Role of Calcium in Pancreatic Islet Cell Death by IFN-γ/TNF-α1

We studied the intracellular events associated with pancreatic β cell apoptosis by IFN-γ/TNF-α synergism. IFN-γ/TNF-α treatment of MIN6N8 insulinoma cells increased the amplitude of high voltage-activated Ca2+ currents, while treatment with IFN-γ or TNF-α alone did not. Cytosolic Ca2+ concentration ([Ca2+]c) was also increased by IFN-γ/TNF-α treatment. Blockade of L-type Ca2+ channel by nifedipine abrogated death of insulinoma cells by IFN-γ/TNF-α. Diazoxide that attenuates voltage-activated Ca2+ currents inhibited MIN6N8 cell death by IFN-γ/TNF-α, while glibenclamide that accentuates voltage-activated Ca2+ currents augmented insulinoma cell death. A protein kinase C inhibitor attenuated MIN6N8 cell death and the increase in [Ca2+]c by IFN-γ/TNF-α. Following the increase in [Ca2+]c, calpain was activated, and calpain inhibitors decreased insulinoma cell death by IFN-γ/TNF-α. As a downstream of calpain, calcineurin was activated and the inhibition of calcineurin activation by FK506 diminished insulinoma cell death by IFN-γ/TNF-α. BAD phosphorylation was decreased by IFN-γ/TNF-α because of the increased calcineurin activity, which was reversed by FK506. IFN-γ/TNF-α induced cytochrome c translocation from mitochondria to cytoplasm and activation of caspase-9. Effector caspases such as caspase-3 or -7 were also activated by IFN-γ/TNF-α treatment. These results indicate that IFN-γ/TNF-α synergism induces pancreatic β cell apoptosis by Ca2+ channel activation followed by downstream intracellular events such as mitochondrial events and caspase activation and also suggest the therapeutic potential of Ca2+ modulation in type 1 diabetes.

A poptosis of pancreatic ␤ cells is now becoming one of the important fields in diabetes research due to recent realization of the mechanism and significance of apoptosis in a variety of physiological or pathological conditions. In autoimmune diabetes, apoptosis of pancreatic ␤ cells is the most critical and final step in the development of autoimmune diabetes (1,2). It may also be important in the pathogenesis of type 2 diabetes, while direct evidence for such a theory has not yet been obtained (3)(4)(5).
Although the importance of pancreatic ␤ cell apoptosis in autoimmune diabetes has been accepted, it is still not clearly elucidated which molecule(s) is the real effector(s) in autoimmune diabetes. The role for Fas ligand as the death effector in autoimmune diabetes has been questioned (6 -8). Others and we have published in vitro and in vivo data suggesting that IFN-␥/TNF-␣ synergism is responsible for pancreatic ␤ cell death (9,10). The role of TNF-␣ as the final death effector molecule is consistent with other articles using genetic ablation models (6,11), while that of IFN-␥ in the development of autoimmune diabetes is not unanimously supported by other studies using the gene targeting technique (12)(13)(14). Intracellular events associated with the apoptosis of pancreatic ␤ cells are also largely unknown. Although the elucidation of intracellular events associated with apoptosis is regarded as one of the most important breakthroughs in the cell biology field, such events in the death of pancreatic ␤ cells have been addressed only in a few reports.
Pancreatic ␤ cells are electrically excitable, displaying action potentials and subsequent rise of cytosolic Ca 2ϩ concentration ([Ca 2ϩ ] c ). 3 Regulation of glucose-induced insulin secretion critically depends on glucose metabolism and the electrical activity controlled by various plasma membrane ion channels, among which the ATP-sensitive K ϩ (K ATP ) channel and the voltage-activated Ca 2ϩ channel (VACC) play key roles. Beside their physiological roles in the regulation of insulin secretion, ion channels of ␤ cells could be targets of various cytokines or other effector molecules. For instance, it has been demonstrated that serum from type 1 diabetes patients activated L-type Ca 2ϩ channels of insulinoma cells while the molecular identity was not characterized (15). In contrast, treatment of ␤ cells with the IL-1␤/IFN-␥ combination has been reported to induce low-voltage activated Ca 2ϩ current (LVA-I Ca ), leading to their death (16). In fact, Ca 2ϩ has been implicated in a variety of cell death modes such as glutamateinduced excitotoxicity of neuronal cells (17) or reperfusion necrosis of cardiac cells (18). Ca 2ϩ is also a prototypic agent inducing mitochondrial permeability transition (MPT) (19), which has been considered as the "point of no return" in several cell death models (20). However, the role of Ca 2ϩ in classical apoptosis is not clearly established, and it remains to be clarified whether MPT precedes and induces cytochrome c translocation or ensues after critical mitochondrial events (21).
This study was conducted to investigate the apoptotic cascade associated with apoptosis of pancreatic ␤ cells by IFN-␥/TNF-␣ synergism with a particular emphasis on the role of Ca 2ϩ .

