Melatonin ameliorates myocardial ischemia/reperfusion injury in type 1 diabetic rats by preserving mitochondrial function: role of AMPK-PGC-1α-SIRT3 signaling

Enhancing mitochondrial biogenesis and reducing mitochondrial oxidative stress have emerged as crucial therapeutic strategies to ameliorate diabetic myocardial ischemia/reperfusion (MI/R) injury. Melatonin has been reported to be a safe and potent cardioprotective agent. However, its role on mitochondrial biogenesis or reactive oxygen species (ROS) production in type 1 diabetic myocardium and the underlying mechanisms remain unknown. We hypothesize that melatonin ameliorates MI/R injury in type 1 diabetic rats by preserving mitochondrial function via AMPK-PGC-1α-SIRT3 signaling pathway. Both our in vivo and in vitro data showed that melatonin reduced MI/R injury by improving cardiac function, enhancing mitochondrial SOD activity, ATP production and oxidative phosphorylation complex (II, III and IV), reducing myocardial apoptosis and mitochondrial MDA, H2O2 generation. Importantly, melatonin also activated AMPK-PGC-1α-SIRT3 signaling and increased SOD2, NRF1 and TFAM expressions. However, these effects were abolished by Compound C (a specific AMPK signaling blocker) administration. Additionally, our cellular experiment showed that SIRT3 siRNA inhibited the cytoprotective effect of melatonin without affecting p-AMPK/AMPK ratio and PGC-1α expression. Taken together, we concluded that melatonin preserves mitochondrial function by reducing mitochondrial oxidative stress and enhancing its biogenesis, thus ameliorating MI/R injury in type 1 diabetic state. AMPK-PGC1α-SIRT3 axis plays an essential role in this process.


Compound C abolished melatonin-induced cardioprotective effect on myocardial ischemia/ reperfusion injury in type 1 diabetic rats.
To determine the effect of melatonin on ischemia/reperfusion injury in type 1 diabetic setting and the underlying mechanisms, we employed Compound C (a specific AMPK signaling blocker) in our in vivo experiment. No significant changes in cardiac function and apoptotic signaling were found between T1D + MI/R + V group and T1D + MI/R + CC group ( Supplementary Fig. S2a-f), indicating that under experimental dosage, Compound C caused no significant effects on diabetic heart. After 3 hours of reperfusion, we found that melatonin treatment significantly improved LVSP and ± dP/dt max (Fig. 2a-c, P < 0.05, compared with the T1D + MI/R + V group). Meanwhile, melatonin-treated group also exhibited reduced myocardial infarction and apoptotic index ( Fig. 2d-f, P < 0.05, compared with the T1D + MI/R + V group). However, these protective actions were abolished by Compound C administration ( Fig. 2a-f, P < 0.05, compared with the T1D + MI/R + Mel group). Additionally, melatonin also significantly inhibited apoptotic signaling pathway by reducing the expressions of caspase-3, bax and cleaved caspase-3 and increasing Bcl-2 level (Fig. 2g-k, P < 0.05, compared with the T1D + MI/R + V group). These effects were also inhibited by Compound C (Fig. 2g-k, P < 0.05, compared with the T1D + MI/R + Mel group). These data all indicated that AMPK signaling played a pivotal role in melatonin's cardioprotective effect.
