Dexamethasone Induces Cell Death in Insulin-Secreting Cells, an Effect Reversed by Exendin-4

Antibodies for poly-(ADP-) ribose polymerase (PARP) (no. 9,542) and BAD (no. 9,292) were from Cell Signaling Technology (Beverly, MA). Antibodies against 155-serine Phospho BAD (sc-12970-R), Bcl-2 (sc-7382), and β-tubulin–horseradish peroxidase (sc-5274HRP) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA), and peroxidase-conjugated anti–guinea pig IgG were from Biomol. Enzymes for RT-PCR were from Qiagen (Hilden, Germany) and Roche Diagnostics (Mannheim, Germany). Cell culture media were purchased from GIBCO (Invitrogen, Karlsruhe, Germany) and FCS from Biochrom (Berlin, Germany). The protein assay was purchased from Bio-Rad Laboratories (München, Germany) and the protein ladder (no. 10748-010) and DNA ladder (no. 15613-011) from Invitrogen (Karlsruhe, Germany). H-89 was purchased from Calbiochem (Merck Biosciences, Bad Soden, Germany). The transferase-mediated dUTP nick-end labeling (TUNEL) assay kit (no. 1 684 795) was purchased from Roche Diagnostics. All other chemicals including 4′6-diamidino-2-phenylindole (DAPI), exendin-4, RU486, dexamethasone, δ-methrin, FK506, caspase-3, and calcineurin assays were from Sigma (Deisenhofen, Germany) and of analytical grade unless otherwise stated.

INS-1 cells (provided by C.B. Wollheim, University of Geneva, Geneva, Switzerland) derived from a rat insulinoma were cultured in HEPES-buffered RPMI-1640 supplemented with 10% FCS, 1 mmol/l HEPES, 1 mmol/l l-glutamine, 1 mmol/l Na pyruvate, 10 μmol/l β-mercaptoethanol, and antibiotics as described previously (6). Mouse islets (from C57BL/6 mice; Charles River Laboratories, Wilmington, MA) were isolated and single cells prepared as described previously using collagenase (Serva, Heidelberg, Germany) (6). They were cultured under the same conditions as INS-1 cells using RPMI-1640 without addition of β-mercaptoethanol. After 2 days, culture cells were treated with dexamethasone, 100 nmol/l dissolved in DMSO, for the indicated time period and proceeded further for DAPI or TUNEL staining as described below.

Mouse β-cells and INS-1 cells, at a density of 2 × 10⁵ cells/ml, were seeded onto poly-l-ornithine–coated glass cover slips in RPMI-1640 medium (6). They were cultured for 2 days and test substances added with fresh medium as indicated. After incubation for the appropriate time period, cells were fixed with ethanol (100%) for 5 min at room temperature and then incubated with DAPI (100 ng/ml ethanol) for 15 min at 37°C. After washing the cells with cold PBS, the morphology of stained nuclei was analyzed under an inverted fluorescence microscope (×40 objective; Nikon, Japan) using an excitation wavelength of 350 nm. Apoptotic nuclei were identified by fractionated and condensed chromatin. Between 100 and 200 cells per condition per experiment were counted and the percentage of apoptotic nuclei determined. For TUNEL staining, cells were fixed and processed as described by the protocol enclosed in the commercial kit.

INS-1 cells were prepared as for DAPI staining and treated with 100 nmol/l dexamethasone for 1 day. Thereafter, cells were loaded with rhodamine-123, 10 μg/ml, for 10 min at 37°C. The fluorescence was excited at 480 nm, and the intensity of the emitted light (filtered through a long-pass filter of 515 nm) was measured using a digital camera and software provided by Till Photonics (Planegg, Germany).

