GADD45β Enhances Col10a1 Transcription via the MTK1/MKK3/6/p38 Axis and Activation of C/EBPβ-TAD4 in Terminally Differentiating Chondrocytes*

GADD45β (growth arrest- and DNA damage-inducible) interacts with upstream regulators of the JNK and p38 stress response kinases. Previously, we reported that the hypertrophic zone of the Gadd45β−/− mouse embryonic growth plate is compressed, and expression of type X collagen (Col10a1) and matrix metalloproteinase 13 (Mmp13) genes is decreased. Herein, we report that GADD45β enhances activity of the proximal Col10a1 promoter, which contains evolutionarily conserved AP-1, cAMP-response element, and C/EBP half-sites, in synergism with C/EBP family members, whereas the MMP13 promoter responds to GADD45β together with AP-1, ATF, or C/EBP family members. C/EBPβ expression also predominantly co-localizes with GADD45β in the embryonic growth plate. Moreover, GADD45β enhances C/EBPβ activation via MTK1, MKK3, and MKK6, and dominant-negative p38αapf, but not JNKapf, disrupts the combined trans-activating effect of GADD45β and C/EBPβ on the Col10a1 promoter. Importantly, GADD45β knockdown prevents p38 phosphorylation while decreasing Col10a1 mRNA levels but does not affect C/EBPβ binding to the Col10a1 promoter in vivo, indicating that GADD45β influences the transactivation function of DNA-bound C/EBPβ. In support of this conclusion, we show that the evolutionarily conserved TAD4 domain of C/EBPβ is the target of the GADD45β-dependent signaling. Collectively, we have uncovered a novel molecular mechanism linking GADD45β via the MTK1/MKK3/6/p38 axis to C/EBPβ-TAD4 activation of Col10a1 transcription in terminally differentiating chondrocytes.

The GADD45 family also includes GADD45␣ and GADD45␥, which regulate apoptosis and differentiation by modulating cascades of stress-responsive mitogen-activated protein kinases (MAPKs), 3 including the p38 and JNK pathways (9). GADD45␤ protein is known to bind to MTK1/MEKK4, a MAP3K, leading to the phosphorylation of MAP2Ks, such as MKK3 or -6 and MKK4 or -7, followed by activation of the p38 and JNK pathways, respectively (10 -12). However, interaction of GADD45␤ with MTK1 has been reported to inhibit or activate MKK7, followed by inhibition or activation of JNK signaling, depending on the cell type and the availability of upstream signals, such as NF-B (13)(14)(15)(16)(17). In chondrocytes, we found that GADD45␤ via JNK activation increased MMP13 promoter activity in synergism with Fra1 or Fra2 together with JunB or JunD (8). Thus, our findings and those of others suggest that GADD45␤ may play an important role in chondrocyte terminal differentiation by modulating both JNK and p38 MAPK signaling cascades.
The MAPK signaling pathways are involved in many cellular processes, such as gene regulation, intracellular metabolism, differentiation, proliferation, mobility, and survival or death (18,19). MAPKs are activated by MAPK kinases (MAP2K) via phosphorylation of conserved threonine and tyrosine residues in their activation loops, followed by phosphorylation of downstream kinases and targets, including transcription factors that regulate a variety of target genes (20). The roles of MAPKs in chondrogenesis have been investigated in vivo as well as in vitro (reviewed in Refs. 21 and 22). Ablation of the MAPK kinase kinase (MAP3K), MEKK4/MTK1, causes skeletal patterning defects in the mouse embryo (23). Moreover, Col2a1 promoterdriven MKK6 transgene overexpression leads to decreased chondrocyte proliferation and delayed terminal differentiation to hypertrophy (24). Constitutive activation of the MAP2K, MEK1, responsible for ERK activation caused persistence of proliferating chondrocytes and delayed hypertrophic maturation (25), whereas in vitro studies using pharmacological inhibition showed that ERK activation is required for chondrocyte hypertrophy (26). Inhibition of the p38 pathway in vitro also leads to impairment of terminal differentiation of chondrocytes to hypertrophy (27,28). These results suggest that the MEK1/ ERK and MKK3/6/p38 pathways are important for regulating chondrogenesis in the embryonic growth plate. However, the precise mechanisms of action of these kinase cascades have not been defined completely because of the complex regulation at different stages of this process, involving several stimuli, as well as the many downstream transcription factors and target genes (5).
