CCAAT/Enhancer-binding Proteins β and δ Mediate the Repression of Gene Transcription of Cartilage-derived Retinoic Acid-sensitive Protein Induced by Interleukin-1β*

Cartilage-derived retinoic acid-sensitive protein (CD-RAP) is a secreted protein expressed by chondrocytes; the expression is repressed by interleukin 1β (IL-1β). To investigate the transcriptional mechanism, by which CD-RAP expression is suppressed by IL-1β, deletion constructs of the mouse CD-RAP promoter were transfected into rat chondrocytes treated with or without IL-1β. The results revealed an IL-1β-responsive element located between −2138 and −2068 bp. As this element contains a CAAT/enhancer-binding protein (C/EBP) motif, the function of C/EBPβ and C/EBPδ was examined. IL-1β stimulated the expression of C/EBPβ and -δ, and the direct binding of C/EBPβ to the C/EBP motif was confirmed. The −2251-bp CD-RAP promoter activity was down-regulated by co-transfection with C/EBP expression vectors. Mutation of the C/EBP motif abolished the inhibitory response to IL-1β. Additionally, C/EBP expression vectors were found to down-regulate the construct containing the promoter and enhancer of the type II collagen gene. Finally, the enhancer factor, Sox9, was shown to bind adjacent to the C/EBP site competing with C/EBP binding. Taken together, these results suggest that C/EBPβ and -δ may play an important role in the IL-1β-induced repression of cartilage-specific proteins and that expression of matrix proteins will be influenced by the availability of positive and negativetrans-acting factors.

inhibitory activity (MIA), are also expressed in various tumor tissues, such as chondrosarcoma, melanoma, and breast cancer (3)(4)(5). CD-RAP was originally cloned as a protein expressed in cartilage and down-regulated in chondrocytes that had been de-differentiated by treatment with retinoic acid (6). The protein structure of CD-RAP/MIA showed that it contains an Src homology 3 domain, which is unique for an extracellular protein. CD-RAP/MIA can interact with the type III repeat of fibronectin suggesting a model whereby CD-RAP/MIA binds to fibronectin and interferes with integrin binding (7). A recent study of CD-RAP-deficient mice showed that the mice have normal expression and distribution of type II collagen, aggrecan, and type X collagen but have increased density, increased diameter, and irregular arrangement of collagen fibrils (8). Thus, CD-RAP may function to organize highly ordered ultrastructural fiber architecture. CD-RAP is regulated by various cytokines or growth factors; insulin and insulin-like growth factor I stimulate and basic fibroblast growth factor and interleukin 1␤ (IL-1␤) suppress CD-RAP expression in bovine articular chondrocytes (9).
IL-1␤ is one of the major cytokines that mediate inflammatory reactions. IL-1␤ promotes the arachidonic acid cascade resulting in production of prostaglandin (10). In joint diseases, such as rheumatoid arthritis and potentially osteoarthritis, IL-1␤ is thought to contribute to degradation of matrix proteins not only by the production of proteases such as matrix metalloproteinases but also by down-regulation of expression of the matrix proteins such as types II, IX, and XI collagen and aggrecan, resulting in the loss of the cartilage structure and prevention of repair (10 -13). Because CD-RAP is thought to be required for formation of the highly ordered ultrastructural fiber architecture (8), down-regulation of CD-RAP gene may contribute further to loss of cartilage integrity.
