|Home | About | Journals | Submit | Contact Us | Français|
Mitochondria form junctions with the sarco/endoplasmic reticulum (SR/ER), which support signal transduction and biosynthetic pathways and affect organellar distribution. Recently, these junctions have received attention because of their pivotal role in mediating calcium signal propagation to the mitochondria, which is important for both ATP production and mitochondrial cell death. Many of the SR/ER-mitochondrial calcium transporters and signaling proteins are sensitive to redox regulation and are directly exposed to the reactive oxygen species (ROS) produced in the mitochondria and SR/ER. Although ROS has been emerging as a novel signaling entity, the redox signaling of the SR/ER-mitochondrial interface is yet to be elucidated. We describe here possible mechanisms of the mutual interaction between local Ca2+ and ROS signaling in the control of SR/ER-mitochondrial function.
The cytoplasm is a highly restrictive medium for the diffusion of charged compounds and ions. Thus, intracellular signaling systems utilizing ionic messengers frequently operate via local communication between the sources and targets of the messenger molecules [1, 2].
The SR/ER and mitochondria are probably the two most extensive intracellular membrane systems. The SR/ER takes shape as a luminally continuous reticular network , while the mitochondria exist as an array of individual organelles that dynamically change their shape and size via fusion and fission capable of forming luminally continuous regional networks . Due to their abundant presence throughout the cytoplasm and dynamic nature, close encounters between SR/ER and mitochondria are predictable. Indeed, close SR/ER-mitochondrial contacts, where the interorganellar gap is ~10–50 nm have been documented in different tissues [5, 6]. However, it has turned out that the contacts are not just stochastic ‘collisions’ but are sturdy junctions secured by protein tethers [6–8]. SR/ER fragments remain associated with the mitochondria after fractionation [6, 9], allowing the isolated SR/ER-mitochondrial complexes to sustain local interorganellar Ca2+ communication [6, 10] and collaboration in phospholipid biosynthesis .
The hunt for the specific proteins that contribute to the SR/ER mitochondrial linkage has recently intensified. The heterogeneous length of the ER-mitochondrial tethers in non-muscle cells, suggests that several proteins might participate in linking the organelles together . Biochemical and functional measurements have implicated a handful of proteins/protein complexes as ER-mitochondrial coupling elements [DLP-1/DRP1-1[12, 13], autocrine motility factor receptor (AMFR) and BAP-31 , PACS2 , grp75, a chaperone forming a complex with type 1 voltage-dependent anion channel (VDAC1) and type 1 IP3 receptor (IP3R1) , type 3 IP3R (IP3R3)  and mitofusin-2 (Mfn2) ]. Furthermore, the main Ca2+ conducting channels of both SR/ER (ryanodine receptors (RyRs) and IP3Rs) and outer mitochondrial membrane (VDACs) form multimolecular complexes (a few protein partners are listed in Figure 1) that can also interact with adjacent membranes. This mechanism might be particularly relevant for the IP3Rs that seem to be concentrated at the ER-mitochondrial interface  but dispensable for RyR2s that concentrate on the sarcolemmal side the SR in the heart . Additional physical support of the SR/ER mitochondrial associations is provided by cytoskeletal proteins, since both SR/ER and mitochondria bind to microtubules/actin filaments (for a recent review see ). Thus, the overlapping spatial arrangements of SR/ER and mitochondria and the interorganellar physical linkage provide a robust platform for local communications between the organelles. However, it is important to note that both SR/ER and mitochondria are continually reorganized, indicating that the SR/ER junctions likely form dynamic structures.
Both SR/ER and mitochondria can accumulate Ca2+. The Ca2+ storage capacity of the SR/ER is limited, but its luminal [Ca2+] ([Ca2+]ER) is close to the millimolar range, promoting the protein folding/processing functions [3, 21]. Due to the 3–4 orders of magnitude [Ca2+] gradient across its membrane, to the presence of messenger-operated Ca2+ release channels, to the robust Ca2+ pumping mechanism and to its cell-wide distribution, the SR/ER can serve the generation of rapid cytoplasmic [Ca2+] ([Ca2+]c) transients. On the other hand, the low-affinity, high-capacity and Ca2+-gated Ca2+ uptake mechanism enables mitochondria to respond effectively to local calcium signals.
is established via the SERCAs. Out of the seven isoforms usually found in mammalian cells, the predominant subtype in non-muscle cells is SERCA2b, in skeletal muscle SERCA1a and in cardiac muscle SERCA2a. The activity of these high affinity (2b>1a~2a) and velocity (1a>2a>2b) Ca2+ pumps is regulated by substrate/product levels ([ATP]/[ADP], [Pi]), pH, [Ca2+]c and [Ca2+]ER, by regulatory proteins (eg. phospholamban, sarcolipin) and by the redox state of critical reactive thiols [22, 23]. The luminal Ca2+ regulation of SERCA2b involves the ER chaperone ERp57 that is also a target for ROS/redox regulation (see under ROS targets: SERCA).
during [Ca2+]c spikes occurs through the RyR and IP3R, which are non-specific cation channels. Both RyRs and IP3Rs exist as tetramers of subunits that contain large N-terminal cytoplasmic domains and luminal loops engaging in numerous regulatory interactions with other proteins [24–26]. RyRs mediate Ca2+-induced Ca2+ release from the SR of skeletal (RyR1) and cardiac (RyR2) muscle in the process of excitation contraction coupling or from the SR/ER of smooth muscle, diaphragm and non-muscle cells (predominantly RyR3) . For IP3Rs three different isoforms have been identified, and most cells appear to express multiple isoforms.
