Transcriptional Regulation of the Endoplasmic Reticulum Stress Gene Chop in Pancreatic Insulin-Producing Cells

Insulin-producing INS-1E cells were cultured in RPMI-1640 medium (Invitrogen, Paisley, U.K.) supplemented with 5% FCS (29,30). Recombinant rat IFN-γ (R&D Systems, Abingdon, U.K.) was used at 0.18 μg/ml (corresponding to 500 units/ml IFN-γ) and human recombinant IL-1β (gift from Dr. C.W. Reinolds, National Cancer Institute, Bethesda, MD), at 10 units/ml. The inducible nitric oxide synthase (iNOS) blocker aminoguanidine (Sigma, Steinheim, Germany) was used at 2.5 mmol/l, and culture medium was collected for nitrite determination (nitrite is a stable product of NO oxidation) using the Griess method (31). Palmitate (Sigma) was solubilized in ethanol and used at a concentration of 0.5 mmol/l. For exposure to palmitate, RPMI-1640 medium containing 1% BSA and 1% FCS was used (6). CPA (Sigma) was dissolved in DMSO and used at a concentration of 25 μmol/l. The respective control conditions were for cytokines, medium alone; for palmitate, 1% ethanol; and for CPA, 0.25% DMSO. These concentrations of ethanol or DMSO do not affect β-cell viability or gene expression (6,12). Cytokines, aminoguanidine, palmitate, and CPA concentrations were selected based on our previous time course and dose-response studies (data not shown) (6,12,30). The presently used INS-1E cells have a well-preserved insulin release in response to glucose (29) and respond to cytokines (5,30), palmitate (6), and CPA (5,12) in a similar way as primary β-cells.

INS-1E cells, treated as described in the figures, were harvested for mRNA isolation and subsequent reverse transcription (32). Chop mRNA expression was determined by real-time RT-PCR using SYBR Green fluorescence on a LightCycler instrument (Roche, Manheim, Germany) using the standard curve method (6). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), whose expression is not modified under the present experimental conditions (5,6,12,30) (data not shown), was used as a housekeeping gene. Results are shown as the expression of the gene of interest divided by GAPDH. Primer sequences for Chop and GAPDH are described elsewhere (5). For RNA stability experiments, INS-1E cells were exposed for 16 h to IL-1β + IFN-γ or palmitate to induce Chop mRNA expression. Medium was then replaced and new RNA formation was blocked with actinomycin D (4 ng/ml); culture was then continued for an extra 240 min with actinomycin D in the presence or absence of cytokines or palmitate.

The following hamster Chop promoter constructs tethered to a luciferase reporter gene were used 1) wild type (−782 to 21 bp); 2) mutants in which the 5′ 345, 553, and 719 bp of the promoter were selectively deleted; and 3) mutants for C/EBP-ATF site and endoplasmic reticulum stress element (ERSE). Oligonucleotide reporter constructs containing two copies of the Chop C/EBP-ATF composite site or mutants at either the C/EBP or ATF portion of the site were also tested. These constructs, originally cloned in the JymCATO plasmid (27), were extracted by enzymatic digestion as follows 1) wild-type and 5′ deletion mutant of the Chop promoter as well as C/EBP-ATF oligonucleotide reporter construct were cut with SmaI and HindIII and 2) C/EBP-ATF and ERSE mutant were cut with NcoI followed by blunt-end formation using Klenow enzyme and finally digested with HindIII. The fragments obtained were subcloned between the ClaI and HindIII sites of the pGL3 basic luciferase construct (Promega, Madison, WI). The substitution mutation of the AP-1 site was made using the QuickChange II site-directed mutagenesis kit (Stratagene, La Jolla, CA). The AP-1 site TGACTCA was changed to gGACTCA using as template the wild-type and C/EBP-ATF mutant Chop promoter constructs. A construct containing the Chop promoter plus the first exon of the Chop gene was generated by PCR amplification of the exon (−13 to +121 bp) using the following primers: 5′-CGCTGCAGAGGTCAGTAGCCTGAGTCTCACACTTC-3′ (forward) and 5′-CTGAAGCTTCACTTCCTTCTGGAACACTCGCTCC-3′ (reverse). A HindIII restriction site was added at the 5′ extremity of the reverse primer (underlined sequence), and the PCR product was then cloned in the Chop promoter luciferase construct between the PstI and HindIII sites. INS-1E cells were transfected using lipofectamine reagent (Invitrogen) with 250 ng of the different luciferase reporter constructs and the pRL-CMV plasmid (50 ng, with Renilla used as internal control for transfection efficiency) (33). Twenty-four hours after transfection, the cells were exposed for 36 h to IL-1β + IFN-γ or palmitate or for 15 h to CPA. Luciferase activities of cell lysates were then determined and expressed as Firefly/Renilla (relative luciferase activity).

