Dissociation of Bcl-2–Beclin1 Complex by Activated AMPK Enhances Cardiac Autophagy and Protects Against Cardiomyocyte Apoptosis in Diabetes

Male Friend virus B (FVB) mice from The Jackson Laboratory (Bar Harbor, ME) were used for the experiments. Eight-week-old mice were rendered diabetic by intraperitoneal injections of STZ (50 mg/kg) on 5 consecutive days, whereas control mice were injected with vehicle (citrate buffer, pH 4.5). One week after the injections, blood glucose was measured by applying tail blood to a glucometer as previously described (22,23). Mice with blood glucose levels >350 mg/dL were considered diabetic. The diabetic mice were randomly assigned to be treated with or without metformin (200 mg/kg/day in drinking water) for 4 months. In addition, 8-week-old control FVB and cardiac-specific transgenic mice that overexpress a dominant-negative (DN) α2 subunit (D157A) of AMPK (DN-AMPKα2; gift of Dr. Rong Tian, University of Washington, Seattle, WA) (24) were treated with STZ and metformin as described above. Four months after the treatment, left ventricular (LV) function was measured using an isolated buffer-perfused heart preparation as described previously (23,25). All animal protocols were reviewed and approved by the University of Oklahoma Institutional Animal Care and Use Committee.

H9c2 cardiac myoblast cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% FBS and were incubated in a humidified atmosphere of 5% CO₂/95% air at 37°C. Upon reaching 50–60% confluence, the cells were incubated with control (5.5 mmol/L) or high-glucose (30 mmol/L) medium for the indicated times as previously described (21,22). Osmotic control groups (to account for medium hyperosmolarity) were exposed to mannitol (24.5 mmol/L) in normal medium containing glucose (5.5 mmol/L).

H9c2 cells and hearts were homogenized in lysis buffer, and protein contents were measured using the Bradford assay. Total lysates were immunoprecipitated with specific antibodies. Immunoprecipitates or cell lysates were subjected to Western blotting as described previously (26,27).

Immunohistochemistry was carried out as previously described (11,28). Apoptosis in heart sections and cultured H9c2 cells was measured by TUNEL staining using a cell death detection assay kit (Roche Applied Science, Indianapolis, IN). To quantify cell apoptosis in cultured H9c2 cells, apoptosis was analyzed using an Annexin-V-FLUOS staining kit (Roche Applied Science). After treatment, H9c2 cells were collected and labeled for 15 min with Alexa 488–conjugated Annexin V and propidium iodide (PI). The apoptotic cells (annexin V⁺/PI⁻ cells) were detected in the FACScan flow cytometer (BD, San Jose, CA).

AMPK activity was measured in ∼20 mg of cardiac tissue as described previously (29,30), using the SAMS peptide as a substrate.

H9c2 cells were plated at 5 × 10⁴ cells per well, allowed to adhere to the plate overnight, and then infected with green fluorescent protein (GFP)–light chain 3 (LC3) adenovirus. After 24 h of transfection, the cells were incubated with chloroquine (CQ; 5 μmol/L) for 16 h in normal or high-glucose media. Fluorescence images were obtained using an inverted fluorescent microscope (Olympus America, Meliville, NY). Autophagy was measured by quantifying the average number of autophagosomes per cell for each sample. A minimum of 100 cells per sample were counted.

We first determined whether high glucose alters autophagy in H9c2 cells. During autophagy, cytosolic microtubule-associated protein LC3-I is lipidated and converted to LC3-II, which is translocated to the autophagosomal membrane (31). Thus, conversion of LC3-I to LC3-II and accumulation of GFP-LC3 provide effective ways to measure the autophagosome (31). H9c2 cells were treated with d-glucose at a final concentration of 20 or 30 mmol/L for 48 h, and cells treated with 5.5 mmol/L d-glucose were used as a control. To exclude the effect of hyperosmolarity, the cells were treated with 5.5 mmol/L d-glucose plus 24.5 mmol/L of mannitol for 48 h, which did not affect AMPK activity and autophagy (data not shown). LC3-II protein levels began to decline at 20 mmol/L and were reduced about threefold at 30 mmol/L (Fig. 1A).

