Cytokine Induction of Fas Gene Expression in Insulin-Producing Cells Requires the Transcription Factors NF-κB and C/EBP

The Fas promoter region was isolated by polymerase chain reaction (PCR) using the Expand High Fidelity PCR system (Roche Molecular Biochemicals, Mannheim, Germany) on 0.5 μg of rat genomic DNA prepared from liver as described (29). The primers used were based on the published rat Fas sequence (28). The forward primer (position −798 to −781) was 5′-CGTCTAGACAGCTCTTCTGGTCAACA with an added XbaI site, and the reverse primer (position +108 to +126) was 5′-CGGGATCCTAGACTGCCTCTGGGTGT with an added BamHI site. The amplified product was cloned into pBluescript KS (Stratagene, La Jolla, CA). Two clones were sequenced and shown to contain small and identical differences with the published sequence (28) so that nucleotide −798 was renumbered −811. The presently cloned Fas sequence was deposited in the GenBank database under accession number AF308477. The promoter region (−811 to +126) and its 5′ deleted fragments were isolated from pBluescript by digestion with BamHI (3′ end) and with the following restriction enzymes for the 5′ end: XbaI (−811), EcoRI (−223), or Nci I (−54), filled in with the Klenow fragment of DNA polymerase I. The fragments were cloned into the SmaI-BglII sites of pGL3-Basic (Promega, Madison, WI) to obtain the pFas luc constructs. Mutations for the NF-κB and C/EBP sites at positions −142 and −130, respectively, were generated on pFas-811 luc by PCR (30) using the Expand High Fidelity PCR system. The pairs of primers were as follows: 5′-ATCAAGCCCTGCTTGGGGCCT and 5′-ATCCTGGGCGCTTCCACAGTC for deletion of nucleotides −143 to −124, 5′-ATCAAGCCCTGCTTGGGGCCT and 5′-TGCATAAATGGGCATTATGCATCCTGGGCGC for mutation in the NF-κB site, and 5′-GGGGAATGCCCATTCTCTAGATCAAGCCCTGCTTG and 5′-ATCCTGGGCGCTTCCACAGTC for mutation in the C/EBP site (mutated bases are in italics). All mutations were confirmed by sequencing. Expression vectors for NF-κB p65 (CMV-p65), NF-κB p50 (CMV-p50), C/EBPβ (CMV-C/EBPβ), and control vector (CMV) were provided by H. Liu and R.M. Pope (31).

RINm5F insulinoma cells were cultured in RPMI-1640 medium with Glutamax-1 (Life Technologies, Paisley, Scotland) supplemented with 10% fetal calf serum. Rat β-cells were FACS purified from islets isolated from male Wistar rats, as previously described (32). The rats were housed according to the guidelines of the Belgian Regulation for Animal Care, and the experimental protocol was approved by the Ethics Committee for Animal Experiments of the Vrije Universiteit Brussel. These preparations contained >95% β-cells and were cultured in HAM’s F-10 medium supplemented with 10 mmol/l glucose, 2 mmol/l glutamine, 0.5% bovine serum albumin, and 50 μmol/l 3-isobutyl-1-methylxanthine (IBMX). The presence of IBMX at this concentration preserves β-cell survival in culture (33) with minimal effects on cAMP formation in the absence of adenylate cyclase activators (34). RINm5F and primary β-cells were cotransfected with the luciferase test plasmids and with pRL-CMV (Promega), as an internal control, by lipofection using lipofectAMINE (Life Technologies, Gaithersburg, MD) as previously described (35). The following modifications were introduced for β-cell transfections: 4 × 10⁴ cells were seeded in 94-well plates and incubated for 3 h with 0.05 μg test plasmid, 0.005 μg pRL-CMV, 0.5 μg PLUS Reagent, and 0.25 μl lipofectAMINE. The day after transfection, the cells were exposed for 16 h to 50 units/ml (for RINm5F cells) or 30 units/ml (for β-cells) of recombinant human IL-1β (provided by Dr C.W. Reynolds, National Cancer Institute, Bethesda, MD) and/or 1,000 units/ml recombinant mouse IFN-γ (Life Technologies). The concentrations of cytokines selected for the experiments were based on previous data from our group (35,36,37,38). In cotransfection experiments, RINm5F cells were transfected with 0.1 μg test plasmid, 0.005 μg pRL-CMV, and the indicated amounts of expression vectors. The total amount of DNA was kept at 0.2 μg by adding the control vector CMV. The cells were harvested 24 h after transfection. Luciferase activities were assayed with the Dual-Luciferase Reporter Assay System (Promega). Values obtained for the promoterless vector pGL3-Basic were 1 and 0.6 light units in RINm5F and β-cells, respectively. The transfection efficiency of primary β-cells was 10–20%. Test values were corrected for the luciferase activity value of the internal control pRL-CMV and expressed as relative luciferase activity.

