Biochemical Journal

Research article

Activation of autophagy through modulation of 5′-AMP-activated protein kinase protects pancreatic β-cells from high glucose

Diana Han, Byungho Yang, L. Karl Olson, Alexander Greenstein, Seung-Hoon Baek, Kate J. Claycombe, John L. Goudreau, Seong-Woon Yu, Eun-Kyoung Kim

Abstract

Chronic hyperglycaemia is detrimental to pancreatic β-cells by causing impaired insulin secretion and diminished β-cell function through glucotoxicity. Understanding the mechanisms underlying β-cell survival is crucial for the prevention of β-cell failure associated with glucotoxicity. Autophagy is a dynamic lysosomal degradation process that protects organisms against metabolic stress. To date, little is known about the physiological function of autophagy in the pathogenesis of diabetes. In the present study, we explored the roles of autophagy in the survival of pancreatic β-cells exposed to high glucose using pharmacological and genetic manipulation of autophagy. We demonstrated that chronic high glucose increases autophagy in rat INS-1 (832/13) cells and pancreatic islets, and that this increase is enhanced by inhibition of 5′-AMP-activated protein kinase. Our results also indicate that stimulation of autophagy rescues pancreatic β-cells from high-glucose-induced cell death and inhibition of autophagy augments caspase-3 activation, suggesting that autophagy plays a protective role in the survival of pancreatic β-cells. Greater knowledge of the molecular mechanisms linking autophagy and β-cell survival may unveil novel therapeutic targets needed to preserve β-cell function.

  • 5′-AMP-activated protein kinase (AMPK)
  • autophagy
  • high glucose
  • pancreatic β-cell

INTRODUCTION

Diabetes is associated with an increased risk for a number of serious complications. Type 2 diabetes is characterized by impaired function of insulin-producing pancreatic β-cells and insulin resistance in peripheral tissues. The ability to secrete adequate amounts of insulin necessary to overcome peripheral insulin resistance depends on the integrity of pancreatic β-cell function and mass.

Glucose, the main regulator of insulin biosynthesis and secretion, exerts negative effects on β-cell function when present in excessive amounts over a prolonged period. The detrimental effect of excessive glucose concentrations, termed glucotoxicity, is particularly evident by its causative reduction in insulin gene expression in pancreatic β-cells and islets [13] and by causing oxidative stress and ER (endoplasmic reticulum) stress, thereby inducing apoptosis, which negatively affects β-cell mass [2,4]. Therefore a better understanding of the mechanisms leading to β-cell demise is essential for the development of novel approaches to prevent β-cell failure associated with glucotoxicity.

In recent years, the importance of autophagy has been appreciated in the aspects of cell survival and death [5]. Autophagy is a process of bulk degradation of intracellular components. During autophagy, the cell engulfs portions of cytoplasm by a cup-shaped membrane (phagophore), followed by the formation of double-membraned vacuoles called autophagosomes. The autophagosome then fuses with lysosomes, generating acidic single-membrane autophagolysosomes where lysosomal proteases degrade the inner autophagosomal membrane and cargo. Autophagy helps to maintain viability by providing amino acids and other intermediates during starvation, as shown in yeasts and many cancer cell lines [6,7]. The autophagic process has been most intensely studied in yeast [6]. The eleven atg (autophagy-related) genes, atg1, atg310, atg12 and atg16, which were first discovered in yeast, have been identified in mammalian cells [8]. Beclin 1 (Atg6) is a component of a complex with the class III PI3K (phosphoinositide 3-kinase) [9] and promotes the formation of autophagosomes in mammalian cells [10]. LC3 (microtubule-associated protein light-chain 3; also known as Atg8) is the only Atg protein that remains associated with the completed autophagosome and thus serves as one of the few markers for autophagy [11]. LC3 protein is processed to LC3-I through cleavage of the C-terminal amino acid residues. LC3-I in the cytosol is recruited to autophagosome membranes in an Atg5-dependent manner and converted into LC3-II after conjugation with phosphatidylethanolamine by Atg7 [11].

Autophagy is activated as a defence mechanism providing metabolic substrates for cell survival under stress conditions, such as starvation or growth-factor deprivation. On the other hand, a prolonged high level of autophagy may lead to a non-apoptotic autophagic cell death [12,13]. Despite recent expansion of our knowledge about autophagy, we have limited knowledge of molecular components and mechanisms of autophagy in mammalian cells, particularly in pancreatic β-cells.

AMPK (5′-AMP-activated protein kinase) has emerged as a potential target for treatment in Type 2 diabetes [1416]. AMPK is a serine/threonine protein kinase that consists of a heterotrimeric complex composed of a catalytic α subunit, and regulatory β and γ subunits [17], and AMPK activation is reflected by phosphorylation of the α subunit (Thr172) [18]. AMPK is activated by an increase in the intracellular AMP/ATP ratio and via phosphorylation by upstream AMPK kinases [e.g. LKB1 (liver kinase B1) and CaMKKβ (calcium/calmodulin-dependent protein kinase kinase β)] [1921]. AMPK serves as a sensor for the energy status of cells and a metabolic master switch. AMPK activation can occur under metabolic stress conditions, including a change in, or exhaustion of, energy substrate. Another important role of the AMPK action is its participation in the regulation of apoptosis and cell proliferation. AMPK activation in β-cells exerts detrimental effects in vitro and in vivo. In murine MIN6 cells, AICAR (5-amino-4-imidazolecarboxamide riboside), an AMPK activator, induces β-cell apoptosis through a pathway involving c-Jun N-terminal kinase and caspase-3 [22]. In addition, overexpression of AMPK impairs β-cell function in vivo [23]. In contrast, inhibition of AMPK protects β-cells from cytokine-induced apoptosis in MIN6 cells and CD1 mouse islets [24]. AMPK activation suppresses cell proliferation by inactivating the mTOR (mammalian target of rapamycin) pathway [25] and induces cell death by increasing p21CIP, p27KIP and p53 [26] in several different cell types. Previously, the possible involvement of AMPK in autophagy was suggested [27,28]. Under stress conditions, the cyclin-dependent kinase inhibitor p27kip1 is phosphorylated at Thr198 by AMPK, thereby increasing p27 stability and inducing autophagy [27]. It has also been suggested that AMPK induces autophagy by regulating the TSC1/2 (tuberous sclerosis complex 1/2)–mTOR pathway [28]. mTOR is known as a negative regulator of autophagy [29]. mTOR activity can be inhibited by TSC1/2, which is activated by AMPK [25]. Whether AMPK participates in autophagy in pancreatic β-cells has yet to be elucidated.

