Essential Role for Signal Transducer and Activator of Transcription-1 in Pancreatic β-Cell Death and Autoimmune Type 1 Diabetes of Nonobese Diabetic Mice Free

All mice were housed in a specific pathogen-free condition in the vivarium of Samsung Medical Center. STAT1^−/− mice (Taconic) were backcrossed to NOD mice for eight generations. Pups were screened by PCR analysis of tail DNA using specific primer sets for STAT1 or neo. At backcross 7, we conducted nucleotide sequencing of cytotoxic T-cell antigen 4 (CTLA-4) exon 2 to select the pups carrying CTLA-4^NOD (A→G at 77). PCR genotyping showed backcross 7 mice were homozygous for 15 Idd NOD alleles, including major histocompatibility complex and CTLA-4 (21). Mice were considered diabetic if blood glucose levels were >16.7 mmol/l once or >13.9 mmol/l on consecutive measurements. All animal experiments were approved by the institutional animal care and use committee of Samsung Medical Center.

Pancreatic islets were isolated by intraductal collagenase P (Roche) injection and Ficoll gradient separation (17). Hand-picked islets were treated with 0.125% Trypsin-0.05% EDTA to make single cells.

After cytokine treatment (R&D Systems), viability and death of single islet cells were measured by 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) assay and Hoechst33342 (Molecular Probes) staining, respectively (17). Nitrate/nitrite as a measure of NO was determined using the Griess method.

Paraffin-embedded pancreatic sections were immunostained using anti-insulin, anti-glucagon (Chemicon), and anti-somatostatin antibodies (Dako) to identify β-, α-, and δ-cells, respectively (17).

Primary pancreatic islet cells were lysed, and cell fractionation was done as described (22). The purity of the nuclear, heavy membrane and cytosolic fraction was confirmed by Western blotting using anti–poly (ADP-ribose) polymerase, anti-Cox4, and anti–inhibitor of κB antibodies, respectively.

Western blotting was conducted using anti–cytochrome c (PharMingen), anti–X-linked inhibitor of apoptosis (XIAP) (MBL), or anti–inducible NO synthase antibodies (BD Biosciences) (17). For XIAP, single islet cells were preincubated in RPMI 0.2% FCS for 24 h to minimize basal expression (23). Cells were then treated with cytokines for 48 h after pretreatment with 50 μmol/l zVAD-fmk (Enzyme Systems) for 30 min to inhibit possible degradation of XIAP protein by activated caspases (23).

Reverse transcription was performed using specific primer sets based on published cDNA sequences at 58°C annealing temperature for 30 cycles.

Splenocytes from diabetic NOD mice (2 × 10⁷) or diabetic NOD/BDC2.5-SCID mice (2 × 10⁶) (kindly provided by Drs. Benoist and Mathis) were infused into the tail vein of 6- to 8-week-old NOD/STAT1^+/− or NOD/STAT1^−/− mice of the same sex after sublethal irradiation (10).

Neonatal pancreata (<3 days old) from NOD/STAT1^+/− or NOD/STAT1^−/− mice were grafted under the kidney capsule of diabetic or nondiabetic NOD/BDC2.5 mice, as described (4). NOD/BDC2.5 mice were made diabetic by intraperitoneal injection of 200 mg/kg cyclophosphamide before transplantation (24). Four weeks after transplantation, the grafts were dissected for hematoxylin and eosin staining and immunohistochemistry using anti-insulin antibody. Point counting over the entire graft after immunohistochemistry was conducted to estimate the relative β-cell mass in the graft (25).

T-cell activation was assessed by adoptively transferring 5,6-carboxylfluorescein diacetate succinimidyl ester (CFSE) (Molecular Probes)–labeled lymphocytes (2 × 10⁶ cells) and assessing in vivo proliferation as described (26). The pancreatic lymph node (PLN) and mesenteric lymph node (MLN) were harvested 66 h after transfer for flow cytometry gated on CD4⁺ and Vβ4⁺ cells.