Cells were seeded into culture dishes at a density of 4 × 10⁶ cells/75-cm² dish and cultured for 3 days before adding test substances as indicated. Thereafter, culture medium was removed, and cells were washed once with ice-cold PBS and assayed for caspase-3 activity using a colorimetric assay (colorimetric caspase-3 assay kit; Sigma). Cells were lysed in lysis buffer containing 50 mmol/l HEPES, pH 7.4, 5 mmol/l 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate, and 5 mmol/l dithiothreitol. Nuclei and organelles were removed by centrifugation at 20,000g. The supernatant was frozen in liquid nitrogen, and 50 μg of total cytosolic protein was used to assess cytosolic caspase-3 activity. To this end, cell homogenates were incubated up to 4 h at 37°C with caspase-3 substrate conjugated to the chromophore p-nitroanilide (Ac-DEVD-pNA; 1 mmol/l) in the presence and absence of the inhibitor Ac-DEVD-CHO (0.1 mmol/l). Cleavage of Ac-DEVD-pNA substrate by caspase-3 releases pNA, which was quantified spectrophotometrically at 405 nm using an enzyme-linked immunosorbent assay reader (Biorad Laboratories, München, Germany). The change in optical density is directly proportional to caspase-3 activity. The activity inhibited by the caspase-3 inhibitor was calculated as specific activity.

INS-1 cells were prepared as for measurements of caspase-3 activity and treated with dexamethasone, 100 nmol/l, and RU486, 1 μmol/l, for 1 day. Thereafter, cells were lysed in 50 mmol/l Tris HCl, pH 7.0, supplemented with 0.1 mmol/l CaCl₂, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 0.5 mmol/l phenylmethylsulfonyl fluoride, and 0.5 mmol/l dithiothreitol. Phophatase activity in cytosolic lysate (4–8 μg/200 μl) was measured using 10 μg/ml calmodulin, 10 mmol/l NiCl₂, and 0.9 mg/ml p-nitrophenylphosphate as substrate with and without 100 μmol/l EGTA. Dephosphorylated p-nitrophenylphosphate (p-nitrophenol) was transformed into p-nitrophenolate with NaOH, a yellow-colored compound absorbing at 405 nm. The activity sensitive to EGTA was calculated as PP-2B specific activity.

INS-1 cells, 4–5 × 10⁶ cells/75-cm² culture dish, were cultured for 3 days. Thereafter, cells were treated with test substances as indicated. Culture medium was removed, and cells were lysed in a solution containing 125 mmol/l NaCl, 1% (vol/vol) Triton X-100, 0.5% sodiumdeoxycholate, 0.1% SDS, 10 mmol/l EDTA, 25 mmol/l HEPES, pH 7.3, 10 mmol/l NaPP, 10 mmol/l NaF, 1 mmol/l Na-vanadate, 10 μg/ml pepstatin A, 10 μg/ml aprotinin, and 0.1 mmol/l phenylmethylsulfonyl fluoride. Total cellular protein, 75 μg, quantified by Coomassie blue G staining (Bradford dye assay; Bio-Rad Laboratories) was subjected to SDS-PAGE (8–12%) and transferred to nitrocellulose membranes (Schleicher and Schuell, Dassel, Germany). Blots were blocked for 3 h at room temperature with 5% nonfat dried milk. Blots were incubated overnight at 4°C with the respective antibody against PARP (diluted 1:1,000), Bcl-2 (1:250), BAD (1:1,000), P-BAD (1:250), and β-tubulin (1:2,000). Bound antibody was visualized using anti-mouse or anti-rabbit secondary antibody, dilution 1:1,000, coupled to horseradish peroxidase and developed using enhanced chemiluminescence reagents (Amersham Biosciences, Freiburg, Germany).

INS-1 cells (4–5 × 10⁶ cells) were cultured in 75-cm² culture flasks and treated with dexamethasone (100 nmol/l) and RU486 (10 μmol/l) for 4 h. Cells were lysed, total RNA isolated (Mini kit; Qiagen), and 1 μg transcribed into cDNA using reverse transcriptase M-MuLV (Roche Diagnostics, Roche Applied Science, Mannheim, Germany). Aliquots of cDNA, corresponding to equal amounts of RNA, were used for quantification of mRNA. Specific primers used were for detection of rat Bcl-2 mRNA: left: 5′-GGAGATCGTGATGAAGTAC-3′, right: 5′-TCAGGTACTCAGTCATCCA-3′; for rat BAD: left: 5′-TGAGGAAGATGAAGGGATGG-3′, right: 5′-GCTTTGTCGCATCTGTGTTG-3′; for rat Bax: left: 5′-GGCGAATTGGAGATGAACTG-3′, right: 5′-TTCTTCCAGATGGTGAGCGA-3′.