Here, we investigated the factors responsible for down-regulation of the CD-RAP gene by IL-1␤, and we show that C/EBP plays a key role in this regulation. Furthermore, we show that C/EBP can down-regulate type II collagen gene transcription, thus suggesting that C/EBP might play an important role in down-regulation of various matrix proteins induced by IL-1␤. We provide the first demonstration that C/EBP down-regulates the expression of a cartilage-specific matrix-associated gene and that it is a novel pathway mediating repressive effects of IL-1␤. Plasmid Constructs-The CD-RAP promoter 5Ј-deletion constructs were made by PCR and subcloned into pGL3-basic vector, as described (34). Mutagenesis of the C/EBP-binding site in the Ϫ2251-bp promoter was performed by PCR using the Site-directed Mutagenesis Kit TM according to the manufacturer's instructions. Human Sox9 expression vector was described before (35). Human Col2A1 promoter construct and C/EBP expression vectors were the kind gifts provided by the following: 4-kb promoter and enhancer construct spanning Ϫ577 to ϩ3426 bp of human COL2A1 in the pGL2-basic vector was obtained from Dr. Mary B. Goldring (36,37); human C/EBP-full-length in the pCDNA3 vector was from Dr. Erika Crouch (38); the rat pCI-neo-LIP was from Dr. Cynthia A. Zahnow (39); the rat pCMV-LAP was from Dr. Mina Bissell (40); and the rat pMSV-C/EBP␦ was from Dr. Alan Friedman (41). The empty expression vectors were made by excision of cDNAs from the corresponding C/EBP expression vectors. All the plasmids were confirmed by sequencing or digestion with restriction enzymes.

Materials
Cell Cultures-Rat chondrogenic cell line, RCJ 3.1C5.18, was maintained in ␣MEM supplemented with 10% FBS, 10 nM dexamethasone, and 2% penicillin/streptomycin, as described before (34). Rat chondrosarcoma (RCS) cells were cultured in DMEM with 10% FBS and 2% penicillin/streptomycin (34). For RNA or nuclear protein extraction, RCS cells were plated in 2 ϫ 10 4 /cm 2 densities and cultured overnight. Human IL-1␤ was then added to the medium at the concentrations indicated. Primary chondrocyte culture was obtained from human articular cartilage. Full thickness slices of human cartilage were dissected from the normal knee joints of donors under aseptic conditions and subjected to sequential Pronase and collagenase P digestion to liberate chondrocytes from tissues. Isolate chondrocytes were plated at a density of 2 ϫ 10 4 /cm 2 and cultured in DMEM/F-12 containing 10% FBS and 2% penicillin/streptomycin for 2 days. IL-1␤ was then added at a 2 ng/ml concentration, and cells were cultured for 24 or 48 h as indicated.
Northern Blot-Total RNA was isolated from cultured cells using Qiagen RNeasy mini kit TM , and 20 g of total RNA/lane were separated on a 1.5% agarose gel and transferred to Hybond-N TM nylon membrane.
The membranes were pre-hybridized for 5 h and then hybridized at 50°C overnight in Ultrahyb TM hybridization medium with the various probes labeled with [␣-32 P]dATP by random priming procedure. To obtain specific probes for C/EBP␤ and -␦, the sequences that have low homology with each other were chosen and digested from the corresponding expression vectors. Probe for CD-RAP was described before (34). The hybridized membranes were then washed twice with 2ϫ SSC and 0.1% SDS buffer at 50°C for 15 min and then by twice with 0.1 ϫ SSC and 0.1% SDS buffer in 50°C for 15 min. The specific signals from probes were obtained by autoradiography.
Transient Transfection and Luciferase Assay-DNA transfections of RCJ3.1C5.18 or RCS cells were performed using FuGENE 6 TM transfection reagent. 1 ϫ 10 5 of RCJ3.1C5.18 cells or 2.2 ϫ 10 5 of RCS cells were cultured on a 12-well plate overnight or for 6 h, respectively. Transfection mixture containing 3 l of FuGENE 6 TM , 375 ng of various promoter constructs, and 125 ng of pCMV-␤gal were then added, and the cells were cultured for 48 h with or without IL-1␤ as indicated. For co-transfection assay using C/EBP expression vector, 500 ng of expression vectors containing empty vector and C/EBP expression vector in various ratios were added to the transfection mixture. The cells were then harvested with Reporter Lysis Buffer TM , and the lysate was analyzed for luciferase activity using Promega Luciferase Assay Reagent TM . The ␤-galactosidase activities were also measured to normalize variations in transfection efficiency. Each transfection experiment was performed in triplicate and repeated at least twice.
