|Home | About | Journals | Submit | Contact Us | Français|
Thioredoxin-interacting protein (TXNIP) is a key regulator of diabetic β-cell apoptosis and dysfunction, and TXNIP inhibition prevents diabetes in mouse models of type 1 and type 2 diabetes. Although we have previously shown that TXNIP is strongly induced by glucose, any regulation by the proinflammatory cytokines tumor necrosis factor α (TNFα), interleukin-1β (IL-1β), and interferon γ (IFNγ) has remained largely unexplored. Moreover, even though this three-cytokine mixture is widely used to mimic type 1 diabetes in vitro, the mechanisms involved are not fully understood. Interestingly, we have now found that this cytokine mixture increases β-cell TXNIP expression; however, although TNFα had no effect, IL-1β surprisingly down-regulated TXNIP transcription, whereas IFNγ increased TXNIP levels in INS-1 β-cells and primary islets. Human TXNIP promoter analyses and chromatin immunoprecipitation studies revealed that the IL-1β effect was mediated by inhibition of carbohydrate response element binding protein activity. In contrast, IFNγ increased pro-apoptotic TXNIP post-transcriptionally via induction of endoplasmic reticulum stress, activation of inositol-requiring enzyme 1α (IRE1α), and suppression of miR-17, a microRNA that targets and down-regulates TXNIP. In fact, miR-17 knockdown was able to mimic the IFNγ effects on TXNIP, whereas miR-17 overexpression blunted the cytokine effect. Thus, our results demonstrate for the first time that the proinflammatory cytokines TNFα, IL-1β, and IFNγ each have distinct and in part opposing effects on β-cell TXNIP expression. These findings thereby provide new mechanistic insight into the regulation of TXNIP and β-cell biology and reveal novel links between proinflammatory cytokines, carbohydrate response element binding protein-mediated transcription, and microRNA signaling.
Pancreatic β-cell dysfunction and death play a central role in the development and progression of diabetes and can be caused by multiple stressors including glucotoxicity, gluco-lipotoxicity, and cytokine toxicity (1, 2). Glucotoxicity induced by chronic exposure of β-cells to high levels of glucose promotes β-cell apoptosis, resulting in a vicious cycle with further worsening of the hyperglycemia (1). Interestingly, we previously discovered thioredoxin-interacting protein (TXNIP),2 a ubiquitously expressed cellular redox regulator (3), as the top glucose-induced gene in a human islet gene expression microarray study (4) and found that TXNIP is a crucial mediator of glucotoxicity-induced β-cell apoptosis (5). We further demonstrated that β-cell TXNIP expression is up-regulated in diabetes and that TXNIP-induced β-cell death is mediated by the intrinsic/mitochondrial death pathway (6). Moreover, we recently revealed that elevated TXNIP levels also contribute to β-cell dysfunction by inducing microRNA expression (miR-204), which in turn targets the insulin transcription factor MafA and thereby inhibits insulin production (7). In contrast, TXNIP deficiency led to an increase in functional β-cell mass and protected against diabetes in mouse models of type 1 and type 2 diabetes (8).
Lipotoxicity and especially combined gluco-lipotoxicity is another key factor contributing to β-cell dysfunction and death especially in the context of type 2 diabetes (9). However, unlike glucose, elevated free fatty acids do not increase β-cell TXNIP expression, and TXNIP deficiency does not effectively protect against fatty acid-induced β-cell death (6, 10). This suggests that different β-cell stressors have distinct effects on TXNIP expression and raises the question of how cytokines might affect β-cell TXNIP.