Compound C impaired mitochondrial function and blunted melatonin-induced suppression on mitochondrial oxidative damage in diabetic myocardium. To assess the mitochondrial function in this setting, we evaluated mitochondrial SOD activity, MDA content, H 2 O 2 formation and ATP content in these experimental groups. As shown in Fig. 3a-d, melatonin effectively preserved mitochondrial function by improving mitochondrial SOD activity and ATP production, and reducing mitochondrial MDA and H 2 O 2 generation (P < 0.05, compared with the T1D + MI/R + V group). Interestingly, inhibition of AMPK signaling also blunted these effects and impaired mitochondrial function (P < 0.05, compared with the T1D + MI/R + Mel group). Furthermore, we assessed the expressions of mitochondrial electron transport chain complexes using western blot analysis. In agreement with the previous data, we found that oxidative phosphorylation (OXPHOS) complex subunits (complexes II, III and IV) were up-regulated in melatonin-treated group (Fig. 3e, compared with the T1D + MI/R + V group). However, Compound C administration also blocked this effect (Fig. 3e, compared with the T1D + MI/R + Mel group). Compound C inhibited myocardial AMPK/PGC-1α signaling in diabetic myocardium. After 3 hours of reperfusion, melatonin significantly increased the expressions of PGC-1α , SIRT3, SOD2, NRF1, TFAM and the p-AMPK/AMPK ratio, decreased cytosolic cytochrome c level ( Fig. 4a-h, P < 0.05, compared with the T1D + MI/R + V group). However, these effects were also abolished by Compound C administration ( Fig. 4a-h, P < 0.05, compared with the T1D + MI/R + Mel group). We next carried out immunohistochemistry analysis. Consistently, we found that melatonin-treated group also showed increased PGC-1α , SIRT3 and SOD2 staining, while inhibition of AMPK by Compound C blunted these effects (Fig. 4i).
Compound C and SIRT3 siRNA transfection blunted melatonin-induced anti-apoptotic effect against SIR injury in high glucose medium treated H9c2 cells. To further confirm the cardioprotective mechanisms of melatonin in promoting mitochondrial function in type 1 diabetic heart, we performed in vitro experiment using H9c2 cardiomyoblasts. Initially, we found that 6 hours of high glucose incubation reduced AMPK phosphorylation and the expressions of PGC-1α , SIRT3 and SOD2 in a glucose concentration-dependent manner ( Supplementary Fig. S1, P < 0.05, compared with the NG+ mannitol group). Then, we evaluated the transfection efficiency of SIRT3 small interfering RNA. As shown in Supplementary Fig. S2g, compared with the SIR+ control siRNA group, SIRT3 siRNA significantly down-regulated SIRT3 expression (P < 0.05). Melatonin not only improved cell viability but also reduced cellular apoptosis by decreasing the percentage of TUNEL positive nuclei and inhibiting the expressions of caspase-3 and cleaved caspase-3 ( Fig. 5a-c,e-g, P < 0.05, compared with the HG + SIR group). However, these cytoprotective effect were blunted by Compound C or SIRT3 siRNA treatment ( Fig. 5a-c,e-g, P < 0.05, compared with the HG + SIR + Mel group). Consistently, melatonin effectively alleviated cellular shrinkage and detachment induced by SIR treatment, while this effect was also inhibited by Compound C or SIRT3 siRNA (Fig. 5d), indicating that AMPK signaling and SIRT3 signaling are both key signaling pathways that mediate melatonin's protective actions.
Compound C and SIRT3 siRNA transfection impaired mitochondrial function and inhibited melatonin-induced suppression on mitochondrial oxidative damage in high glucose medium treated H9c2 cells. As shown in Fig. 6a-d, melatonin markedly increased mitochondrial SOD activity and ATP production, decreased mitochondrial MDA and H 2 O 2 generation (P < 0.05, compared with the HG + SIR  group), while these effects were abolished by either Compound C or SIRT3 siRNA (P < 0.05, compared with the HG + SIR + Mel group). Additionally, western blot analysis showed that OXPHOS complexe I, II, III and IV were increased in melatonin-treated group (Fig. 6e, compared with the HG+ SIR group. However, Compound C or SIRT3 siTNA also blunted this effect (Fig. 6e, compared with the HG + SIR + Mel group). These data suggested that melatonin could also attenuated mitochondrial oxidative stress and preserved mitochondrial function in high glucose cultured H9c2 cells. Importantly, AMPK and SIRT3 signaling mediated this action.
