Proinflammatory Cytokines Activate the Intrinsic Apoptotic Pathway in β-Cells

The clonal rat β-cell line INS-1 (a gift from Prof. Claes Wollheim, Geneva, Switzerland) was cultured at 37°C in a humidified atmosphere of 5% CO₂ in complete RPMI-1640 medium (RPMI-1640 with glutamax [Invitrogen], 11 mmol/l glucose, 50 μmol/l β-mercaptoethanol, 100 units/ml penicillin, and 100 μg/ml streptomycin) supplemented with 10% FBS. When confluent, cells were trypsinized and seeded for culture and experiments. In experiments with INS-1 cells and rat islets, IL-1β (160 pg/ml) and IFN-γ (5 ng/ml) were used, while with human islets IL-1β (2 ng/ml), IFN-γ (100 ng/ml), and TNF-α (100 ng/ml) were used. Recombinant mouse IL-1β was from BD Bioscience PharMingen, and recombinant rat IFN-γ, recombinant human IL-1β, recombinant human IFN-γ, and recombinant human TNF-α were from R&D Systems. The combination and concentrations of cytokines were selected based on their ability to induce a four- to fivefold increase in apoptosis over background levels. The V5 Bax inhibitory peptide VPMLK (University of Calgary Peptide Synthesis, Calgary, Canada, and Calbiochem) was used at 100 μmol/l. FK506 (Tacrolimus) from Sigma was used at 1 μmol/l. In experiments examining the potentially protective effect of a compound, the compound was added 30–60 min prior to cytokines. For all compounds prepared in DMSO, the final concentration of DMSO in the culture media was kept below 0.1%. Appropriate vehicle controls were prepared for all treatments.

Pancreata from 5- to 7-day-old outbred Wistar rats (Taconic) were collagenase digested, and islets were isolated by Percoll gradient purification (GE Healthcare). Five hundred or 1,000 islets were cultured for 5–7 days in complete RPMI-1640 medium with 10% FCS at 37°C in a humidified atmosphere of 5% CO₂ before stimulation in complete RPMI-1640 medium with 0.5% human serum.

Pancreata were retrieved from heart-beating cadaveric donors at the time of multiorgan harvest for transplantation, and islets were isolated as described previously (28). The number of islet equivalents (IEQs) and islet purity were assessed using dithizone (Sigma). Glucose-stimulated insulin secretion was routinely assessed to ensure islet functionality. Isolated human islets (>85% purity) were cultured in CMRL-1066 medium (Life Technologies) containing 10% FBS (Wisent) at 37°C in a humidified atmosphere of 5% CO₂. Medium was changed every other day.

Immunoblotting was performed as described previously (24,28). Briefly, lysate protein concentrations were measured by the Bradford method according to the manufacturer's instructions (Bio-Rad). Equivalent amounts of protein from each condition were immunoblotted as described by the manufacturer (Invitrogen), and the protein of interest was detected by chemiluminescence. Anti–cytochrome c (BD Biosciences) was used at a dilution of 1:1,000; anti–cleaved caspase-3 (Cell Signaling Technology) at 1:250–1,000; anti-Bax (Santa Cruz Biotechnology) at 1:800; anti–cytochrome c oxidase subunit IV (COX4) (BD Biosciences) at 1:500; pBads136 (Biosource) at 1:250; anti–phospho-JNK and total-JNK (Cell signaling) at 1:1,000; anti-actin (Abcam or Chemicon) at 1:10,000 or 1:1.000, respectively; anti-tubulin (Santa Cruz Biotechnology) at 1:1,000; and anti–cleaved caspase-9 (Cell Signaling Technology) at 1:500. Horseradish-peroxidase–conjugated anti-rabbit (1:2,000) and anti-mouse (1:1,000) were purchased from Cell Signaling Technology. Ponceau staining (Fluka biochemika) was used in Fig. 7 E to show equal loading.

To assess caspase-3–like protease activity, cleavage of the colorimetric substrate DEVD-pNA was measured as previously described (28). In short, the supernatant of islet sample lysates (2,000 IEQ) was used for analysis. Following determination of the protein content (Bio-Rad), the release of pNA was analyzed by measuring the absorbance at 405 nm using a Benchmark Microplate Reader (Bio-Rad).

