Induction of Secreted Type IIA Phospholipase A2 Gene Transcription by Interleukin-1β

Secreted type IIA phospholipase A2, which is involved in arachidonic acid release, is abundantly produced by chondrocytes and secreted in the synovial fluids of patients affected by rheumatoid arthritis. Transfection experiments showed that interleukin-1β stimulates the phospholipase A2 [−1614; +20] promoter activity by 6–7-fold and that the [−210; −176] fragment is critical for this stimulation. CAAT enhancer-binding protein (C/EBP) β and C/EBPδ transcription factors bind to this element as shown by bandshift experiments. Interleukin-1β increased the levels of C/EBPδ mRNA as soon as 2 h and up to 24 h whithout affecting those of C/EBPβ. Higher amounts of C/EBPδ proteins correlate with the stimulation of C/EBPδ mRNA. Mutations or 5′ deletions in the upstream [−247; −210] region reduced by 2-fold the basal and interleukin-1β-stimulated transcription activities. Two types of factors bind to overlapping sequences on this fragment: NF1-like proteins and the glucocorticoid receptor. The glucocorticoid receptor is responsible for a moderate stimulation of the promoter activity by dexamethasone and may interact with C/EBP factors to achieve a full transcription activity in basal conditions and in the presence of interleukin-1β. A [−114; −85] proximal regulatory element forms three complexes in bandshift experiments, the slowest mobility one involving the Sp1 zinc finger factor. Mutation of this sequence reduced to 2-fold the stimulation of the promoter activity by interleukin-1β or the C/EBP factors. Induction of the transcription of secreted type IIA phospholipase A2 gene by interleukin-1β in chondrocytes absolutely requires C/EBPβ and C/EBPδ factors but does not involve NF-κB.

The synovial fluid of patients suffering from rheumatoid arthritis or osteoarthritis contains large quantities of prostaglandin E2. Such quantities inhibit collagen synthesis and therefore contribute to joint destruction. These inflammatory lipid mediators are produced by a cascade of enzymes among which phospholipases A 2 (PLA 2 ) 1 play a key role by releasing arachidonic acid from membrane phospholipids. Two calciumdependent PLA 2 are involved in the release of arachidonic acid. Cytosolic PLA 2 , a ubiquitous 85-kDa enzyme, is activated by MAP kinases and translocated from the cytosol to membrane (1,2). Type IIA secreted PLA 2 (sPLA 2 -IIA) was originally purified from the synovial fluid of patients with rheumatoid arthritis, which contains high quantities of this enzyme (3). sPLA 2 -IIA belongs to a large group of 13-15-kDa secreted enzymes present in mammalian fluids and in the venoms of snakes and insects. According to a recent classification (4), the pancreatic version of the enzyme has been included in a type I group whereas the synovial PLA 2 is referred as type IIA sPLA 2 . Three other recently cloned mammalian PLA 2 were classified in IIC, V, and X groups, but their involvement in the arachidonic acid release remains unknown.
Purified or recombinant sPLA 2 -IIA triggers joint inflammation when it is injected intra-articularly in rabbits (5). The number of rheumatoid arthritis-affected joints and the presence of destructive erosion have been correlated with the amount of sPLA 2 -type IIA in the serum of patients (6). Mice with both the TNF␣ and sPLA 2 -IIA transgenes exhibit more joint destruction than do those with TNF␣ alone (7). Interleukin-1␤ (IL-1␤) is the most abundant cytokine in inflammatory synovial fluids. It stimulates the expression of numerous genes in articular cells (8) and increases the level of sPLA 2 -IIA mRNA in chondrocytes. We have previously demonstrated that prostaglandin E2 production by rabbit articular chondrocytes is related to plasma membraneassociated sPLA 2 -IIA activity and that the transcription rate of sPLA 2 -IIA gene is stimulated by IL-1␤ (9 -11).
Three main classes of transcriptional factors have been shown to mediate the effect of IL-1␤ on gene transcription: (i) NF-B is a dimer of p50 and p65 subunits and belongs to the Rel family. (ii) A member of the STAT family of transcription factors, which are activated through phosphorylation by Jak kinases and translocated to the cell nucleus within a few minutes, was characterized as an IL-1␤-stimulated factor but can also be activated by IL-6 and lipopolysaccharide (12). (iii) AP-1 can also be involved in the IL-1␤ pathway because IL-1␤ induces the transcription of the c-Jun and c-Fos genes in some cell models (13).
C/EBP transcription factors are involved in the regulation of gene transcription by IL-6. The C/EBP family includes three main members: C/EBP␣, C/EBP␤, and C/EPB␦. This last member is transcriptionally induced by IL-6 (14,15), whereas C/EBP␤ is mainly regulated at the post-transcriptional level by this cytokine in hepatoma cell lines (16,17), although a tran-scriptional regulation of C/EBP␤ by the inflammatory cytokines has also been described (for review see Ref. 18). In contrast to IL-6, the relationship between IL-1␤ and C/EBP factors has been poorly studied, and some positive regulations in interaction with NF-B have been reported (19,20). However, induction of C/EBP binding to DNA by pro-inflammatory cytokines correlates with the accumulation of prostaglandin E2, and both effects are reversed by anti-inflammatory cytokines (21). C/EBP factors act with NF-B to induce the transcription of many acute phase response genes in response to pro-inflammatory cytokines, and this effect is based on direct proteinprotein interactions (22,23). Similar interactions have also been reported between C/EBP proteins and the glucocorticoid receptor (GR), which may explain the co-induction of the transcription of acute phase response genes by glucocorticoids and cytokines (24). The GR is a member of the steroid/nuclear receptor superfamily and binds to the glucocorticoid-responsive element (GRE) on gene promoters (25).
