Inhibition of AMP-Activated Protein Kinase Protects Pancreatic β-Cells From Cytokine-Mediated Apoptosis and CD8⁺ T-Cell–Induced Cytotoxicity

Collagenase, Histopaque (1,077), and Hoechst (type V) were from Sigma (St. Louis, MO). Mouse recombinant TNF-α, IFN-γ, and IL-1β were from PreproTech EC (London, U.K.). BM Chemiluminescence Blotting Substrate, anti-rabbit and anti-mouse horseradish peroxidase, [³²P]γATP, and chromium sulfate (51) were from Amersham BioSciences (Buckinghamshire, U.K.). Firefly luciferase was from Promega (Madison, WI), and the In Situ Cell Death Detection Kit and tetramethylrhodamine red (TMR Red) was from Roche (Basel, Switzerland). The CD8a⁺ T-cell isolation kit was from MACS (Bergisch Gladbach, Germany). The culture media and cell dissociation buffer were from Invitrogen (Paisley, U.K.). Scintillation fluid Ultima Gold LLT was from Perkin Elmer (Waltham, MA), and the scintillation 96-well plate 1450 MicroBeta counter was from Wallac (Turku, Finland).

Wild-type CD-I mice (20–25 g) and 8- to 12-week-old male NOD mice were used for islet isolation and killed by cervical dislocation immediately before the islet isolation procedure (see cell culture and islet isolation). Insulin-specific T-cell receptor (TCR) transgenic mice were generated from G9C8 cloned T-cells (31), which have previously been shown to have specific reactivity to amino acids 15–23 of the insulin β-chain (32). TCR-α and -β founder lines were intercrossed to produce αβ-TCR transgenic mice (G9.NOD). G9Cα^{−/ −}.NOD mice expressing T-cells monoclonal for the transgenic TCR were generated by crossing the αβ-TCR transgenic mice to NOD.Cα^{−/ −} mice (>20 generations backcross to NOD mice). All animal procedures were in accordance with the British Home Office Animals (Scientific Procedures) Act, 1986.

Rabbit anti–phospho-AMPK (Thr-172) and anti-cleaved caspase-3 (Asp-175) antibodies were purchased from Cell Signaling (Beverly, MA). Rabbit anti-ERK2 was from Santa Cruz (Santa Cruz, CA). Tetra methyl rhodamine isothiocyanate (TRITC)-conjugated secondary antibody against rabbit IgG was purchased from Jackson (West Grove, PA).

Adenoviruses encoding enhanced green fluorescent protein (eGFP) only, hereafter named pAd-GFP (Null), AMPK-CA, or AMPK-DN (AMPKα1 [D¹⁵⁷A]) have been described by Carling and coworkers (33,34). AMPK-CA comprises the NH₂-terminal domain (amino acids 1–312) common to both AMPKα1 and -α2 and is rendered constitutively active by a T¹⁷²D point mutation. Islets and MIN6 cells were infected at a multiplicity of infection of 100 viral particles/cell overnight or 4 h and cultured for 48 h before experiments.

Clonal mouse pancreatic MIN6 β-cells were used between passage numbers 18 and 30 and grown in Dulbecco's modified Eagle's medium containing 25 mmol/l glucose and supplemented with 2 mmol/l l-glutamine, 15% heat-inactivated FCS, 50 μmol/l 2-mercaptoethanol, 100 units/ml penicillin, and 100 g/ml streptomycin. Mouse mastocytoma P815 cells were grown in RPMI-1640 medium, 5% FCS, 2 mmol/l l-glutamine, 50 μmol/l 2-mercaptoethanol, 100 units/ml penicillin, and 100 μg/ml streptomycin. Both were maintained at 37°C in a humidified atmosphere containing 5% CO₂.

Mice were killed by cervical dislocation. Collagenase (1 mg/ml in Krebs Ringer bicarbonate [KRB] medium comprising [in mmol/l]: 120 NaCl, 4.8 KCl, 2.5 CaCl₂, 1.2 MgCl₂, and 24 NaHCO₃ and 1 mg/ml bovine serum albumin, gassed with O₂/CO₂ [95%/5%] to maintain a pH of 7.4) was then injected into the pancreatic duct (3 ml/mouse). The distended pancreas was then placed in a water bath at 37°C for 11 min, and the islets were hand picked. Mouse islets were maintained in RPMI-1640 medium supplemented with 2 mmol/l l-glutamine, 10% FCS, 11 mmol/l glucose, and antibiotics.

