CCAAT/Enhancer-binding Protein β Regulates the Repression of Type II Collagen Expression during the Differentiation from Proliferative to Hypertrophic Chondrocytes*

Background: CCAAT/enhancer-binding protein β (C/EBPβ) promotes hypertrophic differentiation of chondrocytes. Results: C/EBPβ directly represses the expression of type II collagen by interacting with its intronic enhancer. Conclusion: C/EBPβ biphasically functions to repress genes characteristic of proliferative chondrocytes while stimulating genes expressed by hypertrophic chondrocytes. Significance: C/EBPβ is a key regulator to trigger phenotypic conversion from proliferative to hypertrophic chondrocytes. CCAAT/enhancer-binding protein β (C/EBPβ) is a transcription factor that promotes hypertrophic differentiation by stimulating type X collagen and matrix metalloproteinase 13 during chondrocyte differentiation. However, the effect of C/EBPβ on proliferative chondrocytes is unclear. Here, we investigated whether C/EBPβ represses type II collagen (COL2A1) expression and is involved in the regulation of sex-determining region Y-type high mobility group box 9 (SOX9), a crucial factor for transactivation of Col2a1. Endogenous expression of C/EBPβ in the embryonic growth plate and differentiated ATDC5 cells were opposite to those of COL2A1 and SOX9. Overexpression of C/EBPβ by adenovirus vector in ATDC5 cells caused marked repression of Col2a1. The expression of Sox9 mRNA and nuclear protein was also repressed, resulting in decreased binding of SOX9 to the Col2a1 enhancer as shown by a ChIP assay. Knockdown of C/EBPβ by lentivirus expressing shRNA caused significant stimulation of these genes in ATDC5 cells. Reporter assays demonstrated that C/EBPβ repressed transcriptional activity of Col2a1. Deletion and mutation analysis showed that the C/EBPβ core responsive element was located between +2144 and +2152 bp within the Col2a1 enhancer. EMSA and ChIP assays also revealed that C/EBPβ directly bound to this region. Ex vivo organ cultures of mouse limbs transfected with C/EBPβ showed that the expression of COL2A1 and SOX9 was reduced upon ectopic C/EBPβ expression. Together, these results indicated that C/EBPβ represses the transcriptional activity of Col2a1 both directly and indirectly through modulation of Sox9 expression. This consequently promotes the phenotypic conversion from proliferative to hypertrophic chondrocytes during chondrocyte differentiation.

The sequential differentiation process of chondrocytes is observed not only in skeletal formation, but also in fracture healing and development of osteophytes in osteoarthritis (OA) 2 (1)(2)(3)(4). Chondrogenesis initiates when mesenchymal cells condense and differentiate into proliferative chondrocytes that synthesize cartilage-specific extracellular matrix (ECM) including type II collagen (COL2A1) and aggrecan (ACAN). Thereafter, chondrocytes change morphology to become hypertrophic chondrocytes and convert their gene expression profile. They stop expressing COL2A1 and ACAN and start expressing type X collagen (COL10A1), matrix metalloproteinase-13 (MMP13), and VEGF. Finally, osteoblasts migrate into the cartilage, a process that is accompanied by vascular invasion and apoptosis of hypertrophic chondrocytes, and complete formation of bone. In contrast to the developing cartilage of the growing bone, chondrocytes in adult articular cartilage do not pursue the hypertrophic differentiation process and maintain the characteristic ECM structure consisting of COL2A1 and ACAN. However, phenotypic conversion to hypertrophic chondrocytes is also observed in degenerative articular cartilage of OA (2).
This sequential differentiation process is tightly regulated by various transcription factors (1,3). Sex-determining region Y-type high mobility group box 9 (SOX9) is an essential transcription factor for chondrogenesis during mesenchymal condensation and chondrocyte proliferation (5). In humans, its malfunction causes a severe chondrodysplasia called campomelic dysplasia, which is characterized by severe malformations of cartilage-derived structures. Sox9 knock-out mice showed no aggregation of chondroblasts and subsequent expression of ECM (6). Furthermore, SOX9 was reported to directly bind and activate cartilage-specific regulatory elements of Col2a1, resulting in its cartilage-specific gene expression (7,8). Although SOX9 is indispensable for chondrogenesis, it was also reported to work as a negative regulator of hypertrophic differentiation (9). Several studies have reported that Winglesstype murine mammary tumor virus integration site (10), bone morphogenetic protein 2 (BMP2) (11), parathyroid hormonerelated protein (PTHrP) (12), IL-1␤ (13), and Runt-related transcription factor 2 (RUNX2) (14) can regulate SOX9 expression or activity in arthritic chondrocytes or during chondrocyte differentiation.
