Mitochondrial Hyperfusion during Oxidative Stress Is Coupled to a Dysregulation in Calcium Handling within a C2C12 Cell Model

Atrial Fibrillation is the most common sustained cardiac arrhythmia worldwide harming millions of people every year. Atrial Fibrillation (AF) abruptly induces rapid conduction between atrial myocytes which is associated with oxidative stress and abnormal calcium handling. Unfortunately this new equilibrium promotes perpetuation of the arrhythmia. Recently, in addition to being the major source of oxidative stress within cells, mitochondria have been observed to fuse, forming mitochondrial networks and attach to intracellular calcium stores in response to cellular stress. We sought to identify a potential role for rapid stimulation, oxidative stress and mitochondrial hyperfusion in acute changes to myocyte calcium handling. In addition we hoped to link altered calcium handling to increased sarcoplasmic reticulum (SR)-mitochondrial contacts, the so-called mitochondrial associated membrane (MAM). We selected the C2C12 murine myotube model as it has previously been successfully used to investigate mitochondrial dynamics and has a myofibrillar system similar to atrial myocytes. We observed that rapid stimulation of C2C12 cells resulted in mitochondrial hyperfusion and increased mitochondrial colocalisation with calcium stores. Inhibition of mitochondrial fission by transfection of mutant DRP1K38E resulted in similar effects on mitochondrial fusion, SR colocalisation and altered calcium handling. Interestingly the effects of ‘forced fusion’ were reversed by co-incubation with the reducing agent N-Acetyl cysteine (NAC). Subsequently we demonstrated that oxidative stress resulted in similar reversible increases in mitochondrial fusion, SR-colocalisation and altered calcium handling. Finally, we believe we have identified that myocyte calcium handling is reliant on baseline levels of reactive oxygen species as co-incubation with NAC both reversed and retarded myocyte response to caffeine induced calcium release and re-uptake. Based on these results we conclude that the coordinate regulation of mitochondrial fusion and MAM contacts may form a point source for stress-induced arrhythmogenesis. We believe that the MAM merits further investigation as a therapeutic target in AF-induced remodelling.


Introduction
Atrial Fibrillation (AF) is the most common sustained cardiac arrhythmia and worldwide, as populations age, the prevalence of AF will continue to increase [1,2]. A progressive disorder, AF immediately impairs cardiac performance increasing risks of stroke, dementia, heart failure and death [3,4]. Current treatment for AF is sub-optimal and will remain so until fundamental mechanisms underpinning AF development and maintenance are better understood [5]. The recent success of catheter ablation in paroxysmal, but not persistent, AF has refocused attention on the importance of atrial tachycardia remodelling (ATR) [5,6].
Initially paroxysmal, AF provokes functional and structural changes which favour arrhythmia maintenance [6,7]. The high frequency electrical activity in the fibrillating human atrium in vivo is associated in vitro with mitochondrial dysfunction, oxidative stress and calcium overload [6,[8][9][10][11]. If fibrillation persists, surviving atrial myocytes adapt to preserve calcium homoeostasis, a process termed ATR [6,12]. However, this new equilibrium alters sarcolemmal ion currents and compromises sarcoplasmic reticulum (SR) function promoting triggered activity, re-entry and ultimately perpetuation of AF [6]. Despite the experimental evidence implicating impaired calcium handling in these maladaptive processes, the mechanistic links between mitochondria, oxidative stress and AF-induced remodelling remain largely unexplained [6].
Previous investigations of AF have consistently noted changes in mitochondrial morphology and distribution within atrial myocytes [10,13,14]. Confirmation that mitochondria form attachments to SR, the mitochondria associated membrane (MAM), thereby creating calcium microdomains facilitating the rapid uptake and storage of [Ca 2+ ] c has implications for ATR which are, as yet, unknown [15,16]. Mitochondria tether to SR via a protein, mitofusin-2 (Mfn-2), which in addition to regulating mitochondrial fusion and [Ca 2+ ] c uptake, maintains mitochondrial respiratory homoeostasis and mediates the SR stress response [17][18][19][20]. Mitochondria themselves are dynamic organelles responding to short term cellular stress with decreased fission and/or increased fusion, however if stress persists, as occurs during AF, this adaptive response is superceded by mitochondrial fragmentation triggering mitophagy or cellular apoptosis [20,21].
We hypothesised that the stress of rapid stimulation would increase mitochondrial fusion and promote the formation of MAM in vitro. In addition we postulated that altered mitochondrial plasticity with or without oxidative stress would reversibly alter SR calcium handling and result in increased formation of MAM. In order to test our hypothesis we selected the C2C12 murine myotube model as (i) it has previously been used to investigate mitochondrial plasticity and the acute effects of oxidative stress in vitro, (ii) it is capable of responding to electric field stimulation (EFS) and (iii) it exhibits a myofibrillar system broadly similar to atrial myocytes [37,38].

