have nonspecific and even deleterious effects [111]. Notably, Maack et al. [35] showed that in the presence of isoproterenol, this compound reduces Ca²⁺ transient amplitude despite increasing mito-Ca²⁺ load. Furthermore, our data [86] demonstrated that enhanced mito-Ca²⁺ accumulation facilitated by CGP-37157 promotes proarrhythmic spontaneous Ca²⁺ waves in rat hypertrophic myocytes challenged with isoproterenol, likely by increasing mito-ROS emission. Finally, genetic NCLX deletion in mice caused mito-Ca²⁺ overload associated with severe deterioration of myocardial function and sudden onset of HF [43]. On the contrary, NCLX overexpression in mouse hearts had beneficial effects, protecting against ischemia-induced heart failure by augmenting mito-Ca²⁺ clearance in ventricular myocytes.

8.1.3. Mito-Ca²⁺ uptake inhibition

Pioneering work of Garcia-Rivas et al. [112] showed that perfusion with specific MCU inhibitor Ru360 abolishes the incidence of cardiac arrhythmia evoked by ischemia/reperfusion injury in an open chest rat model. The authors suggested that MCU inhibition prevented pathological mito-Ca²⁺ overload and the subsequent activation of mPTP that would result in dissipation of mitochondrial membrane potential. As mentioned above, genetic suppression of MCU expression/function did not substantially affect cardiac function under basal conditions. Given that the MCU complex is not the sole Ca²⁺ influx pathway, these manipulations did not affect basal mito-[Ca²⁺] either. The most important function of MCU-mediated Ca²⁺ uptake is to ensure matching of metabolic output with increased demand during stress. Consequently, diminished MCU complex function reduces cardiac contractility during catecholaminergic surge. However, the benefits of MCU suppression include reduced mito-ROS production, which can adversely affect RyR2s and SERCa2a on the SR [9]. Hamilton et al. [86] showed that application of selective MCU inhibitor Ru360 significantly reduced RyR2 oxidation and pro-arrhythmic spontaneous Ca²⁺ release in rat ventricular myocytes from hypertrophic hearts exposed to isoproterenol. At the whole heart level, Ru360 reduced PVC incidence and VT/VF in hypertrophic hearts induced by exposure to isoproterenol. Consistent with these findings, Xie et al. [96] showed that MCU inhibition with Ru360 or MCU knock down significantly reduced isoproterenol-evoked Ca²⁺-dependent arrhythmia episodes in a non-ischemic mouse HF model induced by hypertension. Notably, β-blockers are a frontline therapy for HF, and their beneficial effects could be at least in part attributed to a reduction in mito-Ca²⁺ uptake and mito-ROS. However, whether MCU inhibition could provide stronger benefits than conventional β-blocker therapy is yet to be determined. Additional evidence that restriction of mito-Ca²⁺ uptake may have antiarrhythmic potential was obtained using SK channel enhancers. In rats with pressure-overload induced hypertrophy, application of selective SK1–3 channel enhancer NS309 attenuated isoproterenol-induced Ca²⁺-dependent arrhythmias in ex vivo optically mapped hearts [75]. At the cellular level, SK channel enhancers or viral-mediated overexpression of SK2 channels in rat ventricular myocytes reduced mito-ROS and mito-Ca²⁺ uptake, thereby reducing RyR2-mediated pro-arrhythmic spontaneous SR Ca²⁺ release [75,113].

Taken together, these data suggest that under certain conditions, facilitation of mito-Ca²⁺ accumulation and retention by enhancing MCU-mediated uptake or inhibiting NCLX-mediated outflow may be beneficial. However this is accompanied by a risk of accelerating damaging mito-ROS production increasing diastolic SR Ca²⁺ release. Conversely, inhibition of mito-Ca²⁺ accumulation reduces mito-ROS and lessens pro-arrhythmic spontaneous SR Ca²⁺ release at the expense of reduced mitochondrial metabolic output under conditions mimicking stress.

9.1. Perspective

In the two last decades, significant progress has been made in delineating mechanisms of bidirectional control of SR Ca²⁺ release and mitochondrial function in CMs. An improved understanding of communication intricacies between mitochondria and SR, as well as identification of the molecular identity of mitochondria Ca²⁺ handling machinery elements, holds great promise in determining key nodes to interrupt the vicious circle of SR-mitochondrial dysfunction. This is expected to foster development of a new class of effective antiarrhythmic therapies, employing more selective pharmacological tools [114], small molecules or genetic gain- or loss-of-function approaches. Rapid progress in the generation of novel Ca²⁺ and ROS biosensors that can be effectively targeted to specific intracellular nanodomains is anticipated to aid in resolving long lasting controversies such as whether mitochondria matrix free [Ca²⁺] is reduced or increased in certain cardiac diseases. More thorough assessment of differences in structural constrains and mitochondrial Ca²⁺ homeostasis between mitochondrial subpopulations may help to reconcile seemingly contradictory data where facilitation of mito-Ca²⁺ accumulation and mito-Ca²⁺ uptake inhibition can be protective. Given apparent structural differences in SR-mitochondrial microdomains in ventricular myocytes vs. atrial myocytes, or elements of conduction systems, caution should be exercised in transferring knowledge obtained in one cell type to another. Likewise, experimentalists using developing or induced pluripotent stem cells derived myocytes must take into account degree of maturation as a limitation. Finally, the majority of information available now was generated using small animal models with fast heart rate and metabolism, and will require further verification using large animal models to be relevant for clinics.