We have previously shown that the activity of the sPLA 2 -IIA promoter is controlled by three regulatory elements in human hepatoma HepG2 cells (26) (see Fig. 1A). The [Ϫ210; Ϫ176] element C is critical for the stimulation of the promoter by IL-6 and binds C/EBP family members, whereas an adjacent [Ϫ247; Ϫ210] element D is recognized by several factors, some of which belonging to the NF1 family (27). The [Ϫ114; Ϫ85] element B is responsible for a high basal activity when the region upstream of the C/EBP-binding site is deleted (28). We have demonstrated that C/EBP factors can mediate the stimulation of transcription by IL-6 in HepG2 cells by suppressing the basal inhibition of transcription, a process that may involve single strand binding activities (27).
In this study, we have identified the sequences and transcription factors involved in the stimulation of the sPLA 2 -IIA promoter by IL-1␤ in chondrocytes. We have found that C/EBP␦ plays a critical role in this stimulation and is transcriptionally induced by IL-1␤ in chondrocytes. We have also shown that the glucocorticoid receptor and the NF1/CTF family members bind to overlapping sites in the previously identified regulatory D element. Moreover Sp1 and other unknown factors bind to the regulatory elements B. These last factors and the glucocorticoid receptor potentiate the transactivation of the sPLA 2 -IIA promoter by C/EBP␤ and C/EBP␦ to achieve full stimulation by IL-1␤.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes, T4 kinase, ligase, and Taq polymerase were purchased from New England Biolabs. Oligonucleotides were provided by Oligoexpress (Paris, France). Materials for cell culture and protein molecular weight markers were provided from Sigma (Dulbecco's supplemented modified Eagle medium, Ham's F-12 medium, fetal calf serum, HEPES, trypsin), Life Technologies, Inc. (Gey medium) and Costar (town, state) (flasks and Petri dishes). The ␤-galactosidase expression vector CMV-␤-gal was obtained from CLONTECH, and poly(dI-dC) and deoxynucleotides were from Amersham Pharmacia Biotech. Radioactive products were supplied by ICN or Amersham Pharmacia Biotech. Collagenase, hyaluronidase, and trypsin for the preparation of monolayers of rabbit chondrocytes were supplied by Roche Molecular Biochemicals. IL-1␤ was provided by Immungenex (Los Angeles, CA).
Plasmid Constructions and Chondrocyte Cultures and Transfections-The various CAT constructs containing wild type and mutant fragments of the sPLA 2 -IIA promoter have been described elsewhere (26,27). PHD expression vectors containing C/EBP␤ and C/EBP␦ were a gift from Dr. Ciliberto (Rome, Italy).
Three-week-old female Fauve de Bourgogne rabbits were killed and the shoulders, knees, and femoral heads were dissected out under sterile conditions as described by Jacques et al. (9). The articular cartilage was removed, cut into small pieces, and digested at 37°C with 0.05% hyaluronidase in Gey medium for 15 min and then with 0.25% trypsin for 30 min and finally with 0.2% collagenase for 90 min. The chondrocytes were then washed with Ham's F-12 medium for 60 min. The suspension of chondrocytes was seeded into 60 mm dishes (1.5 ϫ 10 5 cells per dish) in Ham's F-12 medium supplemented with 10% fetal calf serum. The cells were maintained at 37°C in 5% CO 2 , and the culture medium was changed every 2-3 sday. The cells reached preconfluency within 6 -7 days.
Chondrocytes were transfected using the calcium phosphate DNA co-precipitation method. Cells were changed by Dulbecco's modified Eagle's medium before transfection. Cells were incubated with the transfection mixture containing 12 g of pUC-SH-CAT constructs and 2.5 g of plasmids bearing the ␤-galactosidase gene for 4 h and then shocked with HBS buffer (21 mM HEPES, pH 7.1, 16 mM dextrose, 0.8 mM NA 2 HPO4, 5 mM KCl, and 137 mM NaCl) containing 15% glycerol for 90 s. The cells were incubated 20 h in Ham's F-12 supplemented with 0.2% bovine serum albumin and grown for an additional 24 h in the presence or absence of IL-1␤ (10 ng/ml). The harvested cells were lysed by incubation with 50 l of 100 mM Tris, pH 7.8, 0.7% Nonidet P-40 for 15 min at 4°C. CAT activities were measured by the two-liquid phases method as described by Fan et al. (27). ␤-Galactosidase activities were measured to normalize variations in transfection efficiency. Transfection experiments were performed in duplicate and repeated four times with two different preparations of plasmids.