Compound C and SIRT3 siRNA transfection reduced cellular SOD2, NRF1, TFAM expressions and increased cytosolic cytochrome c expression. Finally, we investigated the signaling molecules of AMPK-PGC-1α pathway, SIRT3 as well as mitochondrial biogenesis related protein expression. As shown in Fig. 7a-d, we found that melatonin significantly increased the p-AMPK/AMPK ratio and the expressions of PGC-1α and SIRT3 (P < 0.05, compared with the HG + SIR group). Compound C markedly reduced p-AMPK/ AMPK ratio and also the expressions of PGC-1α and SIRT3 (P < 0.05, compared with the HG + SIR + Mel group), indicating that AMPK might be the up-stream signaling of PGC-1α and SIRT3. It is consistent with that SIRT3 siRNA did not significantly changed p-AMPK/AMPK ratio and PGC-1α expression (Fig. 7b,c, P > 0.05, compared with the HG + SIR + Mel group). Meanwhile, both SIRT3 siRNA and Compound C markedly inhibited the up-regulation of SOD2, NRF1 and TFAM induced by melatonin treatment (Fig. 7e-g, P < 0.05, compared with the HG + SIR + Mel group). Additionally, melatonin also significantly reduced cytosolic cytochrome c expression ( Fig. 7h, P < 0.05, compared with the HG + SIR group), which was also blocked by Compound C or SIRT3 siRNA (P < 0.05, compared with the HG + SIR + Mel group). These results all indicated that the enhanced mitochondrial biogenesis and reduced oxidative stress were mediated by the activation of AMPK-PGC-1α -SIRT3 signaling (Fig. 7i).

Discussion
There are several major findings in the present study. Firstly, we demonstrated that melatonin treatment is a potential strategy to ameliorate MI/R injury in type 1 diabetic state by enhancing mitochondrial biogenesis and preserving mitochondrial function. Secondly, AMPK-PGC-1α -SIRT3 signaling pathway was found to play a key role in melatonin's cardioprotective action. To the best of our knowledge, this is the first study demonstrating the cardioprotective effect and potential mechanisms of melatonin against MI/R injury in type 1 diabetic state.
During recent years, the global prevalence of diabetes mellitus has grown sharply, which has become one of the most serious health issues worldwide 22 . Importantly, diabetes is associated with higher risk of cardiovascular  complications such as coronary artery disease, congestive heart failure and acute myocardial infarction 23 . To make things worse, the long-term prognosis is also much worse than non-diabetic individuals with higher rates of residual ventricular dysfunction and overall mortality 23,24 . So far, the mechanisms have not been fully elucidated, but prolonged hyperglycemia-enhanced oxidative stress was indicated to be a crucial contributor 25 . Owing to its critical role in generating ATP and ROS, cardiomyocyte mitochondria has been specifically investigated. In diabetic myocardium, mitochondria is deemed as one of the major sources of free radicals. Numerous studies have reported that hyperglycemia markedly impaired mitochondrial morphology and function, causing electron leakage and O 2 •− generation 25,26 . The enhanced oxidative stress has been confirmed to be a key contributor to increased myocardial vulnerablility to MI/R injury 25 . Additionally, reduced mitochondrial biogenesis was also found in diabetic myocardium. Since mitochondrial ATP synthesis is the main source of energy in the heart. Loss of mitochondrial number during MI/R injury can also result in cardiac contractile dysfunction 5 . In fact, plenty of studies have shown that rescuing mitochondria function by reducing mitochondria derived ROS and enhancing its biogenesis might be a promising therapeutic strategy against MI/R injury in diabetic state 5,27 . Consistent with these findings, we found that reperfusion injury significantly impaired cardiac function and enhanced myocardial apoptosis in diabetic animals. Meanwhile, MI/R injured rats also showed increased mitochondrial MDA, H 2 O 2 generation and decreased mitochondrial SOD activity and ATP production, indicating that reperfusion injury also caused mitochondrial dysfunction.