CCAAT/enhancer-binding proteins (C/EBPs) are a family of basic leucine zipper transcription factors with six members as follows: C/EBP␣, ␤, ␦, ⑀, ␥, and . Among them, C/EBP␤ (encoded by CEBPB) was first identified as a nuclear protein that bound to the IL-1␤ response element in the IL-6 promoter region (15), and it has subsequently been reported to regulate various genes involved in cell differentiation, proliferation, survival, immune function, tumor invasiveness, and progression (16 -19). Previously, it was reported that C/EBP␤, in response to IL-1␤, down-regulated cartilage-derived retinoic acid-sensitive protein (Cd-rap), a small secreted protein expressed in cartilage throughout chondrogenesis and in mature chondrocytes (20). C/EBP␤ induced by pro-inflammatory cytokines such as IL-1␤ and TNF-␣ directly binds to the MMP13 and MMP3 promoter regions and stimulates their expression in chondrocytes, resulting in degradation of cartilage in arthritis (21,22). Furthermore, it was reported that hypertrophic differentiation of chondrocytes was delayed in C/EBP␤ knock-out mice through transactivation of cell cycle factor p57 (23). This was associated with decreased expression of Col10a1 (24) and suggested that C/EBP␤ plays an important role in promoting hypertrophic differentiation of chondrocytes. C/EBP␤ is also involved in the hypertrophic changes of articular chondrocytes in OA (23). Although these studies showed that C/EBP␤ stimulates genes expressed in hypertrophic chondrocytes, it is not fully understood whether C/EBP␤ influences expression of genes characteristic of proliferative chondrocytes.
Here, we investigated the direct involvement of C/EBP␤ in regulating ECM of proliferating chondrocytes, specifically Col2a1, during chondrocyte differentiation. We also investigated the regulation of Sox9 expression by C/EBP␤, which indirectly leads to the regulation of Col2a1 during chondrocyte differentiation.
Cell Culture-ATDC5 cells, a mouse chondrogenic cell line, were maintained in DMEM/Ham's F-12 medium supplemented with 5% FBS. To induce chondrogenic and hypertrophic differentiation, subconfluent cultures were changed to medium containing 1% insulin-transferrin-selenium universal culture supplement premix reagent (BD Biosciences, San Jose, CA). Rat chondrosarcoma (RCS) cells and SW1353 human chondrosarcoma cells were cultured in DMEM with 10% FBS. Primary chondrocytes were isolated from the rib cages and sternums of 1-day-old mice as previously described (25) and cultured in DMEM with 10% FBS.
Western Blot-Whole cell lysates were extracted from cells using M-RIPA buffer (Sigma-Aldrich). Nuclear extracts were isolated using nuclear and cytoplasmic extraction reagents (Thermo Scientific, Rockford, IL). Cell lysates were electrophoresed in 4 -12% gradient polyacrylamide gels (Invitrogen) and transferred to nitrocellulose membranes (Amersham Biosciences). After blocking in Tris-buffered saline-Tween containing 3% nonfat milk, the membranes were incubated with primary antibody against C/EBP␤ (C-19; Santa Cruz Biotechnology) diluted 1:500, SOX9 (AB5535; Millipore) diluted 1:1000, or FLAG (F4042; Sigma-Aldrich) diluted 1:1000 in blocking reagent at room temperature for 1 h. We also used ACTIN (MAB1501; Millipore) and LAMIN A/C (H-110; Santa Cruz Biotechnology) antibodies as internal loading controls. Horseradish peroxidase-conjugated secondary antibody (Santa Cruz Biotechnology) diluted in blocking reagent was added and incubated at room temperature for 1 h. The immunoreactivity of the blots was detected using ECL Prime (Amersham Biosciences) and photographed with Ez Capture MG (ATTO, Tokyo, Japan). The band densities were evaluated by densitometric scan using CS Analyzer (ATTO).
Chondrocyte Differentiation Assay-ATDC5 cells were cultured for 3 weeks after transfection with adenovirus vectors. The ATDC5 cells were washed three times with PBS then fixed with 4% formaldehyde for 10 min. Alcian blue staining was performed with 0.3% Alcian blue 8GS (Sigma-Aldrich) in 0.1 N HCl. After staining overnight, the cells were washed with distilled water three times.