Nuclear extracts were prepared from RINm5F cells as described (39). Nuclear proteins (4–5 μg) were preincubated with 1 μg poly(dIdC) in 20 μl containing 10 mmol/l HEPES (pH 7.9), 50 mmol/l KCl, 5 mmol/l MgCl₂, 0.05 mmol/l EDTA, 0.5 mmol/l dithiothreitol, and 10% glycerol for 10 min at 0°C before addition of 50 molar excess of competing oligonucleotide, as indicated, and radiolabeled probe (30,000 cpm). The incubation was continued for 20 min at 0°C for NF-κB electrophoretic mobility shift assays (EMSAs) or at room temperature for C/EBP EMSAs. Where indicated, 1 μl of antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) was added for 20 min at room temperature. The samples were electrophoresed on 5% polyacrylamide gels in 25 mmol/l Tris, 25 mmol/l boric acid, and 0.5 mmol/l EDTA for NF-κB EMSAs or in 1× Tris-glycine-EDTA buffer for C/EBP EMSAs. Oligonucleotides for EMSAs were as follows (upper strand shown): Fas NF-κB (NF)-C/EBP, 5′-agGATGGGGAATGCCCATTTATGCAATCAAG; Fas NF-κB (NF)-C/EBPmut, 5′-agGATGGGGAATGCCCATTCTCTAGATCAAG; Fas C/EBP, 5′-agCCATTTATGCAATCAAG; NF-κB consensus (cons), 5′-agCTTCAGAGGGGACTTTCCGAGA; C/EBP consensus (cons), 5′-agCAGATTGCGCAATCTGCA. The lowercase letters represent added nucleotides to allow end-labeling with the Klenow fragment of DNA polymerase I.

Reverse transcriptase (RT)-PCR was performed on poly(A)+ RNA as described (35). Primers for Fas were 5′-GAATGCAAGGGACTGATAGC (forward primer) and 5′-TGGTTCGTGTGCAAGGCTC (reverse primer). The PCR for Fas and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) primers (as in 35) mRNA detection was performed with 34 and 28 cycles, respectively. These numbers of cycles were selected to allow cDNA amplifications in the linear range.

Nuclear extracts were prepared from 10⁶ RINm5F cells transfected with the expression vectors for p65, p50, or C/EBPβ alone, or for p65 in combination with p50 or C/EBPβ at a one-to-two ratio, as indicated above. Proteins were fractionated by SDS-PAGE and transferred on nitrocellulose. The filters were incubated with anti-p65, anti-p50, or anti-C/EBPβ antibodies and then with a secondary antibody (peroxidase-labeled anti-IgG). Detection was performed with an enhanced chemiluminescence kit (Amersham Pharmacia Biotech, Little Chalfont, U.K.).

Results are given as means ± SE. Multiple comparisons were performed by analysis of variance (ANOVA), followed by group comparisons using the Student’s paired t test with correction of the P values for multiple comparisons by the Bonferroni method.