As an endogenous circadian hormone, melatonin has been proved to be a promising antioxidant against MI/R injury due to its strong anti-oxidative capacity and high safety profile 9,28-31 . Previously, we and others also demonstrated that melatonin exerted a solid protective effect against MI/R injury in non-diabetic animals by activating multiple intracellular signaling pathways such as silent information regulator 1 (SIRT1), Janus kinase 2 (JAK2)/ signal transducer and activator of transcription 3 (STAT3) and Notch1/Hairy and enhancer of split 1 29,32,33 . However, under type 1 diabetic state, its cardioprotective effect was poorly defined. Interestingly, a recent study by Figure 5. Compound C and SIRT3 siRNA transfection blunted melatonin-induced anti-apoptotic effect against SIR injury in high glucose medium treated H9c2 cells. The H9c2 cells were exposed to high glucose medium (33 mmol/l) for 6 hours before the SIR treatment and during the entire reperfusion period (4 hours). HG-treatment was co-administered with or without melatonin (10 μ mol/l) to evaluate its cytoprotective effect. Compound C (3 μ mol/l) was administered for 6 hours before the SIR exposure to inhibit the AMPK signaling. The depicted data are the means ± SEM, n = 6/group. ρρ P < 0.01/ ρ P < 0.05 vs the HG group, ηη P < 0.01/ η P < 0.05 vs the HG + SIR group, θθ P < 0.01/ θ P < 0.05 vs the HG + SIR + Mel group. HG, high glucose. SIR, simulated ischemia reperfusion. Mel, melatonin. CC, Compound C.
Bruno et al. showed that by activating cyclic AMP response element binding protein (CREB)-PGC-1α pathway, melatonin prevented mitochondrial dysfunction in rat skeletal muscle 13 . Moreover, Guo et al. demonstrated that melatonin pretreatment suppressed cadmium-induced hepatotoxicity by enhancing mitochondrial biogenesis 34 . We also demonstrated that in a non-diabetic MI/R rat model, melatonin treatment significantly reduced cardiac damage by reducing mitochondrial oxidative stress 32 . These studies all indicate that mitochondria may be a key target of melatonin in various pathological conditions. Consistent with these results, we found that under diabetic state, melatonin not only reduced mitochondrial oxidative stress markers and increased ATP production, but also significantly up-regulated NRF1 and TFAM expressions. Notably, TFAM is a downstream target of NRF1 that initiates the nuclear genes coding for subunits of the mitochondrial OXPHOS complex by specifically binding to the mitochondrial promoters 35 . As expected, we also observed that melatonin-treated group showed markedly enhanced mitochondrial OXPHOS complex subunits expression.
Another novel finding of this study is that we proved the importance of AMPK-PGC-1α -SIRT3 signaling in melatonin's cardioprotective actions. Sirtuins are a family of class III histone deacetylases and their enzymatic activities are dependent on nicotinamide adenine dinucleotide 20 . So far, seven mammalian homologues of sirtuins (SIRT1-7) have been identified. These sirtuins are found to have distinct subcellular localizations and modulate various metabolic and stress-response signaling 20 . In cardiovascular system, SIRT1 and SIRT3 have been extensively studied. In our previous report, SIRT1 was demonstrated to be an important mediator of melatonin's cardioprotective effect 4,33 . We found that in both diabetic and non-diabetic setting, melatonin suppressed MI/R-induced oxidative damage through activation of SIRT1 signaling. SIRT3 has been proved to be preferentially localized to mitochondria and control mitochondrial oxidative signaling and, consequently, the generation of ROS 20,36 . It has been demonstrated that SIRT3 knockdown in cells enhanced mitochondria ROS formation 37 and cardiomyocytes of SIRT3 knockout mice exhibited markedly increased ROS generation 38 . Importantly, SIRT3 activation has also been demonstrated to reduce MI/R injury. Mohsen et al. demonstrated that losartan ameliorated MI/R injury by activating SIRT3 and thus enhancing thioredoxin-1 and catalase transcription 39 . Meanwhile, He et al. found that SIRT3 knock out impaired post-myocardial ischemia cardiac function by causing coronary microvascular dysfunction 40 . Notably, several studies also demonstrated the key role of SIRT3 in reducing  The depicted data are the means ± SEM, n = 6/group. ρρ P < 0.01/ ρ P < 0.05 vs the HG group, ηη P < 0.01/ η P < 0.05 vs the HG + SIR group, θθ P < 0.01/ θ P < 0.05 vs the HG + SIR + Mel group, ψψ P < 0.01/ ψ P < 0.05 vs the HG + SIR + Mel + CC group. HG, high glucose. SIR, simulated ischemia reperfusion. Mel, melatonin. CC, Compound C.