RyRs are subject to both cytosolic and luminal Ca2+ regulation. In principle, Ca2+ conductance is directly related with both the local [Ca2+]ER and [Ca2+]c and the Ca2+ regulation is established via proprietary Ca2+ binding domains and via interactions with Ca2+ binding proteins eg. calsequestrin/triadin/junctin complex [25, 28, 29]. Activation of the IP3Rs is also controlled by Ca2+ but it requires the binding of IP3, which is directly modulated by the competitive ‘pseudoligand’ IRBIT . In addition to channel opening, IP3 binding leads to time-dependent inactivation . The [Ca2+]c regulation of IP3Rs is biphasic at physiologically occurring [Ca2+]c ranges with isotype and species-dependent differences in the stimulatory and inhibitory range [24, 30]. [Ca2+]ER regulation of the IP3R involves Ca2+ binding and redox-dependent chaperones including calnexin and ERp44 (see later in more details) . Besides Ca2+ regulation, the cytosolic domain of both RyR and IP3R provides a scaffold to a host of regulatory protein interactions giving rise to macromolecular complexes able to receive inputs from most of the major signaling pathways and to sense the metabolic status of the cell [24–26, 29, 30, 33, 34].
Due to the Ca2+-induced activation of both RyR-and IP3R, these channels operate more effectively in clusters than being stochastically dispersed along the SR/ER membranes. Indeed, cardiac and skeletal RyRs are concentrated in dense arrays mostly to the SR interfaces with the T-tubules (junctional SR) . This orderly distribution is ensured by multiple accessory proteins, including FKBP12.6 and is critical for the synchronized activation of the sarcomeres over the excitation-contraction coupling . IP3Rs are also distributed in groups that function as distinct release units  and display enhanced clustering when activated [36, 37]. The exact mechanisms of IP3R positioning and group formation are still yet to be clarified (). Recent data suggest higher prevalence of Ca2+ release units  and chaperone-mediated (Sigma-1 receptor) protection of activated IP3R3s against proteasomal degradation  at the ER-mitochondrial close contacts.
Cytoplasmic Ca2+ has to cross both the outer and inner mitochondrial membranes (OMM, IMM) to enter the matrix. The predominant pathways for Ca2+ diffusion across OMM are the VDACs. Because of their high density and high conductance, VDACs allow rapid Ca2+ diffusion . Still, VDACs may become a Ca2+ transport limiting factor during delivery of microdomain-derived Ca2+ signals to the mitochondria [40, 41]. Evidence has been presented that the VDAC-mediated Ca2+ and solute flux is stimulated by Ca2+ [42, 43]. Additional regulatory mechanisms of the Ca2+ flux are likely to result from the VDAC’s phosphoregulation and direct interaction with several proteins. The primary driving force for Ca2+ entry into the mitochondrial matrix is the inside-negative membrane potential across the IMM maintained by the proton-extruding activity of the electron transport chain (ETC). In most cellular and subcellular paradigms, mitochondrial Ca2+ uptake becomes apparent when the [Ca2+]c is increased to ≥1µM, though [Ca2+]m elevations have also been observed during submicromolar [Ca2+]c increases in some cases . This may be supported by Ca2+-induced allosteric activation of the mitochondrial Ca2+ uptake [44, 45]. The molecular identity of none of the IMM Ca2+ transport mechanisms is known. The most recognized mechanism via which Ca2+ enters the mitochondrial matrix is the Ca2+ uniporter . The uniporter has been functionally characterized by patch-clamping mitoplasts as a ruthenium red-sensitive and highly selective inward-rectifying low-affinity Ca2+-gated Ca2+ channel [47, 48]. Recently, uncoupling proteins 2 and 3 have been implicated in the Ca2+ uniport mostly based on RNA silencing/rescue experiments but this finding remains a subject of discussion [49, 50]. Besides the uniporter, a rapid uptake mode and a mitochondria-located RyR (mtRyR1) have been proposed to participate in the Ca2+ flux across the IMM in cardiomyocytes and neurons but their physiological relevance and regulation is yet to be confirmed [51–53]. An important feature of the mitochondrial Ca2+ uptake is the time-dependent sensitization and desensitization observed in different cellular systems (sensitization: [45, 54]; desensitization: [55–57]) that might have profound role in establishing the relationship between the periodicity and efficacy of mitochondrial delivery of [Ca2+]c oscillations. The mechanism underlying sensitization or desensitization of mitochondrial Ca2+ uptake is yet to be clarified; currenly proposed paradigms involve Ca2+/CaM for the sensitization , PKC-ε , p38 MAPK  and the H+-flux through the F1F0-ATPase as a positive regulator  in the process of desensitization.