Preparation of Nuclear Extracts and Synthesis of Protein in Vitro-Nuclear extracts and cytoplasmic protein from RCJ3.1C5.15 or RCS cells were isolated using Nu-Clear Extraction Kit TM according to the manufacturer's instructions. Sox9 and C/EBP-LIP proteins were synthesized using TNT T7 quick-coupled transcription/translation system with the Sox9 expression vector and the C/EBP-LIP expression vector, respectively.
Electrophoretic Mobility Shift Assay-Fragment A (between Ϫ2138 and Ϫ2037 bp relative to the mouse CD-RAP translation start site) was amplified by PCR. All oligonucleotides were synthesized by Invitrogen, and complementary oligonucleotide was annealed to make doublestranded oligonucleotide. The fragment A or various double-stranded oligonucleotides were end-labeled using T4 polynucleotide kinase and [␥-32 P]dATP. Bandshifts were performed by incubating 4 g of nuclear extracts in the mobility shift buffer (10 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, 50 mM NaCl, 0.5 mM dithiothreitol, 1 mM MgCl 2 , 4% glycerol, 2 g of poly(dI-dC) or poly(dG-dC)) with the DNA probe at room temperature for 25 min. In vitro synthesized proteins were assayed as described before (35,42). For the competition studies, the cold DNA fragments were added at a 100-fold molar excess compared with the probe and incubated for 15 min at room temperature before adding the DNA probe. For the antibody interference experiments, the nuclear extracts and 1 l of antibody were preincubated in the buffer for 1 h at 4°C. The anti-Sox9 antibodies was generous gift from Dr. de Crombrugghe (43). DNA protein complexes were resolved on a 5% polyacrylamide gel at 100 V for several hours. The gels were dried and autoradiographed.
Western Blot-Thirty g of cytoplasmic protein or nuclear extracts were denatured in SDS sample buffer contains 0.1 M dithiothreitol at 100°C for 5 min and separated on a 4 -20% Express Gel TM in Tris/ glycine/SDS buffer. The gels were then transferred to Hybond-C extra TM nitrocellulose membrane in Tris/glycine buffer, pH 8.3, containing 20% methanol. The membranes were saturated in 7% non-fat dry milk in PBS at room temperature for 1 h and hybridized with anti-C/ EBP antibodies diluted to 1:2000 in PBS containing 1% dry milk. The hybridized antibodies were recognized by anti-rabbit IgG antibodies coupled to horseradish peroxidase, and the secondary antibodies were detected by autoradiograph using SuperSignal TM peroxidase substrate.

IL-1␤-responsive Element Is Located between Ϫ2138 and
Ϫ2068 bp of the CD-RAP Promoter-We have shown previously that the Ϫ2251-bp CD-RAP promoter contains elements sufficient for cartilage-specific expression in transgenic mice (2) and that the expression of CD-RAP is down-regulated by IL-1␤ in primary chondrocyte cultures (9). To identify the IL-1␤-responsive element within CD-RAP promoter, three 5Ј-deletion constructs were transiently transfected into RCS cells and incubated in the absence or presence of IL-1␤ (Fig. 1). Regarding basal activities of the constructs in untreated RCS cells, Ϫ2251-bp promoter was stronger than Ϫ2138 bp, and Ϫ2138 bp was stronger than Ϫ2068 bp, suggesting there are activating elements between Ϫ2251 to Ϫ2138 bp and between Ϫ2138 to Ϫ2068 bp. The expressions of the Ϫ2251and Ϫ2138-bp constructs were down-regulated by the IL-1␤ treatment, whereas the response was lost in the Ϫ2068 bp, thus suggesting that the IL-1␤-responsive element is located between Ϫ2138 and Ϫ2068 bp.