To determine which cytokine is the main inducer of Fas expression in rat β-cells, FACS-purified rat β-cells were exposed for 6 or 24 h to IL-1β, IFN-γ, or a combination of both cytokines, and Fas mRNA content was determined by RT-PCR. Unstimulated cells expressed low levels of Fas mRNA (Fig. 1). Fas mRNA expression was induced 15-fold in cells exposed to IL-1β for 6 h, and this increase was not potentiated by IFN-γ. Increased Fas mRNA expression was maintained over a period of 24 h (Fig. 1). To determine the role of NO in Fas expression in rat β-cells, FACS-purified rat β-cells were treated for 24 h with IL-1β in the presence or absence of N^G-methyl-l-arginine ([l-MA] 1 mmol/l), an inhibitor of iNOS activity. The optical density values for Fas mRNA content, corrected for GAPDH mRNA content, were comparable in the presence or absence of l-MA (control, 0.22 ± 0.04; l-MA, 0.05 ± 0.05; IL-1β, 0.79 ± 0.06; IL-1β + l-MA, 1.1 ± 0.17; n = 3), showing that the iNOS blocker did not prevent Fas mRNA induction. This result confirms previous observations in mouse (9,15) and human islets (9) that cytokine-induced Fas expression is independent of NO. On the basis of these results, the subsequent experiments on promoter regulation were performed in the presence of IL-1β alone.

To delineate the Fas promoter region responding to IL-1β, the promoter fragment from nucleotides −811 to +123 relative to the transcription start site and its 5′ deletants (Fig. 2A) were cloned into the luciferase reporter plasmid pGL3 and transfected into FACS-purified β-cells treated with IL-1β or left untreated. In untreated cells, all constructs have a low basal activity (Fig. 2B) consistent with the lack of TATA box or Sp1 binding sites in the rat Fas promoter. In IL-1β–treated cells, the luciferase activity of the constructs pFas-811 luc and pFas-223 luc was induced more than sevenfold, whereas pFas-54 luc was not responsive to IL-1β (Fig. 2B). These results suggest that the element or elements responding to IL-1β are located between nucleotides −223 and −54 and rule out a role for the potential upstream C/EBP and Oct binding sites (Fig. 2A). The same experiments were reproduced in an insulinoma cell line, RINm5F, with similar results as those observed in β-cells (Fig. 3A). Hence (and due to the large number of cells required), subsequent experiments to identify the regulatory elements in the Fas promoter were performed in RINm5F cells.

The region between nucleotides −223 and −54 contains three potential NF-κB binding sites, a C/EBP binding site, and a binding site for Ets factors (Fig. 2A). Contribution of the Ets binding site located at position −196 to the promoter activity was not significant because a construct containing the promoter region up to −153, from which the Ets site is absent, was still responsive to IL-1β (data not shown). NF-κB and C/EBP synergize for transcriptional activation of many proinflammatory genes through cooperative binding (40,41) or direct interaction (42,43). Thus, we chose to mutate the adjacent NF-κB and C/EBP sites (Fig. 2A). A pFas-811 luc construct with nucleotides −143 to −124 deleted, eliminating both binding sites (ΔNF-C/EBP), completely lost responsiveness to IL-1β (Fig. 3B). Similarly, pFas-811 luc, with either the NF-κB (NFmut) or the C/EBP site mutated (C/EBPmut), was unable to respond to IL-1β (Fig. 3B). These results indicate that both binding sites are required for IL-1β induction of the Fas promoter.

