CaMKII induces permeability transition through Drp1 phosphorylation during chronic β-AR stimulation

Mitochondrial permeability transition pore (mPTP) is involved in cardiac dysfunction during chronic β-adrenergic receptor (β-AR) stimulation. The mechanism by which chronic β-AR stimulation leads to mPTP openings is elusive. Here, we show that chronic administration of isoproterenol (ISO) persistently increases the frequency of mPTP openings followed by mitochondrial damage and cardiac dysfunction. Mechanistically, this effect is mediated by phosphorylation of mitochondrial fission protein, dynamin-related protein 1 (Drp1), by Ca2+/calmodulin-dependent kinase II (CaMKII) at a serine 616 (S616) site. Mutating this phosphorylation site or inhibiting Drp1 activity blocks CaMKII- or ISO-induced mPTP opening and myocyte death in vitro and rescues heart hypertrophy in vivo. In human failing hearts, Drp1 phosphorylation at S616 is increased. These results uncover a pathway downstream of chronic β-AR stimulation that links CaMKII, Drp1 and mPTP to bridge cytosolic stress signal with mitochondrial dysfunction in the heart.

A ctivation of b-adrenergic receptors (b-AR) plays an essential role in stimulating cardiac output during fight-or-flight response 1 and a deleterious role in heart failure 2 . Acute stimulation of b-AR augments cardiac contraction by increasing intracellular Ca 2 þ handling 3 , whereas chronic b-AR stimulation exhibits detrimental outcomes, such as myocardial hypertrophy and heart failure, mainly through activating Ca 2 þ /calmodulin-dependent kinase II (CaMKII) 4,5 . Openings of mitochondrial permeability transition pore (mPTP) were involved in cardiotoxicity induced by chronic b-AR stimulation and preventing mPTP openings could attenuate mitochondrial stress and mitochondrial-dependent apoptosis 6 . Consistently, mitochondrial-targeted CaMKII inhibition efficiently prevented isoproterenol (ISO)-induced myocardial injury 7 . It is currently not known whether CaMKII plays a causal role in mPTP openings during chronic b-AR stimulation. Uncovering the molecular mechanism or signalling pathway that mediates chronic b-AR stimulation-induced mPTP openings bears tremendous clinical significance, due to the fact that high catecholamine levels and chronic b-AR activation are hallmarks of human heart failure.
As a crucial gatekeeper, transient or subconductance openings of mPTP occur under physiological conditions and critically maintain mitochondrial homeostasis 8,9 . On the other hand, massive or prolonged mPTP openings, which can be triggered by Ca 2 þ overload, oxidative stress and mitochondrial fission/fusion, uncouple oxidative phosphorylation, induce mitochondrial swelling, promote cytochrome c release, and contribute to the pathology of heart disease such as ischaemia reperfusion injury 10,11 . Tracking mPTP activity in intact cells or in vivo poses a major technical challenge. Recently, we and others have developed the optical recording of short-term and reversible mPTP events by imaging single mitochondrial flashes 12 . Despite debates over the nature of flashes 13 , consensus has been reached that these events are triggered by mPTP and fuelled by mitochondrial electron transport chain in the heart 12,[14][15][16] . This breakthrough discovery provides a tool for evaluating mPTP in the healthy and diseased heart.
By utilizing this state-of-the-art technology, we found that chronic b-AR stimulation induces mPTP through activating CaMKII and subsequent phosphorylation of dynamin-related protein 1 (Drp1) at S616. Inhibiting CaMKII activity, Drp1 phosphorylation or Drp1 activity suppressed mPTP openings, ameliorated mitochondrial dysfunction and myocyte death in vitro, and rescued heart dysfunction in vivo. Taken together, we identify a new mechanism by which chronic b-AR stimulation induces mitochondrial dysfunction through post-translational modification of a fission protein in the heart.

Results
Chronic b-AR stimulation augmented mPTP opening. To evaluate the effect of chronic b-AR stimulation on mPTP, we incubated adult cardiomyocytes with physiological levels of ISO (100 nM) for up to 18 h. ISO treatment augmented flash frequency starting at 12 h (1.5 fold over control) and persisting to 18 h (Fig. 1a,b). Unitary features of flash including amplitude, rising time and decay kinetics were not significantly changed ( Supplementary Fig. 1a-c). The ISO-induced flashes were also coincided with loss of mitochondrial membrane potential (Dc m , Fig. 1a) consistent with previous reports 12 . ISO also dose-dependently increased flash frequency (Fig. 1c). ISO-induced increase in mitochondrial flash frequency was largely blocked by pre-incubation of mPTP inhibitor, cyclosporine A (CsA, 1 mM), knocking down mitochondrial Ca 2 þ uniporter (MCU), or a mitochondria-targeted superoxide scavenger, mitoTEMPO (1 mM) suggesting the mPTP dependence of flash (Fig. 1d-f and Supplementary Fig. 1d). Taken together, chronic b-AR stimulation persistently induced mitochondrial flash-coupled mPTP in a frequency-dependent mode.