For the C/EBP-ATF electrophoretic mobility shift assays (EMSAs), 10 μg nuclear extracts were prepared and electrophoresed on 4% Tris glycine EDTA gels as described previously (27). For the AP-1 EMSAs 5 μg nuclear extracts were prepared and electrophoresed in 5% Tris borate EDTA gels (34). For supershift analysis, 2 μg BSA or antibody (ATF4 [sc-200], C/EBP-β [sc-150], c-Fos [sc-52], c-Jun [sc-45], or Jun-B [sc-73]; Santa Cruz Biotechnologies, Santa Cruz, CA) was used. The double-stranded synthetic oligonucleotides probes (C/EBP-ATF, 5′-AGAAACATTGCATCACCC-3′; and AP-1, 5′-AGCGCGCATGACTCACTCAAAT-3′) were labeled using Klenow enzyme and [γ³²P]CTP. Specificity was assessed using nonradioactive wild-type or mutant oligonucleotides (C/EBP-ATF mutant, 5′-AGAAAcgcgtcccTCACCC-3′; or AP-1 mutant, 5′-AGCGCGCAgGACTCACTCAAAT-3′).

Cells were lysed with either whole-cell extract buffer (1% NP40, 25 mmol/l HEPES, pH 7.8, 50 mmol/l KCl₂, 10 mmol/l NaF, 1 mmol/l Na₃VO₄, 1 mmol/l phenylmethylsulfonyl fluoride [PMSF], and protease inhibitor cocktail) or phospho–whole-cell extract buffer (1% NP40, 1 mmol/l CaCl₂, 1 mmol/l MgCl₂, 150 mmol/l NaCl, 10 mmol/l NaF, 1 mmol/l Na₃VO₄, 1 mmol/l PMSF, and protease inhibitor cocktail [Roche]). Lysates were then subjected to a 10% SDS-PAGE and transferred to a nitrocellulose membrane. Immunoblot analysis was performed with the indicated concentrations of antibodies against Chop (G6916; 4.8 μg/ml; for the cytokines or CPA-treated experiments) from Sigma; ATF4 (sc-200; 0.8 μg/ml), eIF2α (sc-11386; 0.2 μg/ml), and Chop (sc-575; 0.3 μg/ml; for the palmitate Western blot) from Santa Cruz; and phospho-eIF2α (9721; Ser51; 16.7 ng/ml), phospho–c-Jun (9261; Ser63; 9 ng/ml), and β-actin (4967; 8 ng/ml) from Cell Signaling (Danvers, MA). A secondary anti-rabbit horseradish peroxidase–labeled IgG antibody (Santa Cruz Biotechnologies) was used for detection. Western blot pictures were quantified, and values of the protein of interest were corrected for β-actin or total eIF2α considering the maximum value in each experiment as 1.

Data are shown as means ± SE, and comparisons between groups were made by paired t test or by ANOVA followed by t test with the Bonferroni correction, as indicated.