IL-1␤ Stimulates the Expression of C/EBP␤ and C/EBP␦ in RCS Cells-The expressions of C/EBP␤ and -␦ are known to be stimulated by IL-1␤ (18,20). As computer analysis revealed there was a potential binding site for C/EBPs between Ϫ2138 and Ϫ2068 bp, we examined whether IL-1␤ stimulates the expression of C/EBP␤ or C/EBP␦ in RCS cells (Fig. 2, A and B). The expression levels of C/EBP␤ and C/EBP␦ were very low in steady state but were stimulated by IL-1␤ in 4 -24 h increasing up to 48 h. The expression of CD-RAP was repressed by 48 h. Western blots for C/EBP␤ revealed that in the absence of IL-1␤ treatment, a very small amount of C/EBP␤ protein exists in the cytoplasm but not in nuclei (Fig. 2C). The treatment of IL-1␤ stimulated the expression of both of the isoforms of C/EBP␤, LAP (36 kDa), and LIP (20 kDa) that were localized in the nuclei. Protein expression of C/EBP␦ was also stimulated by the IL-1␤ treatment (Fig. 2C).
C/EBP␤ Binds to the IL-1␤-responsive Element of the CD-RAP Promoter-To determine whether C/EBP␤ functions within the IL-1␤-responsive element, EMSA was carried out using nuclear extract from the RCJ3.1C5.18 cells. These cells were chosen because C/EBP␤ is expressed in steady state (Fig.  3A). Fragment A containing the sequence between Ϫ2138 and Ϫ2037 bp was used as a probe, and various cold oligonucleotides were used as competitors (Fig. 3B). Competitor 3, which has a potential binding site for C/EBPs at Ϫ2085/Ϫ2077 bp, competed with the binding of nuclear proteins and the probe (Fig. 3C). As expected, the mutant oligonucleotide 3, in which the C/EBP-binding site was mutated, did not compete with the binding, and supershift analysis confirmed that C/EBP␤ bound to the sequence of native oligonucleotide 3 (Fig. 3D). The antibody for Ikaros, a transcription factor that is also a potent binding protein for the sequence of oligonucleotide 3, did not affect the band shift. EMSA for the nuclear extract of RCS cells treated with 10 ng/ml of IL-1␤ to induce C/EBP␤ (see Fig. 2C) was also carried out using the oligonucleotide 3 as a probe to confirm that the IL-1␤-induced C/EBP␤ bound to the element (data not shown).
C/EBP␤ Functions as a Repressor for the CD-RAP Promoter Activities-The C/EBP-binding site was mutated in the Ϫ2251-bp construct using mutant oligonucleotide 3. The promoter activity of the mutant Ϫ2251-bp construct was about 2-fold stronger than that of wild type Ϫ2251-bp construct after transient transfection into RCJ3.1C5.18 cells, suggesting that C/EBP␤ is acting as a repressor (Fig. 4). To investigate the C/EBP function in detail, the expression vectors for C/EBP␤ and C/EBP␦ were co-transfected with wild type Ϫ2251-bp construct into RCS cells, in which endogenous C/EBP expression is very low (Fig. 5A). Because C/EBP␤ is known to act in different ways depending on the isoforms, we used the following three different expression vectors: pCMV-C/EBP-full-length, pCMV-LAP, and pCMV-LIP. The promoter activity of the Ϫ2251-bp construct was down-regulated in a dose-dependent manner by FIG. 1. IL-1␤-responsive element is located between ؊2138 to ؊2068 bp of the CD-RAP promoter. Various 5Ј-deletion constructs of the CD-RAP promoter were transiently transfected into RCS cells and incubated for a further 48 h in the absence or presence of IL-1␤ (10 ng/ml). Luciferase activities were measured and expressed relative to the activity of promoterless pGL3b (set as 1). Each bar represents the mean Ϯ S.D. all three of the C/EBP␤ expression vectors, no matter what isoform was expressed. C/EBP␦ also down-regulated the activity of the Ϫ2251-bp construct. The presence of each expressed protein was confirmed by Western blot of lysate after the trans-fection (Fig. 5B). The C/EBP␤ or empty expression vectors (500 ng) and the Ϫ2251-bp construct were also co-transfected into undifferentiated ATDC5 mouse chondrogenic cell line. Interestingly, pCMV-C/EBP-FL and pCMV-LAP vectors stimulated promoter activity to 3.44 Ϯ 0.57-and 30.2 Ϯ 4.66-fold, respectively, whereas pCMV-LIP did not (1.03 Ϯ 0.14). These data suggest that the repression induced by C/EBP expression vectors in RCS cells is neither because of nonspecific reaction nor poor function of these expression vectors.
