Reactive oxygen species (ROS) are emerging as centre-stage players in cardiac functional regulation. ROS and Ca2+ signals converge at dyads, the structural and functional units of cardiac excitation–contraction coupling. These two prominent signalling systems are intertwined with ROS modulation of the entire Ca2+-signalling network, and vice versa. While constitutively generated homoeostatic ROS are important in setting the redox potential of the intracellular milieu, dynamic signalling ROS shape microdomain and global Ca2+ signals on both the beat-to-beat and greater time scales. However, ROS effects are complex and subtle, characterized by multiphasic and bidirectional Ca2+ responses; and sustained oxidative stress may lead to compromised contractility and arrhythmogenicity. These new understandings should be leveraged to harness ROS for their beneficial roles while avoiding deleterious effects in the heart.
Reactive oxygen species
Reactive oxygen species (ROS) are a group of chemical species that comprise at least one oxygen atom in each molecule but display stronger reactivity than molecular oxygen. ROS are classified into two categories: free radicals with an unpaired electron (e.g. O2•− and OH•) and non-radical derivatives (e.g. H2O2). Closely related, the principal reactive nitrogen species (RNS) is •NO, a free radical that plays a central role in cardiovascular signalling. Basal, homoeostatic ROS are required to set the redox balance of the cell; when ROS are produced excessively or antioxidant capacity is depleted, indiscriminate oxidation of proteins, lipids, and polynucleotides elicits harmful effects, causing ‘oxidative stress’. The most exciting ongoing development, however, is that ROS and RNS emerge as welcome additions to the repertoire of only a handful of known second messengers (Ca2+, cAMP, IP3, arachidonic acid). That is, dynamic, regulatory and often compartmentalized increases of ROS and RNS are harnessed as powerful, ubiquitous, and indispensable cellular messengers—redox signalling—which is defined as the specific, usually reversible oxidative/reductive modification of cellular signalling pathway components.1 The ROS/RNS-specific targets range from ion channels and transporters to kinases and phosphatases and to transcription factors, and the list continues to grow and permeate throughout pivotal pathways in differentiation and organogenesis,2 cell fate regulation,3 stress responses,4 and wound-healing.5,6
Chemically a single-atom divalent cation, the calcium ion (Ca2+) acts as the most important biological messenger known. Rapid, transient changes in intracellular Ca2+concentration ([Ca2+]i) directly control muscle contraction, hormone secretion, and neurotransmitter release. Sustained [Ca2+]i elevation regulates gene expression, cell growth, and cell death. Paradoxically, and in stark contrast to ROS/RNS signalling, Ca2+ exists in only one biologically relevant form that undergoes neither catabolic degradation nor anabolic synthesis. Its biological information-coding ability derives almost entirely from its binding to and unbinding from target proteins as well as from its charge movement to depolarize membrane potential in the form of Ca2+ currents.
Cardiomyocytes are highly specialized in the use of Ca2+ ions as the most powerful intracellular messenger alongside ROS/RNS. During a cardiac cycle, [Ca2+]i rises and falls spanning a 10-fold dynamic range (100-fold in Ca2+ microdomains). The cyclic rise and fall of [Ca2+]i engages and disengages the molecular machinery of contractile myofilaments to generate force and cell-shortening, incessantly, up to 3.5-billion times over the lifespan of a centenarian. This marvelous robustness of Ca2+ signalling comes with extraordinary stability and efficiency, as well as flexibility for rich modulation, all achieved simultaneously. Recent advances show that ROS and RNS participate in the regulation of cardiac Ca2+ signalling on both the beat-to-beat and greater time scales, with consequences for cardiac contractility, Ca2+-dependent arrhythmogenesis, and other non-contractile cardiac functions and pathologies. Crosstalk between the two prominent signalling systems in the heart has recently been reviewed.7 This review focuses on new findings and insights on how ROS interact with Ca2+ players to modulate microdomain Ca2+ signalling at the dyads, the structural and functional units of cardiac excitation–contraction (EC) coupling and Ca2+ signalling, with RNS signalling briefly discussed in pertinent context. The spotlight is also shed on the mitochondria, given their close proximity to the dyads and their essential role in both ROS and Ca2+ signalling. The overall reciprocal interaction between the ROS and Ca2+ signalling systems has been covered in recent reviews.7,8
2. Dyads: the nexus of Ca2+ and ROS signals
In the mammalian heart, meeting the stringent functional demands apparently hinges on a rather unique microscopic structure, the dyad. On one hand, the sarcolemma forms invaginations or transverse tubules (TTs) 180–280 nm in diameter along the Z-lines of sarcomeres, at longitudinal intervals of ∼1.8 µm and circumferential intervals of 0.5–1.5 µm.9,10 The TTs traverse the entire cell interior while interconnecting with each other. Thus, the surface sarcolemma and TTs constitute one physically continuous membrane system, providing an electrical highway for the rapid spread of the action potential throughout the bulk of the cardiomyocyte, at a velocity too fast to be detected with the 1.5-ms resolution of confocal linescan microscopy,11 for the electrical synchrony of the heart at the cellular level. Meanwhile, it partitions a 104-fold [Ca2+] gradient between the extracellular space (∼mM) and the cytosol (∼100 nM at diastole). On the other hand, the sarcoplasmic reticulum (SR) encloses a continuous subspace inside the cell, which occupies only 1–3% of the cell volume but permeates throughout the cytosol in the form of delicate longitudinal tubules [free SR, (fSR)] and pancake-shaped cisterns at Z-lines [junctional SR, (jSR)], partitioning another 104-fold [Ca2+] gradient between the luminal Ca2+ store (∼0.5 mM) and the cytosolic compartments. Importantly, the electro-chemical signal transduction between the sarcolemmal and SR systems does not occur everywhere; rather, it is restricted to the dyads, where jSR cisterns juxtapose TTs at a distance of 12–20 nm. About 10 000 dyads form a three-dimensional grid inside a cardiac ventricular myocyte. Individual dyads are activated in an all-or-none fashion, giving rise to discrete, quantal ‘Ca2+ sparks’.12 Spatiotemporal summation of them results in global [Ca2+]i transients. From the viewpoint of the control theory, such a binary ‘digital’ design is quintessential in conferring the ability for cardiac EC coupling to realize simultaneous high-gain amplification and stability.13,14 The tale of cardiac EC coupling is thus essentially a tale of microdomain Ca2+ signalling and regulation at the dyads (Figure 1).