Plasmid Preparation and Reporter Assay-Mouse Col2a1 sequences spanning Ϫ952 to ϩ73 and ϩ2038 to ϩ2678 (including promoter and intron 1) were subcloned into the pGL-4.10 (luc2) vector (Promega, Madison, WI). Deletion sequences of promoter and enhancer were also generated in various combinations using PCR. Mutation constructs were made with the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). RCS or SW1353 cells were seeded 1 day before transfection in 24-well tissue culture plates at 3 ϫ 10 4 cells/cm 2 . 0.4 g of reporter plasmids were co-transfected with expression vectors like pCMV-LAP (an expression vector of rat C/EBP␤) and an A-C/EBP vector tagged with FLAG (a dominant-negative C/EBP expression vector kindly provided by Dr. Charles R. Vinson) (28) using Lipofectamine Plus reagent (Invitrogen). Lucif-erase activity was measured 48 h after transfection using the Dual-Luciferase reporter assay system (Promega).
Electrophoretic Mobility Shift Assay-Nuclear protein was extracted from ATDC5 cells that had been transfected with C/EBP␤ using nuclear and cytoplasmic extraction reagents (Thermo Scientific). Complementary oligonucleotides were end-labeled with the Biotin 3Ј end DNA labeling kit (Thermo Scientific) and then annealed to obtain double-stranded oligonucleotides. EMSA was performed using the LightShift chemiluminescent EMSA kit (Thermo Scientific). Twenty fmol of biotin-labeled probes were incubated with nuclear protein in 1ϫ binding buffer (including 2.5% glycerol, 5 mM MgCl 2 , 50 ng/l poly(dI-dC)) at room temperature for 20 min. For competition experiments, the cold probes were added at a 200-fold molar excess. For antibody interference experiments, the nuclear extract and 1 l of C/EBP␤ antibody (Santa Cruz Biotechnology) were preincubated for 1 h at 4°C. Binding samples were subjected to electrophoresis in a 6% DNA retardation gel (Invitrogen) and run in 0.5ϫ TBE buffer at 100 V for 1 h, then transferred to a positively charged membrane (Invitrogen), and cross-linked. Detection was performed using streptavidinhorseradish peroxidase conjugate and chemiluminescent substrate. The oligonucleotides were as follows: wild type, 5Ј-CAG-ATGGGCTGAAACCCTGCC-3Ј (sense) and 5Ј-GGCAGGG-TTTCAGCCCATCTG-3Ј (antisense); and mutant, 5Ј-CAGA-TGGGCTGACCCCCTGCC-3Ј (sense) and 5Ј-GGCAGGGG-GTCAGCCCATCTG-3Ј (antisense).
Chromatin Immunoprecipitation Assay-ChIP assays were performed with a ChIP assay kit (Millipore). ATDC5 cells were differentiated for 3 weeks to induce hypertrophic differentiation or cultured for 4 days after transfection with adenovirus vectors expressing C/EBP␤ or LacZ control. The ATDC5 cells were fixed with 4% formaldehyde and sonicated. For immunoprecipitation experiments, C/EBP␤ antibody (Santa Cruz Biotechnology), SOX9 antibody (Millipore), and normal rabbit IgG (Santa Cruz Biotechnology) were used. Primers used in PCR were as follows: amplified between ϩ2003 and ϩ2198 bp for the Col2a1 enhancer including the C/EBP␤-binding site, between ϩ2136 and ϩ2292 bp for the Col2a1 enhancer including the SOX9-binding site, and between Ϫ1891 and Ϫ1611 bp as a negative control.
Ex Vivo Organ Culture-Tibias were isolated from hind limbs of E14.5 mouse embryos and cultured in organ culture medium (29). One day after dissection, each tibia obtained from identical mouse embryos were transfected with adenovirus vectors expressing C/EBP␤ or LacZ control and cultured at 37°C in a humidified 5% CO 2 incubator for 4 days. Simultaneously, tibias dissected from different embryos were cultured without transfection of adenovirus vectors for 4 days. Safranin O and immunofluorescent staining was performed. Histological analysis was repeated at least twice for each sample from six pairs of limbs, respectively.
Statistical Analysis-The data are reported as means Ϯ S.D. of three independent experiments, each performed in duplicate. Statistical analysis was performed using JMP 9 statistical software (SAS Institute, Inc., Cary, NC). The Mann-Whitney U test was used for two-group comparisons. p Ͻ 0.05 was considered statistically significant.

