Exposure to Chronic High Glucose Induces β-Cell Apoptosis Through Decreased Interaction of Glucokinase With Mitochondria: Downregulation of Glucokinase in Pancreatic β-Cells

MIN6N8 cells, which are SV40 T-transformed insulinoma cells derived from NOD mice, were kindly provided by Dr. M.S. Lee (Sungkyunkwan University School of Medicine, Seoul, Korea). These cells were grown in DMEM containing 15% fetal bovine serum, 2 mmol/l glutamine, 100 IU/ml penicillin, and 100 μg/ml streptomycin (Life Technologies, Gaithersburg, MD). All antibodies were obtained from Cell Signaling Technology (Beverly, MA) or Santa Cruz Biotechnology (Santa Cruz, CA), and chemicals were purchased from Sigma (St. Louis, MO). Bismaleimidohexane was obtained from Pierce Biotechnology (Rockford, IL).

Mouse GCK cDNA was amplified by PCR and cloned into pcDNA3 and pEGFP C2 vectors (Clontech). The pEGFP GCK plasmid was generated using forward 5′-gggaagtctgggctacttctg-3′ and reverse 5′-ctagtggactgggagagcatttg-3′ primers to produce oligonucleotides that were subcloned within EcoRI/BamH1 sites in the pEGFP vector. The pcDNA GCK plasmid was generated by cutting pEGFP mGCK with HindIII/BamH1 restriction enzymes and inserted into the HindIII/BamH1 site of the pcDNA3.1C-V5His vector (Invitrogen, Carlsbad, CA).

Islets were isolated from overnight-fasted ICR mice (weight 20–25 g) by the collagenase digestion technique, as previously described (22,23).

MIN6N8 cells were transfected with green fluorescent protein (GFP) GCK or pCDNA wild-type GCK and Neo vector control DNA using a lipofectin reagent (Life Technologies). After a 16-h incubation period, growth media was replaced and cells were grown for an additional 48 h. Cells were then exposed to a selective concentration of 400 μg/ml G418 sulfate (Life Technologies) to isolate stably transfected cells.

Cells were lysed in RIPA buffer (23) at 4°C and then vortexed and centrifuged at 16,000 rpm for 10 min at 4°C. The supernatant was mixed in Laemmli loading buffer, boiled for 4 min, and then subjected to SDS-PAGE. For endogenous complexes, mitochondrial lysates (1 mg) fractionated from the cells were immunoprecipitated with 2 μg antibody and immunoblotted. For exogenous complexes, whole-cell lysates (500 μg) of transfected cells were used.

Apoptotic cells were detected using Apop Tag, an in situ apoptosis detection kit from Oncor (Gaitherburg, MD) as previously described (23).

Immunocytochemistry for anti-Bax, anti-GCK, anti-p53, and anti-inuslin was performed as previously described (23).

The Bax oligomerization assay was adapted as previously described (17,23). Briefly, the mitochondrially enriched fraction from isolated cells (21) was subjected to protein cross-linking by incubation in a freshly prepared mixture of 10 mmol/l bismaleimidohexane (Pierce), 16.8% DMSO, and PBS for 30 min at room temperature with occasional mixing. For the Bax-VDAC–binding assay, mitochondrial fractions isolated from total cellular extracts were immunoprecipitated with anti-VDAC antibody, washed twice in lysis buffer, and subjected to Western blotting.

For in vitro–binding assays, 500 μg lysates were incubated with 3.0 μg glutathione S-transferase (GST) or fusion GST GCK proteins coupled to glutathione sepharose beads in 300 μl lysis buffer overnight at 4°C with continuous rocking, as previously described (24).

To determine insulin release in response to glucose stimulation, MIN6N8 cells or isolated islet cells stimulated with glucose for 4 days were washed in Krebs-Ringer bicarbonate buffer (KRBB) containing 3.3 mmol/l glucose and preincubated for 30 min in the same buffer. The KRBB was then discarded and replaced by fresh buffer containing 3.3 mmol/l glucose for 1 h, followed by an additional 1-h incubation in KRBB containing 16.7 mmol/l glucose. Supernatants were collected and frozen for insulin assays. Thereafter, cells were washed with PBS and extracted with 0.18 N HCl in 70% ethanol for 24 h at 4°C. The acid-ethanol extracts were collected for determination of insulin content. Insulin was determined by radioimmunoassay using mouse insulin as the standard (14). ATP levels in MIN6N8 cells and islets were determined using a luminometric assay kit from Promega as previously described (14).