For immunocytochemistry, islets were dissociated with cell dissociation buffer before plating the liberated cells onto glass coverslips.

G9Cα^{−/ −}.NOD mice express monoclonal insulin-reactive CD8 T-cells. Cells were extracted from the spleen and activated overnight in the presence of insulin B15–23 peptide, in RPMI-1640 medium containing 5% FCS, 2 mmol/l l-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin, at 37°C in a humidified atmosphere containing 5% CO₂. The day after, activated CD8⁺ T-cells were purified from the splenocytes using a CD8⁺ T-cell isolation kit (MACS) according to the manufacturer's instructions. CD8⁺ T-cells obtained by this method were 90–95% pure.

MIN6 cells or dispersed islets were washed three times with PBS before fixation with 4% (vol/vol) paraformaldehyde in 0.1 mol/l NaH₂PO₄, pH 7.4. Cells were then permeabilized with 0.3% (vol/vol) Triton X-100 for 20 min and incubated for 1 h with primary rabbit polyclonal anti-cleaved caspase-3 antibody, at 1:200 dilution, at 4°C. Primary antibody was revealed using TRITC-conjugated secondary antibody against rabbit IgG (1:500 dilution). Nuclear staining was achieved by incubating the cells in wash buffer containing Hoechst for 10 min at room temperature. Apoptotic cells were imaged either on a Leica SP2 laser-scanning confocal or a Leica SP5 MP/FLIM inverted optics confocal microscope, using a ×40 or a ×63 oil immersion objective with excitation at 488 nm (Ar) and 543 nm (He-Ne) and emission detected at >515 (green, eGFP) and >560 nm (red, TRITC). A terminal deoxynucleotidyl transferase biotin dUTP nick-end labeling (TUNEL) assay was also performed using the in Situ Cell Death Detection Kit, TMR red, to visualize the DNA strand breaks of apoptotic cells by fluorescence microscopy, according to the manufacturer's recommendations. Briefly, the cells were treated as described above for fixation and permeabilization, and the TUNEL reaction mixture was added for 1 h. Imaging was performed with an excitation wavelength range of 520–560 nm, detecting in the range of 570–620 nm. Results were quantified by densitometry.

MIN6 cells or mouse islets were incubated and lysed similarly as for measurements of caspase activity. Whole cellular extract (50 μg) was denatured for 5 min at 100°C in 2% (wt/vol) SDS, 5% mercaptoethanol, and resolved by 10% or 7.5% SDS-PAGE before transferring to PVDF membranes and immunoblotting. Secondary antibodies were revealed using BM Chemiluminescence's blotting substrate. Intensities were measured by digital scanning of gels and quantified using Scion Image (www.scioncorp.com/).

For total ATP assay, MIN6 cells were infected with adenoviruses and incubated with the indicated cytokines in culture medium and then at the given glucose concentrations in KRB medium for 30 min before extraction into perchloric acid (10%, vol/vol). ATP was quantitated in extracts neutralized with Hepes-buffered KOH, using partially purifed firefly luciferase and photon counting, as described before (35).

A chromium release assay was performed similar to that previously described (31). Islets from 8- to 12-week-old NOD male mice were infected with either the Null adenovirus or the AMPK-DN for 48 h. They were then labeled with chromium sulfate (51) for 1.5 h, and washed (Fig. 8A). The purified activated CD8⁺ T-cells were added to the islets for 16 h, with or without added insulin B15–23 peptide (1 μg/ml) at an effector-to-target ratio of 20:1 (assuming 1,000 cells/islet). Cytotoxicity using P815 cells, labeled with chromium sulfate (51) for 1 h and coated with insulin B15–23 peptide, at an effector-to-target ratio of 10:1 was used as a positive control. Results were expressed as the percentage of specific lysis determined as [(cytotoxic release − min)/(max − min)] × 100, where the spontaneous lysis corresponds to the minimal release (min), and the lysis provoked by addition of hydrochloric acid corresponds to the maximal lysis (max).