We next determined the time course of autophagy suppression in a high-glucose environment. H9c2 cells were exposed to high glucose for 12–48 h. A significant reduction in LC3-II protein levels was observed at 24 h, and this effect persisted throughout the study period (Fig. 1B). To study the effect of high glucose on the autophagic flux, H9c2 cells were treated with high glucose in the presence or absence of CQ, a lysosome and autophagosome fusion inhibitor (32). Administration of CQ increased LC3-II accumulation under normal glucose conditions. This increase was attenuated by high-glucose treatment (Fig. 1C). Immunofluorescence analysis demonstrated that elevated glucose levels reduced the number and distribution of GFP-LC3 punctate staining, suggesting a reduction in autophagosome formation. Under normal glucose conditions, CQ treatment enhanced the number and distribution of the GFP-LC3 spots. Notably, this increase was depressed in a high-glucose environment, indicating that elevated glucose levels reduce autophagic flux (Fig. 1D and E).

Metformin induces autophagy through activation of AMPK (11). To explore the mechanisms by which this kinase regulates both autophagy and apoptosis during the development of diabetic cardiomyopathy, we first examined whether AMPK activation by metformin induces autophagy. High glucose inhibited AMPK activity, as estimated by decreased phosphorylation of AMPK and acetyl-CoA carboxylase (ACC), and conversion of LC3-I to LC3-II was decreased (Fig. 1F). Administration of metformin activated AMPK and enhanced LC3-II accumulation under normal glucose conditions. The recovered AMPK activity that occurred after metformin treatment restored the autophagic capacity in high glucose–treated H9c2 cells (Fig. 1F). Under this condition, administration of CQ caused a further increase in LC3-II accumulation. Immunofluorescence analysis revealed that high glucose reduced the number and distribution of GFP-LC3 punctate staining, which was reversed after administration of metformin. Addition of both metformin and CQ to cells treated with high glucose caused an additional increase in the number and distribution of GFP-LC3 puncta. Taken together, these data suggest that metformin restores the autophagic flux in a high-glucose environment (Fig. 1G and H). The effect of AMPK on autophagy was further assessed using genetic means. Transfection of cells with AMPK small interfering RNA (siRNA) reduced LC3-II protein levels under basal conditions. Importantly, metformin treatment restored autophagy capacity in the cells transfected with control siRNA but failed to do so in the cells transfected with AMPK siRNA (Fig. 1I). Conversely, transfection of constitutively active AMPK (CA-AMPK) adenovirus in H9c2 cells enhanced LC3-II protein levels. In a high-glucose environment, metformin treatment induced more autophagy in the cells transfected with CA-AMPK adenovirus than in the cells transfected with GFP adenovirus (Fig. 1J). Taken together, our results implicate that AMPK activation by metformin stimulates autophagy.

We next investigated whether activation of AMPK prevents high glucose–induced apoptosis in H9c2 cells. As compared with normal glucose and osmotic control conditions, elevated glucose levels increased TUNEL-positive cells (Fig. 2A and B) and apoptotic markers, such as cleavage of caspase-3 and PARP (Fig. 2C). Flow cytometric analysis revealed that high glucose induced more apoptotic cell death (11.7 ± 2.2%) than normal glucose (3.6 ± 1.2%) or osmotic control (3.9 ± 1.1%) (Fig. 2D and E). In the cells exposed to high glucose, metformin treatment prevented the reduction in phosphorylation of AMPK and ACC (Fig. 2F), reduced the cleavage of caspase-3 and PARP (Fig. 2G), and diminished the number of apoptotic cells (Fig. 2H).

To determine the functional role of autophagy in the development of diabetic cardiomyopathy, we assessed apoptosis after manipulation of autophagic activity in H9c2 cells. Under normal glucose conditions, metformin treatment enhanced LC3-II accumulation (Fig. 3A), and the conversion of LC3-I to LC3-II was blocked by the autophagy inhibitor 3-methylademine (3-MA) (33). In a high-glucose environment, metformin prevented the reduced LC3-II protein levels; however, this effect was attenuated by administration of 3-MA (Fig. 3A). In addition, metformin attenuated the enhancement of apoptotic markers (cleavage of caspase-3 and PARP) by high glucose, and this protective effect was abrogated by 3-MA treatment (Fig. 3B).

