Molecular Regulation of Monocyte Chemoattractant Protein-1 Expression in Pancreatic β-Cells

The MCP-1 promoter and enhancer region were isolated by PCR using the Expand High Fidelity PCR system (Roche Molecular Biochemicals, Mannheim, Germany) on 0.5 μg rat genomic DNA prepared from liver as described (36). The primers used were based on the Genbank sequence (accession number AF079313). For the promoter region, the forward primer (position −515 to −499) was 5′-CGAATTCAAGCTCTTCGGTTTGG with an added EcoRI site, and the reverse primer (position +31 to +53) was 5′-CGGGATCCGGCTTCAGTGAGAG with an added BamHI site. For the enhancer region, the forward primer (position −2,478 to −2,459) was 5′-GCTCTAGACCAGTAGAGGCTCAATCC with an added XbaI site, and the reverse primer (position −2,464 to −2,180) was 5′-CGGAATTCACTTTACTCTGTCAAC with an added EcoRI site. The amplified products were cloned into pBluescript KS (Stratagene, La Jolla, CA). Clones were sequenced and shown to be 99% identical to the published sequence. The promoter region (−514 to +53) was isolated from pBluescript by digestion with EcoRV (5′ end) and BamHI (3′ end). It was later cloned into the SmaI-BglII sites of pGL3-Basic (Promega, Madison, WI) to obtain the pMCP-1-514 luc construct. To build the pMCP-1-514(enh) luc construct, the promoter region was isolated by EcoRI (5′ end)-BamHI (3′ end) digestion, and the enhancer region was isolated by XbaI (5′ end)-EcoRI (3′ end) digestion from the pBluescript. The promoter and the enhancer fragments were ligated into the BglII and NheI sites of pGL3-Basic. Mutations for the two NF-κB sites at positions −2,250 and −2,276 were generated on pMCP-1-514(enh) luc by PCR (37) using the Expand High Fidelity PCR system. The pairs of primers were as follows: 5′-CTGAAGGGTCTGC̅C̅AACTTCCAATAC and 5′-GAGAGATATTTAGCTGGAATATGAGC for mutation in NF-κB site A and 5′-CAATACTGCCTCAGAATC̅C̅GAATTTCCACAC and 5′-GAAGTTCCCAGACCCTTCAGGAGAG for mutation in NF-κB site B (mutated bases are underlined). All mutations were confirmed by sequencing.

Insulin-producing INS-1E cells, a kind gift from Prof. C. Wollheim, Center Medical Universitaire, Geneva, Switzerland, were cultured in RPMI-1640 (with Glutamax-1) supplemented with 10 mmol/l HEPES, 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, 100 μg/ml streptomycin, 1 mmol/l sodium pyruvate, and 50 μmol/l 2-mercaptoethanol (38). RINm5F insulinoma cells were cultured in RPMI-1640 medium with Glutamax-1 supplemented with 10% FCS. Islets were isolated from the pancreas of male Wistar rats by collagenase digestion in cold Hank’s salt solution followed by hand picking under a stereomicroscope. Rat β-cells were FACS purified from the islets, as previously described (39). These preparations contained >95% viable β-cells and were cultured in Ham’s F-10 medium supplemented with 10 mmol/l glucose, 2 mmol/l glutamine, 0.5% BSA, and 50 μmol/l 3-isobutyl-1-methylxanthine. The presence of 3-isobutyl-1-methylxanthine at this concentration preserves β-cell survival in culture (40) with minimal effects on cAMP formation in the absence of adenylate cyclase activators (41). INS-1E and primary β-cells were cotransfected with the luciferase test plasmids, and pRL-CMV (Promega) as an internal control, by lipofection with lipofectAMINE (Invitrogen, Baesley, Scotland), as previously described (42), and were exposed for 16 h to recombinant human IL-1β (10–50 units/ml; kindly provided by Dr. C. W. Reynolds, National Cancer Institute, Bethesda, MD) and recombinant murine IFN-γ (100–1,000 units/ml; Invitrogen) in various combinations. The concentrations of cytokines used in the experiments with primary β-cells and INS-1E cells were selected based on dose-response studies (2 and B.K. and D.L.E., unpublished data). Luciferase activities were assayed with the Dual-Luciferase Reporter Assay System (Promega). Values obtained for the test plasmids in unstimulated cell extracts were at least 90-fold higher than values for the promoterless vector pGL3-Basic (0.1 light unit). Test values were corrected for the luciferase activity value of the internal control plasmid, pRL-CMV. The results for cytokine-exposed cells are expressed as a fold-induction of the luciferase activity of the same construct in control condition, taking control (no cytokine added) value as 1.