Cytokine toxicity, resulting from an increase in proinflammatory cytokines such as IL-1β, IFNγ, and TNFα, has also been implicated as a critical factor in the β-cell loss of type 1 diabetes (11, 12), but the pathways by which cytokines induce β-cell death are still controversial. Cytokines have been reported to induce apoptosis (11) but also to promote β-cell death by a combination of apoptosis and necrosis that is sensitive to BCL2 (13). In addition, cytokines have been demonstrated to induce endoplasmic reticulum (ER) stress by inhibiting the sarcoendoplasmic reticulum pump calcium ATPase (SERCA) (14). However, whereas knock down of C/EBP homologous protein (CHOP), a key factor in ER stress-induced apoptosis, has been suggested to prevent cytokine-induced β-cell apoptosis (15), there have also been reports indicating that cytokine-induced β-cell death is not dependent on ER stress (16). Proinflammatory cytokines have also been shown to activate the intrinsic mitochondrial apoptosis pathway via disruption of the mitochondrial membrane potential, Bax and caspase-9 activation, and cytochrome c release and to thereby induce β-cell death (13, 17, 18). Moreover, proinflammatory cytokines have been demonstrated to impair β-cell function through reduction of insulin content (19).
Interestingly, TXNIP is induced by diabetes and various stressors and, similar to cytokines, has been shown to promote β-cell apoptosis and dysfunction, but the potential link between proinflammatory cytokines and pro-apoptotic TXNIP has remained largely unexplored. Therefore, the aim of the present study was to investigate whether cytokines might regulate β-cell TXNIP expression and to determine the molecular mechanisms involved.
Rat INS-1 β-cells were cultured in RPMI 1640 medium (Life Technologies) containing 10% FBS, 11.1 mm glucose, 1% penicillin/streptomycin, 1 mm sodium pyruvate, 2 mm l-glutamine, 10 mm HEPES, and 0.05 mm 2-mercaptoethanol, and after overnight incubation at 5 mm glucose, cells were incubated at 25 mm glucose (unless noted otherwise) with or without recombinant rat TNFα (5 ng/ml; R&D system, Minneapolis, MN), mouse IL-1β (1 ng/ml; BD Pharmingen), and/or rat IFNγ (5 ng/ml; R&D system) for 24 h. The generation, culture, and characterization of INS-1 β-cells stably expressing human TXNIP (INS-TXNIP) or LacZ (INS-LacZ) has been described previously (20).
Mouse studies were approved by the University of Alabama at Birmingham Animal Care and Use Committee and comply with the NIH Guide for the Care and Use of Laboratory Animals. Primary pancreatic islets from wild-type C57BL/6 mice were isolated by collagenase digestion method (5), and RNA extraction was performed after incubation with recombinant mouse IL-1β (10 ng/ml; BD Pharmingen).
Isolated human pancreatic islets were obtained from the Integrated Islet Distribution Program (IIDP) and after overnight incubation at 5 mm glucose was gently dispersed by using 0.05% of trypsin-EDTA (7) and cultured with recombinant human IFNγ (1000 units/ml; Peprotech, Rocky Hill, NJ) for 24 h at 25 mm glucose.
Total RNA was extracted using a miRNeasy Mini (Qiagen, Valencia, CA) according to the manufacturer's instructions and reverse-transcribed to cDNA using the first strand cDNA synthesis kit (Roche Diagnostics). Quantitative real-time PCR was carried out on a LightCycler 480 System using SYBR Green (Applied Biosystems, Foster City, CA). 18S ribosomal subunit (Applied Biosystems), TXNIP, rat BCL2, and rat CHOP primers have been described previously (5, 20). Additional primer sequences used were as follows: rat carbohydrate response element-binding protein (ChREBP) forward, 5′-CAAAGCAACCACGCTTCAGA-3′, and reverse, 5′-CCCGCTCCTGCTGTAGCAT-3′; rat liver-type pyruvate kinase (L-PK) forward, 5′-CTTTGATCCAGGCTCTGCAGAC-3′, and reverse 5′-TGAGTCCTGGTTAAAGTATAACC-3′; mouse L-PK forward, 5′-GAAGTGGAACACGGTGGTTTC-3′, and reverse, 5′-TCGGCATTTGGCAAGTTCA-3′; rat Bcl-2-like protein 11 (Bim) forward, 5′-GTCTTCCGCCTCTCGGTAAT-3′, and reverse, 5′-AGAGATACGGATCGCACAGG-3′; human Bim 5′-TTCTTGCAGCCACCCTGC-3′, and reverse, 5′-CTTGCGTTTCTCAGTCCGA-3′. All samples were normalized by 18S run as an internal control. miR-17 expression was assessed in duplicate using a TaqMan microRNA Assay (Applied Biosystems) and U6 (Life Technologies) was used as an internal control.