Changes in mRNAs were quantified using a light cycler system (Roche Diagnostics, Roche Applied Science). PCRs were performed in a final volume of 20 μl containing 2 μl cDNA, 3 mmol/l MgCl₂, 0.5 μmol/l of each primer, 2 μl cDNA Master Sybr Green I mix (Roche Molecular Biochemicals, Mannheim, Germany). The transcript levels of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase of each sample were taken as reference using a commercial primer kit (Search, Heidelberg, Germany). Amplification of the target DNA was performed during 35 cycles, each 10 s at 95°C, 10 s at 68°C (declining by 0.5°C each cycle), and 16 s at 72°C. Melting curve, agarose gel analyses, and product sequences confirmed the specificity of amplified products (not shown). Results were calculated as a ratio of the target versus housekeeping gene transcripts.

Data are provided as means ± SE; n represents the number of independent experiments. All data were tested for significance using ANOVA. Results with P < 0.05 were considered statistically significant.

To examine whether dexamethasone induces cell death of insulin-secreting cells, isolated mouse islet cells were cultured with 100 nmol/l of dexamethasone for 1–4 days. Programmed islet cell death was evaluated by the quantity of DAPI-stained condensed nuclei and TUNEL assay. The number of condensed nuclei was doubled from 1.9 ± 0.2 to 3.9 ± 0.45% (n = 9) after 1-day dexamethasone treatment. A 4-day dexamethasone treatment increased the number of condensed nuclei from 1.8 ± 0.2 to 7.0 ± 0.7% (n = 5; Fig. 1A and B), while under control conditions the number of dead cells did not exceed 2% at any time point. TUNEL staining, a more specific marker for apoptotic cell death, revealed that treatment with dexamethasone (100 nmol/l) for 4 days significantly increased the amount of TUNEL-positive cells (Fig. 1A).

Since the analysis of proteins that are involved in dexamethasone-mediated changes of signal transduction pathways requires the treatment of large quantities of cells, further experiments were performed on the insulin-secreting cell line INS-1. The amount of condensed nuclei in control INS-1 cell culture was almost constant over 4 days (2.0 ± 0.2 to 2.8 ± 0.2%; Fig. 2B). Treatment with dexamethasone (100 nmol/l) increased the number of condensed nuclei twofold after 1 day to sevenfold after 4 days of treatment (to 4.7 ± 0.12 and 13.9 ± 0.6%, respectively; Fig. 2A and B). In parallel, the number of TUNEL-stained cells increased dramatically after dexamethasone treatment but not in control INS-1 cells (Fig. 2A). The effect of dexamethasone was dose-dependent and significant at 10 nmol/l (Fig. 2C). Mifepristone (RU486), 1 μmol/l, an antagonist of the glucocorticoid receptor, abrogated dexamethasone-induced apoptosis (Fig. 2D).

Accordingly, dexamethasone triggered apoptotic cell death in insulin-secreting INS-1 and in mouse islet cells. Apoptotic markers, caspase-3 activity, and the cleavage of the caspase-3 substrate PARP were measured in INS-1 cells. After 2 days of dexamethasone treatment, caspase-3 activity increased threefold (from 30.5 ± 1.5 to 93.2 ± 4.6 μmol cleaved Ac-DEVD-pNA/mg protein), an effect antagonized by 1 μmol/l RU486 (Fig. 3A). Western blot analysis disclosed the significantly increased abundance of cleaved PARP (89 kDa) in INS-1 cell lysates after culture with dexamethasone as compared with control conditions (Fig. 3B). The effect of dexamethasone was inhibited by RU486. The amount of uncleaved PARP (116 kDa, Fig. 3B) remained high, reflecting the high amount (90%) of intact cells during dexamethasone treatment. These data indicate that dexamethasone induces apoptotic cell death in INS-1 cells via activation of the glucocorticoid receptor.