To exclude nonspecific effects of 3-MA (34), we suppressed autophagy by gene silencing of Atg7. Transfection of Atg7 siRNA not only prevented the conversion of LC3-I to LC3-II under basal conditions but also inhibited metformin-enhanced LC3-II accumulation (Fig. 3C). Importantly, metformin reduced apoptotic cell death and cleavage of caspase-3 and PARP in the cells transfected with control siRNA, whereas this protective effect was attenuated in the cells transfected with Atg7 siRNA (Fig. 3C and D).

To confirm that autophagy is necessary for H9c2 cell survival, we assessed apoptosis in H9c2 cells that overexpressed Atg7 adenovirus. Overexpression of Atg7 caused an increase in autophagic flux, as the addition of CQ induced more LC3-II accumulation in the cells infected with Atg7 adenovirus than in those infected with GFP adenovirus (Fig. 3E). In addition, overexpression of Atg7 attenuated high glucose–induced apoptotic cell death and cleavage of caspase-3 and PARP (Fig. 3F and G).

Reports that starvation and a Bcl-2 homology domain 3 molecular mimetic stimulate autophagy through dissociation of the Beclin1–Bcl-2 complex (9,35) prompted us to investigate the effect of AMPK on the interaction between Beclin1 and Bcl-2. Immunoprecipitation of Beclin1 followed by probing for Bcl-2 (or the reverse experiment) indicated that high glucose enhanced the association between Beclin1 and Bcl-2 and that this interaction was disrupted by metformin treatment (Fig. 4A and B). Together with the findings that elevated glucose levels reduced the autophagic flux and metformin restored autophagy in high glucose–treated H9c2 cells, our data suggest that activation of AMPK may stimulate autophagy by dissociation of the Beclin1–Bcl-2 complex.

Since activation of Jun NH₂-terminal kinase 1 (JNK1) results in phosphorylation of Bcl-2 (36,37) and disruption of the Beclin1–Bcl-2 complex (38), we first analyzed JNK1 phosphorylation in H9c2 cells. The cells were exposed to high glucose for 24–36 h. Reduced JNK1 phosphorylation was observed at 24 h and reached its minimum at 36 h. This reduction was prevented by metformin treatment (Fig. 4C). In line with the alteration in JNK1 activation, high glucose decreased Bcl-2 phosphorylation at Ser70, and this effect was depressed by administration of metformin (Fig. 4D).

Next, using gain- and loss-of-function approaches, the role of JNK1 in regulating the interaction between Beclin1 and Bcl-2 was determined. Under normal and high-glucose conditions, the JNK1 inhibitor SP600125 reduced JNK1 phosphorylation (Fig. 4E). Coimmunoprecipitation experiments indicated that metformin reduced the association between Beclin1 and Bcl-2 in normal glucose medium, whereas high glucose enhanced the association between these two proteins and this interaction was disrupted by metformin treatment. Administration of SP600125 abrogated the effects of metformin on the Beclin1–Bcl-2 complex, as Beclin1 and Bcl-2 associated under these conditions (Fig. 4E). As a result, metformin-enhanced autophagy was attenuated (Fig. 4F). Transfection of H9c2 cells with CA-JNK1 plasmid disrupted the association between Beclin1 and Bcl-2 and recovered the accumulation of LC3-II under high-glucose conditions (Fig. 4G and H). Importantly, inhibition of JNK1 by SP600125 blocked the reduction of apoptotic cell number (Fig. 4I) and apoptotic markers (i.e., cleavage of caspase-3 and PARP) caused by metformin (Fig. 4J). These results suggest that activation of JNK1–Bcl-2 signaling is critical for AMPK to induce autophagy to protect against apoptotic cell death.