Whole-cell patch clamp study
Cells were transferred into a bath on an inverted microscope (IX-70; Olympus, Tokyo, Japan). The bath was superfused with Tyrode's solution containing 140 mM NaCl, 5 mM KCl, 10 mM HEPES, 1 mM MgCl 2 , 1.3 mM CaCl 2 , and 5 mM glucose (pH 7.4), and voltage clamp experiments were performed at room temperature. Patch pipettes were connected to the head stage of a patch-clamp amplifier (Axopatch-1D; Axon Instruments, Foster City, CA). Liquid junction potentials were corrected with an offset circuit before each experiment. Voltage and current data were low-pass filtered (5 kHz) and analyzed using Origin version 6.1 software (Microcal Software, Northampton, MA). The membrane potential was monitored under the zero-current clamp mode of whole-cell configuration using KCl pipette solution (135 mM KCl, 5 mM NaCl, 0.5 mM MgCl 2 , 3 mM MgATP, and 10 mM HEPES (pH 7.3) titrated with KOH). Ca 2ϩ currents were recorded using CsCl pipette solution containing 140 mM CsCl, 0.5 mM MgCl 2 , 3 mM MgATP, and 10 mM HEPES (pH 7.3). After making a whole-cell configuration, Ca 2ϩ currents were elicited by step-depolarizing pulses. The membrane voltage was held at Ϫ90 mV and various levels of step-like depolarizing pulses were applied. In another series of experiments, K ATP current in each group of cells was measured. ATP-free KCl pipette solution was used for the whole-cell patch clamp, and cells were superfused with glucose-free KCl bath solution, where the K ϩ equilibrium potential was located at 0 mV. The membrane conductance was monitored as brief current/voltage (I/V) curves obtained by ramp-like pulses. The superfusion with a glucose-free KCl condition alone slowly increased the membrane conductance. When the membrane conductance reached a steady state, 50 -100 M diazoxide was added, which further increased the membrane conductance. Then, 0.2 M glibenclamide (GC) was applied, and the size of K ATP current in each cell was measured as a GC-sensitive current at Ϫ80 mV (see Fig. 2C).

Measurement of [Ca 2ϩ ] c
Cells were loaded with the acetoxymethyl ester form of 2 M fura-2 (Molecular Probes, Eugene, OR) in Tyrode's solution at room temperature for 20 min and then rinsed twice with fresh solution. The recording of [Ca 2ϩ ] c was performed with a microfluorometric system comprising an inverted fluorescence microscope (Olympus IX-70) with a dry-type fluorescence objective lens (ϫ40; aperture, 0.85), a type R1527 photomultiplier tube (Hamamatsu, Japan), and a Deltascan illuminator (Photon Technology International, Lawrence Ville, NJ). [Ca 2ϩ ] c was calculated from the fluorescence emission ratio R at 340 nm/380 nm excitation according to a reported formula (22).