Since activation of the glucocorticoid receptor leads to changes in nuclear transcription, further experiments were performed to analyze the effects of dexamethasone on mRNA levels in INS-1 cells. In other cell systems it has been shown that dexamethasone changes the transcription of proteins of the Bcl-2 family, which are involved in apoptosis (29,30). Thus, in the following the mRNAs of the early anti-apoptotic protein (Bcl-2) were analyzed using semiquantitative real-time PCR. Already after 4 h of treatment with 100 nmol/l dexamethasone, the relative mRNA concentration of the anti-apoptotic marker Bcl-2 was reduced 3.5-fold (from 0.21 ± 0.04 to 0.06 ± 0.04 arbitrary units of mRNA; Fig. 4A). The effect of dexamethasone was blunted by 80% in the presence of RU486. According to Western blotting, this reduction of mRNA was paralleled by a decrease in protein abundance (Fig. 4B).

In a next step, we analyzed whether dexamethasone treatment affects the mitochondrial potential. Since glucose-induced [Ca²⁺]_i oscillations have been observed after dexamethasone treatment, (3,6) it is not surprising that glucose hyperpolarizes the mitochondrial membrane in 72% of control and 83% of dexamethasone-treated cells (Fig. 4C). However, the basal rhodamine-123 fluorescence in relation to the maximal possible depolarization induced by carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone (1 μmol/l) was significantly lower in control than in dexamethasone-treated cells, indicating that mitochondria are depolarized in dexamethasone-treated cells as compared with control cells (Fig. 4C).

In a further series of experiments, we determined expression and phosphorylation of BAD. While Bax mRNA was unchanged by dexamethasone treatment (4 and 24 h), BAD mRNA was significantly increased after 24 h (from 9.9 ± 2.3 to 17.1 ± 3.1 arbitrary units of mRNA; Fig. 5A and B), an effect reversed by RU486. However, the protein concentration of BAD after 1 day’s dexamethasone treatment was not significantly changed (Fig. 5C). Since only dephosphorylated BAD stimulates apoptosis, the amount of phosphorylated BAD was examined. Dexamethasone (100 nmol/l) for 1 day significantly reduced BAD phosphorylation at serine 155, an effect blunted by RU486 (n = 4; Fig. 5D). These results indicate that dexamethasone triggers dephosphorylation of BAD.

It has previously been proposed that glucocorticoids activate calcineurin, which in turn dephosphorylates the apoptotic protein BAD (31). Therefore, we analyzed whether dexamethasone-induced cell death is antagonized by calcineurin inhibitors. Since cyclosporine A does not only inhibit calcineurin but also binds to the mitochondrial transition pore, we used the calcineurin inhibitors FK506 and deltamethrin (32). Indeed, deltamethrin (1 μmol/l) and FK506 (0.1 μmol/l) significantly inhibited dexamethasone-induced cell death by 86 (Fig. 6A) and 44% (Fig. 6B), respectively. Next, we examined the effect of dexamethasone treatment on calcineurin activity. Ca²⁺-dependent phosphatase activity was significantly increased threefold by dexamethasone, an effect blunted by RU486 (by 55%), deltamethrin (by 55%), and FK506 (by 100%) (Fig. 6C).

These data indicate that dexamethasone induced cell death in INS-1 cells at least partially through the activation of the glucocorticoid receptor with subsequent activation of calcineurin and dephosphorylation of BAD at serine 155. The effect favors the interaction of BAD with mitochondria, thus promoting the depolarization of the mitochondrial membrane. The effect is compounded by an inhibitory effect of dexamethasone on the transcript levels of anti-apoptotic Bcl-2.