To assess caspase-9 activity, luciferase activity dependent on caspase-9–mediated generation of luciferase substrate was measured. Briefly, 25 rat islets or 1 × 10⁶ INS cells were exposed to cytokines for 24 h. Caspase-Glo reagent (Caspase-Glo; Promega) was added directly to islet cultures or to INS cells following removal of the medium, and light emission was detected after 30 min using a VictorX Light (PerkinElmer).

Aliquots containing 500 IEQ in 500 μl of medium were placed in sterile Eppendorf tubes, and 50 μl of stock 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (5 mg/ml) (Sigma) was added to each sample. The samples were incubated at 37°C for 2 h, washed twice with cold PBS, and lysed with 200 μl of DMSO (Sigma). Aliquots of 100 μl from each sample were transferred to a 96-well plate, and the absorbance was measured at 595 nm using a Benchmark Microplate Reader (Bio-Rad). Four independent samples were analyzed per experiment, and each experiment was repeated at least three times.

For measurement of glucose-stimulated insulin release, cultured human islets (100 IEQ per group in duplicate) were washed with CMRL-1066 and incubated in Hank's buffered saline containing 2.2 mmol/l glucose for two consecutive periods of 60 min at 37°C. Next, islets were incubated for 30 min with 22 mmol/l glucose and then another 30 min with 22 mmol/l glucose and 50 μmol/l 3-isobutyl-1-methylxanthine (IBMX), an inhibitor of cyclic AMP phosphodiesterase. Finally, islets were washed with Hank's buffered saline and incubated for 1 h in 2.2 mmol/l glucose. The supernatants were kept following each incubation and analyzed for insulin content using a commercially available insulin enzyme-linked immunosorbent assay kit (Crystal Chem). Insulin release was normalized to the protein content of the pellet determined using a Bio-Rad Protein Assay Dye Reagent. For measurement of accumulated insulin release, 150 rat islets were cultured in 1 ml of medium. Following 24 h of cytokine exposure, accumulated insulin in the incubation medium was measured as previously described (29).

Cytosolic and mitochondria-enriched fractions were prepared using an ApoAlert Cell Fractionation Kit according to the manufacturer's directions (BD Biosciences). Briefly, islet samples (8,000 IEQ) were resuspended in fractionation buffer (BD Biosciences) and incubated on ice for 10 min. Cells were then homogenized on ice using 100 passes with a dounce tissue grinder. The resulting homogenate was centrifuged at 3,000 rpm for 10 min, and the supernatant was kept. The supernatant was centrifuged at 12,000 rpm for 25 min, and the resulting supernatant was the cytosolic fraction. The mitochondria-enriched pellet was resuspended in 10 μl fractionation buffer. The protein content was determined using a Bio-Rad Protein Assay Dye Reagent.

As a marker for apoptosis, histone/DNA complexes released from the nucleus to the cytosol were measured according to the manufacturer's instructions (Cell Death Detection ELISA^PLUS; Roche). In short, cell lysates were incubated with anti–DNA-peroxidase and anti–histone-biotin and transferred to streptavidin-coated wells. Absorbance was measured after addition of the peroxidase substrate ABTS (2′,2-azino-bis-3-ethylbenzthiazoline-6-sufonate).

Data are presented as means ± SE. Statistical significance was determined by Student's paired t test (on the planned comparisons reported) or by two-way ANOVA with Dunnett's post hoc test (time course experiments) (Figs. 3,A and 5 A), with P < 0.05 as the level of significance.

To investigate if cytokines induce mitochondrial stress in human islets, we investigated the effects of cytokine exposure on mitochondrial cytochrome c release and mitochondrial metabolic activity. We used cell fractionation followed by immunoblotting to observe a movement of cytochrome c from the mitochondrial fraction into the cytosolic fraction after cytokine exposure (Fig. 1,A and B). Immunoblotting for COX4 was used to confirm that the cytosolic fractions were not contaminated with mitochondrial proteins. Cytokine exposure also led to a significant decrease in mitochondria metabolic activity examined with the MTT assay (Fig. 1 C).