Mutagenesis of C/EBP-binding Site Eliminates the Response to IL-1␤ Treatment-To determine whether the C/EBP-binding site on the CD-RAP promoter was required for the response to IL-1␤, the mutant Ϫ2251-bp construct, in which the C/EBPbinding site was mutated, was transfected into RCS cells and incubated in the absence or presence of IL-1␤ for a further 48 h. As shown Fig. 6A, the mutant Ϫ2251-bp construct did not respond to IL-1␤ treatment. Furthermore, co-transfection of the pCMV-C/EBP-full-length vector did not result in repression of activity of the mutant Ϫ2251-bp construct. These results indicate that the C/EBP-binding site at Ϫ2085/Ϫ2077 bp of the CD-RAP promoter is required for the functional response to IL-1␤ and suggest that C/EBPs are key factors in this response.
Sox9 Binds to the Element Overlapping with the C/EBP Motif-Because the IL-1␤-responsive element seems to contain an activating element as well (see Fig. 1), we hypothesized that C/EBP may interact with another activator protein conferring an indirect repressive effect. Because both of the flanking sequences of the C/EBP motif contain an HMG-like motif, the binding of Sox9 to this element was tested (Fig. 7). EMSA using

FIG. 5. The promoter activity of the ؊2251-bp construct is down-regulated by co-transfection with C/EBP expression vectors. A, the Ϫ2251-bp construct was co-transfected into RCS cells with various C/EBP expression vectors. Expression vectors for C/EBP␤-fulllength (FL)
, C/EBP␤-LAP, C/EBP-LIP, or C/EBP␦ were added to the transfection mixture as indicated. Total amounts of expression vectors were adjusted to 500 ng using the empty expression vector in each transfection. The activity of Ϫ2251-bp constructs co-transfected with 500 ng of the empty expression vector is set at 100. Each bar represents the mean Ϯ S.D. B, Western blot of the lysate from the co-transfection assay for C/EBP␤ confirming that the expression vectors form the protein products.
RCS nuclear extract revealed the Sox9 binding to the sequence of oligonucleotide 3 (Fig. 7B). To see the interaction between Sox9 and C/EBP␤ within this element, EMSA was carried out using Sox9 and C/EBP-LIP protein synthesized by in vitro translation. The binding of Sox9 protein to the sequence of oligonucleotide 3 was confirmed by supershift using anti-Sox9 antibody (Fig. 7C). To screen the binding site, each of two different competitors was preincubated with the binding reaction mix. The competitor b, which has the sequence TTCAAAA that is close to the Sox protein motif (A/T)(A/T)CAA(A/T)G, competed with the Sox9 binding to the probe more strongly than the competitor a, suggesting Sox9 prefers the motif in b for its binding (Fig. 7C). When C/EBP␤ and Sox9 proteins were incubated together with the probe, two major independent bands were observed to be the same size as Sox9 alone and C/EBP alone (Fig. 7D). The addition of antibodies for either Sox9 or C/EBP␤ generated supershift of the related band and did not affect the other band (Fig. 7D). These results indicate that the supershifted complex did not contain the other protein, suggesting that Sox9 and C/EBP␤ cannot bind to the probe in the same time probably because the binding sites partially overlap. Therefore, C/EBP␤ may compete with Sox9 binding to this element thus eliminating the enhancer effect of Sox9.
Type II Collagen Promoter-Enhancer Construct Is Down-regulated by C/EBPs-Because the expression of type II collagen is also down-regulated by IL-1␤ and the data base analysis revealed that the first intron of type II collagen gene has multiple C/EBP motifs, we examined whether C/EBPs also regulate the promoter activity of the type II collagen gene. Co-transfection of the type II collagen promoter-enhancer construct, pGL2-COL2-577/ϩ3426, into RCS cells with the expression vectors for C/EBP␤-LAP, C/EBP␤-LIP, or C/EBP␦ also down-regulated the reporter gene activity (Fig. 8). Similar to the response of the CD-RAP promoter (see Fig. 5A), overexpression of C/EBP␤-LAP produced more potent inhibition than C/EBP-LIP or C/EBP␦. These results indicate that C/EBP␤ could mediate, at least partially, the inhibitory effect of IL-1␤ on the type II collagen gene as well.