Chronic b-AR stimulation-induced myocyte dysfunction. Chronic ISO treatment increased the sensitivity of mitochondria to oxidative stress as shown by a shorter time to laser-induced Dc m loss and decreased resting Dc m suggesting mitochondrial damage ( Supplementary Fig. 2a-d). Meanwhile, ISO also induced myocyte dysfunction as evidenced by increased oxidative stress at 24 h, decreased Ca 2 þ transient amplitude, compromised cardiac contraction, and increased both necrosis and apoptosis at 48 h ( Supplementary Fig. 2e-i). Thus, chronic b-AR stimulation elicited sequential events in mitochondrial and myocyte dysfunction in adult cardiomyocytes ( Supplementary Fig. 2j). Despite the significant impact on mitochondrial and myocyte function, mitochondrial biogenesis and autophagy were not significantly altered by ISO stimulation in vitro (100 nM, 12-48 h) or in vivo (15 mg kg À 1 , 2 weeks) ( Supplementary Fig. 3).
To test whether mPTP underlies ISO-induced mitochondrial and myocyte dysfunction, we blocked mPTP opening by CsA, at 12 h after ISO treatment when increased flash-coupled mPTP was detected, and found that CsA prevented laser-induced Dc m loss suggesting it decreased mPTP sensitivity, maintained Dc m , and decreased cellular oxidative stress at 24-48 h (Fig. 2a-d). Moreover, CsA added at 12 h after ISO stimulation also enhanced Ca 2 þ transient amplitude and contractility and rescued myocyte death at 48 h ( Fig. 2e-g). To obtain molecular evidence on the mPTP dependence of ISO-induced mitochondrial and myocyte dysfunction, we treated adult cardiomyocytes from cyclophilin D knockout (CypD KO) mice with ISO. CypD KO myocytes were resistant to chronic ISO-induced cell death (Fig. 2h) consistent with previous report 17 . Taken together, increased mPTP openings are responsible for mitochondrial and cardiac dysfunction in chronic b-AR stimulation.
To further explore the molecular mechanism on how CaMKII induces mPTP, we overexpressed the dominant-negative CaMKII (CaMKII DN) 19 , which prevented ISO-induced flash activity (Fig. 3d) Fig. 4c). Taken together, ISO-induced mPTP is through CaMKII pathway. In addition, CaMKII DN overexpression prevented ISO-induced mitochondrial dysfunction, as indicated by a longer time of laser-induced Dc m loss, and rescued myocyte death (Fig. 3e,f). These results suggest that CaMKII mediates ISO-induced mPTP and cardiotoxicity.
Next, we manipulated CaMKII activity through a gain of function approach. Overexpression of wild-type CaMKII (CaMKII WT) or a constitutively active CaMKII (CaMKII CA) increased flash activity (Fig. 4a,b) and PLN (ref. 17) phosphorylation ( Supplementary Fig. 4c). CaMKII specific inhibitor (KN93, 0.5 mM) or mPTP inhibitor (CsA, 1 mM) blocked the effect of CaMKII WT (Fig. 4c,d). Overexpression of CaMKII also potentiated the effect of ISO on laser-induced Dc m loss and cardiomyocyte death as indicated by the earlier onset of these events (Fig. 4e,f). Thus increased CaMKII activity is sufficient to induce mPTP and myocyte dysfunction.
Finally, infusing mt-cpYFP transgenic mice with ISO (15 mg kg À 1 for 14 days), a well-established model for heart hypertrophy, resulted in significantly increased flash frequency detected in the perfused intact heart (Fig. 5a). Importantly, administration of KN93 (10 mM kg À 1 ) together with ISO prevented this increase (Fig. 5b). Moreover, propranolol or KN93 reversed the increased heart/body weight ratio and hypertrophic markers (ANP and BNP) ( Fig. 5c-e). Thus, the ISO-CaMKII-mPTP pathway exists in in vivo conditions.