Twenty-four–and 36-h treatment of INS-1E cells with cytokines led to a 30- and 15-fold increase, respectively, of Chop mRNA expression (Fig. 1A), whereas palmitate induced a 6- and 2-fold upregulation of Chop transcript, respectively (Fig. 1B). A 22-fold induction of Chop was observed after 15 h of exposure to CPA (Fig. 1C). Western blot time course experiments were also performed to analyze Chop protein induction by cytokines, palmitate, or CPA, indicating clear upregulation in protein expression after 8–16 h (Supplementary Fig. S1 in the online appendix [available at http://dx.doi.org/10.2337/db06-1253]). INS-1E cells were also exposed to cytokines with or without the iNOS blocker aminoguanidine. Aminoguanidine did not affect basal NO production, but it prevented to a large extent cytokine-induced NO formation (Supplementary Fig. S2). Aminoguanidine completely blocked cytokine-induced Chop mRNA expression (Fig. 1A), confirming that the effect of cytokines is mediated via NO formation. Because neither palmitate nor CPA trigger NO production in INS-1E cells (5,6), we did not study the role of NO in Chop induction by these agents. Neither cytokines (Fig. 1D) nor palmitate (Fig. 1E) increased Chop mRNA stability, and after 4 h of actinomycin exposure, Chop mRNA expression was similar in both treated and untreated cells. Thus, Chop mRNA induction by cytokines or palmitate in INS-1E cells results mainly from transcriptional activation.

To identify the regions that are responsible for the activation of the Chop promoter by cytokines/NO, palmitate, or CPA, we performed transient transfections with luciferase reporter constructs containing progressive 5′ deletions of the Chop promoter (Fig. 2A). There was a basal Chop activity with the −782-bp wild-type promoter that increased two- to threefold after cytokine (Fig. 2B), palmitate (Fig. 2C), or CPA (Fig. 2D) exposure. As observed for Chop mRNA expression (Fig. 1A), activation of the Chop promoter by cytokines was abolished by aminoguanidine (Fig. 2B). Aminoguanidine alone did not modify Chop promoter activity (Supplementary Fig. S3).

It was previously described that the first exon of the Chop gene contains regulatory elements (28). Hence, we prepared a luciferase reporter construct containing the −782 bp of the hamster Chop promoter plus the first exon (120 bp). We did not observe, however, any effect of exon 1 in the induction by IL-1β + IFN-γ or palmitate (Supplementary Fig. S4).

Removal of nucleotides −782 to −437 bp (deletion 1) had no effect in Chop induction by the tested agents (Fig. 2), but deletion of nucleotides −437 to −229 bp (deletion 2) led to a decrease in basal activity (Fig. 2), a partial impairment of CPA induction (Fig. 2D), and abolished induction of the Chop promoter by cytokines (Fig. 2B) and palmitate (Fig. 2C). Deletion of downstream nucleotides −229 to −63 bp (deletion 3) further decreased basal activity (Fig. 2) and completely abolished CPA induction of the Chop promoter (Fig. 2D). Based on these data, we identified one promoter region (−437 to −229 bp) necessary for CPA, cytokine, and palmitate induction of the Chop promoter and another region (−229 to −63 bp) that is mostly involved in induction by CPA. The first region contains a C/EBP-ATF composite and an AP-1 binding site, whereas the second contains an ERSE (Fig. 2A) (24,27). Mutation of the C/EBP-ATF composite site (see scheme on Fig. 3A) decreased basal promoter activity and Chop promoter induction by cytokines (Fig. 3B), palmitate (Fig. 3C), or CPA (Fig. 3D). Mutation of the AP-1 site (Fig. 3A) led to a small increase in basal promoter activity and impaired Chop induction by cytokines (Fig. 3B) and palmitate (Fig. 3C) but had no effect of on promoter induction by CPA (Fig. 3D). A double mutant for both C/EBP-ATF and AP-1 binding sites (Fig. 3A) abolished induction of luciferase activity by cytokines (Fig. 3B) or palmitate (Fig. 3C). Deletion of the ERSE binding site (Fig. 3A) specifically impaired induction by CPA (Fig. 3D), whereas double mutation of C/EBP-ATF site and ERSE greatly decreased CPA-induced promoter activity (Fig. 3D). These results suggest that CPA induces the Chop promoter mainly via the C/EBP-ATF and ERSE binding sites but not through the AP-1 site. Mutation of the ERSE site did not affect induction of Chop promoter activity by cytokines or palmitate (Fig. 3B and C).