Calpain and calcineurin activity
Calpain activity was measured using t-BOC-Leu-Met-CMAC fluorochrome (Molecular Probes) as a substrate. After incubating cells in 0.15 M NaCl-5 mM HEPES-2 mM EDTA (pH. 7.35) containing the substrate, fluorescence was measured at 530 nm using a fluorescence microplate reader (Bio-Tek, Winooski, VT). Serine/threonine phosphatase activity was determined as a measure of calcineurin activity using a commercial kit (Promega, Madison, WI). Briefly, cell lysate was added to the reaction premix in each well of 96-well plates. After the reaction was stopped with premade molybdate dye/additive mix, absorbance at 600 nm was measured with a microplate reader.

Cell fractionation
Cells were fractionated according to a previously published protocol (23). Briefly, cells were lysed in a buffer containing 10 mM HEPES (pH 8.0), 250 mM sucrose, 1 mM EDTA, 1 mM EGTA,1 mM DTT, 2 mM PMSF, and 100 g/ml leupeptin. After scraping of the cells and douncing, unbroken cells and nuclear fraction were discarded after centrifugation at 600 ϫ g for 10 min. After further centrifugation at 15,000 ϫ g for 10 min, pellet was suspended in heavy membrane (HM) buffer containing 10 mM HEPES, 5 mM MgCl 2 , 42 mM KCl, and 1% Triton X-100 (HM fraction). Supernatant was divided into the light membrane fraction and cytosolic fraction after ultracentrifugation at 100,000 ϫ g for 60 min. Purity of the HM fraction and cytosolic fraction was confirmed by immunoblot using anti-Cox4 and -IB Ab, respectively.

Measurement of mitochondrial potential
Mitochondrial potential was measured using JC-1 (Molecular Probes). After incubating 5 ϫ 10 6 cells in 0.5 ml of PBS containing 10 g/ml JC-1 at 37°C for 10 min, emission at 510 and 590 nm was measured using a fluorescence microplate reader. The ratio of A 590 :A 510 was calculated as an indicator of mitochondrial potential.

Isolation of mouse pancreatic islets
Islets were isolated from overnight-fasted ICR mice by the collagenase digestion technique as previously described (10). Briefly, pancreas was gently pulled out after 2.5 ml of collagenase P (0.8 mg/ml) was injected into the bile duct. The pancreas was then digested in collagenase P solution at 37°C for 15 min with gentle shaking. After stopping the digestion with cold HBSS, tissue was then passed through a 400-m screen and centrifuged on 25, 23, 21.5, and 11.5% Ficoll gradients, respectively. Islets collected from the interface were washed with M199 medium, and individual islets were hand-picked using micropipettes. The islets were treated with trypsin-EDTA for 5 min to yield single islet cells before treatment with test reagents. All animal experiments in this work were done in accordance with the institutional guidelines of Samsung Medical Center Animal Facility, an Association for Assessment and Accreditation for Laboratory Animal Care International-accredited facility.

Statistical analysis
All values were expressed as means Ϯ SEM. Student's t test was used to compare the differences between the groups. Values of p Ͻ 0.05 were regarded as statistically significant.