Cardiac dyads as the nexus of Ca2+ and ROS/RNS signalling. The main structural feature is the trio of transverse tubule (TT), junctional sarcoplasmic reticulum (jSR) connected to free SR (fSR), and mitochondrion that makes contacts with the jSR through nanoscopic cleft and tethering. Molecular players in ROS/RNS generation are located on the TT, in the mitochondrial outer and inner membranes, in the jSR membrane, and in the cytosol. NOX2, NADPH oxidase isoform 2; eNOS, nNOS, iNOS, endothelial, neuronal, and inducible nitric oxide synthases; MAO, monoamine oxidase; XO, xanthine oxidase; CYP, cytochrome p450s; COX, cyclooxygenase; LOX, lipoxygenase. Players for microdomain Ca2+ signalling, all being ROS/RNS targets, include Ca2+ channels and transporters: LTCC, L-type Ca2+ channel; NCX, Na+/Ca2+ exchanger operating in either reverse or forward mode; RyR, cardiac ryanodine receptor; SERCA, SR/ER Ca2+-ATPase; CaM, calmodulin; CaMKII, Ca2+/CaM-dependent protein kinase II. Mitochondrial channels, including the inner membrane anion channel (IMAC), mitochondrial permeability transition pore (MPTP) and voltage-dependent anion channel (VDAC), may offer routes for the passage of ROS/RNS across different compartments. Phosphorylation-mediated post-translational modifications, mostly mediated by G-protein-coupled receptor (GPCR)-activated protein kinase A (PKA) and CaMKII, may interplay with the ROS/RNS-mediated redox modifications. For clarity, reciprocal Ca2+ -to-ROS signalling pathways, many phosphorylation-mediated pathways, and mitochondrial Ca2+ transporters are not shown.
A highly energy-demanding cell type, the cardiomyocyte contains ∼6000 mitochondria that occupy 30–40% of the cell volume. Importantly, mitochondria are present in close proximity (40–180 nm) to TT-SR dyads,15 and form junctional contacts or even tether with the jSR.16–18 While supplying local ATP on demand, they are also the major source of local ROS production and play significant roles in local Ca2+ handling (see below). In an extended view, trios of dyadic TT-SR and their associated mitochondria are recognized as the main structural, functional, and regulatory units of microdomain Ca2+ and ROS signalling in the heart (Figure 1).
3. Microdomain Ca2+ signalling at the dyads
The past two decades have witnessed a revolutionary advance in our understanding of microdomain Ca2+ signalling at the dyads, evidenced by the prescient prediction from ‘local control theory’14 and confocal visualization of the trilogy of elemental Ca2+ signalling events—Ca2+ sparks,12 Ca2+ sparklets,19 and Ca2+ blinks20. Specifically, during the excitation of a TT-SR dyad by an action potential, extracellular Ca2+ enters through the L-type Ca2+ channel (LTCC) and elevates [Ca2+] in the subspace to ∼10 µM while effusing into the ambient cytosol. This LTCC-mediated Ca2+ sparklet serves as a physiological trigger to activate an array of ∼100 ryanodine receptors (RyR) Ca2+-release channels, also known as a Ca2+-release unit,13,21 via the Ca2+-induced Ca2+ release (CICR) mechanism. The resultant store Ca2+ release creates an even greater microdomain Ca2+ in the form of a Ca2+ spark. Concurrently, the jSR Ca2+ is depleted, manifesting as a Ca2+ blink in the lumen of the cistern. During restoration, cytosolic Ca2+ is sequestered into the intracellular Ca2+ store through the sarcoplasmic/endoplasmic reticulum Ca2+ ATPase (SERCA) or is extruded to the extracellular space by the sarcolemmal Na+/Ca2+ exchanger (NCX) and the plasma membrane Ca2+-ATPase (PMCA) (Figure 1). As their regulatory elements, the dyads also contain players in the well-known protein phosphorylation and dephosphorylation mechanism. For example, the subspace of a dyad contains cAMP-dependent protein kinase (PKA), which transmits signals from the sarcolemmal G protein-coupled receptors, and Ca2+/calmodulin-dependent protein kinase II (CaMKII), which is dually activated by Ca2+ and ROS,22 as well as protein phosphatases. These kinases and phosphatases form macromolecular complexes with the RyR23. For recent reviews, please see references13,24,25.