C2C12 mouse myoblast differentiation
Differentiated C2C12 myotubes ( Figure 1) were obtained as recently described by Kuo et al [39]. Briefly, C2C12 myoblasts were cultured at 37uC in an atmosphere of 10% CO 2 in 'complete' growth medium (DMEM supplemented with 20% FBS, penicillinstreptomycin and L-glutamine).Once cells reached 70-80% confluence, differentiation was induced by replacing complete medium with low-serum medium (DMEM supplemented with 2% horse serum, penicillin-streptomycin and L-glutamine) and cultured for 24-96 h at 37uC in an atmosphere of 10% CO 2 . Differentiated myotubes, grown onto coverslips (12 mm diameter, 0.16-0.19 mm thickness), were mounted in media with 10 mM Hepes (pH 7.4), 2.5 mg/ml Hoechst, 0.1 mM MitoFluorRed 633, and 5 mM dihydroethidium (hET) 24-96 h post differentiation. Upon oxidation hET (a free radical sensor dye) is cleaved, and the resulting ethidium intercalates with Mitochondrial DNA and, upon high rates of conversion, the dye also fills the cytosol and nucleus as it binds chromosomal DNA. Live images were captured at 405 nm (to visualize DNA, blue), 515nm (to visualize oxidized hET, green) and 633 nm (to visualize MitoFluorRed 633, red) with a 100X NA 1.4 oil immersion objective (Olympus) at 1 airy U on a laser-scanning confocal microscope (IX80; Olympus) operated by FV1000 software version 1.4a (Olympus).

Electrical Field stimulation (EFS) as a model of ''fibrillatory stress''
Monolayers of cultured C2C12 myocytes were placed on gelatin/fibronectin coated coverslips suspended in serum free normal Tyrode solution in a custom designed chamber incubated for 24 hours with 5% CO 2 at 37uC as previously described [37,38]. Parallel carbon electrodes connected via platinum wire to a GRASS electrical stimulator continuously applied monophasic square-waves of pulse duration 2 ms, field gradients ,8 V/cm without harmful chemical reaction [40]. In order to investigate the effects of ''fibrillatory stress'', stimulation protocols include control conditions (0 Hz), simulated physiological conditions (1 Hz) and fibrillatory stress (5 Hz), adapted from Tandon et al [40]. Three days post differentiation, nascent C2C12 myotubes were subjected to stimulation protocols for 24 h at 37uC in an atmosphere of 10% CO 2 . To study SR colocalization, differentiated myotubes were incubated (overnight, 4uC) with mouse anti-PDI antibody (ER marker, 1:100 dilution) and rabbit anti-Tom20 antibody (mitochondria marker, 1:1000 dilution), followed by 3 washes with blocking solution. Differentiated myotubes were next incubated (1:1000 dilution, 45 min, room temperature) with goat anti-mouse (514 nm) or goat anti-rabbit (647 nm) conjugated Alexa Fluor secondary antibodies, followed by 3 washes with PBS. Myotubes were visualized using the confocal microscope as above.

