The Transcription Factor C/EBP delta Has Anti-Apoptotic and Anti-Inflammatory Roles in Pancreatic Beta Cells

In the course of Type 1 diabetes pro-inflammatory cytokines (e.g., IL-1β, IFN-γ and TNF-α) produced by islet-infiltrating immune cells modify expression of key gene networks in β-cells, leading to local inflammation and β-cell apoptosis. Most known cytokine-induced transcription factors have pro-apoptotic effects, and little is known regarding “protective” transcription factors. To this end, we presently evaluated the role of the transcription factor CCAAT/enhancer binding protein delta (C/EBPδ) on β-cell apoptosis and production of inflammatory mediators in the rat insulinoma INS-1E cells, in purified primary rat β-cells and in human islets. C/EBPδ is expressed and up-regulated in response to the cytokines IL-1β and IFN-γ in rat β-cells and human islets. Small interfering RNA-mediated C/EBPδ silencing exacerbated IL-1β+IFN-γ-induced caspase 9 and 3 cleavage and apoptosis in these cells. C/EBPδ deficiency increased the up-regulation of the transcription factor CHOP in response to cytokines, enhancing expression of the pro-apoptotic Bcl-2 family member BIM. Interfering with C/EBPδ and CHOP or C/EBPδ and BIM in double knockdown approaches abrogated the exacerbating effects of C/EBPδ deficiency on cytokine-induced β-cell apoptosis, while C/EBPδ overexpression inhibited BIM expression and partially protected β-cells against IL-1β+IFN-γ-induced apoptosis. Furthermore, C/EBPδ silencing boosted cytokine-induced production of the chemokines CXCL1, 9, 10 and CCL20 in β-cells by hampering IRF-1 up-regulation and increasing STAT1 activation in response to cytokines. These observations identify a novel function of C/EBPδ as a modulatory transcription factor that inhibits the pro-apoptotic and pro-inflammatory gene networks activated by cytokines in pancreatic β-cells.


Introduction
Type 1 diabetes (T1D) is a multi-factorial disease where a chronic autoimmune assault results in a progressive b-cell loss and increased circulating blood glucose levels [1,2]. The recent discovery of numerous T1D-associated susceptibly genes [3,4], as well as T1D-predisposing environmental factors [5,6], added new layers of complexity to our understanding of the disease. Pancreatic islet infiltration by activated immune cells and the development of an aberrant islet inflammation (insulitis) are assumed to represent common events in early T1D [1,2,7]. A detailed understanding of early insulitis, during which infiltrating autoimmune cells induce b-cell apoptosis and inflammation [1,8], may indicate novel and rational approaches for therapeutic interventions [9][10][11].
Further analysis of our microarray data pointed out to an early induction of the transcription factor CCAAT/enhancer binding protein delta (C/EBPd) in cytokine-treated b-cell via NF-kB and STAT1 activation [13,15,16]. The role for this transcription factor in b-cell, however, remains to be clarified. The C/EBP family consists of six transcription factors (a, b, c, d, e and f) sharing a highly conserved basic leucin zipper domain at the C-terminal region of the protein; this domain is involved in homo-or heterodimerization and in DNA binding activity [28]. C/EBPd expression is induced in other cell types in response to various stimuli, including mitogens, hormones, toxins and cytokines (IL-1b, IL-6, IFN-c), and is mostly regulated at the transcriptional level [28]. Unlike C/EBPa, b and e that exist as different splicing variants displaying diverse functions [29,30], only one C/EBPd isoform has been identified in rodents and humans [28]. C/EBPd dimerises with several members of the C/EPB family (a, b and f) but also with NF-kB1 p50, RelA, and the Ets family member PU.1. [31][32][33][34], allowing it to exert various functions in different cell types. C/EBPd activities have been associated with adipocytes differentiation [35], learning and memory processes in neurons [36], tumor suppressor activities in mammary gland epithelial cells [37,38] and with Toll-like Receptor-mediated production of proinflammatory cytokines in macrophages [39], but much less is known about this transcription factor as compared to other members of the C/EBP family [28].
We presently report that C/EBPd is expressed in rat insulinoma cells, primary rat b-cells and human islets, and that its expression is up-regulated upon exposure to IL-1b+IFN-c. Using several single and combined siRNA-mediated knockdown approaches, we demonstrate that C/EBPd deficiency exacerbates cytokineinduced b-cell demise by promoting pro-apoptotic and proinflammatory signalling pathways. Likewise, C/EBPd overexpression partially protects b-cells against cytokine-induced apoptosis. All together, these results identify C/EBPd as a new transcription factor that exerts unique and non-redundant functions in b-cells by negatively regulating the deleterious effects of pro-inflammatory cytokines.

