cFLIP Protein Prevents Tumor Necrosis Factor-α–Mediated Induction of Caspase-8–Dependent Apoptosis in Insulin-Secreting βTc-Tet Cells

CDM3D cells are βTc-Tet cells (38) that have been modified to overexpress Bcl-2 (39). They were grown in Dulbecco’s modified Eagle’s medium (Life Technologies, Basel, Switzerland) containing 25 mmol/l glucose and supplemented with 15% horse serum (Amimed; BioConcept, Allschwil, Switzerland), 2.5% fetal bovine serum (Life Technologies), 10 mmol/l HEPES, 1 mmol/l Na-pyruvate, and 2 mmol/l glutamine, at 37°C with 5% CO₂.

The human cFLIP cDNA, provided by Dr. J. Tschopp (Department of Biochemistry, University of Lausanne) (29), contains an octapeptide tag (FLAG) epitope at the 5′ end. It was subcloned into a modified SIN-18/phosphoglycerate kinase/woodchuck hepatitis virus (SIN-18-PGK-WHV) vector (42,43), which contains a neomycin resistance gene downstream of an internal ribosome entry site from encephalomyocarditis virus, provided by Dr. N. Déglon (University Hospital, Lausanne, Switzerland). High-titer stocks of lentiviral vectors carrying the PGK-driven lacZ or cFLIP genes packaged by the multiply attenuated lentivirus CMVΔR8.91, and pseudotyped with the vesicular stomatitis virus-G envelope protein (plasmid pMD-G), were prepared by transient transfection of 293T cells as described (39,44,45). cFLIP and lacZ virus titers were determined by p24 enzyme-linked immunosorbent assay (ELISA) (NEN Life Science Products, Boston, MA) according to the manufacturer’s instructions. CDM3D cells were transduced with a multiplicity of infection of 10–20. Selection of the pool of infected cells was initiated 48 h after infection by adding 800 μg/ml G418 (an analog of neomycin) for 1 week, followed by 400 μg/ml of the drug for an additional week. At the end of the selection period, all surviving and growing cells expressed the transgene in various amounts.

Total RNA was isolated and analyzed by Northern blot as described previously, using specific probes prepared by random-primer labeling (46). Densitometry scanning of the blots was performed using the Image FX phosphorimager (Bio-Rad). Cell lysates were immunoassayed as previously described (47), and the cFLIP protein level was determined in cell lysates prepared in 5% SDS, 5 mmol/l EDTA, and 80 mmol/l Tris (pH 6.8), with 50 μg/ml phenylmethylsulfonyl fluoride and 2 μg/ml aprotinin. cFLIP protein was detected in immunoblot using the anti–Flag-M2 mouse monoclonal antibody (Sigma).

Cells were plated in 24-well dishes at a density of 10⁵ cells/well for 48 h before incubation with cytokines. After cytokine exposure for 48 h, cells were then incubated for 1 h in HEPES-buffered Krebs-Ringer bicarbonate buffer (KRBH), pH 7.4, containing 0.5% BSA with 2.8 mmol/l glucose and 200 μmol/l isobutylmethylxanthine (IBMX) (Sigma Chemie, Buchs, Switzerland). The medium was changed and cells were incubated again for 1 h in KRBH/0.5% BSA containing 2.8 or 16.7 mmol/l glucose and IBMX. Secreted insulin was quantitated by radioimmunoassay (Linco Research, Labodia, Yens, Switzerland) as described (39,48). Intracellular insulin was measured in acid-ethanol cell lysates. Briefly, cells were lysed in 250 μl of 75% ethanol and 1.5% concentrated hydrochloric acid. Aliquots of cell lysates were also analyzed for DNA content (49) to normalize the secretion data. Lysates in acid ethanol were neutralized with 1/10 volume of 1 mol/l Na₂CO₃, and DNA content was determined by fluorescence using a Fluoroskan-II microplate fluorometer (Labsystems, Helsinki, Finland) with excitation filter set at 355 nm and emission filter set at 460 nm.