Among the transcription factors implicated in regulating genes related to the terminal hypertrophic chondrocyte phenotype, the Runt domain transcription factor, Runx2, or CBFA1, is a critical factor for the expression of Col10a1 and Mmp13 in vivo and in vitro (29 -32). Col10a1 gene activation requires a Runx2 binding site located in its distal promoter (33,34), as well as a non-consensus Runx2 binding site located in the proximal promoter region (35). In addition, the Fra2 knock-out phenotype is characterized by impaired hypertrophic differentiation of chondrocytes in the embryonic growth plate, along with reduced Col10a1 mRNA expression levels (36). A recent study showed in the CCAAT/enhancer-binding protein ␤ (C/ebp␤) knock-out embryo a delay in hypertrophic differentiation in the growth plate, associated with decreased Col10a1 and Mmp13 mRNA (37), a phenotype similar to that of the Gadd45␤ Ϫ/Ϫ embryo (8). Fra2 and C/EBP␤ are members of the bZIP family of transcription factors, characterized by a basic region and a leucine zipper domain (38,39). Transcriptional control by bZIP family members is complex because not only the expression level of each factor but also post-translational modifications determine their effects. Indeed, specific bZIP phosphorylations regulate nuclear translocation, the formation of homo-or heterodimers that determines binding to single or composite sites, and the recruitment of the general transcription machinery to polymerase II transcription initiation complexes (39).
Here, we demonstrate that GADD45␤, which co-localizes with C/EBP␤ and collagen X in growth plate chondrocytes, is targeted to C/EBP␤ and activates the Col10a1 proximal promoter dependent on MKK3/6/p38 signaling. Importantly, we identify the evolutionarily conserved fourth transactivation domain (TAD4) of C/EBP␤ as the target of the GADD45␤enhancing effect on Col10a1 promoter activity. Together, our findings indicate that enhancement of C/EBP␤ transactivation by GADD45␤ is one of the mechanisms underlying Col10a1 transcriptional control in terminally differentiating chondrocytes during skeletal development.
Transfection and Reporter Assays-Transient transfection experiments were performed in ATDC5 cells using Lipofectamine PLUS TM reagent (Invitrogen), as described previously (8). Cells were seeded 24 h prior to transfection in 24-well tissue culture plates at 1.25 ϫ 10 4 cells/cm 2 in Dulbecco's modified Eagle's medium/F-12 containing 5% fetal bovine serum. Transfections were carried out in serum-free medium with a total of no more than 475 ng of plasmid DNA, including 300 ng of luciferase reporter construct per well.
Chromatin Immunoprecipitation (ChIP) Assay-The ChIP-IT Express enzymatic kit (catalog number 53009, Active Motif) was used to perform ChIP assays, according to the manufacturer's instructions with minor modifications. Briefly, ATDC5 cells infected with lentiviral vectors containing siRNA-GFP (siGFP) or siRNA-GADD45␤ (si45␤) (8) were plated on 150-cm dishes and differentiated in Dulbecco's modified Eagle's medium/Ham's F-12 medium containing 5% fetal bovine serum and 1% ITSϩ Premix (catalog number 354352, BD Biosciences) for 1 and 2 weeks. Cells were cross-linked with 1% formaldehyde for 10 min at room temperature; nuclei were isolated, and chromatin was enzymatically sheared for 8 min at 37°C, resulting in chromatin fragments of 250 -1000 bp. Chromatin was precleared by incubation with 25 l of protein G magnetic beads and 5 g of nonspecific (control) rabbit IgG (catalog number 2729; Cell Signaling) for 2 h at 4°C with rotation. After preclearing and removal of the protein G magnetic beads, the lysates were incubated at 4°C for 16 h with 5 g of rabbit anti-C/EBP␤ antibody (catalog number sc-150X, Santa Cruz Biotechnology, Inc. (Santa Cruz, CA)) or normal rabbit IgG (catalog number 2729, Cell Signaling), and 10 l of the precleared chromatin was stored to be used as assay input. After reverse cross-linking of the DNA-protein complexes, the DNA was purified using DNA minicolumns (catalog number 28104, Qiagen). The final DNA preparations were subjected to PCR analysis using 5 l of the eluted DNA and the following set of primers: 5Ј-CCGTTAGGACTTCCCACCAT-3Ј (sense) and 5Ј-CAGGTAAGCCTCGTCTGAGG-3Ј (antisense), spanning from Ϫ114 to ϩ91 of the proximal mouse Col10a1 promoter, containing the C/EBP binding motif. The PCR products were resolved on a 2.5% agarose gel and confirmed by DNA sequencing at the Cornell University Life Sciences Core Laboratories Center. Additional real-time PCR analysis was performed using the Opticon 2 real time PCR detector system (Bio-Rad) in order to better detect changes of the C/EBP␤ binding to the amplified Col10a1 promoter region.