The cytokines IL-1b and IFN-c up-regulate C/EBPd mRNA and protein expression in INS-1E cells, primary rat b-cells and human islets
Our previous microarray data suggest that C/EBPd expression is modified by cytokine treatment in rat b-cells [13,15,16,40]. In order to validate these observations, we evaluated C/EBPd mRNA and protein expression in the rat insulinoma INS-1E cells, in primary FACS-sorted rat b-cells and human islets exposed to cytokines. C/EBPd mRNA expression was up-regulated already after 4 h of IL-1b+IFN-c treatment in INS-1E cells, reaching a peak at 8 h (6-fold increase) and remaining increased until 24 h. On the other hand, the combination of TNF-a+IFN-c did not induce C/EBPd expression at any of the tested time points ( Figure 1A). Treatment of INS-1E cells for 8 h with each individual cytokine identified IL-1b as the key inducer of C/ EBPd expression; IFN-c had no effect on its own, but it amplified IL-1b-induced C/EBPd up-regulation ( Figure 1B). Western blot analysis in INS-1E cells corroborated these observations, since C/ EBPd proteins were up-regulated by respectively 2-and 5-fold after 8 h of IL-1b and IL-1b+IFN-c exposure ( Figure 1C; densitometry in Figure S1A). These observations were confirmed in FACS-purified primary rat b-cells, in which IL-1b+IFN-c induced a 3-and 10-fold up-regulation of C/EBPd expression after 12 and 24 h respectively ( Figure 1D) while its expression was unaffected by TNF-a+IFN-c treatment ( Figure S1B). In human islet cells, a 24 h exposure to IL-1b+IFN-c up-regulated both C/ EBPd mRNA and protein expressions by 2-3 fold ( Figure 1E-G). We also confirmed the role of the transcription factors NF-kB and STAT1 in cytokine-induced C/EBPd up-regulation in b-cells [15,16]. Thus, NF-kB blockade using a super-repressor IkBa inhibited IL-1b+IFN-c-induced C/EBPd transcription until 24 h ( Figure S1C), while siRNA-mediated STAT1 knockdown affected mostly the early (12 h) up-regulation of C/EBPd by cytokines ( Figure S1D).

siRNA-mediated C/EBPd silencing exacerbates cytokineinduced apoptosis in INS-1E cells, primary rat b-cells, and dispersed human islets
We next used a siRNA-mediated knockdown approach to evaluate a putative role for C/EBPd in IL-1b+IFN-c-induced bcell apoptosis. INS-1E cells were transfected with an irrelevant control siRNA (siCtrl) or with two siRNAs targeting C/EBPd (siC/EBPd #1 and #2). Cells were subsequently left untreated, or treated for 2 to 24 h with IL-1b+IFN-c. Cytokines induced C/ EBPd protein up-regulation, reaching a peak at 8 h and slowly decreasing until 24 h (Figure 2A and 2B). The two siRNAs targeting C/EBPd accurately silenced its expression in both untreated and IL-1b+IFN-c-treated cells (Figure 2A and 2B). As previously reported [8,16], treatment with IL-1b+IFN-c induced apoptosis in siCtrl-transfected INS-1E cells after 24 h and in primary b-cells after 48 h ( Figure 2C & 2D), while inducing necrosis in few cells under all experimental conditions tested (,2% necrotic cells -data not shown). C/EBPd silencing with the two siRNAs exacerbated INS-1E cell and primary b-cell apoptosis after exposure to IL-1b+IFN-c ( Figure 2C & 2D). This increased apoptosis was confirmed in INS-1E cells by a second method that detects cytoplasmic fragmented DNA ( Figure 2E). In line with these observations, quantification of the remaining living cells by neutral red confirmed that C/EBPd deficiency hampered the ability of the cells to survive the IL-1b+IFN-c treatment ( Figure 2F). On the other hand, C/EBPd mRNA silencing did not induce apoptosis in untreated INS-1E cells or after treatments with IL-1b or IFN-c alone ( Figure S2A), and it neither exacerbated apoptosis ( Figure S2B) nor decreased the survival ( Figure S2C) of INS-1E cells exposed to the combination of TNF-a+IFN-c. The enhanced apoptosis in IL-1b+IFN-ctreated C/EBPd-silenced b-cells could not be explained by a higher production of nitric oxide (NO), since nitrite was similarly produced by siCtrl-and siC/EBPd-transfected cells after IL-1b+IFN-c exposure in primary rat b-cells ( Figure 2G) and INS-1E cells ( Figure S2D). Furthermore, the production of intracellular reactive oxygen-or nitrogen species (ROS or RNS) was also similar in cytokine-treated siCtrl-and siC/EBPd-transfected INS-1E cells ( Figure S2E). These data suggest that the increased cytokine-induced apoptosis in C/EBPd-silenced cells is probably not due to an amplified oxidative/nitrosative stress. Viability assays in dispersed human islet cells also demonstrated that efficient silencing of C/EBPd with three different siRNAs ( Figure 2H & Figure S2F) resulted in increased apoptotic cell death after 48 h of IL-1b+IFN-c exposure ( Figure 2I). Cytokinetriggered b-cell apoptosis occurs through the mitochondrialdependent intrinsic pathway of cell death, involving the translocation of the pro-apoptotic protein Bax to the mitochondria, cytochrome c release from the mitochondria to the cytoplasm, and subsequent cleavage and activation of caspases 9 and 3 [17,23,24]. As shown in Figure S2G, IL-1b+IFN-cinduced apoptosis in C/EBPd-silenced cells followed the same pathway, with Bax staining showing strong co-localization with the mitochondrial marker ATP synthase in apoptotic cells (arrow), while it was weak and diffuse in living cells ( Figure  S2G). Moreover, cytochrome c staining co-localized with the mitochondrial marker AIF in siC/EBPd-transfected living cells, while apoptotic cells had diffuse cytoplasmic cytochrome c staining (arrow), suggesting cytochrome c release from the mitochondria ( Figure S2H). Western blot analysis demonstrated that cytokine-induced expression of cleaved caspase 9 and caspase 3 were enhanced in siC/EBPd-transfected cells as compared to siCtrl-transfected counterparts ( Figure 2J -densitometries are shown in Figures S2I & S2J), supporting the observations from the viability assays ( Figure 2C & 2D) and confirming the involvement of the mitochondrial pathway. We next performed glucose oxidation tests to evaluate the putative role of C/EBPd on the functional inhibition of b-cells induced by cytokines [8,13,16]. Cytokine treatment greatly reduced the ability of b-cells to oxidize glucose ( Figure S2K). The inhibition of glucose oxidation was, however, similar in cytokine-treated siCtrl-and siC/EBPd-transfected cells ( Figure S2K), suggesting that C/ EBPd is not involved in inhibition of b-cell function.
We also evaluated a putative role of C/EBPb in cytokineinduced b-cell death. This other member of the C/EBP family was previously shown to be up-regulated by cytokines in array analysis of cytokine-treated b-cells [13,16] and to modulate mouse b-cell susceptibility to ER stress [41]. The expression of C/EBPb was up-regulated in INS-1E and primary b-cells after 12-24 h of IL-1b+IFN-c exposure ( Figure S3A) [13,16]. A 36-60% silencing of C/EBPb using two different siRNAs ( Figure S3B) did not affect cytokine-induced apoptosis in INS-1E cells ( Figure S3C) while it slightly decreased NO production ( Figure S3D).