The activation of the caspase-8 protease was determined using a commercially available kit (ApoAlert Caspase-8 colorimetric assay kit; Clontech Laboratories, Palo Alto, CA) according to the manufacturer’s specifications. This method is based on spectrophotometric detection of the chromophore p-nitroanilide (pNA) after cleavage from the protease substrate IETD-pNA, characteristic of the caspase-8 cleavage site. Briefly, crude lysates from cytokine-stimulated or untreated cells were recovered, and total proteins (100 μg) were incubated in the presence of 200 μmol/l IETD-pNA. After 2 h of incubation at 37°C, the absorbance at 405 nm was measured using an ELISA plate reader. Comparison of the colorimetric reaction from cytokine-treated samples with an uninduced control allowed quantification of the increase in protease activity.

Cells were seeded in 24-well dishes at a density of 10⁵ cells/well 48 h before transfection with the indicated plasmids using the LipofectAmine-2000 reagent (Roche Molecular Biochemicals, Rotkreuz, Switzerland). A total of 1 μg DNA was transfected, which consisted of 0.8 μg of a NF-κB-luciferase reporter plasmid (provided by Dr C. Widmann, Institute of Cellular Biology and Morphology, University of Lausanne, Switzerland) or 0.8 μg of an iNOS-luciferase reporter plasmid (piNOS-1002luc), containing 1,002 bp of the rat iNOS promoter linked to luciferase (provided by Dr. D. Eizirik, Diabetes Research Center, Vrij Universiteit Brussel, Brussels, Belgium), and 0.2 μg of a β-galactosidase reporter plasmid (driven by the cytomegalovirus promoter) that was used to correct for transfection efficiency. Seventy-two hours after transfection, cells were stimulated with cytokines for determined times, and relative activity of luciferase and β-galactosidase was determined as described (40).

Two days before induction of apoptosis, cells were seeded in a polylysine-treated 96-well microtiter plate (10⁴ cells/well). The medium was changed and cells were treated or not for 36 h with a combination of TNF-α, IL-1β, and IFN-γ (10³ units/ml each). The percentages of viable and apoptotic cells were assessed as described previously (39). Medium was removed from the wells and replaced with the same volume of medium containing 20 μg/ml Hoechst 33342 (Fluka, Buchs, Switzerland) and 10 μ g/ml propidium iodide (Sigma). After 5 min at room temperature, the cells were examined with an inverted fluorescence microscope with ultraviolet excitation at 340–380 nm. In each experimental condition, at least 500 cells were counted. A control plate was analyzed in parallel to determine spontaneous cell death, which was deduced from the experimental values.

To evaluate whether it was possible to interfere with the TNF-α apoptotic response, we transferred the cFLIP cDNA in βTc-Tet and CDM3D cells using a recombinant lentivirus. Pools of transduced cells were selected in the presence of G418 and cFLIP transcripts, and proteins were detected by Northern (Fig. 1A) and Western (Fig. 1B) blot analysis, respectively. In transiently transfected 293T cells, the full-length 55-kDa form of cFLIP (cFLIP_L), as well as a COOH-terminally truncated form of the recombinant protein (43 kDa), were detected. In transduced CDM3D cells, only the full-length cFLIP was observed. We further assessed whether these cells were still able to secrete insulin in response to glucose. At high glucose concentrations, a three- to sevenfold stimulation of insulin secretion was observed (data not shown; see also Fig. 6).

To investigate whether cytokine-induced apoptosis in mouse β Tc-Tet cells was correlated with caspase activation, and to assess the role of cFLIP in the TNF-α–induced apoptotic pathway, the level of caspase-8 activity was evaluated in CDM3D and cFLIP CDM3D cells exposed for 5 h to TNF-α (10³ units/ml) and the protein synthesis inhibitor cycloheximide (CHX) (0.5 μg/ml) or to a combination of 10³ units/ml of TNF-α, IL-1β, and IFN-γ. Figure 2 shows that in control and lacZ-expressing CDM3D cells, a ∼1.6- to 2-fold increase in the activity of caspase-8 was observed upon treatment with TNF-α/CHX or the three cytokines. In contrast, in cFLIP-expressing cells, cytokine-induced activation of caspase-8 was completely abolished.