Autophagy has received attention as an alternative mechanism of cell survival not only by providing intracellular nutrients, but also by eliminating damaged organelles and misfolded proteins [5]. In the present study, we determined the roles of autophagy in the survival of pancreatic β-cells exposed to high glucose and examined whether AMPK regulates autophagy in INS-1 (832/13) cells and rat islets.

EXPERIMENTAL

Cell culture

INS-1 (832/13) cells (generously given by Dr Christopher Newgard, Department of Pharmacology and Cancer Biology, Duke University Medical Center, Durban, NC, U.S.A.) were maintained in INS-1 medium [RPMI 1640 medium containing 11 mM glucose supplemented with 10% (v/v) fetal bovine serum, 2 mM L-glutamine, 10 mM Hepes, 1 mM sodium pyruvate, 0.05 mM 2-mercaptoethanol, 100 units/ml penicillin, and 100 μg/ml streptomycin]. Fetal bovine serum and 2-mercaptoethanol were from HyClone and Sigma, respectively. The other reagents were from Invitrogen.

Pancreatic islet culture

Rat islets were isolated by modifications (collagenase digestion, histopaque gradient and manual handpicking) as previously described [30] from adult male Sprague–Dawley rats (200–250 g; Charles River). Animal experimentation was done in accordance with guidelines on care and use as established by the Michigan State University Institutional Animal Care and Use Committee. Islets were cultured in INS-1 medium containing the indicated glucose concentration.

Antibodies and chemical reagents

Antibodies against phospho-Akt (Ser473), phospho-mTOR (Ser2448), phospho-AMPK (α1/α2:Thr172), AMPK (α1/α2) and activated caspase-3 were from Cell Signaling. Antibodies against Beclin 1 were from BD Biosciences, and antibodies against LC3, β-tubulin, and p62 were from Sigma. Chloroquine, 3-methyladenine, LiCl, Li2CO3, pepstatin A and E64d were from Sigma. AICAR and compound C were purchased from Toronto Research Chemicals and Calbiochem, respectively.

MTS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium] assay

INS-1 cells were plated on to 96-well plates at a density of 1.2×105 cells per cm2 in 11 or 30 mM glucose medium and incubated for up to 72 h. At 24, 48 and 72 h, cell viability was assessed by the ability of metabolically active cells to reduce tetrazolium salt to formazan compounds using MTS reagents (Promega). The absorbance of the samples was measured with a microplate reader at 450 nm wavelength after a 3 h incubation with MTS solution (0.19 mg/ml). Results shown correspond to mean ±S.E.M. values from three independent experiments.

Trypan Blue exclusion assay

Cells were plated on to 12-well plates at a density of 1.2×105 cells per cm2 in 11 or 30 mM glucose medium and collected, and stained with Trypan Blue (Invitrogen). Numbers of total cells and Trypan Blue-stained cells were counted using a haemocytometer.

Measurement of mitochondrial membrane potential

Mitochondrial membrane potential was assessed with TMRE (tetramethylrhodamine ethyl ester; Sigma–Aldrich), a lypophilic fluorescent probe that accumulates in mitochondria in a membrane-potential-dependent manner, following the method of Plotnikov et al. [31] with modifications. Cells were plated on to a 96-black-well plate (Corning) to remove fluorescence interference between measurements. At the end of the incubation time, cells were washed twice with assay buffer (130 mM NaCl, 5.5 mM KCl, 1.8 mM CaCl2, 1.0 mM MgCl2, 25 mM glucose and 20 mM Hepes, pH 7.4) and immediately stained with 100 nM TMRE in assay buffer for 15 min at room temperature (25 °C) in the dark. Stained cells were washed twice with assay buffer and TMRE fluorescence was measured with a fluorescence microplate reader (Fluoroskan Ascent FL; Thermo Fisher) at 538 nm for excitation and 590 nm for emission. CCCP (carbonyl cyanide m-chlorophenylhydrazone) is a decoupler of the mitochondrial respiratory chain and thereby dissipates mitochondrial membrane potential. CCCP was included as a positive control in this assay. The cells cultured in 11 mM glucose were treated with CCCP (10 mM) for 30 min before TMRE staining.