To identify the factors binding to the IL-1β–responsive element, we performed EMSAs using a probe spanning the NF-κB and C/EBP sites from nucleotides −148 to −116 (probe Fas NF-C/EBP), and nuclear extracts from RINm5F cells untreated or exposed to IL-1β for 30 min and 4 h. A major complex a and a minor complex b were induced in cells treated with IL-1β, with a higher binding activity at 30 min (Fig. 4A). Complex a was similarly observed when the C/EBP site was mutated in the probe (probe Fas NF-C/EBPmut, Fig. 4A). Complex b was more clearly detected when using the Fas NF-C/EBPmut probe. This finding may be due to the absence of a nonspecific complex, observed with the wild-type probe, which interferes with the visualization of complex b. These results are consistent with NF-κB binding. Indeed, we have previously shown that binding activity of NF-κB is maximal at 30 min in IL-1β–exposed RINm5F cells (38). The nature of complexes a and b was further characterized by competition and supershift experiments. The two complexes were specifically competed by an excess of unlabeled probe (Fig. 4B, lane 2) and by a consensus NF-κB oligonucleotide (lanes 4 and 13) but not by an unrelated oligonucleotide (lane 5). An antibody directed against the NF-κB family member p65 supershifted complex a (lanes 6 and 14), whereas an antibody directed against another NF-κB family member, p50, affected both complexes a and b (lanes 10 and 16), suggesting that complex a contains p65 homodimers and/or heterodimers of p65 and p50 and that complex b contains homodimers of p50. With use of the Fas NF-C/EBP probe, no complex could be competed by a consensus C/EBP oligonucleotide (Fig. 4B, lane 3) and antibodies against C/EBP family members, C/EBPβ and -δ, had no effect (lanes 7 and 8), indicating that C/EBP did not bind the probe and that NF-κB-C/EBP heteromers did not interact with the probe under these experimental conditions. Similarly, no complex could be detected when a probe spanning the C/EBP site from nucleotides −134 to −116 (probe Fas C/EBP) was used (data not shown). This was unexpected because mutation of the C/EBP site abolished IL-1β induction of the promoter activity (Fig. 3B), suggesting that this site is functional. To further clarify this issue, we performed EMSAs, changing the electrophoresis buffer Tris-borate to a buffer of higher ionic strength, Tris-glycine (44). Under these conditions, however, two complexes (c and d) were similarly formed with the probe Fas C/EBP in untreated cells and in cells treated with IL-1β for 30 min up to 8 h (Fig. 5 and data not shown). These two complexes were specifically competed by the unlabeled probe (Fig. 5, lane 2) and by a consensus C/EBP oligonucleotide (lane 3). Antibodies directed against C/EBP family members were used in supershift experiments to further characterize the complexes. An antibody directed against C/EBPα had no effect (lane 4), whereas an antibody directed against C/EBPβ altered the mobility of both complexes (lane 5), and an antibody directed against C/EBPδ disrupted complex d (lane 6). This suggests the presence of C/EBPβ in complex c and the presence of C/EBPβ and -δ in complex d. These experimental conditions, however, were not adequate for NF-κB detection either with the Fas NF-C/EBP probe or a NF-κB consensus probe (data not shown).

To evaluate whether NF-κB and C/EBP factors are able to transactivate the Fas promoter, we performed cotransfection experiments with Fas promoter constructs and expression vectors for NF-κB p65, NF-κB p50, and C/EBPβ. First, the ability of p65 to activate the wild-type pFas-811 luc construct and its mutants was examined. NF-κB p65 transactivated the Fas promoter in a dose-dependent manner (Fig. 6A). At the suboptimal dose (0.01 μg), p65 induced the activity of the promoter construct with the C/EBP site mutated (C/EBPmut) similarly as the wild-type construct (Fig. 6B). The activity of the promoter constructs with either the NF-κB site mutated (NFmut) or the NF-κB and C/EBP sites deleted (ΔNF-C/EBP) was marginally induced compared with the wild-type construct (Fig. 6B), indicating that p65 induction is specific and that the NF-κB site at position −142 is responsible for induction of the Fas promoter activity by NF-κB. To determine the effect of overexpression of p65 together with p50 on Fas promoter activity, the two expression vectors were cotransfected, keeping p65 constant and increasing the amount of p50. Overexpression of p50 alone had no effect on the wild-type promoter activity (Fig. 6C). However, when cotransfected with p65, it inhibited the p65-induced promoter activity in a dose-dependent manner (Fig. 6C).

Overexpression of C/EBPβ transfected at doses up to 0.02 μg had no effect on the activity of pFas-811 luc (Fig. 6D). At higher doses (up to 0.1 μg), C/EBPβ induced the luciferase activity in a nonspecific way since the activity of the promoterless vector pGL3 was also increased (data not shown). When cotransfected at increasing amounts (up to 0.02 μg) with a constant amount of p65, C/EBPβ inhibited the p65-induced promoter activity in a dose-dependent manner (Fig. 6D). Western blot analysis for p65, p50, and C/EBPβ showed efficient nuclear expression of the different proteins (Fig. 7). Moreover, the expression of p65 was not affected by the expression of p50 or C/EBPβ (Fig. 7).