Since CaMKII can be activated by oxidation 20,21 , we tested whether oxidative stress underlies ISO-induced CaMKII activation. Through monitoring mitochondrial reactive oxygen species (ROS) production by MitoSOX red or a mitochondria-targeted genetic H 2 O 2 indicator, mtHyper 22,23 , we found no increase in mitochondrial ROS 12-24 h after ISO treatment ( Supplementary Fig. 5). CaMKII is activated and flash-coupled mPTP openings are increased at 6-12 h after ISO treatment. Thus, it seems that oxidative stress cannot be account for the initial activation of this pathway. We do detected increased cytosolic ROS 24 h after ISO treatment (Fig. 2d), which is consistent with previous reports 6,24 . Therefore, it is likely that oxidative stress is downstream of mPTP and may act in a positive feedback manner to support sustained CaMKII activation and mPTP openings at the later stage 21 .
CaMKII induced mPTP via Drp1 S616 phosphorylation. CaMKII is a cytosolic kinase and we hypothesized that CaMKII may phosphorylate a cytosolic protein, which travels to   mitochondria and modulates mPTP. Drp1 is a cytosolic protein that translocates to mitochondria upon activation by Ca 2 þ or phosphorylation 25,26 . A number of kinases can phosphorylate Drp1 at different sites [25][26][27][28] , such as by a CaMK family member, CaMKI at S637, in the brain 29 . Since in vivo phospho-proteomics analysis reveals proteins with serine/threonine were the cardiac targets of b-AR signalling 30 , we hypothesized that CaMKII may induce Drp1 phosphorylation and through which facilitates mPTP opening 11 . Indeed, by using site-specific phosphorylation antibodies, we found that 2 weeks of ISO infusion significantly increased Drp1 phosphorylation at S616 site (Drp1 S616 , 1.8 fold increase over control) in the mouse heart (Fig. 6a). Further, KN93 (10 mM kg À 1 ) or b1-AR antagonist (propranolol, 10 mg kg À 1 ), efficiently prevented Drp1 S616 phosphorylation (Fig. 6a). In contrast, Drp1 phosphorylation at S637 site (Drp1 S637 ) was only detected in vitro in ISO-treated (12 h) adult myocytes in a cAMP/PKA-dependent but CaMKII-independent manner as previously reported 25 , but not in vivo in the mouse heart after 2 weeks of ISO infusion ( Supplementary Fig. 6a,b). In addition, CaMKII DN did not affect the phosphorylation status of Drp1 at S637 at baseline or after ISO treatment ( Supplementary Fig. 6c), Overexpression WT Drp1 or S616A mutation equally increased S637 phosphorylation due to increased total Drp1 levels, suggesting the phosphorylation status at S616 has no effect on S637 phosphorylation ( Supplementary Fig. 6d). In cultured adult cardiomyocytes, ISO treatment led to Drp1 S616 phosphorylation as early as 6 h and can be abolished by KN93 (Fig. 6b,c). Further, overexpression of CaMKII WT promoted Drp1 S616 phosphorylation (Fig. 6d). These results strongly suggest that CaMKII is upstream of Drp1 S616 phosphorylation during chronic b1-AR stimulation. Next, whether Drp1 S616 phosphorylation promotes its translocation to mitochondria was determined by measuring Drp1 protein levels in mitochondrial and cytosolic fractions of the myocyte after 2 weeks of ISO infusion. Indeed, Drp1 was significantly enriched in mitochondrial fraction (1.84 fold) while decreased in cytosolic fraction (Fig. 6e). KN93 prevented these changes (Fig. 6e). In adult cardiomyocytes treated with ISO, increased 'punctate' Drp1 pattern that colocalized with mitochondrial marker was found by immunofluorescent staining (Fig. 6f). Moreover, we found phosphorylated Drp1 (at S616 site) was enriched in the mitochondrial fraction of the heart preparations after 2 weeks of ISO infusion (Fig. 6g). These results provide strong support that Drp1 phosphorylation at S616 site promotes its mitochondrial translocation.
Finally, we tested whether the increased mitochondrial translocation of Drp1 affects mitochondrial morphology. In H9C2 cardiac myoblast cells, chronic ISO treatment (1 mM for 24 h) shifted the balance of fission/fusion towards fission as indicated by decreased aspect ratio (AR) and form factor (FF) (Fig. 6h). We have used these two parameters to quantify mitochondrial fission/fusion in other cells 31 . As positive control, we found that the dominant-negative Drp1 mutation, Drp1 K38A, increased AR and FF ( Fig. 6h and Supplementary  Fig. 7). Overexpression of CaMKII WT mildly increased, whereas overexpression of CaMKII CA significantly facilitated mitochondrial fission. CaMKII WT also enhanced ISO-induced mitochondrial fission ( Fig. 6i and Supplementary Fig. 7). Calcineurin has been shown to dephosphorylate Drp1 at S637 site and promote its mitochondrial translocation 26 . Therefore, we tested whether CsA, also a calcineurin inhibitor, can affect Drp1 phosphorylation at S616 or S637 site and the mitochondrial translocation of Drp1 during ISO treatment. CsA had no effect on ISO-induced S616 and S637 phosphorylation in H9C2 cardiac myoblast cells, Drp1 translocation to mitochondria and mitochondrial fission ( Supplementary Figs 7 and 8). Taken together, CaMKII promotes fission through Drp1 phosphorylation at S616 and its mitochondrial translocation.