C/EBPd knockdown enhances cytokine-induced CHOP expression and transcriptional activity in b-cells
To clarify the molecular pathways underlying the exacerbation of cytokine-induced apoptosis in C/EBPd-silenced cells, we first evaluated the expression of several genes induced during ER stress, a cellular response associated with IL-1b+IFN-c-induced apoptosis in b-cells [18,42]. XBP-1s mRNA expression was similarly up-regulated by cytokine treatment in siCtrl-and siC/ EBPd-transfected INS-1E cells and C/EBPd silencing did not induce nuclear translocation of XBP-1s in cytokine-treated INS-1E cells (data not shown). Cytokine treatment equally decreased Bip mRNA expression at 24 h in the three transfected conditions. On the other hand, C/EBPd deficiency enhanced IL-1b+IFN-cinduced CHOP mRNA up-regulation after 8-and 16 h in INS-1E  Figure 3C). Since CHOP is a member of the C/EBP family (C/EBPf) [28] and may hetero-dimerize with C/EBPd [43], we next evaluated a putative role of CHOP in the exacerbation of cytokine-induced apoptosis in C/EBPd-silenced cells. To this end, we interfered in parallel with C/EBPd and CHOP expressions, in a double knockdown approach, and performed viability assays. As described above, C/EBPd deficiency exacerbated IL-1b+IFN-cinduced apoptosis in INS-1E cells, while CHOP silencing had no effect on cell apoptosis after 24 h of cytokine treatment ( Figure 3F). Interestingly, the concomitant knockdown of CHOP abrogated the exacerbating effect of C/EBPd silencing on apoptosis ( Figure 3F). These results were confirmed by Western blot for cleaved caspase 3: the simultaneous inhibition of C/EBPd and CHOP reversed the increased expression of cleaved caspase 3 observed after C/EBPd silencing ( Figure 3G). These results identify CHOP as a contributory factor in the exacerbation of cytokine-induced apoptosis in C/EBPd-silenced b-cells. Since CHOP may also regulate the mitochondrial Unfolded Protein Response (UPR mt ) [44], we evaluated whether C/EBPd silencing exacerbates the UPR mt . This was done by evaluating the expression of LONP1, ClpP and HSP60, well-known markers of the UPR mt [44]. There were not, however, noticeable differences in the regulation of these three UPR mt markers in cytokine-treated C-EBPd-deficient INS-1E cells as compared to their control counterparts ( Figure S4C), making it unlikely that the UPR mt play a major role in the pro-apoptotic effect of C/EBPd-silencing.
The expression of the pro-apoptotic protein BIM is upregulated in C/EBPd-silenced cells   Figures S5D et S5E). As a whole, the data described above suggest that BIM plays a key role in the sensitization of C/EBPd-silenced cells to the pro-apoptotic effects of cytokines. To test this hypothesis, we performed viability assays after parallel knockdown of C/EBPd and BIM. BIM inhibition partially prevented IL-1b+IFN-c-induced apoptosis in INS-1E cells ( Figure 4D), while the double inhibition of C/EBPd and BIM abrogated the potentiating effects of C/EBPd knockdown on cytokine-induced apoptosis ( Figure 4D), suggesting that increased BIM activity contributes for the exacerbation of apoptosis in cytokine-treated C/EBPd silenced cells.
C/EBPd inhibition increases cytokine-induced b-cell chemokine production through defective control of STAT1 transcriptional activity b-cells produce many chemokines from the CXC and CC families in response to IL-1b+IFN-c exposure, hence contributing to attract and activate immune cells during insulitis [1]. Taking this into account, we evaluated whether C/EBPd also modulates the inflammatory response in IL-1b+IFN-c-exposed cells. C/ EBPd silencing exacerbated the expression of CXCL1, 9, 10 and CCL20 mRNAs after 16 h of cytokine exposure ( Figure 5A-E). These data were at least in part confirmed at the protein level, since C/EBPd-silenced cells secreted higher amounts of the chemokines CXCL1 and CXCL9 as compared to control cells after 16-and 24 h of cytokine treatment ( Figure 5F & 5G). Since we previously identified that the STAT1/IRF-1 signalling pathway exerts a key role in cytokine-induced chemokine production in b-cells [16], we performed time course analysis of the expression of these transcription factors in Ctrl-and C/EBPddeficient INS-1E cells after cytokine exposure. Interestingly, C/ EBPd silencing hampered cytokine-induced IRF-1 expression ( Figure 5H & 5J). As previously shown [16], this impaired IRF-1 expression was accompanied by exacerbated STAT1 phosphorylation after 8 h of treatment with IL-1b+IFN-c, but also by increased expression of total STAT1 after 8, 16 and 24 h of cytokine treatment ( Figure 5H, 5K & 5L). This prolonged STAT1 activation in C/EBPd-silenced cells enhanced STAT1 transcriptional activity, as shown by IL-1b+IFN-c-induced augmented  activation of a STAT1 reporter in C/EBPd-silenced cells as compared to controls ( Figure 5I). This effect is specific, since C/ EBPd knockdown had an opposite effect on an NF-kB reporter, inducing a 20-50% decrease ( Figure S6). We next evaluated the expression of SOCS-1, a negative regulator of STAT1 activity that is modulated by IRF-1 activation in b-cells [16]. As shown in Figure 5M, cytokine-induced SOCS-1 up-regulation was also impaired in C/EBPd-silenced cells as compared to Ctrl, reflecting the impaired activation of IRF-1 in C/EBPd-deficient cells. All together, these results demonstrate that C/EBPd exacerbates cytokine-induced chemokine production by interfering with IRF-1 up-regulation, hence disturbing the negative regulatory feedback loop by which IRF-1 modulates STAT1 activation through the induction of the inhibitory protein SOCS-1 [16].
The transcription factors CHOP and STAT1 contribute to the enhancement of BIM expression in C/EBPd-silenced cells We described above that the activation of the transcription factors CHOP and STAT1 are increased or prolonged in C/ EBPd-silenced cells ( Figure 3D-E and Figure 5H-L). Since both CHOP and STAT1 have been reported to mediate BIM transcriptional regulation in thymocytes and b-cells respectively [26,45], we next evaluated a putative role of these transcription factors in the exacerbation of BIM expression in C/EBPd-silenced b-cells. To this end, INS-1E cells were transfected with previously validated siRNAs targeting C/EBPd (Figure 2A-B), CHOP ( Figure 3G) and STAT1 [16,46] alone or in combination, and the expression of BIM mRNA evaluated. As shown in Fig. 6A-C, the individual siRNAs inhibited their target gene without affecting the expression of the other non-targeted genes evaluated. As described in Figures 3 and 4, C/EBPd silencing exacerbated CHOP and BIM up-regulation as compared to controls after 16 h of cytokine treatment ( Figure 6A & 6D), while the knockdown of CHOP alone had no effect on cytokine-induced BIM expression ( Figure 6D). Interestingly, the double knockdown of C/EBPd and CHOP reversed to a large extent the exacerbation of BIM expression induced by C/EBPd silencing alone ( Figure 6D). STAT1 silencing, alone or in combination with C/EBPd inhibition, potently inhibited IL-1b+IFN-c-induced BIM upregulation ( Figure 6D). The double knockdown of CHOP and STAT1 did not further inhibit cytokine-induced BIM expression as compared to STAT1 silencing alone ( Figure 6D). We next studied the regulation of the rat BIM promoter using a luciferase reporter construct containing the complete sequence of the this promoter (22454/+2658) [47]. To this end, INS-1E cells were successively transfected with specific siRNAs as indicated and then with the rat BIM luciferase reporter. C/EBPd silencing augmented the up-regulation of the BIM promoter induced by IL-1b+IFN-c treatment as compared to siCtrl-transfected cells, while the concomitant knockdown of C/EBPd and CHOP reversed the exacerbating effect of C/EBPd silencing on the induction of the BIM promoter ( Figure 6E). The siSTAT1, alone or associated with C/EBPd or CHOP inhibition also prevented cytokine-induced upregulation of the BIM promoter ( Figure 6E). Viability assays corroborated these observations, with CHOP silencing inhibiting the exacerbating pro-apoptotic effects of C/EBPd inhibition in cytokine-treated INS-1E cells, while STAT1 silencing both protected the cells against cytokine-induced apoptosis and counteracted the pro-apoptotic effects of C/EBPd silencing ( Figure 6F). Theses results identify STAT1 as a key mediator of IL-1b+IFN-c-induced BIM up-regulation in b-cells, and suggest that the transcription factor CHOP mediates the ''exacerbating effect'' of C/EBPd silencing on cytokine-induced BIM upregulation.