Preparation of Nuclear Extracts and Cell Lysates-Chondrocytes nuclear extracts were prepared as described previously (27). Briefly, confluent cells from 3 P100 dishes were grown in Ham's F-12 without fetal calf serum and then incubated in the presence or absence of IL-1␤ (10 ng/ml) for 24 h in Ham's F-12 containing 10% fetal calf serum. They were then washed and scraped off into phosphate-buffered saline. The cells were centrifuged at 1500 ϫ g for 5 min, and the pellet was suspended in 500 l of buffer A (5 mM HEPES, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.5% Nonidet P-40, 0.5 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 50 mM NaF). The cells were incubated at 4°C for 15 min, centrifuged at 6000 ϫ g for 10 min, and the pellet was suspended in 100 l of buffer C (20 mM HEPES, pH 7.9, 25% glycerol, 0.5 M NaCl, 1.5 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 50 mM NaF). The nuclei were lysed by pipetting up and down four times and incubating for 30 min at 4°C. The lysates were centrifuged at 100,000 ϫ g for 30 min at 4°C in a TLC centrifuge (Beckman). The supernatants were collected, and the protein concentrations were measured according Olivier et al. (26). The nuclear protein batches were stored at Ϫ80°C. Lysates of COS-1 cells were prepared 40 h after transfection of the cells with the C/EBP expression vectors according to Olivier et al. (26).
Bandshift Assays-We used the double-stranded CWT and DWT as wild type probes, corresponding to the [Ϫ210; Ϫ176] sequence of the element C and to the [Ϫ247; Ϫ210] sequence of the element D. These double-stranded oligonucleotides (100 ng) were labeled using T4 kinase and 50 Ci of [␥-32 P]ATP. Free nucleotides were separated from the labeled probe on a Sephadex G50 column. The specific activity of the probe was estimated by spotting 1 l of the labeling mixture (before the G50 column) on to a TLC plate, separating the labeled probe and free nucleotide by chromatography, and counting them. The specific activities were 1-2 10 8 cpm/g. Chondrocyte nuclear extracts (6 -9 g) were incubated at 4°C for 15 min in 20 ml of solution containing 25 mM HEPES, pH 7.6, 8% Ficoll, 40 mM KCl, 5 mM MgCl 2 , 1 mM dithiothreitol, and 3 g of double-stranded poly(dI-dC). When double-stranded competitor oligonucleotides were used, they were added in a volume of 1 l, and the reaction mixture was incubated with the nuclear extracts at 4°C for 15 min before adding the probe. As competitors in bandshift assays, we used the Sp1 oligonucleotide 5Ј-GCAGTGTTTCTCCGC-CCCGATACGCGTAT-3Ј (29), the consensus glucocorticoid responsive element 5Ј-AGCTGCTCAGCTGGTACACTCCGTCCTCTACT (30), the HNF3 oligonucleotide 5Ј-GTTGACTAAGTCAATAATCAGA-3Ј corresponding to the HNF3-binding element of the transthyretin promoter (31), the NF1 oligonucleotide 5Ј-ACAATTTTTTGGCAAGAATATTAT-3Ј and the C/EBP oligonucleotide 5Ј-TGGTATGATTTTGTAATGGGG-TAGGA-3Ј, which correspond to the elements E and D, respectively, of the murine albumin promoter (32). The sequences of the other oligonucleotides used as competitors are indicated on the corresponding figures. Double-stranded oligonucleotide probes (60,000 cpm) were then added, and the incubation was continued for 30 min at 4°C. Free DNA and DNA-protein complexes were resolved by electrophoresis in 5% or 7% polyacrylamide gels in 6.7 mM Tris-HCl, 3.3 mM sodium acetate, 1 mM EDTA, pH 7.9. The gels were dried and used to expose X-OMAT® films (Eastman Kodak Co.). In supershift experiments, chondrocyte nuclear extracts were preincubated for 15 min at 4°C with 1 l of antibodies raised against Sp1, C/EBP␣, C/EBP␤, or C/EBP␦ (Santa Cruz Biotechnology Inc., Santa Cruz, CA).
Northern Blotting-Total RNA was isolated from cultured chondrocytes using guanidinium isothiocyanate. The total RNA content was measured by spectrophotometry, and its integrity was assessed by agarose gel electrophoresis; 15 g of total RNA/lane were separated on 1% agarose/2.2 M formaldehyde gels and transferred to nylon filters (Hybond N; Amersham Pharmacia Biotech). The membranes were prehybridized for 15 min, and then hybridized at 65°C for 2 h with the various probes in the rapid hyb-buffer medium (Amersham Pharmacia Biotech). The specific C/EBP␤ and C/EBP␦ probes were obtained by digestion of the corresponding expression vectors with PstI and XhoI (New England Biolabs, Boston, MA) respectively. The digestion products were separated on 1% agarose gels, the 500-and 850-bp-length bands corresponding to fragments of the C/EBP␦ and C/EBP␤ cDNA, respectively, were sliced and extracted using the Gene-Clean kit (Bio101, La Jolla, CA). The C/EBP probes were labeled using the random-primed labeling system (Amersham Pharmacia Biotech) and [␣-32 P]dCTP (3000 Ci/mmol). An oligonucleotide hybridizing to the 28 S RNA was labeled with [␥-32 P]ATP by T4 kinase and used as probe to take into account the variations in loaded and transferred RNA. The hybridized filters were washed twice in 2ϫ SSC (150 mM NaCl, 17 mM trisodium citrate) 0.1% SDS at room temperature for 15 min and then twice in 0.1ϫ SSC, 0.1% SDS for 15 min at room temperature and at 50°C. Autoradiography was performed for 10 days for C/EBP probes and 24 h for the 28 S RNA probe. The blots were successively hybridized with the C/EBP␦, C/EBP␤, and 28 S RNA probes. The filters were washed in 0.1ϫ SSC, 0.1% SDS at 85°C for 10 min before rehybridization.