4. ROS sources and dynamics: evidence for ROS microdomains
In cardiomyocytes, potential enzymatic sources of ROS include mitochondrial respiratory oxidases, NADPH oxidases in the sarcolemma and TT, and the cytoplasmic arachidonic acid pathway enzymes lipoxygenase26 and cyclooxygenase,27 cytochrome p450s,28 and xanthine oxidase.29,30 Often, O2•− is the primal ROS species produced and is subsequently converted into H2O2 (through spontaneous or superoxide dismutase (SOD)-catalysed dismutation) and OH• (derived from O2•− and H2O2 through the Haber–Weiss reaction). For RNS, enzymatic production of •NO from l-arginine, NADPH and O2 is catalysed by one of the three •NO synthase (NOS) isoforms—nNOS, iNOS, and eNOS, all expressed in cardiomyocytes.31 Peroxynitrite (ONOO−), a ROS and RNS species, is formed when O2•− and •NO react. In terms of locality of action, OH• has an extremely short lifetime and thus acts only at zero-distance; O2•− is charged (pKa for H+ ∼ 4) and membrane-impermeable, so its action is limited to the same membrane-delineated compartment where it is produced. Often as an O2•− derivative, H2O2 is more stable and membrane-permeable. In this way, newly formed ROS and RNS can affect effectors both locally and globally (Figure 1).
Despite continuous ROS generation, intracellular redox balance is safeguarded by a powerful endogenous antioxidant system comprising enzymatic ROS scavengers and non-enzymatic antioxidants. The ROS-scavenging enzymatic SODs catalyse the dismutation of O2•− into H2O2 and catalases convert H2O2 to water. The non-enzymatic antioxidants include glutathione, peroxiredoxin, thioredoxin, and NADPH. They collectively form an antioxidant pool, with associated enzymes such as glutathione peroxidase and thioredoxin reductase that promote inter-conversion towards an equilibrium between the reduced/oxidized species and among different antioxidants.31–35 Under experimental conditions, exogenous ROS scavengers may also be present, such as the SOD mimic MnTyP36 and the mitochondrial-targeted antioxidants, mitoTEMPO,37 and the synthetic tetrapeptide SS31.38 Local ROS gradients and thus the distance of action are collectively determined not only by the flux, the lifetime, and the compartmentalization of the ROS of interest, but also by the local ROS-scavenging capacity which is in part determined by the redox status of the antioxidant pool.
4.1 Constitutive mitochondrial ROS production
The mitochondrial respiratory chain efficiently transfer electrons along intermolecular and intramolecular pathways of increasing redox potential (Eh), from Eh = −320 mV at the entry point of Complex I (NADH dehydrogenase), all the way to Eh = +390 mV at Complex IV (cytochrome c oxidase), where four electrons are donated to O2 forming two molecules of H2O. However, with its Eh = −160 mV, O2 also snaps up 0.15–2% of the respiratory chain electrons, one at a time, at places prior to Complex IV, and undergoes one-electron reduction to form O2•−.39,40 There are two major respiratory chain complexes in the mitochondria which produce ROS: Complex I and Complex III (ubiquinol cytochrome c reductase).41 The O2•− formation occurs in the mitochondrial matrix (Complexes I and II) and on both sides of the inner mitochondrial membrane (Complex III).32 This mode of ROS production is obligatorily linked to mitochondrial respiration, and appears to be constitutive, continuous, and relatively constant. Mitochondrial O2•− is quickly converted to H2O2 by SOD2 abundantly present in the matrix and cytochrome c in the intermembrane space.42 In addition, monoamine oxidases (MAOs) on the outer mitochondrial membrane also generate H2O2 as they catalyse the breakdown of neurotransmitters such as catecholamines.
4.2 Mitochondrial superoxide flashes
With the serendipitous finding that a circularly permutated fluorescent protein, cpYFP, is a reversible O2•− biosensor, we recently discovered a novel mode of mitochondrial O2•− production: intermittent, quantal, 10-s bursting O2•− formation, namely the ‘superoxide flash’, in single mitochondria.43 A superoxide flash and its hypothetical companion H2O2 burst create a high-ROS microdomain affecting all ROS effectors at its neighbouring dyads. The visualization of superoxide flashes has thus provided direct evidence for the existence of ROS microdomains under physiological conditions, and their subsequent characterization has further revealed extensive resemblance between superoxide flashes and Ca2+ sparks (see below). Mechanistically, superoxide flashes are triggered by the opening of a physiologically relevant mitochondrial permeability transition pore (mPTP),43,44 and each coincides with an electrophysiological excitation of the mitochondrion, i.e. depolarization of mitochondrial membrane potential (Ψm). Careful analysis of flash ignition suggested that it is a regenerative process dominated by the stochastic recruitment of a limited number of participating units (e.g. mPTPs).45 Surprisingly, superoxide flashes are abolished by all electron transfer chain (ETC) blockers tested, including antimycin A that actually enhances constitutive mitochondrial ROS production.43 Albeit paradoxical, this result has been reproduced in different laboratories,46,47 and it is in general agreement with the notion that signalling ROS and homoeostatic ROS are regulated by distinctly different mechanisms. To date, the exact mechanisms underlying superoxide flash formation and termination remain open questions.