Protein lysates as well as nuclear and cytoplasmic protein lysates were prepared as detailed previously (5). Protein concentrations were measured by Pierce BCA protein assay (Thermo Fisher Scientific, Hudson, NH), and equal amounts of protein (20–50 μg/lane) were loaded. The following antibodies were used: mouse anti-TXNIP IgG (JY2; catalog no. K0205-3; lot no. 022; 1;1000; MBL, Woburn, MA), rabbit anti-ChREBP IgG (M300; catalog no. sc-33764; lot no. D2412; amino acids 401–700 mapping within an internal region of ChREBP; 1:500; Santa Cruz, Dallas, TX); mouse anti-USF2 IgG (C20; catalog no. sc-862; lot no. A1711; carboxyl terminus of USF2; 1:3000; Santa Cruz), rabbit anti-pIRE1α IgG (catalog no. NB100-2323; lot no. AH-1; IRE1α phosphospecific (Ser-724); 1:500; Novus Biologicals, Littleton, CO); rabbit anti-total IRE1α IgG (catalog no. ab37073; lot no. GR82255-18; 16 amino acids near the carboxyl terminus of IRE1α; 1:1000; Abcam, Cambridge, MA); rabbit anti-BCL2 IgG (catalog no. 2870; lot no. 5; carboxyl terminus of BCl2α 1:3000; Cell Signaling, Danvers, MA); mouse anti-actin IgG (ACTN05 (C4); catalog no. ab3280 lot no. GR207161–1; 1:5000; Abcam); goat anti-mouse IgG-HRP (catalog no. sc-2005; lot no. I1114; 1:5000; Santa Cruz); goat anti-rabbit IgG-HRP (catalog no. sc-2004, lot no. F2215; 1:5000; Santa Cruz). Bands were visualized by ECL plus (GE Healthcare, Buckinghamshire, UK) and quantified by ImageQuant.
Construction of the TXNIP promoter reporter constructs has been described previously (20). INS-1 cells were transfected with promoter reporter constructs containing the full-length human TXNIP promoter as well as different deletions/mutations (0.4 μg/well; Promega, Madison, WI) along with the pRL-TK control plasmid using DharmaFECT Duo transfection reagent (1 μl/well; GE Dharmacon, Lafayette, CO). Luciferase activities were assessed in duplicates using the Dual Luciferase Reporter assay system (Promega).
DharmaFECT1 transfection reagent (GE Dharmacon) was used and for miR-17 loss or gain of function experiments, INS-1 cells were transfected with 25 nm miR-17 inhibitor or hairpin inhibitor negative control 2 or with miR-17 precursor or pre-miR negative control 2 (Applied Biosystem), respectively. For Bim knockdown, cells were transfected with specific rat Bim siRNA or with negative control oligonucleotide (Life Technologies).
INS-1 β-cells were incubated with IL-1β for 6 h, and chromatin immunoprecipitation (ChIP) was performed using 4 μg of anti-ChREBP antibody (P13; Santa Cruz) as described previously (21). After immunoprecipitation, enriched DNA was purified and quantified by quantitative PCR using primers for the TXNIP promoter, forward 5′-CCCAAGAGGAGTCCCCTGGATG-3′ and reverse 5′-GTCAAGCGGCTGCCGGAAACGG-3′.
INS-1 cells were fixed in 4% formaldehyde and permeabilized with 0.2% Triton X-100, PBS for 5 min. The DeadEnd Fluorometric TUNEL System (Promega) was used to detect apoptotic nuclei according to the manufacturer's protocol. The Vectashield with DAPI mounting solution (Vector, Burlingame, CA) was used for nuclei staining.
Caspase 3/7 activity was assessed using the Caspase-Glo 3/7 assay kit and a GLOMAX 20/20 luminometer (Promega) according to the manufacturer's instructions.
The significance of a difference between two groups was calculated using two-tailed Student's t tests, and differences with a p < 0.05 were considered statistically significant. To compare data sets of more than two groups, we used one-way analysis of variance calculations followed by Holm-Sidak tests for multiple comparisons.