Whereas cytokine-induced apoptosis of βTc-Tet cells was partially prevented by expression of Bcl-2, the additional presence of mutant proteins blocking the IL-1β signaling pathway conferred an increased resistance to the induction of cell death (39,40). To address the question of whether interfering with TNF-α signal transduction would also provide an increased protection against cytokine-induced apoptosis, cells were exposed for 36 h to a combination of IL-1β, TNF-α, and IFN-γ (10³ units/ml each). As shown in Fig. 3, whereas viability of CDM3D cells was reduced by 60–70%, viability of cFLIP CDM3D cells was reduced by only ∼35%.

In addition to the protective effect of cFLIP on cytokine-induced caspase-8 activation and apoptosis, we addressed the question of whether cFLIP per se was sufficient to modify the basal transcriptional activity of NF-κB. We therefore evaluated whether the stable expression of cFLIP in βTc-Tet and CDM3D cells would result in an increase in the transcriptional activity of NF-κB. Cells were transiently transfected with a NF-κB-luciferase reporter gene, and luciferase assays were performed 72 h posttransfection. As shown in Fig. 4A, NF-κB–driven luciferase activity was increased by 10-fold in βTc-Tet cells expressing cFLIP and by 2-fold in cFLIP CDM3D cells compared with the respective parental cell lines. We also observed that the basal NF-κB–dependent transcriptional activity was enhanced in CDM3D cells stably expressing Bcl-2 compared with βTc-Tet cells, as reflected by a higher NF-κB–driven luciferase activity. Furthermore, upon 6-h exposure to IL-1β (100 units/ml) and IL-1β plus TNF-α (100 units/ml each), cFLIP CDM3D cells displayed a higher stimulation of NF-κB transcriptional activity compared with cytokine-treated CDM3D cells (Fig. 4B).

We have shown previously (40) that the upregulation of iNOS gene and NO production in our cellular model was dependent on the presence of both IL-1β and IFN-γ, as has also been described in human and mouse primary β-cells, and no additional effect of TNF-α was observed. In addition, we also reported that iNOS expression and nitrite production were completely suppressed by blocking either the IL-1β or IFN-γ intracellular signaling pathway (40,41). To further address the question whether the stable expression of cFLIP may affect the ability of CDM3D cells to induce iNOS expression and NO secretion, the cells were transiently transfected with an iNOS-luciferase reporter gene. The cells were then treated for 6 h with IL-1β (10 units/ml) and IFN-γ (150 units/ml) before measuring luciferase activity. As shown in Fig. 5A, the activation of the iNOS reporter construct (∼1.7-fold) in response to cytokines was similar in cFLIP-expressing cells and CDM3D cells. In agreement with the above results, 19-h exposure to IL-1β combined with IFN-γ resulted in an increase in nitrite production that did not differ between cells expressing cFLIP or not (Fig. 5B).

To determine whether expression of cFLIP would protect cytokine-treated cells against impaired GSIS, secretion experiments were performed with cells exposed to cytokines. Cells were first treated for 48 h with IFN-γ (150 units/ml), TNF-α (10³ units/ml), or a combination of IL-1β (10 units/ml) plus TNF-α (100 units/ml) and IFN-γ (150 units/ml) and then incubated for 1 h at 2.8 mmol/l glucose before being exposed to 2.8 or 16.7 mmol/l glucose for 1 h. Figure 6A shows that insulin secretion, at low and high glucose concentrations, was markedly reduced in CDM3D cells after exposure to IFN-γ alone or the three cytokines and that cFLIP did not prevent this inhibitory effect. At 16.7 mmol/l glucose, exposure of CDM3D cells to a high concentration of TNF-α resulted in a 25 ± 4% decrease in GSIS, whereas no significant reduction was observed in cFLIP-expressing cells. As shown in Fig. 6B, impaired secretion in CDM3D and cFLIP CDM3D cells was correlated with a reduction of intracellular insulin levels, which reached 50–60% following treatment with IFN-γ or the combined cytokines. Similarly, insulin content was decreased by 18 ± 5% in CDM3D cells after exposure to high amounts of TNF-α, and the presence of cFLIP prevented this reduction (Fig. 6B). When secretion was expressed as a percentage of the total intracellular insulin content, the secretion rate was not affected in cytokine-treated cells compared with untreated cells (Fig. 6C).