Annexin V/PI (propidium iodide) staining

INS-1 cells were stained with both Annexin V and PI using the Annexin-V-FLUOS staining kit (Roche) in accordance with the manufacturer's instructions. Briefly, cells plated on to 24-well plates at a density of 1.2×105 cells per cm2 were treated with lysosomal proteases or staurosporine and stained with incubation buffer containing Annexin V and PI for 15 min at room temperature after removal of the medium. Apoptotic cells were imaged using a fluorescence microscope (Nikon Eclipse TE2000-U) with an excitation wavelength in the range of 450–500 nm and detection in the range of 515–565 nm. For quantification, nuclear staining was achieved by incubating the cells with Hoechst 33342 (Invitrogen) for 10 min at room temperature.

Measurement of DNA synthesis using BrdU (5-bromo-2′-deoxyuridine) incorporation

The rate of DNA synthesis was evaluated by detection of incorporated BrdU (10 μM; Roche) using a peroxidase-coupled anti-BrdU antibody (100 μl) and quantified in an ELISA reader (Molecular Devices Corporation). Cells (1.2×105 cells per cm2) were plated on to three 96-well plates and collected separately at 24, 48 and 72 h. The average rate of DNA synthesis in INS-1 cells was calculated as the mean of the values measured by BrdU incorporation after 6 h of incubation with BrdU.

Immunoblot analysis

INS-1 cells were plated on to six-well plates at a density of 1.2×105 cells per cm2. Isolated islets were cultured in 60 mm2 plates. Immunoblotting was performed with total protein extracts (15 μg) separated by SDS/PAGE on a 4–15% or 4–20% gradient gel. Antibodies were incubated at an appropriate concentration as recommended by the manufacturers. To assess the equal protein amount, membranes were stripped and reprobed with anti-β-tubulin antibody.

Transfection

Lipofectamine™ 2000 transfection reagent (Invitrogen) was used to transfect INS-1 cells, which were plated on to 24-well plates (0.6×105 cell per cm2) and cultured in 11 mM glucose INS-1 medium for 24 h before transfection. Cells were transiently transfected with 250 ng of GFP (green fluorescent protein)–LC3 plasmid subcloned from LC3 cDNA clone (generously given by Dr Noboru Mizushima, Department of Physiology and Cell Biology, Tokyo Medical and Dental University, Tokyo, Japan). After 24 h of transfection, cells were exposed to 11 or 30 mM glucose for up to 72 h.

GFP–LC3 stable cell lines

A stable INS-1 cell line (GFP–LC3/INS-1) expressing GFP–LC3 from INS-1 (832/13) cells was generated by transfection with GFP–LC3 plasmids subcloned into the pIRESpuro3 vector (Clontech). After a 48 h transfection with Lipofectamine™ 2000, the cells were selected with 1 μg/ml puromycin for 4 weeks. GFP–LC3-expressing colonies were isolated and expanded.

Gene silencing by siRNAs (small interfering RNAs)

INS-1 cells plated on to six-well plates (0.6×105 cells per cm2) were cultured in 11 mM glucose INS-1 medium for 24 h before transfection. Cells were transiently transfected with siRNAs targeting AMPKα1, α2 (5 μM), Atg7 (5 μM) or control siRNAs (5 μM) (Dharmacon) using Lipofectamine™ 2000. After 24 h of transfection, cells were exposed to 11 or 30 mM glucose for up to 72 h. Time courses and dose effects of each siRNA were determined by immunoblot analysis using the corresponding antibodies.

Data analysis

All results were presented as mean±S.E.M. values. Statistical significance was determined by one-way (Tukey test) or two-way ANOVA using Prism software when applicable by comparisons of individual treatments. A two-tailed P value of <0.05 was considered statistically significant.

RESULTS

Chronic exposure to high glucose decreases cell proliferation and induces apoptosis

RPMI 1640 medium containing 11 mM glucose has been used as a standard culture condition for the rat pancreatic β-cell line INS-1 (832/13), which has an insulin content similar to normal islets and exhibits glucose-stimulated insulin secretion [32]. We have adopted 30 mM glucose for hyperglycaemic treatment because this glucose concentration has been shown to reduce insulin mRNA levels [33] and induce ER stress in INS-1 cells through glucotoxicity [34].

To study the cytotoxicity mechanisms of high glucose, we first measured cell viability using the MTS assay, which is dependent on the metabolic activity of viable cells. Interestingly, INS-1 cells cultured in 30 mM glucose for 24 h at a density of 1.2×105 cells per cm2 showed increased viability. After 72 h of culture in 30 mM glucose, however, there was a significant fall (26%) in viability when compared with 11 mM glucose (Figure 1A). When we tested the effects of different densities of plated cells on the MTS assays over 3 days, there was a decrease in viability compared with 11 mM glucose; for example, a decrease of 20% (0.6×105 cells per cm2) and 26% (1.8×105 cells per cm2) at 72 h (results not shown). The decreases in MTS values observed in 30 mM glucose may be due to reduced proliferation or increased cell death. When cell death was assessed by the Trypan Blue exclusion assay, an increased number of dead cells (7.4%) was seen at 72 h in 30 mM glucose (Figure 1B). When DNA synthesis rates were assessed by BrdU incorporation rates, 30 mM glucose exposure led to an increase in BrdU incorporation (15.1%) at 24 h, whereas it led to a decrease (14.9%) in BrdU incorporation at 72 h (Figure 1C). The loss of viability that occurred between 48 and 72 h in cells cultured in 30 mM glucose was associated with activation of the common executioner caspase, caspase-3 (Figure 1D), suggesting a role for apoptosis. Consistent with this, another hallmark of apoptosis, loss of mitochondrial membrane potential, was observed in cells cultured in 30 mM glucose (Figure 1E). A general caspase inhibitor, Z-VAD-FMK (benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone) (40 μM), partially reversed the high-glucose-induced decrease in viability at 72 h (Figure 1F). Since the Akt–mTOR pathway is involved in cell proliferation, we determined whether mTOR was deactivated by high glucose (Figure 1D). Akt-induced phosphorylation at Ser2448 on mTOR was monitored as a measure of mTOR activity. A decrease in mTOR phosphorylation level along with decreased phospho-Akt (Ser473) in high glucose suggests that inactivation of the mTOR pathway may contribute to decreases in β-cell growth or in the DNA synthesis rate following chronic exposure to high glucose. Taken together, these results show that 30 mM glucose initially stimulates INS-1 cell proliferation, but long-term exposure (72 h) to high glucose leads to decreased proliferation and increased apoptosis, which eventually exerts detrimental effects on β-cell function and mass.