CaMKII bound Drp1 and directly phosphorylated it at S616. CaMKII has many intracellular targets, including kinases and Drp1 S616 phosphorylation can be modulated by other kinases/phosphatases as well. Therefore, it is critical to determine whether CaMKII can directly bind and phosphorylate Drp1 at this site. Co-immunoprecipitation analysis showed that endogenous Drp1 and CaMKII in adult rat cardiomyocytes bound with each other (Fig. 7a). Moreover, we purified CaMKII from H9C2 cells and murine WT Drp1 or the S579A mutation (equivalent to human Drp1 S616A) from Escherichia coli (Supplementary Fig. 9) and incubated them together in the presence of calmodulin and Ca 2 þ . The results showed that anti-P-S616 antibody detected phosphorylation in WT Drp1, but not the mutated Drp1, when incubated with CaMKII (Fig. 7b). Interestingly, in this in vitro phosphorylation reaction system, we found CaMKII and Drp1 S616A can form a more stable enzyme-substrate intermediate compared with CaMKII and WT Drp1 (Fig. 7c). Since, after being phosphorylated, the protein tends to dissociate with its kinase to facilitate subsequent binding and phosphorylation of a new protein, these results support that CaMKII directly binds and phosphorylates Drp1 at S616 site.
Blocking Drp1 ameliorated myocyte and heart dysfunction. Whether Drp1 S616 phosphorylation contributes to CaMKIIinduced mPTP is determined. We found that overexpression of non-phosphorylatable Drp1 mutation (Drp1 S616A) prevented ISO-induced flash activity (Fig. 8a). Moreover, cardiomyocytes overexpressing Drp1 S616A were resistant to laser-induced Dc m loss and showed less cell death after 48 h ISO treatment (Fig. 8b,c). Additionally, overexpressing Drp1 K38A prevented ISO-induced mPTP and cell death ( Fig. 8d-f), CaMKII-induced mitochondrial flashes, and ISO's effect on laser-induced Dc m loss ( Supplementary Fig. 10). To gain therapeutic insights over the role of Drp1 in ISO-induced heart dysfunction, we blocked Drp1 activity in vivo by Mdivi-1 (50 mg kg À 1 ), a compound inhibiting Drp1 activity, and found that ISO infusion-induced flash activity was decreased and cardiac hypertrophy attenuated (Fig. 8g,h). These results demonstrate that increased Drp1 activity likely due to Drp1 S616 phosphorylation contributes to heart hypertrophy.
Increased Drp1 S616 phosphorylation in human failing hearts. Chronically elevated b1-AR activity is a hallmark of heart failure in human. To further test the clinical relevance of the CaMKII-Drp1-mPTP pathway, we determined Drp1 S616 phosphorylation in human failing hearts. Ventricular samples from dilated cardiomyopathy or ischaemic heart failure patients 32 showed significantly increased Drp1 S616 phosphorylation with no change in total Drp1 levels (Fig. 8i). These data strongly support that increased Drp1 S616 phosphorylation is associated with human heart failure.

Discussion
In this report, we demonstrate a novel signalling pathway that mediates chronic b-AR stimulation-induced mitochondrial dysfunction (Fig. 9). Specifically, CaMKII directly phosphorylates a fission protein, Drp1 at a novel site and this post-translational modification enhances Drp1 translocation to mitochondria, mPTP openings and eventually myocyte death. Massive or irreversible mPTP openings are final executors of mitochondrial pathway of cell death and have been linked to chronic b-AR stimulation 6,33 . The mPTP is also known to open transiently or in subconductance state under physiological conditions 8,9 , which may bear essential signalling roles 34,35 . It is largely unknown whether these two modes of mPTP openings are connected and how the pathological mPTP is induced in the stressed heart. Results from this study show that the physiological mPTP openings, if activated persistently, are detrimental. Further, the CaMKII-Drp1-mPTP pathway could be a key modulator of stress-induced heart pathology and a potential target for heart failure therapy. mPTP is a non-specific and large pore that can be triggered by pathological perturbations [9][10][11]33 . In the heart, massive and irreversible mPTP openings are the culprit of ischaemia reperfusion injury 9 . However, transient or subconductance mPTP openings are detected under resting conditions 36,37 , and may have important physiological functions 38 . Several pieces of evidence shown here suggest that the physiologically relevant mPTP openings (for example, mitochondrial flash-coupled) could be detrimental under stress conditions. mPTP openings was persistently increased after 12-18 h of ISO treatment and preceded oxidative stress, loss of Dc m and myocyte death. Blocking the persistent mPTP openings at the early phase (12 h) attenuated mitochondrial damage and myocyte death at the later stage. Increased mPTP was downstream of CaMKII pathway, which has been linked to cell death during chronic b-AR stimulation 4,5,7,19 . Finally, increased mPTP was observed in the heart after chronic ISO infusion. Thus, mPTP openings, if persistently activated, could induce mitochondrial and myocyte dysfunction.