C/EBP␤ and -␦ Are Stimulated in Normal Chondrocytes from Human Articular Cartilage in Response to IL-1␤ Treatment-
Our studies have been focused on two chondrocytic cell lines producing low (RCJ3.1C5.18) and high (RCS) amounts of CD-RAP. In order to examine this response in normal chondrocytes, the levels of C/EBP␤ and -␦, CD-RAP, and type II colla-gen mRNAs were investigated in IL-1␤-treated chondrocytes isolated from human articular cartilage (Fig. 9). Cells were treated with 2 ng/ml IL-1␤ for up to 48 h. By using the real time quantitative RT-PCR method, the expression of C/EBP␤ and -␦ were shown to increase at 24 h up to 3-and 2-fold, respectively, and returned to basal levels by 48 h. Type II collagen mRNA significantly decreased to 8% within 24 h, whereas CD-RAP mRNA decreased relatively slowly to 56% compared with the controls within 48 h. Consistent with the results from cell lines, these results suggest that C/EBP␤ and -␦ can regulate the expression of CD-RAP and type II collagen in articular cartilage in the same manner. DISCUSSION The mechanism by which the proinflammatory cytokines IL-␤ and TNF-␣ stimulate inflammation and degradative enzymes as well as repress synthesis of structural matrix is FIG. 6. Mutation of the C/EBP-binding site abolishes the responsiveness of the ؊2251-bp promoter to IL-1␤ treatment. A, the mutant Ϫ2251-bp construct, in which the C/EBP-binding site was mutated, was transfected into RCS cells and incubated with or without IL-1␤ (10 ng/ml). B, the mutant Ϫ2251-bp construct was co-transfected into RCS cells with empty or C/EBP-FL expression vector. The promoter activity of the construct without IL-1␤ and that co-transfected with empty vector were set at 100. Each bar represents the mean Ϯ S.D.
FIG. 7. Sox9 and C/EBP␤ bind to ؊2094/؊2069 bp and compete for binding to the probe. A, the sequences of oligonucleotide 3 and competitors a and b. The C/EBP consensus and HMG-like motifs are underlined. B, EMSA was carried out using oligonucleotide 3 as a probe and RCS nuclear extract (N.E.). The addition of anti-Sox9 antibody generated a supershift. The specificity of the anti-Sox9 antibody has been confirmed before in RCS cell nuclear extract (43) and in vitro translated protein (35). C, EMSA was performed using Sox9 protein synthesized by in vitro translation. Binding of Sox9 was confirmed by supershift using anti-Sox9 antibody (lanes 1 and 2). Competitor b competes with the binding of Sox9 to the probe more strongly than a (lanes 3 and 4). D, EMSA using Sox9 and C/EBP␤ proteins synthesized by in vitro translation. When both Sox9 and C/EBP␤ are incubated with the probe (lane 4), two major independent bands were observed to be the same size as the bands obtained by Sox9 (lane 1) or C/EBP␤ (lane 2) alone. The addition of antibody for Sox9 (lane 5) or C/EBP␤ (lane 6) generated supershift of the related band and did not affect to the other band. thought to occur via independent signal transduction and transcriptional mechanisms (44,45). In this study, we show that C/EBP is a critical regulator for repression of CD-RAP transcription induced by IL-1␤. In contrast, C/EBP␤ is known to activate expression of phospholipase A 2 , cyclooxygenase-2, and manganese superoxide dismutase in response to IL-1␤ (17,21,29) and thus promote inflammatory reactions or catalytic effects in tissues. This is the first report showing that C/EBP mediates repressive effects of IL-1␤ for matrix-associate proteins, including type II collagen, thereby elucidating a common pathway in chondrocytes for IL-1-induced activation of degradation and repression of new matrix synthesis.