Increasing evidence supports the notion that superoxide flashes constitute elemental local ROS signalling events in physiological, stressful, and pathophysiological conditions.48 They are richly regulated in a frequency-modulated manner, by diverse signals converging onto the ETC and mPTP. As is the case for Ca2+ sparks, superoxide flash modulation occurs predominantly in a frequency-modulated manner, with only minor-to-moderate changes in the amplitude and duration. This ‘digital’ feature of ROS signalling bears a close resemblance to that of cardiac Ca2+ signalling depicted above, which is crucial to achieving signalling efficiency, specificity, and stability. Functionally, superoxide flash activity in skeletal muscle decodes the status of cellular and mitochondrial energy metabolism both in vivo and in vitro,44,46,47,49 and is elevated in the skeletal muscle of RyR1Y522S/WT malignant hyperthermic mice which exhibit marked temperature-dependent increases in ROS and RNS generation.46 It also responds to oxidative stress induced by hypoxia- and anoxia-reoxygenation in cardiomyocytes.43,50 In HeLa cells with elevated mitochondrial basal ROS production and Ca2+ uniporter activity, flash activity displays a reversible 20-fold increase that contributes to the activation of JNK and p38, essential signals for adaptive cell-survival responses.51 In addition, superoxide flashes are involved in numerous biological and pathological process, such as cortical neural progenitor proliferation and cerebral cortical development,52 Huntington's disease,53 and cancer cell apoptosis.54 It is noteworthy that, owing to their low frequency, brief duration and spatial confinement, superoxide flashes make only a negligible contribution to global ROS homoeostasis,52,54 unless at very high frequencies (Wei-LaPierre et al., personal communication).
4.3 ROS-induced ROS release
While superoxide flashes do not normally propagate, inter-mitochondrial ROS-induced ROS release (RIRR) occurs under extreme stress conditions. Zorov et al.55,56 have shown that intense photostimulation-elicited ROS triggers mPTP openings that cause mitochondrial ROS bursts. Unlike superoxide flashes of the physiologically relevant mPTP, photostimulated RIRR is likely triggered by peroxide and singlet oxygen (1O2) from photon-induced triplet state fluorochromes, and occurs within a narrow time-window prior to permanent cell injury, somewhat analogous to the mPTP opening at the commitment to cell death.57,58 The O'Rourke group found that local laser flash-induced mPTP-dependent ROS can trigger self-sustained oscillations in ΔΨm and ROS in cardiomyocytes.59,60 The whole-cell oscillations depend on the mitochondrial inner membrane anion channels rather than the mPTP.59,60 Using this protocol, our own study also revealed a spectrum of ROS signals, from local mitochondrial ROS bursts, to self-sustaining cell-wide ROS oscillations, and to ROS waves that propagate at a velocity of ∼4 μm/s.61 To extend the spark-flash analogy, global ROS oscillations and waves may consist of superoxide flashes as elemental events; however, the transition from discrete solitary events to RIRR may require the ROS system to be poised beyond a ‘criticality’ permissive for regenerative ROS production.62,63 Moreover, the criticality more likely reflects the ROS excitability of the mitochondrial network, rather than the magnitude of the trigger ROS, for we failed to observe propagating ROS waves during synchronous superoxide flashes involving extensive perinuclear mitochondrial networks of >100 µm3.64
It should be noted that the mitochondria can directly access the Ca2+ microdomains at their associated dyads. The ruthenium red-sensitive Ca2+ uniporter mediated by the MCU and MCU1 complexes65–68 accounts for the rapid beat-to-beat changes, while additional Ca2+ transport mechanisms involving families of the mitochondrial NCX and Ca2+/H+ antiporter69 perhaps participate in the slow, pacing-dependent accumulation of mitochondrial matrix Ca2+.70 Conversely, mitochondrial Ca2+ plays a profound role in regulating the aforementioned modes of mitochondrial ROS production, whether constitutive or regulatory or pathophysiological. For instance, the frequency of superoxide flashes is synergistically regulated by the MCU-MCU1-mediated Ca2+ uniporter and constitutive ROS formation in HeLa cells.51
4.4 ROS generation by NADPH oxidase: X-ROS
NADPH oxidases and the related dual oxidase (Duox) family has seven members, Nox1, Nox2, Nox3, Nox4, Nox5, Duox1, and Duox2, but only the Nox2, Nox1, and Nox4 isoforms are expressed in the heart, Nox2 being the predominant isoform in the adult cardiomyocyte.71 Nox2 is localized in the TTs of cardiac and skeletal muscles,72,73 making it an important player at the dyads. Similarly, eNOS is preferentially localized at the caveolae in TTs and the surface sarcolemma74,75 while nNOS is localized not only at the caveolae76,77 but also at the SR membrane (Figure 1).78 In an elegant approach involving MyoTak for cell adhesion and programmable devices for precise control, Prosser et al.72 have shown that a stepwise stretch of cardiomyocyte produces a ∼10-s long ROS transient that derives from Nox2, named ‘X-ROS’, and this response is sensitive to Nox inhibitors and absent in Nox2−/− cardiomyocytes. Remarkably, this X-ROS transient rises immediately upon stretch, making it suitable for beat-to-beat mechanochemical transduction.72,79 In an earlier related study, however, the Sollott group has demonstrated that stretch-activated, l-NAME-sensitive •NO production can be detected by 4,5-diaminofluorescein fluorescence at 10 min after a single-step static 10% stretch. These data show differential temporal dynamics of ROS and RNS signalling at the dyads. With the advent of the new cell-stretch technique,80 it would be of interest to determine how X-ROS respond to cyclic stretch in beating cardiomyocytes and even beating hearts and to further characterize the differential dynamics of X-ROS and RNS signalling at the dyads.