Earlier studies (27,33–35) indicated that incretins protect islet cells against apoptosis. Thus, we examined whether exendin-4, a stable agonist of the GLP-1 receptor, protects mouse islet cells and INS-1 cells against dexamethasone-induced apoptosis. Exendin-4 (10 nmol/l) did not change the rate of cell death under control condition but significantly reduced dexamethasone-induced cell death by 82% (Fig. 7A and B). In INS-1 cells exendin-4 reduced dexamethasone-mediated cleavage of PARP (Fig. 7C). In addition, this effect was accompanied by a significant reduction of caspase-3 activation by 76% (Fig. 7D). To examine whether PKA is involved in the protective effects of exendin-4, two PKA inhibitors, H89 and KT5720, have been applied. H89 (10 μmol/l) increased cell death about twofold in control INS-1 cells (from 3.1 ± 0.3 to 6.0 ± 0.5%; Fig. 8A). In the presence of H89, exendin-4 did not reverse cell death in control cells nor did it protect against dexamethasone-induced cell death. Comparable results were obtained with the PKA inhibitor KT5720 (5 μmol/l; Fig. 8B). High concentrations of forskolin, 10 μmol/l, mimicked the protective effect of exendin-4. The effect was, again, reversed by PKA inhibition (Fig. 8C). To rule out whether GEF might be involved in exendin-4 effects, we used 8CPT-Me-cAMP (50 μmol/l) for specific activation of GEF. Since the agonist did not affect dexamethasone-induced cell death (Fig. 8D), the stimulation of GEF seems not to mediate anti-apoptotic effects of exendin-4. These results further suggest that inhibition of PKA, which itself induces cell death, antagonizes the protective effect of exendin-4.

This study demonstrates that dexamethasone induces cell death in insulin-secreting INS-1 cells and mouse β-cells through activation of the glucocorticoid receptor. Cell death is accompanied by caspase-3 activation and increased accumulation of cleaved PARP, a caspase-3 substrate, indicating that apoptosis is activated by dexamethasone. This conclusion is further supported by the increased amount of TUNEL-positive cells after dexamethasone treatment. Interestingly, glucocorticoids have been described to induce apoptosis in prolactin-stimulated rat islets, a mechanism that might downregulate β-cell function during late pregnancy (36).

The reduction of the antiapoptotic protein Bcl-2, the stimulation of calcineurin, and the dephosphorylation of BAD indicate that dexamethasone activates a mitochondrial pathway of apoptotic cell death. Dexamethasone-induced cell death was indeed inhibited by calcineurin inhibitors. Furthermore, activation of PKA by GLP-1 receptors counteracted dexamethasone-induced cell death.

The Bcl-2 protein family comprises anti- and proapoptotic proteins, which are involved in programmed cell death. Some of these proteins (such as Bcl-2 and Bcl-XL) are antiapoptotic and located at the outer mitochondrial membrane, while others (such as BAD or Bax) are proapoptotic and cytosolic proteins. The proapoptotic proteins act as sensors of cellular damage or stress. BAD is bound to the cytosolic adaptor protein 14-3-3 in its phosphorylated form (Fig. 9). Cellular stress induced by irradiation, viral infection, or chemicals, e.g., dexamethasone, may result in an increase of cytosolic Ca²⁺ activity, which in turn activates the phosphatase calcineurin. Calcineurin is known to dephosphorylate BAD, which triggers its release from the cytosolic protein 14-3-3 (37). This enables BAD to bind to Bcl-2 at the surface of the mitochondrial membrane. The interaction between BAD as pro- and Bcl-2/Bcl-XL as antiapoptotic protein at the mitochondrial membrane disrupts the normal function of the antiapoptotic Bcl-2 proteins. The binding of BAD leads to the formation of permeability transition pores in the mitochondria and the release of cytochrome c and other proapoptotic molecules from the mitochondrial intermembrane space (37,38). This in turn leads to the formation of the apoptosome and the activation of the caspase cascade (39). Active cytosolic caspase-3 is a key enzyme in the apoptotic process, which accomplishes the degradation of nuclear DNA (40,41).