Studies of mitochondrial plasticity, SR colocalization and calcium handling
Two-days post differentiation, nascent C2C12 myotubes were infected (500 MOI) with dominant-negative dynamin-related protein 1 (DRP1K38E), a mutant form of the mitochondrial fission factor that promotes an elongated mitochondrial reticulum. Empty adenoviral vectors served as control. In order to observe mitochondrial plasticity, differentiated myotubes were loaded with 100 nM tetramethyl-rhodamine ethyl ester (TMRE). Images were captured at 440 nm (to visualize DRP1K38E-CFP expression) and 543 nm (to visualize TMRE, red) using the confocal microscope as above.
To investigate effects of altered mitochondrial plasticity on calcium handling, differentiated myotubes were washed with C2C12 myoblasts were cultured and induced to differentiate for 24-96 h at 37uC in an atmosphere of 10% CO 2 . Panel A: C2C12 myoblasts differentiate into myotubes. Cells were plated onto coverslips and allowed to reach 70-80% confluence before inducing differentiation. Live images were captured at 405 nm (to visualize DNA, blue), 515 nm (to visualize oxidized hET, green) and 633 nm (to visualize mitochondira, red) on a laserscanning confocal microscope. Panel B: Immunoblots confirming upregulated expression of mitochondrial fission protein DRP1 and contractile protein myosin heavy chain during differentiation in C2C12 myotubes. Cells were harvested, and 50 mg of protein (pooled from 5 separate experiments) were processed for gel electrophoresis. Primary antibodies used were against DRP1, MF-20, and HSP70. Panel C: Translocation of DRP1 to the mitochondrial membrane in C2C12 myotubes. Differentiated C2C12 myotubes were fixed, and incubated with primary antibodies against MF20, DRP1 and/or Tom20 and then corresponding Alexa Fluor secondary antibodies. Confocal microscopy images were captured as described before. doi:10.1371/journal.pone.0069165.g001 modified Hanks' buffered saline solution, loaded with 2.5 mM Fluo-4 and 100 nM TMRE, incubated at 37uC in an atmosphere of 10% CO 2 for 30 min, and then washed with warmed buffer. The methylxanthine caffeine is recognised at millimolar concentration to release Ca 2+ from intracellular stores during diastole, primarily in a ryanodine receptor (RyR) mediated process involving the SR in skeletal muscle preparations [41,42]. We used caffeine (4 mM) added to the superfusate to elicit Ca 2+ waves in C2C12 myotubes during laser scanning confocal imaging to detect release of Ca 2+ from intracellular stores.

Studies of oxidative stress and altered redox state
Differentiated C2C12 myotubes were treated with the glutathione synthase inhibitor buthionine sulphoximine (BSO, 200 mM, 24 h) or with the thiol-oxidizing agent diamide (100 mM, 2 h). To assess the effect of oxidative stress on mitochondrial plasticity, differentiated myotubes were loaded with 100 nM TMRE, and images were captured using the confocal microscope as above. In a parallel set of experiments, differentiated myotubes were loaded with 100 nM TMRE and 2.5 mM Fluo-4, and calcium release events were triggered by 4 mM caffeine and detected by laser scanning confocal imaging as above. In order to determine the degree to which oxidative stress was reversible, and to establish if a baseline level of oxidation exists, these protocols were repeated using C2C12 myotubes cultured with the glutathione precursor N-Acetyl Cysteine (NAC 200 mM, 24 h), both pre and post exposure to BSO or DRP1K38E expression.

Electron Microscopy Studies
C2C12 myoblasts were grown and induced to differentiate on 22 mm glass coverslips. During differentiation nascent C2C12 myotubes were either infected with DRP1K38E (500 MOI) exposed to oxidative stress or EFS-induced ''fibrillatory stress'', as previously described, for the final 24 h period prior to processing. Differentiated myotubes were fixed in 2% glutaraldehyde in 0.1M PBS, pH 7.4, post-fixed in 1% osmium tetroxide in PBS, en bloc stained in 3% uranyl acetate, dehydrated in an ascending series of ethanol, and then processed and embedded in Spurr epoxy resin. Thin sections were cut on a Leica UC 6 ultramicrotome, collected on copper grids and counter-stained with lead citrate. Samples were viewed and images taken with a Jeol 1230 TEM equipped with AMT software.

Statistical Analysis
Results are presented as mean6SE of the mean with the exception of the electron microscopy studies. An unpaired, twotailed Student's t-test was used to evaluate the significance of difference between the calculated means of two groups; multiple comparisons were assessed using one way analysis of variance (ANOVA) via the Prism statistical software package (Graphpad Inc, La Jolla, USA).Some confocal and electron microscopy studies were assessed using one way analysis of variance (ANOVA) or Fisher's exact test where indicated, via Prism as before. A probability value of ,0.05 was considered significant.