The [Ca2+]m increase enhances oxidative metabolism via activating Ca2+ sensitive matrix dehydrogenases (CSMDH) [61–63] that feed reducing equivalents to the ETC and the F1F0-ATPase itself . Therefore, the RyR/IP3R-mediated [Ca2+]m signal is rapidly followed by an increase in mitochondrial [ATP] [65, 66]. With the increased ETC activity mitochondrial superoxide (·O2−) generation also increases and as a compensatory mechanism, Ca2+ also activates ·O2−- scavenging enzymes (SOD2). However, after a point, the [Ca2+]m increase activates the opening of the permeability transition pore (PTP), allowing indiscriminate exchange of ions and small solutes between the mitochondrial matrix and the cytosol. The PTP opening dissipates mitochondrial membrane potential, might cause swelling and release of apoptotic factors from the intermembrane space (IMS) and initiation of the mitochondria-dependent apoptotic cascade . Importantly, mitochondria can hold elevated [Ca2+]m steadily below the toxic level even during extended mitochondrial Ca2+ accumulation by means of a Pi-dependent dynamic Ca2+ buffering/precipitating system [68, 69]. PTP opening seems to occur when Ca2+ uptake is very fast or when massive amounts of Ca2+ are accumulated.
is carried out mostly via cation exchangers [3Na+/Ca2+-exchanger (NCX), 3H+/Ca2+-exchanger] and perhaps transient openings of the PTP. The Na+-independent pathways display increased activity under oxidizing conditions in the mitochondrial matrix perhaps as a defense mechanism against the [Ca2+]m rise-associated enhancement of ROS production .
Depending on the location of the RyR and IP3R, the released Ca2+ can enter directly the SR/ER-OMM cleft or can spread in by diffusion when released via nearby channels. The former scenario is likely if the IP3R and VDAC are physically coupled . However, only the latter mechanism is plausible in skeletal and cardiac muscle, where RyRs are clustered at the triadic/dyadic clefts instead of the SR/OMM interface. Both mechanisms are competent to expose the mitochondrial Ca2+ uptake sites to high local [Ca2+]c since RyR/IP3R-dependent calcium signal propagation to the mitochondria could persist under [Ca2+]c clamping using a slow Ca2+ chelator (EGTA) in permeabilized cardiac and non-muscle cells [71–73]. Also, single mitochondrial [Ca2+]m elevations (“Ca2+ marks”) have been detected in cardiac cells during “elementary” Ca2+ release events (sparks) that fail to significantly elevate the global [Ca2+]c ). Nevertheless, synchronous activation of a large number of IP3Rs delivers Ca2+ to the mitochondria more effectively than random openings . During [Ca2+]c oscillations that peak at ≤1 µM in the bulk cytosol, the mitochondrial side of the interface can be exposed to >100µM [Ca2+] if the RyRs/IP3Rs are also in the interface area, and to ~10µM [Ca2+] if the RyRs/IP3Rs are located within 100nm from the contact region.
Disruption of the ER-mitochondrial physical linkage by limited proteolysis uncoupled mitochondria from IP3-induced Ca2+ release  and genetic interference with the tethering proteins grp75 and Mfn2 also reduced the efficacy of IP3R-dependent [Ca2+]m signal generation [16, 18]. Conversely, strengthening of the ER-mitochondrial linkage by synthetic linkers and Ca2+-mediated inhibition of mitochondrial motility close to the sites of Ca2+ mobilization led to more effective IP3R-mitochondrial Ca2+ transfer [6, 75]. Notably, Ca2+ has been shown to affect several aspects of ER and mitochondrial dynamics, including mitochondrial motility and fission and ER motility and restructuring, which may also mediate Ca2+-dependent changes in the ER-mitochondrial coupling [75–82]. Once Ca2+ entered the mitochondrial matrix, [Ca2+]m increases and Ca2+ removal mechanisms are activated to restore the ‘resting state’. Mitochondrial Ca2+ efflux might locally support ER Ca2+ reloading . Hyperactivity of Ca2+ extrusion might attenuate the beneficial stimulatory effect of the [Ca2+]m rise on oxidative ATP production as it has been observed in association with increased [Na+]c in failing cardiomyocytes [84, 85].
Enhanced production or external exposure to ROS and reactive nitrogen species (RNS) has been proven hazardous to cells causing various types of damage and accelerating cell death or senescence via multiple mechanisms collectively known as oxidative/nitrosative stress. However, a multitude of molecules participating in different cellular signaling or regulatory networks have redox active moieties (e.g. cysteine, methionine, tyrosine) at functional regions endowing them a ROS/RNS-sensor feature. Consequently, endogenously generated ROS and RNS have been emerging not just as harmful products but diffusible modulators of a range of different signaling cascades and even more, as messengers themselves [23, 86]. Many ROS/RNS sources and sensors are localized in the mitochondria or SR/ER and are relevant for calcium signaling. In this review, we focus on ROS as local messengers between the SR/ER and mitochondria (for a review on RNS see ). The main ROS sources and targets and pathways of ROS dynamics are depicted in Figure 2.