The mechanism of C/EBP␤ activation in response to IL-1␤ and TNF-␣ is complex and still controversial. It is reported that IL-1␤ or TNF-␣ stimulate C/EBP␤ expression at the mRNA level in liver, spleen, and kidney in vivo (18), in MC3T3-E1 osteoblastic cells (19) and in J774.2 macrophage cells (20). In contrast, no change in the mRNA expression is observed in transformed hepatocyte (25), in 3T3-L1 adipocyte (28), and in rabbit articular chondrocyte (21). IL-1, TNF-␣, or antioxidant activates C/EBP␤ in hepatocytes at the post-translational level by phosphorylation of serine within the transcriptionactivating domain (23) and DNA binding domain (24,26), followed by nuclear translocation (25,26), enhancing transcrip-tional efficacy (23,24). In contrast to the nuclear translocation, it is also reported that TNF-␣ induces nuclear export of C/EBP␤ by phosphorylation of serine resulting in repression of albumin gene in hepatocyte (46). In the current study, the amounts of C/EBP␤ proteins were very low in cytoplasm and nuclei of RCS cell without IL-1␤ treatment. IL-1␤ stimulated both of C/EBP␤ and -␦ mRNA expression resulting in increase of both proteins in the nuclei. The mRNA in human articular chondrocytes also responded to the IL-1␤ treatment. It is likely that the mechanism of C/EBP␤ activation varies depending on cell types and conditions.
In the current study, the repression of CD-RAP mRNA occurred more than 24 h later than the activation of C/EBPs in RCS cells and human chondrocyte. This time lag may be due to the relatively high stability of CD-RAP mRNA. We performed RNA stability assays for CD-RAP using the RNA polymerase II inhibitor, 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole (60 M) in RCS cell and in bovine articular chondrocyte (9). CD-RAP mRNA was still present at 90% after 6 h of treatment and at 60 (RCS cell) to 40% (bovine chondrocyte) after 24 h of treatment.
C/EBP␤ is an unusual transcription factor in that it can activate or repress gene transcription. For example, C/EBP␤ activates the gene encoding the ␣1 chain of type I collagen (30), whereas it represses the gene encoding the ␣2 chain of type I collagen (22) and albumin genes (47). We found an HMG-like motif partially overlapping with the C/EBP motif that can bind to Sox9. As binding of one protein appears to preclude binding of the other, the balance between C/EBP and Sox proteins may regulate the transcription of CD-RAP. The interaction of C/EBPs with other transcription factors has been observed in other genes, although the mechanisms of binding and the results of binding are specific to the transcription factor and gene. For example, in the CD11c and rat cytochrome CYP2D5 genes, C/EBP acts synergistically with the transcription factor Sp1 to enhance gene expression through a mechanism whereby Sp1 facilitates the binding of C/EBP to a low affinity site in the DNA (48,49). C/EBP also interacts with retinoblastoma protein and various other transcriptional activators or co-activators such as c-Myb, PU.1, and ATF-2 (41,50). In the osteocalcin gene, C/EBP␤ binds to DNA and, in turn, binds to the transcription factor Runx2/Cbfa1 to greatly enhance gene expression (32).
We have found that C/EBP also repressed the transcription of the type II collagen gene. There are multiple C/EBP motifs in the first intron of type II collagen gene. Interestingly, analysis revealed that there is a C/EBP motif within the 48-bp core enhancer element defined by Lefebvre et al. (42,43) that regulates cartilage-specific expression of type II collagen gene; the C/EBP motif is next to the Sox9-binding motif in the element. A zinc finger transcription factor, ␣A-crystallin-binding protein 1 (CRYBP1), that is expressed in a reciprocal pattern compared with that of type II collagen is reported to bind to the sequence overlapping the C/EBP motif and compete with Sox9 binding thereby repressing gene expression of type II collagen (51). C/EBP may participate in IL-1␤-induced repression of type II collagen by taking place of the CRYBP1 binding and competing with Sox9 binding in a similar manner. It has been reported that Sox9 is down-regulated by IL-1␤ and TNF-␣ (52). In that report, the expression of Sox9 was decreased within a few hours via direct interaction of IL-1␤-induced NF-B with the Sox-9 promoter, whereas the decrease of type II collagen takes much longer. Because no further Sox9 is then expressed but the existing Sox9 remains bound to the enhancer of type II collagen gene, C/EBP might be involved in this regulation by displacing Sox9 from the element. Further studies will be performed to ascertain the involvement of C/EBP in the regulation of type II collagen gene.
