Role of Interferon Regulatory Factor-1 in Double-stranded RNA-induced iNOS Expression by Mouse Islets*

Environmental factors, such as viral infection, have been implicated as potential triggering events leading to the initial destruction of pancreatic (cid:1) cells during the development of autoimmune diabetes. Double-stranded RNA (dsRNA), the active component of a viral infection that stimulates antiviral responses in infected cells, has been shown in combination with interferon- (cid:2) (IFN- (cid:2) ) to stimulate inducible nitric oxide synthase (iNOS) expression and nitric oxide production and to inhibit (cid:1) cell function. Interferon regulatory factor-1 (IRF-1), the activation of which is induced by dsRNA, viral infection, and IFN- (cid:2) , regulates the expression of many antiviral proteins, including PKR, type I IFN, and iNOS. In this study, we show that IRF-1 is not required for dsRNA (cid:3) IFN- (cid:2) -stimulated iNOS expression and nitric oxide production by mouse islets. In contrast to islets, dsRNA (cid:3) IFN- (cid:2) fails to induce iNOS expression or nitric oxide production by macrophages isolated from IRF-1 (cid:4) / (cid:4) mice; however, dsRNA (cid:3) IFN- (cid:2) induces similar levels of IL-1 release by macrophages isolated from both IRF-1 (cid:4) / (cid:4) and IRF-1 (cid:3) / (cid:3) mice. Importantly, we show that dsRNA- or dsRNA (cid:3) IFN- (cid:2) -stimulated IRF-1 expression by mouse islets and peritoneal macrophages is independent of PKR. These results indicate that IRF-1 is required for dsRNA (cid:3) IFN- (cid:2) -induced iNOS expression and nitric oxide production by mouse peritoneal macrophages but not by mouse islets. These findings suggest that dsRNA (cid:3) IFN- (cid:2) stimulates iNOS expression by two distinct PKR-independent mechanisms; one that is IRF-1-dependent in macrophages and another that is IRF-1-independent in islets. GGTCCGACAG-3 (cid:4) (PCR product size, 297 base pairs). A standard 25- (cid:5) l PCR reaction of 30 cycles was performed (7), and PCR products were separated on 1.5% agarose gels containing 0.5 (cid:5) g/ml ethidium bromide and visualized by UV light exposure. Statistical Analysis— Statistical comparisons were made between groups using one-way analysis of variance. Significant differences between groups ( p (cid:5) 0.05) were determined by Newman-Keuls posthoc analysis. macrophages. that islet macrophages islet nitric to dsRNA (cid:1) IFN- (cid:3) This in concordance with previous studies showing that dsRNA (cid:1) IFN- (cid:3) induces expression and nitric oxide production by rat islets de-pleted of resident macrophages

Autoimmune diabetes is characterized by a local inflammatory reaction in and around the pancreatic islets of Langerhans, followed by selective destruction of insulin-producing ␤ cells (1,2). This inflammatory reaction consists of macrophages, monocytes, T and B lymphocytes, and natural killer cells that infiltrate into the islet after an initial triggering event to cause ␤ cell destruction (2,3). Although the mechanism associated with the autoimmune reaction leading to the development of diabetes has been studied in detail, few studies have examined the precipitating events that trigger the initial destruction of ␤ cells leading to this autoimmune inflammatory reaction. We and others have shown that IL-1 1 induces an inhibition of insulin secretion and the subsequent destruction of rat islets that is mediated by the local production of nitric oxide by ␤ cells (4,5). One important source of IL-1 within the islet is the resident macrophage, which produces IL-1 on activation by such stimuli as TNF-␣ and LPS (6). Studies have shown that activated intraislet macrophages release IL-1 in sufficient quantities within the microenvironment of the islet to stimulate ␤ cell production of nitric oxide and nitric oxidedependent ␤ cell dysfunction, events that are potentiated by the presence of exogenous IFN-␥ (6,7). Viral infection has been implicated as one environmental factor that may trigger the initial autoimmune reaction that targets and destroys ␤ cells in genetically susceptible individuals (8 -10). Viruses have been isolated from the pancreata of acutely diabetic deceased patients, and viral-specific IgM antibodies have been isolated from newly diagnosed diabetic patients (8,10,11). Autoimmune diabetes can also be induced in genetically susceptible strains of rats and mice by viral infection (8,10,12). Kilham rat virus-induced diabetes in diabetesresistant BioBreeding rats is dependent on the presence of macrophages and associated with the increased expression of the macrophage-derived cytokines IL-12, IL-1␤, and TNF-␣, as well as the T-cell cytokine IFN-␥ (13). In addition, encephalomyocarditis virus-induced diabetes in DBA/2 mice can be attenuated by macrophage depletion (14), daily administration of neutralizing antisera specific for IL-1␤ and TNF-␣, or selective inhibition of iNOS using aminoguanidine (AG) (15). This evidence suggests that viral infection can stimulate diabetes in genetically susceptible rodents and that disease development is dependent on the production of macrophage-and T-cell-derived cytokines and nitric oxide.