To examine if cytokine-mediated mitochondrial dysfunction is mediated by Bax, we used the Bax inhibitory peptide V5. The V5 peptide is derived from the Bax-binding domain of human Ku70 (30). Ku70 binds Bax in the cytosol and prevents the proapoptotic functions of Bax by inhibiting Bax translocation to the mitochondria (31,32). In analogy, V5 binds Bax, inhibits Bax activation (proapoptotic conformational change), and suppresses apoptosis (30,33). To confirm that V5 can efficiently enter cells within an islet, isolated human islets were treated with 5-FAM (6-carboxy-fluorescein)–labeled V5. When treated islets were washed and dispersed into single cells, all cells were found to contain the labeled peptide localized mainly to the cytosol (confocal microscopy) (data not shown). Immunoblot analysis of cytosolic and mitochondria-enriched fractions showed that V5 decreased cytokine-induced cytochrome c release from the mitochondria into the cytosol (Fig. 1,A). V5 was also able to protect human islets against cytokine-induced inhibition of mitochondrial function (Fig. 1 B), whereas a negative control peptide had no effect (Calbiochem) (data not shown). Taken together, these findings demonstrate that inflammatory cytokines induce Bax-mediated mitochondrial dysfunction and cytochrome c release.

Perturbations in mitochondrial function and integrity can lead to defective insulin secretion (34). Overexpression of Bcl-X_L has previously been shown to impair insulin secretion and was associated with altered mitochondrial polarization and Ca²⁺ signaling in response to glucose (35). We therefore examined whether V5 affected glucose-stimulated insulin release in isolated human islets. Freshly isolated islets were cultured for 48 h in the presence of 100 μmol/l V5, and glucose-stimulated insulin release was assessed. V5 had no effect on basal insulin release at low glucose levels or on insulin release in response to high glucose (Fig. 2,A). Addition of IBMX, an inhibitor of cyclic AMP phosphodiesterase, causes elevated intracellular cyclic AMP levels that lead to increased exocytosis of insulin (36). IBMX-induced insulin release was unaffected by V5 treatment (Fig. 2,A). Furthermore, V5 did not affect regulation of insulin secretion following a return to low glucose levels (Fig. 2 A). Therefore, V5 did not adversely affect glucose-stimulated insulin secretion from isolated human islets.

It has been demonstrated previously that cytokines inhibit insulin release (1). Having established that V5 did not affect glucose-stimulated insulin release, we next asked whether V5 affected cytokine-mediated inhibition of insulin release from rat islets. Thus, the effect of cytokines on accumulated insulin in the culture medium released from rat islets during 24 h incubation in the presence or absence of V5 was measured. As seen in Fig. 2 B, the inhibitory effect of cytokines on accumulated insulin release was reduced by V5.

To verify the cytokine-induced activation of caspase-9 in INS-1 cells and rat islets, we used an antibody specific for cleaved caspase-9 and a caspase-9 activity assay. The combination of IL-1β and IFN-γ caused a time-dependent cleavage of caspase-9 in INS-1 cells, which was significant after 12 h of cytokine exposure (Fig. 3,A). In rat islets, caspase-9 was cleaved with slower kinetics than in INS-1 cells (Fig. 3,C). In both INS-1 cells (Fig. 3,B) and rat islets (Fig. 3 D), cleavage of caspase-9 was associated with increased caspase-9 activity. In contrast, we observed that caspase-8 was activated not by the combination of IL-1β and IFN-γ but only when TNF-α was added to this combination (data not shown).

We observed caspase-3 cleavage after 24 h of cytokine exposure in INS-1 cells (Fig. 4,A) and in human islets (Fig. 4,B). To assess if caspase-3 activation depended upon Bax activity, we examined the effect of the V5 inhibitor on caspase-3 cleavage and DEVDase activity, indicative of caspase-3–like protease activity. In both human islets and INS-1 cells, cytokine-induced processing of caspase-3 to its cleaved (active) form was inhibited by pre-exposure to V5 for 1 h (Fig. 4,A and B). Cytokine treatment for 24 h resulted in a nearly fourfold increase in DEVDase activity, and this effect was inhibited by V5 pretreatment (Fig. 4,C). Furthermore, pre-exposure to V5 decreased cytokine-induced oligonucleosomal fragmentation after 24 h—a marker of apoptosis (Fig. 4 D). Taken together, these results suggest that Bax-mediated mitochondrial permeabilization is important in cytokine-induced islet cell death.

To examine the role of Bad in cytokine-induced Bax activation and apoptosis signaling, we investigated the effects of cytokine exposure on Bad Ser136 dephosphorylation. In INS-1 cells, IL-1β plus IFN-γ induced a time-dependent dephosphorylation of Bad Ser136 (Fig. 5,A) without affecting Bad protein expression (data not shown). In rat islets, exposure to IL-1β plus IFN-γ induced Bad Ser136 dephosphorylation (Fig. 5,B) with slower kinetics than in INS-1 cells. We were unable to detect phospho-Bad Ser136 in human islets (data not shown). In line with the fact that Bad acts upstream of Bax activation (3), we found that cytokine-induced Bad Ser136 dephosphorylation was unaffected by V5 (Fig. 5 C).