One of the unique observations of CD-RAP gene is that its physiological expression is primarily restricted to cartilage. Various transcription factors have been shown to function in regulating the CD-RAP promoter activity. For example, AP-2 binds to the sequence from Ϫ463 to Ϫ456 bp and regulates transcription in a biphasic manner, activating at a low concentration of AP-2 and repressing at a high concentration (34). Sox9 binds at Ϫ410 to Ϫ404 bp (a different site from the one shown in this study) and activates transcription (35). Upstream stimulatory factor and delta-EF1 were also found to function in an E box located at Ϫ488 to Ϫ482 bp and activate or repress CD-RAP depending on the relative population of upstream stimulatory factor and delta-EF1 in the nuclei (53). Studies of transgenic mice, which harbor various lengths of CD-RAP 5Ј-flanking sequence linked to ␤-galactosidase gene (lacZ), revealed that Ϫ2251-bp promoter was sufficient to direct cartilage-specific expression of reporter gene, but Ϫ2068 bp was not (2). These results in transgenic mice indicate that the regulation of tissue-specific expression of the CD-RAP gene cannot be explained completely by the factors found binding to the sequence surrounding Ϫ400 bp and that there may be important elements between Ϫ2251 and Ϫ2068 bp needed to control tissue-specific expression.
In the current study, we show that there is a C/EBP-binding site within the domain located on Ϫ2091 to Ϫ2066 bp, which functions as a repressor, and the HMG-like motif, adjacent to the C/EBP site, which could be a potent activator domain by binding of Sox proteins. Another HMG-like motif at Ϫ2152 to Ϫ2245 bp has been shown from our laboratory (54), which is consistent with the presence of another activating element between Ϫ2251 and Ϫ2138 bp. These positive and negative regulators within the tissue-specific domain may exert the tissue-specific regulation of the CD-RAP gene. C/EBP expression is negatively correlated with CD-RAP synthesis. For example, in undifferentiated RCJ3.1C5.18 cells C/EBP␤ is present, and expression of CD-RAP is very low, whereas in RCS, whose expressions of C/EBP␤ and ␦ are very low, CD-RAP expression is very high. In skeletal tissue, C/EBP␣ is reported to be expressed in the germinal cell but not in mature chondrocytes of the growth plate (55). C/EBP␤ and -␦ are expressed in osteoblasts and activate osteocalcin gene transcription in cooperation with Runx2 (32). The reciprocal expression of C/EBP and CD-RAP suggests that C/EBP may also participate in tissue-specific regulation of CD-RAP by repressing its expression in undifferentiated chondrocytes, osteoblasts, or other tissues. In preliminary studies, we have generated transgenic mice harboring the Ϫ3345-bp promoter of CD-RAP linked to lacZ gene. This promoter directs cartilage-specific expression of the reporter gene. In another transgenic line, in which the sequence from Ϫ2251 to Ϫ2068 bp (including the C/EBP-binding site) was removed from the Ϫ3345-bp construct, the reporter gene was expressed widely. 2 These results support the hypothesis that C/EBP may be responsible for tissue-specific repression of CD-RAP gene expression.
In summary, the data presented here show that C/EBP is a critical factor for IL-1␤-induced repression of CD-RAP gene. This effect may be coordinated with binding competition between C/EBP and the enhancer protein Sox9. C/EBP also down-regulated the type II collagen gene transcription. This is the first report that has shown that C/EBP regulates a cartilage-specific gene. These results strongly suggest that C/EBP is the common factor accounting for the disparate effects of IL-1␤: transcriptional repression of chondrocyte extracellular matrix production in addition to the promotion of inflammatory cascade and catabolic processes in joint diseases, such as rheumatoid arthritis and osteoarthritis.