Rapid stimulation results in increased mitochondrial fusion and SR colocalisation
Exposing differentiated C2C12 cells to sudden and persistent rapid stimulation for 24 h resulted in the formation of elongated mitochondria which colocalised with SR and was associated with an increased frequency of SR:mitochondrial contact sites ( Figure 2). There was no significant difference between control cells and those stimulated at 1 Hz in the parameters assessed. However, mitochondrial 2D size increased approximately 5-fold in cells which were stimulated at 5 Hz ( Figure 2B)   As compared to control myotubes which had been transfected with empty adenoviral vector, cells expressing DRP1K38E released threefold more calcium upon application of 4 mM of caffeine (Amplitude of caffeine induced calcium release in control myotubes 5646181 fluorescence arbitrary units (fau) v. 22216397 fau in DRP1K38E myotubes, p,0.001 Student's t-test, n = 40 myotubes. Figure 6B, C). There was a significant delay in caffeine response time ( Figure 6E i) but not in time to peak caffeine effect ( Figure 6E ii) in DRP1K38E myotubes as compared to control (caffeine response time control myotubes 863 s v. 1462 s in DRP1K38E expressing myotubes, p,0.001; Time to peak caffeine effect control myotubes 862 s v. 1062 s in DRP1K38E expressing myotubes, p = ns, n = 40 myotubes for all comparisons). However, the first derivative of the caffeine response (DF/DT, Figure 6E iii) in DRP1K38E myotubes was significantly greater than that observed in control myotubes (caffeine response control myotubes 70611 fau/s v. 245640 fau/s in DRP1K38E myotubes, n = 40 myotubes p,0.001). In addition to altered calcium release characteristics, subsequent calcium clearance from the cytosol was also significantly prolonged in myotubes expressing DRP1K38E. Early (time to 25% clearance, t 25 ), mid (time to 50% clearance, t 50 ), and late (time to 90% clearance, t 90 ) responses ( Figure 6D) were significantly delayed in myotubes with a hyperfused mitochondrial reticulum (Early Ca 2+ clearance time (t 25 Figure 5C) and increased SR proteins present in mitochondrial pellets isolated from C2C12 cells following transfection with DRP1K38E ( Figure 6F).

Oxidative stress promotes mitochondrial fusion and results in altered SR calcium handling
It has previously been shown that increased cellular levels of oxidized glutathione, which is common during cellular stress, can lead to the activation of the mitochondrial fusion machinery [43]. Consistent with this, inhibition of GSH synthesis with 200 mM of BSO and/or induction of GSH oxidation via diamide (100 mM) altered mitochondrial plasticity promoting a fused reticulum in differentiated C2C12 myotubes (Figures 3, 5 and 7A) [44][45][46][47]. Oxidative stress was associated with an almost fourfold increase in fluorescence following caffeine application (amplitude of caffeine induced calcium release in control myotubes 534684 fau v. 19966269 fau in myotubes incubated with BSO and 19436438 fau in myotubes treated with diamide, p,0.001, ANOVA, n = 40 myotubes Figure 7B,C). Similarly there was a pronounced increase in the caffeine response time ( Figure 7D i) Figure 7E).