Figure 1 Chronic high glucose decreases cell proliferation and induces apoptosis

(A) Cell viability was assessed by MTS assays in INS-1 cells cultured in INS-1 medium containing 11 or 30 mM glucose. Cells were plated on to 96-well plates at a cell density of 1.2×105 cells per cm2 (0 h) and the MTS assay was performed at the time points indicated. Cell viability is shown as a percentage of the MTS values measured in cells cultured in 11 mM glucose medium at each time point (n=9). Statistical significance is represented as ***P<0.001 compared with the corresponding 11 mM glucose at each time point. (B) The Trypan Blue (TB) exclusion assay was performed as a dead-cell measurement. Statistical significance is represented as ***P<0.001 compared with 11 mM glucose (1.2×105 cells per cm2) at each time point (n=12). (C) BrdU incorporation rates in INS-1 cells (1.2×105 cells per cm2 at 0 h) are shown as a relative value of 11 mM glucose at each time point (n=6). **P<0.01 compared with 11 mM glucose at each time point. (D) Protein levels of the cleaved active form of caspase-3 (cas-3), phosphorylated Akt (pAkt), phosphorylated mTOR (p-mTOR) and β-tubulin (tubulin) as an internal control were determined in INS-1 cells cultured in 11 or 30 mM glucose medium. (E) Mitochondrial membrane potential was assessed with TMRE, a lypophilic fluorescent probe that accumulates within mitochondria in a membrane-potential-dependent manner. CCCP (10 μM), a positive control for the dissipation of membrane potential, was added for 30 min to INS-1 cells cultured in 11 mM glucose before TMRE staining. Relative mitochondrial membrane potential was presented as a percentage of the relative fluorescence intensity measured in cells cultured in 11 mM glucose (n=6). **P<0.01 or ***P<0.001 compared with 11 mM glucose. (F) Effects of z-VAD-FMK (40 μM), a general caspase inhibitor, on the MTS assay in INS-1 cells (1.2×105 cells per cm2 at 0 h) following 30 mM glucose exposure (n=9). **P<0.01 compared with the control (30 mM glucose).

Chronic exposure to high glucose activates the autophagy pathway

Accumulating evidence indicates that autophagy plays an important role in cell survival and death in response to cellular stress. Autophagy provides a protective role by removing damaged cytoplasmic components caused by oxidative stress. Therefore we examined whether autophagy occurs in pancreatic β-cells exposed to high glucose. To determine whether autophagic pathways respond to high glucose, we monitored two biochemical markers of autophagy: (i) the protein level of Beclin 1/Atg6; and (ii) the conversion of type I of LC3 (LC3-I) into type II (LC3-II).

Beclin 1 is involved in the early step of autophagosome formation. LC3-II is tightly associated with the autophagosomal membranes and migrates faster (16 kDa) than LC3-I (18 kDa) on SDS/PAGE [35]. Therefore detection of the LC3-II band by immunoblotting can be used to measure the autophagic activity in mammalian cells. When INS-1 cells were cultured in 30 mM glucose, we observed considerable increases in Beclin 1 and LC3-II at 48 and 72 h compared with cells cultured in 11 mM glucose (Figure 2A).

Figure 2 Chronic high glucose activates the autophagy pathway in INS-1 cells and islets

(A) Protein levels of Beclin 1 and LC3 in INS-1 cells cultured in 11 or 30 mM glucose. The LC3-II level can be monitored by immunoblotting using an anti-LC3 antibody after normalization relative to β-tubulin. LC3-I (I) and LC3-II (II) are indicated on the immunoblot (4–15% polyacrylamide gels). (B) GFP–LC3 puncta observed in INS-1 cells transiently transfected with the GFP–LC3 plasmid and cultured in 30 mM glucose at 72 h compared with cells in 11 mM glucose. The numbers of GFP–LC3 puncta were counted among GFP–LC3-positive cells and the relative numbers were shown as a percentage compared with those of puncta in GFP–LC3-positive cells in 11 mM glucose (n=20). Statistical significance is represented as **P<0.01. (C) Immunoblot analysis with pancreatic islets of Sprague–Dawley rats cultured in 11 or 30 mM glucose for 72 h. (D) Treatments with lysosomal protease inhibitors, pepstatin A (20 μg/ml) and E64d (20 μg/ml) (+), increased the accumulation of LC3-II and the activation of caspase-3 (cas-3) compared with non-treated (−) INS-1 cells and pancreatic islets cultured in 30 mM glucose. Immunoblot analysis was performed by SDS/PAGE using 4–20% polyacrylamide gels. (E) Annexin V/PI/Hoechst 33342 (Hoechst) staining was performed in INS-1 cells cultured up to 72 h in 11 mM glucose (11), 30 mM glucose (30), pepstatin A and E64d (20 μg/ml each) in 30 mM glucose (30+P/E), or staurosporine (0.5 μM) in 30 mM glucose (30+S). Staurosporine was included as a prototypic inducer of apoptosis. Lysosomal protease inhibitors and staurosporine were added 8 h and 3 h prior to staining respectively. Nuclear staining was achieved by the Hoechst 33342 dye. For quantification, 350 cells on average were randomly counted for each condition (three independent counts with duplicates). *P<0.01 compared with 11 mM glucose; ##P<0.01 or ###P<0.001 compared with 30 mM glucose.