In principle, ROS are products of consecutive stepwise reduction of molecular oxygen:
where enzyme E1-3 represent the catalyzing enzymes. When E3 are saturated and Fe2+ is available, H2O2 is reduced to H2O via non-enzymatic Fenton reaction yielding a hydroxyl radical OH., the most aggressive ROS:_
E1 represents mitochondrial and non-mitochondrial oxidases, out of which the most significant ·O2− sources are complex I and III of the electron transport chain (ETC) and NADPH oxidases associated with the plasma membrane or the SR/ER ([88, 89]). In the highly reducing environment of mitochondria, a number of potential e− donors are capable of reducing O2 to ·O2−. However, it is mostly the e− leak via complexes I and III that cannot be completely overcome by the powerful mitochondrial ROS scavenging mechanisms (). The relative contribution of respiratory complex I and III to the mitochondrial ·O2− generation appears to be cell/tissue type-and respiratory status-dependent. In cardiac muscle, with fully respiring (state 3) mitochondria and in the lung the ubiquinone (Q) cycle of complex III is the primary source of ·O2− generation. Since autooxidation of Q can happen at either side of IMM, it can increase ·O2− in both the matrix and the IMS. In the brain and in state 4 respiration (absence of ADP) the NADH oxidase complex I is the dominant ·O2− source at the inner side of IMM (). Importantly, complex I has been reported to work as a redox sensor, having highly reactive thiol groups that become S-gluthationylated when GSH/GSSG ratio decreases (e.g. under oxidative stress) causing decreased e− flow to complex III but increased ·O2− production [91, 92]. The rate of mitochondrial ·O2− production is determined in principle by simple mass action with an inverse relation to the electron flow over the ETC and with a direct proportionality with the available O2 and electron donor (R·).
In vitro, about 1–4% of the O2 entering the ETC is reduced to ·O2− [88, 90], though some groups have published even smaller fraction (~0.1% [94, 95]). The activity of the tricarboxylic acid cycle (TCA), which provides a vast source of e− to the ETC in the form of NADH is a crucial determinant of the rate of mitochondrial ·O2− production.
The highly regulated TCA is an ideal platform to connect mitochondrial ·O2 − production with signaling systems; in particular with Ca2+ signaling via the CSMDH. In addition, the [4Fe-4S] iron-sulfur cluster of aconitase is reversibly inactivated by ·O2− (to [3Fe-4S]) serving as a ‘rheostatic’ limit switch on oxidative metabolism and fuel (ATP) production that keeps radical ‘byproducts’ in a non-toxic range [96, 97]. Frataxin, an iron chaperone protein mutated in Friedrich’s ataxia appears to play an important role in maintaining the reversibility of the oxidation of the iron-sulfur cluster of aconitase . Importantly, if sufficient glutamate is present, the activity of aconitase will not limit the NAD(P)H output of the TCA as α- ketoglutarate dehydrogenase (α-KGDH), one of the Ca2+-sensitive rate-limiting enzymes of the TCA cycle can bypass the segment between oxaloacetate and α-KG . Nevertheless, α-KGDH is also inhibited by ROS but with less sensitivity than aconitase. Moreover, the α-KGDH complex can generate significant amount of ·O2− upon excess in [NADH] [100–102] making α-KGDH a main regulatory node for mitochondrial energy metabolism and ROS production. ROS regulation of and ROS production by α-KGDH appears to be particularly significant in neuronal cells and is suspected to contribute to ROS-related neuronal degeneration [102, 103].
NADPH oxidases (Nox) have been recognized as the ‘superoxide guns’ of phagocytic immune cells to kill bacteria but later other isoforms of the enzyme have been discovered in the plasma membrane and SR/ER of non-phagocytic cells [104, 105]. It has been recently suggested that the ·O2− generation by an SR-associated Ca2+-sensitive Nox is regulated by cADP-ribose activation of RyR in coronary artery smooth muscle cells and functions as a local calcium signal amplifier [106, 107]. Along this line, an SR-associated NADH oxidase has been found in skeletal muscle that produced ·O2− upon substrate addition and was proposed to be responsible for the sensitization of RyR1 by NADH . However, thus far there is no experimental evidence that ROS derived from SR/ER-resident Nox would affect mitochondria.
Nitric oxide synthase (NOS) is present in various organelles, including the plasma membrane, SR/ER and mitochondria. NOS is an NADPH/FAD oxidoreductase that produces ·NO from the substrate L-arginine. This process requires the co-factor (6R)-5,6,7,8,-tetrahydrobiopterin (BH4). In the absence of the substrate or co-factor, the oxidoreductase function becomes uncoupled from ·NO generation and produces ·O2− [87, 109]. A recent paper indicates that a mitochondrial NOS can be activated by Ca2+ and when uncoupled, it may promote opening of the PTP and apoptosis in cardiomyocytes .
Direct H2O2 generation: Besides ·O2− dismutation, H2O2 can also be directly generated in the mitochondria. The 66kDa isoform of the growth factor adapter Shc (p66Shc) has been recently recognized as a key promoting factor in mammalian aging . Upon oxidative stress, p66Shc accumulates to the IMS via a PKCβ-and PIN1 (a peptidyl prolyl isomerase) dependent manner where it displays cytochrome c oxidase activity generating H2O2 and so promotes IMM permeabilization and mitochondria-dependent apoptosis  . The OMM-bound monoamino oxidase (MAO) is another local source of H2O2 as a byproduct of the oxidation of biogenic amines and is believed to play a significant role in the mitochondrial damage and PTP activation in several neurological disorders .