ROS inhibition alters both mitochondrial plasticity and SR calcium handling
The physiological role of low levels of intracellular ROS as signalling molecules is well recognised [22]. In order to compare oxidative stress with reducing stress we devised protocols whereby myotubes were first incubated with oxidizing agents then with reducing agents in addition and vice versa. In this way we aimed not only to determine whether the observed effects of oxidative stress on calcium handling and mitochondrial plasticity could be reversed, but also whether reducing agents themselves had any reversible effects on mitochondrial plasticity and calcium handling.
Inhibition of GSH synthesis with 200 mM of BSO promoted mitochondrial fusion as before (Figures 3, 5, 7A & 8A). However, as compared to myotubes similarly cultured but with the addition of the glutathione precursor NAC (200 mM) for the final 24 h period of incubation, a less fused reticulum was observed, suggesting reversibility ( Figure 8A). In contrast, myotubes cultured solely in the presence of NAC appeared to display a more disconnected reticulum with possibly greater numbers of fragmented mitochondria, suggestive of reduced fusion or increased fission or both. Subsequent to the addition of BSO to culture media for 24 h the mitochondria appeared to return to a more fused network ( Figure 8A). Inhibition of glutathione synthase by BSO was again associated with an almost fourfold increase in fluorescence which was reversed by the addition of NAC (amplitude of caffeine induced calcium release in control myotubes 5136164 fau v. 18986258 fau in myotubes incubated with BSO (p,0.001) and 8656300 fau in myotubes treated with BSO+ NAC (p = ns), ANOVA, n = 40 myotubes Figure 8B, C). In contrast, the addition of NAC to myotube media resulted in an approximate three to fivefold potentiation of the delay in caffeine response ( Figure 8C ii) and the time to peak ( Figure 8C   Addition of the glutathione precursor NAC to myotubes media during culture was associated with an approximate 80% increase in fluorescence which was further doubled by the addition of BSO to culture media (amplitude of caffeine induced calcium release in control myotubes 4986152 fau v. 8946285 fau in myotubes incubated with NAC (p,0.01) and 19026387 fau in myotubes treated with NAC+ BSO (p,0.001), ANOVA, n = 60 myotubes, Figure 8E i). Again, the addition of NAC to media resulted in an approximate fivefold potentiation of the delay in caffeine response ( Figure 8E ii) and a sevenfold increase in the time to peak ( Figure 8E iii) caffeine response (caffeine response time control myotubes 1063 seconds (s) v. 49610 s in NAC myotubes (p,0.001) and to 6768 s in NAC+ BSO treated myotubes (p,0.001); Time to peak caffeine effect control myotubes 662 s v. 44612 s in NAC myotubes (p,0.001) and 4367 s in NAC+ BSO treated myotubes (p,0.001), n = 40 myotubes per group). There was no significant effect of NAC on DF/DT of the caffeine response ( Figure 8E iv) and the expected BSO mediated increase in DF/DT was not observed in the NAC pre-treated myotubes (caffeine response control myotubes 63610 fau/s v. 43628 fau/s in NAC myotubes and 44620 fau/s in NAC+ BSO myotubes (p = ns for all comparisons), n = 40 myotubes per group). Addition of NAC to culture media resulted in a significant delay to clearance of calcium from the cytosol (Figure 8F ii), and it was not possible to observe complete recovery in myotubes exposed to Altering mitochondrial plasticity, fibrillatory and oxidative stress similarly result in enlarged mitochondria with increased frequency of SR: mitochondrial contacts Transfection of dominant-negative mutant DRP1K38E resulted in a more fused mitochondrial reticulum composed of fewer, larger mitochondria as compared to control (Figures 3, 4, 5, 6 and 9). Inhibiting mitochondrial fission in this way resulted in a greater than fourfold increase in mitochondrial 2-D cross-sectional area with a significant increase in observed SR: mitochondrial contact points ( Figure 5). Similarly, oxidative stress resulted in a significant enlargement of mitochondria, associated with a more fused reticulum (Figures 3, 7, 8, 9) and promoted formation of a greater number of visible SR: mitochondrial contacts (Figure5C). Interestingly, the addition of the reducing agent NAC had no effect on mitochondrial fusion under control conditions but reversed not only the fusion induced by BSO but also infection with DRPK138E (Figures 3, 5, 8, and 9). In addition, NAC reversed the effect of 'forced fusion' on caffeine induced calcium release (Figure 9).
Addition of the glutathione precursor NAC to myotubes previously infected with DRP1K38E was associated with a reduction in fluorescence to control levels (amplitude of caffeine induced calcium release in control myotubes 5646181 fau v. 22216397 fau in myotubes incubated with DRP1K38E (p,0.001) and 7956350 fau in myotubes treated with DRP1K38E+ NAC (p = ns), ANOVA n = 40 myotubes Figure 9B, C). Again, the addition of NAC to media resulted in an approximate fourfold  Figure 9D iii) of the caffeine response (caffeine response control myotubes 70611 fau/s v. 245640 fau/s in DRP1K38E myotubes and 35620 fau/s in DRP1K38E+ NAC myotubes (p,0.001 for all comparisons), n = 40 myotubes per group). Addition of NAC to culture media resulted in a significant delay to clearance of calcium from the cytosol ( Figure 9E). Indeed, it was not possible to observe complete recovery in myotubes exposed to DRP1K38E+ NAC (t 25