SCiENTiFiC REPORTS | 7:41337 | DOI: 10.1038/srep41337 myocardial injury in diabetic state. For example, SIRT3 knockout mice which were fed with 16 weeks of high-fat diet (HFD) exhibited increased myocardial ROS formation, impaired HIF signalling and reduced capillary density compared with HFD-fed wild type mice 41 . Also, Zeng et al. found that the SIRT3 expression in diabetic mice heart was markedly reduced 42 while SIRT3 activation is the key mechanism of apelin-induced cardioprotective effect against myocardial injury in diabetic mice 43 . Consistently, we found that myocardial SIRT3 expression were significantly reduced after 1 month of STZ injection. Moreover, SIRT3 signaling was further impaired after the MI/R insult. Importantly, both our in vivo and in vitro data showed that melatonin significantly enhanced SIRT3 signaling and suppressed mitochondrial oxidative stress in diabetic cardiomyocytes. However, knockdown of SIRT3 inhibited the cytoprotective actions of melatonin on SIR-injured cardiomyoblasts in hyperglycemic state. These results indicated that reduced SIRT3 signaling might contribute to diabetic MI/R injury. Park et al. showed that the SIRT3 activity in mouse liver and muscle was up-regulated by PGC-1α , suggesting that PGC-1α functions as an upstream regulator of SIRT3 44 . In fact, PGC1α -SIRT3 signaling has been demonstrated as essential for the regulation of mitochondrial oxidative stress and biogenesis 20,45 . Reduced PGC-1α expression have been linked to the pathogenesis of diabetes mellitus 46 . Meanwhile, PGC-1α has been reported to function as the critical downstream molecule of AMPK [47][48][49] . What is more, in the present experiment, we found AMPK-PGC-1α signaling was markedly down-regulated in diabetic myocardium. Therefore, we hypothesized that reduced AMPK-PGC-1α signaling resulted in decreased SIRT3 expression and melatonin might enhance SIRT3 signaling via the activation of AMPK-PGC-1α . As expected, our in vivo data showed that inhibition of AMPK significantly down-regulated PGC-1α and SIRT3 signaling. And the cellular experiment revealed that SIRT3 siRNA inhibited the up-regulation of SOD2, NRF1 and TFAM induced by melatonin treatment without affecting p-AMPK and PGC-1α protein levels. Therefore, we come to the conclusion that melatonin preserves mitochondrial function by reducing mitochondrial oxidative damage and enhancing its biogenesis, thus ameliorating MI/R injury in type 1 diabetic state. AMPK-PGC1α -SIRT3 axis plays an essential role in this process (Fig. 7i).
Previously, we and others demonstrated that melatonin membrane receptors mediated the cardiac protective actions of melatonin 28,33 . Moreover, we recently demonstrated that melatonin reduced flow shear stress (FSS)-induced bone marrow mesenchymal stem cell injury by activating melatonin membrane receptors and AMPK signaling 50 . Kunst et al. also reported that PGC-1α was the target genes of melatonin in murine retina cells. Additionally, melatonin membrane receptors participated in this action 51 . However, the potential role of melatonin membrane receptors in this study was unknown. In diabetic myocardium, whether melatonin modulates AMPK-PGC1α -SIRT3 signaling pathway via its membrane receptors and the detailed mechanisms remain further study.