5. ROS regulation of microdomain Ca2+ signalling
Recent advances have shown that ROS play important roles in reshaping local and global Ca2+ signal amplitudes and kinetics. This redox-mediated functional regulation is often multiphasic and bidirectional, depending on the species of oxidants, the concentration, and the history of exposure. Mechanistically, ROS-dependent modifications permeate through virtually all Ca2+ signalling components at the dyads. In particular, ROS regulate LTCC, NCX, and PMCA in the sarcolemma and TT; and RyR and SERCA in the SR. Of these, we focus here on ROS regulation of dyadic RyR and Ca2+ sparks (Figure 1). See references7,8,31,81,82 for comprehensive reviews.
5.1 ROS regulation of Ca2+ sparks
5.1.1 Effects of mitochondrially derived ROS
Application of reducing agents or SOD mimics and inhibition of the mitochondrial respiratory chain with myxothiazol, a Complex III inhibitor, decrease the cytosolic ROS level and suppress Ca2+ spark frequency and amplitude,61 suggesting that basal cellular ROS production and redox balance are responsible for a significant portion of the spontaneous Ca2+ spark activity. Exogenously applied H2O2 modulates microdomain Ca2+ signalling in a dose-dependent manner: it has no effect at 1 μM but elicits a marked, irreversible increase of resting [Ca2+]i, at supraphysiological levels (100 μM).83 H2O2 elicits sustained Ca2+ spark activation at 50 μM in intact cardiomyocytes and permeabilized skeletal muscle fibres,61,84 but a bidirectional effect at 200 μM—a transient augmentation of cardiac Ca2+ sparks within 1 min followed by a suppression after 10 min exposure.61 A biphasic, bidirectional response of the Ca2+ transient to even higher doses of H2O2 (1 mM) has been reported in guinea-pig cardiomyocytes.85 In permeabilized rat cardiomyocytes, exogenous O2•− generated from the xanthine/xanthine oxidase system stimulates Ca2+ sparks in the first 3 min followed by a depression of spark frequency and amplitude.86
It has been demonstrated that photostimulated RIRR enhances local Ca2+ spark incidence.55 More recently, the Wang group has shown that pacing of cardiomyocytes (0.5–2 Hz) stimulates mitochondrial superoxide flash activity in an Ru360-sensitive manner,87 probably because the mitochondrial Ca2+ uniporter is a synergistic regulator of superoxide flash production.51 More spontaneous Ca2+ sparks are clustered near the flashing mitochondria than elsewhere,61,87 demonstrating an elegant reciprocal regulation between Ca2+ sparks and superoxide flashes at the dyads, under physiologically relevant conditions. During a mitochondria-derived cell-wide ROS transient elicited by photostimulation or antimycin A, Ca2+ sparks are initially increased61,88 followed by sustained depression.61 Regulation of Ca2+ sparks by cyclic ROS oscillations of RIRR appears to be intriguing: the first ROS oscillation produces a two-fold phasic Ca2+ spark activation, whereas the subsequent oscillations produce small phasic depression, suggesting a ‘mode-switch’ of the RyR response to ROS.61 In another study, cyclic small-amplitude Ca2+ spark frequency oscillations were found to coincide with oscillations of NADH, GSH, and ROS, with increased Ca2+ spark activity at the peaks of ROS production (troughs of the TMRM and NADH signals).88
5.1.2 Effects of stretch-activated X-ROS and RNS
Under physiological conditions, X-ROS play a significant role in sensitizing CICR. Specifically, an immediate, short-lived (10 s) 30% increase in the spark rate in response to a static 8% stretch, an effect absent in Nox2−/− cardiomyocytes and exaggerated in mdx cardiomyocytes,72 is consistent with the stretch-induced X-ROS production described above. Likewise, stretch increases the number of Ca2+ sparks recruited during EC coupling (in the presence of submaximal LTCC blockers). It also promotes Ca2+ wave production, more so in mdx cardiomyocytes.72 In a related study, Belmonte and Morad89 found that ‘pressure flow’ (PF) triggers a slowly developing (300 ms), 1.7 s [Ca2+]i transient. The response is independent of Ca2+ entry and sarcolemmal Ca2+ transporters and channels, but accompanies a mitochondrial Ca2+ release. The involvement of SR and RyR has not been fully explored, however. It remains unclear whether PF- and X-ROS-induced Ca2+ signals are of the same or similar origin.