C/EBPd overexpression decreases BIM expression and partially protects INS-1E cells against cytokine-induced apoptosis
We next tested whether C/EBPd overexpression affects BIM expression in INS-1E cells. Interestingly, a strong overexpression of C/EBPd ( Figure 7A) reduced both basal and cytokine-induced BIM expression (Figure 7A & 7B) and also decreased IL-1b+IFNc-induced up-regulation of BIM and CHOP mRNA expressions ( Figure 7C & 7D). The co-tranfection of the C/EBPd overexpression vector with the rat BIM promoter also reduced cytokineinduced up-regulation of this promoter as compared to untranfected-and control-transfected counterparts ( Figure 7E). Finally, viability experiments demonstrated that C/EBPd overexpression induced a moderate protection against IL-1b+IFN-c-induced apoptosis ( Figure 7F), and increased the survival of INS-1E cells as compared to control-transfected cells ( Figure 7G). Taken together, these experiments support the hypothesis that C/EBPd inhibits BIM expression and thus partially counteracts cytokineinduced apoptosis.

Discussion
A better understanding of the signalling pathways involved in cytokine-induced b-cell apoptosis during insulitis may help to define potential therapeutic targets to interfere with T1D development [9]. We presently demonstrate, using silencing and overexpression approaches, that the transcription factor C/EBPd is up-regulated by IL-1b+IFN-c in rat b-cells and human islets and exerts regulatory functions in these cells, inhibiting pro-apoptotic and pro-inflammatory signals (Figure 8). NF-kB plays an important role in cytokine-induced C/EBPd expression, as IL-1b (a known NF-kB inducer) up-regulates C/EBPd expression in bcells, while NF-kB blockade inhibits cytokine-induced C/EBPd mRNA up-regulation (present data; [15]). These results are in line with previous observations describing NF-kB recruitment at the C/EBPd promoter after 1 h of LPS exposure in macrophages [48]. The transcription factor STAT1 is also involved in cytokineinduced C/EBPd expression, since IFN-c enhances IL-1binduced-C/EBPd up-regulation and interfering with STAT1 activity hampers cytokine-induced C/EBPd expression at early time points ( Figure S1D; [16]). It is surprising, however, that the combination of TNF-a+IFN-c does not up-regulate C/EBPd expression in b-cells, and likewise, that C/EBPd silencing or overexpression does not affect TNF-a+IFN-c-induced cells death in b-cells ( Figure S2B-C and data not shown). Indeed, TNF-a is also an inducer of NF-kB activation in these cells, albeit to a lesser extent than IL-1b [49]. C/EBPd was shown to auto-regulate its expression by binding to sites located downstream of the C/EBPd gene [43,50]. Unlike IL-1b, TNF-a alone does not up-regulate C/ EBPd expression ( Figure 1B-C). It is thus conceivable that C/ EBPd transcription in b-cells is initiated by NF-kB and other transcription factors induced by IL-1b (but not by TNF-a) and that this initial synthesis of C/EBPd is required to allow further C/ EBPd-mediated auto-transcription. Interestingly, IL-1b+IFN-cinduced C/EBPd mRNA up-regulation was sustained until 24 h of cytokine treatment while the expression of C/EBPd protein returned to basal after 16 h (Figs. 1 & 2). This may be explained by the progressive increase of cytokine-induced ER-stress, previously suggested to inhibit mRNA translation of many proteins in b-cells [42].
Inhibiting C/EBPd activity in b-cells exacerbates IL-1b+IFN-cinduced apoptosis by increasing the expression of the proapoptotic factors CHOP and BIM, while overexpressing C/EBPd down-regulates CHOP and BIM expression and partially protects b-cells against the deleterious effects of cytokines (Figure 3, 4 & 7). This increased cytokine-induced apoptosis in C/EBPd-silenced cells is independent of increased oxidative or nitrosative stress ( Figure 2G & S2D-E), and our preliminary experiments did not indicate a role of modified autophagy in this process (data not shown). Autophagy has recently been suggested to be involved in b-cell death in both type 1 & type 2 diabetes [51,52]. This negative regulatory role of C/EBPd on cytokine-induced apoptosis in bcells appears to be critical and non-redundant, since it is conserved between rodents and humans (Figures 1 and 2). This adds a new dimension to C/EBPd functions, as previous reports described C/ EBPd mostly as a cell growth regulator and tumor suppressor. Indeed, C/EBPd expression is down-regulated in acute myeloid leukemia [53] and in cervical, breast and liver cancers [54,55], while C/EBPd knockout mice display mammary gland ductal hyperplasia [37]. These anti-proliferative activities in mammary gland epithelial cells are linked to C/EBPd-induced Cdc27 expression, resulting in proteasome-mediated destruction of the cell cycle promoting factor Cyclin D1 [38,56]. Putative tumorsuppressing activities of C/EBPd may, however, depend on the hormonal background, since C/EBPd is highly expressed in androgen-dependent but not androgen-independent prostate carcinomas [57]. These data underline the context dependency of the function of regulatory transcription factors in different cell types or hormonal backgrounds, and indicate two novel functions for C/EBPd in primary non-dividing cells namely as an important anti-apoptotic and anti-inflammatory regulator. Our preliminary data on C/EBPb silencing suggest that C/EBPb does not play a key role in IL-1b+IFN-c-induced b-cell death. This differs from the previously reported effects of C/EBPb in ER-stress-induced bcell and tumoral cell death [41,58], palmitate-induced b-cell apoptosis [59], and NO-dependent hepatocyte apoptosis [60]. On the other hand, C/EBPb knockdown slightly decreases NO production in INS-1E cells (present data), which is in accordance with the reported roles of C/EBPa and C/EBPb for iNOS induction in hepatocytes [60]. We cannot exclude that the moderate (60%) C/EBPb silencing achieved in our experiments was not sufficient to disclose the putative effect of the transcription factor in cytokine-induced apoptosis. Additional experiments are required to clarify this issue.