Not only the increase in LC3-II, but also the change in the intracellular localization of LC3 collectively provide the best method for the detection of autophagy. LC3 tagged at its N-terminus with GFP (GFP–LC3) provides a tool to visualize autophagosome formation by its ring-shaped or punctate fluorescence signal after translocation to autophagosomes, compared with a diffuse cytosolic fluorescence signal in the absence of autophagosome formation [11]. To monitor LC3 puncta formation, INS-1 cells were transiently transfected with GFP–LC3 plasmid constructs and, on the following day, exposed to 30 mM glucose. Transfected INS-1 cells displayed increased GFP-labelled puncta formation in 30 mM glucose at 72 h (277%) compared with 11 mM glucose, demonstrating that autophagy is induced by 30 mM glucose (Figure 2B).

High-glucose-induced autophagy was not limited to INS-1 cells. When primary Sprague–Dawley rat islets were cultured for 72 h in 30 mM glucose, there were increased levels of LC3-II compared with islets cultured in 11 mM glucose (Figure 2C).

Inhibition of lysosomal proteases blocks the degradation of LC3-II and increases the activation of caspase-3

An increased level of LC3-II correlates with increased autophagy. Accumulation of LC3-II at static states, however, could be due to enhanced autophagic flux (on-rate) or reduced autophagic rate through blockage of autophagic degradation (off-rate). These two possibilities can be distinguished by preventing lysosomal degradation of autophagosomes using lysosomal protease inhibitors, since LC3-II is degraded following fusion of autophagosomes with lysosomes. If LC3-II accumulates further in the presence of lysosomal protease inhibitors, this indicates increased autophagic flux. If LC3-II levels remain unchanged, the increase in LC3-II is due to failure of fusion of autophagosomes with lysosomes or impaired autophagic degradation in autolysosomes. We confirmed the autophagic flux by measuring a change in the level of LC3-II using the lysosomal protease inhibitors pepstatin A and E64d (20 μg/ml of each), in INS-1 cells or pancreatic islets of Sprague–Dawley rats cultured in 30 mM glucose (Figure 2D). In both INS-1 cells and islets, inhibition of lysosomal activity led to increased LC3-II, indicating that the increase in LC3-II induced by high glucose reflects increased flux through the autophagic pathway. Interestingly, inhibition of lysosomal activity also resulted in caspase-3 activation (Figure 2D). An increase in apoptosis by lysosomal protease inhibitors was confirmed by the Annexin V/PI fluorescence assay (Figure 2E). Increased staining with Annexin V is indicative of the early stage (Annexin V-positive only) or late stage (Annexin V- and PI-positive) of apoptosis. A negligible population of PI-positive-only cells (demonstrating necrosis) was observed. Pepstatin A and E64d treatment augmented apoptosis in INS-1 cells compared with cells cultured in 30 mM glucose alone. Increased caspase-3 activation and Annexin V-positive cell numbers suggest that autophagy inhibition may accentuate apoptosis.

Inhibition of autophagy diminishes β-cell survival

Autophagy can be inhibited pharmacologically by targeting the class III PI3K involved in autophagosome formation with 3-MA (3-methyladenine) [36] or targeting the autophagosome–lysosome fusion step with lysosomotropic alkalines such as chloroquine [8]. In contrast, activation of autophagy can be achieved with lithium by lowering levels of inositol 1,4,5-trisphosphate, which suppresses autophagy [37]. Pharmacological inhibition of autophagy by 3-MA or chloroquine decreased MTS values more significantly in 30 mM glucose than in 11 mM glucose, whereas the activation of autophagy by LiCl or Li2CO3 increased MTS values in 30 mM glucose (Figure 3A). These results indicate that autophagy plays a protective role in the survival of pancreatic β-cells following high glucose exposure. In addition, this result suggests that a basal level of autophagy is necessary to maintain INS-1 cell viability in 11 mM glucose.

Figure 3 Inhibition of autophagy decreases β-cell survival and increases caspase-3 activation under high glucose

(A) An MTS assay was performed in cells plated at a density of 1.2×105 cells per cm2 and treated with 3-MA (10 mM), chloroquine (CQ; 20 μM), LiCl (0.5 mM) or Li2CO3 (1 mM) in 11 or 30 mM glucose. Statistical significance is represented as *P<0.05, **P<0.01 or ***P<0.001 compared with non-treated control cells at each time point (n=15). (B) Immunoblotting with antibodies against cleaved active caspase-3 (cas-3) and p62 from 3-MA (M)- or LiCl (L)-treated cells and non-treated cells (−) in 30 mM glucose. (C) INS-1 cells transfected with siRNAs targeting Atg7 (A7) or control siRNAs (ctl) for 24 h in 11 mM glucose were exposed to 30 mM glucose for an additional 48 and 72 h. (D) An MTS assay was performed in the siRNA-transfected cells at the same time points as for (C) (n=9).

p62/SQSTM1 (SQSTM1 is sequestosome 1) (p62) mediates the specific recognition of ubiquitinated aggregates by binding to LC3 and is degraded by autophagy [38]. Therefore accumulation of p62 is a good indicator of suppressed autophagy. Accordingly, p62 levels accumulated upon autophagy inhibition by 3-MA, whereas LiCl-induced autophagy led to a decrease in p62 levels (Figure 3B). Of note, inhibition of autophagy by 3-MA increased caspase-3 activation and reduced cell viability (Figures 3A and 3B), suggesting that inhibition of autophagy makes INS-1 cells more susceptible to hyperglycaemic toxicity. Conversely, LiCl decreased caspase-3 activation with an increase in cell viability (Figures 3A and 3B).