Results are expressed as means ± SEM of at least three independent experiments. Statistical significance was evaluated using the Student's t test for unpaired comparison with Bonferroni correction as appropriate. P < 0.05 was considered statistically significant.

It has previously been reported that incubation of MIN6 cells (36) or islets (37,38) with different cytokine combinations can lead to the induction of apoptosis. Apoptosis can be prompted by several different mechanisms, but the one usually implicated in islet cell death in type 1 diabetes and allograft transplantation involves cytotoxic cytokines secreted by macrophages and possibly also β-cells. Here, we observed that incubation of MIN6 cells for 12, 24, or 48 h with a combination of the three cytokines, IL-1β, TNF-α, and IFN-γ (Fig. 1), induced the cleavage of caspase-3, indicating the establishment of apoptosis in clonal MIN6 cells.

Previous studies have shown that AMPK activators such as 5-amino-4-imidazolecarboxamide riboside (AICAR) (17) or chronic incubation of MIN6 cells or islets with low concentrations (0–3 mmol/l) of glucose (39) or metformin (40) can lead to β-cell apoptosis. Consistent with these earlier findings, when MIN6 cells were infected with Null/GFP (Null) or the AMPK-CA adenoviruses and cultured for 24–96 h, we observed a fourfold increase in the level of activated caspase-3 after 96 h of incubation with AMPK-CA compared with Null virus-expressing cells (Fig. 2). Thus, AMPK activation alone is sufficient to induce apoptosis in MIN6 cells.

Measured in either MIN6 cells (Fig. 3A) or islets (Fig. 3B), AMPKα phosphorylation at Thr-172 was markedly higher after incubation for 30 min in KRB containing low (0–3 mmol/l) than at elevated (17 mmol/l) glucose concentrations, consistent with previous findings (15). Likewise, incubation for 48 h with a combination of IL-1β, TNF-α, and IFN-γ increased by 15- and 4-fold the level of phosphorylation on AMPK Thr-172 at 17 mmol/l glucose in MIN6 cells and mouse islets, respectively.

To determine whether cytokine-mediated apoptosis required AMPK activation, we used an adenovirus to express AMPK-DN (41). MIN6 cells were infected with the AMPK-DN adenovirus for 48 h and then treated with cytokines for a further 48 h. As expected, the increased phosphorylation of AMPK induced by 3 mmol/l glucose or by cytokines was reduced in cells and islets infected with AMPK-DN (data not shown). Inhibition of AMPK also reduced the pro-apoptotic effect of the cytokines. Thus, apoptosis was reduced about four- to fivefold in MIN6 cells infected with the AMPK-DN adenovirus and treated 48 h with the three cytokines, as shown by TUNEL assay (Fig. 4) or measuring anti-cleaved caspase 3 either in single cells by immunochemistry (where infection with adenovirus could be confirmed on a cell basis by the presence of eGFP; Fig. 5A and B) or in cell populations by Western (immuno-) blotting (Fig. 5C).

To examine the role of AMPK in primary β-cells, mouse islets were dispersed into single cells before infection with adenoviruses. This approach ensured high levels of infection such that >80% of cells were positive as judged by eGFP fluorescence. After treatment of the cells for a further 24 and 48 h with cytokines, apoptosis was assessed by TUNEL assay (Fig. 6A and B). The level of apoptosis was between 60 and 70% after the incubation with cytokines, and a significant decrease was apparent in cells expressing AMPK-DN versus those infected with Null adenovirus. Moreover, activation of AMPK by the AMPK-CA adenovirus during 72 h also led to an increase in apoptosis in primary mouse islets, reaching 70% (Fig. 6A and B). To determine whether caspase-3 was also involved in the latter process, we measured caspase-3 cleavage (Fig. 6C and D). Interestingly, the percentage of cells expressing cleaved caspase-3 reached 20% after 24 h in presence of cytokines. These results confirm the involvement of caspase-3 in cytokine-induced apoptosis of primary β-cells. Nevertheless, this number is less than the percentage of apoptosis measured with the TUNEL assay (Fig. 6A and B), even after 24 h of incubation with cytokines, and thus implies the involvement of other (noncaspase 3–dependent) pathways.