Our data suggest that increased CHOP and BIM expression in C/EBPd-silenced cells contributes for the exacerbation of apoptosis upon exposure to IL-1b+IFN-c. Indeed, concomitant silencing of CHOP and C/EBPd or BIM and C/EBPd abrogates the exacerbating effects of C/EBPd silencing on cytokine-induced up-regulation of the rat BIM promoter and on cytokine-induced apoptosis (present data). CHOP is a well known ER-stress-induced transcription factor that may trigger apoptosis through perturbation of intracellular pH and cellular cytoskeleton, down-regulation of the anti-apoptotic factor Bcl-2 and increase of reactive oxygen species production [61,62]. CHOP expression is regulated by other C/EBP family members, including C/EBPa and C/EBPb [63]. CHOP has been shown to hetero-dimerize with C/EBPa, b and d [28,43,64], hence targeting these transcription factors to specific DNA binding sequences [65]. The nature of the interaction between C/EBPd and CHOP in cytokine-treated bcells and how C/EBPd deficiency enhances CHOP expression, remain to be determined. Increased CHOP expression in C/ EBPd-silenced cells is associated with increased CHOP-regulated transcription, since the expression of the CHOP-regulated proapoptotic gene GADD34 [66] is also exacerbated in C/EBPddeficient cells (present data). Moreover, increased CHOP expression enhances the expression of the pro-apoptotic Bcl-2 family member BIM in cytokine-treated C/EBPd-silenced cells (see below). Previous reports indicate that BIM has an important role in high glucose- [27], virus- [67] and cytokine-induced b-cell death [26,68], which occurs through the mitochondrial apoptotic pathway [23]. We presently show that the exacerbation of cytokine-induced up-regulation of BIM mRNA and BIM promoter activity in C/EBPd-silenced cells is at least in part mediated by the transcription factors STAT1 and CHOP. BIM up-regulation was already shown to be mediated by STAT1 in IL-21-treated lymphocytic leukaemia cells [69] and TNF-a+IFN-cexposed b-cells [26], while CHOP activity is required for BIM upregulation and apoptosis induction in thapsigargin-treated thymocytes [45] and growth factor-deprived lymphoid precursors [70]. Although BIM mRNA was up-regulated by cytokines at all time points studied in siCtrl-tranfected cells, we observed a transient decrease in BIM protein expression after 8 h of IL-1b+IFN-c exposure ( Figure 4C). This is not due to an off-target effect of the siCtrl, as BIM protein also decreased after 8 h of cytokine exposure in untransfected cells ( Figure S5D-E). This transient decrease in BIM protein expression was not observed in siC/ EBPd-silenced cells, in which exacerbated up-regulation of BIM mRNA expression seems to allow constant BIM protein expression throughout cytokine exposure. BIM has been reported to be targeted to proteasome-mediated degradation during rapid ischemic tolerance in neurons [71,72] and subjected to caspase 3-mediated degradation as a feedback mechanism in MEFs and osteoclasts [73], but the putative pathways leading to BIM degradation and re-synthesis in IL-1b+IFN-c-exposed b-cells remain to be clarified. As a whole, our data suggest that IL-1b+IFN-c-induced BIM up-regulation in b-cells is mediated mostly via STAT1 activation, while CHOP seems to contribute for the exacerbating effects of C/EBPd silencing on BIM expression. Nonetheless, while C/EBPd overexpression repressed to a large extent BIM expression in INS-1E cells, it only moderately protected these cells against IL-1b+IFN-c-induced apoptosis (Figure 7). These results emphasize the complexity of the pro-apoptotic pathways induced by cytokines in b-cells and the relative complementarity/redundancy of function of the proapoptotic Bcl-2 family members in cytokine-exposed b-cells [23].
Cytokine-induced up-regulations of the chemokines CXCL1, 9, 10 and CCL20 mRNAs are all exacerbated in C/EBPd-silenced cells, as is the release of the tested chemokines CXCL1 and CXCL9, suggesting that C/EBPd elicit anti-inflammatory activities in b-cells. These observations contrast with previous reports, describing C/EBPd as a pro-inflammatory transcription factor that promotes LPS-induced activation of microglial cells and astrocytes [74], participates in TLR-induced pro-inflammatory cytokines production [39] and discriminates between transient and persistent LPS-mediated signalling in macrophages [48]. On the other hand, the anti-inflammatory agent dexamethasone was shown to induce C/EBPd expression in skeletal muscle [57], while amyloid-b fibrils down-regulate C/EBPd in astro-microglial cells during Alzheimer's disease [75]. We have previously shown that cytokine-induced chemokine production is increased by the prolongation of STAT1 activation in b-cells; IRF-1 controls a negative regulatory feedback loop that limits STAT1 activation through induction of the STAT1 regulator SOCS-1 [16]. In the present experiments, prolonged IL-1b+IFN-c-induced STAT1 activation in C/EBPd-silenced cells was also associated with defective induction of IRF-1 and its downstream gene SOCS-1, and augmented chemokine production. Of note, C/EBPd deficiency decreases NF-kB-dependent transcription in cytokinetreated INS-1E cells. Since NF-kB is involved in cytokine-induced IRF-1 expression [16,76] it is likely that C/EBPd activation promotes IRF-1 expression by either participating directly to IRF-1 transcription by dimerizing with NF-kB1 p50 or RelA, or by inhibiting a negative regulator of NF-kB activation. The relation between C/EBPd and IRF-1 expression remains to be clarified, and the co-immunoprecipitation experiments that we performed did not indicate a direct interaction between IRF-1 and C/EBPd (data not shown), as it was previously reported for C/EBPb in IFN-c-treated HepG2 cells [77].
We and others observed that b-cells have in place regulatory mechanisms devoted to control excessive activation of deleterious factors induced by cytokines. These include PTPN2-and IRF-1/ SOCS-1-mediated attenuation of STAT1 activation [16,78], JunB-mediated attenuation of NO production and ER stress [79], A20-and SOCS-3-mediated NF-kB inactivation [80,81] and HSP70 synthesis upon cytokine-treatment [82]. It is likely that C/ EBPd contributes to these signalling pathways that provide ''local'' regulation to inflammatory processes in order to protect poorly proliferating and long-lived cells, such as b-cells, against excessive damage. It remains to be clarified why these regulatory mechanisms are overruled in T1D, but the identification of novel agents that selectively promote protective pathways in b-cells may represent an alternative and ''local'' approach to protect b-cells against the relentless destructive assault initiated by autoimmune cells.