RESULTS
Expression Patterns of C/EBP␤, COL2A1, and SOX9 during Chondrocyte Differentiation-To confirm the expression pattern of C/EBP␤ and chondrocyte differentiation markers in vivo, immunohistochemistry of mouse embryos was performed (Fig. 1A). Consistent with previous reports, SOX9 was strongly expressed from proliferative to prehypertrophic chondrocytes, but not by hypertrophic chondrocytes. COL2A1 was expressed in chondrocytes from the proliferative zone and then gradually decreased toward the hypertrophic zone. C/EBP␤ was weakly detected in proliferative chondrocytes, but strongly expressed by prehypertrophic and hypertrophic chondrocytes where it co-localized with COL10A1. These results indicated an opposite expression pattern between C/EBP␤ and COL2A1 and SOX9 during chondrocyte differentiation.
To investigate the relationship of expression patterns between C/EBP␤ and various chondrocyte differentiation markers in vitro, ATDC5 cells were used. To confirm that ATDC5 cells possess the ability of chondrogenic and hypertrophic differentiation, the expression of each marker was compared between ATDC5 cells (at the 14th day) and primary chondrocytes obtained from mouse ribs and sternums cartilage. The mRNA expression profile of these differentiation markers in ATCD5 cells was similar to that in primary chondrocytes ( Table 1), indicating that ATDC5 cells represent the chondrocytic nature. During differentiation of ATDC5 cells, expression of Col2a1 mRNA gradually increased to a maximum at the 14th day and then decreased (Fig. 1B). This pattern was consistent with Sox9 mRNA expression. In contrast, Western blot analysis of nuclear proteins extracted from ATDC5 cells revealed that C/EBP␤ was elevated at a later stage, similar to Col10a1 mRNA expression (Fig. 1, B and C). These findings suggested that C/EBP␤ could be involved in the repression of Col2a1 and Sox9 during hypertrophic differentiation of chondrocytes. were subject to immunohistochemistry with COL2A1, SOX9, COL10A1, and C/EBP␤ antibodies. Tissue stained with IgG is shown as a negative control. Hematoxylin was used as a counterstain. Red, green, and blue bars indicate the proliferative, prehypertrophic, and hypertrophic zones, respectively. Scale bar, 500 m. The data are representative of two independent experiments performed in duplicate. B, time course of Col2a1, Sox9, Col10a1, and Col1a1 mRNA was measured by real time RT-PCR in differentiating ATDC5 cells cultured for 4 weeks. Each value was normalized to 18S in the same sample. The value of mRNA expression at each stage relative to that on the 4th day was indicated. Means Ϯ S.D. of duplicates from three independent experiments are shown. C, time course of C/EBP␤ protein was determined by Western blot of nuclear extracts of ATDC5 cells cultured for 4 weeks. The data are representative of two independent experiments performed in duplicate.

C/EBP␤ Represses Expression of Col2a1 and Acan, while
Stimulating Expression of Hypertrophic Markers-To investigate the effect of C/EBP␤ on chondrocyte differentiation markers, ATDC5 cells were transfected with adenovirus vectors expressing C/EBP␤ or LacZ control, and the cells were subsequently differentiated. Hypertrophic markers such as Col10a1 and Mmp13, as well as Runx2, which is known to be a major regulator of chondrocyte maturation and osteoblast differentiation (30), were significantly increased along with ectopic Cebpb expression (Fig. 2A). These findings confirmed that C/EBP␤ promotes hypertrophic differentiation of chondrocytes. Conversely, the expression of both Col2a1 and Acan was significantly repressed in ATDC5 cells transfected with C/EBP␤ at all stages of differentiation (Fig. 2B). The expression of Col1a1 was decreased at a later stage during differentiation of ATDC5 cells (Fig. 1B) and was also reduced by overexpression of C/EBP␤ ( Fig. 2A), demonstrating that repression of Col2a1 and Acan is not the cause of chondrocyte dedifferentiation toward fibroblastic phenotype. These results were further confirmed by Alcian blue staining of ATDC5 cells cultured for 3 weeks (Fig. 2C). These findings indicate that C/EBP␤ may biphasically function to repress the expression of Col2a1 and Acan, while stimulating hypertrophic markers during hypertrophic differentiation of chondrocytes.
C/EBP␤ Indirectly Repressed Expression of Col2a1 through Interacting with SOX9-We investigated the effect of C/EBP␤ on the expression of Sox9 because SOX9 is known to be a crucial transcription factor for activating Col2a1. The expression of Sox9 mRNA was significantly repressed by C/EBP␤ at all stages of differentiation along with ectopic Cebpb expression (Fig. 3A). Western blot analysis also revealed that SOX9 proteins extracted from both cytoplasm and nucleus were strongly repressed by C/EBP␤ (Fig. 3B). Furthermore, a ChIP assay using  ATDC5 cells transfected with adenovirus vectors revealed that the binding of SOX9 to Col2a1 intronic enhancer was decreased by overexpression of C/EBP␤ (Fig. 3C). These results indicated that C/EBP␤ may repress the expression of Col2a1 through interacting with SOX9 during hypertrophic differentiation of chondrocytes.