Nuclear extracts were prepared from INS-1E cells as described (43). Nuclear proteins (4 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 4°C before addition of 50 molar excess of competing oligonucleotide, as indicated, and radiolabeled probe (40,000 cpm). The incubation was continued for 20 min at 4°C. Where indicated, 1 μl antibodies specific for p50, p65, or Stat1 (Santa Cruz Biotechnology, Santa Cruz, CA) were added for 20 min at room temperature. Stat1 antibody was used as a control for nonspecific binding. The samples were electrophoresed on 5% polyacrylamide gels in 25 mmol/l Tris, 25 mmol/l boric acid, and 0.5 mmol/l EDTA. Oligonucleotides for electrophoretic mobility shift assays (EMSAs) were as follows (upper strand shown): MCP-1 enhancer NF-κB site A, 5′-agAAGGGTCTGGGAACTTCCAATA; MCP-1 enhancer NF-κB site B, 5′-agAGAATGGGAATTTCCACACTCTT; and NF-κB consensus, 5′-agCTTCAGAGGGGACTTTCCGAGA. The lowercase letters represent added nucleotides to allow end labeling with the Klenow fragment.

Cross-linked chromatin from 5.10⁷ RINm5F cells, left untreated or exposed to IL-1β for 30 min, 2 h, or 4 h, was prepared as described (44), with the following modifications: β-glycerophosphate and sodium butyrate were omitted and 1 mmol/l phenylmethylsulfonyl fluoride, 1 μg/ml leupeptin, and 1 μg/ml aprotinin were added as protease inhibitors. Samples were sonicated with a Branson Sonifier 250 for four 20-s pulses. SDS was added at a final concentration of 1%, and samples were rotated for 1 h at room temperature and centrifuged at 12,000 rpm for 15 min. Supernatants were further sonicated for two 20-s pulses to an average size of 500–1,500 bp. For immununoprecipitation, 60 μl chromatin corresponding to 5.10⁶ cells was diluted 10-fold in 0.01% SDS, 1.1% Triton X-100, 155 mmol/l NaCl, 1 mmol/l EDTA, and 10 mmol/l Tris-HCl (pH 8) with protease inhibitors. The chromatin solution was precleared with 20 μl 50% protein A-Sepharose containing 250 μg/ml BSA and 100 μg/ml sonicated herring sperm DNA for 45 min at 4°C. Precleared chromatin (20 μl saved as total input chromatin) was rotated without or with 3 μg rabbit polyclonal antibody overnight at 4°C. The antibody was directed against p65. As a negative control, cross-linked chromatin was incubated overnight in the absence of antibody. Protein A-Sepharose (50 μl) containing BSA and herring sperm DNA was added, and the incubation was continued for 3 h at 4°C. Immune complexes were washed twice with each of the following buffers: low salt buffer (0.1% SDS, 1% Triton X-100, 150 mmol/l NaCl, 2 mmol/l EDTA, and 20 mmol/l Tris-HCl, pH 8), high-salt buffer (0.1% SDS, 1% Triton X-100, 500 mmol/l NaCl, 2 mmol/l EDTA, and 20 mmol/l Tris-HCl, pH 8), LiCl buffer (0.25 mol/l LiCl, 1% Igepal, 1% deoxycholate, 1 mmol/l EDTA, and 10 mmol/l Tris-HCl, pH 8), TE buffer (1 mmol/l EDTA and 10 mmol/l Tris-HCl, pH 8). Immune complexes were eluted with two times 200 μl 0.1 mol/l NaHCO₃, 1% SDS for 15 min. Cross-links were reversed by addition of NaCl at a final concentration of 200 mmol/l and incubation overnight at 65°C. Samples were treated with 40 μg/ml proteinase K for 1 h at 42°C, extracted with phenol-chloroform and chloroform, and ethanol-precipitated. Pellets were resuspended in 25–30 μl H₂O.