Culture of primary FACS-sorted rat b-cells, human islets and INS-1E cells
Male Wistar rats (Charles River Laboratories, Brussels, Belgium) were housed and used according to the guidelines of the Belgian Regulations for Animal Care; all experiments performed have been approved by the CEBEA Ethical Committee (Commission d'Ethique du Bien-Etre Animal, Universite Libre de Bruxelles, Permit Number LA 1230351, valid until 2014). For islet isolation, pancreases were digested by collagenase and islets were hand picked under a stereomicroscope. B-cells were purified by autofluorescence-activated cell sorting (FACSAria, BD Bioscience, San Jose, CA, USA) [83]. The preparations contained 9562% bcells (n = 9). b-cells were cultured for 2 days in Ham's F-10 medium containing 10 mM glucose, 2 mM glutaMAX, 50 mM 3isobutyl-1-methylxanthine, 5% FBS, 0.5% charcoal-absorbed BSA (Boehringer, Indianapolis, IN, USA), 50 U/ml penicillin and 50 mg/ml streptomycin [83]. During cytokine exposure, cells were cultured in the same medium but without serum.
Human islets were isolated from 10 non-diabetic organ donors (age 6564 years; body mass index 26.161 kg/m 2 ) in Pisa, Italy, with the approval of the Ethics Committee of the University of Pisa. A written consent was obtained from each organ donor for the use of the pancreas for research purposes. Islets were isolated by enzymatic digestion, density-gradient purification [84], and cultured in M199 medium containing 5.5 mM glucose. The human islets were shipped to Brussels within 1-5 days of isolation. After overnight recovery in Ham's F-10 containing 6.1 mM glucose, 10% FBS, 2 mM GlutaMAX, 50 mM 3-isobutyl-1methylxanthine, 1% BSA, 50 U/ml penicillin and 50 mg/ml streptomycin, islets were dispersed for viability assays or exposed to cytokines in the same medium without FBS for 24 h. The percentage of b-cells, examined in the 10 dispersed islet preparations by staining with anti-insulin antibody (1:1000, Sigma, Bornem, Belgium) and donkey anti-mouse IgG rhodamine (1:200, Lucron Bioproducts, De Pinte, Belgium), was 6365%; only preparations containing .40% b-cells were used in the experiments.
The rat insulin-producing INS-1E cell line (a kind gift from Dr. C. Wollheim, Centre Medical Universitaire, Geneva, Switzerland) was cultured as previously described [85] and used between passages 52 and 72.