Western Blotting-Aliquots (50 g) of nuclear proteins were separated on a 12% SDS-polyacrylamide gel electrophoresis in 0.38 M Tris-HCl, 0.1% SDS, pH 8.8, and electroblotted on to Protran BA83 nitrocellulose membranes (Schleicher & Schull). The membranes were saturated in 10 mM Tris, pH 7.5, 100 mM NaCl, 0.1% Tween 20, 5% nonfat milk at 4°C overnight, hybridized with anti-C/EBP␤ or anti-C/ EBP␦ antibodies (Santa Cruz) for 1 h at 4°C, washed in the saturation buffer, and then developed using the ECL system (Amersham Pharmacia Biotech). The abundance of C/EBP␦ protein in nuclear extracts from untreated and IL-1␤-treated chondrocytes was calculated by quantitative scanning of autoradiograms using a CCD video camera and the Densylab system (Quantum Bioprobe, Montreuil, France).

RESULTS
The Regulatory Element C of the sPLA 2 -IIA Promoter Is Critical for the Stimulation of Its Activity by IL-1␤ and Binds C/EBP␤ and C/EBP␦-Rabbit primary culture chondrocytes were transfected with CAT constructs containing various 5Ј deleted fragments of the sPLA 2 -IIA promoter. A 24-h treatment of the cells by IL-1␤ increased by 6.6 Ϯ 2 the activity of the [Ϫ1614; ϩ20] fragment of the sPLA 2 -IIA promoter (Fig. 1B). Deletion of the [Ϫ247; Ϫ225] fragment decreased by 2-fold the basal and IL-1␤-stimulated transcription activities. The relative stimulation of the transcription activities by IL-1␤ was not significantly modified by the 5Ј deletions from the Ϫ225 to the Ϫ203 positions because the basal and stimulated activities were similarly decreased. Deletion downstream from the position Ϫ195 completely suppressed the stimulation of the transcription activity by IL-1␤. Deletion of the Ϫ159/Ϫ138 fragment produced an additional 2-fold reduction of the transcription activity, and the promoter activity was suppressed by a further deletion to the Ϫ87 position (Fig. 1B).
Electrophoretic mobility shift assays showed that chondrocyte nuclear extracts formed two complexes, C1 and C2, with the CWT probe, which corresponds to the [Ϫ210; Ϫ176] ( Fig.  2B, lanes 1 and 7). The slowest electrophoretic mobility complex C1 could be clearly observed only when the chondrocytes were treated with IL-1␤ prior the extraction of the nuclear proteins (Fig. 2B, compare lanes 1 and 7). The C2 complex was formed when the extracts of both the untreated and IL-1␤stimulated chondrocytes were used, but its intensity was higher when the cells were treated by IL-1␤ (Fig. 2B, compare  lanes 1 and 7). Because this region was previously shown to correspond to a C/EBP-binding site in hepatocytes (26), we used as competitors the C/EBP-binding oligonucleotide corresponding to the D element of the murine albumin promoter (32) and the Cmut oligonucleotide in which the Ϫ199/Ϫ197 5Ј-TTT-3Ј triplet was mutated into an 5Ј-GCC-3Ј sequence ( Fig.  2A). All the complexes were suppressed when the chondrocyte nuclear extracts were preincubated with the C/EBP-binding oligonucleotide prior to the addition of the CWT probe (Fig. 2B,  lanes 2 and 8). By contrast, none of the complexes were com- The regulatory elements were previously identified by footprint, bandshift, and transfection assays (26 -28). The P60 dishes of 60% confluent chondrocytes were transfected by the calcium phosphate co-precipitation method using 12 g of the various CAT constructs and 2 g of the CMV-␤-gal expression vector. 24 h after transfection, the cells were incubated with or without IL-1␤ (10 ng/ml) for a further 24 h and then harvested for the measurement of CAT and ␤-galactosidase activity. The activities of untreated (white bars) or IL-1␤-stimulated (solid bars) chondrocytes were calculated relative to that induced by the [Ϫ326; ϩ20] fragment in untreated cells. Results are expressed as the means Ϯ S.E. of four independent transfections performed in duplicate.
peted out by the Cmut oligonucleotide (Fig. 2B, lanes 3 and 9). An antibody raised against C/EBP␣ did not alter any of the complexes formed between the CWT probe and the chondrocyte nuclear extracts (Fig. 2B, lanes 4 and 10). Because all the C/EBP family members share the same binding site, it is likely that rabbit articular chondrocytes lack C/EBP␣ expression. Antibodies to C/EBP␤ and C/EBP␦ supershifted the C1 complex (Fig. 2B, lanes 5, 6, 11, and 12). The upper part of the C2 complex was displaced by the antibody to C/EBP␤ complex (Fig. 2B, lanes 5 and 11), whereas its lower part was shifted by the antibody to C/EBP␦ (Fig. 2B, lanes 6 and 12). Because C/EBP␤ and C/EBP␦ bind to DNA by forming heterodimers as well as homodimers, these data indicate that the C1 complex is formed by C/EBP␤-C/EBP␦ heterodimers, whereas the C2 complex is heterogeneous and corresponds to co-migrating C/EBP␤-C/EBP␤ and C/EBP␦-C/EBP␦ homodimers.