[ROS] in the intracellular compartments is tightly controlled via scavenging mechanisms to maintain subtoxic levels (Fig 2). ·O2− scavenging is established via dismutation to H2O2 by superoxide dismutases with Mn2+ (SOD2 in the mitochondrial matrix) or Zn2+/Cu2+ (SOD1 in the IMS and cytosol) centers. H2O2 scavenging is established via reduction to H2O using glutathione dependent and independent pathways; the former one being dominant. Both pathways utilize NADPH as e− donor provided by the pentose phosphate cycle (P5P in Fig2) in the cytosol or by transhydrogenation from NADH in the mitochondria . Thus, maintaining the level of reducing equivalents in the mitochondrial matrix is critical for both, the oxidative ATP production as well as the balance between oxidized and reduced thiols in the mitochondria. In the glutathione dependent pathway, gluthatione reductase (GR) transfers the electron from NADPH to oxidized glutathione (GSSG) keeping most of it in the reduced state (GSH) under physiological conditions. GSH can then transfer its reducing equivalent to glutathione peroxidase (GPx) that catalyses the reduction of H2O2 to water . Notably, glutathione is synthesized in the cytosol and it is transported into organelles via glutathione transferases. Mitochondrial glutathione is 10–15% of the total cellular glutathione, but its concentration is similar to the cytosolic one (5–10 mM in rat liver) .
The glutathione-independent H2O2 scavenging system transfers the electron from NADPH to H2O2 an electron transport chain comprised of thioredoxin reductase (TrxR1 cytosol, TrxR2 mitochondria), thioredoxin (Trx1 and Trx2) and peroxiredoxin (Prx1 and Prx2) in the order TrxR→Trx→Prx→ H2O2. This system has less scavenging power than the glutathione-dependent system; however, the redox state of Trx limits its capability of recovering S-glutathionylated proteins (see below) [23, 90, 115]. Another powerful H2O2 scavenging enzyme is catalase that is predominantly confined to peroxisomes  and as such, is probably less relevant in the ER-mitochondrial local redox signaling.
Importantly, ·O2− also readily reacts with nitric oxide (NO) when present and NO· has been even suggested as a ·O2− scavenger in the late ‘80s . However, the reaction product peroxinitrite (ONOO−) anion has been proven highly reactive, bearing more oxidative power than its parent molecules; and even more, its solute interactions may further produce aggressive radicals (e.g. hydroxyl and carboxyl radicals) . Moreover, ONOO− is membrane permeant thus easily crosses compartmental barriers. ONOO− is the center figure in nitrosative stress and has been progressively ‘claiming’ responsibility for effects attributed to ROS. Thus, a part of oxidative effects of ·O2− is actually mediated by ONOO−.
Because of its negative charge and high reactivity (redox potential against H2O2 Eh~ +0.94V), ·O2− cannot permeate through lipid membranes and the bulk of it is rapidly dismutated to H2O2 by SODs. The steady state [·O2−] in the mitochondrial matrix has been estimated being in the range of 100–200pM based on calculations considering ·O2− production by the ETC and consumption by SOD2 in isolated liver and heart mitochondria . ·O2− can leave mitochondria via anion channels (the VDAC in the OMM  and, perhaps through the inner membrane anion channel (IMAC) and VDAC as part of the peripheral benzodiazepine receptor complex at the contact points of IMM and OMM [121, 122]) or via the PTP. Because of its electroneutrality and relatively moderate reactivity (Eh~ +0.38V) H2O2, is considered freely permeable to biological membranes (although see [123, 124] for potential limitations) and thus widely recognized as a diffusible messenger in ROS-associated redox signaling. Thus, the combination of the two ROS intermediates enables both tightly confined short-lasting (microdomain) operations as well as longer lasting more distant signaling actions .
In summary, mitochondrial ·O2− production is tightly controlled at membrane-bound main sources utilizing ‘rheostatic’ feedback loops on the reducing equivalent supply (aconitase, αKGDH and powerful clearance/buffer systems (SOD2, GPx) to secure a subtoxic basal setpoint. Both, ·O2− production and clearance are regulated in part, by Ca2+ and thus can be targeted by calcium signaling . The recent visualization of ·O2− ‘flashes’ in single mitochondria indicates that ·O2− in individual mitochondria can be rapidly and independently regulated . However, visualization of cell-wide synchronized oscillations and waves of mitochondrial ROS has also provided evidence for regenerative mechanisms of propagation [122, 128, 129]. Expansion of the local ROS events is likely to utilize rapid “metabolism” of ·O2− to H2O2 that spreads effectively among the individual mitochondria. In contrast with Ca2+, the ROS ‘messengers’ are consumed by their targets.
The most common molecular targets of ROS are the Cys thiol groups. Protein Cys thiols can be divided into four groups based on their prevalent interactions, which are determined by their immediate intra-and perimolecular environment: those that form permanent structural disulfide bonds, those that coordinate metals, those that are permanently in the reduced state, and those that are reversibly oxidized . The latter group is often referred to as (hyper)reactive thiols and plays the lead role in the oxidative/redox regulation of proteins. Upon exposure to ROS, these reactive thiols may undergo multiple reversible and irreversible oxidative alterations forming sulfenic (R-SOH), sulfinic (R-SO2H) and sulfonic (R-SO3H) acid residues. The formation of the sulfenic goup is reversible but this residue is an instable intermediate that quicly becomes further oxidized to the stable sulfinic or sulfonic groups or forms intra/intermolecular disulphide bonds (R-SS-R) or mixed disulfide bond with GSH (S-gluthationylation→ R-SSG) . The latter pathway dominates under physiological coditions in the cytoplasm because of the abundance of cytoplasmic reduced glutathione (GSH:GSSG~100:1 at millimolar concentrations). Formation of the more oxidized sulfinic and sulfonic acid residues is generally irreversible with the only known exception being the H2O2 scavenger peroxiredoxin that after being oxidized and inactivated by H2O2 (via sulfinylation) is slowly recovered by a novel ATP-dependent enzyme sulfiredoxin . On the other hand, depending on the overall redox tone, the reactive Cys residues of the S-gluthationylated proteins can be reduced back enzymatically by glutaredoxin or thioredoxin. Thus, proteine S-glutathionylation is a reversible oxidative alteration of the reactive Cys thiols triggered by ROS production and controlled by redox enzymes and small molecular weight thiols. As such, it can function as the premier effector pathway in ROS signaling [23, 115, 132]. Indeed, most of the reversible functional effects of ROS on enzymes, ion transporting proteins and other factors involved in intracellular signaling is associated with thiol oxidation [23, 132, 133].
ROS targets can be arbitrarily divided into two groups: ROS scavengers comprising of enzymes that catalyze ROS reduction to water, and ROS effectors that alter their function owing to interaction with ROS. Some ROS scavengers (e.g. GPx4) also interact with ROS effectors and regulate their thiol oxidation. GPx4 (an isoform present both in the cytosol and mitochondrial matrix) has been proven a vital molecular decoder of glutathione-dependent redox signaling as its deletion is embryonic lethal (killing mouse embryos at the same developmental stage as GSH deficiency) [134, 135]. Via metabolizing lipid hydroperoxides (e.g. cardiolipin hydroperoxide in the mitochondria) and counteracting with lypoxygenases as well as cyclooxygenase, GPx4 exerts anti-inflammatory and anti-apoptotic actions [116, 136]. Trx and Grx are also important modulators in the glutathione-dependent redox signaling as they regulate the duration and degree of reversible S-glutathionylation of target signaling molecules [115, 116].
Cell-wide molecular targets of ROS are reviewed elsewhere , here we focus on ROS effectors that participate in the local Ca2+ homeostasis between the ER/SR and mitochondria:
SERCA pumps contain numerous free Cys residues (24 in SERCA1, out of which 6 displayed sensitivity to oxidation by peroxynitrite ). The reactive thiol groups and disulfide bonds of SERCA affect the pump activity differently, according to their position, accessibility and reactivity . Cys674, a low-pKa reactive thiol at the hinge region of the protein  appears to play a key and dual role in the oxidative regulation of SERCA. Mild oxidative conditions lead to S-glutathionylation of Cys674, and increase pump activity, which might be a significant contributor to NO-dependent vasodilatation . Extensive or extended exposure to oxidants that lead to oxidation of other Cys residues and sulphonylation of Cys674 irreversibly inhibit the pump [139–141]. The distinct functional changes of the SERCA evoked by mild and harsh oxidative conditions illustrate that graded ROS elevations can serve both as a signaling event and a harmful stress condition acting through a single protein. In addition, the luminal regulation of SERCA2b also appears to involve a redox component: its C-terminal tail and 4th luminal loop forms a complex with the Ca2+-binding chaperone calreticulin and the ERp57 oxidoreductase, respectively, in a [Ca2+]ER and redox dependent manner. At high [Ca2+]ER and oxidizing conditions ERp57 promotes disulfide bridge formation between the L4 thiols that reduces pump activity, while when [Ca2+]ER decreases to <50 µM, ERp57 dissociates from L4 leaving the –SH groups reduced and increasing pump activity . This mechanism increases the energy-efficiency of SERCA2b and protects from Ca2+ depletion-induced ER stress. Thus, ROS production by adjacent mitochondria is likely to facilitate ER store Ca2+ refilling by SERCA pumps but massive ROS generation would engage Ca2+ depletion-induced ER stress that may synergize with other effectors of ROS to initiate apoptosis.
RyR/IP3Rs contain multiple reactive Cys thiols that influence channel gating or assembly. Moreover, many of the regulatory proteins that form complex with the RyR/IP3R also have reactive thiols that affect their interaction with RyR/IP3R [24, 33, 143]. RyR1: out of the three RyR isoforms, RyR1 have the most reactive thiols: 29 free –SH groups out of the 100 Cys moieties in one subunit . Thiol oxidation in general increases channel activity via enhancing intersubunit binding and preventing the binding of the negative regulator calmodulin (both apoCaM and CaCaM) to the receptor . The reactive Cys thiols of the RyR1 act as both cytosolic and luminal (regulating the interaction with triadin) redox sensors [146, 147]. Apparently, the different reactive groups have different sensitivity to oxidation, S-nitrosylation and S-glutathionylation, enabling the RyR1 to distinguish among different redox inputs . RyR2 carries 21 reactive thiols out of 89 Cys moieties per subunit . Similarly to RyR1, the net effect of thiol oxidation is increased/sensitized channel activity [29, 149]. Sensitization of RyR3 by ROS is relevant for the glutamate-induced mitochondrial ROS-dependent long-term potentiation in hippocampal neurons .
IP3R On IP3R1 (monomer), ~70% of the 60 Cys residues are kept in reduced state with variable accessibility and variable regulatory significance . Oxidation of certain –SH groups via exposure to thimerosal or GSSG sensitizes IP3R activation by IP3; however, oxidation of other thiol groups, which can be accessed by mersalyl but not the other thiol reagents, sensitize only the agonist binding but inhibit channel function [151, 152]. The sensitizing effect to the agonist has also been reproduced in bilayer systems using purified reconstituted receptors [153–155].