One common feature of a viral infection is the formation of dsRNA, which accumulates during viral replication (16). dsRNA is an active component of a viral infection that stimulates host antiviral responses, including the production of cytokines and nitric oxide (17)(18)(19)(20). The synthetic dsRNA molecule poly(I-C) also activates the antiviral response (16) and has been shown to stimulate the development of diabetes in diabetesresistant BioBreeding rats and to accelerate disease development in diabetes-prone BioBreeding rats (21,22). We have shown that dsRNA, in combination with IFN-␥, induces islet dysfunction and destruction by a mechanism that is dependent on the expression of iNOS and production of nitric oxide by ␤ cells (23,24). In addition, dsRNA ϩ IFN-␥ has been shown to stimulate macrophage activation, as evidenced by increased iNOS expression, nitric oxide production, and IL-1 release (23). ␤ cells also produce IL-1 in response to dsRNA ϩ IFN-␥, and ␤-cell production of IL-1 appears to participate in dsRNA ϩ IFN-␥-induced iNOS expression and inhibition of islet function (5). These findings show that dsRNA can modulate islet function by stimulating the production of inflammatory molecules, such as IL-1 and nitric oxide. However, the mechanisms by which dsRNA stimulates the production of these potentially destructive molecules in islets have not been fully examined.
NF-B and interferon regulatory factor-1 (IRF-1) are two transcription factors that participate in gene activation in response to viral infection or dsRNA (25)(26)(27)(28). These transcription factors regulate cellular antiviral responses, including the expression of antiviral proteins (type I IFN, PKR, and major histocompatibility complex class I), control of cell cycle progression, and induction of apoptosis (29 -31). Binding elements for both IRF-1 and NF-B are found in the promoter regions of the rat, mouse, and human iNOS genes (32)(33)(34). In addition, IRF-1 and NF-B have been shown to physically interact at the iNOS promoter region, potentially leading to the synergistic activation of iNOS expression observed when both transcription factors are activated (35). dsRNA-stimulated activation of both NF-B and IRF-1 have been shown to require activation of PKR in mouse embryonic fibroblasts (36). NF-B appears to be required for dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by rat and human islets (37), as well as for iNOS and IL-1 expression by mouse peritoneal macrophages (38).
IRF-1 is a member of a large family of transcription factors activated in response to IFNs. All members of the IRF family of transcription factors bind to the IFN-stimulated response element located in the promoters of many IFN-inducible genes. Some IRFs, such as IRF-1 and ISGF3, activate mRNA transcription, whereas other IRFs, including IRF-2 and ICSBP, repression transcription (39). IRF-1 is an inducible transcription factor expressed in response to IFN-␥, viral infection, dsRNA, and the cytokines IL-1, IL-6, and TNF-␣ (40). The IRF-1 gene promoter region contains binding elements for both NF-B and STAT1 (␥-activated site). Binding of either transcription factor to the IRF-1 promoter region is sufficient to induce IRF-1 expression; however, the presence of both NF-B and STAT1 binding greatly increases IRF-1 transcription (41,42). IRF-1 is required for LPS ϩ IFN-␥-induced iNOS expression by macrophages (40) and appears to participate in IL-1induced nitric oxide production by mouse islets (43,44).