Dexamethasone-mediated cell death is known to activate the mitochondrial apoptotic pathway (16–20,42). However, dexamethasone has been found to have opposing effects in distinct cells. In granulosa cells, dexamethasone protects against serum deprivation–and cAMP-induced apoptosis and increases expression of Bcl-2 (43). In contrast, in LTR-6 cells (mouse M1 myeloid leukemia cells) dexamethasone induces cell death and reduces Bcl-2 expression. In thymocytes, lymphoid cells and airway epithelial cells apoptosis by dexamethasone is blocked by the overexpression of Bcl-2 (42,44,45). Indeed, the expression of Bcl-2 seems to determine the pro- or antiapoptotic effect of glucocorticoids (45). Thus, one prerequisite for dexamethasone-induced apoptosis might be the downregulation of Bcl-2 protein. In insulin-secreting cells, Bcl-2 overexpression has been described to protect against cytokine-induced cell damage (46). We show here that Bcl-2 expression is reduced by dexamethasone in INS-1 cells (Fig. 4A and B). The effect is antagonized by the glucocorticoid receptor antagonist RU486 and, thus, mediated by the glucocorticoid receptor. Activated glucocorticoid receptor enters the nucleus and modifies gene transcription by binding to glucocorticoid response element sites (47).

Glucocorticoid receptor activation by glucocorticoids results in release of HSP90 from the glucocorticoid receptor complex (48). HSP90 has been described to activate calcineurin (31). Calcineurin activation involves increases in [Ca²⁺]_i and Ca²⁺-calmodulin–mediated activation of the enzyme independent of transcriptional changes (49). Dexamethasone stimulated calcineurin activity in INS-1 cells (Fig. 6C). The increased phosphatase activity was sensitive to chelating Ca²⁺ with EGTA and to FK506, a specific blocker of calcineurin activity (50). The effect of dexamethasone on calcineurin is reversed by the glucocorticoid receptor antagonist RU486. Thus, dexamethasone may activate calcineurin by a factor released from the glucocorticoid receptor complex, possibly HSP90.

Dexamethasone rather blunts glucose-induced [Ca²⁺]_i oscillations in insulin-secreting cells, making it unlikely that dexamethasone increases the [Ca²⁺]_i activity (3,6). The treatment of dexamethasone has been performed in cell culture in the presence of 11 mmol/l glucose, a glucose concentration that induces [Ca²⁺]_i oscillations in INS-1 cells. This condition mimics hyperglycemia induced by hepatic glucose mobilization during glucocorticoid treatment. Since inhibition of calcineurin counteracted dexamethasone-induced apoptosis in INS-1 cells (Fig. 6A), it can be concluded that glucocorticoid receptor–dependent activation of the enzyme is crucial for the apoptotic effect of dexamethasone.

In other cells, BAD and Bax were found to have an increased expression or a shift of the cellular location from a soluble to a membrane-bound form during dexamethasone-induced apoptosis (19,42). Three different phosphorylation sites of BAD have been described. Protein kinase B phosphorylates at serine 136, PKA at serine 155, and mitogen-activated protein kinase–AP-K1 at serine 112, preventing release of BAD from the chaperone protein 14-3-3 (28). Our data suggest that dexamethasone promotes dephosphorylation of BAD at serine155, while incretins might stabilize phosphorylation via activation of PKA in insulin-secreting cells.

Specific PKA-induced phosphorylation of BAD on serine 155 has been demonstrated to inhibit binding of Bcl-2 (28). We show here that the protective effect of exendin-4 against dexamethasone-induced apoptosis is antagonized by PKA inhibitors, which themselves promote apoptosis. However, unspecific effects of those inhibitors cannot be ruled out, and further experiments are needed to analyze the role of PKA-dependent phosphorylation of BAD, cAMP/PKA-dependent activation of Bcl-2 transcription via cAMP response element–binding protein, and cAMP-dependent activation of Akt/protein kinase B, which have been shown to be involved in incretin-mediated protection against apoptosis (23,24).

Our observations suggest that glucocorticoids may affect normal β-cell function not only by inhibition of insulin secretion but also by promotion of apoptosis, which may in the long term reduce β-cell mass (9,51). Calcineurin activation might play a key role in this process. The observed effects may well be involved in the development of steroid-induced diabetes, and GLP-1 receptor activation could be used for therapeutic protection.

This study was supported by the Deutsche Forschungsgemeinschaft, grants La 315/6-1, La 315/11-1/2, and LA 315/13-1, and a grant from the Tübingen fortüne program 1346-0-0.

The authors acknowledge the skilled technical assistance of U. Schueler, E. Faber, and S. Heuser and the preparation of the manuscript by T. Loch.