Increased Ca 2ϩ flux and intracellular Ca 2ϩ contents by IFN-␥/ TNF-␣ combination
Because recent articles reported the expression of the LVA-Ca 2ϩ channel in islet ␤ cells from NOD mice (25) and activation of the LVA-Ca 2ϩ channel in pancreatic islet cell death by cytokines (16), we first studied whether LVA-Ca 2ϩ channel was activated in our model of pancreatic islet cell death by 100 U/ml IFN-␥ in combination with 10 ng/ml TNF-␣ (IFN-␥/TNF-␣) using MIN6N8 insulinoma cells. Under the whole-cell patch clamp condition, an inward Ca 2ϩ current appeared at above Ϫ40 mV and reached its peak at 0 mV. The peak inward current to voltage relation showed an inverted bell-shaped I/V curve, suggesting the dominance of high-voltage activated (HVA)-Ca 2ϩ channels (Fig. 1A). Current tracing showed that the inward current was markedly augmented by 1 M BayK8644 and blocked by 10 M nifedipine, the L-type Ca 2ϩ channel agonist and antagonist, respectively (Fig. 1B), again indicating that the L-type Ca 2ϩ channels showing high threshold voltage are predominantly expressed in MIN6N8 cells. Recording of membrane current response to step-depolarizing pulse and I/V curve analysis showed that treatment of MIN6N8 cells with IFN-␥/TNF-␣ for 6 h increased the amplitude of inward Ca 2ϩ currents through HVA-Ca 2ϩ channels (HVA-I Ca ) by 43% compared with control cells (Fig. 1, A, C, and D). The increase in Ca 2ϩ current was due to the synergistic action of IFN-␥ and TNF-␣ because treatment with MIN6N8 cells with either IFN-␥ or TNF-␣ alone did not increase Ca 2ϩ current (Fig. 1A). I/V curves from 22 cells showed that there is no clear evidence demonstrating the activation of LVA-Ca 2ϩ channel current (Fig. 1A).
Next, we compared [Ca 2ϩ ] c of control and IFN-␥/TNF-␣treated cells using the fura-2 method. Treatment of MIN6N8 cells with IFN-␥/TNF-␣ for 4 h induced a significant increase in [Ca 2ϩ ] c . The increase in [Ca 2ϩ ] c reached the peak 6 h after IFN-␥/TNF-␣ treatment and was sustained until 10 h ( p Ͻ 0.05 for all comparisons; Fig. 2A ] c could be due to an increased membrane depolarization in addition to increased Ca 2ϩ current through HVA-I Ca , we next measured the membrane potential of MIN6N8 cells. The mean of initial resting membrane potential (RMP) within 30 s after making a whole-cell configuration was Ϫ45 Ϯ 1.9 mV (n ϭ 23) and Ϫ43 Ϯ 1.3 mV (n ϭ 24) for control and IFN-␥/TNF-␣-treated cells, respectively. There was no significant difference between them ( p Ͼ 0.1), suggesting that the increased [Ca 2ϩ ] c after treatment of MIN6N6 cells with IFN-␥/TNF-␣ is due not to an increase in membrane depolarization but to an increased Ca 2ϩ influx (Fig. 2B). The steady-state RMP 5 min after making a whole-cell configuration was also not significantly different between control (Ϫ50.6 Ϯ 2.47 mV, n ϭ 23) and IFN-␥/TNF-␣treated cells (Ϫ48 Ϯ 1.45 mV, n ϭ 24) ( p Ͼ 0.1; Fig. 2B). We also compared the size of K ATP current before and after treatment with IFN-␥/TNF-␣ because the activity of the K ATP channel plays a crucial role in the determination of RMP of ␤ islet cells and in the release of insulin (26). K ATP current in MIN6N8 cells measured as GC-sensitive current under the glucose-free condition or as full activation by diazoxide was not significantly changed by IFN-␥/ TNF-␣ treatment ( p Ͼ 0.1 for both comparisons; Fig. 2, C and D). As a whole, the positive modulation of HVA-Ca 2ϩ channel current with no significant changes in RMP and K ATP current suggests that the increase in [Ca 2ϩ ] c after IFN-␥/TNF-␣ treatment is due to an increased Ca 2ϩ influx through the HVA-Ca 2ϩ channel during action potentials.

Death of insulinoma cells due to increased intracellular Ca 2ϩ
Because the above change in [Ca 2ϩ ] c and HVA-I Ca by IFN-␥/ TNF-␣ raised the possibility that the observed increase in [Ca 2ϩ ] c might cause death of MIN6N8 cells, we next studied whether blockade of Ca 2ϩ influx in MIN6N8 cells could affect their death  after IFN-␥/TNF-␣ treatment. Death of MIN6N8 cells by IFN-␥/ TNF-␣ was a classical apoptosis associated with DNA fragmentation/nuclear condensation by Hoechst staining and sub-G 1 peak in DNA ploidy assay as previously reported (10) (data not shown). Ten micromolar nifedipine, a classical blocker of the L-type HVA-Ca 2ϩ channel, significantly inhibited death of MIN6N8 cells by IFN-␥/TNF-␣ (Fig. 3A), suggesting that the increase in Ca 2ϩ influx after IFN-␥/TNF-␣ treatment induced or triggered their death ( p Ͻ 0.01). One hundred micromolar NiCl 2 , a nonspecific blocker of Ca 2ϩ channels, also inhibited MIN6N8 cell death by IFN-␥/ TNF-␣ ( p Ͻ 0.01; Fig. 3A). Similar effects were observed when VACC was indirectly modified. The application of 50 M diazoxide, a K ATP channel opener that hyperpolarizes the membrane and prevents the activation of VACC (Fig. 2C), significantly inhibited the death of MIN6N8 cells after treatment with IFN-␥/TNF-␣ (Fig.  3B), further suggesting the role of increased Ca 2ϩ influx in MIN6N8 cell death ( p Ͻ 0.01). In contrast, GC that blocks K ATP channel activity and augments the activation of VACC (Fig. 2C) significantly accentuated the death of MIN6N8 cells by IFN-␥/ TNF-␣ in a dose-dependent manner ( p Ͻ 0.05; Fig. 3B).
To study the mechanism of the increase in [Ca 2ϩ ] c by IFN-␥/ TNF-␣ synergism, we investigated the effect of protein kinase C (PKC) inhibitors because the PKC activation by growth factors or cytokines is reportedly related to an increased Ca 2ϩ influx (27,28). One micromolar chelerythrine, a PKC inhibitor, substantially attenuated the increase in [Ca 2ϩ ] c by IFN-␥/TNF-␣ treatment for 6 h ( p Ͻ 0.01), suggesting that the activation of PKC is probably involved in the activation of the Ca 2ϩ channel and the increase in Ca 2ϩ influx (Fig. 4A). Consistent with these results, 1 M chelerythrine partly inhibited death of MIN6N8 cells by IFN-␥/TNF-␣ treatment for 48 h ( p Ͻ 0.05; Fig. 4B).

Calcineurin-mediated dephosphorylation of BAD in insulinoma cell death by IFN-␥/TNF-␣
Because these results indicated the role of Ca 2ϩ influx as a triggering event in MIN6N8 cell apoptosis by IFN-␥/TNF-␣, we next asked whether calpain, an abundant Ca 2ϩ -activatable protease, was activated in MIN6N8 cells after treatment with IFN-␥/TNF-␣. Calpain substrate assay showed a significantly increased fluorescence 3-24 h after IFN-␥/TNF-␣ treatment ( p Ͻ 0.05 for all comparisons), indicating calpain activation by increased intracellular Ca 2ϩ (Fig. 5A). Thus, we next studied whether calpain inhibitors could block MIN6N8 cell death by IFN-␥/TNF-␣. Both 1 M calpeptin and 1 M ALLN, specific and cell-permeable inhibitors of calpain I and II, inhibited the increase in calpain activity by IFN-␥/TNF-␣ treatment for 24 h ( p Ͻ 0.05 for both comparisons; Fig. 5A) and also significantly attenuated MIN6N8 cell death by IFN-␥/TNF-␣ treatment for 48 h ( p Ͻ 0.01 for both comparisons), suggesting that activated calpain plays a role in MIN6N8 cell death by IFN-␥/TNF-␣ (Fig. 5B). Because previous articles reported that calpain activates calcineurin, leading to Ca 2ϩ -induced apoptosis (29,30), we next conducted a serine/threonine phosphatase assay as a measure of calcineurin activity after treatment of MIN6N8 cells with IFN-␥/TNF-␣. Calcineurin activity began to rise 3 h after IFN-␥/TNF-␣ treatment, which became significant 6 h after cytokine treatment and reached the peak at 12 h ( p Ͻ 0.05 at 6, 12, and 24 h; Fig. 5C). Furthermore, inhibition of calcineurin activity with 1 M FK506 or 1 M cyclosporin A (CsA) significantly inhibited death of MIN6N8 cells after IFN-␥/TNF-␣ treatment for 48 h ( p Ͻ 0.01 for both comparisons), again supporting the role of increased calcineurin activity as an important step in MIN6N8 cell death by IFN-␥/TNF-␣ (Fig. 5D). We next studied the phosphorylation status of BAD using Ab specific for phospho-BAD because previous articles described calcineurin-dependent dephosphorylation of BAD as a mechanism of Ca 2ϩ -induced apoptosis (30,31). As hypothesized, a decreased phosphorylation of Ser 112 of BAD was observed between 3 and 24 h after treatment of MIN6N8 cells with IFN-␥/TNF-␣. In addition, the decrease in BAD phosphorylation was reversed by FK506, suggesting that calcineurin activation after treatment of MIN6N8 cells with IFN-␥/TNF-␣ was responsible for BAD dephosphorylation (Fig. 5E). In contrast, the expression of total BAD was not changed by IFN-␥/TNF-␣ (Fig. 5F).

Early cytochrome c translocation and delayed loss of mitochondrial potential by IFN-␥/TNF-␣
Because dephosphorylated BAD could translocate to mitochondria and lead to mitochondrial events critically involved in the apoptotic pathway (32, 33), we next studied the possible translocation of cytochrome c and change in mitochondrial potential. As expected from the dephosphorylation of BAD, cytochrome c was translocated from the HM fraction to the cytosolic fraction 3 h after IFN-␥/TNF-␣ treatment (Fig. 6A). Cytochrome c translocation was obvious at 6 h and became maximal 48 h after cytokine treatment when almost all cytochrome c of the HM fraction was translocated to cytosol. Cytosolic cytochrome c was not observed 72 h after IFN-␥/TNF-␣ treatment probably because cytochrome c passed through damaged plasma membrane during the secondary necrosis as observed in other studies (34). In contrast to the early cytochrome c translocation, notable dissipation of mitochondrial potential was observed only 48 h after IFN-␥/TNF-␣ treatment when overt cell damage was apparent (Fig. 6B).

Cleavage of caspases by IFN-␥/TNF-␣
We next studied the postmitochondrial events occurring in the course of MIN6N8 cell death by IFN-␥/TNF-␣. We studied whether caspase-9 was cleaved after cytochrome c translocation. Immunoblot analysis showed that caspase-9 was cleaved 36 h after treatment of MIN6N8 cells with IFN-␥/TNF-␣ (Fig. 7A), suggesting that caspase-9 was cleaved by apoptosome following cytochrome c translocation from the mitochondria to the cytoplasm. We then asked whether effector caspases were cleaved as the final executors of insulinoma cell apoptosis. Immunoblot analysis using CM1 Ab recognizing cleaved caspase-3 but not its proform demonstrated that caspase-3 was cleaved yielding a 15-kDa fragment by IFN-␥/TNF-␣ treatment of MIN6N8 cells for 36 h (Fig. 7B). IFN-␥ or TNF-␣ alone did not induce cleavage of caspase-3. Caspase-7, another effector caspase, was also cleaved 36 h after IFN-␥/TNF-␣ treatment (Fig. 7C).