In the C/EBP-ATF composite site, the C/EBP (5′-CATTGC) and ATF (5′-ATCATC) binding sequences are present in tandem. To further investigate the role of these two binding sites, we used luciferase reporter constructs containing two copies of the C/EBP-ATF composite site or mutants at either the C/EBP or ATF sites (Fig. 4A). The wild-type construct was induced around twofold in response to cytokines (Fig. 4B) or palmitate (Fig. 4C). Mutations of either the C/EBP or ATF portions of the site severely decreased basal promoter activity (Fig. 4B and C). Mutation of the ATF site, but not of the C/EBP site, inhibited activation of the construct by cytokines (Fig. 4B) or palmitate (Fig. 4C). These results suggest that the ATF portion of the C/EBP-ATF composite site is involved in Chop induction and point to ATF4 as an important transcription factor for Chop activation. Because induction of ATF4 is secondary to eIF2α phosphorylation (4), we evaluated eIF2α phosphorylation and ATF4 induction by cytokines, palmitate, or CPA. Phosphorylation of eIF2α in response to IL-1β + IFN-γ was already observed after 4 h and remained present up to 24 h (Fig. 4D). The induction of ATF4 by cytokines was detected after 16 h of exposure and was still present at 24 h (Fig. 4D). Blocking cytokine-induced NO production by aminoguanidine prevented eIF2α phosphorylation and decreased ATF4 and Chop protein induction (Fig. 4E). Palmitate induced a prolonged (up to 24 h) upregulation in eIF2α phosphorylation and ATF4 expression, already detectable after 4 h and increasing further in the subsequent hours (Fig. 4F). CPA induced an early eIF2α phosphorylation and ATF4 induction that remained detectable until 24 h (Fig. 4G).

To identify the cytokine- and palmitate-induced protein complexes binding to the C/EBP-ATF composite site of the Chop promoter, EMSA was performed using a probe containing the wild-type C/EBP-ATF site. Under control conditions, two different complexes were identified (Fig. 5, complexes α and β). Competition assays with 20-fold excess cold wild-type or mutated probes demonstrated the specificity of these complexes (Fig. 5A, lanes 11–14, and D, lanes 11 and 12). Complex β was induced in response to cytokines or palmitate after 16 and 24 h of exposure (Fig. 5A and D, lanes 8 and 10). Blocking NO production with aminoguanidine prevented cytokine-mediated induction of complex β (Fig. 5B, compare lanes 2 and 4). An ATF4 antibody completely supershifted the cytokine- and palmitate-induced complex β (Fig. 5C and E, lane 4). On the other hand, the C/EBPβ antibody partially supershifted the constitutive complex α but did not affect complex β (Fig. 5C and E, lanes 5 and 6).

Both cytokines and palmitate induced a specific complex that binds to the radiolabeled probe containing the AP-1 binding site (Fig. 6A and C). This complex is barely detectable under control conditions but is induced after 4 h of cytokine or palmitate exposure (Fig. 6A and C). In cytokine- or palmitate-treated cells, there was a maximum AP-1 binding at 8 and 4–8 h, respectively, that remained detectable up to 24 h (Fig. 6A and C). Specificity of the complex was confirmed by competition assays using a 20-fold excess of cold wild-type or mutant AP-1 cold probes (Fig. 6A, lanes 11 and 12, and C, lanes 11–14). Cytokine-induced formation of the AP-1 complex was mildly reduced (∼20–30%) when the cells were treated with cytokines plus aminoguanidine (Fig. 6B). In INS-1E cells treated with either cytokines or palmitate, the c-Fos antibody induced a shift in the AP-1 complex (Fig. 7A and B, lane 4). The c-Jun antibody induced a shift in cells treated with cytokines (Fig. 7A, lane 10) but not in cells treated with palmitate (Fig. 7B, lane 8). As observed in Fig. 7A (lane 8) and B (lane 10), Jun-B antibody shifted both the cytokine- and palmitate-induced AP-1 complexes. To confirm the activation of c-Jun by cytokines, we examined c-Jun phosphorylation on serine 63 (Fig. 7C). Cytokines induced c-Jun phosphorylation already after 4 h of exposure, with a peak at 8 h and return to basal levels by 24 h. In agreement with the supershift experiments (Fig. 7B), palmitate did not trigger c-Jun phosphorylation in INS-1E cells (data not shown). These results indicate that cytokines induce AP-1 dimers composed of c-Fos, c-Jun, and/or Jun-B, whereas palmitate induces AP-1 dimers composed preferentially of c-Fos and Jun-B.