In this study, we report that exposure of CDM3D cells to TNF-α, in the presence of CHX or IL-1β and IFN-γ, leads to the activation of caspase-8. We show that lentivirus-mediated stable expression of cFLIP in CDM3D cells suppressed TNF-α−induced caspase-8 activation and protected the cells against cytokine-induced apoptosis. Furthermore, cFLIP overexpression is associated with an increase in the basal and cytokine-induced transcriptional activity of NF-κB, whereas it does not affect IL-1β/IFN-γ– mediated iNOS gene upregulation and NO secretion. Finally, the reduction in GSIS induced by TNF-α was prevented by cFLIP but not that induced by the combined action of TNF-α, IFN-γ, and IL-1β, indicating that other cytokine-activated intracellular signaling pathways induce the secretory dysfunction.

We previously observed that the induction of apoptosis in CDM3D cells required the combined action of TNF-α, IFN-γ, and IL-1β (39,40). Here we show that cytokine-induced apoptosis is correlated with the activation of caspase-8, an initiator caspase participating in the TNF-α apoptotic cascade. The major role of caspases in inducing apoptosis is well established (50), and their role in cytokine-dependent apoptosis in β-cells has also been suggested by a number of studies (6,51). For instance, in human and mouse islets, as well as in β-cell lines, IFN-γ on its own induces the upregulation of IL-1–converting enzyme (ICE/caspase-1), and caspase-1 (52) is able to process and activate the effector caspase-3.

TNF-α–dependent activation of caspase-8 in CDM3D cells requires the inhibition of protein synthesis, indicating that short-lived proteins may protect against apoptosis. Alternatively, caspase-8 activation and apoptosis of CDM3D cells can be induced by the action of TNF-α associated with IFN-γ and IL-1β. These results therefore suggest that these two cytokines may participate in the repression of anti-apoptotic proteins or, conversely, in the induction of pro-apoptotic molecules, or both. We recently demonstrated that cytokine-induced apoptosis in mouse βTc-Tet cells could also be suppressed by inactivating the IFN-γ– dependent activation of the Janus kinase-signal transducer and activation of transcription (JAK/STAT) pathway by overexpressing the SOCS-1 signaling inhibitor (41). The signaling pathways activated by TNF-α and IFN-γ and leading to apoptosis are complex and highly interactive. Several reports have described that STAT-1–mediated induction of apoptosis in response to IFN-γ was dependent on the expression of caspases such as caspase-1, -3, and -8 (51,53,54). IFN-γ may also prime the cells to respond to the cytotoxic action of TNF-α in part by inducing caspase-8 but also, possibly, by upregulating STAT-1 and IFN regulatory factor (IRF)-1 (55). On the other hand, TNF-α was shown to activate the JAK/STAT pathway by increasing both JAK and STAT-1 tyrosine phosphorylation (56,57). A role of TNF-α in controlling the JAK/STAT pathway was also supported by data (51) showing that STAT-1–null cells were resistant to apoptosis induced by TNF-α and that reintroduction of STAT-1 restored both TNF-α –induced apoptosis and expression of caspases. A further interaction between the two cytokines’ signaling pathways (58) is through binding of STAT-1 to the TNFR1-TRADD signaling complex, which favors the formation of the DISC required for induction of apoptosis.

If cFLIP acts as an inhibitor of caspase-8 activation, its stable overexpression in CDM3D cells should block TNF-α–induced caspase-8 activation and apoptosis. We indeed demonstrated that the presence of cFLIP prevented cytokine-induced activation of caspase-8 and was correlated with increased resistance to apoptosis. Activation of the NF-κB and apoptotic pathways downstream of the TNF-α receptor depend on FADD and TRADD binding to the receptor death domain. cFLIP interaction with the DISC may therefore favor activation of the other pathway, leading to increased NF-κB activity, and act as an important determinant of cell survival/apoptosis balance. Indeed, we observed that stable expression of cFLIP in βTc-Tet and CDM3D cells resulted in a marked increase (2- to 10-fold) in NF-κB transcriptional activity. Therefore, our results indicate that overexpressed cFLIP may act not only by competing with caspase-8 recruitment and activation at the DISC, but also by triggering the activity of NF-κB. On the other hand, whether cFLIP-dependent activation of NF-κB leads to upregulation of anti- or pro-apoptotic genes is not known. Indeed, NF-κB may have opposite effects on apoptosis in a single cell type, depending on the stimulus (59). However, there is evidence showing that NF-κB participates in pro-apoptotic events in β-cells (60,61). Furthermore, we also showed that the presence of cFLIP correlated with increased IL-1β– mediated transcriptional activity of NF-κB. This is in contrast, however, to the observation that cFLIP did not enhance cytokine-induced iNOS gene transcription and nitrite production. Thus, these results seem to indicate that the fraction of IL-1β– induced NF-κB activity dependent on cFLIP does not have an impact on the regulation of iNOS and NO in CDM3D cells in response to cytokines.

Exposing CDM3D cells to cytokines strongly reduces GSIS and intracellular insulin content (∼60%). We previously observed that this effect could be explained by the sole action of IFN-γ, which could be completely prevented by blocking the JAK/STAT pathway by overexpression of SOCS-1 (41). Upon exposure to high concentrations of TNF-α, a weak inhibitory effect (∼ 20%) on cellular insulin content and secretion was nevertheless observed, which was suppressed in the presence of cFLIP. In primary mouse islets, exposure to TNF-α leads to a 40–50% reduction in intracellular insulin level (7,62). TNF-α may therefore also negatively affect insulin response in mouse β-cells. Although the mechanism by which this occurs is not known, it may possibly involve activation of the JAK/STAT pathway, as described above.

Transplantation of encapsulated, insulin-secreting β-cells may provide the most efficient therapy for type 1 diabetes. In this therapeutic approach, however, the transplanted cells have to face several challenges. First, they are exposed to unfavorable oxygen and nutrient gradients due to high cell density and the presence of an immunoisolation barrier. Second, they need to withstand the action of cytokines released by inflammatory and reactivated immune cells. We previously showed that transfer of the Bcl-2 gene into βTc-Tet (CDM3D) cells conferred protection against stress- and hypoxia-induced apoptosis (39). Stable expression of dominant-negative mutants of MyD88 (40) and overexpression of SOCS-1 (41) blocked the IL-1β and IFN-γ intracellular signaling pathways, respectively, and both increased the resistance to cytokine-induced apoptosis. In addition, blocking the IFN-γ pathway with SOCS-1 prevented decreased cellular insulin content and impaired GSIS. Here we demonstrate that cFLIP abolishes cytokine-induced caspase-8 activation and apoptosis and, in addition, might favor activation of the protective NF-κB pathway.

Together, our data indicate that cytokines can activate distinct pathways to induce β-cell dysfunction and apoptosis. On the one hand, IFN-γ action is sufficient to markedly reduce insulin gene expression and GSIS. On the other hand, cell death requires the combined presence of TNF-α, IL-1β, and IFN-γ, even though blocking either intracellular signaling pathway is sufficient to increase resistance to apoptosis, indicating that these three pathways must converge to induce apoptosis. Transplantation of cells with inhibition of selected cytokine intracellular signaling pathways in autoimmune diabetic mice may allow us to determine the contribution of each pathway in graft rejection. Furthermore, these studies may allow us to evaluate whether genetic modifications, alone or in combination with cellular encapsulation, can permit cell survival and function. These studies will be important in determining cellular parameters required for successful cell transplantation therapy of type 1 diabetes.

This work was supported by Juvenile Diabetes Foundation International Grant 4-1999-844 and Swiss National Science Foundation Grant 31-46958.96 (to B.T.). We thank Dr. J. Tschopp and Dr. K. Burns for providing the Flag-cFLIP cDNA and Dr. C. Widmann for providing the NF-κB-luciferase reporter plasmid.