Changes of [Ca 2ϩ ] c in primary islet cells
To confirm that the change in Ca 2ϩ concentration by cytokine treatment is an important step in the death of primary islet cells as well as insulinoma cells, we treated islet cells from ICR mice with IFN-␥/TNF-␣ and measured [Ca 2ϩ ] c . [Ca 2ϩ ] c in primary islet cells after IFN-␥/TNF-␣ treatment for 4 h was significantly increased compared with that of untreated islet cells (Fig. 8A). Single cytokine was ineffective. Ten micromolar nifedipine and 1 M chelerythrine substantially inhibited the death of primary islet cells by IFN-␥/TNF-␣, suggesting that IFN-␥/TNF-␣ treatment induces death of islet cells by essentially the same mechanism involving Ca 2ϩ (Fig. 8B).

Discussion
Although TNF-␣ induces apoptosis by activating apical caspases in most cell death models, several previous articles have shown that TNF-␣ could induce apoptosis by increasing [Ca 2ϩ ] c after mobilization of Ca 2ϩ from an extracellular or intracellular source (35)(36)(37). Consistent with those reports, we also observed an increased Ca 2ϩ current and increased [Ca 2ϩ ] c in the current investigation studying the mechanism of IFN-␥/TNF-␣-induced apoptosis of pancreatic ␤ cells (10). At first, we expected a possible increase in Ca 2ϩ current through the LVA-Ca 2ϩ channel because previous articles reported that the LVA-Ca 2ϩ channel was abnormally expressed in ␤ cells of NOD mice and the Ca 2ϩ current through the LVA-Ca 2ϩ channel mediates pancreatic ␤ cell death by the IL-1␤/IFN-␥ combination (16,25). In contrast to those reports, we could not find evidence supporting the role of LVA-Ca 2ϩ channels in MIN6N8 cell death by IFN-␥/TNF-␣. A single peak of the I/V curve with the threshold voltage of Ϫ40 mV and an inward current markedly augmented by BayK8644 and blocked by nifedipine favored the presence of HVA-Ca 2ϩ channels rather than LVA-Ca 2ϩ channels. This discrepancy might be due to the difference in the types of cells or cytokines used. Although the RMP was not different between control insulinoma cells and IFN-␥/TNF-␣treated ones, an increased Ca 2ϩ influx through the HVA-Ca 2ϩ channel during action potentials could lead to an increased [Ca 2ϩ ] c observed in this study. The increase in Ca 2ϩ influx and [Ca 2ϩ ] c occurring 4 h after IFN-␥/TNF-␣ treatment seems to be responsible for the following apoptotic cascade because an L-type Ca 2ϩ channel blocker inhibited the death of insulinoma cells by IFN-␥/ TNF-␣. The inhibition and exaggeration of IFN-␥/TNF-␣-induced MIN6N8 cell death by diazoxide and GC, respectively, further support the role of Ca 2ϩ influx in MIN6N8 cell apoptosis by IFN-␥/TNF-␣, and are also consistent with a previous report showing  the inhibition of tolbutamide-induced ␤ cell death by a calcium channel blocker or diazoxide (38).
Activation of calpain was also observed in MIN6N8 cells after IFN-␥/TNF-␣ treatment. Furthermore, calpeptin or ALLN significantly decreased MIN6N8 cell death after IFN-␥/TNF-␣ treatment, consistent with previous reports showing a critical role of calpain in Ca 2ϩ -mediated apoptosis (39). Calcineurin, a serine/ threonine phosphatase activated by Ca 2ϩ , was also activated in MIN6N8 cells after IFN-␥/TNF-␣ treatment. Its abrogation by FK506 or CsA also inhibited insulinoma cell death by IFN-␥/ TNF-␣, suggesting an important role of calpain and calcineurin activation following Ca 2ϩ influx in this model. Although CsA could affect MPT, the inhibition of insulinoma cell death by another inhibitor of calcineurin without such an effect on MPT, FK506, strongly suggests that calcineurin activation plays a role in MIN6N8 cell death by IFN-␥/TNF-␣. Although the relationship between calpain and calcineurin activation remains largely unknown, a recent article suggested a possible role of cain as a missing link between those two processes in Ca 2ϩ -triggered cell death (29). Calpain-induced degradation of cain that