Western Blotting Analysis-After incubation without or with BMP-2 at 100 ng/ml for 15 and 30 min, the siGFP-and si45␤-ATDC5 cells were collected by scraping, and total protein was extracted as described previously (8). The cell lysates were analyzed on Western blots using antibodies against phospho-p38 (catalog number 9211, Cell Signaling) or total p38 (catalog number 9212, Cell Signaling) and ␤-actin (catalog number A5441, Sigma) was used as an additional loading control.
Vectastain Elite ABC kit (Vector Laboratories, Inc.) was used as described by the manufacturer.
Statistical Analysis-Data are reported as mean Ϯ S.D. of at least three independent experiments, each performed in triplicate. Statistical analysis was performed by Student's t test with p values of Ͻ0.05 considered significant.

RESULTS
Conserved Half-sites for AP-1, CREB/ATF, and C/EBP Binding in the Col10a1 Proximal Promoter-Comparative analyses of DNA sequences from multiple species using the University of California Santa Cruz Genome Browser (Human Assembly, March 2006) identified two evolutionarily conserved regions in the Col10a1 promoter, one in the distal promoter region and the other in the proximal region (Fig. 1). The conserved regions contained a Runx2 binding site in the distal promoter, previously described to be functional in vivo (33,51), and a proximal Runx2 binding site, reported to be functional by in vitro assay (35). In addition, we found conserved half-sites for binding AP-1 (TGA(C/G)TCA), CREB/ATF (TGACGTCA), and C/EBP (ATTGCGCAAT) in the proximal promoter ( Fig. 1).
Previously, we demonstrated that GADD45␤ enhanced AP-1 reporter activity with either JunD or JunB in synergism with Fra1 or Fra2 (8). Therefore, we examined the effects of GADD45␤ on gene reporters driven by consensus sequences recognized by AP-1, CREB/ATF, and C/EBP family members, respectively. Fra2/JunD transactivation of the AP-1 reporter was further enhanced by GADD45␤ (Table 1), as expected from our previous report (8). GADD45␤ also enhanced c-Jun-dependent transactivation but not in the presence of Fra2 (Table 1, top). Moreover, GADD45␤ increased ATF2-driven CRE reporter activity and C/EBP␣,  -␤, and -␦ transactivation of the C/EBP reporter by almost 2-fold (Table 1, middle and bottom). Collectively, these in vitro results using consensus reporters indicate that GADD45␤ can synergize with not only Fra2/JunD but also with ATF and C/EBP family members that could interact with conserved sites in the Col10a1 proximal promoter region.

GADD45␤ Promotes C/EBP Transactivation of the Col10a1 and MMP13
Promoters-To examine whether GADD45␤ could participate in transactivating the Col10a1 promoter, we used a Luc reporter construct driven by 4.6 kb of Col10a1 promoter sequence, which produces strong expression in hypertrophic chondrocytes (33,34,40,51,52), and another Luc reporter with 1.6 kb of the MMP13 promoter as a reference control (8). First, cotransfection of Fra2 and GADD45␤, alone or in combination, revealed little or no response of the Col10a1 promoter compared with the positive response of MMP13-Luc ( Table  2, top). ATF2 overexpression alone had no effect, whereas GADD45␤ induced both Col10a1 and MMP13 promoter activation in synergism with ATF2 (Tables 2, third and bottom parts). Importantly, co-transfection of GADD45␤ with C/EBP␣, -␤, or -␦ also enhanced the activities of both promoters compared with transfections with C/EBP alone ( Table 2, third and bottom parts). These results prompted us to investigate the C/EBP family as candidate downstream targets of GADD45␤ in the regulation of Col10a1 gene expression during chondrocyte terminal differentiation.