In this study, we examined the role of the inducible transcription factor IRF-1 in dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by both mouse islets and peritoneal macrophages. We show that although IRF-1 is required for dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by mouse macrophages, dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by mouse islets are IRF-1-independent. In addition, the antiviral protein PKR, which is thought to regulate IRF-1 expression in response to dsRNA, is not required for dsRNA ϩ IFN-␥-stimulated IRF-1 expression by mouse islets or peritoneal macrophages. These results suggest that dsRNA ϩ IFN-␥ stimulates iNOS expression by two distinct PKR-independent mechanisms, one that is IRF-1-dependent in macrophages and another that is IRF-1independent in islets.

EXPERIMENTAL PROCEDURES
Materials and Animals-CMRL-1066 tissue culture medium, L-glutamine, penicillin, streptomycin, and rat recombinant IFN-␥ were from Invitrogen. Fetal calf serum was from Hyclone (Logan, UT). RINm5F cells and RPMI 1640 medium (containing 10% fetal calf serum and 1ϫ L-glutamine) were obtained from the Washington University Tissue Culture Support Center (St. Louis, MO). Human recombinant IL-1␤ and mouse recombinant IFN-␥ were from R&D Systems, Inc. (Minneapolis, MN). Poly(I-C) and collagenase type XI were obtained from Sigma. [␥-32 P]dCTP and enhanced chemiluminescence reagents were from Amersham Biosciences, Inc. Horseradish peroxidase-conjugated donkey anti-rabbit IgG was from Jackson ImmunoResearch (West Grove, PA), and rabbit antiserum specific for the C-terminal 27 amino acids of mouse macrophage iNOS was a generous gift from Dr. Thomas Misko (Searle and Co.). Rabbit anti-human IRF-1 was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). iNOS and cyclophilin cDNAs were gifts from Dr. Charles Rodi (Monsanto Corporate Research, St. Louis, MO) and Dr. Steve Carroll (Department of Pathology, University of Alabama, Birmingham, AL), respectively. IL-1␣ and IL-1␤ cDNAs were gifts from Dr. Clifford Bellone (St. Louis University) and have previously described (45). Male C57BL/6(J) (IRF ϩ/ϩ ) mice, male and female IRF-1 Ϫ/Ϫ mice, and male B6129 (C57BL/6 ϫ 129) mice (7-8 weeks) were purchased from The Jackson Laboratory (Bar Harbor, ME). Male and female PKR Ϫ/Ϫ mice on a C57BL/6 ϫ 129 background (46) were a generous gift from Dr. Randal J. Kaufman (University of Michigan Medical Center, Ann Arbor, MI). All other reagents were from commercially available sources.
Islet Isolation-Islets were isolated from male or female mice by collagenase digestion as previously described (47). After isolation, islets were cultured overnight in complete CMRL-1066 (CMRL-1066 containing 2 mM L-glutamine, 10% heat-inactivated fetal calf serum, 100 units/ml penicillin, 10 g/ml streptomycin) under an atmosphere of 95% air and 5% CO 2 at 37°C. Before each experiment, islets were washed three times in complete CMRL-1066, counted, and cultured for 2 additional h at 37°C. No differences in islet function or cytokine signaling were observed between male and female PKR Ϫ/Ϫ or IRF-1 Ϫ/Ϫ mice.
Cell Culture and Peritoneal Macrophage Isolation-RINm5F cells were removed from flasks by trypsin treatment with 0.05% trypsin and 0.02% EDTA for 5 min at 37°C. Cells were washed twice with media and plated at the indicated concentrations. The cells were allowed to adhere to the plates for 2 h under an atmosphere of 5% CO 2 and 95% air before the initiation of experiments. Peritoneal exudate cells were obtained from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ mice by lavage as previously described (48). After isolation, cells were cultured in complete CMRL-1066 and incubated for 2 h under an atmosphere of 95% air and 5% CO 2 . The cells were washed three times with complete CMRL-1066 to remove nonadherent cells before initiation of experiments.
Nitrite and IL-1 Determinations-Nitrite formation was determined by mixing 50 l of culture medium with 50 l of Griess reagent (49). Absorbance was measured at 540 nm, and nitrite concentrations were calculated from a sodium nitrite standard curve. IL-1 release was measured using the RINm5F cell bioassay as previously described (50).
Statistical Analysis-Statistical comparisons were made between groups using one-way analysis of variance. Significant differences between groups (p Ͻ 0.05) were determined by Newman-Keuls posthoc analysis.

Effects of Cycloheximide on dsRNA-induced IL-1␤ mRNA
Accumulation by Mouse Islets-Heitmeier et al. (5) have shown that new protein synthesis is required for dsRNA ϩ IFN-␥induced iNOS mRNA accumulation in rat islets, an event that is mediated, in part, by intraislet expression and release of IL-1. To determine whether new protein synthesis is also required for dsRNA ϩ IFN-␥-induced IL-1 expression, we examined the effects of cycloheximide on IL-1␤ mRNA accumulation in mouse islets by RT-PCR. Islets were treated for 30 min with or without 10 M cycloheximide and then incubated an additional 18 h with 400 g/ml poly(I-C) and 150 units/ml IFN-␥. As shown in Fig. 1, both poly(I-C) and poly(I-C) ϩ IFN-␥ stimulate IL-1␤ mRNA accumulation in mouse islets, and this expression is prevented by cycloheximide. Alone, IFN-␥ does not stimulate accumulation of IL-1␤ mRNA. Treatment of islets with 20 g/ml LPS ϩ 150 units/ml IFN-␥ for 18 h was used as a positive control for IL-1␤ mRNA expression (38). These results indicate that new protein synthesis is required for dsRNA-induced IL-1␤ expression by mouse islets.
Time Dependence of dsRNA-stimulated IRF-1 Expression by Mouse Islets-Unlike many constitutively expressed members of the IRF family of transcription factors, IRF-1 expression is regulated at the level of mRNA transcription (39), and viral infection, dsRNA, and inflammatory cytokines have been shown to stimulate IRF-1 expression (40). Because dsRNA-and dsRNA ϩ IFN-␥-induced IL-1 expression require new protein synthesis, the effects of dsRNA on IRF-1 expression by islets were examined. Alone, 400 g/ml poly(I-C) stimulates IRF-1 expression, with maximal accumulation of the transcription factor after a 3-h incubation. After 7-9 h, poly(I-C)-stimulated IRF-1 accumulation is reduced to control levels (Fig. 2a). The level of expression is similar in magnitude to that seen in response to a 3-h treatment of mouse islets with 150 units/ml IFN-␥, a positive control of IRF-1 expression. Importantly, treatment of islets with both poly(I-C) and IFN-␥ for 3 h stimulates IRF-1 expression to levels that appear to be equivalent to the additive effects of poly(I-C) and IFN-␥ alone (Fig. 2b). These findings suggest that poly(I-C) and IFN-␥ stimulate IRF-1 expression by distinct pathways, an observation that is in agreement with the synergism between STAT1 (activated by IFN-␥) and NF-B (activated by poly(I-C)) in the expression of IRF-1 by HepG2 cells (42).
PKR Is Not Required for dsRNA-induced IRF-1 Expression by Mouse Islets-dsRNA-stimulated IRF-1 expression has been shown to require activation of PKR in mouse embryonic fibroblasts (36). To determine whether PKR is required for dsRNAinduced IRF-1 expression by islets, we examined IRF-1 expression by islets isolated from mice devoid of PKR. Alone, dsRNA and IFN-␥ stimulate IRF-1 expression to similar levels in islets isolated from PKR Ϫ/Ϫ and PKR ϩ/ϩ mice (Fig. 3). In addition, the combined effect of dsRNA and IFN-␥ on IRF-1 expression is not altered in islets isolated from PKR Ϫ/Ϫ mice compared with PKR ϩ/ϩ mice. These results indicate that PKR is not required for dsRNA-induced IRF-1 expression in mouse islets, a finding consistent with studies showing that PKR is not required for dsRNA-stimulated NF-B activation or dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by mouse islets (37).
Effects of dsRNA ϩ IFN-␥ on iNOS Expression and Nitric Oxide Production by Islets Isolated from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ Mice-The role of IRF-1 in dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production was examined using islets isolated from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ mice. Islets isolated from IRF-1-deficient mice produce nitrite in response to 400 g/ml poly(I-C) plus 150 units/ml IFN-␥, and the levels are similar in magnitude to the levels produced by islets isolated from IRF-1 ϩ/ϩ mice (Fig. 4a). Consistent with nitrite production, poly(I-C) ϩ IFN-␥ stimulates iNOS mRNA accumulation (data not shown) and protein expression (Fig. 4b) to similar levels in islets isolated from both IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ mice. These levels are similar in magnitude to those induced in response to 15 units/ml hIL-1␤ plus 150 units/ml IFN-␥, a positive control for iNOS expression. As expected, neither poly(I-C) nor IFN-␥ alone stimulates nitrite production or iNOS expression by mouse islets (37). Fig. 4c shows that IRF-1 is not expressed by islets isolated from IRF-1 Ϫ/Ϫ mice in response to poly(I-C), IFN-␥, or poly(I-C) ϩ IFN-␥. These results suggest that IRF-1 is not required for dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by mouse islets.
Effects of dsRNA ϩ IFN-␥ on IL-1␤ Expression by Islets Isolated from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ Mice-We have shown that ␤ cells, in addition to resident macrophages, are a source of IL-1␤ within islets in response to dsRNA (5). We therefore examined IL-1␤ mRNA accumulation in islets isolated from

PKR Is Not Required for dsRNA ϩ IFN-␥-stimulated IRF-1 Expression by Peritoneal Macrophages-Previous studies have
shown that dsRNA ϩ IFN-␥ stimulates macrophage activation, inducing iNOS expression, nitric oxide production, and IL-1 release, events that are PKR-independent in primary macrophages (38,54). Therefore, the effects of dsRNA ϩ IFN-␥ on IRF-1 expression by peritoneal macrophages isolated from PKR Ϫ/Ϫ mice were examined. Treatment of macrophages with 150 units/ml IFN-␥ stimulates maximal IRF-1 expression after 90 min (Fig. 6); however, IRF-1 is not expressed in response to poly(I-C) alone. This finding is in contrast to the stimulating action of poly(I-C) on IRF-1 expression by mouse islets (Fig. 2). Similar to mouse islets, treatment of macrophages with poly(I-C) ϩ IFN-␥ stimulates an increase in IRF-1 expression that is greater than the expression seen in response to IFN-␥ alone. Importantly, IFN-␥ and poly(I-C) ϩ IFN-␥ stimulate IRF-1 expression to similar levels in both PKR Ϫ/Ϫ and PKR ϩ/ϩ mice. These results indicate that PKR is not required for dsRNA ϩ IFN-␥-induced IRF-1 expression by mouse peritoneal macrophages, a finding that is in contrast to studies showing that PKR is required for dsRNA-stimulated IRF-1 expression by mouse embryonic fibroblasts (36). These results also indicate that although dsRNA does not stimulate IRF-1 expression in mouse peritoneal macrophages, dsRNA does appear to potentiate the effects of IFN-␥ on IRF-1 expression.
Effects of dsRNA ϩ IFN-␥ on iNOS Expression and Nitric Oxide Production by Peritoneal Macrophages Isolated from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ Mice-Peritoneal macrophages isolated from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ mice were used to determine whether IRF-1 is required for dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by macrophages. As shown in Fig. 7a, neither poly(I-C) nor IFN-␥ alone induces nitrite production by mouse macrophages, consistent with a requirement for two signals for iNOS expression by primary macrophages (38,54). Importantly, poly(I-C) ϩ IFN-␥ fails to stimulate nitrite production by macrophages isolated from IRF-1 Ϫ/Ϫ mice compared with an ϳ8-fold increase in nitrite production by macrophages isolated from IRF-1 ϩ/ϩ mice. Sim- FIG. 3. Effects of poly(I-C) and IFN-␥ on IRF-1 expression by islets isolated from PKR ؊/؊ and PKR ؉/؉ mice. Islets isolated from PKR Ϫ/Ϫ and PKR ϩ/ϩ mice (120/400 l of complete CMRL-1066 medium) were incubated with 400 g/ml poly(I-C) or 150 units/ml IFN-␥ for 3 h at 37°C as indicated. The islets were isolated, and IRF-1 protein expression was determined by Western blot analysis. Results for IRF-1 protein expression are representative of three independent experiments.

FIG. 4. Effects of poly(I-C) ؉ IFN-␥ on nitrite production and iNOS expression by islets isolated from IRF-1 ؊/؊ and IRF-1 ؉/؉ mice.
Islets isolated from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ mice (120/400 l of complete CMRL-1066 medium) were incubated with the indicated concentrations of poly(I-C), IFN-␥, and IL-1␤ for 40 h at 37°C. After treatment, the medium was removed for nitrite determination (a), and iNOS expression was determined by Western blot analysis (b). Islets isolated from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ mice (120/400 l of complete CMRL-1066 medium) were treated with the indicated concentrations of poly(I-C) and IFN-␥ for 3 h at 37°C, and IRF-1 protein expression was examined by Western blot (c). Results for nitrite are the average Ϯ S.E. of three independent experiments, and iNOS and IRF-1 protein expression are representative of three and two independent experiments, respectively. ‫,ء‬ p Ͻ 0.05 versus untreated control.

FIG. 5. Effects of poly(I-C)
؉ IFN-␥ on IL-1␤ mRNA accumulation by islets isolated from IRF-1 ؊/؊ and IRF-1 ؉/؉ mice. a, islets isolated from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ mice (120/400 l of complete CMRL-1066 medium) were incubated with the indicated concentrations of poly(I-C), IFN-␥, or LPS for 18 h at 37°C. After treatment, total RNA was isolated from islets, cDNA was generated by first-strand synthesis, and IL-1␤ and GAPDH (loading control) mRNA accumulation was determined by RT-PCR. b, islets isolated from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ mice (500/ml of complete CMRL-1066 medium) were incubated with the indicated concentrations of poly(I-C), IFN-␥, or LPS for 18 h at 37°C. After treatment, total RNA was isolated from islets, and IL-1␤ and cyclophilin (loading control) mRNA accumulation was determined by Northern blot. c, islets isolated from IRF-1 Ϫ/Ϫ mice (120/400 l of complete CMRL-1066 medium) were treated with 10 M cycloheximide (CHX) for 30 min and then incubated for an additional 18 h with the indicated concentration of poly(I-C), IFN-␥, or LPS at 37°C. After treatment, total RNA was isolated from islets, cDNA was generated by first-strand synthesis, and IL-1␤ and GAPDH (loading control) mRNA accumulation was determined by RT-PCR. LPS ϩ IFN-␥-induced IL-1␤ mRNA accumulation by resident macrophages was used as a positive control for IL-1␤ mRNA accumulation (a-c). Results for IL-1␤ mRNA accumulation as examined by RT-PCR are representative of three independent experiments, and results for IL-1␤ mRNA accumulation as examined by Northern blot are the average Ϯ S.E. of three independent experiments.
ilarly, treatment of macrophages with poly(I-C) ϩ IFN-␥ fails to stimulate iNOS mRNA accumulation or protein expression in macrophages devoid of IRF-1, as determined by Northern blot and Western blot analyses, respectively (Fig. 7, b and c). These results suggest that IRF-1 is required for dsRNA ϩ IFN-␥induced iNOS expression and nitric oxide production by mouse peritoneal macrophages.

DISCUSSION
Both genetic and environmental determinants are thought to participate in the development of autoimmune diabetes. Viral infection has been implicated as one environmental factor that may trigger the initial destruction of ␤ cells during the development of diabetes (8 -10). Analysis of animal models suggests that inflammatory cytokines such as IL-1, TNF-␣, and IFN-␥, as well as nitric oxide, may participate in the development of viral-induced diabetes (13). dsRNA is an active component of a viral infection that stimulates antiviral responses in infected cells (16). Treatment of rat and human islets and primary ␤ cells with dsRNA ϩ IFN-␥ results in a potent inhibition of glucose-stimulated insulin secretion and islet degeneration, events that require ␤ cell production of nitric oxide (23). However, the signaling mechanisms required for dsRNA ϩ IFN-␥induced iNOS expression and nitric oxide-dependent islet dysfunction have yet to be defined.
In this study, we examined the role of the inducible transcription factor IRF-1 in dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by mouse islets. We show that dsRNA (in the form of poly(I-C)) stimulates IRF-1 expression by mouse islets to levels similar to those stimulated by IFN-␥. Importantly, the level of IRF-1 expressed in response to the combination of dsRNA and IFN-␥ appears to be equivalent to the additive effects of either dsRNA or IFN-␥ alone on IRF-1 expression. These findings suggest that dsRNA and IFN-␥ stimulate IRF-1 expression by distinct pathways. Indeed, the promoter region of the IRF-1 gene contains binding elements for both NF-B (B site) and STAT1 (␥-activated site; Refs. 41,42). These binding elements overlap, creating a composite ␥-activated site-B promoter element to which NF-B and STAT1 bind in a mutually exclusive and independent manner (42). dsRNA has been shown to activate NF-B in several cell types, including islets (37,38,46), whereas STAT1 is activated in response to IFN-␥ (55). Simultaneous activation of NF-B and STAT1 (by TNF-␣ and IFN-␥, respectively) has been shown to stimulate the synergistic induction of IRF-1 that is mediated by the composite ␥-activated site-B element in HepG2 cells (42). Our observations with dsRNA and IFN-␥ are consistent with these findings, suggesting cooperation between NF-B and STAT1 that leads to increased IRF-1 expression by islets.
Islets contain 10 -15 resident macrophages that may contribute to ␤ cell destruction by producing the inflammatory cytokine IL-1 and nitric oxide (56). Arnush et al. (6,7) showed that activation of resident macrophages within the microenvironment of the islet leads to iNOS expression and nitric oxide production by ␤ cells by a mechanism that is dependent on macrophage IL-1 release. In addition, dsRNA ϩ IFN-␥ stimulates macrophage activation, as evidenced by increased iNOS FIG. 6. Effects of poly(I-C) and IFN-␥ on IRF-1 expression by mouse peritoneal macrophages. Peritoneal macrophages isolated from PKR ϩ/ϩ and PKR Ϫ/Ϫ mice (4 ϫ 10 5 /400 l of complete CMRL-1066 medium) were treated with 50 g/ml poly(I-C) and 150 units/ml IFN-␥ as indicated for 90 min at 37°C. Macrophages were collected, and IRF-1 expression was determined by Western blot. Results for IRF-1 protein expression are representative of three independent experiments.

FIG. 7. Effects of poly(I-C) ؉ IFN-␥ on iNOS expression and nitrite production by peritoneal macrophages isolated from
IRF-1 ؊/؊ and IRF-1 ؉/؉ mice. Peritoneal macrophages (4 ϫ 10 5 /400 l of complete CMRL-1066 medium) were treated with 50 g/ml poly(I-C) and 150 units/ml IFN-␥ for 24 h at 37°C as indicated. Nitrite production was determined on the culture media (a), and iNOS protein expression was determined by Western blot (c). b, peritoneal macrophages (4 ϫ 10 5 /400 l of complete CMRL-1066 medium) were treated with 50 g/ml poly(I-C) and 150 units/ml IFN-␥ for 6 h at 37°C as indicated. Total RNA was isolated from macrophages, and iNOS mRNA accumulation was determined by Northern blot. 28 S RNA levels were detected by ethidium bromide staining as an RNA loading control. Results for nitrite production are the average Ϯ S.E. of three independent experiments. Results for iNOS protein expression and mRNA accumulation are representative of three independent experiments. ‫,ء‬ p Ͻ 0.05 versus untreated control.

IRF-1-independent iNOS Expression by Mouse Islets
expression, nitric oxide production, and IL-1 release (38,54). In this study, we also examined the role of IRF-1 in the activation of mouse peritoneal macrophages. We show that dsRNA fails to stimulate IRF-1 expression by macrophages. This is in contrast to mouse islets, where IRF-1 expression is greatly increased after a 3-h treatment. However, dsRNA appears to potentiate IFN-␥-stimulated IRF-1 expression by macrophages. This is similar to findings in mouse islets, where maximal IRF-1 expression requires the activation of two transcription factors (NF-B and STAT1).