C/EBP␤ Knockdown by shRNA Increased Expression of Chondrocyte Differentiation
Markers-To further examine the relationship between C/EBP␤ and chondrocyte differentiation markers, shRNA-transfected ATDC5 cells targeting Cebpb were differentiated. Nuclear extracts and mRNA expression of C/EBP␤ were effectively reduced by shRNA compared with control (Fig. 4, A and C). The expression of Col10a1 was reduced at a later stage (Fig. 4A), whereas the expression of Col2a1 and Acan was significantly increased by C/EBP␤ knockdown along with chondrocyte differentiation (Fig. 4B). The expression of Sox9 mRNA was significantly increased on the 4th day in ATDC5 cells transfected with shRNA (Fig. 4B). Furthermore, nuclear protein of SOX9 was markedly increased on the 4th day, and this was also confirmed by densitometric scan (Fig. 4, C and D). Together, these findings suggest that C/EBP␤ is involved in repression of differentiation markers at the endogenous level during hypertrophic differentiation of chondrocytes.
C/EBP␤ Repressed Transcriptional Activity of Col2a1 in Chondrocytes-We confirmed that the expression of Col2a1 is regulated by C/EBP␤. However, it is unclear whether C/EBP␤ directly regulates the expression of Col2a1 because SOX9 could be involved in this regulation. To investigate the direct regulation of Col2a1 by C/EBP␤, a Col2a1 reporter construct was generated (Fig. 5A). We used two chondrocytic cell lines, RCS and SW1353 cells. RCS cells constitutively express Sox9, leading to stronger expression of Col2a1 than SW1353 cells. On the other hand, SW1353 cells strongly express endogenous CEBPB compared with RCS cells (Fig. 5B). Luciferase activity of Col2a1 reporter construct in RCS cells was significantly higher than in SW1353 cell (Fig. 5C), indicating the validity of this reporter construct. Col2a1 promoter activity was down-regulated by C/EBP␤ in a dose-dependent manner even in RCS cells that express Sox9 (Fig. 5D). In contrast, A-C/EBP, which inhibits binding of C/EBP family members to specific binding sites by forming a heterodimeric complex (28), reversed the down-regulation of Col2a1 promoter activity caused by C/EBP␤ in a dose-dependent manner (Fig. 5E). A-C/EBP also increased Col2a1 promoter activity in a dose-dependent manner in SW1353 cells (Fig. 5F).
C/EBP␤ Directly Down-regulated Col2a1 through Its Intronic Enhancer-To identify the C/EBP␤ response element in the Col2a1 gene, deletion and mutation analyses were performed. C/EBP␤ was reported to recognize T(T/G)NNGNAA(T/G) as a binding sequence (15), and we found two conserved C/EBP␤binding motifs within 1 kb of the Col2a1 promoter (Fig. 6A). Although Col2a1 promoter activity gradually decreased along with deletion of a series of 5Ј promoter elements, transcriptional repression by C/EBP␤ was observed for all deletion constructs (Fig. 6A). This indicated that the functional C/EBP␤binding element was located within 640 bp of the Col2a1 enhancer.
Next, we generated deletion constructs of the enhancer combined to the shortest promoter, which did not contain C/EBPbinding motifs, to investigate the function of C/EBP␤ on the enhancer element. A reporter assay revealed that only the reporter construct, which included the first intron sequence between ϩ2038 and ϩ2249 bp, was down-regulated by C/EBP␤ (Fig. 6B), suggesting that there were functional C/EBP␤-binding sites in this construct. We identified one conserved C/EBP␤-binding motif within this enhancer sequence. A mutation, from GNAA to GNCC, which is essential for C/EBP␤ binding, was introduced (Fig. 6B). An inhibition of luciferase activity by C/EBP␤ was not observed in this mutation construct (Fig. 6B). These results suggest that C/EBP␤ bound to the binding element located between ϩ2144 bp to ϩ2152 bp in the  Fig. 2 were used. Each value was normalized to 18S in the same sample. The value of each mRNA expression relative to that of LacZ on the 4th day was indicated. Means Ϯ S.D. of duplicates from three independent experiments are shown. *, p Ͻ 0.05 versus LacZ. B, cytoplasmic and nuclear extracts were prepared from ATDC5 cells cultured for 4 days after transfection with adenovirus vectors. Western blot was performed using SOX9 and C/EBP␤ antibodies. The data are representative of two independent experiments performed in duplicate. C, a ChIP assay was performed using ATDC5 cells cultured for 4 days after transfection of adenovirus vectors. SOX9 antibodies were used to immunoprecipitate. Semiquantitative RT-PCR was performed using primers that amplified between ϩ2136 and ϩ2292 bp to detect binding of SOX9 to the Col2a1 enhancer. The data are representative of two independent experiments performed in duplicate.