Pancreatic β-cells are susceptible to glucotoxicity, lipotoxicity, and cytokine toxicity, all of which can lead to β-cell death and contribute to the loss of functional β-cell mass in the context of diabetes (1, 2). We previously discovered that TXNIP is a key mediator of β-cell apoptosis induced by glucotoxicity (5), whereas TXNIP expression is not induced by free fatty acids or in the context of lipotoxicity (6). Although a combination of the proinflammatory cytokines TNFα, IL-1β, and IFNγ typically used to mimic the conditions of type 1 diabetes in vitro has been shown to induce β-cell apoptosis (11), it has remained largely unknown how cytokines might affect β-cell levels or pro-apoptotic TXNIP. To now address this question we exposed INS-1 β-cells to high (25 mm) glucose and the TNFα/IL-1β/IFNγ cytokine mixture and indeed observed a small but significant increase in TXNIP expression (Fig. 1A). However, the unexpected small effect size raised the question of whether individual cytokines might have different effects on TXNIP expression. We, therefore, next examined the effects of each individual cytokine on TXNIP expression. We observed that TNFα treatment had no effect on β-cell TXNIP expression (Fig. 1B). Surprisingly, we further found that IL-1β treatment resulted in a ~70% decrease in TXNIP mRNA levels (Fig. 1C), whereas IFNγ significantly increased TXNIP mRNA levels >2-fold (Fig. 1D), which is consistent with the mild net increase observed in response to the cytokine combination. Overall, very similar results were also obtained at 5 mm glucose. Exposure to the TNFα/IL-1β/IFNγ cytokine mixture led again to a significant increase in TXNIP expression (Fig. 1E), whereas TNFα again did not increase TXNIP expression (Fig. 1F). Although at 5 mm glucose IL-1β led to an increase in TXNIP expression in INS-1 cells, typically cultured at higher glucose (Fig. 1G), IL-1β again significantly decreased TXNIP expression in primary mouse islets (Fig. 1H) consistent with the findings at 25 mm glucose. Furthermore, as at high glucose, exposure to IFNγ resulted in a >2-fold increase in TXNIP expression in INS-1 cells as well as in primary human islets (Fig. 1, I and J). Taken together, these findings reveal that individual cytokines have differential effects on β-cell TXNIP expression and that IFNγ is the main cytokine contributing to the induction of TXNIP expression.
Because the fact that IL-1β decreased TXNIP expression was surprising, we also performed time-course and dose-response experiments to rule out any potential biphasic effect. However, IL-1β consistently reduced TXNIP expression irrespective of timing or dosing (Fig. 2, A and B). Furthermore, we also confirmed the IL-1β-mediated reduction of TXNIP expression in primary mouse islets (Fig. 2C) and found that TXNIP protein levels were also significantly decreased in response to IL-1β (Fig. 2D).
To identify the cis-acting element responsible for the effect of IL-1β on TXNIP expression, we performed human TXNIP promoter deletion studies and found that mutation of the E-box repeat carbohydrate response element (ChoRE) completely abolished IL-1β-mediated reduction of TXNIP promoter activity (Fig. 3A), indicating that this ChoRE was necessary for the observed inhibition of TXNIP in response to IL-1β. Because we previously identified this ChoRE as the binding site for ChREBP, which is also the transcription factor conferring glucose-induced TXNIP expression (21), these findings suggested that IL-1β might act via inhibition of ChREBP action. ChREBP function is primarily regulated by nuclear entry (22, 23), and whereas IL-1β did not significantly alter total ChREBP protein levels (Fig. 3B), it significantly reduced nuclear ChREBP levels as measured after nuclear fractionation (Fig. 3C). In addition, ChREBP ChIP assays revealed high occupancy at the TXNIP promoter and showed that ChREBP binding to the TXNIP promoter was significantly reduced in response to IL-1β treatment (Fig. 3D), whereas ChREBP binding to the GAPDH negative control was negligible. Together this suggests that IL-1β inhibits nuclear entry of ChREBP and thereby suppresses TXNIP transcription.