For comparing values obtained in three or more groups, one-factor ANOVA was used followed by Tukey’s post hoc test, and P < 0.05 was taken to imply statistical significance.

To examine the cytotoxic effect of high glucose on a pancreatic insulinoma cell line, MIN6N8 cells were treated with glucose at different concentrations for varying time periods. Treatment with 33.3 mmol/l glucose induced marked genomic DNA fragmentation in a time- and dose-dependent manner and caused a significant increase in the number of terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL)-positive MIN6N8 cells relative to cells treated with 5.5 mmol/l glucose, concomitant with cleavage of poly(ADP-ribose) polymerase (PARP) similar to caspase-3 cleavage (Fig. 1A). Pretreatment of the cells with a specific caspase-3 inhibitor, z-DEVD-CHO, completely reduced 33.3 mmol/l glucose–induced PARP cleavage (Fig. 1C) and apoptosis (data not shown).

Exposure to high glucose resulted in a time- and dose-dependent decrease in Bcl-2 and Bcl-xL expression, whereas expression of Bax significantly increased, thereby increasing Bax/Bcl-2 or Bcl-xL ratios. The expression of Fas and p53, potent proapoptotic proteins in the mitochondrial apoptotic pathway (25), also significantly increased (Fig. 1B). Moreover, a significant time-dependent increase in mitochondrial release of cytochrome C into the cytosol was observed after 2 days. Cytochrome C oxidase IV (mitochondrial protein) was used to confirm whether mitochondrial fraction was isolated purely from cell extract (Fig. 1C). Release of cytochrome C from the mitochondria into the cytosol is mediated by Bax translocation to the mitochondrial outer membrane and its subsequent oligomerization (14,20). Normally residing within the cytosol, Bax levels were significantly reduced within cytosolic fractions after 2 days of exposure and increased in the mitochondrial fraction of 33.3 mmol/l glucose–treated cells (Fig. 1C). Translocation of Bax was further confirmed by immunostaining using fluorescein isothiocyanate (FITC) Bax and Mito-Tracker CMXRos. While FITC Bax was localized primarily in the cytosol of 5.5 mmol/l glucose–treated cells, treatment with 33.3 mmol/l glucose increased Bax translocation to the mitochondria by 4.3-fold (Fig. 1D). To confirm that Bax translocates to the mitochondria through binding with VDAC, we examined the interaction of Bax with VDAC in isolated mitochondrial fractions. Bax interaction with VDAC (Fig. 1E) and Bax oligomerization (Fig. 1F) substantially increased 2 days after treatment with 33.3 mmol/l glucose.

To examine whether chronic exposure to high-glucose–induced apoptosis occurs as a direct result of glucotoxicity, we studied the effects of high glucose on genes associated with glucose metabolism and insulin content. Chronic exposure to 33.3 mmol/l glucose strongly reduced expression of GCK in a time- and dose-dependent manner but not in hexokinase I (Fig. 2A and B). To confirm these results, immunocytochemistry on MIN6N8 cells was performed using anti-GCK. As shown in Fig. 2C, glucose reduced GCK expression dose dependently, decreasing significantly by 40% in 33.3 mmol/l glucose–treated cells. Consistent with decreased GCK expression, chronic exposure of the cells for 4 days abolished the ability of acute glucose to stimulate insulin content and inhibited ATP production (Fig. 2D). However, in cells treated with 16 mmol/l glucose, insulin content and ATP production increased slightly compared with cells treated with 5.5 mmol/l glucose. Furthermore, expression of other major glucose metabolism proteins, including Glut2, peroxisome proliferator–activated receptor γ coactivator 1, and sterol regulatory element–binding protein 1, was also significantly reduced, whereas considerable changes in the phosphatidylinositol 3-kinase subunit p85, VDAC, and superoxide dismutase were not detected.