The mutations of the whole [Ϫ204; Ϫ181] sequence or the Ϫ199/Ϫ197 triplet abolished the stimulation of the promoter activity by IL-1␤ (Fig. 3). The 6-fold stimulation of the wild type [Ϫ326; ϩ20] sPLA 2 -IIA promoter activity by IL-1␤ was mimicked by co-transfecting the C/EBP␤ and C/EBP␦ expression vectors with the [Ϫ326; ϩ20]-pUC-SH-CAT construct in the absence of cytokine treatment (Fig. 3). By contrast co-transfection had no effect with the mutated CAT constructs for the element C. Treatment of the cells with IL-1␤, after co-transfection of the C/EBP expression vectors with the wild type [Ϫ326; ϩ20]-pUC-SH-CAT construct, did not increase the stimulation of the [Ϫ326; ϩ20] promoter activity by the overexpressed C/EBP␤ and C/EBP␦ factors (Fig. 3), suggesting a pivotal role for the C/EBP factors in the induction of the sPLA 2 -IIA promoter activity by IL-1␤.
IL-1␤ Stimulates the Expression of the C/EBP␦ Gene-Messenger RNA levels of C/EBP␤ and C/EBP␦ were measured in chondrocytes at various times after treatment by IL-1␤. The C/EBP␦ mRNAs were barely detectable by Northern blot in absence of IL-1␤. The C/EBP␦ mRNA levels were increased by IL-1␤ after a 2-h treatment, reached maximal values between 14 -16 h of treatment, and were maintained over a 24-h period. By contrast IL-1␤ did not affect the levels of C/EBP␤ mRNAs (Fig. 4). Western blot experiments confirmed that IL-1␤ increased by 3-4-fold the concentration of C/EBP␦ proteins (Fig. 5).
The Ϫ224] 5Ј-TGTTTT-3Ј sequence, which is homologous to the consensus 5Ј half-site for the glucocorticoid receptor 5Ј-TGTTCT-3Ј (33) (Fig. 6A). The DWT probe (Fig. 6A) formed, with nuclear extracts from untreated chondrocytes, two specific complexes, D1 and D2 (Fig. 6B, lanes 1 and 2). The formation of complex D1 was suppressed by the addition of a 250-fold excess of a consensus GRE (Fig. 6B, lane 4), which suggests that this complex correspond to the binding of the glucocorticoid receptor. Complex D2 was competed out by the same excess of a NF1-binding oligonucleotide, which indicates the involvement of CTF/NF1 family members in the formation of this complex (Fig. 6B, lane 3).
To further delineate the binding sites of the NF1 proteins and the glucocorticoid receptor, four oligonucleotides corresponding to a mutation of the element D (Fig. 6A) were used in bandshift experiments. The Dm1 oligonucleotide, in which the [Ϫ224; Ϫ220] CTF/NF1 hemi-site was mutated (Fig. 6A), competed out the D1 complex but did not affect the formation of the D2 one (Fig. 6B, lane 5), confirming that this latter involves CTF/NF1 family members. The Dm2 oligonucleotide which contains two mutations, respectively, on the Ϫ228 G nucleotide and on the Ϫ238/Ϫ236 sequence (Fig. 6A), competed out the D1 complex but not the D2 one (Fig. 6B, lane 6). The involvement of the [Ϫ229; Ϫ224] 5Ј-TGTTTT-3Ј sequence in the formation of the D1 complex was confirmed by the lack of competition of this  1-6) or IL-1␤-stimulated (lanes 7-12) chondrocytes. Treatment with 10 ng/ml IL-1␤ was performed for 24 h prior to the extraction of nuclear proteins. Incubation of the CWT probe was carried out for 30 min at 4°C in a mixture containing 6 g of rabbit chondrocyte nuclear extracts and 3 g of poly(dI-dC) in 16 mM HEPES, pH 7.6, 5% Ficoll, 40 mM KCl, 5 mM MgCl 2 , 1 mM dithiothreitol, and 0.1 mM EDTA. 250-fold excesses of C/EBP and Cmut oligonucleotides over the CWT concentration were incubated with the nuclear extracts for 15 min before the addition of the probe (lanes 2, 3, 8, and 9). Antibodies specifically raised against C/EBP␣ (lanes 4 and 10) C/EBP␤ (lanes 5 and 11) or C/EBP␦ (lanes 6 and 12) were incubated with chondrocyte nuclear extracts for 15 min at 4°C prior to the addition of poly(dI-dC). NS indicates a nonspecific band, and the asterisk represents a variable complex. C1 and C2 complexes are indicated by the arrows. Electrophoresis was run on a 5% 30:1 bisacrylamide/acrylamide gel. complex by the Dm3 oligonucleotide (Fig. 6B, lane 7) in which the Ϫ229/Ϫ228 TG nucleotides were mutated into CA (Fig. 6A). In addition the 250-fold excess of Dm3 oligonucleotide did not fully displaced the D2 complex, indicating that the sequence upstream from the [Ϫ224; Ϫ220] CTF/NF1 hemi-site plays a role in the binding of the CTF/NF1 family members to the element D (Fig. 6B, lane 7). Finally the mutation of the whole [Ϫ231; Ϫ224] 5Ј-TGTGTTTT-3Ј sequence abolished the ability of the Dm4 oligonucleotide to compete with the DWT probe for the formation of both the D1 and D2 complexes (Fig. 6B,  lane 8). The Dm1 oligonucleotide was also used as probe and formed the D1 complex but not the D2 one (Fig. 6C, lane 3). The GRE probe formed two complexes with the nuclear extracts from untreated and IL-1␤-stimulated chondrocytes. The upper complex, which was the most intense, co-migrated with the D1 complex formed with the DM1 probe (Fig. 6C, compare  lanes 1 and 2 with lane 3). This result supports the involvement of the glucocorticoid receptor in the formation of the D1 complex and also indicates that this receptor is present in nuclei in absence of dexamethasone treatment. The Dm3 oligonucleotide used as probe did not form either the D1 or the D2 complex, confirming the involvement of the sequence upstream the [Ϫ224; Ϫ220] CTF/NF1 hemi-site in the binding of the CTF/NF1 family members to the D element (Fig.  6C, lane 4). Treatment of chondrocytes with IL-1␤ for 24 h did not affect the formation or the intensity of the complex formed with the GRE probe (Fig. 6C, lane 2).