Mechanical stretch of single cardiomyocytes embedded in agarose-filled polyethylene tubing increases the Ca2+ spark activity by up to 400% after a 10 min, 10% stretch.90 Meanwhile, there is a reversible increase in the amplitude of electrically evoked [Ca2+]i, transients.90 These Ca2+ spark and global Ca2+ responses are reversible and graded by the degree of stretch and are absent in eNOS−/− cardiomyocytes.90 A stretch activation of the eNOS-Akt-PI3K pathway has been suggested to underlie these responses.90
As is the case with stretch-activated RNS but in contrast to the phasic and bidirectional effects of ROS and X-ROS, exogenous RNS effects appear to be excitatory and more enduring in general. For instance, S-nitroso-N-acetylpenicillamine, an •NO donor, reversibly increases Ca2+ spark frequency.90 Angeli's salt as a donor of nitroxyl (HNO), the one-electron reduction product of •NO, increases the RyR open probability in planar lipid bilayers and the ATP-dependent Ca2+ uptake in SR vesicle preparations. In intact cardiomyocytes, it markedly increases Ca2+ spark frequency and enhances the [Ca2+]i transient while accelerating its relaxation, in a redox-sensitive and protein kinase G-independent manner.91
That ROS regulate Ca2+ spark activity and reciprocally, Ca2+ regulates ROS generation, allows for an intimate co-ordination and integration between the Ca2+ and ROS signalling systems. The coupling of ROS/RNS to stretch adds yet another dimension to the regulation of cardiac Ca2+ and ROS signalling and biomechanics. A pull or push of tissue generates two acoustic waves with different properties, the P-wave, which is compressional or tensile and the S-wave which is a shear wave. The former propagates at ∼1550 m/s and the latter at 1–5 m/s (∼2 m/s in skeletal muscle).92 It is clear that the P-wave (i.e. the stretch caused by preload in a cardiac cycle) traverses the entire myocardium in a fraction of a millisecond, faster than the propagation of action potentials (∼1.5 m/s in the bundle of His). Assuming the X-ROS mechanism operates in the intact heart, the mechanical stretch precedes waves of electrical excitation, and triggers instantaneous X-ROS production so to prime the myocardium by sensitizing the CICR mechanism, for stronger contraction when excited. In this regard, the newly described stretch-activated NOX2-dependent ROS (X-ROS) may provide a mechanism for the ‘biomechanical synchrony’ of the heart, to ensure homogenous force development and workload partition on a beat-to-beat basis.
The Frank–Starling law of the heart states that increasing end-diastolic pressure (hence stretch of the myocardium) induces a more powerful peak systolic pressure (hence a more powerful contraction) (Figure 2). A large component of the stretch/length dependence has been attributed to the rearrangement of contractile filaments and the enhancement of their Ca2+ sensitivity.93 However, it is conceivable that X-ROS-enhanced local Ca2+ signalling also makes a significant contribution, provided that it outweighs the direct inhibitory effects of ROS on myofilament contractility.72,79 When compared with the X-ROS, RNS might be a slowly developing signal, because the immediate Ca2+ spark response (within 10 s) is insensitive to the inhibition of NOS by l-NAME. Thus, RNS appear to have little to do with the hypothetical biomechanical synchrony. Nevertheless, they are enduring signals, and exert a greater effect on Ca2+ sparks (∼400 vs. ∼30% increases in the Ca2+ spark rate at ∼10% stretch, for X-ROS and RNS, respectively).90 This prominent RNS effect may provide an important cellular mechanism to explain the Anrep component of the Frank–Starling effect: an enhanced [Ca2+]i transient and further enhanced contractility develop over 10 min after a stretch (Figure 2).94,95 Future investigations are warranted to demonstrate stretch-activated X-ROS and RNS and their Ca2+ and biomechanical consequences in the beating heart.
Possible physiological roles of stretch-activated X-ROS and RNS. Rapidly-responding X-ROS may enhance biomechanical synchrony during beat-to-beat contraction (A), while X-ROS and stretch-activated RNS may underlie the Anrep effect of the heart (A and B).
5.2 ROS Regulation of RyR: a complex story
The cardiac RyR is a homotetramer with each subunit containing ∼21 free cysteine residues,96 amounting to ∼84 thiols in a channel, and ∼8000 in a Ca2+ release unit . Some of the hyperactive SH groups on the RyR appear to have a redox-sensing function, conferring on it the ability to act as a redox sensor.97–100 The oxidation of critical SH bonds within the RyR complex is responsible for opening the channel pore.101 In purified cardiac RyR channels incorporated into planar lipid bilayers, H2O2 increases the channel open probability and this effect is reversed by the SH-reducing agent dithiothreitol (DTT).102 Similarly, OH• and SH-oxidizing reagents react with some SH groups of the RyR and increase the open probability of the channel.103 Reagents that specifically oxidize free-SH groups and promote the formation of disulfide bonds within the RyR complex, such as 2,2′-dithiodipyridine, activate the RyR, and this effect is reversed by reagents that reduce disulfides to thiols (e.g. DTT and glutathione).104 These findings suggest that ROS react with the active SH groups of the RyR and increase Ca2+ release from the SR. In addition to direct regulation of SH, ROS can regulate RyR Ca2+ release through indirect pathways by altering the calmodulin binding.105,106 Calmodulin reduces RyR Ca2+ release by direct interaction.107 ROS, especially O2•−, trigger Ca2+ release by displacing the binding of calmodulin from RyR.108
Given the diversity of ROS species and ROS effector proteins and the promiscuity of the ROS-reactive sites on the RyR, it can be understood that the regulation of Ca2+ sparks by ROS can be both stimulatory and inhibitory, depending on the dose, the history and duration, and the cellular context. In particular, the aforementioned ‘mode-switch’ response differs from conventional channel inactivation or ‘adaptation’. Instead, it may be accounted for by a simple four-state hypothetical model in which the RyR contains two ROS-reactive sites of different kinetics and affinity (Figure 3). Oxidation of the fast but low-affinity site alone increases the channel open probability, while oxidation of the slow but the high-affinity site is functionally neutral by itself but brings the channel to a state such that subsequent oxidative modification negatively regulates the channel activity, manifesting as a modality switch. This model resembles the model of Abramson and Salama101 with one fundamental difference. Their model for redox modification of the RyR involves three sulfhydryl groups in close proximity: thiol oxidation that forms a disulfide bond opens the channel, and subsequent thiol-disulfide interchange closes the channel. Our model, however, hypothesizes that a slow ROS modification primes the channel to render the opposite response to subsequent ROS modification (Figure 3). Experimental testing of these models would provide new insights into the complexity and subtlety of ROS regulation of effector proteins in general and RyR in particular.