Col2a1 enhancer. To verify the direct binding of C/EBP␤ to this element in the Col2a1 enhancer, EMSA was performed (Fig.  6C). C/EBP␤ bound strongly to the wild-type probe, but binding to the mutant probe was weak. Nonlabeled wild-type probe inhibited the binding of C/EBP␤ to labeled wild-type probe, but nonlabeled mutant probe could not block it. Supershift was observed by addition of a C/EBP␤ antibody, indicating the specificity of C/EBP␤ binding. A ChIP assay also confirmed that C/EBP␤ bound to the enhancer region of Col2a1 (Fig. 6D). These results suggest that C/EBP␤ directly represses transcriptional activity of Col2a1 by interacting with the enhancer region.
Ectopic Expression of C/EBP␤ Represses the Expression of COL2A1 and SOX9 in ex Vivo Organ Culture-Finally, we performed an ex vivo organ culture of mouse tibias (Fig. 7). Gene expression of tibias that had been transfected with the adenovirus vector expressing LacZ control was equivalent to those of tibias without transfection, indicating that the LacZ-infected tibias were the proper controls for C/EBP␤-infected tibias. Immunofluorescent staining with a ␤-gal antibody showed that transfection using adenovirus vector was efficiently performed, and increased expression of C/EBP␤ induced by the infection was also confirmed. Although the transfection of adenovirus vector expressing LacZ slightly reduced the expression of COL2A1 compared with nontreated tibias, it was strongly reduced in C/EBP␤-infected tibias along with the ectopic expression of C/EBP␤. The expression of COL10A1, RUNX2, and MMP13 was misexpressed through the tibias that were transfected with C/EBP␤, compared with LacZ control. Forced expression of C/EBP␤ may lead the ectopic expression of these genes even in the regions that do not show the morphological hypertrophy because C/EBP␤ is reported as a direct regulator of them. Moreover, the expression of SOX9 was also decreased and restricted to a small upper area of the growth plate by overexpression of C/EBP␤, similar to the expression of COL2A1. Together, these results further confirmed that C/EBP␤ could be involved in regulation of phenotypic conversion from proliferative to hypertrophic chondrocytes by repressing the genes characteristic of proliferative chondrocytes during chondrocyte differentiation.

DISCUSSION
Endochondral ossification is tightly regulated by various factors. During this process, differentiation from proliferative to hypertrophic chondrocytes is accompanied by transition of ECM gene expression from COL2A1 to COL10A1. C/EBP␤ has been reported to promote hypertrophic differentiation by transactivation of cell cycle factor p57 (23), and it stimulates expression of hypertrophic markers such as COL10A1 and MMP13 (24). However, the regulation of genes characteristic of proliferative chondrocytes by C/EBP␤ remains to be elucidated. The present study is the first to show that C/EBP␤ represses transcriptional activity of Col2a1 both directly and indirectly through modulation of SOX9 expression during hypertrophic differentiation of chondrocytes. We also report that Acan is repressed along with SOX9 and that Runx2 is stimulated along with Col10a1.
We used ATDC5 cells as a model system of chondrocyte differentiation, because ATDC5 cells represent equivalent expression of chondrocyte differentiation markers like Col2a1, Acan, and Sox9 with primary chondrocytes ( Table 1). The expression of hypertrophic markers such as Mmp13, Runx2, and Cebpb was higher in primary chondrocytes. Because the primary chondrocytes were obtained from mouse ribs and sternums, they might contain more hypertrophic chondrocytes than ATDC5 cells at the 14th day. Meanwhile, the expression of Col1a1 was higher in ATDC5 cells. ATDC5 cells show chondrogenic and hypertrophic differentiation in nodules, and undifferentiated ATDC5 cells remain fibroblastic morphology around nodules. Subsequently, the cells form numerous nodules, and hypertrophic differentiation of ATDC5 cells progresses (31). Thus, mixing the cells at various differentiation stages may cause higher expression of Col1a1 in ATDC5 cells than in primary chondrocytes. Together, it is considered that ATDC5 cell line is one of the useful models to study chondrogenic differentiation and hypertrophic transformation of chondrocytes in vitro.