How to differentiate the physiological and pathological mPTP openings is an important question. The mitochondrial flash is such an optical readout that may be used to answer this question. We have shown that flash-coupled mPTP in the normal heart is at a low frequency and is moderately induced by physiological stimulations such as Ca 2 þ transients during excitationcontraction coupling 22 . On the other hand, dramatic increases in flash-coupled mPTP openings such as by ischaemia reperfusion, excessive oxidants, or persistently high Ca 2 þ , are detrimental 12,17 . Therefore, the magnitude and duration of mPTP openings could determine the eventual outcome. This study supports this idea by showing that chronic b-AR stimulation persistently increased the frequency of mPTP openings, and over time, caused myocyte death. It is conceivable that b-AR stimulation-induced mPTP openings may be a compensatory response to either stimulate mitochondrial respiration or limit matrix Ca 2 þ overload in the short term. However, in the long run, persistently increased mPTP openings could compromise bioenergetics, induce oxidative stress and disturb Ca 2 þ homeostasis. In agreement with this speculation, increased mPTP openings have been linked to oxidative insultinduced apoptosis and neural excitotoxicity 17,39 .
This study also provides evidence to show the causal role of CaMKII in chronic b-AR-induced mPTP openings and cell death. Chronic b-AR stimulation causes mitochondrial swelling, irreversible mPTP openings and mitochondrial-dependent apoptosis 6,33 . Whether these effects are mediated by CaMKII is not known. Here, we show that CaMKII activation is both sufficient and necessary to stimulate mPTP. First, ISO-induced mPTP openings was blocked by CaMKII inhibitors and a dominant-negative CaMKII mutation. Moreover, suppressing or enhancing CaMKII activity prevented or facilitated mPTP openings, mitochondrial damage and myocyte death. CaMKII also modulates cytosolic and sarcoplasmic reticulum (SR) Ca 2 þ handling and gene expression through phosphorylating various targets 40,41 . Specifically, the cytosolic Ca 2 þ derangement, such as increased SR leak and resting Ca 2 þ levels, may synergistically stimulate mPTP opening and contribute to cell death. This speculation is consistent with recent reports underscoring the role of CaMKII in mitotoxicity, heart dysfunction and hypertrophy 7,42 . A recent report showed that CaMKII knockout mouse has lower myocyte death but still develops heart hypertrophy after chronic ISO infusion 43 . We speculate that, unlike the pharmacological and acute inhibition of CaMKII in this study, total ablation of CaMKII in the heart may activate compensatory mechanisms that mediate the hypertrophy phenotype.