Taken together, this study investigated the potential cardioprotective effect of melatonin against MI/R injury in diabetic state. We found that melatonin treatment effectively enhanced mitochondrial biogenesis and preserved mitochondrial function, thus reducing MI/R injury in type 1 diabetic state. More importantly, we demonstrated the critical role of AMPK-PGC-1α -SIRT3 signaling pathway in melatonin's cardioprotective actions. These results showed that melatonin treatment might be a promising therapeutic strategy against MI/R injury in diabetic state. Further studies using SIRT3 deficient animals are warranted to confirm these findings.

Methods
Ethics Statement and Aniamls. Eight-week-old male Sprague-Dawley (SD) rats (weighing 180-220 g) were obtained from the General Hospital of Shenyang Military Area Command Experimental Animal Center and kept in adequate-size cages with standard rat chow and water ad libitum. The room was kept at constant temperature and humidity with a light/dark cycle (12/12 h). The animal care and experimental procedures of this study were approved by the General Hospital of Shenyang Military Area Command Committee on Animal Care. All experiments were carried out in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication, 8th Edition, 2011).
Establishment of type 1 diabetic model. Type 1 diabetic model was induced as described previously 52 .
Briefly, the SD rats were fasted overnight and received streptozotocin (STZ, Sigma-Aldrich, MO, USA, 50 mg/kg/d, i.p., dissolved in 0.1 mol/l citrate buffer, pH 4.5) injection for 3 consecutive days. 7 days after the injection, the rats with fasting plasma glucose (PG) above 11.1 mmol/l were classified as diabetic. Intraperitoneal glucose tolerance test (IPGTT) and oral glucose tolerance test (OGTT) were performed by administering glucose (2 g/kg) by gastric lavage or intraperitoneal injection to further confirm the diabetic model 4 .
Myocardial ischemia/reperfusion model and experimental design. MI/R rat model was established as described previously 53 . The rats were anesthetized with 1% sodium pentobarbital (40 mg/kg). Myocardial ischemia surgery was performed by exposing the heart with a left thoracic incision. A silk ligature was placed around left anterior descending coronary artery. After 30 min of ischemia, the ligature was released and the myocardium was reperfused for 3 hours. Sham-operated group were subjected to the same procedures except that the ligature was left untied. During the whole ischemia and reperfusion period, the left ventricular function was measured by a hemodynamic monitoring system (Chengdu Taimeng technology Co., Ltd., China). Left ventricular systolic pressure (LVSP), first derivative of left ventricular pressure (+ dP/dt max and − dP/dt max ) were directly obtained by computer algorithms at baseline, ischemia for 30 min and reperfusion for 1, 2, and 3 hours.
SCiENTiFiC REPORTS | 7:41337 | DOI: 10.1038/srep41337 Determination of myocardial infarct size and apoptosis. Myocardial infarct size were measured after 6 h of reperfusion by Evans blue/TTC double-staining technique (Solarbio Technology, Beijing, China) as described previously 54 . The area at risk (AAR), infarct size (INF) and viable area were assessed digitally using Image-Pro Plus software (Media Cybernetics, MA, USA). The infarct size was expressed as a percentage of infarct area (INF) over total AAR (INF/AAR × 100%).
Myocardial apoptosis was measured with an in situ cell death detection kit (Roche Molecular Biochemicals, Mannheim, Germany) as previously described 55 . The apoptotic index was expressed as the number of apoptotic nuclei/the total number of nuclei counted × 100%. Immunohistochemistry analysis. Immunohistochemistry staining was performed as previously described 56 . The cardiac tissue was fixed with 10% formalin and embedded in paraffin. Anti-PGC-1α antibody (Cell Signaling Technology, MA, USA, 1:100 dilution), anti-SIRT3 antibody and anti-SOD2 antibody (Santa Cruz, CA, USA, 1:100 dilution) were used as the primary antibody. The positive area was detected by 3,3′ -Diaminobenzidine (DAB) staining (Zhongshan biotechnology, Beijing, China). IgG antibody was used as a negative control. The images were photographed at 200× magnification (Olympus BX-63, JAPAN).