Discussion
With these studies we provide further evidence of the intimate relationship between mitochondrial fusion, oxidative stress, and altered SR calcium handling in murine myocytes. Increased MAM formation, via inhibition of mitochondrial fission or triggered by oxidative stress, was associated with increased caffeine induced calcium release with preservation of release kinetics and delayed cytosolic clearance of calcium. Further corroboration of the redox sensitive nature of intracellular calcium transporters is provided by the slowing of SR calcium release and the potentiation of delayed calcium clearance in the presence of the glutathione precursor NAC.
Mitochondria are increasingly recognised to perform a number of vital functions in mammalian cells. In addition to their long appreciated role in cellular respiration, the discovery of MAM has linked mitochondrial plasticity, ECM coupling and cellular stress responses. The central role of rapid bidirectional calcium signalling between SR and mitochondria at the MAM offers tremendous opportunities to improve our understanding of various cardiomyocyte pathophysiologies. We sought to investigate the relationship between fibrillatory and oxidative stress, mitochondrial fusion and calcium handling with simple experiments in order to identify a possible link between AF-induced cellular stress and subsequent atrial tachycardia remodelling.
As mitochondria cannot be generated de novo, they are perpetually being repaired and recycled in a dynamic equilibrium between opposing processes of fission and fusion [20]. This plasticity is required in order to preserve both mitochondrial and myocyte integrity [18]. Fusion produces elongated interconnected mitochondria and formation of mitochondrial networks facilitates the transmission of calcium signals and membrane potential across individual cells [20]. Mitochondrial fission generates numerous morphologically and functionally distinct isolated mitochondria and can be physiological, as a prelude to mitophagy, or pathological in response to rapid increases in [Ca 2+ ] [20,48]. Dysfunctional mitochondrial plasticity results in increased sensitivity to apoptotic stimuli and is pro-arrhythmogenic, however the importance of mitochondrial plasticity in human heart disease is only just being recognised [18,49,50].
The Mitochondria-Associated Membrane is the physical association of juxtaposed SR and mitochondrial membranes which facilitates privileged inter-organelle communication performing several physiological functions [15,[51][52][53]. Both interorganelle proximity and formation of MAM tethers are dynamic, remodelling in response to local [Ca 2+ ] c [53][54][55]. The first direct MAM tether identified in the mammalian system was Mitofusin-2 (Mfn-2), a large trans-membrane protein residing in the outer mitochondrial membrane (OMM) pivotal to mitochondrial fusion [17]. As SR structural integrity and bidirectional calcium signalling at the MAM are Mfn-2 dependent, the MAM may also perform critical roles in calcium buffering and the regulation of cellular respiration during stress [17,[56][57][58].
Mitochondria generate ATP by oxidative phosphorylation with calcium and ROS acting as signalling molecules in a bioenergetic homoeostasis involving mitochondria, SR and the nucleus [22,24]. Mitochondrial respiration is also the primary source of ROS within myocytes which, under normal conditions, remain counterbalanced by calcium-dependent production of reducing agents [22,59]. Accumulation of ROS impairs myofibrillar calcium handling and has been implicated in contractile dysfunction and AF-remodelling [30,36]. Altered phosphorylation of sarcolemmal ion currents modulates whole cell calcium entry and extrusion, while oxidation of the SR calcium transporters RyR and SERCA modify SR calcium release and re-uptake [26,27,31]. Experimentally induced mitochondrial ROS production induces proarrhythmic SR calcium release [60]. If metabolic stress is not relieved, oxidative stress uncouples mitochondria, disrupts cytosolic proteins and nuclear DNA resulting in cellular dysfunction and ultimately apoptosis [22,25,26,61].
Confirmation of the existence of microdomains of Ca 2+ and ROS at the MAM suggest the possibility of crosstalk between respiration, oxidative stress, mitochondrial plasticity and calcium handling [15,53,60,62,63]. Recent descriptions of stress induced mitochondrial hyperfusion (SIMH) provide further credence to the possibility of such crosstalk [21]. Mitochondria initially respond to a variety of cellular stresses by forming a fused network in an apparent protective effect [20]. Mitochondrial function is maintained temporarily under adverse circumstances; however if stress persists mitochondrial fragmentation and ultimately cell death occur [20,21]. In related work, ROS induced disulphide switching of Mfn-2 promoted mitochondrial fusion in a GTP dependent process directly linking oxidative stress and mitochondrial fusion [43]. Recent data suggests that stress induced phosphorylation of drp1 by PKA inactivates fission, promoting formation of a fused mitochondrial network [20,64]. This hyperfused reticulum appears resistant to mitophagy, conferring temporary protection to cells from death triggers, 'buying time' for other homoeostatic mechanisms to alleviate stress and thus preserve cellular integrity [20]. It is not known which combination(s) of factors promote switching from SIMH to fragmentation nor the circumstances which determine the threshold for death triggers, although one should not be surprised to learn that emerging data supports a regulatory role for Mfn-2 mediated bidirectional calcium signalling in SR stress responses [17,19].
Our results support a role for hyperfusion in response to stress and broadly speaking appear in agreement with those of the Dorn laboratory [18,65]. Similarly, altered calcium handling by the SR and/or mitochondria associated with increased MAM formation, either as a result of oxidative stress or unopposed fusion, suggests that Mfn-2 is critical to crosstalk. Caffeine induced SR calcium release into mitochondria has previously been reported and is related to proximity of the MAM and the rate of SR emptying [15,66,67]. Opposing sides of the MAM preferentially locate with anchoring of outer and inner mitochondrial membrane permitting privileged access to the respiratory complexes [68]. Prevention of mitochondrial calcium overload has previously been shown to rescue in vitro models of pathological tachycardia and cardiomyopathy [69][70][71][72]. Mitochondrial calcium overload eventually necessitates calcium release back into the cytosol [73][74][75]. Currently it is not known to what degree this mitochondrial calcium efflux may contribute to SR refilling or diastolic triggering of CICR, particularly if mitochondria have fused into networks [76][77][78][79][80][81].
Redox potential would appear to have a hermetic response curve suggesting various affinities for stimulatory and inhibitory regulatory pathways ultimately with ROS becoming toxic at high concentration. As the outcome of an intracellular calcium signal ordinarily depends on the strength, localisation, duration and pattern of that signal, SR Ca 2+ release channeled to a tethered mitochondrion could act as a point source, propagating calcium signals via the mitochondrial network throughout the cell, obviating ''local control'' [57,[82][83][84][85]. Hence we postulate that Mfn-2 performs multiple, yet related, roles: tethering mitochondria to the SR, creating and maintaining calcium signalling at the MAM, modulating mitochondrial plasticity in response to local [Ca 2+ ] c , transducing SIHM if stress can be ameliorated, and if not, initiating the death trigger [86][87][88][89][90][91]. This possibility raises the question ''Could the MAM be a point source for arrhythmogenesis?''

Limitations/Suggested further work
The C2C12 myotube model is derived from murine skeletal myoblasts and has the advantage of being capable of differentiation thus affords the opportunity to observe induced changes in mitochondrial plasticity while developing myofibrillar apparatus. This advantage of the C2C12 model readily permits reproducible interruption of ongoing fission/fusion processes while simultaneously offering the ability to investigate both fibrillatory and oxidative stress. In addition, the study of isolated myocytes permits the observation of autonomous intracellular processes, indicating that ATR can, at least in part, occur in the absence of cellular communication and does not require extracellular factors. However this inherent malleability is in direct contrast to that of terminally-differentiated human atrial myocytes. Unfortunately only relatively simple experiments can be designed examining scarce human atrial tissue from consenting patients.
We believe a more practical solution is to seek to rapidly identify potential candidate factors and regulatory pathways for investigation in the C2C12 myotube model, design and develop techniques for investigating these processes in differentiated atrial myocytes such as HL-1 myocytes and then perform simple corroborative experiments with human tissue.

Conclusions
Long overlooked in cardiovascular pathology, the dynamic equilibrium of mitochondrial plasticity remains in equipoise under basal conditions. Cellular stress, whether as a result of rapid stimulation or from ROS promotes a hyperfused mitochondrial network, increased formation of mitochondria associated membrane and Mfn-2 mediated alterations in SR calcium handling with, as yet unknown, implications for arrhythmogenesis. Further work is necessary to determine the intermediate factors and ultimately the relevance of oxidative stress, mitochondrial plasticity and altered calcium handling in ATR.

Acknowledgments
Dr. Peter Backx for critical appraisal of the manuscript. Mr. Peter Rippstein for technical assistance with the electron microscopy studies.