To address how CaMKII signalling pathway modulates mPTP, we show that mitochondrial fission protein Drp1 is a novel mediator. Drp1 is a cytosolic protein that can be    25 . It is possible that the phosphorylation of Drp1 at PKA site (S637) versus CaMKII site (S616) serves as a switch or checkpoint to determine distinct outcomes during acute versus chronic b-AR stimulation. Finally, we show that Drp1 S616 phosphorylation or Drp1 activity is required for b-AR stimulation-or CaMKII-induced mPTP openings, since mutating this phosphorylation site (S616A) or the GTP-binding site (K38A) of Drp1 efficiently  blocked increased mPTP openings, mitochondrial damage and cardiomyocyte death. However, how Drp1 S616 phosphorylation or Drp1 activity stimulates mPTP is not fully understood. Previous reports indicates that mitochondrial fission is essential for mPTP openings in response to stresses such as hyperglycaemia, and inhibiting mitochondrial fission protects against ischaemia reperfusion injury probably through suppressing mPTP (refs 11,31). Another possibility is that Drp1 may promote mitochondrial fission, which could cause membrane fractures or form lipidic pores 45 . Finally, the mitochondrial outer membrane proteins, such as the Bcl-2 family proteins, may interact with Drp1 and modulate mPTP (for example, through outer membrane permeability) 46-48 . In summary, two novel findings were reported in this study. First, chronic b-AR stimulation persistently increased mPTP openings through CaMKII pathway. Second, CaMKII directly interacted and phosphorylated Drp1 at S616 site and through which increased mPTP openings. Therefore, activation of the cytosolic kinase CaMKII modulates mitochondrial function through phosphorylation of a fission protein. This pathway could contribute to mitochondrial dysfunction and myocyte death during chronic b-AR stimulation, a clinically important condition often seen in heart failure patients. Recombinant adenovirus vectors construction. Construction of recombinant adenovirus vectors containing: mitochondrial-targeted circularly permuted yellow fluorescent protein (Ad-mt-cpYFP) 12 , HA-tagged CaMKIId C DN, CAMKIId C WT and CaMKIId C CA were generated previously 19 . Recombinant adenovirus expressing MCU short hairpin RNA and Drp1 S616A were generated by Vector BioLabs (Malvern, PA, USA) with the original construct obtained from Addgene (Cambridge, MA, USA). Adenovirus containing Drp1 K38A or mtHyper were generated previously 31,49 . Concentration of the viruses was determined to be at B1 Â 10 11 viral particles per ml. All the viruses were aliquoted and stored at À 80°C.
Adult cardiomyocyte culture and gene transfer. All the animal procedures were approved by the Institutional Animal Care and Use Committee at the University of Washington. Cardiomyocytes were isolated from the heart of female Sprague Dawley rats (200-250 g, Harlan) using standard enzymatic technique 22 . Briefly, rat was anaesthetized by intraperitoneal (i.p.) injection of pentobarbital. The heart was quickly removed, cannulated via the ascending aorta, and mounted on a modified Langendorff perfusion system. The heart was perfused with oxygenated Krebs-Henseleit Buffer (KHB) solution supplemented with collagenase II Animal experiments. Pan-tissue mt-cpYFP transgenic mice were generated using C57BL/6 mice (Charles River) and with the pUC-CAGGS-mt-cpYFP vector as previously reported 22,50,51 . The mice were housed under standard conditions with a 12-h light-dark cycle and a constant temperature (22 ± 2°C). Food and water were provided ad libitum. To establish the cardiac hypertrophy model, transgenic mice of both genders at the age of 8-10 weeks (littermates) were randomly assigned to surgery or sham operation groups. ISO (15 mg kg À 1 d À 1 ) with or without KN93 (10 mM kg À 1 d À 1 ), Prop (10 mg kg À 1 d À 1 ) and Mdivi-1 (50 mg kg À 1 d À 1 ) were delivered by mini-osmotic pumps (2002, Alzet Corporation) implanted subcutaneously under anaesthesia with avertin (2.5% wt/vol, 15 ml g À 1 body weight, i.p.), respectively. After 2 weeks, mice were euthanized and hearts were excised for perfused heart confocal imaging or weighed, and frozen in liquid nitrogen for Western blot and real-time PCR assay. In addition, male CypD KO mice and its WT controls (B6;129-Ppiftm1 J mol J À 1 , 3 months old) were purchased from the Jackson Laboratory. The adult cardiomyocytes isolating from these mice were cultured and treated with 100 nM ISO to determine the role of CypD in ISO-induced cell death.
Confocal imaging of mitochondrial flashes. Single mitochondrial flash was detected by using a Zeiss LSM 510 Meta confocal microscope equipped with a 40 Â 1.3 NA oil immersion objective 12 . For detecting mitochondrial flashes in adult cardiomyocytes, cells infected with Ad-mt-cpYFP virus were incubated in modified KHB buffer (138 mM NaCl, 3.7 mM KCl, 1.2 mM KH 2 PO 4 , 5 mM Glucose, 20 mM HEPES and 1 mM CaCl 2 ) at room temperature. The cells were cultured on 25 mm coverslips and tightly attached to a custom designed perfusion chamber mounted on confocal microscope stage. Dual excitation images of mt-cpYFP were taken by alternating excitation at 405 and 488 nm and collecting emissions at 4505 nm. Time-lapse x,y images were acquired at 1,024 resolution for 100 frames and at a sampling rate of 1 s per frame. During the scan, the spatiotemporal occurrence of flash is stochastic and no apparent laser or time-dependent induction of flashes was observed. In a subset of experiments, mitochondrial membrane potential indicator, TMRM (20 nM, Invitrogen) was loaded into cells and tri-wavelength imaging of mt-cpYFP and TMRM were taken by sequential excitation at 405, 488 and 543 nm, and emissions collected at 505-550, 505-550 and 4560 nm, respectively 22 .