Additional studies using Atg7-knockdown with siRNAs indicate that Atg7 gene down-regulation causes increased caspase-3 activation and decreased cell viability (Figures 3C and 3D). Collectively, these results demonstrate that autophagy plays a protective role in the survival of pancreatic β-cells following high glucose exposure.

Inhibition of AMPK enhances high-glucose-induced autophagy

Besides the role of AMPK as a master regulator of cellular metabolism, emerging evidence implicates AMPK in the regulation of autophagy [27,28]. Given this idea, we investigated whether AMPK is critical in the regulation of autophagy in INS-1 cells. We used compound C and AICAR [39] as a pharmacological inhibitor and activator of AMPK respectively. Compound C (6.25 μM) increased the viability of INS-1 cells at 24, 48 and 72 h (Figure 4A). In contrast, AICAR (0.5 mM) significantly decreased cell viability.

Figure 4 Inhibition of AMPK enhances high-glucose-induced autophagy

(A) Effects of AICAR and compound C (C.C) on MTS assays in INS-1 cells (1.2×105 cells per cm2) following 30 mM glucose exposure (n=9). *P<0.05 or ***P<0.001 compared with the control (30 mM glucose) at each time point. (B) Immunoblot analysis with non-treated (−), 0.5 mM AICAR (A)- or 6.25 μM compound C (C)-treated INS-1 cells in 30 mM glucose. (C) Immunoblot analysis for caspase-3 activation (cas-3) with INS-1 cells or pancreatic islets of Sprague–Dawley rats cultured in 11 or 30 mM glucose with non-treatment (−), 0.5 mM AICAR (A) or compound C (C) for 72 h. (D) Immunoblot analysis with pancreatic islets of Sprague–Dawley rats cultured in 30 mM glucose with non-treatment (−), 0.5 mM AICAR (A) or 6.25 μM compound C (C) for 72 h. (E) INS-1 cells were transfected with control siRNAs (ctl) or siRNAs targeting α1 and α2 in 11 mM glucose for 24 h and exposed to 30 mM glucose for an additional 48 and 72 h. pAMPK, phospho-AMPK.

Next, we determined whether inhibition or activation of AMPK affected autophagy flux in high glucose. AMPK activation was assessed by phosphorylation levels of AMPK at Thr172 of the α subunit using immunoblot analysis with phospho-AMPK antibodies. As expected, 6.25 μM compound C and 0.5 mM AICAR decreased and increased the phosphorylation of AMPK respectively (Figures 4B and 4D). Pharmacological inhibition of AMPK led to an increase in LC3-II in INS-1 cells cultured in 30 mM glucose, but had no effect on the level of Beclin 1 (Figure 4B). Consistent with this, compound C increased LC3-II in rat islets cultured in 30 mM glucose without a significant change in Beclin 1 levels (Figure 4D). These results suggest that AMPK inhibition might enhance autophagy by other pathways than up-regulation of Beclin 1 in high glucose. In contrast, AICAR reduced LC3-II without a change in Beclin 1 levels at 72 h in INS-1 and islets (Figures 4B and 4D). Modulation of AMPK changed the levels of caspase-3 activation with more profound effects at high glucose in INS-1 cells and islets, suggesting that the changes in caspase-3 activation might be due to stimulation and inhibition of autophagy by compound C and AICAR respectively (Figure 4C). This is consistent with our finding that inhibition of autophagy increases caspase-3 activation.

To confirm that AMPK negatively regulates autophagy, AMPK activity was modulated using α-subunit-targeted siRNAs (Figure 4E). Immunoblot analyses with α1- or α2-isoform-specific antibodies revealed that α1 was expressed in INS-1 cells, whereas α2 was barely detectable (results not shown). When INS-1 cells were transiently transfected with siRNAs targeting α1 or α2 for isoform-specific knockdown, only α1 siRNA dramatically reduced the total AMPK level and phosphorylation at Thr172 of the α subunit to an undetectable level, suggesting that α1 is the dominant form in INS-1 cells. Consequently, α1 siRNA-mediated suppression of AMPK expression increased LC3-II without a change in Beclin 1 levels, consistent with the result obtained by pharmacological inhibition of AMPK. In addition, α1 siRNA-mediated suppression of AMPK decreased p62 accumulation, indicating an increase in autophagy.

To date, TEM (transmission electron microscopy) is the gold standard for monitoring and counting the formation of autophagic vacuoles, which are the ultrastructural hallmark of autophagy. Compared with INS-1 cells cultured in 11 mM glucose (Figure 5A), an increase in autophagic vacuoles such as autophagosomes (arrowhead) and autolysosomes (arrows) was observed in TEM images 48 h after exposure to 30 mM glucose (Figure 5B). More autophagic vacuoles were observed in compound-C-treated INS-1 cells in 30 mM glucose (Figure 5C). Quantitative analysis by counting the autophagic vacuoles showed that high glucose increases autophagy and compound C stimulates autophagy more significantly (Figure 5D). Collectively, these results indicate that AMPK inhibition enhances autophagy in INS-1 cells.