Cell treatment and NO measurement
The following cytokine concentrations were used, based on previous dose-response experiments [8,85,87]: recombinant human IL-1b (specific activity 1.8610 7 units/mg; a gift from C.W. Reinolds, National Cancer Institute, Bethesda, MD) at 10 units/ ml (for rat cells) or 50 units/ml (for human cells); recombinant murine TNF-a (specific activity: 2610 8 units/mg; Innogenetics, Gent, Belgium) at 1,000 units/ml; and recombinant rat or human IFN-c (specific activity: 2610 7 units/mg; R&D Systems, Abingdon, U.K.) at 100 and 1,000 units/ml for rat cells and human islets respectively. Culture supernatants were collected for nitrite determination (nitrite is a stable product of NO oxidation) at OD 540 nm using the Griess method.

Evaluation of intracellular ROS/RNS content
Cells were transfected with siCtrl, siC/EBPd #1 or siC/EBPd #2 and left untreated or treated as described above for 2-24 h with IL-1b+IFN-c. Cells were then lysed in PBS by freeze/thaw cycles followed by sonication. Cell lysates were centrifuged to discard cellular debris and lysates were quantified using the Bradfod method. Intracellular ROS/RNS content was then evaluated in an equivalent amount of cell extract for each condition using the OxiSelect in vitro ROS/RNS assay kit (Green Fluorescence -Cell Biolabs, San Diego, CA, USA).

mRNA extraction and real time PCR
Poly(A) + mRNA was isolated from INS-1E cells or rat primary b-cells using the Dynabeads mRNA DIRECT TM kit (Invitrogen), and reverse transcribed as previously described [83]. The real time PCR amplification reaction was done as described [83], using SYBR Green and compared to a standard curve. Expression values were corrected for the housekeeping genes glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and b-actin for rat and human assays respectively; we have previously shown that cytokines do not modify GAPDH and b-actin expressions in these species [78]. The primers used in this study are listed in Table S1.

Assessment of apoptosis and viability
The percentage of viable, apoptotic and necrotic cells was determined after staining with the DNA-binding dyes Propidium Iodide (PI, 5 mg/ml, Sigma) and Hoechst 33342 (HO, 5 mg/ml, Sigma) [83]. This method is quantitative, and has been validated by systematic comparison against electron microscopy [88] and several other well-characterized methods, including fluorometric caspase 3 & 7 assays and determination of histone-complexed DNA fragments by ELISA [24,25,[89][90][91]. A minimum of 500 cells was counted in each experimental condition. Viability was evaluated by two independent observers, one of them being unaware of sample identity. The agreement between findings obtained by the two observers was .90%. Results are expressed as percent apoptosis, calculated as (number of apoptotic cells/total number of cells)6100. Apoptosis was confirmed in some experiments by the Cell Death Detection ELISAplus kit (Roche Diagnostics, Vilvoorde, Belgium), which detects cytoplasmic fragmented DNA. For some key experiments, cellular viability after cytokine treatment was assessed using the Neutral Red-based in vitro toxicology assay kit (Sigma) as follows: neutral red (final concentration 33%) was added to the culture medium for 2 h before the end of cytokine treatment. The culture medium was then removed and cells were washed once with the fixative solution to discard detached cells and remove unincorporated neutral red. The neutral red incorporated into living cells was then eluted using the solubilisation solution. ODs were read at 540 nm; the magnitude of the OD 540 nm is directly proportional to the number of remaining living cells for each condition.

Glucose oxidation assay
D-[U-14 C] glucose (specific activity: 300 mCi/mmol, concentration: 1 mCi/ml, Perkin Elmer, Waltham, MA) was used to evaluate glucose oxidation in untreated and IL-1b+IFN-c-treated siCtrl-or siC/EBPd-transfected cells. Cells were trypsinized and resuspended in KRBH solution without glucose at 5610 6 living cells/ml. 20 ml of the cell suspension (1.10 5 cells) was transferred to glass vials containing 20 ml KRBH buffer supplemented with 0.19 mCi of D-[U-14 C] glucose and non-radioactive glucose to a final concentration of 1.67, 10 or 16.7 mM of glucose. The vials were inserted into 20 ml glass scintillation flasks, gassed with 95% CO 2 and 5% O 2 , airtight sealed with rubber membranes and shaken continuously for 2 h at 37uC. After incubation, 0.02 ml of metabolic poison (400 mM citrate buffer -pH 4.9, 3 mM KCN, 10 mM Antimycin A, 20 mM Rotenone; pH 7.4) was injected through the rubber cap into the vials containing the cells and 0.2 ml hyamine hydroxide was injected outside of the vials to absorb the released CO 2 . Following 1 h incubation at 37uC, the vials were removed and scintillation fluid (6 ml) was added to the hyamine. After 14 h at 4uC, the radioactivity was measured in a TriCarb 2100TR Liquid scintillation analyzer (Perkin Elmer, Waltham, MA). The rate of glucose oxidation was expressed as mmol/120 min.1000 cells.

Infection with recombinant adenoviruses
Cells were infected either with Ad-Luc (luciferase-expressing adenovirus) or Ad-srIkB (a virus expressing an NF-kB superrepressor IkBa protein) [40]. Cells were infected for 2 h at 37uC with a multiplicity of infection (MOI) of 10. The MOI was selected based on lowest toxicity by viral infection combined with highest blockade of NF-kB activation. After infection (24 h), cells were treated with cytokines. We have previously shown that infection of b-cells with Ad-srIkB at the MOI used in the present study does not change its function or viability [40].

Overexpression of rat C/EBPd
The expression vectors pCMV-Ctrl [16] and pCMV-C/EBPd (TrueORF cDNA Clones -OriGene, Rockville, MD, USA) were transfected in INS-1E cells using Lipofectamine 2000 (Invitrogen) as previously described [49]. After overnight incubation, the medium was changed and cells were exposed to cytokines as indicated.

Evaluation of chemokine accumulation in the medium by ELISAs
Cells were transfected with siCtrl, siC/EBPd #1 or siC/EBPd #2 and subsequently left untreated or treated with IL-1b+IFN-c as described above. Supernatants were collected after 2, 4, 8, 16 & 24 h of treatment for determination of CXCL1 and CXCL9 chemokines secretion using the commercially available ELISA kits for rat CXCL1 (R&D Systems) and rat CXCL9 (Uscn Life Science, Wuhan, China).

Statistical analysis
Data are presented as mean 6 SEM. Comparisons were performed by two-tailed paired Student's t-test or by ANOVA followed by Student's t test with Bonferroni correction as indicated. A p value,0.05 was considered statistically significant.