AMPK is an AMP-sensitive enzyme whose activity is expected, at least in large part, to be regulated by changes in intracellular ATP-to-AMP ratio. Given that the adenylate kinase reaction is likely to be at near equilibrium in β-cells, we measured the total cellular ATP content as a guide to ATP-to-AMP ratio (Fig. 7). As anticipated, cellular ATP content was increased as glucose concentrations were elevated, consistent with a lowering of AMP levels and inhibition of AMPK (15,41). Interestingly, when cells were incubated for 12 h to 48 h with the cytokine combination, the cellular ATP content was decreased at both three and 17 mmol/l glucose. These results thus suggest that the action of the cytokines on AMPK is likely to be due to a fall in ATP-to-AMP ratio and LKB1-mediated phosphorylation of AMPKα (13). By contrast, AMPK-DN had no significant impact on the action of the cytokine combination to lower cellular ATP content (Fig. 7).

CD8⁺ T-cell cytotoxicity toward islets probably occurs via a combination of effects including cytokine-mediated killing, Fas/FasL apoptosis, and lysis by perforin/granzymes. To determine whether AMPK might be involved in the cytotoxicity of CD8⁺ T-cells against islets during type 1 diabetes, we used a cytotoxic assay (Fig. 8A) involving insulin B15-23–reactive CD8⁺ T-cells as effectors and islets isolated from NOD mice and infected either with the Null or the AMPK-DN adenoviruses as targets. The insulin-specific T-cells used in this study are low avidity and require a relatively high peptide concentration to stimulate them (42). Although NOD β-cells do express endogenous peptide, the concentration is low. However, β-cells (both wild-type and NOD) also express major histocompatibility complex class I molecules so the added insulin peptide can be presented at their surface. In other words, β-cells have a dual role in this assay, serving both as the targets of insulin-specific cytolytic T-cells and as antigen-presenting cells for these T-cells. Inhibition of AMPK led to a 50% decrease of CD8⁺ T-cell cytotoxicity toward the islets (Fig. 8C). Lysis of B15–23–coated P815 target cells was used as a positive control for CD8⁺ T-cell cytotoxicity (Fig. 8B).

The main aim of the present study was to determine whether AMPK activation may be involved in β-cell death induced by cytokines or cytotoxic T-cells and might therefore represent an interesting therapeutic target to enhance the efficacy of transplantation protocols.

The nature of the immunological effectors that induce apoptosis in β-cells, and may thus lead to β-cell destruction and type 1 diabetes, is still debated. However, involvement of autoreactive T-cells and an inflammatory response in which perforin, granzyme B, FasL, TNF-α, IL-1β, IFN-γ NO, or a combination of all of the above, has been implicated in the destruction of pancreatic β-cells (2,5,6). These molecules are likely to act in synergy to induce apoptotic signaling cascades through a caspase-3–dependant cell death (6,30). Nevertheless, it is also now recognized that endoplasmic reticulum stress, possibly triggered by IL-1α and acting via NO formation and caspase-12 activation (43), as well as stress-activated kinases through reactive oxygen species production and loss of mitochondrial membrane potential, possibly triggered by IFN-γ/TNF-α (36), are also key factors in primary β-cell apoptosis.

In our hands, 12- to 48-h treatment of MIN6 β-cells with the cytokines TNF-α, IL-1β, and IFN-γ led to increased apoptosis as demonstrated by enhanced caspase-3 cleavage (Fig. 1). These findings are in accordance with previous results in rat pancreatic islets (44) and clonal insulin-secreting cells (45). Although still clearly apparent (Fig. 6C and D), cytokine-induced caspase-3 cleavage in primary mouse islet (expected to be largely β-) cells was less markedly increased than TUNEL staining (Fig. 6A and B), suggesting that different apoptotic mechanisms may be triggered in primary cells versus clonal MIN6 cells. Kefas et al. (17) have previously shown that chronic treatment of MIN6 cells or rat islets with low glucose concentrations, or a sustained activation of AMPK with AICAR, also leads to programmed cell death. Correspondingly, our results show an increase in AMPK activity in response to cytokines (Fig. 3) and in apoptosis in MIN6 cells after an infection with a recombinant adenovirus expressing AMPK-CA (Fig. 2).