Several reports have shown that Chop may be induced at the transcriptional and/or posttranscriptional level by a variety of agents that cause cellular stress (15). The elements controlling Chop transcription vary depending on the cell type and the kind of stress induced (24–28). This report provides a comprehensive analysis of the regulation of Chop mRNA expression in β-cells by different pro-apoptotic endoplasmic reticulum stressors, namely a chemical (CPA) and two physiological endoplasmic reticulum stress inducers (cytokines and palmitate).

In agreement with previous experiments (5,6,12), we observed that Chop mRNA expression is induced in insulin-producing cells by IL-1β + IFN-γ (via NO formation), palmitate, and CPA. This increase is due to augmented gene transcription, because both cytokines and palmitate increased activity of the Chop promoter without modifying Chop mRNA stability. We identified the C/EBP-ATF composite and AP-1 sites as key regulatory elements for Chop induction in response to cytokines/NO and palmitate. Differently, Chop regulation by the chemical endoplasmic reticulum stress inducer CPA involves the C/EBP-ATF and ERSE binding sites but not the AP-1 site. The SERCA blocker thapsigargin was previously shown to regulate Chop mRNA transcription in fibroblasts in a similar fashion (27). By using a luciferase reporter construct containing wild-type and mutated forms of the C/EBP-ATF composite site, we identified the ATF portion of the site as responsible for both cytokine and palmitate induction. In agreement with this, EMSA identified ATF4 as the main transcription factor of the inducible complex activated by both agents. In endoplasmic reticulum–stressed fibroblasts, two inducible complexes involving ATF4 were observed: one formed by ATF4 only and another containing both ATF4 and C/EBPβ transcription factors (27). At variance with these observations, we did not observe C/EBPβ in the complexes induced by cytokines or palmitate: C/EBPβ was only involved in a constitutive complex (Fig. 5, complex α). There was a good correlation between the kinetics of ATF4 protein induction and binding to the C/EBP-ATF site in response to cytokines, but we observed a discrepancy between the early ATF4 protein induction and its late binding to the C/EBP-ATF site in response to palmitate. This suggests that another transcription factor besides C/EBPβ is necessary for the binding of ATF4 to the C/EBP-ATF site. One possible candidate is ATF2, which binds to this site in amino acid–starved Hela cells (35). Another possibility is that posttranslational regulation of ATF4, such as phosphorylation (36), contributes to this time lag.

Induction of ATF4 protein occurs downstream of eIF2α phosphorylation (4,37). We have presently shown that cytokines (Fig. 4D), palmitate (Fig. 4F), and CPA (Fig. 4G) induce eIF2α phosphorylation in INS-1E cells. Four kinases, namely PERK, general control nonderepressible-2, heme-regulated inhibitor, and double-stranded RNA–induced protein kinase can phosphorylate eIF2α protein (37), but PERK is the main mediator of this effect during endoplasmic reticulum stress. Palmitate activates PERK kinase leading to eIF2α phosphorylation and ATF4 induction (7). Of note, cytokine-induced eIF2α phosphorylation (present data) may explain the early decrease in (pro)insulin and total protein biosynthesis in β-cells exposed to IL-1β + IFN-γ (38).