Because ChREBP is known to mediate glucose-induced transactivation of L-PK in liver and β-cells (24, 25), we also measured L-PK expression in response to IL-1β. The results showed that IL-1β also down-regulated L-PK expression in INS-1 β-cells and primary mouse islets (Fig. 4, A and B). It also blocked ChREBP binding to the L-PK promoter as assessed using ChREBP ChIP assays (Fig. 4C), indicating that the IL-1β effects on ChREBP-mediated transactivation were not restricted to TXNIP and included other ChREBP target genes.
To further study the effects of IFNγ, we again performed time-course experiments which revealed that TXNIP expression was significantly induced already after 6 h of IFNγ treatment and continued to increase in a time-dependent manner (Fig. 5A). We also confirmed the IFNγ-induced TXNIP expression observed in INS-1 β-cells in primary human islets (Fig. 5B). Consistent with the mRNA expression, we also found that IFNγ significantly increased TXNIP protein levels (Fig. 5C).
To further explore the molecular mechanism underlying IFNγ-mediated up-regulation of TXNIP, we first again performed luciferase assays using the full-length human TXNIP promoter reporter construct in the absence or presence of IFNγ. Surprisingly, IFNγ treatment resulted in a significant reduction in TXNIP promoter activity (Fig. 5D). Consistent with this decrease in TXNIP transcription, and as observed with IL-1β, IFNγ also led to a significant decrease in ChREBP protein levels (Fig. 5E). Together with the observed increase in TXNIP mRNA expression in response to IFNγ (Fig. 5A), these findings indicated that IFNγ-induced TXNIP expression was not conferred at the transcriptional level but might rather be mediated by post-transcriptional regulation, such as RNA stability.
Because RNA stability is often regulated by microRNAs (miRs), and miR-17 in particular has been shown to target TXNIP mRNA (26, 27), we investigated the possibility that IFNγ, by down-regulating miR-17, might increase TXNIP expression. Indeed, using a TaqMan microRNA assay, we found that IFNγ significantly inhibited miR-17 expression levels in INS-1 β-cells (Fig. 6A) as well as in human islets (Fig. 6B). Furthermore, knockdown of miR-17 with anti-miR-17 resulted in a highly significant increase in TXNIP expression, supporting the proposed role of miR-17 (Fig. 6C). Most importantly, overexpression of miR-17 was able to blunt IFNγ-induced TXNIP expression (Fig. 6D), suggesting that IFNγ up-regulated TXNIP expression via inhibition of miR-17.
Based on recent reports that miR-17 is rapidly degraded in response to phosphorylation/activation of IRE1α (28) and that this results in TXNIP induction (26), we hypothesized that IFNγ might induce IRE1α activation leading to the observed decrease in miR-17 and increase in TXNIP. Indeed, phospho-IRE1α protein levels were up-regulated in INS-1 β-cells exposed to IFNγ, whereas total IRE1α levels remained unchanged (Fig. 7A). Because activation of IRE1α is part of the unfolded protein response (UPR) and associated with ER stress, we also measured CHOP, a well established transcriptionally induced marker of ER stress (29). We found that IFNγ also significantly increased CHOP expression in INS-1 β-cells (Fig. 7B). Combined, these findings suggest that IFNγ induces TXNIP expression via induction of ER stress, activation of IRE1α, and degradation of miR-17.
Consistent with the induction of pro-apoptotic TXNIP and CHOP, we also observed that IFNγ down-regulated anti-apoptotic BCL2, as shown by a decrease in BCL2 protein levels (Fig. 7C) and reduced BCL2 expression in INS-1 β-cells (Fig. 7D) as well as primary human islets (Fig. 7E). In addition, TUNEL staining (Fig. 7F) as well as caspase 3/7 activity assays (Fig. 7G) revealed that β-cell apoptosis was significantly increased in response to IFNγ treatment, indicating that this cytokine was able to promote β-cell apoptosis on its own.