The Fas pathway may contribute to β-cell death in type 1 diabetes. Fas expression is induced by cytokines on β-cells and renders them susceptible to apoptosis mediated by FasL expressed on islet-infiltrating T-cells and macrophages (13,15,16,25). The mechanism or mechanisms by which Fas gene expression is induced by cytokines in β-cells or in other cell types remain to be investigated.

In the present study, we show that Fas mRNA expression is induced by IL-1β alone in rat β-cells, whereas IFN-γ has no effect. We identified the transcription factors NF-κB and C/EBP as key regulators of rat Fas promoter activity. First, mutation of either the NF-κB binding site located at −142 bp or the C/EBP binding site located at −130 bp abolished IL-1β induction of Fas promoter activity. Second, we detected by EMSAs and supershift experiments the binding of NF-κB p65 and p50 factors as well as C/EBPβ and -δ factors to their respective sites. Third, in cotransfection experiments, NF-κB p65 transactivated the Fas promoter. Coexpression of NF-κB p50 or C/EBPβ with p65 resulted in a dose-dependent inhibition of the p65-induced Fas promoter activity.

Interactions between the members of the NF-κB family have been characterized in detail (45). Thus, p65/p50 heterodimers, the most common dimers, and p65 homodimers are activators of gene transcription, whereas p50 homodimers, which lack a transactivation domain, are transcriptionally inactive (46). Our cotransfection experiments with p65 and p50 did not show synergistic activation by the two factors but resulted instead in inhibition of p65-driven Fas promoter activity. This suggests that p65 homodimers, and not p65/p50 heterodimers, are the main activators of Fas gene transcription. Similar observations were made for the regulation of monocyte chemoattractant protein-1 promoter in mesangial cells (47) and fibroblasts (48), and intercellular adhesion molecule-1 (ICAM-1) promoter in astrocytes (49). p50 Homodimers may be formed at higher p50:p65 ratios and function as repressor of Fas promoter activity by interfering with p65 binding. In line with this possibility, we observed by EMSAs that p50 homodimers may bind to the NF-κB site of the promoter (Fig. 4).

Cooperative binding and physical association between NF-κB and C/EBP proteins have been described as mechanisms for synergistic transactivation of proinflammatory (40,41,42,43) and viral genes (50). This was not observed in the present experiments. Both factors bound independently to their respective sites, and no NF-κB-C/EBP heteromers were detected in EMSAs. This could, however, be due to the different experimental conditions required to detect NF-κB and C/EBP binding. There was no synergistic activation of the Fas promoter when NF-κB p65 and C/EBPβ were coexpressed in transfection studies. Instead, overexpression of C/EBPβ had an inhibitory effect on p65-induced promoter activity. These findings are similar to those observed for the transcriptional regulation of ICAM-1 in astrocytes (49) and of TNF-α in macrophages A(31). The transactivation capacity of C/EBPβ is enhanced by phosphorylation of its activation domain (51). Thus, it cannot be excluded that IL-1β induces posttranslational modification of C/EBPβ, preventing the presently observed inhibitory effect of C/EBPβ on Fas promoter activity. Furthermore, cofactors may be required for functional synergism between NF-κB and C/EBP on the Fas promoter. For example, CREB-binding protein/p300 are required for maximal transcriptional activity of E-selectin promoter by NF-κB and ATF-2/c-Jun (52) and IL-6 promoter by NF-κB, C/EBP, CREB, and AP-1 (53). High-mobility group I(Y) proteins are required for proper regulation of the IL-2R α gene by NF-κB and Elf-1 (54) and of the IFN-β gene by NF-κB, ATF-2/c-Jun, and IRF (55).

In conclusion, we have demonstrated for the first time that NF-κB plays a critical role in the regulation by cytokines of the rat Fas promoter in insulin-producing cells. This novel information provides the rationale to target NF-κB activation, by genetic intervention, in future attempts to prevent β-cell destruction in animal models of type 1 diabetes and allograft rejection.

This work was supported by grants from the Juvenile Diabetes Foundation International, the Fund for Scientific Research Flanders, and by a Shared Cost Action in Medical and Health Research of the European Community.

We thank the personnel involved in β-cell purification: E. Quartier for sequencing, M.-C. Chen for providing some RNA samples, and H. Liu and R.M. Pope for providing the expression vectors.