We next determined if AMPK directly caused JNK1 phosphorylation. As depicted in Fig. 5A, metformin increased AMPK phosphorylation in a dose-dependent manner, which was accompanied by an increase in JNK1 phosphorylation. Incubation of purified recombinant AMPK with recombinant JNK1 dose-dependently increased JNK1 phosphorylation (Fig. 5B). Exposure of H9c2 cells to elevated glucose levels, which inhibited JNK1 phosphorylation, disrupted the interaction of AMPK and JNK1. Consistently, the disruption was depressed by metformin treatment, indicating a direct interaction of AMPK with JNK1 (Fig. 5C).

We recently reported that hyperglycemia activated the mammalian target of rapamycin (mTOR) signaling pathway and inhibited cardiac autophagy, and activation of AMPK by metformin was associated with inhibition of mTOR signaling and restoration of autophagy (11). In this experiment, we further determined if mTOR affects the Bcl-2–Beclin1 interaction. As shown in Fig. 5D, inhibition of mTOR did not disrupt the association between Beclin1 and Bcl-2 induced by elevated glucose levels.

Four months after diabetes induction, diabetic mice exhibited significantly higher blood glucose levels than their control counterparts (469 ± 27 vs. 108 ± 5 mg/dL; P < 0.001, n = 11 per group). Administration of metformin failed to normalize blood glucose (451 ± 28 mg/dL; n = 9) in diabetic mice. In FVB mice, metformin treatment increased AMPK and ACC phosphorylation (Fig. 6A), activated JNK1 and Bcl-2, disrupted the association between Beclin1 and Bcl-2 (Fig. 6B), and elevated LC3-II protein levels (Fig. 6C). Compared with FVB controls, diabetes decreased phosphorylation of AMPK and ACC (Fig. 6A), and the phosphorylation levels were restored with metformin treatment (Fig. 6A). Hyperglycemia reduced JNK1 and Bcl-2 phosphorylation and enhanced the interaction between Beclin1 and Bcl-2. Chronic metformin stimulated JNK1 and Bcl-2, leading to dissociation of Beclin1 and Bcl-2 (Fig. 6B). As a result, administration of metformin enhanced LC3-II protein levels in diabetic mice (Fig. 6C).

To establish the role of AMPK in the regulation of cardiac autophagy in vivo, we induced diabetes in FVB (WT) and DN-AMPKα2 (DN) transgenic mice. Four months after diabetes induction, the diabetic mice had higher blood glucose than their nondiabetic control subjects (WT: 477 ± 21 vs. 113 ± 2 mg/dL, P < 0.001, n = 8 in each group; DN: 483 ± 15 vs. 125 ± 9 mg/dL, P < 0.001, n = 9 in each group). AMPKα2 activity was inhibited by overexpression of DN-AMPKα2 and STZ-induced diabetes. Diabetes did not cause a further reduction of AMPKα2 activity in DN hearts. Chronic administration of metformin restored AMPK activity in WT STZ mice but this effect was absent in DN STZ mice (Fig. 6D). Although DN mice have a normal cardiac phenotype, their hearts exhibited lower LC3-II levels (Fig. 6E), indicating a reduction in cardiac autophagy. The hyperglycemia reduced LC3-II accumulation in both WT and DN mice (Fig. 6E). Although metformin failed to normalize blood glucose levels in diabetic mice, this treatment restored LC3-II protein levels in WT diabetic mice. Notably, overexpression of DN-AMPKα2 in the heart depressed the protective effect of metformin on autophagy (Fig. 6F). We also determined if activation of AMPK by metformin improves insulin signaling. As shown in Fig. 6G, metformin treatment prevented diabetes-reduced Akt phosphorylation.

We next determined the effect of autophagy induction on cardiac apoptosis in vivo. In FVB mice, metformin treatment had no effect on the number of TUNEL-positive cells and cleavage of caspase-3 and PARP (Fig. 7A–C). TUNEL staining demonstrated that TUNEL-positive cells were seldom identified in control mouse hearts, but numerous TUNEL-positive cells were observed in diabetic mouse hearts. The increase was attenuated by chronic metformin treatment (Fig. 7A and B). Consistently, Western analysis of cardiac homogenates revealed a significant increase in cleavage of caspase-3 and PARP in diabetic mouse hearts, and metformin treatment depressed the observed increase in apoptosis markers in diabetic mice (Fig. 7D). Overexpression of DN-AMPKα2 did not increase these apoptotic markers in nondiabetic mice nor did this mutant exaggerate apoptosis in diabetic DN hearts compared with that in WT hearts (Fig. 7E). Administration of metformin prevented diabetes-enhanced cleavage of caspase-3 and PARP in diabetic WT mice but not in diabetic DN mice (Fig. 7F).