IRF-1 is required for LPS ϩ IFN-␥-induced iNOS expression by macrophages, because LPS ϩ IFN-␥ fails to induce iNOS expression in macrophages isolated from IRF-1-deficient mice (40). We show that dsRNA ϩ IFN-␥ also fails to induce iNOS expression and nitric oxide production by peritoneal macrophages isolated from IRF-1 Ϫ/Ϫ mice. This finding is in contrast to mouse islets, where dsRNA ϩ IFN-␥ induces iNOS expression and nitric oxide production in the absence of IRF-1. These results suggest that dsRNA ϩ IFN-␥ stimulates iNOS expression by two distinct mechanisms, one that is IRF-1-dependent (macrophages) and another that is IRF-1-independent (islets).
Although IRF-1 is required for dsRNA ϩ IFN-␥-induced nitric oxide production by macrophages, we do not see a decrease in dsRNA ϩ IFN-␥-induced nitric oxide production by islets isolated from IRF-1 Ϫ/Ϫ mice, which contain resident macrophages. These results imply that resident islet macrophages do not contribute significantly to total islet nitric oxide production in response to dsRNA ϩ IFN-␥. This finding is in concordance with previous studies showing that dsRNA ϩ IFN-␥ induces iNOS expression and nitric oxide production by rat islets depleted of resident macrophages (23).
One mechanism by which macrophages may contribute to dsRNA ϩ IFN-␥-induced islet dysfunction is by the production and release of IL-1, which may then stimulate ␤ cells to express iNOS and produce nitric oxide. We show that dsRNA ϩ IFN-␥ induces IL-1 release to similar levels in peritoneal macrophages isolated from IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ mice. ␤ cells also produce and release IL-1 in response to dsRNA ϩ IFN-␥. Recent studies have shown that ␤ cell production of IL-1␤ partially mediates dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by rat islets and primary ␤ cells (5). In addition, dsRNA ϩ IFN-␥-induced nitric oxide production is reduced by ϳ50% in islets isolated from mice deficient in the type I IL-1 receptor. 2 We show that dsRNA ϩ IFN-␥ induces IL-1 expression by peritoneal macrophages and islets and IL-1 release by peritoneal macrophages isolated from both IRF-1 Ϫ/Ϫ and IRF-1 ϩ/ϩ mice. Importantly, cycloheximide prevents dsRNA ϩ IFN-␥-induced IL-1␤ mRNA accumulation by islets isolated from IRF-1 Ϫ/Ϫ mice. This implies that an inducible factor other than IRF-1 is required for dsRNA ϩ IFN-␥-induced IL-1 expression by islets. Potential factors whose expression may be induced in islets by dsRNA are currently being examined by our laboratory.
We have shown that dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by mouse islets and peritoneal macrophages do not require the antiviral kinase PKR (37,54). In addition, we have shown that dsRNA-stimulated NF-B activation in both islets and peritoneal macrophages is PKRindependent (37,54). Consistent with these findings, we show that dsRNA and dsRNA ϩ IFN-␥-stimulated IRF-1 expression is also independent of PKR. These results are in contrast to previous studies showing that PKR is required for dsRNAinduced IRF-1 and NF-B activation by mouse embryonic fibroblasts (36,46). The mechanisms responsible for the differ-ences in dsRNA responsiveness of embryonic fibroblasts compared with islets or macrophages are unknown. However, recent studies by Iordanov et al. (57) question the role of PKR in dsRNA-stimulated NF-B activation in mouse embryonic fibroblasts.
In summary, we have examined the potential role of the inducible transcription factor IRF-1 in dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by both mouse islets and peritoneal macrophages. We show that dsRNA (or dsRNA ϩ IFN-␥) stimulates IRF-1 expression by mouse islets and peritoneal macrophages and that this expression is PKR-independent. Importantly, we show that IRF-1 is not required for dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by mouse islets. In contrast, dsRNA ϩ IFN-␥-induced iNOS expression and nitric oxide production by mouse peritoneal macrophages requires IRF-1 expression. We also show that IRF-1 is not required for dsRNA-induced IL-1 expression by mouse peritoneal macrophages or islets. These results suggest that IRF-1 plays a minimal role in mediating the destructive effects of dsRNA and IFN-␥ on mouse islets.