CD4⁺ T-cells prepared from the spleens using MACS bead column (Dynal) were stimulated with 5 μg/ml immobilized anti-CD3ε antibody (PharMingen) and 1 μg/ml soluble anti-CD28 antibody (PharMingen) in the presence of 5 ng/ml IL-12 (Sigma) and 10 μg/ml anti–IL-4 antibody (11B11) (for Th1 differentiation) or 1,000 units/ml IL-4, 5 μg/ml anti–IFN-γ antibody (XMG1.2), and 3 μg/ml anti–IL-12 antibody (C-178) (for Th2 differentiation) (27). After 24 h, 50 units/ml IL-2 (R&D Systems) was added for additional culture for 6 days. Equal numbers of cells were restimulated with immobilized anti-CD3ε antibody for 24 h. The supernatant was harvested for enzyme-linked immunosorbent assay of IFN-γ and IL-4 (PharMingen).

Islet cells were pretreated with AG490 (Calbiochem) dissolved in DMSO for 1 h before cytokine treatment. For in vivo experiments, AG490 dissolved in 5% ethanol-40% polyethylene glycol 400 (Sigma)-15% cremophor EL (Sigma)-40% PBS was injected intraperitoneally into NOD/BDC2.5-SCID mice. Control mice were treated with solvent alone.

All results were expressed as means ± SD from more than three independent experiments performed in triplicate. Student's t test was used to compare cell death, NO production, or β-cell mass between groups. ANOVA was used to compare values at multiple points. Scheffé test was used to compare the two groups once ANOVA showed a significant difference. The incidence of diabetes in NOD/STAT1^+/+, NOD/STAT1^+/−, and NOD/STAT1^−/− mice was plotted according to the Kaplan-Meier method, and the development of diabetes between the two groups was compared using the log-rank test. Diabetes incidence at specific time points and the fraction of apoptotic cells were compared using the binominal test. P values <0.05 were considered to represent statistically significant differences.

We have reported that transfection of a dominant-negative STAT1 mutant rendered MIN6N8 insulinoma cells resistant to cell death by IFN-γ/TNF-α (17). Hence, we first studied whether STAT1 is also involved in IFN-γ/TNF-α–induced death of primary pancreatic islet cells in vitro, by treating primary pancreatic islet cells from STAT1^−/− C57BL/6 mice with 100 units/ml IFN-γ + 10 ng/ml TNF-α for 5 days. MTT assay showed that STAT1^−/− islets were remarkably resistant to death by IFN-γ/TNF-α as hypothesized, while primary islets from STAT1^+/− or STAT1^+/+ C57BL/6 mice were susceptible to IFN-γ/TNF-α (P < 0.001 compared with STAT1^+/− or STAT1^+/+ islets) (Fig. 1A). Morphologically, death of primary islet cells from STAT1^+/− or STAT1^+/+ mice by IFN-γ/TNF-α was a classical apoptosis characterized by nuclear condensation/fragmentation (online appendix Fig. 1 [available at http://dx.doi.org/10.2337/db06-1372]), similar to our previous reports (17). Hoechst33342 staining demonstrated that the fraction of cells showing nuclear condensation/fragmentation after treatment with IFN-γ/TNF-α for 5 days was also significantly lower in STAT1^−/− islet cells compared with STAT1^+/− islet cells, consistent with the MTT assay (8.9 ± 1.6% vs. 40.4 ± 2.6%; P < 0.001) (Fig. 1B). Combined immunofluorescent/Hoechst33342 staining showed that most apoptotic cells showing nuclear condensation/fragmentation among STAT1^+/+ islet cells after IFN-γ/TNF-α treatment were insulin⁺ β-cells (>90%) (online appendix Fig. 1). After treatment with IFN-γ/TNF-α for 5 days, 44.7% of β-cells from STAT1^+/+ mice and only 3.7% of β-cells from STAT1^−/− mice underwent apoptosis.