A hypothetical four-state model for a mode-switch response of the RyR to oscillatory ROS signals. (A) Oxidation of the fast and low-affinity moiety increases the channel open probability from P0 to 20P0 if the slow but high-affinity moiety is in its reduced state. Oxidation of the slow but high-affinity moiety per se does not alter the channel activity if the fast moiety is in its reduced state, but closes the channel (P = 0) if the other moiety is also oxidized. As such, the slow moiety acts as a switch, oxidation of which converts the oxidation of the fast moiety from spark-activating to spark-suppressing. (B) RyR activity in response to a two-fold step increase in ROS and a subsequent ROS oscillation. Note that the basal ROS level ([ROS]b) maintains RyR activity at a level (∼3 P0) about double that if the RyR were fully reduced (P0).
6. Pathological relevance
While clinical interventions with antioxidants have thus far proven to have little or no impact on cardiac disease risk and progression,109–111 this failure may stem in part from the fact that a complete picture of ROS involvement in heart diseases is still lacking. On the other hand, increasing evidence shows that the redox state of the failing heart is altered towards greater oxidation which directly or indirectly affects Ca2+ handling. The antioxidant defence, such as SOD and thioredoxin, is decreased in heart failure (HF);112 ROS production by mitochondria,113–120 NOX121–124 and NO generation by NOS76,125–128 are increased in hypertrophy, HF or ischaemia.
In HF, elevated oxidation of the RyR tends to increase its open probability, enhance SR Ca2+ leakage during diastole, deplete SR Ca2+ content, and decrease [Ca2+]i transient amplitude, hallmarks of HF.129 The depressed SR Ca2+ load depends not only on RyR hyperactivity, but also on SERCA oxidation, which tends to inhibit its activity130 (see also91 for RNS enhancement of pump activity). The fact that these defects are partly normalized by reducing agents129,131,132 points to the importance of RyR oxidation in HF. Clinical data have also shown that inhibition of xanthine oxidase in HF patients ameliorates the ejection fraction after 1 month of treatment.133 Taken together, these data strongly suggests that ROS are culprits driving the disease progression in HF.
In addition to undermining systolic [Ca2+]i transients and contractile function, diastolic Ca2+ leakage through oxidized RyRs is also arrythmogenic. Excessive Ca2+ release during diastole, particularly in the form of Ca2+ waves, causes membrane depolarization when it is extruded through the electrogenic NCX, and therefore underlies arrhythmogenic afterdepolarizations. When the background ROS level is high and the redox balance is shifted towards oxidation, as in mdx cardiomyocytes,134 stretch-activated X-ROS oxidize the RyR to generate a greater Ca2+ spark response and trigger more Ca2+ waves,72 revealing the double-edged-blade nature of X-ROS (as it is true for any second messenger). Elevated RyR oxidization in an HF model with prolonged β-adrenergic stimulation135 has also been linked to pro-arrhythmogenic Ca2+ waves. Intriguingly, the ROS mechanism seems to be intertwined with the classic phosphorylation mechanism of post-translational modification of the RyR. The effect of RyR oxidation is synergistically enhanced by its PKA- and CaMKII-dependent phosphorylation, because both kinases are activated by elevated oxidation136 and both contribute to the RyR leak in cardiomyopathy.23,137 In a dog model of sudden death, ROS-mediated oxidation of RyR is involved in arrhythmogenic Ca2+ alternans by increasing the fractional SR release, RyR sensitivity to Ca2+ and diastolic leak, defects that are normalized by reducing agents.138 Arrhythmogenic early after depolarizations may also be activated by oxidation, acting either directly on the LTCC139 or though CaMKII activation, which enhances Ca2+ entry through LTCC.140–142
Besides oxidizing RyR and other key Ca2+-handling proteins, ROS also contribute to cardiac pathology upstream and through indirect mechanisms. In fact, ROS have been suggested to contribute to angiotensin II-induced cardiac hypertrophy by NF-AT translocation,143 thus participating in excitation–transcription coupling. All the pathological effects of oxidation may involve CaMKII, which is both pro-hypertrophic and pro-arrhythmogenic,144 and is activated by oxidation.22
In conjunction with oxidative stress, enhanced SR Ca2+ leak and elevated phosphorylation-mediated signalling, conspicuous structural remodelling at the levels of TT-SR dyads and/or TT-SR-mitochondria trios has also been reported in the hypertrophied and failing heart.145,146 The TT-SR dyadic remodelling, accompanied by changes in the structural proteins such as junctophilin, has been confirmed in various HF models,147–149 resulting in orphaned RyRs and dyssynchronous Ca2+ sparks during EC coupling.149 These functional changes lead to eroded safety margin of EC coupling147 or overtly defective local EC coupling and increased propensity for arrhythmogenic Ca2+ release.148,149 Future investigations are warranted to unravel how such dyadic remodelling affects local production of signalling ROS (X-ROS and superoxide flashes), overall ROS homoeostasis (constitutive ROS production), and ROS stability (i.e. RIRR and ROS oscillations), and vice versa.