These changes were matched by significant decreases in ATP content and hence probably increases in AMP level, which are likely to underlie the activation of AMPK (Fig. 7). Nevertheless, it is also conceivable that increases in intracellular free Ca²⁺ concentration, and hence activation of CaMKK1β (11,45), may also contribute.

We demonstrate for the first time that cytokine-mediated β-cell death is inhibited when the activation of AMPK is blocked by overexpression of a dominant-negative form of the enzyme (Figs. 4, 5, and 6). Interestingly, AMPK-DN had no apparent effect on the cytokine-induced changes in intracellular ATP content, indicating that the pro-apoptotic pathways activated in response to AMPK act downstream of the decline in ATP. Several mechanisms could explain the link between AMPK and apoptosis. First, activation of AMPK may cause cell cycle arrest, as observed in prostate cancer and smooth muscle cells (46). This observation is consistent with the fact that the upstream kinase LKB1 (47) is a tumor suppressor mutated in Peutz-Jeghers syndrome. Importantly, recent work demonstrates that AMPK activation induces phosphorylation of the tumor suppressor p53, an event required to initiate cell-cycle arrest in G1 phase (48). While this arrest was reversible in the short term, persistent activation of AMPK led to cell death. Providing alternative mechanisms, Kefas et al. (16) showed that AICAR treatment of β-cells lead to a c-jun NH₂-terminal kinase (JNK) and caspase-3–dependant apoptosis, whereas Jambal et al. and Inoki et al. (49,50) showed an inhibition of protein kinase B and mTOR, respectively, by AMPK and thus an inhibition of the anti-apoptotic pathway and protein synthesis. AMPK activation is also associated with increased mitochondrial superoxide-derived radicals (reactive oxygen species) production and decreased mitochondrial activity (51,52).

There has been much debate on the mode of cytotoxity toward islet β-cells mediated by CD8⁺ T-cells, and it is likely that cytokines, the Fas/FasL system, and granzyme/perforin all play a role in the attack (53). We demonstrate here that AMPK may be involved in cytokine-induced apoptosis, and we also show that the inhibition of AMPK decreases the cytotoxic effect of CD8⁺ T-cells (Fig. 8), although this finding alone is insufficient to pinpoint which aspect of T-cell action was affected. This question will need further elucidation, for example by blocking each of these pathways using cells from mice expressing targeted mutations, or with blocking antibodies. It has recently been shown by Suzuki et al. (54) that ARK5, a novel AMPK catalytic subunit family member of AMPK, whose activation is directly regulated by Akt, leads to the resistance of colorectal cancer cells to Fas-induced apoptosis by negatively regulating procaspase-6. Taken with our own findings, the latter result suggests that activation of different AMPK family members may exert quite distinct effects on cell survival.

In summary, we demonstrate for the first time that AMPK is involved in regulating β-cell apoptosis induced by cytokines and that AMPK is involved in CD8⁺ T-cell cytotoxicity toward NOD mouse islets. These data suggest that inhibition of AMPK activity may enhance the survival of islets in diabetes or after islet transplantation.

This study was supported by a fellowship from ALFEDIAM (France) (to A.R.-C.) and grants to G.A.R. from the Medical Research Council (G0401641) the Wellcome Trust (Programmes 067081/Z/02/Z and 081958/Z/07/Z), and Diabetes UK. G.A.R. was a Wellcome Trust Research Leave Fellow, and F.S.W. was a Wellcome Trust Senior Fellow in Clinical Science.

We thank Drs. Gabriela da Silva Xavier, Isabelle Leclerc, and Richard Smith for discussion and Dr. Sarah Richards, Dr. Laura Parton, Rebecca Rowe, Dr. Magalie Ravier, Stephen Chapman, Alan Leard (Bristol MRC Imaging Facility), and Dr. Martin Spitaler (Imperial College London, Facility for Imaging by Light Microscopy) for their assistance.