Chronic administration of metformin had no effect on collagen I deposition (Fig. 8A and B) and cardiac functions (Fig. 8C–E) in nondiabetic mice. Compared with control hearts, diabetic hearts exhibited increased deposition of collagen I, and this effect was attenuated by metformin treatment (Fig. 8A and B). LV systolic and diastolic functions were determined by analysis of maximal LV-developed pressure, calculated dp/dt_max (maximal change in pressure per unit time), and dp/dt minimum (dp/dt_min). In comparison with FVB control mice, diabetic mice exhibited a significant reduction in LV-developed pressure (Fig. 8C) and dp/dt_max (Fig. 8D). Chronic metformin restored both LV-developed pressure and dp/dt_max (Fig. 8C and D) to wild-type levels. In addition, dp/dt_min, a parameter that reflects LV diastolic function, was impaired in diabetic mice relative to FVB control mice. This impairment was prevented by chronic administration of metformin (Fig. 8E).

In the current study, we demonstrated that hyperglycemia inhibits JNK–Bcl-2 signaling and promotes the interaction between Beclin1 and Bcl-2. Concomitantly, high levels of glucose induce apoptosis and suppress cardiac autophagy. Activation of AMPK stimulates JNK1, which mediates Bcl-2 phosphorylation and subsequent Beclin1–Bcl-2 disassociation, leading to the restoration of cardiac autophagy and protection against cardiac apoptosis. As a result, cardiac structure and function are improved in diabetic mice. Our findings suggest that activation of JNK1–Bcl-2 signaling and subsequent disruption of the Beclin1–Bcl-2 complex may be an essential mechanism for AMPK to regulate the switch between autophagy and apoptotic cell death pathways under diabetic conditions.

Cardiomyopathy in diabetes is characterized by decreased muscle mass, interstitial fibrosis, and impaired ventricular function. Since adult cardiomyocytes rarely proliferate, the loss of cardiomyocytes would eventually lead to compromised cardiac function. In STZ-induced diabetic mice, hyperglycemia enhanced the number of TUNEL-positive cells and cleavage of caspase-3 and PARP, which is associated with increased deposition of collagen I and impaired systolic and diastolic cardiac function. Chronic administration of metformin, a well-known AMPK activator, protected against apoptotic cell death, resulting in reduced cardiac fibrosis and improved cardiac function. More important, cardiac autophagy was suppressed in diabetic mice. Activation of AMPK by chronic metformin treatment prevented diabetes-suppressed cardiac autophagy. These observations provide strong evidence in support of the hypothesis that cardiomyocyte apoptosis is a critical event in the development of diabetic cardiomyopathy. These findings also suggest that the interplay between autophagy and apoptotic cell death pathways is important in the pathogenesis of diabetic cardiomyopathy.

Although in theory autophagy may lead to cell death through excessive self-digestion and degradation of essential cellular constituents, this process has actually been reported to protect cells from apoptosis (39). For example, increased autophagy promotes cell survival under conditions of nutrient deprivation or growth factor withdrawal through inhibition of apoptosis (40,41). However, a recent study demonstrates that high glucose directly inhibits autophagic flux in neonatal rat cardiomyocytes, and in these cells, the reduction of autophagy appears to be an adaptive response that functions to limit high glucose–induced cardiomyocyte injury (42). Neonatal cardiomyocytes have been reported to behave substantially different from adult cardiomyocytes (43). Especially, autophagy is upregulated in the neonatal cardiac tissue during the perinatal period of relative starvation (44). Thus, autophagy could be either protective or detrimental depending on the cell type and cellular environment. In our study, enhanced autophagy by metformin or overexpression of Atg7 protein attenuated hyperglycemia-induced apoptotic cell death, cardiac fibrosis, and cardiac dysfunction, but these protective actions of metformin are absent in mice deficient in AMPKα2. These data indicate that induction of autophagy by activated AMPK constitutes a protective mechanism against hyperglycemic injury to cardiomyocytes.