A more significant repression of SOX9 protein compared with its mRNA by C/EBP␤ was observed (Fig. 3, A and B). One possible mechanism is that SOX9 protein is degraded posttranscriptionally. In fact, it was reported that ubiquitination of SOX9 protein determined its transcriptional activity (14,32). A significant decrease of SOX9 nuclear protein resulted in downregulation of Col2a1 (Fig. 3C). These findings suggest that indi- rect repression of Col2a1 through interaction with Sox9 is another mechanism by which C/EBP␤ regulates Col2a1 during hypertrophic differentiation of chondrocytes. Although this study did not explore the inhibitory mechanism of Acan during hypertrophic differentiation of chondrocytes, the repression of Sox9 by C/EBP␤ may also be involved in regulating the expression of Acan because SOX9 is known to stimulate ACAN (33).
Although a reporter assay and EMSA revealed that C/EBP␤ directly bound to an intronic enhancer region of Col2a1 to repress its transcription, a point mutation of the C/EBP␤-binding site within the Col2a1 intronic enhancer also decreased Col2a1 luciferase activity (Fig. 6B). Moreover, the Col2a1 reporter was no longer active when the enhancer sequence was deleted despite the existence of binding sites for SOX trio (Fig.  6B). One possible reason is that this sequence may be essential for another unknown transcription factor, which enhances transactivation of Col2a1, and a competition in binding between the unknown factor and C/EBP␤ may be the mechanism of the repression of Col2a1 by C/EBP␤. Further studies will be necessary to examine the detailed mechanism of the interaction between C/EBP␤ and other transcriptional factors.
Meanwhile, it is interesting that a functional C/EBP␤-binding site is located in the Col2a1 enhancer because the site is located nearby the binding elements for SOX trio. SOX9 was reported to interact with the co-activators cAMP response element-binding protein (CBP) and p300 and regulate Col2a1 through histone acetylation around its enhancer region (34). A previous report also demonstrated that Cd-rap, a cartilage-specific ECM molecule, was activated by SOX9 and that CBP and p300 were involved in this mechanism by inhibiting binding of C/EBP␤ to Cd-rap (35). Considering these reports, it is possible that C/EBP␤ represses Col2a1 by competing with SOX9 binding via co-activators like CBP and p300 in the Col2a1 enhancer region.
Regulation of Col2a1 occurs not only during skeletal development, but also in arthritic articular cartilage. C/EBP␤ is known to be induced by pro-inflammatory cytokines such as IL-1␤, IL-6, and TNF-␣, and it is expressed in arthritic cartilage (15,20,22). It has been suggested that C/EBP␤ also plays a role in the repression of COL2A1 in response to inflammatory signals (20). Some other transcription factors including c-Fos, c-Jun, Jun B, and Egr-1, have been reported to be induced by FIGURE 6. C/EBP␤ directly down-regulated transcriptional activity of Col2a1 through its enhancer region. A, a series of 5Ј Col2a1 promoter deletion constructs were generated (left panel). Black and gray boxes indicate SOX9-binding elements reported by previous studies and conserved C/EBP␤-binding motifs, respectively. These reporter constructs were co-transfected with 0.2 g of pCMV-LAP or GFP into RCS cells, and a reporter assay was performed (right panel). Means Ϯ S.D. of duplicates from three independent experiments are shown. *, p Ͻ 0.05 versus GFP. B, reporter constructs were generated that combined Ϫ185 to ϩ73 bp of the Col2a1 promoter that had no conserved C/EBP␤-binding motifs with various enhancer sequences (left panel). Black and gray boxes indicate the SOX9-binding element and the conserved C/EBP␤-binding motif, respectively. A point mutation was also introduced to the C/EBP␤-binding motif within the enhancer region. These reporter constructs were co-transfected with 0.2 g of pCMV-LAP or GFP into RCS cells, and a reporter assay was performed (right panel). Means Ϯ S.D. of duplicates from three independent experiments are shown. *, p Ͻ 0.05 versus GFP. C, EMSA for specific binding of C/EBP␤ to the Col2a1 intronic enhancer. Consensus oligonucleotide (lanes C), wild-type (lanes Wt), and mutant (lanes Mt) probes were incubated with nuclear extract from C/EBP␤-transfected ATDC5 cells. Competition and supershift experiments were also performed. The data are representative of two independent experiments performed in duplicate. D, A ChIP assay for C/EBP␤ using ATDC5 cells cultured for 3 weeks. Semiquantitative RT-PCR was performed using primers as follows: enhancer region of Col2a1 (from ϩ2003 to ϩ2198) and negative control (from Ϫ1891 to Ϫ1611 bp). The data are representative of two independent experiments performed in duplicate.