Interestingly, the BH3-only BCL2 family protein Bim has recently been reported to play a critical role in IFNγ+TNFα-induced β-cell death (30, 31) and is also known to inhibit BCL2 (32), raising the possibility that Bim might mediate the observed β-cell apoptosis in response to IFNγ. Indeed, Bim expression was significantly induced by IFNγ in INS-1 β-cells as well as in human islets (Fig. 8, A and B), and Bim knockdown completely rescued β-cells from IFNγ-induced apoptosis and resulted in cleaved caspase activity levels that were indistinguishable from baseline (Fig. 8C). Of note, this protective effect was associated with a dramatic decrease in IFNγ-induced TXNIP expression (Fig. 8D). To counteract this decrease in TXNIP expression, we performed parallel experiments in our stably transfected INS-TXNIP cells with constitutive TXNIP expression or in control INS-LacZ cells (Table 1). In control cells, we again observed a 2-fold increase in cleaved caspase 3/7 activity in response to IFNγ or a cytokine mixture, and Bim knockdown again completely blunted this effect, resulting in caspase activity levels identical to those in the no cytokine control. In contrast, although Bim knockdown also resulted in some reduction in apoptosis in INS-TXNIP cells, a highly significant, close to 2-fold increase in cleaved caspase activity in response to IFNγ or cytokine mixture remained in the context of constitutive TXNIP expression (Table 1), suggesting that a decrease in TXNIP is critical for effective protection against cytokine-induced apoptosis.
Taken together, the results of our studies demonstrate that proinflammatory cytokines regulate β-cell TXNIP expression but, surprisingly, show that IL-1β and IFNγ have opposite effects and act via distinct mechanisms, revealing novel links between IL-1β and regulation of gene transcription as well as IFNγ and microRNA expression.
TXNIP is up-regulated in diabetes and plays a key role in β-cell biology, promotes oxidative stress and β-cell apoptosis, and inhibits insulin production and β-cell function (5, 7, 8). Appropriately, its expression has also been shown to be tightly regulated by metabolic stimuli such as glucose (20, 33), by various stresses including ER stress (26), and now, based on the current studies, by proinflammatory cytokines.
The proinflammatory cytokines TNFα, IL-1β, and IFNγ have been implicated in β-cell dysfunction and destruction in type 1 diabetes (11), and a combination of these cytokines has been widely used to mimic the conditions of type 1 diabetes in vitro (34,–36). Our finding that this cytokine mixture-induced β-cell expression of TXNIP is, therefore, consistent with this notion as well as with a previous report of TXNIP expression being mildly increased in response to cytokines (37). On the other hand, the discovery that individual cytokines have different effects on β-cell TXNIP expression has been rather unexpected but helps explain the small net effect size of the cytokine mixture and highlights once again the intricacies of TXNIP regulation. In addition, the findings are in alignment with the previous demonstration that different cytokine combinations, such as IL-1β/IFNγ and IL-1β/TNFα, differentially modify β-cell gene expression profiles (38) and that there are differences in the effects of individual cytokines.
In combination with other cytokines, TNFα has been demonstrated to activate resident macrophages within intact islets and to induce them to secrete IL-1β and thereby to exert some cytotoxic effects on β-cells (39, 40). However, although TNFα has also been shown to contribute to insulin resistance in rat islets, it has not been reported to have any direct toxic effects on β-cells by itself (41), consistent with our observation that TNFα did not increase β-cell expression of pro-apoptotic TXNIP.
In contrast to the induction observed with the cytokine mixture, we surprisingly found that exposure of β-cells or primary islets to IL-1β led to a significant decrease in TXNIP expression irrespective of timing or concentration. Although IL-1β has been implicated in β-cell destruction (42, 43), it has been suggested that IL-1β itself is not sufficient to cause β-cell death and that low concentration or short term treatment with IL-1β may even have beneficial effects on β-cell function and survival (44, 45), which is in alignment with the observed reduction in pro-apoptotic TXNIP expression. Interestingly, TXNIP has previously been shown to be involved in inflammasome activation and to thereby induce IL-1β production (46). It is, therefore, tempting to speculate that the reduction in TXNIP expression in response to IL-1β may represent a negative feedback loop especially in inflammatory cells such as macrophages. If so, one would expect that activation of these cells, which results in IL-1β production as part of the inflammatory response, would lead to a reduction in TXNIP expression. Indeed, TXNIP expression has been demonstrated to be down-regulated in macrophages in response to lipopolysaccharide stimulation or IL-1β exposure, and this effect was found to be mediated by MondoA (47). Again, this is consistent with our current findings revealing that IL-1β acts via inhibition of ChREBP, which is the paralog of MondoA and the major transcription factor regulating TXNIP expression in pancreatic β-cells. Moreover, we found that the effects of IL-1β were not restricted to TXNIP, as IL-1β also down-regulated L-PK, another bona fide downstream target of ChREBP playing an important role in glycolysis. This suggests that IL-1β modulates ChREBP-mediated transcription and signaling and thereby may affect β-cell biology and metabolism in previously unappreciated ways.