The association of hexokinase I and II with mitochondria prevents the interaction of Bax with VDAC and inhibits Bax-induced apoptosis (15,16,18). Since GCK is a unique hexokinase family member whose expression is largely restricted to the liver and pancreatic β-cells, the ability of GCK to associate with mitochondria through VDAC was examined. Therefore, we first examined the localization of GCK using Mito-Tracker CMXRos after transient transfection with GFP GCK. GCK green fluorescence was coexpressed in both cytosol and mitochondria of control cells, while 1 day after glucose treatment, a large amount of the GFP GCK translocated to the mitochondria. However, the colocalization of GFP GCK in the mitochondria began diffusing into the cytosol after 2 days, and the GFP GCK appeared in the cytoplasm separate from the mitochondria 4 days after treatment, concomitant with decreased GFP GCK expression at the same time (Fig. 3A). Next, to characterize whether GCKs directly bind VDAC and whether 33.3 mmol/l glucose affects GCK binding to VDAC, the GST pull-down assay was conducted using purified GST or GST GCK. As shown in Fig. 3B, VDAC bound to GST GCK but not to control GST beads. Furthermore, the levels of GCK binding to VDAC decreased slightly in 33.3 mmol/l glucose–treated cells compared with 5.5 mmol/l glucose, but this decrease was not due to changes in GST GCK or VDAC expression in total cell lysates. Concomitant with reduced GCK interaction with VDAC, Bax was overexpressed in 33.3 mmol/l glucose–treated cells (Fig. 3C). In contrast, chronic high glucose significantly inhibited Bad phosphorylation observed in control and cells treated with 33.3 mmol/l glucose for 2 h (Fig. 3D), suggesting that inhibition of Bad phosphorylation and Bax overexpression by chronic high glucose may be involved in the reducing GCK-VDAC interaction.

To clarify whether GCK endogenously interacts with VDAC on the mitochondrial outer membrane and if interactions are reduced by chronic exposure to glucose, GCK-VDAC complexes were immunoprecipitated with anti-GCK. Whereas a basal level of GCK and VDAC interaction was observed in control and cells treated acutely with glucose (1 day and 16 mmol/l), GCK-VDAC interactions were significantly inhibited by chronic high glucose in a time- and dose-dependent manner (Fig. 3E). Similar to this, translocation of GCK to the mitochondria was also inhibited by chronic exposure to glucose in a dose-dependent manner (Fig. 3F). GCK expression in both cytosolic and mitochondrial fractions was significantly decreased by 33.3 mmol/l glucose, whereas no changes were observed in hexokinase I expression in both fractions, indicating that inhibition of GCK translocation to the mitochondria by chronic high glucose may be due to downregulation of GCK.

To confirm that the association between GCK and mitochondria is an important antiapoptotic determinant, we investigated the effects of clotrimazole, an agent known to dissociate hexokinase from mitochondria (21), on apoptosis and the interaction of Bax with VDAC induced by high glucose. Addition of 20 μmol/l clotrimazole to the cells almost completely inhibited the association of GCK and hexokinase I with mitochondria induced by 16 mmol/l glucose in MIN6N8 cells, similar to inhibition of GCK and hexokinase II in HepG2 cells (Fig. 4A). Next, whether detaching hexokinase I or GCK from mitochondria by clotrimazole affects the ability of Bax to induce apoptosis and release cytochrome C in 33.3 mmol/l glucose–treated cells was examined. Cells were stimulated with 50 μmol/l indomethacin, a nonsteroidal anti-inflammatory agent that induces Bax-dependent apoptosis (21,26). Cells treated for 2 h with either indomethacin or clotrimazole induced little DNA fragmentation, but when combined, DNA fragmentation markedly increased. Furthermore, apoptosis induced by chronic exposure to 33.3 mmol/l glucose for 2 days was potentiated by pretreatment with clotrimazole (Fig. 4B). When 33.3 mmol/l glucose was administered to cells pretreated with clotrimazole, Bax binding increased significantly with complete release of cytochrome C compared with 33.3 mmol/l treatment alone (Fig. 4C), similar to the combination of indomethacin and clotrimazole. These results demonstrate that detachment of GCK, as well as hexokinase I, from mitochondria increase the susceptibility of cells to apoptosis induced by chronic high glucose by increasing Bax binding to mitochondria in MIN6N8 cells.

To examine the essential role of GCK in chronic high-glucose–induced apoptosis, MIN6N8 cells were transfected with wild-type GCK cDNA and selected GCK-overexpressed clones (Fig. 5A), followed by treatment with 33.3 mmol/l glucose. DNA fragmentation induced by chronic exposure of cells to 33.3 mmol/l glucose was significantly inhibited by GCK overexpression (Fig. 5C, left). Similarly, GCK overexpression significantly, but not completely, prevented TUNEL-positive apoptotic cells induced by 33.3 mmol/l glucose (Fig. 5C, right). Furthermore, reductions in insulin content and ATP production induced by glucotoxicity were also restored in GCK-overexpressing cells (Fig. 5D). The recovery observed in GCK-overexpressing cells may be due to the sustenance of GCK compared with the complete reduction of GCK by chronic exposure to high glucose in Neo control cells (Fig. 5B). We also have obtained the similar tendentious results in other Neo- or GCK-overexpressing clones (data not shown).

GCK translocation was further confirmed by immunocytochemistry in GCK-overexpressing cells (Fig. 5E). The results show that reduced GCK translocation induced by 33.3 mmol/l glucose in Neo-transfected cells was attenuated in GCK-overexpressing cells. Additionally, GCK was still localized in the mitochondria of GCK-overexpressing cells exposed to 33.3 mmol/l glucose, with a stronger expression compared with Neo-transfected cells (Fig. 5E). Next, we examined whether the strong localization of GCK in the mitochondria induced by GCK overexpression is mediated by the association of GCK with mitochondrial VDAC. As shown in Fig. 5F, reduced GCK association with VDAC induced by 33.3 mmol/l glucose in cells transfected with Neo vector was significantly attenuated in GCK-overexpressing cells. GCK overexpression almost completely inhibited the increase of interaction between Bax and VDAC induced by 33.3 mmol/l glucose in Neo-transfected cells (Fig. 5G). Consistent with these results, Bax oligomerization, which was increased by 33.3 mmol/l glucose in Neo-transfected cells, was also completely inhibited in GCK-overexpressing cells (Fig. 5H), indicating that GCK overexpression inhibits apoptosis induced by high glucose through increased GCK interaction with mitochondria, thereby inhibiting Bax translocation and oligomerization.

Recently, it was reported that phosphorylated Bad preserves mitochondrial integrity by forming a complex with GCK and limiting Bax-induced apoptosis through prevention of Bax interaction with mitochondria and enhancement of glucose metabolism (15,18). Therefore, we have characterized the effects of GCK on the regulation of Bad phosphorylation by chronic high glucose in both Neo- and GCK-overexpressing cells. Acute exposure to 33.3 mmol/l glucose for 2 h increased Bad phosphorylation in both cell groups, with more phosphorylated Bad found in GCK-overexpressing cells (Fig. 6A and B). However, chronic exposure for 48 h significantly decreased Bad phosphorylation in Neo vector–transfected cells, whereas Bad phosphorylation was significantly prolonged in GCK-overexpressing cells. In contrast to Bad phosphorylation, the decrease of phosphorylated Akt observed at 48 h after treatment in Neo vector–transfected cells did not recover in GCK-transfected cells (Fig. 6B). Furthermore, decreases in Bcl-2 and increases in Bax, p53, and p21 proteins by high glucose in Neo-transfected cells were nearly prevented in GCK-overexpressing cells (Fig. 6C). On the other hand, GCK overexpression also restored decreases in Glut2, peroxisome proliferator–activated receptor γ coactivator 1, and sterol regulatory element–binding protein 1 similar to GCK (Fig. 6D).

To determine whether decreases in glucose metabolism and alterations in apoptotic proteins are also responsible for apoptosis of pancreatic islet cells, isolated islet cells were treated with 33.3 mmol/l glucose. Synergistic apoptosis was observed in single-islet cells, which was confirmed by TUNEL assay and DNA fragmentation (Fig. 7A and B). Similar to MIN6N8 cells, GCK significantly decreased in response to 33.3 mmol/l glucose (Fig. 7D), which correlated with decreased insulin content and ATP production (Fig. 7C). Most proapoptotic proteins, including p53, p21, and Bax, increased in 33.3 mmol/l glucose–treated islet cells, whereas Bcl-2 was significantly decreased, correlating with increased release of cytochrome C into the cytosol (Fig. 7D). The decrease in GCK and increase in p53 levels in insulin-positive cells were also confirmed by double immunostaining of islets with anti-GCK, anti-p53, and anti-insulin antibodies (Fig. 7E and F). To clearly define the role of GCK on apoptosis induced by 33.3 mmol/l glucose in isolated islet cells, we also examined the localization and interaction of GCK with mitochondrial VDAC using FITC GCK and MitoTracker CMXRos. As shown in Fig. 8A, GCK green fluorescence was highly expressed and colocalized in mitochondria in 5.5 mmol/l glucose–treated cells, while significantly decreased in 33.3 mmol/l glucose–treated cells. Furthermore, the interaction of GCK with VDAC decreased significantly more in 33.3 mmol/l glucose–treated cells than in 5.5 mmol/l glucose–treated cells (Fig. 8B), which may be due to decreased GCK expression, as its expression was decreased in both mitochondrial and cytosolic fractions (data not shown). In contrast, Bax binding with mitochondrial VDAC increased from treatment with 33.3 mmol/l glucose (Fig. 8C), correlating with increased Bax translocation to mitochondrial fractions (Fig. 8D). Colocalization of Bax with mitochondria significantly increased in cells treated with 33.3 mmol/l glucose compared with 5.5 mmol/l, which was confirmed by immunostaining (Fig. 8E). Similar to these results, Bax oligomerization also increased highly after treatment with 33.3 mmol/l glucose (Fig. 8F), consistent with results obtained from MIN6N8 cells. These results show that GCK may play an important role in regulating apoptosis induced by chronic exposure of pancreatic β islet cells to high glucose.

Reports suggest that elevated glucose concentrations have a dual function on β-cell turnover depending on duration of exposure and the genetic background of the islets (7,14). Although the effects of elevated glucose on β-cell proliferation and apoptosis are still controversial, several studies have demonstrated that chronic exposure of β-cells to high glucose results in β-cell dysfunction and ultimately β-cell death (7,27). During progression of type 2 diabetes, glucotoxicity is likely an important factor that contributes to advancing β-cell failure and development of overt diabetes (28). However, the exact molecular mechanism involved in glucotoxicity-induced β-cell dysfunction and apoptosis is not clearly understood. In this study, we demonstrated that chronic exposure to high glucose induces β-cell apoptosis through decreasing GCK protein expression and interactions with VDAC in the mitochondrial outer membrane, correlating with decreases in Bad phosphorylation. Decreased binding of GCK with mitochondria promotes the binding of Bax with VDAC and, subsequently, Bax oligomerization, cytochrome C release, and apoptosis along with decreased cellular ATP production and insulin secretion. Therefore, based on the pivotal role of GCK on glucotoxicity-induced apoptosis, we believe that GCK involvement is integral for the relationship between glucose metabolism and cell apoptosis in pancreatic β-cells. Although the effects of GCK expression in β-cells has been widely studied, defining the functional role of GCK and the mechanisms regulated by GCK in glucotoxicity-induced β-cell apoptosis appears to be worthwhile.

GCK is the proximal and rate-limiting step in the utilization of glucose and is therefore critical in regulating insulin secretion by β-cells (5,9). Recently, it was suggested that GCK is involved in apoptosis associated with glucose metabolism (15,16), but the exact mechanisms by which GCK is involved in apoptosis are not known. As with hexokinase I, our data show that GCK also endogenously interacts with VDAC on the mitochondrial outer membrane; this was supported by the GST pull down (Fig. 3). Our in vitro–binding GST pull-down assay revealed that GST GCK binds to mitochondrial VDAC (Fig. 3B) and that the interaction between GST GCK and VDAC was reduced by 33.3 mmol/l glucose and correlated with Bax overexpression, but the pull-down GST and total VDAC levels were unaffected by 33.3 mmol/l glucose (Fig. 3C), suggesting that reduced interaction between GCK and VDAC by 33.3 mmol/l glucose is not due to changes in protein level. In contrast to GCK, the expression of hexokinase I and its interaction with VDAC or hexokinase I translocation to the mitochondria were not affected by chronic high glucose (Fig. 3F). Based on these results, GCK binding to VDAC may play a critical role in chronic high-glucose–induced apoptosis, although GCK lacks the NH₂-terminal domain necessary for binding of hexokinase I to VDAC. A different NH₂-terminal domain of GCK from that of hexokinase I may be involved in its interaction with VDAC.