Figure 5 Inhibition of AMPK increases the formation of autophagic vacuoles

(AC) TEM analysis reveals autophagic vacuoles, such as autophagosomes (arrowheads) and autolysosomes (arrows), in INS-1 cells in (B) 30 mM glucose and in (C) 6.25 μM compound-C-containing 30 mM glucose at 48 h compared with (A) 11 mM glucose. Scale bars 1 μm. (D) To quantify autophagic vacuoles, at least three cells from each section were randomly selected from three sections prepared from cells in 11 mM glucose (11), 30 mM glucose (30) or compound-C-containing 30 mM glucose (30+C). Statistical significance is represented as **P<0.01 compared with 11 mM glucose control; ##P<0.01 compared with 30 mM glucose control (n=9).

Inhibition of AMPK increases GFP–LC3 puncta in GFP–LC3expressing INS-1 cells

To facilitate the monitoring of autophagy and eliminate artifacts resulting from the exposure to transfection reagents, we generated a stable INS-1 cell line (GFP–LC3/INS-1) expressing GFP–LC3 in INS-1 (832/13) cells. Compared with 11 mM glucose, treatment of GFP–LC3/INS-1 cells with 30 mM glucose increased the number of cells containing GFP–LC3 puncta by up to 23% and 36% at 48 h and 72 h respectively (Figures 6A and 6B). Compound C increased further the number of GFP–LC3 puncta-positive cells by up to 42% and 46% at 48 h and 72 h respectively. In contrast, AICAR significantly decreased puncta-positive cell numbers at 72 h. Moreover, compound C significantly increased the numbers of GFP–LC3 puncta within puncta-positive cells compared with cells in 30 mM glucose at 48 h (160%) and 72 h (151%), whereas AICAR significantly decreased those numbers of puncta (30% at 48 h and 15% at 72 h) (Figure 6C). However, in 11 mM glucose there was no significant effect of AICAR on either the number of cells containing GFP–LC3 puncta or the numbers of GFP–LC3 puncta within puncta-positive cells at 48 h and 72h (Figures 6B and 6C; images not shown). In contrast, compound C in 11 mM glucose had a significant effect on the increase in the numbers of GFP–LC3 puncta at both time points (Figure 6C; images not shown). Taken together, these results are best explained by a negative role of AMPK in high-glucose-induced autophagy in INS-1 cells.

Figure 6 Inhibition of AMPK increases GFP–LC puncta formation

(A) A stable cell line (GFP–LC3/INS-1) expressing GFP–LC3 was treated with 0.5 mM AICAR (30+A) or 6.25 μM compound C (30+C) in 30 mM glucose. The formation of GFP–LC3 puncta (arrowheads) was observed in cells in 30 mM glucose and in compound-C-containing 30 mM glucose. The GFP–LC3/INS-1 cells cultured in 11 mM glucose are shown as a control (11). (B) Quantification of GFP–LC3 puncta in GFP–LC3/INS-1 cells. The numbers of cells containing GFP–LC3 puncta were counted individually among 150–200 cells for each experimental condition using a Nikon Eclipse TE2000-U fluorescence microscope. The ratio of counted GFP–LC3 puncta-positive cells to GFP-positive cells is shown as a percentage at 48 h and 72 h. Statistical significance is represented as *P<0.05, ##P<0.01 or ***P<0.001 at each time point. NS, not significant. (C) The numbers of GFP–LC3 puncta were counted among GFP–LC3 puncta-positive cells and the relative numbers are shown as a percentage compared with those of puncta-positive cells in 11 mM glucose at the same time points. Statistical significance is represented as #P<0.05, ###P<0.001, **P<0.01 or ***P<0.001 at each time point as indicated. NS, not significant.

DISCUSSION

The balance between survival and death of pancreatic cells plays a crucial role in the pathogenesis of diabetes. In the present study, we characterized high-glucose-induced toxicity in pancreatic β-cells by means of several complementary assays, including cell viability, proliferation and cell-death assays. Importantly, we showed that chronic exposure of β-cells and pancreatic rat islets to elevated glucose induced autophagy.

Autophagy occurs at a basal level in most cells, and it can be induced by a variety of conditions including nutrient deprivation, growth-factor depletion and hypoxia; however, little is known about the glucose-induced autophagy in pancreatic β-cells. Pharmacological modulation of autophagy demonstrates that autophagy plays a protective role against glucotoxicity in INS-1 cells. High glucose generates oxidative stress, ER stress and protein aggregates. Under hyperglycaemic conditions, it is likely that autophagy constitutes a stress-adaptation pathway that promotes cell survival through the elimination of damaged proteins and organelles. Indeed, it has been shown that oxidative stress caused by high glucose induces ubiquitinated protein aggregates of pancreatic β-cells, which seem to be disposed of by autophagy [40]. It has been suggested that p62 may facilitate autophagy-mediated clearance of protein aggregates by binding to LC3 with proteins containing Lys63-linked polyubiquitin [41]. However, the mechanism underlying autophagy-mediated degradation for ubiquitinated protein aggregates remains unknown. The emerging evidence indicating that induction of autophagy is a protective mechanism against palmitate-induced β-cell death suggests that NEFA (non-esterified fatty acid; ‘free’ fatty acid)-induced oxidative or ER stress mediates the induction of autophagy [42]. Similarly, our results support the hypothesis that elevated glucose can induce autophagic pathways to protect β-cells from subsequent oxidative damage.