The Ca²⁺-activated Ser/Thr phosphatase calcineurin dephosphorylates Bad, leading to induction of apoptosis (11). To investigate if IL-1β + IFN-γ–induced dephosphorylation of Bad Ser136 was mediated by calcineurin, we pre-exposed β-cells to the calcineurin inhibitor FK506 1 h before cytokine stimulation for 24 h. As seen in Fig. 6, FK506 inhibited IL-1β + IFN-γ–induced dephosphorylation of Bad Ser136 in both INS-1 cells (Fig. 6,A) and rat islets (Fig. 6 B), suggesting that calcineurin is important in IL-1β + IFN-γ–induced dephosphorylation of Bad Ser136. Since increased intracellular Ca²⁺ concentrations induce calcineurin activity (37), we examined the effect of Ca²⁺ influx induced by KCl. We found that a 24-h exposure to KCl (50 mmol/l) induced dephosphorylation of Bad Ser136 equal to that induced by IL-1β + IFN-γ in INS-1 cells but did not affect IL-1β + IFN-γ–induced Bad Ser136 dephosphorylation (data not shown). These data support a role of calcineurin in Bad Ser136 dephosphorylation.

To investigate if calcineurin-mediated Bad dephosphorylation causes cytokine-induced cell death, we examined whether pre-exposure to FK506 inhibited cytokine-induced β-cell death. In rat islets, FK506 inhibited cytokine-induced cleavage of caspase-3 (Fig. 7,A). Accordingly, FK506 decreased cytokine-induced DEVDase activity in human islets (Fig. 7,B). FK506 also improved mitochondrial function and decreased apoptosis caused by cytokine exposure in human islets (Fig. 7,C and D). To demonstrate that the protective effect of FK506 was not due to inhibition of JNK, as has been observed elsewhere (38), we examined the effect of FK506 on JNK phosphorylation, which is necessary for IL-1β–induced apoptosis (1). As shown in Fig. 7 E, FK506 did not affect cytokine-induced JNK phosphorylation or JNK expression. Taken together, these findings demonstrate that proinflammatory cytokines induce calcineurin-mediated dephosphorylation of Bad Ser136, leading to cell death.

In this study, we demonstrate that cytokines induce 1) dephosphorylation of Bad on Ser136 in INS-1 and rat islet cells, 2) Bax-dependent release of cytochrome c and inhibition of mitochondrial metabolic activity in human islets, 3) cleavage/activation of caspase-9 in INS-1 cells and rat islets and cleavage/activation of caspase-3 in INS-1 cells and rat and human islets, and finally 4) β-cell apoptosis in INS-1 cells and rat and human islets. Taken together, our findings strongly suggest that cytokines induce β-cell apoptosis through the canonical intrinsic mitochondrial pathway.

It has been suggested that cytokine-induced β-cell death is independent of Bax or Bak since siRNA-mediated knockdown of Bax or Bak failed to affect cytokine-induced cell death (39). In that study, the combination of IL-1β and IFN-γ was found to induce cell death in INS-1–derived cell lines and rat islets via nonapoptotic killing, which is in contrast to our present observations. However, the INS-1–derived cells used previously (39) were stably transfected with the human proinsulin gene and were selected for G418 resistance (40), which may well affect the sensitivity to cytokines, as observed by Chen et al. (41). Another discrepancy between the studies is the markedly higher concentrations of cytokines used by Collier et al. (39) than in our study, which may preferentially induce nonapoptotic β-cell killing (e.g., via reactive oxygen species/nitric oxide [NO]), while β-cell apoptosis may primarily be induced at lower concentrations. Thus, Bax could be required for cytokine-induced β-cell apoptosis but not for cytokine-induced nonapoptotic β-cell killing. Further studies are needed to clarify the relative importance of apoptosis versus necrosis, which may depend on the model system used and the particular combination and concentration of proinflammatory cytokines.