We next studied intracellular events associated with apoptosis. Combined immunofluorescent/Hoechst33342 staining showed that cytochrome c was translocated from mitochondria to cytosol in most apoptotic islet cells showing nuclear condensation/fragmentation after treatment of STAT1^+/+ primary islet cells with IFN-γ/TNF-α for 3 days. In contrast, cytochrome c translocation was rarely observed after the same treatment of STAT1^−/− islet cells (online appendix Fig. 2). Western blotting also showed translocation of cytochrome c from mitochondria to cytoplasm in STAT1^+/− primary islet cells but not in STAT1^−/− primary islet cells after IFN-γ/TNF-α treatment for 3 days (Fig. 1C). To further dissect the mechanism of the resistance of STAT1^−/− islet cells against IFN-γ/TNF-α, we studied the expression of XIAP that could regulate mitochondrial events during apoptosis as a strong inhibitor of caspase-3 or -7 (28,29). The expression of XIAP in islet cells is controlled by IFN-γ and TNF-α as a downstream of nuclear factor-κB (23,30). The expression of XIAP protein in STAT1^+/− islet cells was increased by TNF-α, similar to our previous results utilizing insulinoma cells (Fig. 1D, upper panel) (23,31). When STAT1^+/− islet cells were treated with IFN-γ/TNF-α for 48 h, IFN-γ inhibited the induction of XIAP protein by TNF-α (Fig. 1D, upper panel), which could be due to the inhibitory effect of IFN-γ on translation (32) rather than XIAP cleavage by activated caspases, because islet cells were pretreated with a pancaspase inhibitor, zVAD-fmk, before cytokine treatment (23). When STAT1^−/− islet cells were used, the induction of XIAP protein by TNF-α was not inhibited by IFN-γ (Fig. 1D, upper panel), which indicates that IFN-γ inhibits XIAP protein induction through STAT1, consistent with previous studies (33,34) showing STAT1-mediated inhibition of translation. These results suggest that undiminished XIAP in STAT1^−/− islet cells inhibits caspase-3 and -7, acting as both effector caspases (17) and regulators of mitochondrial events (29), and explains the absence of cytochrome c translocation and cell death in STAT1^−/− islet cells after IFN-γ/TNF-α treatment. In contrast to the protein expression, XIAP mRNA induction after TNF-α treatment was not decreased by IFN-γ in both STAT1^+/− and STAT1^−/− islet cells, indicating that IFN-γ and STAT1 activation act on the translation step (32) (Fig. 1D, lower panel).

We next tested whether STAT1 is involved in islet cell death by IFN-γ/IL-1β, which has been implicated as an effector cytokine combination in type 1 diabetes (11,35). Primary islet cells from STAT1^−/− mice showed a remarkable resistance to treatment with 5 ng/ml IL-1β + 100 units/ml IFN-γ for 5 days that effectively induced STAT1^+/− islet cell death (P < 0.001 compared with STAT1^+/− islet cells) (Fig. 1E). Combined immunofluorescent/Hoechst33342 staining demonstrated that 53.7% of insulin⁺ β-cells from STAT1^+/+ mice underwent apoptosis by IFN-γ/IL-1β treatment for 5 days compared with 4.6% of STAT1^−/− β-cells (online appendix Fig. 1).

Consistent with the abrogation of cell death, inducible NO synthase induction and NO production by IFN-γ/IL-1β were abolished in STAT1^−/− islet cells (P < 0.001 compared with STAT1^+/− islet cells) (Fig. 1F and G). These ex vivo results suggest that STAT1 plays a key role in the susceptibility of primary islet cells to both IFN-γ/TNF-α and IFN-γ/IL-1β combinations that have been implicated as important death effectors.

Because the in vivo role of STAT1 could differ from its in vitro effect, we next produced STAT1-deficient NOD mice by backcrossing STAT1^−/− mice onto the NOD background. During backcrossing, pups homozygous for NOD-type major histocompatibility complex (I-A^g7) and also for NOD-type CTLA-4 allele were selected (data not shown) because CTLA-4 polymorphism has been reported to affect the development of type 1 diabetes in NOD mice as Idd5.1, and the STAT1 locus is tightly linked to the CTLA-4 locus (36). Selected pups were also homozygous for other 13 Idd NOD alleles (data not shown) (21). These mice were intercrossed to obtain STAT1^−/−CTLA-4^NOD/NOD NOD mice (NOD/STAT1^−/− mice). Hematoxylin and eosin staining and immunohistochemistry showed no defect in the development of pancreatic islets and apparently normal distribution of islet cells expressing insulin, glucagon, or somatostatin in STAT1^−/− mice of both C57BL/6 (data not shown) and NOD background at 5 weeks of age (Fig. 2A). Microscopic analysis showed no insulitis or periinsulitis in NOD/STAT1^−/− mice of both sexes until 40 weeks of age (Fig. 2B). Consistent with the absence of insulitis, diabetes never developed in 18 female NOD/STAT1^−/− mice, while it occurred in 66.7% and 53.6% of female NOD/STAT1^+/+ and NOD/STAT1^+/− mice, respectively (P < 0.001 compared with both NOD/STAT1^+/+ and NOD/STAT1^+/− mice) (Fig. 2C). The incidence of diabetes was not significantly different between female NOD/STAT1^+/+ and NOD/STAT1^+/− mice (P > 0.1). Diabetes was also not observed in male NOD/STAT1^−/− mice, in contrast to the development of diabetes in 40.0% and 43.4% of male NOD/STAT1^+/+ and NOD/STAT1^+/− mice, respectively (P < 0.05 compared with both NOD/STAT1^+/+ and NOD/STAT1^+/− mice) (Fig. 2D). These results indicate essential roles for STAT1 in the development of type 1 diabetes in vivo, consistent with the resistance of STAT1^−/− islet cells against cytokines in vitro.