Inhibition of AMPK may provide a protective role in the regulation of apoptosis and autophagy in β-cells exposed to high glucose. Conversely, the activation of AMPK in pancreatic β-cells has the opposite effect on apoptosis and autophagy, leading to β-cell demise. Of note, we identified AMPK as a negative regulator of autophagy in INS-1 cells and islets. This finding reverses the known positive role of AMPK in autophagy. AMPK inhibition enhanced autophagy and led to increased cell viability after chronic high-glucose exposure. In addition, knockdown of AMPK activity using siRNAs confirmed the pharmacological effects.

However, mechanisms by which AMPK regulates autophagy in β-cells remain poorly understood. It is likely that AMPK regulates autophagy positively or negatively depending on cell type and context. A recent study shows that mTOR positively regulates autophagy through its negative feedback inhibition of Akt in human cancer cells [43]. It now becomes controversial that the mTOR pathway always functions as a negative regulator of autophagy. It is therefore possible that AMPK inhibits autophagy via inactivation of the mTOR pathway through TSC1/2-complex activation if mTOR plays a positive role in β-cells. Our findings suggest that AMPK might regulate autophagy through Beclin-1-independent pathways, or through modulating the protein interaction with Beclin 1 or downstream of Beclin 1. It has recently been reported that AICAR decreases class III PI3K binding to Beclin 1 [44]. Even though it remains elusive whether the effect of AICAR on class III PI3K is AMPK-dependent, AMPK might serve as a negative regulator of autophagy by inactivating the class III PI3K or interfering with its interaction with Beclin 1 in β-cells. Beclin 1 also interacts with other proteins such as the Bcl-2 family, Vps15 (vacuolar protein sorting-associated protein 15), UVRAG (UV irradiation resistance-associated gene), Atg14L (Atgl4-like protein) and Rubicon [4548]. Therefore, AMPK inhibition might relieve the inhibitory protein interactions or enhance other positive-regulatory protein interactions with Beclin 1 in a manner independent of the up-regulated Beclin 1 level in high glucose.

Recently, it has been shown that pancreatic β-cell-specific Atg7-knockout mice have increased apoptosis and decreased proliferation in β-cells, leading to impaired glucose tolerance and decreased levels of serum insulin [49]. Interestingly, a high-fat diet caused profound deterioration of glucose tolerance in Atg7-deficient mice in a different study [50]. These findings are supportive of the hypothesis that autophagy is essential for the maintenance of normal β-cell mass and function and plays a protective role in the pathogenesis of diabetes. Consistent with these findings, the present study of Atg7 knockdown using siRNAs implicates autophagy as a protective mechanism.

In the present study, we clearly demonstrated that pharmacological or siRNA-mediated inhibition of autophagy resulted in caspase-3 activation, suggesting that impairment or dysregulation of autophagy causes apoptotic cell death. Apoptotic cell death may not be the only mechanism responsible for a decrease in β-cell mass. Prolonged inhibition of cell-cycle progress or cell proliferation may also contribute to β-cell failure. The activation of autophagy may increase β-cell mass by decreasing apoptosis and/or increasing cell proliferation. Future studies on the precise mechanism should address the potential interplay among autophagy, apoptosis and proliferation. It would also be interesting to investigate whether autophagy is regulated functionally and whether AMPK is involved in any possible alterations in autophagy in diabetic animal models or humans in the future study. These lines of direction will lay the groundwork for the systemic analysis of different mechanisms regulating survival of pancreatic β-cells and islets, and should contribute to therapeutic designs for diabetes prevention or treatment.

FUNDING

This work was supported by a grant from the Intramural Research Grant Program sponsored by the Michigan State University [grant number 07-IRGP-1187 to E.-K. Kim].

AUTHOR CONTRIBUTION

Diana Han, Seong-Woon Yu and Eun-Kyoung Kim designed the experiments and analysed data. Diana Han, Byungho Yang and Alexander Greenstein performed the cell viability/death assay, Western blot analysis, puncta assay, generation of the stable cell line and siRNA experiments. Seung-Hoon Baek measured mitochondrial membrane potential and assisted on fluorescence microscopy. Karl Olsen, Kate Claycombe and John Goudreau provided the method and the animals for islet culture. Seong-Woon Yu and Eun-Kyoung Kim wrote the paper, and Diana Han, Karl Olson, Kate Claycombe and Seong-Woon Yu edited the manuscript prior to submission.

Acknowledgments

We thank Mr Ralph Common (Electron Microscopy Facility, Division of Human Pathology, Michigan State University, East Lansing, MI, U.S.A.) for technical assistance for TEM.

Abbreviations: AICAR, 5-amino-4-imidazolecarboxamide riboside; AMPK, 5′-AMP-activated protein kinase; atg, autophagy-related; BrdU, 5-bromo-2′-deoxyuridine; CCCP, carbonyl cyanide m-chlorophenylhydrazone; ER, endoplasmic reticulum; GFP, green fluorescent protein; LC3, microtubule-associated protein light-chain 3; 3-MA, 3-methyladenine; mTOR, mammalian target of rapamycin; MTS, 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium; PI, propidium iodide; PI3K, phosphoinositide 3-kinase; siRNA, small interfering RNA; TEM, transmission electron microscopy; TMRE, tetramethylrhodamine ethyl ester; TSC1/2, tuberous sclerosis complex 1/2

References

View Abstract