For imaging single mitochondrial flash activity in intact hearts from mt-cpYFP transgenic mice, the hearts were perfused with modified KHB buffer containing 0.5 mM pyruvate and 10 mM glucose at 37°C and bubbled with 95% O 2 and 5% CO 2 . The fluorescence acquisition plane was focused about 30 mm into the myocardium from the heart surface. Blebbistatin (10 mM) and gentle pressure were applied when taking time lapse two-dimensional (2D) images of the myocardium. In some experiments, TMRM (Dc m indicator, 100-500 nM) was included in the perfusion solution 50 .  N ¼ 17,714, 17,692, 9,567 and 17,035 mitochondria in the groups of Vehicle, CaMKII WT, WT þ ISO and CaMKII CA, respectively. Data are mean ± s.e.m. *Po0.05 versus Vehicle, # Po0.05 versus ISO. The data were analysed using One-way ANOVA followed by Turkey post test.
Laser-induced Dw m loss and determination of resting Dw m . Laser-induced abrupt Dc m loss was used as an indication of the sensitivity of mitochondria to oxidative stress-induced irreversible mPTP opening in intact myocytes as previously described 52 . Briefly, the adult cardiomyocytes were incubated with TMRM (20 nM) in KHB buffer at 37°C for 20 min. Confocal linescan (x, t) was set-up along the long axis of the myocytes and encompass 10-15 mitochondria that showed clear pattern and TMRM loading. TMRM was excited at 543 nm and at a slow speed (35 ms per line) to trigger oxidative stress and Dc m loss, which is shown as a sudden loss of TMRM signal in that mitochondrion. The time from the start of the scan to the complete loss of TMRM signal was calculated as the time of mPTP openings. A shorter mPTP time indicates that the pore is more sensitive to laser-induced openings. The intensity of the TMRM fluorescence at beginning of the scan was used as the basal Dc m in individual mitochondrion.  Western blot analysis. The heart tissues and adult cardiomyocytes were lysed with cell lysis buffer in the presence of a cocktail of proteinase/phosphatase inhibitors (Cell Signaling Technology, Inc., Danvers, MA, USA) and centrifuged at 12,000g for 30 min at 4°C. Isolation of mitochondrial and cytosolic fractions from whole heart was performed using the Mitochondria/cytosol Fractionation Kit (Beyotime Institute of Biotechnology, Shanghai, China). The human man heart samples were collected in a previous study 32 and the study protocol was approved by local ethics committee. The extracted proteins were separated by NuPAGE Novex 4-12% Bis-Tris Gels (Life Technologies). After transferring the proteins on to nitrocellulose membranes or polyvinylidene difluoride (PVDF) membranes, the membranes were blocked and incubated with various primary antibodies at 4°C overnight. We used the following primary antibodies for the Western blots: anti- . The appropriate anti-rabbit or antimouse secondary antibodies were applied. The immunoblots signals were detected and quantified using an Odyssey Infrared Imaging System (LI-COR, Lincoln, NE, USA) or a ChemiDoc XRS þ System with Image Lab Software (Bio-Rad). All the uncropped scannings are provided in Supplementary Fig. 11.
Real-time PCR analysis. Quantitative real-time PCR was applied to detect the hypertrophy-related genes by using SYBR green (Bio-Rad). Total ribonucleic acid was isolated from fresh heart tissue using the RNeasy Kit, and reverse transcripted to complementary DNA by using Omniscript reverse synthase (Qiagen). The primers for ANP and BNP are kindly gift from Dr Stephen C. Kolwicz (University of Washington). The primers for ANP were 5 0 -ATTGACAGGATTGGAGCC CAGAGT-3 0 and 5 0 -TGACACACCACAAGG GCTTAGGAT-3 0 ; for BNP were 5 0 -GCCAGTCTCCAGAGCAATTCA-3 0 and 5 0 -GGGCCATTTCCT CCGAC TT-3 0 . The real-time PCR result for the messenger RNA levels of each gene was repeated four times and was normalized to 18S ribosomal RNA levels.    31 . Individual mitochondria (particles) were subjected to particle analyses to acquire values for the FF (the reciprocal of circularity value) and AR (major axis/minor axis). High values for FF represent elongated tubular mitochondria, and increased AR values indicate an increase of mitochondrial complexity (length and branching).