Our current studies further demonstrated that IFNγ leads to a potent induction of TXNIP expression in INS-1 β-cells as well as in human islets. Taking into account the lack of an increase by TNFα and the observed inhibition by IL-1β, this suggests that IFNγ is the main contributor to the up-regulation of TXNIP in response to the three-cytokine mixture.
Surprisingly, IFNγ seemed to induce β-cell TXNIP expression at a post-transcriptional level as indicated by the observed increase in mRNA expression but a lack in TXNIP promoter activation. Indeed we found that the effects were mediated by a decrease in miR-17, a microRNA that is capable of targeting and down-regulating TXNIP (26, 27). This notion is further supported by the fact that miR-17 overexpression completely blunted the IFNγ effects on TXNIP expression. Interestingly, we also found that IFNγ effectively activated IRE1α, which has previously been shown to lead to degradation of miR-17 (28), suggesting that IFNγ decreased miR-17 levels via IRE1α. Furthermore, IFNγ also led to a significant increase in CHOP, a downstream factor in the IRE1α UPR (unfolded protein response) signaling pathways and a key marker of ER stress. This is in alignment with previous findings suggesting that proinflammatory cytokines induce ER stress (16). Moreover, exposure to IFNγ also resulted in a marked increase in β-cell apoptosis, consistent with the pro-apoptotic effects of TXNIP (8, 20) and with reports that IFNγ promotes apoptotic signaling (48,–52) and potentiates ER stress-induced β-cell death (53). Interestingly, blockage of IFNγ signaling has been demonstrated to protect non-obese diabetic (NOD) mice against diabetes, further supporting the role of this cytokine in inflammatory β-cell loss (54).
In addition to the induction of pro-apoptotic TXNIP, we discovered that IFNγ also increases Bim expression, a powerful factor that directly induces the apoptotic pathway (32). Our studies further established a critical role for Bim in IFNγ and cytokine-induced β-cell apoptosis, which is consistent with recent reports implicating Bim in β-cell death (30, 31). Although this supports the notion that cytokine-induced β-cell death is mediated by different pathways, some of which may act independently of TXNIP signaling, our results also indicate that for any complete protection against cytokine-induced apoptosis, effective inhibition of TXNIP expression is crucial (Fig. 8, C and D, and Table 1).
In summary, our results reveal for the first time that the proinflammatory cytokines TNFα, IL-1β, and IFNγ each have distinct effects on the pancreatic β-cell and on TXNIP expression. Although IL-1β down-regulated the activity of a major transcription factor, ChREBP, and resulted in decreased TXNIP and L-PK expression, IFNγ inhibited the expression of a microRNA, miR-17, and thereby promoted β-cell expression of TXNIP. Thus, the results of the current studies not only provide novel mechanistic insight into the regulation of TXNIP expression but also reveal several previously unappreciated effects of proinflammatory cytokines on cell signaling and β-cell biology and underline the importance of considering the unique actions of individual cytokines.
K. H. performed and analyzed most of the experiments and drafted the manuscript. G. X. made some of the original observations and helped with the microRNA related studies. T. B. G. isolated the mouse islets and provided technical assistance. A. S. conceived and designed the study and revised the manuscript. All authors reviewed the results and approved the final version of the manuscript.
*This work was supported by National Institutes of Health Grants R01DK-078752 and UC4DK104204. This work was also supported by JDRF Grant 3-SRA-2014-302-M-R. The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
2The abbreviations used are: