Proapoptotic BH3-Only Protein Bid Is Essential For Death Receptor–Induced Apoptosis of Pancreatic β-Cells

All mice used were bred and maintained under specific pathogen-free conditions at the St. Vincent's Institute animal facility (Fitzroy, Victoria, Australia). C57BL/6 and NOD/Lt mice were obtained from the Walter and Eliza Hall Institute animal breeding facility (MuriGen, Victoria, Australia). Bid-deficient (bid^−/−) mice were generated on an inbred C57BL/6 background using C57BL/6-derived embryonic stem cells (17). Homozygous H-2^bm1 RIP-Bcl-2 and NOD.RIP–Bcl-2 transgenic mice (backcrossed ≥10 generations), which express human Bcl-2 in β-cells under control of the rat insulin promoter, have previously been described (19,20). The generation of bax^−/− (21), bak^−/− (22) and bak^−/−bax^+/− (23) mice has previously been described, and those used here had been backcrossed for >10 generations onto a C57BL/6 background. Perforin-deficient OT-I T-cell receptor transgenic mice have been described (24). All animal experiments were approved by the institutional animal ethics committee.

Recombinant murine γ-interferon (IFN-γ) and tumor necrosis factor (TNF)-α were obtained from Genentech (San Francisco, CA) and used at 1–100 units/ml and 1,000–2,400 units/ml, respectively. Human recombinant interleukin-1α from Dr. C. Reynolds (National Cancer Institute, Bethesda, MD) was used at 100 units/ml. IL-1β from R&D systems (Minneapolis, MN) was used at 10–150 units/ml. MegaFasL (APO-O10), provided by Dr. M. Dupuis (Apoxis, Lausanne, Switzerland), was used at 100 nmol/l. N^G-monomethyl-l-arginine (NMMA) (Sigma-Aldrich, St. Louis, MO) and z-Val-Ala-Asp-fluoromethylketone (zVAD.fmk) (Enzyme Systems Products, Livermore, CA) were used at 2 mmol/l and 100 μmol/l, respectively. Cycloheximide (CHX) (Sigma-Aldrich) was used at 5 μg/ml. Staurosporine (Sigma-Aldrich), etoposide (Pfizer, Kalamazoo, MI), tunicamycin (Sigma-Aldrich), and thapsigargin (Calbiochem, San Diego, CA) were used at the concentrations indicated.

Islets were isolated from mice using collagenase P (Roche, Basel, Switzerland) and histopaque-1077 density gradients (Sigma-Aldrich) as previously described (24). Islets were handpicked and cultured overnight at 37°C and 5% CO₂ in CMRL medium-1066 (Gibco products; Invitrogen, Grand Island, NY) supplemented with 100 units/ml penicillin, 100 μg/ml streptomycin, 2 mmol/l glutamine, and 10% fetal calf serum (KS; JRH Biosciences) (referred to below as complete CMRL).

Isolated islets from age-matched male and female mice were cultured overnight and then washed in complete CMRL to remove cellular debris. Islet yields were determined by counting the total number of islets isolated per mouse. For quantification of size, isolated islets were blind scored under 40× magnification with a graduated eyepiece to determine islet diameter. A reference grid was placed under each dish to ensure that each islet was analyzed only once.

The expression of insulin in islet β-cells was determined as previously described (20). Freshly isolated islets were trypsinized, fixed, and permeabilized. Cells were then stained with either guinea pig anti-insulin antibody (diluted 1:1,000; DakoCytomation, Carpinteria, CA) or isotype-matched control guinea pig IgG (diluated 1:2,000; Sigma-Aldrich) followed by biotinylated donkey anti-guinea pig IgG antibodies (diluted 1:400; Jackson ImmunoResearch Laboratories) plus streptavidin allophycocyanin (diluted 1:600; BD Pharmingen, San Diego, CA).

Mice were injected intraperitoneally with 2 mg d-glucose/g body wt. Blood samples were obtained from tail-tip bleeds and blood glucose concentrations determined using Advantage II Glucose Strips with Advantage glucometer (Roche).

The day after isolation, 50–100 uniformly sized islets (excluding very large or necrotic islets) per sample were handpicked into 3.5-cm Petri dishes containing 1.1 ml complete CMRL. Islets were cultured with the appropriate stimuli to induce cell death. For assays with FasL, 50 islets (1.8 × 10⁴ ± 0.3 × 10⁴ cells) ranging in size from 50 to 200 μm were used. For γ-radiation, islets were irradiated at the indicated doses and then washed and cultured. At the end of the culture period, nonattached cells and islets were transferred into polypropylene tubes and washed in PBS. Islets were then dispersed (0.1 mg/ml bovine trypsin [Calbiochem] and 2 mmol/l EDTA in PBS) for 5 min at 37°C. Islets were mechanically dispersed using a pipette, washed in PBS, and recovered in complete CMRL medium for 1 h at 37°C in 5% CO₂. Cells were then washed in PBS and resuspended in 250 μl hypotonic buffer containing 50 mg/ml propidium iodide (Sigma-Aldrich), 0.1% sodium citrate, and 0.1% TritonX-100, which stains nuclear DNA. Cells were then analyzed on a FACSCalibur (Becton Dickinson, Franklin Lakes, NJ) using the FL3 channel. Apoptotic cells were identified by their subdiploid DNA content as previously described (25). DNA fragmentation of untreated cells was always <10%.

Activated perforin-deficient OT-I cytotoxic T-lymphocytes (CTLs) were generated in vitro as previously described (24). Wild type and Bid-deficient islets were pulsed with specific ovalbumin (OVA_257–264) or nonspecific herpes simple virus (HSV_498–505) peptides and used in ⁵¹Cr release assays as previously described (24).

Islets were cultured with IFN-γ, IL-1β, and FasL in complete CMRL. After 3 days, islets were dispersed into single cells using trypsin, and cytochrome c release was measured as previously described (26). This technique relies on the principle that cytochrome c diffuses through cells with mitochondrial outer membrane damage. After plasma membrane permeabilization, diffused cytochrome c leaks out of the cell, resulting in low fluorescence when cells are stained with anti–cytochrome c antibody. Cells with intact mitochondria retain cytochrome c and have high fluorescence after staining (26). Cells were permeabilized and fixed and then stained overnight at 4°C with mouse anti–cytochrome c antibody (Clone 6H2B4, BD Pharmingen) followed by phycoerythrin-conjugated sheep anti-mouse IgG (Silenus, Hawthorn, Australia) in blocking buffer for 1 h at room temperature. Cells were analyzed by flow cytometry on a FACSCalibur.

Islets were cultured with IL-1β and IFN-γ in complete CMRL for 2 days. Islets dispersed with trypsin were allowed to recover in complete CMRL for 1 h before staining with hamster anti-mouse Fas antibodies (Jo2; BD Pharmingen) followed by biotinylated anti-hamster Ig (BD Pharmingen) and phycoerythrin-conjugated streptavidin (Caltag, Burlingame, CA). Dead cells and leukocytes were excluded by staining with propidium iodide (3.3 μg/ml) and anti–CD45-Percp-Cy5.5 (30-F11; BD Pharmingen) respectively. Cells were analyzed on a FACSCalibur.

Analyses of data were performed using GraphPad Prism (GraphPad Software, San Diego, CA). Data are represented as means ± SE. Data were analyzed by one-way ANOVA with Bonferroni's posttest for comparison of multiple columns. ⁵¹Cr release assays were analyzed with paired t tests.

Islet development and function were studied in Bid-deficient mice. We observed no significant difference in the islet yields from Bid-deficient and wild-type mice (Fig. 1A), and islets from both sets of animals displayed a similar size distribution (Fig. 1B). Moreover, intracellular insulin expression, measured by flow cytometry, was similar in freshly isolated islets from Bid-deficient and wild-type mice (Fig. 1C). There was also no significant difference in intraperitoneal glucose tolerance tests between Bid-deficient and wild-type mice (Fig. 1D). These results demonstrate that Bid is not required for normal development and function of islets or apoptosis due to physical stress that occurs during the isolation process.

Bid has been reported to be essential for apoptosis induced by a broad range of cytotoxic stimuli (27) and direct DNA damage (28,29) or endoplasmic reticulum (ER) stress (30). In addition, Bid can be activated by effector caspases and has therefore been proposed to function as a general amplifier of apoptosis signaling, but whether this is essential for cell killing has not been determined. Islets from Bid-deficient and wild-type mice were exposed to stimuli that activate the mitochondrial apoptotic pathway and cell death measured by intracellular DNA content analysis (apoptotic cells exhibit a sub-G1 DNA content). Bid-deficient islets underwent apoptosis to the same extent as wild-type islets in response to treatment with staurosporine, a broad-spectrum kinase inhibitor (Fig. 2A). Bid-deficient islets were also normally sensitive to DNA damage induced with the topoisomerase inhibitor etoposide (Fig. 2B) or γ-radiation (Fig. 2C), consistent with previous analysis of fibroblasts and hemopoietic cells (17,31). Islets are thought to be particularly susceptible to ER stress because of their high rate of protein synthesis and secretion (32). We found that Bid-deficient and wild-type islets were equally susceptible to treatment with the ER stressors tunicamycin (Fig. 2D) or thapsigargin (Fig. 2E). While as many as 40% of the cells in islets represent non-β cells, apoptosis observed in these assays always exceeded 40%, suggesting that β-cells at least in part contributed to the cell death. These results demonstrate that Bid is not required for mitochondrial apoptosis signaling by a broad range of cytotoxic stimuli.

Overexpression of the caspase-8 inhibitor CrmA (33) or dominant-negative FADD (dnFADD) (20) both inhibit FasL-induced apoptosis of β-cells. However, the connection between the death receptor pathway and the mitochondrial pathway, and Bid in particular, in FasL-induced β-cell killing, has not yet been explored. Islets were cultured with IFN-γ plus IL-1β to induce Fas expression (which is not constitutively expressed on islet cells [34]) and were cotreated with FasL. Because high concentrations of IFN-γ plus IL-1β are toxic to islets even without adding FasL, the concentrations of these cytokines were optimized to obtain efficient Fas upregulation with minimal cytotoxicity (1 unit/ml IFN-γ and 150 units/ml IL-1β) (data not shown). After a 4-day incubation with IFN-γ, IL-1β, and FasL, apoptotic cells were identified by flow cytometry (Fig. 3A). Significant killing was seen in wild-type islets (65.3 ± 6.6%) and, as expected, this was inhibited by the broad spectrum caspase inhibitor zVAD.fmk (34). In contrast, Bid-deficient islets were highly resistant to FasL, exhibiting only 12.6 ± 1.5% (bid^−/− vs. wild type; P < 0.001) (Fig. 3B). As previously reported (17), Bid-deficient thymocytes (type I cells) were normally sensitive to FasL (Fig. 3C).

We also examined whether Bid was essential for FasL-induced cytochrome c release from mitochondria in islet cells. Upon treatment with IFN-γ, IL-1β, and FasL, wild-type islets released cytochrome c (46.6 ± 4.5% cytochrome c low), whereas Bid-deficient islets showed only little release of cytochrome c above background (25.5 ± 5.9%) (Fig. 4A and B). These data demonstrate that Bid is essential for FasL-induced mitochondrial cytochrome c release and apoptosis of islet cells, indicating that they display a type II phenotype.

Because Bid needs to activate Bax/Bak to trigger apoptosis and because Bax and Bak are controlled by the antiapoptotic members of the Bcl-2 family (35), we tested whether Bcl-2 overexpression can inhibit FasL-induced killing of islets. Islets from RIP-Bcl-2 transgenic or wild-type mice were cultured in the presence of IL-1β, IFN-γ, and FasL and cell death examined after 4 days. Bcl-2–overexpressing islets were as resistant to FasL as those from bid^−/− mice, and this was seen on two separate genetic backgrounds: bm1 (21.4 ± 1.8% RIP.Bcl-2 vs. 69.8 ± 9.3% wild type) and NOD (25.5 ± 3.0% NOD.RIP.Bcl-2 vs. 71.5 ± 7.9% wild type) (Fig. 5A and B).

Fas upregulation on Bid-deficient and Bcl-2–overexpressing islets treated with cytokines was similar to that observed in wild-type islets (Fig. 5C and D), excluding the possibility that they are abnormally resistant to FasL as a result of a defect in Fas expression. These results show that Bcl-2 overexpression can protect islet β-cells from FasL-induced apoptosis.

The protection of Bid-deficient islets from FasL-induced killing prompted us to test whether they were protected from CTL that kill in a FasL-dependent manner. To do this, we used in vitro–activated, perforin-deficient, ovalbumin-specific OT-I CTL as effector cells and peptide-pulsed islets from wild-type or Bid-deficient mice as targets. We have previously shown that perforin-deficient OT-I cells kill peptide-pulsed islets in a FasL-dependent manner (24). OVA_257–264-pulsed Bid-deficient islets were significantly protected from killing at both 10:1 and 20:1 effector:target ratios (supplementary Fig. 1, available in an online appendix at http://dx.doi.org/10.2337/db07-1692).

To assess whether Bid and linkage to the mitochondrial apoptotic pathway are also required for apoptosis triggered by other death receptors, we treated islets from Bid-deficient, Bcl-2–overexpressing and wild-type mice with TNF- in the presence of CHX. In most primary cells, including β-cells, TNF-α does not cause cell death on its own because its receptors, TNF-R1 and TNF-R2, activate the Rel/nuclear factor-κB pathway that promotes expression of inhibitors of apoptosis such as FADD-like IL-1β–converting enzyme inhibitory protein (FLIP), which blocks caspase-8. However, when new protein synthesis is blocked by CHX or when the Rel/NFκB pathway is incapacitated (e.g., by deficiency of a critical component such as RelA), TNF-α elicits cell death by recruitment of TRADD (TNF receptor–associated DD protein) and FADD to TNF-R1, resulting in the activation of caspase-8 (36). Figure 6A and B shows that, in comparison to wild-type islets, those lacking Bid or overexpressing Bcl-2 were resistant to treatment with TNF-α plus CHX (24.4 ± 2.0% bid^−/−, 67.4 ± 7.5% wild type [Fig. 6A]; 19.4 ± 3.1% RIP.Bcl-2, 53.6 ± 1.1% wild type [Fig. 6B]). These results demonstrate that Bid is essential not only for FasL- but also for TNF-α–induced apoptosis of islet cells.

Islets treated with the combination of IFN-γ, IL-1α/β, and TNF-α (in the absence of CHX) undergo cell death in culture. The main mechanism for this cell death is believed to be dependent on generation of nitric oxide (NO) (2). We treated islets from wild-type and Bid-deficient mice for 4 days with IL-1β, IFN-γ, and TNF-α and measured cell death by flow cytometry. Combined treatment with IFN-γ, IL-1β, and TNF-α (Fig. 7A) or IFN-γ plus TNFα (Fig. 7B) killed 59.6 ± 1.9 or 23.6 ± 2.6%, respectively, of the wild-type islet cells. Bid-deficient islets showed significantly less killing: 44.8 ± 3.1% for treatment with IFN-γ, IL-1β plus TNF-α (Fig. 7A), and 17 ± 2.2% for stimulation with IFN-γ plus TNF-α (Fig. 7B). Bid-deficient islets were, however, normally sensitive to treatment with IFN-γ plus IL-1α or IL-1β (Fig. 7C). Similar to loss of Bid, addition of the caspase inhibitor zVAD.fmk afforded partial protection from apoptosis induced by treatment with IFN-γ, IL-1β, and TNFα (41.5 ± 3.2%) (Fig. 7A) or IFN-γ plus TNFα (18.7 ± 0.9%) (Fig. 7B) but had no impact on killing by IFN-γ plus IL-1α or IL-1β (data not shown). Addition of the inducible NO synthase inhibitor NMMA almost completely prevented cell death induced by any of these three treatments (7.9 ± 1.1% IFN-γ, IL-1β, and TNFα; 9.0 ± 1.4% IFN-γ plus TNF-α). These data indicate that TNF-α in the cytokine cocktail triggers an apoptotic pathway that requires caspase activity and Bid, which can operate in parallel to the cytokine (IFN-γ plus IL-1α or IL-1β)-induced cell-killing pathway, which is mediated by NO. IL-1α and IL-1β, on the other hand, only induce NO-dependent cell death, which can be blocked by the inducible NO synthase inhibitor NMMA and appears to be independent of zVAD.fmk-sensitive caspases and Bid.

The multi-BH domain proapoptotic Bcl-2 family members Bax and Bak are released from antiapoptotic Bcl-2 family members when BH3-only proteins such as Bid are activated and have essential, albeit largely overlapping roles in developmentally programmed and cytotoxic stress–induced apoptosis (10,37). Islets deficient in either Bax or Bak displayed reduced apoptosis after treatment with cytokines plus FasL (Fig. 8A and B) (45.9 ± 6.0% bax^−/− and 80.7 ± 2.2% wild type [Fig. 8A]; 45.3 ± 3.5% bak^−/− and 64.4 ± 3.3% wild type [Fig. 8B]) or TNF-α plus CHX (Fig. 8C and D) (62.6 ± 3.0% bax^−/− and 83.6 ± 1.1% wild type [Fig. 8C]; 57.9 ± 3.2% bak^−/− and 87.1 ± 0.4% wild type [Fig. 8D]). Bak-deficient islets with only a single allele of Bax showed no additional protection over Bak deficiency alone. This indicates that at least one allele of either Bax or Bak is required for death receptor killing in islets and that without one of these proteins apoptosis is reduced.

β-cell destruction underlies or at least contributes to development of type 1 as well as type 2 diabetes (2). Multiple cell death pathways, including perforin/granzyme, Fas/FasL, TNF-α, and other cytokines, as well as excessive ER stress (2), have been implicated in β-cell killing, but the mechanisms are still unclear. We have found that the proapoptotic BH3-only protein Bid is critical for FasL- and TNF-α–induced apoptosis of islet cells and also plays a minor role in their killing by inflammatory cytokines (IFN-γ, IL-1β, and TNF-α). This shows that, like in hepatocytes (16,17), death receptor–induced apoptosis requires amplification of the caspase cascade through Bid-mediated activation of the mitochondrial pathway for efficient killing of β-cells. In other cell types, such as thymocytes or mature T-cells, ligation of death receptors leads to efficient cell killing solely through activation of caspase-8 and the downstream effector caspases (−3, −6, and −7) without the need for Bid- and mitochondrial pathway–mediated amplification of the caspase cascade. Therefore, islets are one of only a handful of primary cell types in which Fas-induced apoptosis is dependent on Bid (16,17,38). No pathology in the pancreas was seen when mice were injected with agonistic anti-Fas antibodies. This may be due to the fact that β-cell killing takes longer than the killing of hepatocytes, which leads to death of the animals. Alternatively, and probably more likely, β-cells are not killed because, in contrast to hepatocytes, they do not constitutively express Fas in vivo.

Like Bid deficiency, overexpression of Bcl-2 also protected islet β-cells from death receptor–induced killing, and loss of either Bak or Bax could also partially inhibit this process. This further confirms that the connection between the death receptor and the mitochondrial pathway is essential for FasL-induced apoptosis in β-cells. Curiously, although Bcl-2 overexpression can protect β-cells from FasL, it is unable to do so in hepatocytes (39,40), which also require Bid for Fas-mediated apoptosis (16,17). There are two possible explanations. Since activated Bid binds only relatively weakly to Bcl-2, the very high levels of transgene expression achieved in β-cells may be sufficient for protection, whereas the levels achieved in hepatocytes might be inadequate. Alternatively, since Bcl-2 was shown to be able to control only Bax but not Bak (23), it is possible that FasL-induced apoptosis is mostly dependent on Bak in hepatocytes, while either Bax or Bak can mediate this death in β-cells.

Loss of Bid did not protect β-cells against a range of apoptotic stimuli that can be countered by Bcl-2 overexpression (19), such as ER stress or DNA damage. It appears likely that these death stimuli activate other BH3-only proteins, such as Bim in the case of ER stress (41) or Puma in the case of DNA damage (42). Since β-cell killing induced by IFN-γ plus IL-1β, which requires NO production, cannot be inhibited by either loss of Bid, caspase blockade, or Bcl-2 overexpression (20), it appears likely that this death occurs through a process independent of this pathway.

Apoptotic β-cell death is central to the pathogenesis of type 1 diabetes and islet graft rejection. β-cell killing in type 1 diabetes is likely to involve at least the perforin/granzyme and death receptor pathways. The importance of the perforin pathway is probably best documented by the finding that loss of perforin greatly reduces the incidence of diabetes in NOD mice (43). Experiments with transgenic NOD mice expressing dnFADD in their β-cells indicated that death receptor signaling plays a partial role in diabetes development (20). Antigen-specific autoreactive cytotoxic T-cells kill β-cells in vitro using the Fas/FasL pathway in the absence of perforin (24,44), indicating that this mechanism may still contribute to diabetes development. Therefore, simultaneous inhibition of multiple cell killing pathways will be needed to prevent β-cell destruction and diabetes development. For this reason, genetic strategies to inhibit intracellular pathways of β-cell apoptosis have been generally less successful than hoped in blocking diabetes or allograft destruction. Inhibition of extrinsic (i.e., dnFADD, cFLIP, or A20) or intrinsic (i.e., Bcl-2 or Bcl-X_L) signals of apoptosis has only limited impact on islet cell apoptosis and diabetes development. Overexpression of XIAP, which prevents activation of effector caspases late in apoptosis, has shown promising results (45,46). However, caspase inhibition is generally not sufficient for survival following mitochondrial outer membrane permeabilization (47,48). Collectively, the observations that loss of Bid protects β-cells against FasL, TNF-α, inflammatory cytokines (albeit only partially), and (at least in vitro) perforin plus granzyme B indicate that pharmacological inhibition of Bid may be able to prevent or limit apoptosis induced by multiple pathways without interfering with normal physiological β-cell function. Inhibitors targeted at Bid have been recently described (49), making this a feasible approach for preserving β-cell survival and prevention of diabetes.

This work was supported by grants and fellowships from the Juvenile Diabetes Research Foundation, the National Health and Medical Research Council of Australia, the National Cancer Institute, the Leukemia and Lymphoma Society of America, and the Swiss National Science Foundation.

We thank Dr. Marc Dupuis from Apoxis for the MegaFasL and Rochelle Ayala and Stacey Fynch for excellent technical assistance.