A 24-h treatment of transfected chondrocytes by dexamethasone (10 Ϫ7 M) induced a moderate stimulation of the [Ϫ326; ϩ20] sPLA 2 -IIA promoter activity by 160 Ϯ 12%, and this stimulation was suppressed by co-incubation with the synthetic glucocorticoid analog and inhibitor RU486 (10 Ϫ5 M) (Fig. 7A). Co-incubation with dexamethasone (10 Ϫ7 M) had no effect on the stimulation of the [Ϫ326; ϩ20] sPLA 2 -IIA promoter activity by IL-1␤ (Fig. 7A). Similarly, RU486 did not modify the induction of transcription activity by IL-1␤ (Fig. 7A). Dexamethasone elicited a 5-fold induction of a promoter containing a GRE sequence upstream of the thymidine kinase promoter, and this induction was inhibited by the anti-glucocorticoid RU486. The dexamethasonestimulated transcription was equivalent in the case of the GRE-TK and [Ϫ326/ϩ20] CAT construct, whereas the basal activities of these two plasmids were different.
Stimulation of the sPLA 2 -IIA transcription by dexamethasone in transient transfection experiments of chondrocytes was suppressed by the deletion of the region upstream the Ϫ225 position (Fig. 7B), which altered the GRE (Fig. 6A). This last 5Ј deletion also decreased the basal transcription activity as previously shown in Fig. 1B. The substitution of the [Ϫ231; Ϫ224] 5Ј-TGTGTTTT-3Ј fragment overlapping the NF1 and GR hemisites by the nonspecific sequence 5Ј-GGTACCCG-3Ј drastically reduced the basal transcription activity of the resulting ⌺[Ϫ231; Ϫ224]-pUC-SH-CAT construct to 18 Ϯ 9% of that of the wild type promoter and suppressed its stimulation by IL-1␤ and dexamethasone (Fig. 7B). The mutation of the 5Ј-TG-GCA-3Ј CTF/NF1 hemi-site in the ⌺[Ϫ223; Ϫ218]-pUC-SH-CAT construct did not affect either the basal transcription activity or its stimulation by IL-1␤ but abolished that by dexamethasone (Fig. 7B). Taken together, these results indicate that the stimulation of the sPLA 2 -IIA transcription by IL-1␤ and dexamethasone are not mediated in the same way, although both involve the GR hemi-site on the element D. The GRE located on the element D is required to achieve full basal and IL-1␤-stimulated transcription activities.
The Ϫ99] 5Ј GACCACGCC-3Ј sequence is critical for the sPLA 2 -IIA promoter activity in HepG2 cells (28). The BWT probe, which corresponds to the [Ϫ114; Ϫ87] sequence of the sPLA 2 -IIA promoter (Fig. 8A), formed three complexes with the chondrocyte nuclear proteins (Fig. 8B, lane 1). A 500-fold excess of the unlabeled BMut oligonucleotide, in which the [Ϫ107; Ϫ99] sequence was mutated (Fig. 8A), did not displace any of the complexes B1, B1Ј, or B2 (Fig. 8B, lane 3). The complex B1 was supershifted by a specific antibody to Sp1 (Fig. 8B, lane 5), indicating the involvement of Sp1 in the formation of this complex. The pretreatment of the chondrocytes by IL-1␤ did not change the mobility or the intensity of any of the three complexes formed with the BWT probe (Fig. 8B, lane 6).