Analysis of the AP-1 site by EMSA revealed an early binding (4 h) induced by both cytokines and palmitate; this was maintained up to 24 h but at a reduced level. Supershift experiments indicated that the AP-1 dimer binding to the Chop promoter after palmitate exposure is composed of Jun-B/c-Fos, whereas cytokine exposure induces Jun-B/c-Fos and c-Jun/c-Fos in INS-1E cells. Jun-B/Jun-B and Jun-B/c-Jun are mostly transcription repressors, and c-Jun/c-Jun homodimers bind weakly to DNA (39). IL-1β induces a rapid (30 min) increase in c-Jun and c-Fos expression peaking at 3–5 h (40–42); this peak coincides with the AP-1 binding to the Chop promoter observed in the present experiments. The transactivating potential of c-Jun is regulated both at the level of expression and by posttranslational modifications (39), such as phosphorylation of c-Jun on Ser-63 and Ser-73 by JNK, which enhances its transactivating activity (43). We observed that cytokines, but not palmitate, induce Ser-63 phosphorylation of c-Jun in INS-1E cells. This lack of palmitate effects on Ser-63 phosphorylation correlates with the lack of c-Jun supershift. Phosphorylation of c-Jun at Ser-63 is pro-apoptotic in neuroblastoma cells (44), and IL-1β–induced JNK activation contributes to β-cell death (45,46). During prolonged endoplasmic reticulum stress, IRE1 activates JNK via TRAF2 (47,48). Cytokine-induced JNK activation in INS-1E cells, however, is an early (1 h) and NO-independent event, whereas the endoplasmic reticulum stress response induced by cytokines in these cells is an NO-dependent and late (>4–6 h) phenomenon (5). Additional studies are required to clarify the role of the JNK–AP-1 pathway in cytokine-induced β-cell apoptosis.

Cytokines, via NO production, induce a progressive increase of Chop transcript in INS-1E cells, starting at 8 h and peaking at 24 h (5). Chop mRNA induction by palmitate increases later, ∼12 h (4.9 ± 1.1–fold increase vs. control, n = 4), with peak at 24 h (6; present data). Both cytokines and palmitate induce AP-1 binding at early time points (starting at 4 h), whereas binding of ATF4 to the C/EBP-ATF occurs later (16 h). Based on the present findings, Fig. 8 provides an overview of our current understanding of Chop induction in insulin-producing cells. The early induction of Chop mRNA by cytokines is mediated by AP-1 transcription factors, whose effects are enhanced by the subsequent binding of ATF4 to C/EBP-ATF site. The early binding of AP-1 to the Chop promoter is NO independent, whereas the subsequent cytokine induction of ATF4 and its binding to the C/EBP-ATF binding site is NO dependent. In the case of palmitate, occupation of both AP-1 and C/EBP-ATF sites are necessary for promoter activation.

In conclusion, we show that both AP-1 and ATF4 transcription factors are involved in Chop induction in INS-1E cells in response to cytokines and palmitate and that at variance with CPA treatment, neither of these agents uses the ERSE. Although ATF4 is probably activated via the UPR/PERK pathway for both cytokines and palmitate, induction of AP-1 by cytokines seems to be endoplasmic reticulum stress independent (Fig. 8).

It is puzzling that a branch of the UPR, namely the PERK/ATF4 pathway, usually considered as a β-cell survival mechanism (4) also induces the pro-apoptotic transcription factor Chop. Chop action occurs essentially via nuclear translocation and subsequent regulation of target genes (15,17–19). Chop cannot homodimerize (18), suggesting that a lack of suitable partners in the nucleus might prevent its effects. Posttranslational modification of Chop can also affect its action, because phosphorylation by p38 kinase enhances Chop transcriptional activity (49), whereas phosphorylation by Casein kinase-2 has an inhibitory effect (50). Additional studies are thus required to clarify the activation and the role of Chop in endoplasmic reticulum stress–induced β-cell apoptosis.

The present observations indicate similarities but also subtle differences in the way endoplasmic reticulum stress caused by different agents activates Chop in β-cells. Further understanding of the signal transduction of cytokines and palmitate will hopefully allow the development of specific maneuvers to modulate the pro-apoptotic components of the UPR response while boosting the UPR “defense” components.

A.K.C. is the recipient of a postdoctoral fellowship from the Juvenile Diabetes Research Foundation and is presently a Research Associate of the Fonds National de la Recherche Scientifique (FNRS; Belgium). This work has received grants from the European Union (Integrated Project EuroDia LSHM-CT-2006-518153 in the Framework Programme 6 of the European Community), the FNRS, and the Communauté française de Belgique-Actions de Recherche Concertées.

We thank the personnel from the Laboratory of Experimental Medicine, ULB, especially M.A. Neef, G. Vandenbroeck, and R. Leeman for excellent technical support and Drs. Z. Dogusan and P. Hekerman for advice and help with phospho–Western blot analysis.