Co-immunoprecipitation. Adult cardiomyocytes from rats were collected in PBS with 0.1% Triton X-100. After three freeze-thaw cycles, the cell lysates were centrifuged at 13,000 r.p.m. for 15 min. The supernatant were incubated with Protein A/G agarose beads (Thermo Fisher) for 2 h to remove IgG in the samples. The samples were centrifuged (5,000 r.p.m.) and the protein concentrations in the supernatant were determined by Pierce BCA protein assay kit. The protein samples were incubated with either mouse anti-Drp1 antibody (BD Biosciences, 611113) or mouse IgG (Sigma, A0919) for 12 h at 4°C. The Protein A/G agarose beads were added and incubated overnight at 4°C. After binding, the beads were washed with PBS containing 0.1% Triton X-100 for five times. Proteins were eluted by using 0.1 M glycine (pH ¼ 2.7) and loaded on SDS-polyacrylamide gel electrophoresis (PAGE) for Western blot. The antibody used are rabbit anti-Drp1 (Novus Biologicals, NB11055288, 1:2,000), rabbit anti-CaMKII (GeneTex, GTX111401, 1:1,000) and mouse anti-rabbit light chain second antibody (Jackson ImmunoResearch Laboratories, 1:10,000).
In vitro Drp1 phosphorylation and binding assay. The complementary DNA of mouse WT Drp1 or the S579A (equivalent to S616A in human) mutation was cloned into pET30a vector between the Nde1 and Not1 sites (Novagen) by using the forward primer: 5 0 -CAT ATG CAC CAT CAT CAT CAT CAT GAG GCG CTG ATC CCG GTC ATC-3 0 , and the reverse primer: 5 0 -GCG GCC GCT CAT CAC CAA AGA TGA GTC TCT CGG ATT TCA-3 0 . The forward primer contains 6 Â His tag sequence. E. coli Rosetta (DE3) cells were used for the prokaryotic expression of Drp1 proteins. When the cell density reached an OD 600 of B1.0, protein expression was induced by adding 0.5 mM isopropyl 1-thio-b-D-galactopyranoside and the cells were cultured at 4°C for 7 days. The resulting poly-His-tagged WT Drp1 or S579A mutation was purified by Nickle resin (Ni Sepharose 6 Fast Flow, GE Healthcare) followed by imidazole elution. The purified recombinant Drp1 or Drp1 S579A was evaluated by SDS-PAGE and coomassie blue staining which revealed a purity of B90% and a molecular weight of B80 kDa (Supplementary Fig. 9). HA-tagged CAMKIId C WT was purified from H9C2 myoblast cells after adenovirus-mediated gene expression. Calmodulin was purchased from Millipore. For in vitro Drp1 phosphorylation assay, anti-HA magnetic beads (Pierce) were used to concentrate and immobilize HA-CaMKII. The immobilized HA-CaMKII beads (5 ml) were incubated with purified Drp1 or Drp1 S616A (2 mM) in a 100 ml reaction buffer containing 50 mM Tris-HCl, 150 mM NaCl and 25 mM ATP (pH7.4) for 24 h at 30°C in the presence or absence of 1 mM CaCl 2 and/or 1 mM calmodulin. After the incubation, the supernatant was used to detect Drp1 phosphorylation by Western blots using anti-P-S616 antibody (1:1,000, Cell Signaling) and rabbit anti-Drp1 antibody (1:2,000, Cell Signaling). The beads were washed, and immobilized HA-CaMKII and Drp1 complex was eluted by boiling, and the samples were used for Western blot analysis using rabbit anti-HA (1:5,000, Sigma) and rabbit anti-Drp1 (1:2,000, Cell Signaling) antibodies.
Statistical analysis. All experimental results are expressed as mean ± s.e.m. Each experiment was conducted at least three times. When multiple experiments using different numbers of animals were pooled for the statistical analysis, the range of number of animals was indicated in the figure legend. Data comparisons among the groups were performed using One-way ANOVA and unpaired Student's t-test when appropriate. A P value o0.05 was considered statistically significant.
Data availability. All data supporting the findings of this study are available within the article, its Supplementary Information File, or from the corresponding authors on request. Sustained ISO treatment activates CaMKII pathway, a downstream kinase of b1-AR, and subsequently increases the phosphorylation of Drp1 at S616 (Drp1 S616 ), which activates Drp1. After translocating to the outer membrane of mitochondria, the phosphorylated Drp1 triggers fission and mPTP openings, which are recorded by mitochondrial flash events. Finally, chronic activation of this pathway leads to mitochondrial and myocyte dysfunction. Abolishing CaMKII activity (CaMKII DN, KN93 or AIP), inhibiting Drp1 activity (Drp1 K38A or Mdivi-1), preventing Drp1 S616 phosphorylation (Drp1 S616A), or blocking mPTP openings (CsA or CypD KO) efficiently prevented myocyte damage and cardiac hypertrophy during chronic b1-AR stimulation.