For quantification of the precipitated sequences, real-time PCR of MCP-1 enhancer sequence was performed with a single-step PCR on a LightCycler instrument (Roche Diagnostics, Mannheim, Germany) as described previously (45,46). The rat MCP-1 enhancer region was specifically detected using the same primers as those used for the plasmid construction (see above). PCR amplification using a LightCycler instrument was carried out in 20 μl reaction containing 3.5 mmol/l MgCl₂, 0.5 μmol/l forward and reverse primers, 2 μl FastStart SYBR Green 1 mix (Roche), and 2 μl template genomic DNA/plasmid in a glass capillary tube. The amplification of the MCP-1 enhancer was started with an initial activation step at 95°C (10 min) followed by 45 rounds of amplification at 95°C (10 s) for denaturation, 58°C (10 s) for annealing, and 72°C (14 s) for extension.

External standards for MCP-1 enhancer sequence were prepared by 10-fold serial dilutions of a plasmid containing the rat MCP-1 enhancer region; each dilution was ascribed an arbitrary copy number. Each experimental run consisted of six dilutions of the external standard, a negative control without a template, and the samples, containing total input DNA or immunoprecipitated products. The real-time PCR assays were repeated twice. The MCP-1 enhancer content of the immunoprecipitated reactions were determined as follows: the values obtained for each p65 antibody immunoprecipitation were corrected by substracting their corresponding background values (no Ab added) and further normalized by their individual input chromatin values to equalize 100%. The results were expressed in arbitrary units, defined by comparisons with the serial dilutions of the standard plasmid preparation (see above).

β-Cells were cultured overnight, as described above, and then infected with control virus AdGFP, encoding the green fluorescence protein gene (47), or with AdIκB^(SA)2, expressing a NF-κB-super-repressor (see below). As an additional control, a group of β-cells was left noninfected. The recombinant replicative-deficient adenovirus containing a mutated nondegradable IκBα (AdIκB^[SA]2) was prepared as previously reported (48) and was a kind gift from Dr. C. Jobin, Division of Digestive Diseases and Nutrition, University of North Carolina, Chapel Hill, North Carolina. β-cells were infected at a multiplicity of infection 7.5 for 2 h at 37°C. We have previously demonstrated that AdIκB^(SA)2, at this multiplicity of infection, prevents cytokine and double-stranded RNA- induced NF-κB activation, inducible nitric oxide synthase and Fas mRNA expression, and β-cell death (47,49) without affecting β-cell insulin content, glucose-induced insulin release (47), or the expression of genes involved in differentiated β-cell functions (8). In additional experiments, rat islets were pretreated with 100 μmol/l PDTC (Sigma, Belgium) for 1 h and then exposed for 6 or 24 h to IL-1β plus IFN-γ in the continuous presence of PDTC (same concentrations and culture conditions as above). We have previously shown that 100 μmol/l PDTC prevents cytokine-induced NF-κB activation in both human and rat pancreatic islets (50), a finding confirmed in the present series of experiments (data not shown).

RT-PCR was performed on poly(A)⁺ RNA as described (8). Primers for MCP-1 were 5′-ACCTGCTGCTACTCATTCAC-3′ and 5′-CTACAGAAGTGCTTGAGGTG-3′ (325 bp). The primers for glyceraldehyde 3-phosphate dehydrogenase were described previously (7). The ethidium bromide-stained agarose gels were photographed under ultraviolet transillumination using a Kodak Digital Science DC 120 camera (Eastman Kodak, Brussels, Belgium). Abundance of the PCR products were assessed by Biomax one-dimensional image analysis software (Kodak), and MCP-1 mRNA contents are expressed as optical densities corrected for GAPDH. Cytokines do not modify GAPDH expression in primary β-cells (8,51 and present data).

Results are given as means ± SE. Comparisons versus the respective control groups were performed using the Student’s paired t test.