C/EBP␤ Co-localizes with GADD45␤ and Collagen Type X in the Embryonic Growth Plate-Because C/EBP␤ expression has been linked previously to chondrocyte terminal differentiation   6 kb) promoter activity with ATF2, Fra2, and C/EBP family members ATDC5 cells were transfected with reporter plasmid (300 ng) of Col10a1 (4.6 kb)-Luc (top three parts) or MMP13 (1.6 kb)-Luc (first, second, and fourth parts) and plasmids encoding Fra2 (first part), ATF2 (second part), or C/EBP␣, C/EBP␤, or C/EBP␦ (third and fourth part), alone or together with GADD45␤ in a total of 400 ng/well. Values of firefly luciferase activities are normalized to values of Renilla luciferase activities of pRL-TK and shown as -fold induction compared with control. Means Ϯ S.D. from three or more independent experiments are shown. In the top part, p Ͻ 0.01 was considered as significant.  (37,53), we next examined the in situ expression of the C/EBP family and GADD45␤ in the embryonic growth plate by immunohistochemistry. C/EBP␤ was highly expressed in late proliferative and prehypertrophic growth plate chondrocytes at embryonic day 15.5, consistent with the findings of Hirata et al. (37) at embryonic day 16.5, but also in the late stage hypertrophic zone (Fig. 2, A and D). Furthermore, C/EBP␤ co-localized with both GADD45␤ (Fig. 2, B and E) and collagen type X (Fig. 2, C and F) in the early and late hypertrophic zones. Strong nuclear staining for C/EBP␤ was evident, whereas GADD45␤ protein was present in both cytoplasm and nuclei. Interestingly, C/EBP␣, C/EBP␦, and C/EBP⑀ were also expressed with variable expression levels in this region (data not shown), suggesting that other C/EBP members may also participate in regulating Col10a1 expression. Taken together with our in vitro data, the co-localization of GADD45␤ with C/EBP␤ further supports the view that Col10a1 may be a downstream target of GADD45␤ and C/EBP␤ during chondrocyte maturation in the embryonic growth plate.

Col10a1 Activation by GADD45␤ via C/EBP␤-TAD4
The Synergistic Effect of GADD45␤ and C/EBP␤ Requires a C/EBP␤ Binding Motif in the Col10a1 Proximal Promoter-To identify the DNA elements responsible for Col10a1 transcriptional control by GADD45␤ and C/EBP␤, Col10a1 promoter mutants were constructed and tested in reporter assays. A Col10a1-(Ϫ60/ϩ52)-Luc reporter construct lacking the 5Ј promoter region spanning bp Ϫ2128 to Ϫ60 still responded strongly to GADD45␤ and C/EBP␤ synergism compared with the fulllength Col10a1 promoter (Fig. 3A). However, the activity of the Col10a1-(Ϫ34/ϩ52)-Luc reporter with deletion to Ϫ34 bp was strongly reduced, indicating that the sequence spanning Ϫ60 to Ϫ34 bp contains sites responsive to GADD45␤ and C/EBP␤ (Fig. 3A). We, therefore, made point mutations to delineate the C/EBP␤ binding sites (Fig. 3B). Whereas there was no significant difference between wild type and mutants without co-transfection with GADD45␤ and C/EBP␤, mutations E and F in one C/EBP binding site strongly disrupted the activation of the Col10a1 promoter (80% reduction) in response to C/EBP␤ and GADD45␤. In addition, mutation D in a non-C/EBP consensus site also caused a significant reduction (Fig. 3B).