The resistance of STAT1^−/− islet cells against IFN-γ/TNF-α or IFN-γ/IL-1β combination in vitro and the absence of diabetes in NOD/STAT1^−/− mice in vivo suggested that the former result is responsible for the latter. To prove the causality between these, we conducted adoptive transfer of lymphocytes from diabetic NOD mice into irradiated NOD/STAT1^−/− mice. The development of diabetes after adoptive transfer was significantly delayed in NOD/STAT1^−/− mice compared with NOD/STAT1^+/− mice, as judged by Kaplan-Meier curve analysis (P < 0.01). However, the final incidence of diabetes was not significantly different as assessed by the binominal test (P > 0.1) (Fig. 3A). Because the susceptibility of STAT1^−/− islet cells to diabetogenic lymphocytes in vivo despite their resistance to cytokine combinations in vitro could be due to CD8⁺ T-cells equipped with STAT1-independent death effectors such as perforin, we next utilized BDC2.5 CD4⁺ T-cells that carry rearranged T-cell receptor-α and -β genes of a diabetogenic T-cell clone and exert cytotoxicity using contact-independent soluble death mediators (37). When we transferred highly diabetogenic BDC2.5 CD4⁺ T-cells from NOD/BDC2.5-SCID mice (3) into irradiated NOD/STAT1^+/− mice, diabetes occurred 7 days after adoptive transfer, and the incidence of diabetes reached 88.9% (8 of 9) 4 weeks after transfer. In contrast, only 20% of NOD/STAT1^−/− mice (2 of 10) developed diabetes 4 weeks after adoptive transfer of BDC2.5 CD4⁺ T-cells, indicating that STAT1 deficiency in islet cells render them resistant to cell death by diabetogenic CD4⁺ T-cells in vivo (P < 0.005) (Fig. 3B).

To further prove the role of STAT1 deficiency in the resistance against CD4⁺ T-cell–mediated death, neonatal pancreata from NOD/STAT1^−/− mice were transplanted under the kidney capsule of diabetic or nondiabetic NOD/BDC2.5 mice. While control pancreata from neonatal NOD/STAT1^+/− mice were destroyed with extensive inflammation 4 weeks after transplantation, STAT1^−/− neonatal pancreata showed only minimal inflammation around the graft. Immunohistochemistry also showed almost complete absence of insulin⁺ β-cells in STAT1^+/− neonatal pancreas graft but substantial β-cells in STAT1^−/− neonatal pancreas graft (Fig. 3C). Point counting demonstrated that relative β-cell mass in STAT1^−/− neonatal pancreas grafts (n = 4, including those grafted to two diabetic and those to two nondiabetic NOD/BDC2.5 mice) was significantly higher compared with that in STAT1^+/− neonatal pancreas grafts (n = 10, including those grafted to four diabetic and those to six nondiabetic NOD/BDC2.5 mice) (0.0062 ± 0.0044 vs. 0.0002 ± 0.0001) (P < 0.05). Distribution of β-cells was apparently not different between diabetic and nondiabetic recipient mice.

In NOD/STAT1^−/− mice, diabetes never occurred spontaneously; however, islet cells were susceptible to diabetogenic T-cells from NOD mice. Thus, total absence of insulitis may not be explained by the resistance of STAT1^−/− islets against cell death alone and suggests the possible effect of STAT1 deficiency on the immune system, which led us to investigate whether priming of β-cell–specific T-cells by dendritic cells is affected by STAT1 deficiency. We transferred CFSE-labeled naïve BDC2.5 CD4⁺ T-cells into 20 ∼25-day-old NOD/STAT1^−/− or littermate NOD/STAT1^+/− mice and measured T-cell proliferation in response to spontaneous islet cell death during islet development by assessing dilution of the CFSE label (26). As previously reported, prominent BDC2.5 CD4⁺ T-cell proliferation was detected in the PLN but not in the MLN of NOD/STAT1^+/− mice, showing β-cell–specific T-cell priming by dendritic cells in STAT1^+/− mice. In STAT1^−/− mice, the priming of transferred naïve BDC2.5 CD4⁺ T-cells occurred in the PLN but not in the MLN, suggesting that T-cell priming by dendritic cells or the efficiency of dendritic cells to present islet-specific antigens to T-cells is not affected by STAT1 deficiency (Fig. 4A). Next, we investigated Th1/Th2 differentiation, which is critical for the development of type 1 diabetes (38,39) because the induction of T-bet, a key transcription factor for Th1 initiation, is dependent on STAT1 (40,41). When CD4⁺ T-cells from NOD/STAT1^−/− mice were incubated with immobilized anti-CD3ε antibody and soluble anti-CD28 antibody in the presence of IL-12 (Th1) or IL-4 (Th2) and restimulated with anti-CD3ε antibody, IFN-γ production from Th1 cells was significantly decreased compared with NOD/STAT1^+/− mice, suggesting that Th1 differentiation is impaired in NOD/STAT1^−/− mice. On the other hand, IL-4 production from Th2 cells was increased in NOD/STAT1^−/− mice compared with NOD/STAT1^+/− mice (Fig. 4B).

Because these results showed an essential role for STAT1 in type 1 diabetes and potential therapeutic value of STAT1 blockade in the treatment of type 1 diabetes, we tested the effect of AG490, an inhibitor of janus kinase (JAK) 2 upstream of STAT1 (42). As expected, pretreatment with AG490 significantly attenuated the death of wild-type islets by IFN-γ/TNF-α or IFN-γ/IL-1β (P < 0.001 by ANOVA and individual Scheffé test for both cytokine combinations) (Fig. 5A). When administered in vivo, an intraperitoneal injection of 150 μg AG490 14 days after birth for 14 consecutive days significantly delayed the development of diabetes in NOD/BDC2.5-SCID mice, as judged by Kaplan-Meier curve analysis (P < 0.005); however, the final incidence of diabetes was not decreased, as determined by the binomial test (P > 0.1) (Fig. 5B).

Our ex vivo data showing the resistance of STAT1^−/− islet cells against IFN-γ/TNF-α or IFN-γ/IL-1β suggest that STAT1 plays a crucial role in the sensitization of islet cells to TNF-α or IL-1β, which alone does not have significant effect on islet cell viability. These results are consistent with our previous data (17) showing that a dominant-negative STAT1 mutant transfection rendered insulinoma cells resistant to IFN-γ/TNF-α and other studies (43) showing that STAT1^−/− islets were resistant to IFN-γ/IL-1. These results are also compatible with previous studies reporting a reduced diabetes incidence in mice overexpressing suppressor of cytokine signaling 1 (SOCS-1), a negative regulator of STAT1 (44) and increased susceptibility of SOCS-1^−/− islets to cytokines (45). Hence, STAT1 activation in islet cells appears to induce the susceptibility to TNF-α or IL-1β no matter which cytokine combination, IFN-γ/TNF-α or IFN-γ/IL-1β, plays a dominant role in type 1 diabetes in vivo. However, these results are inconsistent with previous studies reporting a protective role of STAT1 against IFN-γ/IL-1β (18). These discrepancies could be due to differences in the types and species of the cells employed.

As a mechanism of the resistance of STAT1^−/− islet cells against IFN-γ/TNF-α, we observed that the inhibition of TNF-α–mediated XIAP protein induction by IFN-γ was abrogated in STAT1^−/− islet cells. The inhibition of translation by nuclear complex comprising STAT1 has been reported (33,34). Because XIAP inhibits effector caspases such as caspase-3 and -7 besides -9 (28), undiminished XIAP protein expression could inhibit the effector phase of islet cell apoptosis, as reported (23). Additionally, the inhibition of caspase-3 or -7 by XIAP could inhibit cytochrome c translocation and other mitochondrial events, as was reported in a recent report (29) demonstrating essential roles for caspase-3 and -7 in the mitochondrial events of apoptosis.

Our in vivo observation that NOD/STAT1^−/− mice never developed diabetes underscores critical roles for STAT1 in type 1 diabetes. The absence of diabetes in NOD/STAT1^−/− mice was not due to the linkage between STAT1 and CTLA-4, a diabetes susceptibility gene on the Idd5 locus (36), because we sequenced the CTLA-4 gene and selected mice with homozygous NOD-type CTLA-4 for further breeding. Our data using a natural type 1 diabetes model of NOD mice differ from the results obtained by multiple low-dose streptozotocin treatment of STAT1^−/− mice of the C57BL/6 background that may cause diabetes by a mechanism distinct from spontaneous type 1 diabetes of NOD mice (43).

Complete absence of insulitis and diabetes in NOD/STAT1^−/− mice could be partly explained by the in vivo resistance of STAT1^−/− islets against cell death because NOD/STAT1^−/− mice were resistant to the development of diabetes after transfer of highly diabetogenic CD4⁺ T-cells from NOD/BDC2.5-SCID mice. However, the resistance of STAT1^−/− islet cells against CD4⁺ effector T-cells alone may not explain the total absence of insulitis and diabetes in NOD/STAT1^−/− mice because, while delayed in onset, diabetes developed after adoptive transfer of lymphocytes from diabetic NOD mice. Indeed, insulitis was unperturbed in perforin^−/− or “class I MHC β-bold” NOD mice, in which CD8⁺ T-cell–dependent cytotoxicity was abrogated (7,46). The development of diabetes in NOD/STAT1^−/− mice after transfer of lymphocytes from diabetic NOD mice could be due to CD8⁺ T-cells that use contact-dependent death arms such as perforin. Full activation of CD8⁺ T-cells in vivo may need islet cell damage by CD4⁺ T-cells or CD4⁺ T-cell help acting through STAT1. Thus, the pancreatic environment in NOD/STAT1^−/− mice may not support the generation or maintenance of the full repertoire of activated effector T-cells, explaining the complete absence of diabetes in NOD/STAT1^−/− mice. However, complete abrogation of insulitis and diabetes in NOD/STAT1^−/− mice might be better explained by the effect of STAT1 on immunity. We have reported that STAT1 plays a critical role in death of macrophages (47) that play an essential role in the development of type 1 diabetes (48). However, we observed that priming of naïve CD4⁺ diabetogenic T-cells occurred in the PLN of NOD/STAT1^−/− mice, suggesting that dendritic cell function is not impaired in NOD/STAT1^−/− mice. Instead, Th1 differentiation was impaired in NOD/STAT1^−/− mice, which can explain the absence of diabetes and insulitis in NOD/STAT1^−/− mice because Th1 environment is important for the development of type 1 diabetes (38,39).

In contrast to the abrogated diabetes in NOD/STAT1^−/− mice, STAT1 deficiency aggravates EAE by reducing the number and function of CD4⁺CD25⁺ Treg-cells (19). The mechanism of the different in vivo effects of the same STAT1 deficiency, depending on the types of autoimmune diseases, is not clearly understood. Differences in the critical effector cytokines inducing target cell death between type 1 diabetes and EAE might account for the dissimilar phenotypes. For instance, the effect of IFN-γ or STAT1 inducing susceptibility to other effector cytokines such as TNF-α may not be critical in oligodendrocyte death (49). In contrast to STAT1 deficiency, targeted disruption of T-bet, downstream of STAT1, prevented EAE, reflecting the complexity of feedback immune regulation in vivo (20). Incomplete protection against diabetes in NOD/BDC2.5-SCID mice by AG490 could be due to its short half-life (50) or aggressive nature of diabetes in NOD/BDC2.5-SCID mice. If other JAK2 inhibitors or STAT1 inhibitors with a longer half-life become available, they may have therapeutic values in the prevention/treatment of type 1 diabetes.

In summary, we have shown unequivocal evidence indicating an essential role for STAT1 in islet cell death in vitro, T-cell immunoregulation, and in the development of type 1 diabetes in vivo. These results suggest a therapeutic potential of STAT1 inhibitors or its upstream JAK inhibitors in the prevention/treatment of type 1 diabetes.

This work was supported by the grants from the KBRDG Initiative Research Program and the 21C Frontier Functional Proteomics Project from the Korean Ministry of Science and Technology (FPR05C1-160). M.-S.L. is an awardee of Science Research Council grants (R01-2005-000-10326-0) from the Korea Science and Engineering Foundation and a Juvenile Diabetes Research Foundation grant (5-2007-361).