The detachment of GCK or hexokinase I from mitochondria using clotrimazole significantly potentiated apoptosis in high–glucose–and indomethacin-treated cells and correlated with an increase in Bax translocation and cytochrome C release from mitochondria (Fig. 4). This suggests that GCK or hexokinase I protects against Bax-dependent apoptosis and that the imbalance between hexokinases, including GCK and Bax, may regulate cell survival or death of pancreatic β-cells. However, in our system, GCK, rather than hexokinase I, may play a more prominent role in mediating susceptibility to apoptosis by chronic high glucose, since chronic high glucose decreases GCK expression and GCK binding with VDAC but not hexokinase I (Figs. 2 and 3). In further support of a relationship between GCK and protection of apoptosis, the relative expression of GCK in GCK-transfected cells closely corresponded to the ability of GCK overexpression to resist 33.3 mmol/l glucose–induced apoptosis, reductions of GCK-VDAC interactions, and Bax binding to mitochondria and its oligomerization (Fig. 5). However, we cannot exclude the possibile involvement of hexokinase I itself and alternate pathways in regulating apoptosis induced by chronic high glucose, since GCK overexpression did not completely prevent TUNEL-positive apoptotic cells induced by 33.3 mmol/l glucose in Neo-transfected cells (Fig. 5C).

Reportedly, Bad phosphorylation is needed for the formation of a mitochondrially located complex, consisting of GCK and associated with phosphorylated Bad, that enhances glycolysis and prevents apoptosis in liver cells (15,16). Therefore, Bad phosphorylation may also play an important role in glucose metabolism and apoptosis of pancreatic β-cells by regulating the interaction of GCK with mitochondria. Supporting this hypothesis, Bad phosphorylation was significantly increased and prolonged in GCK-overexpressing cells than in Neo vector–transfected cells (Fig. 6). Moreover, hyperglycemia-induced apoptosis and GCK downregulation were significantly inhibited in constitutively phosphorylated Bad-transfected cells but potentiated by constitutively nonphosphorylated Bad (W.-H.K., J.W.L., Y.H.S., M.H.J., unpublished data), suggesting that both Bad phosphorylation and GCK activation may be essential for regulating glucose metabolism in pancreatic β-cells and may be autoregulated by each other rather than have a cause-and-effect relationship. On the other hand, some studies demonstrated that phosphorylation of Bcl-2 family proteins, including Bad, is attenuated by reactive oxygen species (ROS) (29). In our unpublished data, GCK overexpression inhibited ROS production induced by hyperglycemia, concordant with an increase in Bad phosphorylation, suggesting that ROS produced under a hyperglycemic condition may regulate Bad phosphorylation. Interestingly, Bad phosphorylation more sustained in GCK-overexpressed cells than in Neo cells was still dephosphorylated after 4 days (Fig. 6A), indicating that other pathways may possibly be involved in the restoration of GCK overexpression in pancreatic β-cells.

Reportedly, the expression of several genes essential for β-cell function, including GCK, insulin, and GLUT2, are regulated by oxidative stress and stress-activated protein kinase/c-Jun NH₂-terminal kinase activation in obesity (30,31). Similar to this report, oxidative stress triggers p21 induction, leading to suppression of GCK and insulin gene expression in pancreatic cells (32,33), whose expression is inhibited by GCK overexpression. Furthermore, activation of the c-Jun NH₂-terminal kinase pathway may also affect the activity and expression of GCK through pancreatic duodenal homeobox 1, since stress-activated protein kinase/c-Jun NH₂-terminal kinase activation decreases pancreatic duodenal homeobox 1 activity and subsequent suppression of GCK transcription (34). Finally, induction of glycation reactions by hyperglycemia has also been shown to suppress GCK gene expression (35,36), and AMPK, although previously controversial, may be an additional candidate factor in regulating GCK activity and expression in pancreatic β-cells (37,38). Studies on the potential involvement of these signaling proteins are currently under investigation.

Finally, we provide novel insight into mechanisms contributing to chronic hyperglycemia-induced β-cell apoptosis, showing that GCK plays an important role in regulating apoptosis of pancreatic β-cells via competition with Bax to bind with VDAC and suggesting the essential role of GCK on glucose metabolism and cell apoptosis in β-cells.

This study was supported by research grants from the Korean National Institutes of Health (347-6111-211-207).

We thank Dr. M.S. Lee for providing insulinoma cell lines (MIN6N8 cells) and S.S. Kim for technical assistance with isolated pancreatic islet cells. We also thank Dr. H.J. Kim for providing ICR mice and Dr. Van-Anh Nguyen for peer reviewing the manuscript.