To determine whether cytokines induce MCP-1 mRNA expression in INS-1E cells, as previously described for primary β-cells and RINm5F cells (4,5), these cells were exposed for 6 or 24 h to IL-1β, IFN-γ, or a combination of both cytokines, and MCP-1 mRNA content was measured by RT-PCR. There was no detectable MCP-1 expression in unstimulated cells or cells exposed to IFN-γ alone (Fig. 1), but MCP-1 expression was induced in cells exposed to IL-1β for 6 and 24 h, with higher value at 6 h. This increase was not potentiated by IFN-γ. These findings are similar to our previous observations indicating that IL-1β-induced MCP-1 expression in rat β-cells and human islets is not modified by IFN-γ (4,5). Thus, most of the subsequent experiments on MCP-1 promoter regulation were performed in the presence of IL-1β alone.

To determine the MCP-1 promoter regions responsive to IL-1β, the promoter fragment from nucleotides −514 to +53 relative to the transcription start site, with or without the enhancer fragment between −2,180 to −2,478 relative to the transcription start site (Fig. 2), were cloned into the luciferase reporter plasmid pGL3 and transfected into FACS-purified β-cells treated or not with IL-1β. In IL-1β-treated cells, the luciferase activity of the constructs pMCP-1-514(enh) luc was induced threefold, whereas pMCP-1-514 luc did not significantly respond to IL-1β (Fig. 3). This lack of response of the promoter region alone, even though there is a potential NF-κB binding site in this region, indicates a role for the potential NF-κB binding sites in the enhancer region (Fig. 2). Similar results were observed in INS-1E cells transfected with the same constructs. Thus, IL-1β induced a significant increase in the activity of pMCP-1-514[enh] but not of pMCP-1-514 luc (Fig. 4). Due to the large number of cells required, the subsequent experiments to elucidate the regulatory elements in the MCP-1 enhancer were performed in INS-1E cells.

The enhancer region contains two potential NF-κB binding sites (Fig. 2). Therefore, we mutated each NF-κB site separately and denoted the mutated constructs as pMCP-1-514[enh]mutA luc and pMCP-1-514[enh]mutB luc. The fold induction of MCP-1 promoter activity in response to IL-1β was reduced by 75 and 80% for mutA and mutB, respectively (Fig. 4). These results indicate that both binding sites are important for IL-1β-induced MCP-1 promoter activity.

To characterize the factors binding to the IL-1β-responsive elements in the enhancer region, we performed EMSAs using probes spanning the NF-κB sites from nucleotides −2,272 to −2,297 containing the first NF-κB motif of the enhancer (site A) and from −2,242 to −2,266 containing the second NF-κB motif of the enhancer (site B). Based on our previous observation that the maximal IL-1β-induced NF-κB binding activity in vitro in insulin-producing cells takes place after 30 min (42,52), this time point was selected for in vitro determination of nuclear factor binding. Two complexes, denoted by “a” and “b,” were induced in cells stimulated with IL-1β, with a higher affinity for the probe of site B (Figs. 5A and B). The nature of complexes a and b was further identified by competition and supershift experiments. The two complexes were specifically competed by an excess of unlabelled probe (Fig. 5A, lane 3, and Fig. 5B, lane 10) and by a consensus NF-κB oligonucleotide (Fig. 5A, lane 4, and Fig. 5B, lane 11). An antibody directed against the NF-κB family member p50 supershifted complex b (Fig. 5A, lane 5, and Fig. 5B, lane 12), whereas an antibody directed against the NF-κB family member p65 supershifted both complexes a and b (Fig. 5A, lane 6, and Fig. 5B, lane 13). An unrelated antibody against Stat1 had no effect on the migration of the two complexes (Fig. 5A, lane 7, and Fig. 5B, lane 14). These results suggest that complex a contains p65 homodimers, whereas complex b consists of p65 and p50 heterodimers.