IP3-induced ER Ca2+ mobilization can be sensitized with the use of thiol-oxidizing agents (e.g. thimerosal, GSSG) and this sensitization alone is sufficient to generate spontaneous Ca2+ release events and [Ca2+]c oscillations [156, 157]. In addition to the pharmacological thiol reagents, endogenous ·O2− derived from NADPH oxidase activity has also been shown to amplify IP3R-mediated calcium signaling via sensitization of the IP3- induced Ca2+ release in aortic endothelial cells . Also, pharmacological evidence has suggested that mitochondrial ·O2− production is necessary to maintain IP3R-linked [Ca2+]c oscillations in pancreatic acinar cells . Exogenous ROS derived from macrophages augmented IP3R-mediated Ca2+ signaling and Ca2+-dependent apoptosis in pulmonary microvascular endothelial cells, which mechanism might be of relevance in development of atherosclerotic lesions . Similarly, externally added H2O2 enhanced IP3R-dependent GABA-ergic synaptic activity in spinal cord nociceptive interneurons . It has been also suggested that ROS might interfere with IP3 degradation since in vascular SMC the enhancement of IP3-induced Ca2+ release by ·O2− donors could not be observed when nonhydrolysable IP3-analogues were used .
RyR/IP3R-associated proteins: FKBP12.6 has been reported to stabilize the RyRs and synchronize their function . FKBP12 association with both RyR2 and RyR1 is sensitive to H2O2, involving a specific thiol group on the RyR-FKBP interaction site . ERp44, a thioredoxin family protein previously known to be involved in oxidative protein folding via reinforcing the Ero1α/oxidoreductase system has been shown to directly interact with the reduced Cys residues on the 3rd luminal loop (L3V) of IP3R1 in a Ca2+ and redox state dependent manner and inhibit the channel . This inhibitory effect has been proposed to support ER Ca2+ refilling and in turn, to protect protein folding when the redox balance in the ER shifts toward reducing conditions. Since silencing experiments revealed a basal inhibition of IP3R1 by ERp44 , one could expect that exposure to ROS and oxidizing the free –SH groups in the L3V luminal loop would be a way to enhance IP3R1 activity.
The ROS sensitivity of both RyR and IP3R indicates that feedback amplification of the Ca2+ release at the SR/ER mitochondrial interface might be induced by mitochondria-derived ROS that is induced by the calcium signal propagation to the mitochondria. This mechanism can be particularly important during pathological conditions when the PTP opens and the discharge of the ER Ca2+ pool further supports rapid execution of the cell .
Other targets at the SR/ER-mitochondrial interface: Cytoplasmic kinases and phosphatases are among the most established ROS-dependent regulators of the SR/ER and mitochondrial Ca2+ transporting proteins. Practically all the major cytosolic kinase signaling pathways are modulated by ROS  but here we discuss only the cAMP-dependent kinases (PKA) and the Ca2+/calmodulin-dependent kinases (CaMK) that can associate with the Ca2+ transporting proteins via anchoring proteins (AKAPs [165, 166]) or CaM. Moreover, SR/ER Ca2+ release channels can form macromolecular complexes with phosphatases PP1, PP2A [165, 167, 168]. Kinases and their counteracting phosphatases have different sensitivities to ROS modulation (mostly inhibition) and often the phosphatases are the more sensitive targets. Therefore, moderately oxidizing conditions inhibit only phosphatases to increase net phosphorylation, while massive exposure to ROS also inhibits kinases to suspend phosphorylation [169, 170]. For example, calcineurin (CaN), a Ca2+-activated protein phosphatase associates with RyR/IP3R via the FKBP12 proteins . Ca2+-activated CaN undergoes time-dependent inactivation that can be suspended by SOD1, suggesting a ·O2− -dependent mechanism . This mechanism has been implicated in the pathogenesis of amyotrophic lateral sclerosis . Thus, CaN can integrate Ca2+ and ROS signals and mediate both Ca2+-activated stabilization of RyR/IP3R through FKBP dephosphorylation and ROS-induced increase in channel activation through phosphorylation [23, 174]. Furthermore, H2O2 also promotes dissociation CaN from RyR/IP3R suspending the local interaction with the channels under oxidative stress [163, 171].
Mitochondrial targets: ROS (feedback) regulation of proteins/protein complexes that contribute to mitochondrial ROS production (αKGDH, aconitase, respiratory complex I) have been discussed earlier under ‘ROS sources’. It is important to note however, that while the direct ·O2− producer complex I increases its ROS generating potency upon ROS exposure (diverting the electrons to this direction from the ETC), the production of reducing equivalents by TCA decreases. Thus, exposure to excessive ROS will lead to progressive decay of ΔΨm and matrix acidification promoting PTP activation (see below).
While considerable data argues for the physiological redox regulation of the SR/ER Ca2+ transporting proteins, much less is known about ROS-regulation of mitochondrial Ca2+ transporters. VDAC has been shown to contribute to the ·O2− -induced increase in OMM permeability, which leads to the loss of cytochrome c from the IMS , however no evidence has been presented so far for the VDACs physiological redox regulation as a Ca2+ pore. A recent report showing that silencing of the cytosolic glutaredoxin-1 caused oxidative modification of –SH groups of VDAC gives a hint that probably VDAC is actively protected from oxidation and could be a subject of a local redox regulation . The Ca2+ uniporter and the Ca2+ exchangers have no known redox regulation.