A certain functional redundancy exists between Bax and Bak (7). Thus, while we show that inhibition of Bax alone decreased cytokine-induced β-cell death, the combined targeting of Bax and Bak may afford greater protection. In addition to the intrinsic pathway investigated in this study, apoptosis may also be induced by activation of the extrinsic (death receptor induced) pathway through caspase-8–mediated caspase-3 cleavage. In β-cells, TNF-α, but not IL-1β, induces activation of caspase-8 (15,42), indicating that TNF-α is able to activate the extrinsic pathway. In agreement with this, we observed that caspase-8 is activated by TNF-α + IL-1β + IFN-γ but not by IL-1β + IFN-γ only (data not shown). In some cell types, the extrinsic pathway must activate the intrinsic pathway to induce cell death through caspase-8–mediated cleavage of the BH3-only Bcl-2 protein Bid. Bid is required for the potentiating effects of TNF-α on IL-1β + IFN-γ–induced DNA fragmentation in primary mouse islets (43), indicating that the intrinsic pathway is important for TNF-α–mediated β-cell death. Overall, the relative dependence of cytokine-induced cell death on the extrinsic and intrinsic pathways remains unclear, mostly due to an incomplete understanding of how these pathways interact.

Because of its regulation by both prosurvival and apoptotic signals, Bad may act as a survival switch between β-cell death and survival. Thus, in other cell systems, antiapoptotic Akt signaling induces Bad Ser136 phosphorylation and inactivation (13), whereas proapoptotic JNK signaling and increased intracellular Ca²⁺ concentration induce Bad Ser136 dephosphorylation and activation (11,12). This notion may be of special relevance in cytokine-induced β-cell death because cytokines decrease Akt activity (27), induce JNK activity (28), and increase intracellular Ca²⁺ concentration (26). In this study, the FK506 inhibitor is used to abrogate cytokine-induced Bad dephosphorylation mediated by calcineurin. In other cell systems, FK506 inhibits nuclear factor κB–induced expression of inducible NO synthase (44). In rodent islets, l-arginine analogues that block inducible NO synthase activity inhibit cytokine-induced β-cell death (1). However, NO plays only a minor role in cytokine-induced apoptosis in human islets (1), and the molecular mechanism underlying the protective effect of FK506 against cytokine-induced apoptosis in human islets is therefore most likely independent of inhibition of NO synthesis.

Exposure of β-cells to a combination of IL-1β and IFN-γ increases cytosolic Ca²⁺ levels via low-voltage–activated Ca²⁺ channels (26). Antagonists of Ca²⁺ channels prevent the rise in cytosolic Ca²⁺ and reduce cell death in cytokine-treated β-cells (16,26), suggesting a key role of Ca²⁺ in cytokine-induced cell death. Indeed, previous findings suggested that IFN-γ– and TNF-α–induced cell death may be partially mediated by Ca²⁺-activated proteins such as calpain or calcineurin (16). Our findings suggest that calcineurin-mediated dephosphorylation of Bad Ser136 may provide a link between the rise in cytosolic Ca²⁺ and the core death machinery in β-cells upon cytokine exposure.

Calcineurin-based immunosuppressants are known to impair insulin production by blocking calcineurin-mediated activation of the transcription factor nuclear factor of activated T-cells (NFAT), and long-term exposure is associated with decreased β-cell survival (45,,,–49). We have previously suggested that the long-term negative effect of calcineurin-based inhibitors on β-cell survival may be due to the inhibition of autocrine insulin–mediated survival (50). In the current study, we exposed β-cells to FK506 for a relatively short time period (25 h), and the rationale for using FK506 in this study was solely to examine its role on calcineurin-mediated Bad dephosphorylation. Thus, it is entirely possible that the short-term protective effects of FK506 against cytokine-mediated apoptosis are overshadowed by long-term toxicity.

In summary, our findings demonstrate that proinflammatory cytokines induce β-cell apoptosis through the canonical mitochondrial pathway and that this effect is dependent on calcineurin-mediated Bad dephosphorylation and Bax activity. These results enhance our understanding of proapoptotic signaling induced by proinflammatory cytokines in β-cells and may provide the basis for the development of future strategies for intervention in type 1 diabetes and islet graft failure.

This work was supported by the Danish Medical Research Council, the Danish Diabetes Foundation, the Juvenile Diabetes Research Foundation, the Canadian Institutes for Health Research, and the Canadian Diabetes Association. L.R. is a National Scientist supported by the Fonds de la Recherches en Santé du Quebec.

This work was also supported by Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

Parts of this study were presented at the 42nd European Association for the Study of Diabetes Meeting, 2006 (O418), and at the 12th European Cell Death Organization Meeting, 2004 (Poster 1).

H. Foght, A.-S. Hillesøe, T. Drengsgaard, G. Singh, M. Lipsett, S. Hanley, and M. Castellarin are thanked for excellent technical assistance.