Transfection experiments showed that the mutation of the native [Ϫ107; Ϫ99] sequence in the ⌺[Ϫ107; Ϫ99]-pUC-SH-CAT plasmid reduced the stimulation of the [Ϫ326; ϩ20] promoter activity by IL-1␤ 2-fold (Fig. 9). This mutation had a moderate effect on the basal activity of the sPLA 2 -IIA promoter  4. IL-1␤ induces expression of the C/EBP␦ gene but does not affect that of C/EBP␤. Confluent rabbit primary culture chondrocytes in P100 dishes were incubated with 10 ng/ml IL-1␤ for 2 to 24 h. In control experiments (time 0), the cells were cultivated for an additional 24 h without the cytokine. After extraction, 10 g of total RNA were used for Northern blot experiments, as indicated under "Experimental Procedures." Hybridization to the 28 S ribosomal fraction was used to normalize any variations in mRNA quantity and integrity. activity, i.e. a decrease by 30% was observed (Fig. 9). When the expression vectors of C/EBP␤ was co-transfected with the constructs containing the wild type and the mutant promoters, respectively, the stimulation of the transcription activity dropped from 7-8-fold to 2.5-3.5-fold. By contrast, overexpressed C/EBP␦ stimulated the activity of the wild type promoter 7-8-fold and that of the mutant promoter 5-fold (Fig. 9). These results show that the factors bound to the regulatory element B are involved in the stimulation of the sPLA 2 -IIA promoter activity by IL-1␤ and reinforce the transactivation by the C/EBP factors. DISCUSSION We present evidence that C/EBP factors are central to the control of sPLA 2 -IIA gene transcription by IL-1␤. The major complex C2, formed by C/EBP␤ and C/EBP␦ homodimers, was observed with extracts from both untreated and IL-1␤-treated chondrocytes, whereas a second complex C1 was formed with extracts from IL-1␤-treated cells. The amounts of C/EBP␦ mRNA and protein in rabbit chondrocytes are increased by IL-1␤, which does not, in contrast, modify the amounts of C/EBP␤ mRNA. The newly synthesized C/EBP␦ protein can form higher amounts of homodimers and heterodimers with C/EBP␤, and this could explain the fact that the intensities of both the C1 and C2 complexes were reinforced when the nuclear proteins were extracted from IL-1␤-treated chondrocytes. The stimulation of sPLA 2 -IIA transcription by IL-1␤ is therefore mediated through the increase of the C/EBP␦ gene expression itself. The increase of C/EBP␦ mRNA levels after 2 h is FIG . 5. IL-1␤ increases the quantities of C/EBP␦ protein in primary culture chondrocyte. A, confluent rabbit primary culture chondrocytes in P100 dishes were incubated with 10 ng/ml IL-1␤ or the carrier (phosphate saline buffer) for 24 h. 50 g of nuclear proteins were separated by electrophoresis and transferred on to a nitrocellulose membrane as indicated under "Experimental Procedures." The membrane was incubated with antibodies raised against C/EBP␦. A lysate (5 g proteins) of COS-I cells expressing C/EBP␦ was used as a control of migration and hybridization. NS indicates a nonspecific band cross-hybridizing with the antibodies against C/EBP␦. The intensity of this band was measured and taken as reference. MW, molecular mass. B, densitometric analysis of the intensities of the nonspecific band (NS) and the band corresponding to C/EBP␦ in the absence or in presence of IL-1␤. consistent with the delayed increase of the sPLA2 mRNA levels, which we previously measured in rabbit primary culture chondrocytes (9).
Although the element C is the key element for the regulation Ϫ99] sequence. This suggests that C/EBP␤ and C/EBP␦ may interact differently with the general transcription machinery in terms of their affinity for the various co-activators such as CBP/P300 (34) or Nopp140 (35). We have shown that the 5Ј-TGTTTT-3Ј sequence of element D is critical for the formation of a slow electrophoretic mobility complex D1 and for full sPLA 2 -IIA basal and dexamethasoneand IL-1␤-stimulated transcription activity. The D1 complex is competed out by a consensus GRE and co-migrates with the main complex formed between the GRE probe and the chondrocyte nuclear extracts. The glucocorticoid receptor typically binds to two hexameric inverted repeats separated by 3 bp. The [Ϫ229; Ϫ224] 5Ј-TGTTTT-3Ј sequence of element D is highly homologous to the consensus 3Ј hemi-site 5Ј-TGTTCT-3Ј for GREs. The G and T nucleotides on positions 2 and 3, which interact with Arg-447 and Val-443 residues of GR, are conserved, and we have observed in a previous study that the Ϫ228 G nucleotide is a major site of interference with a liver protein forming an intense complex with a mobility similar to that of the D1 complex (26). However, the C nucleotide on position 5, which interacts with the Lys-442 residue of GR in consensus GRE (36) was replaced by a T. The element D lacks a typical 5Ј half-site, but the [Ϫ241; Ϫ236] 5Ј-CTGCCT-3Ј sequence, which is 6 bp upstream of the 5Ј-TGTTTT-3Ј site, is weakly homologous to the consensus sequence (36). Furthermore, variations of orientation and spacing between the two hemi-sites up to 8 -9 bp have been described (37,38), and many GRE lack 5Ј hemi-sites, such as those of mouse IL-2 receptor ␣, human elastin, and rat chromogranin promoters (39 -41).