IL-1␤ and involved in the down-regulation of COL2A1 during OA (36). It was also reported that Egr-1 inhibited interactions between CBP, Sp1, and TATA-binding proteins by interfering with Sp1 binding to the COL2A1 proximal promoter (37). ESE-1 was also reported to repress COL2A1 by Egr-1 through binding to the COL2A1 promoter (38). In Col2a1 regulation by C/EBP␤, C/EBP␤ may also interact with these co-factors, consequently inhibiting initiation of mRNA transcription in response to the inflammatory signal.
This study also showed that exogenous C/EBP␤ stimulated expression of hypertrophic markers such as Col10a1 and Mmp13, but also Runx2, even in the early differentiation stages of ATDC5 cells ( Fig. 2A). RUNX2 directly binds to the COL10A1 promoter and enhances its transcription (39). RUNX2 is also reported to regulate MMP13 and osteocalcin expression synergistically with C/EBP␤ (24,40). Our results imply that C/EBP␤ may enhance expression of Runx2 to further promote hypertrophic differentiation of chondrocytes.
It was reported that hypertrophic chondrocytes also appear in OA articular cartilage (2). Chondrocyte hypertrophy occurs in association with the expression of RUNX2 and MMP13 in a mechanically induced OA model (41), and C/EBP␤ is also observed in the OA model (23,24). However, the reason for the hypertrophic changes of articular chondrocytes in OA is not fully understood. Because pro-inflammatory cytokines such as IL-1␤ are involved in early OA (42), C/EBP␤ may transmit the inflammatory signal to trigger hypertrophic conversion by repressing Col2a1 and stimulating Runx2, Col10a1, and Mmp13 expression.
Finally, ex vivo organ cultures revealed that C/EBP␤ overexpression markedly repressed the genes expressed by proliferative chondrocytes such as COL2A1 and SOX9 (Fig. 7). MMP13 is known to be a major degrading enzyme of COL2A1, and we previously reported that C/EBP␤ is a direct inducer of MMP13 (22). Increase in the expression of MMP13 could be involved in the degradation of COL2A1 in C/EBP␤-infected tibias. Meanwhile, it is possible that the synthesis of COL2A1 was directly repressed by C/EBP␤, as in vitro experiments suggest. Furthermore, it was reported that the proliferative zone of C/EBP␤ knock-out mice was significantly enlarged compared with that of wild-type mice (23). These gain and loss of function experiments also confirmed that C/EBP␤ could be involved in the regulation of phenotypic conversion from proliferative to hypertrophic chondrocytes. However, the difference of phenotype between C/EBP␤ knock-out mice and wild-type mice was slight, and the dwarfism of C/EBP␤ knock-out mouse gradually disappeared as the mice grew up. Presumably other C/EBP families may compensate for the lack of C/EBP␤. In fact, we reported that not only C/EBP␤, but also C/EBP␦ is induced during OA development (21), and C/EBP␦ induced by cytokines also represses the transcriptional activity of Cd-rap (20). Although C/EBP␤ could be a therapeutic target for OA, we have to create an inhibitor that specifically blocks C/EBP family activities in arthritic articular cartilage.
Together, our results suggest that C/EBP␤ directly represses transcription of Col2a1 during differentiation from proliferative to hypertrophic chondrocytes. In addition, C/EBP␤ represses Sox9 to indirectly inhibit transactivation of Col2a1. C/EBP␤ also stimulates expression of Mmp13, Col10a1, and Runx2, and preliminary results obtained in our laboratory suggest that C/EBP␤ also stimulates Indian hedgehog. Combined, these results suggest that C/EBP␤ is a key regulator that triggers the phenotypic conversion from proliferative to hypertrophic chondrocytes by turning off genes characteristic of proliferative chondrocytes and turning on genes associated with hypertrophic chondrocytes. Therefore, C/EBP␤ plays multiple roles in matrix degradation and hypertrophic differentiation of chondrocytes during bone development, as well as in arthritic cartilage. Safranin O staining and immunofluorescent staining were performed to localize COL2A1, COL10A1, SOX9, C/EBP␤, RUNX2, and MMP13. ␤-Galactosidase antibody (␤-gal) was used to confirm the transfection efficiency of adenovirus vectors. DAPI was used as a counterstain. Red, green, and blue bars indicate the proliferative, prehypertrophic, and hypertrophic zones, respectively. Scale bar, 500 m. Histological analysis was repeated at least twice for each sample from six pairs of limbs, respectively.