The most extensively studied mitochondrial ROS target that conducts Ca2+ is the PTP. In highlights, the PTP opening is promoted by high [Ca2+]m, ROS, adenine nucleotide depletion, decreased ΔΨm, and elevated phosphate levels in the mitochondrial matrix [67, 177, 178]. Increased electron flow through complex I of the ETC associated with enhanced ·O2− generation promotes PTP via a quinone-mediated regulation [179, 180]. The molecular composition of the PTP is still uncertain (see [67, 177, 178]). Cyclophyilin D (CypD), a matrix peptidyl-prolyl cis-trans isomerase modulates the Ca2+ sensitivity of the PTP [181–183] and it is the target of the so far most specific pharmacological PTP inhibitor, cyclosporine A (CSA). However, PT could be evoked by quinones or ROS even after genetic ablation of CypD . The actual pore forming component(s) of the PTP are also elusive. The long-proposed adenine nucleotide translocator (ANT) that modulates the sensitivity of the PTP to Ca2+ and adenine nucleotides have been proven dispensible for CSA sensitive Ca2+-or ROS-dependent PTP activation based on genetic knockout studies . Most recently, the mitochondrial phosphate carrier (mPiC) has been emerging as a pore candidate (perhaps in conjunction with ANT), since it also binds CypD in a CSA-sensitive manner and displays sensitivity to quinones; however this role of mPiC requires further establishment [178, 185, 186].
The PTP has at least two different redox sensing sites; one that reacts to the glutathione oxidation level (decreased matrix GSH/GSSG ratio) and another that senses the oxidation level of matrix pyridine NAD(P) [187–190].
Importantly, opening of the PTP is reversible ; however, depending on the nature/strength of the activating stimuli it may become fixed in the open conformation ultimately disabling, osmotically destroying the mitochondrion and deliberating mitochondrial pro-apoptotic factors to the cytosol [177, 178]. Reversible (transient) openings of PTP have been suggested to be central players in the ROS-induced ROS release mechanisms underlying the regenerative ROS waves observed in cardiac muscle . Transient opening of the PTP does not allow escape of significant amount of pyridine nucleotides but causes a temporary uncoupling, giving a boost to ROS production by the ETC and giving way out to the superoxide from the matrix. In a similar but rather progressive manner, enhanced NOX-derived ROS-and Ca2+-induced mitochondrial ROS production has been shown to be a central player in the neuronal degeneration associated with the deficiency of PINK1, a protein regulating mitochondrial Ca2+ extrusion and mutated in familiar Parkinson’s disease . A striking ROS effect on mitochondrial fragmentation indicates that mitochondria shaping proteins or their regulators also represent a ROS target [193, 194].
Sources as well as targets of Ca2+ and ROS are brought to close distances at focal contacts of the organelles, making Ca2+ and ROS effective local messengers. However, testing the local crosstalk between these signaling systems is technically challenging, hence to date only a few studies have provided direct evidence in this direction. In astrocytes, ROS-induced transient focal mitochondrial depolarizations (flickers) reflecting transient PTP openings have been shown to depend on Ca2+ stored in the ER. The persistence of these depolarizations at increased [Ca2+]c buffering suggested that local Ca2+ delivery from the ER to the mitochondria might have caused them and that this local Ca2+ transfer became sensitized by the ROS derived from the mitochondria. Although the observed phenomenon was based on an artificial mitochondrial ROS source (the membrane potential probe TMRE) it depicted a scenario where mitochondria-derived ROS could induce local ER Ca2+ release events that in turn, ‘hit back’ on the mitochondria to promote PTP activation . The same group has also reported increased spark frequency in cardiomyocytes at regions of mitochondrial ROS generation (TMRE illumination), which was consistent with local redox sensitization of RyR2 . A very recent study has given detailed account on the effect of different ROS on the spatiotemporal parameters of sparks. Using pharmacological inhibitors (respectively TMPyP and myxothiazol), it also provided evidence that basal cytosolic and mitochondrial ROS production are determinants of basal spark parameters in cardiomyocytes .
Compelling evidence supports a mutual relationship between calcium and ROS signaling, which affects Ca2+ and ROS metabolism and the activation of their respective effectors. SR/ER is the main intracellular source of Ca2+ and mitochondria are important modulators of the Ca2+ dynamics. On the other hand, mitochondria and in a lesser extent, SR/ER represents a source and breakdown site of ROS. The present description of the structural and functional coupling between mitochondria and SR/ER illustrates that the SR/ER mitochondrial junction might serve as the center stage for Ca2+-ROS interactions. However, the testing of this idea will require new approaches to target and visualize the components of the local interorganellar coupling. This task is further complicated by the permanent reorganization of SR/ER and mitochondria, which likely involves several Ca2+ and ROS sensitive factors. Further consideration and studies of the local involvement of RNS will be also required. Currently available data have provided evidence that the Ca2+-ROS interplay is relevant for both physiological homeostasis and for a range of pathological conditions when either the Ca2+ or ROS regulations are disturbed. In several grave human disorders from acute ischemia-reperfusion conditions of the heart and brain to the slowly progressing neurodegenerative diseases, the local interaction of Ca2+ and ROS at the mitochondria is gaining notice. These observations give reasons for studying the derangement of the Ca2+ and ROS signaling and structure of the SR/ER-mitochondrial junction as a contributor in the development of a range of diseases.
We would like to thank Drs. Suresh K. Joseph and Pál Pacher for helpful discussions.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.