To confirm in vivo our in vitro observations indicating NF-κB binding to the enhancer region, we performed ChIP assays with an antibody directed against the NF-κB p65 protein. ChIP assay allows for detailed analyses of protein-DNA interactions in a native chromatin environment (53), and the detection of association of rare transcription factor complexes can be achieved by real-time PCR quantification (45). We currently observed IL-1β-induced NF-κB binding in vivo by quantification of the rat MCP-1 enhancer region with real-time PCR (see research design and methods). NF-κB binding was detected as early as 30 min following IL-1β exposure, with continuously increasing values at 2 and 4 h (Fig. 6). The rat MCP-1 enhancer was undetectable in unstimulated control cells.

Overexpression of the IκB super-repressor by an adenovirus, AdIκB^(SA)2, with serine 32 and 36 substituted by alanine residues, specifically blocks NF-κB activation without interfering with other transcription factors (47,48). We currently observed that AdIκB^(SA)2 partially blocked cytokine-induced MCP-1 mRNA expression (Fig 7A). Thus, the optical density (OD) values of AdIκB^(SA)2 plus cytokines were 55 ± 9% of the values observed in noninfected cells exposed to the same cytokines (n = 3; P < 0.05), while infection with the control virus AdGFP did not modify the effects of cytokines (Fig. 7A; OD 86 ± 33% of the values observed in noninfected cells exposed to cytokines; n = 3). To further evaluate the role of NF-κB for cytokine-induced MCP-1 expression, rat islets were exposed to cytokines in the presence of the NF-κB blocker PDTC (50; Fig. 7B). There was no detectable MCP-1 expression in unstimulated cells or cells exposed to PDTC alone (Fig. 7B). MCP-1 mRNA was induced by cytokines after 6 and 24 h, and this induction was, to a major extent, prevented by PDTC (Fig. 7B). The results, expressed in OD after correction for GAPDH mRNA, for experiments 1 and 2, respectively, were as follows: cytokines-6 h, 1.96 and 1.61; PDTC plus cytokines-6 h, 0.54 and 0.47 (inhibition 72 and 71%, respectively); cytokines-24 h, 1.09 and 1.17; and PDTC plus cytokines-24 h, below detection limit (Fig. 7B).

MCP-1 expression is induced by cytokines in pancreatic β-cells and probably plays a role in attracting monocytes and T-cells to the site of insulitis (5,7,54). Indeed, the temporal pattern of MCP-1 expression, along with other CC chemokines, in the islets of NOD mice is well correlated with the progression of insulitis and β-cell destruction (5,54), and transgenic MCP-1 expression in β-cells induces mononuclear cell infiltration (28). In this context, it is of relevance to understand the molecular regulation of MCP-1 expression in the β-cells.

In the present study, we identified the transcription factor NF-κB as a key regulator of MCP-1 promoter activity. Mutation of either of the two NF-κB binding sites, located in the enhancer region at −2,242 and at −2,272, respectively, abrogated IL-1β-induced MCP-1 promoter activity. Moreover, we detected the binding of NF-κB p50 and p65 factors to these two sites by EMSAs and supershift experiments. In vivo binding of NF-κB to the enhancer region was confirmed by ChIP assays. Finally, overexpression of an IκB super-repressor, AdIκB^(SA)2, or preculture in the presence of the NF-κB blocker PDTC decreased cytokine-induced MCP-1 mRNA expression. The proximal promoter region by itself does not respond to cytokines, suggesting no role in insulin-producing cells for neither the putative NF-κB and AP1 binding sites (33), nor for the IRIS and GAS elements (34) located in this region.

Stereospecific alignment between adjacent cis elements has been observed in several genes, including those containing κB motifs (55). The two NF-κB binding sites in the enhancer region of MCP-1 promoter, site A and site B, are not identical in sequence, suggesting that they may serve distinct functions. Although higher affinity of NF-κB binding was observed for site B than A (Fig. 5), the fold induction of MCP-1 promoter activity in response to IL-1β was reduced by 75 and 80% for mutated site A and B, respectively (present data). The presence of these sites was also described as necessary to maintain TNF-induced MCP-1 gene promoter activity in mouse fibroblasts (30). On the other hand, insertion of a half or full helical turn between the κB sites, or site reversion between the two NF-κB binding sites in the enhancer region, did not affect TNF-induced MCP-1 gene promoter activity in these cells (30). Taken together with our findings, these results suggest that both binding sites for NF-κB are important for the IL-1β-induced MCP-1 promoter activity in insulin-producing cells.