As one of the most exciting developments in cardiac signalling, recent advances show that ROS as well as RNS modulate microdomain and global [Ca2+]i transients under physiological, stressful, and pathophysiological conditions. The excitement is four-fold. First, the realization that virtually all the molecular players of EC coupling and Ca2+ handling are authentic or potential effectors of ROS/RNS signalling. Secondly, the visualization of dynamic, regulatory, and local signalling ROS with their microdomains encompassing the dyads. These include mitochondrial superoxide flashes with regulatory mechanisms distinctly different from that for constitutive, homoeostatic ROS generated as byproducts of respiration. Moreover, mechanical stretch triggers X-ROS from the dedicated ROS-generator NOX2 in TTs as well as RNS from eNOS in cardiomyocytes. Thirdly, the demonstrated roles of physiological signalling ROS/RNS in modulating dyadic Ca2+ signalling events—Ca2+ sparks. Fourthly, the increasing appreciation of similarities and distinctions between ROS and Ca2+ signalling in terms of the underlying design principles. With these, it is safe to conclude that ROS signalling adds a new layer of regulation of cardiac EC coupling, in addition to the classic phosphorylation mechanism mediated by kinases and phosphatases.
Although powerful, ROS signalling appears to operate with great stability, such that local ROS are promptly terminated and spatially well-confined. However, when cardiomyocytes are challenged with extreme stressors, inter-mitochondrial RIRR takes place and gives rise to hierarchical spatiotemporal ROS dynamics, revealing intrinsic instability of the ROS signalling system. Dysfunctional and instable ROS signalling, in turn, causes dysfunction and instability of Ca2+ signalling. In fact, oxidative stress has never been so intimately intertwined with Ca2+ overload and dysregulation. Combined with deranged phosphorylation-mediated signalling,150,151 it leads to compromised contractility and exaggerated arrhythmogenicity.
With a large array of interconvertible molecular players, dozens of possible redox-dependent target protein modifications, and only a few specific target sites amidst at least 10-fold more ‘non-specific’ ones on a given target protein, the ROS signalling is notoriously complex—apparently much more complex than the Ca2+ signalling that we know. As a rule of thumb, ROS effects are multiphasic and bidirectional, depending on the species of oxidants, their concentrations, history of exposure, and cellular context. By analogy to Ca2+ signalling, however, we hold the strong belief that under the surface of apparent complexity, there is simplicity at the heart, awaiting elucidation by future investigations. Mastery of the underlying design principles will not only deepen our understanding on how ROS are harnessed to play signalling roles, but also give us the ability to constrain deleterious ROS effects. For instance, rather than the proven failure of indiscriminate scavenging of ‘good’ and ‘bad’ ROS,152 targeted antioxidant therapy could be developed based on precise knowledge of the offending ROS species, compartments, concentrations, and modes of action. Towards that goal, the chasm of insufficient knowledge remains wide and deep, signifying tremendous opportunities for future endeavours.
This work was supported by the 973 Program of China (2013CB531200, 2011CB809102) and the National Natural Science Foundation of China (31130067 and 31221002), the Agence National de la Recherche (ANR-09-Geno-012 and ANR-09-Geno-034), CODDIM (2010) and FRM (programme cardiovasculaire 2011). U-769 is a member of the Laboratory of excellence LERMIT, supported by an ANR grant ‘Investissements d'avenir’.
We thank Drs Iain C. Bruce for editing and Wang Wang for valuable discussion.
Conflict of interest: none declared.
This article is part of the Spotlight Issue on: T-tubules and ryanodine receptor microdomain signalling in cardiac hypertrophy and failure.
. Modulation of the beta-adrenergic response of cardiomyocytes by specific lipoxygenase products involves their incorporation into phosphatidylinositol and activation of protein kinase C. J Biol Chem 1994;269:29055-29060.
. Localization of endothelial nitric-oxide synthase phosphorylated on serine 1179 and nitric oxide in Golgi and plasma membrane defines the existence of two pools of active enzyme. J Biol Chem 2002;277:4277-4284.
. Direct evidence for the existence and functional role of hyperreactive sulfhydryls on the ryanodine receptor-triadin complex selectively labeled by the coumarin maleimide 7-diethylamino-3-(4′-maleimidylphenyl)-4-methylcoumarin. Mol Pharmacol 1994;45:189-200.
. Effects of hydroxyl radical and sulfhydryl reagents on the open probability of the purified cardiac ryanodine receptor channel incorporated into planar lipid bilayers. Biochem Biophys Res Commun 1998;249:938-942.
. Cooperative interaction between reactive oxygen species and Ca2+ signals contributes to angiotensin II-induced hypertrophy in adult rat cardiomyocytes. Am J Physiol Heart Circ Physiol 2012;302:H901-H909.