Cell treatment and in vitro experimental protocol. H9c2 cardiomyoblast cells (Tiancheng Technology, Shanghai, China) were cultured as described previously 53 . Simulated ischemia/reperfusion treatment was carried out as described by our previous study 53 . Briefly, the cells were initially incubated with the ischemic buffer [0.49 MgCl 2 , 0.9 CaCl 2 , 12 KCl, 137 NaCl, 4 HEPES, 0.75 sodium dithionate, 10 deoxyglucose and 20 lactate (in mmol/L)] in a normal cell culture incubator. After 2 hours of simulated ischemia, the cells were returned to normal culture medium for 4 hours of simulated reperfusion. Before and after the SIR treatment, the cells received different treatment.
As previously described 28,29 , H9c2 cells were incubated in high-glucose medium (33 mmol/l) for 6 hours before the SIR treatment and during the entire reperfusion period (4 hours) to mimic the in vivo diabetic setting. Normal medium (containing 5.5 mmol/l glucose) was used as a control. HG-treatment was co-administered with or without melatonin (10 μ mol/l) to evaluate its cytoprotective effect. Compound C (3 μ mol/l) was administered for 6 hours before the SIR exposure to inhibit the AMPK signaling.
Small interfering RNA transfection. SIRT3 siRNA duplex solution, transfection reagent and medium were all obtained from Santa Cruz biotechnology (CA, USA). The transfection of SIRT3 and control siRNA was carried out in accordance with the manufacture's instruction as described previously 28 . The knockdown capacity of SIRT3 siRNA was evaluated by western blot analysis. Cellular viability assessment. Cellular viability was evaluated by 3-(4,5-dimethylthiazol-2-yl)-2,5-diph enyltetrazolium bromide (MTT) assay kit (Solarbio Technology, Beijing, China). The results was read using a microtiter plate reader (SpectraMax 190, Molecular Device, USA) and calculated by dividing the optical density of samples by that of the control group 55 .
Isolation of myocardial mitochondria and cytosol fraction. The isolation of myocardial mitochondrial was performed strictly as described by our previous study 32 . In brief, the cardiac tissue or cells was washed and homogenized using buffer A [0.3 phenylmethylsulfonyl fluoride, 1 orthovanadate, 1 NaF, 1 EDTA, 10 Tris-HCl, 250 sucrose (in mmol/l), added with 5 μ g/ml each of leupeptin, aprotinin, and pepstatin A], followed by serial centrifugations (1000, 10000 and 100000 g). The 100000 g supernatant was collected and defined as the cytosolic fraction, while the 10000 g pellet was washed, collected and resuspended in buffer B [0.3 PMSF, 1 NaF, 1 orthovanadate, 10 EDTA, 20 Tris-HCl, 150 NaCl (in mmol/l), 1% NP-40, added with 0.5 lg/mL pepstatin A and 5 lg/mL each of leupeptin and aprotinin], followed by 10 min of centrifugation (21000 g). Then, the supernatant was carefully collected and defined as the mitochondrial fraction.
Mitochondrial oxidative stress and functional evaluation. The mitochondrial oxidative stress markers (SOD activity, MDA content, H 2 O 2 formation) and ATP production was measured using commercially available kits (Beyotime biotechnology, Shanghai, China) as described previously 32 . Additionally, we also measured the cytosolic cytochrome c expression to determine the mitochondrial function and apoptosis.

Statistical analysis.
Values are presented as mean ± SEM and were subjected to a 2-tailed Student's t test between two groups or by ANOVA followed by post-hoc Bonferroni's multiple comparison tests (GraphPad Prism software version 5.0, GraphPad Software, CA, USA). Any P value < 0.05 was considered statistically significant.