Mammalian cells contain five NF-κB family members: p50, p52, p65, c-Rel, and RelB. Interactions between these proteins have been characterized in detail (56). P65/p50 heterodimers, the most frequent dimers, and p65 homodimers are activators of gene transcription, whereas p50 homodimers, which lack a transactivation domain, are transcriptionally inactive (57). Our supershift experiments with antibodies directed against p65 and p50 proteins suggest a role for p65/p50 and p65/p65 dimers in the regulation of MCP-1 in insulin-producing cells. The differential induction and binding of activated transcription factors to the promoter region of the MCP-1 gene provides a critical regulatory step, allowing expression of the chemokine in a cell- and stimulus-specific manner (15). Of note, we have observed the presence of the same active dimers of NF-κB in the context of cytokine regulation of Fas expression in primary β-cells (42).

In vitro determination of transcription factor binding to a promoter does not necessarily reflect the complex environment and the three-dimensional conformation of chromatin in a living cell. Indeed, it has been shown that the compact nature of chromatin can modulate accessibility to some DNA binding sites (45,53). Therefore, to confirm in vivo our observations of NF-κB binding to the rat MCP-1 promoter, we have performed ChIP assay, a novel method that allows detection of transcription factor binding in situ (53). We detected NF-κB binding in vivo at 30 min after IL-1β stimulation, with subsequent increases at 2 and 4 h (present data). Interestingly, it has been previously reported the presence of two distinct waves of early and late NF-κB recruitment to target promoters in a murine macrophage cell line, depending on the acetylation state of chromatin (58). NF-κB binding to the MCP-1 enhancer was detected only after 90 min of LPS exposure to macrophage cells (58). Our data suggest an early and sustained IL-1β-induced binding of NF-κB to the MCP-1 promoter in β-cells (Fig. 6), helping to explain the long-lasting effect of IL-1β on MCP-1 mRNA expression (Figs. 1 and 4,5,7,8).

The results described above suggest that NF-κB translocation to the nucleus is necessary for MCP-1 mRNA induction in β-cells exposed to cytokines. Indeed, rat β-cells infected with an adenovirus overexpressing the nondegradable IκBα super-repressor protein (AdIκB^[SA]2) have a 50% lower induction of MCP-1 mRNA expression in response to IL-1β. The partial blocking observed (Fig. 7A) may be due to the limited efficiency of infection by the adenoviral system used, which is ∼70–80% (47). These data are in good agreement with our previous microarray results, showing a 50–60% inhibition of cytokine-induced MCP-1 mRNA expression in β-cells previously infected with the AdIκB^(SA)2 (8) and with the present observations that PDTC, a potent inhibitor of NF-κB, blocked cytokine- induced MCP-1 mRNA expression >70% in cytokine-exposed islets. However, the fact that the observed inhibitory effect of NF-κB blockers on MCP-1 expression in vitro is not complete suggests that other transcription factors may also contribute to MCP-1 regulation in β-cells.

As a whole, these findings suggest an essential role for NF-κB in the regulation of MCP-1 in β-cells. Of note, in vivo overexpression of IκBα, a natural inhibitor of NF-κB, significantly decreases local MCP-1 and ICAM-1 expression, as well as recruitment of macrophages following iliac artery balloon angioplasty in rabbits, thus reducing arterial restenosis (59). This, and the fact that high basal MCP-1 production by human islets correlates with a poor clinical outcome following islet transplantation in patients with type 1 diabetes (6), suggests that adenovirus- or lentivirus-mediated NF-κB inhibition might be a useful approach to decrease mononuclear cell infiltration and graft dysfunction/destruction following islet transplantation.

This work was supported by grants from the Juvenile Diabetes Foundation International, the Funds voor Wetenschappelijk Onderzoek (FWO), and the Fonds National de la Recherche Scientifique (FNRS), Belgium.

We thank the personnel from the Diabetes Research Center, VUB, involved in β-cell purification; M.A. Neef, J. Schoonheydt, and M. Urbain, Laboratory of Experimental Medicine, ULB, for islet isolation; and E. Quartier for sequencing.