We have shown that both the GRE and DWT probes form complexes with nuclear extracts from untreated chondrocytes. This indicates that the glucocorticoid receptor was present in chondrocyte nuclei in the absence of treatment by dexamethasone or IL-1␤. Several studies have reported the localization of unliganded glucocorticoid receptors inside the nuclei of several cell types (42)(43)(44)(45). More recently, two types of glucocorticoid receptor have been described in humans, hGR␣ and hGR␤, which differ at their C-terminal end beyond amino acid 727; hGR␣, which has 50 additional residues; and hGR␤, which contains 15 nonhomologous residues that are generated through alternative splicing of the last exon (46). hGR␣ is sequestered in the cytosol by heat shock protein hsp90 in basal conditions, binds glucocorticoids or RU486, and is translocated into the nuclei; hGR␤ in the nuclei does not bind dexamethasone or RU486 but can bind to GRE (46,47). This factor has been described as a putative repressor of hGR␣ (47)(48)(49), but this hypothesis is a subject of debate (50). Furthermore RU486 does not impede the binding of GR to DNA (51, 52) but inhibits the transactivation ability of GR in the presence of dexametha-sone. The glucocorticoid receptor(s) expressed in chondrocytes have been poorly studied, and their regulation remains to be examined in further detail in this cell model. Dexamethasone decreases the level and activity of sPLA 2 -IIA mRNA in vascular smooth muscle cells (53), but conflicting data have been reported in mesangial cells (54,55). Furthermore these data have been obtained in rats, and the rat promoter does not contain any sequence homologous to a GRE in the region corresponding to the human element D. The only study in humans was performed in the hepatoma HepG2 cell line by Haselmann and Goppelt-Struebe (56), who described a partial decrease of oncostatin M-stimulated expression in the presence of dexamethasone. Modulation of sPLA 2 -IIA gene by glucocorticoids are therefore cell-and species-specific and may act at different levels in opposite ways. Regarding the regulation of the sPLA 2 -IIA promoter, we assume that unliganded GRs bind to the 5Ј-TGTTTT-3Ј sequence and increase the basal transcription activity without affecting the relative stimulation by IL-1␤, as shown by deletions through the element D. The suppression of the IL-1␤-induced stimulation of the transcription activity, which was observed by using the [Ϫ326; ϩ20] promoter mutated on the [Ϫ231; Ϫ224] 5Ј-TGTTTT-3Ј sequence, might be due to the activity of inhibitory factors such as those described in HepG2 cells (27). The binding of GR to the element D also explains the stimulation of the promoter activity by dexamethasone, which is abolished by mutation of the 5Ј-TGTTTT-3Ј sequence; the low level of homology with the consensus GRE and the absence of repeated GRE may explain the low level of stimulation.
Mutation of the [Ϫ224; Ϫ220] 5Ј-TGGCA-3Ј CTF/NF1 hemisite suppressed the stimulation of sPLA 2 -IIA transcription activity by dexamethasone without affecting its basal and IL-1␤stimulated levels. The CTF/NF1 family members are encoded by four genes, NFI-A, NFI-B, NFI-C, and NFI-X, and their diversity is increased by alternative splicing or cleavage of larger polypeptides (57)(58)(59)(60). The products of the four genes have different transactivation abilities and generate heterodimers with intermediate activation potentials (61). Stimulations of the human papillomavirus type 16 and aspartate aminotransferase promoters by glucocorticoid require the presence of intact NF1-binding sites (62,63). The glucocorticoid receptor favors binding of CTF/NF1 factors to the murine mammary tumor virus promoter in vivo by stimulating the nucleosome disrupting activity of the SWI/SNF complex (64). In the context of the sPLA 2 -IIA promoter, its main action would be to allow the C/EBP factors to achieve full transcription activities. Such effects are conditioned by interactions between the GR bound to the element D and the C/EBP factors bound to the element C. It is interesting to note that the GR recruits C/EBP␤ to the rat ␣1 acid glycoprotein promoter (65) and that the downstream GRE is separated from the upstream C/EBP-binding sites by 23 bp in this promoter (24), the same distance as that between the GR and C/EBP-binding sites in the sPLA 2 -IIA promoter. Boruk et al. (66) demonstrated that the GR interacts with C/EBP␤ through their AF2 domain and an intermediary factor. Chang et al. (67) have suggested that this intermediary factor is, in the context of the ␣1 acid glycoprotein promoter, the co-activator TIF1␤, which belongs to the RING protein family. The identification of the factor(s) ensuring the interaction between the GR and the C/EBP proteins bound to sPLA 2 -IIA promoter in chondrocytes requires further studies. The hypothesis of the putative regulatory mechanisms of the various factors that bind to human sPLA 2 -IIA promoter are summarized in Fig. 10.
Increased synthesis of prostaglandin E2 by chondrocytes and synoviocytes in response to IL-1␤ is a key event of the inflammatory process in joints. Prostaglandin E2 is produced through the functional interaction between cyclooxygenase-2 and sPLA 2 in cell types that express both of these genes (68,69). We have shown in this study that C/EBP␦ and C/EBP␤ play a critical role in the regulation of the secreted type IIA sPLA 2 promoter activity by IL-1␤. We and others have demonstrated that these factors are also essential for the regulation of cyclooxygenase-2 transcription. C/EBP␤ and C/EBP␦ may therefore be considered as putative targets of therapeutic strategies to inhibit icosanoid synthesis in chondrocytes.