binds to and inhibits calcineurin (40,41) could lead to calcineurin activation, while such a possibility was not tested in this study. The diabetogenic effect of FK506 or CsA reported in patients with islet transplantation is observed only after prolonged administration (42) and appears to be due to their effect on insulin secretion (43) resulting from the blockade of FK506-binding protein-cADP ribose interaction or K ATP (44,45). Such an observation is unrelated to islet cell death and not incompatible with our finding of the beneficial effect of FK506 or CsA on islet cell viability after a short-term treatment.
Our data showing dephosphorylation of BAD at Ser 112 probably by calcineurin activation is consistent with previous articles showing a critical role of BAD dephosphorylation as a downstream event following calcineurin activation Ca 2ϩ -mediated cell death models (30,31). As a downstream event of BAD dephosphorylation, we observed translocation of cytochrome c. In contrast to the early cytochrome c translocation, dissipation of mitochondrial potential was not apparent until 48 h after IFN-␥/TNF-␣ treatment, suggesting that the loss of mitochondrial potential is not a cause of cytochrome c translocation observed in this study. The causal relationship among cytochrome c translocation, loss of mitochondrial potential, and MPT is yet to be solved (21). Subsequent to mitochondrial events, we observed activation of caspase-9, which was evident 36 -48 h after IFN-␥/TNF-␣ treatment of MIN6N8 cells. As the final effector molecules in the apoptotic cascade, caspase-3 and caspase-7 were also cleaved. Because we have reported that IFN-␥/TNF-␣ synergism is one of the in vivo effector mechanisms of pancreatic ␤ cell death in autoimmune diabetes (10), the apoptotic process observed in this study may reflect that occurring in vivo, while other effector molecules such as IL-1␤ or NO would play some role in the destruction of islet ␤ cells. However, MIN6N8 insulinoma cell death by IFN-␥/ TNF-␣ does not entail NO or Fas as previously reported, and Fas expression or NO production was not induced by IFN-␥/TNF-␣ (Refs, 10 and 46; I. Chang and M.-S. Lee, unpublished results).
The mechanism of Ca 2ϩ channel activation is obscure. Protein kinases have been reported to modulate Ca 2ϩ current or [Ca 2ϩ ] c as downstream mediators of cytokines or hormones (27,47,48). Consistent with those reports, we observed a significant inhibition of IFN-␥/TNF-␣-induced increase in [Ca 2ϩ ] c and insulinoma cell death by a PKC inhibitor. Death of primary islet cells by IFN-␥/ TNF-␣ was also inhibited by the PKC inhibitor. Ca 2ϩ has diverse roles in pancreatic ␤ cells beside its involvement in islet cell death and well-recognized role in insulin secretion. Ca 2ϩ influx through the HVA-Ca 2ϩ channel is involved in the growth (49) or survival of islet cells (50). Thus, islet cells seem to maintain an intricate balance of Ca 2ϩ influx and [Ca 2ϩ ] c to achieve optimal growth and survival while avoiding a dangerous level of [Ca 2ϩ ] c .
Taken together, we report a mechanism of IFN-␥/TNF-␣-induced apoptosis of insulinoma and pancreatic islet cells involving Ca 2ϩ channel activation, increased Ca 2ϩ influx, calpain activation, calcineurin activation, BAD dephosphorylation, cytochrome c translocation, and the activation of effector caspases. Such a pathway involving Ca 2ϩ influx does not appear to play a role in Fas ligand-mediated apoptosis of target cells like Jurkat cells, indicating the uniqueness of this Ca 2ϩ pathway in islet cell death (I. Chang and M.-S. Lee, unpublished results). Although the mechanism of Ca 2ϩ channel activation by cytokine combination is not addressed further in this study, our model provides a new insight into the apoptotic cascade in pancreatic islet cells leading to autoimmune diabetes and suggests the therapeutic potential of Ca 2ϩ modulation in the prevention of type 1 diabetes.