The interaction between the antiapoptotic protein Bcl-2 and the autophagy protein Beclin1 has been shown to regulate the switch between the autophagic and apoptotic machinery (38). Beclin1 is part of the class III PI3K complex that is required for the formation of the autophagic vesicle, and interference with Beclin1 prevents autophagy induction (45). Beclin1 is an interacting protein of Bcl-2 (46). Binding of Bcl-2 to Beclin1 inhibits Beclin1-mediated autophagy via sequestration of Beclin1 away from class III PI3K (35,47). Here, we observed a strong interaction between Beclin1 and Bcl-2 in H9c2 cells treated with elevated glucose levels and in hearts from diabetic animals. Treatment of cells and animals with metformin to activate AMPK resulted in disruption of the association between Beclin1 and Bcl-2, and the free Beclin1 bound to class III PI3K to form the kinase complex (9), which is essential for the induction of autophagy. The induction of autophagy degrades and recycles cytoplasmic components and selectively removes damaged mitochondria, serving as a cytoprotective mechanism. In addition, phosphorylated Bcl-2 could preserve the integrity of the mitochondrial outer membrane and prevent proapoptotic proteins from releasing into the cytoplasm (48,49), thus protecting against apoptosis. Therefore, dissociation of Bcl-2 from Beclin1 may be an important mechanism by which AMPK activation restores autophagy and protects against apoptosis under diabetic conditions.

In agreement with the finding that disruption of the association between Beclin1 and Bcl-2 by starvation is dependent on JNK1-mediated phosphorylation of Bcl-2 on multiple residues of the protein (38,50), JNK1–Bcl-2 signaling also regulates the interaction between Beclin1 and Bcl-2 in the development of diabetic cardiomyopathy. Since AMPK can directly phosphorylate JNK1, inhibition of AMPK in the presence of hyperglycemia decreases phosphorylation of JNK1 and Bcl-2, enhances the interaction between Beclin1 and Bcl-2, and suppresses autophagic activity. Activation of AMPK by metformin or overexpression of constitutively active JNK1 stimulated the JNK1–Bcl-2 pathway, leading to dissociation of the Beclin1–Bcl-2 complex and restoration of autophagy. These effects were blocked by inhibition of JNK1. More importantly, suppression of autophagy by inhibition of JNK1 prevented the apoptotic cell death induced by high-glucose levels, indicating that JNK1–Bcl-2 signaling is a crucial pathway involved in AMPK promotion of cardiomyocyte survival under diabetic conditions.

In conclusion, we demonstrated that AMPK activation attenuates diabetic cardiomyopathy through regulation of the switch between autophagy and apoptotic machinery. This effect is attributable to JNK-mediated Bcl-2 phosphorylation and subsequent Beclin1–Bcl-2 disassociation. Our results provide new insight into the regulation of autophagy in the development of diabetic cardiomyopathy and suggest that specific modulation of autophagy, perhaps by stimulation of AMPK, may represent a novel approach for the treatment of heart failure in diabetic patients.

This study was supported by funding from the following sources: the National Institutes of Health (HL-079584, HL-080499, HL-074399, HL-089920, HL-096032, and HL-110488 to M.Z. and 1P20-RR-024215-01 to Z.X. and M.Z.), the American Heart Association Scientist Development Grant (Z.X.), and the Oklahoma Center for the Advancement of Science and Technology (Z.X.). M.H.Z. is a recipient of the National Established Investigator Award of the American Heart Association.

No potential conflicts of interest relevant to this article were reported.

C.H. researched data and prepared the manuscript. H.Z. and H.L. researched data. M.-H.Z. conceived the project and wrote the paper. Z.X. designed the experiments, analyzed data, and wrote the manuscript. Z.X. is the guarantor of this work, had full access to all the data, and takes full responsibility for the integrity of data and the accuracy of data analysis.

The authors thank Dr. Rong Tian (University of Washington, Seattle, WA) for providing DN-AMPKα2 cardiac-specific transgenic mice.