To investigate if GADD45␤ affected the binding of C/EBP␤ to the endogenous Col10a1 proximal promoter, ChIP assays were performed with ATDC5 cells, in which we previously described the ablation of GADD45␤ expression by stable transduction with lentiviral vectors expressing an siRNA targeted to FIGURE 3. Identification of the Col10a1 promoter element responsive to the synergistic action of C/EBP␤ and GADD45␤. A, ATDC5 cells were co-transfected with deletion mutants of the Col10a1 promoter (Ϫ2128/ ϩ201, Ϫ60/ϩ52, and Ϫ34/ϩ52) in the pGL3 reporter vector (300 ng) and 25 ng of each expression vector encoding mouse C/EBP␤ (C␤) or GADD45␤ (45␤), alone or in combination, in a total amount of 375 ng/well. B, the regions with homologies in the mouse and human Col10a1 promoter region (Ϫ60/Ϫ34 bp) responsive to the synergistic action of C/EBP␤ and GADD45␤ are indicated by a line above. The two half-sites for CREB or AP-1 and the C/EBP half-site are indicated as square boxes. Sites in the mouse sequence that are highlighted in boldface type correspond to substituted sites in the mutant constructs used in the transfections shown in the lower panel. Col10a1 promoter constructs (300 ng) containing mutations in the proximal region were cotransfected with empty vector or 25 ng of expression vector encoding mouse C/EBP␤ or GADD45␤, alone or in combination, in a total amount of 375 ng/well in ATDC5 cells. Luciferase activities were normalized to pRL-TK reporter activities and shown as -fold induction compared with control (empty vector). Experiments were done in triplicate with data shown as means Ϯ S.D.

Col10a1 Activation by GADD45␤ via C/EBP␤-TAD4
GADD45␤ versus the siGFP-matched control cell population (8). As shown in Fig. 4A, no difference was noted in C/EBP␤ binding to its proximal site in the endogenous Col10a1 promoter in siGFP control versus GADD45␤ knockdown ATDC5 cells after culturing them under differentiating conditions for 1 or 2 weeks, and the same result was also verified by real-time PCR analysis (not shown). Similar to the results that we observed previously in micromass cultures of primary mouse rib chondrocytes, the expression of Col10a1 was reduced in the 2-week differentiated ATDC5 cells stably expressing siRNA-GADD45␤ (Fig. 4B). These results indicate that the C/EBP binding site (GCAAT) is responsible for Col10a1 promoter activation by GADD45␤ and C/EBP␤ and also implies the involvement of the flanking region through homo-or heterodimerization (54 -57).
Because the binding of C/EBP␤ to the endogenous Col10a1 proximal promoter was not affected by GADD45␤ knockdown, we then investigated whether GADD45␤-dependent phosphorylation events could modulate C/EBP␤ activity independent of DNA binding. Because C/EBP␤ is known to be a direct substrate of p38 kinase (58 -60), we analyzed whether p38 activation was differentially modulated in siGFP control versus si45␤-ATDC5 cells after BMP-2 treatment. Indeed, although BMP-2 treatment enhanced p38 phosphorylation in siGFP control cells, it failed to do so in the GADD45␤ knockdown cells (Fig.  4C). Taken together, the above results are consistent with GADD45␤-mediated Col10a1 modulation working through p38 to enhance the transcriptional activity of DNA-bound C/EBP␤. GADD45␤ Enhances C/EBP␤ Activity via MTK1/MKK3/6/ p38 Signaling-To better delineate the mechanisms underlying the enhancement of C/EBP␤ transactivation by GADD45␤, we compared the activities of wild type, dominant negative, or dominant active kinase mutants in co-transfections with the Col10a1-(Ϫ60/ϩ52)-Luc reporter. First, wild type MTK1 (Mw) overexpression further enhanced the activation of the Col10a1 reporter co-transfected with GADD45␤ and C/EBP␤ expression plasmids, whereas the kinase-dead MTK1K/R mutant (Mm) did not (Fig. 5A). To identify the required MAP2K that delivers the signal from GADD45␤ to C/EBP␤, the MKK6KD mutant or the MKK3ala, MKK4ala, and MKK7ala mutants with phosphoacceptor amino acid threonine or tyrosine changed to alanine were used in reporter assays. Interestingly, overexpression of MKK6KD (M6K; Fig. 5B) or MKK3ala (data not shown) reduced the ability of GADD45␤ to enhance C/EBP␤-mediated Col10a1 transactivation in a dose-dependent manner, whereas overexpression of MKK4ala (data not shown) or MKK7ala (M7a; Fig.  5B) did not. Furthermore, constitutively active mutants, MKK6glu (M6g; Fig. 5B) and MKK3 (M3; Fig. 5D), but not MKK7glu (M7g; Fig. 5B), mimicked the GADD45␤-dependent enhancement of C/EBP␤ transactivation of the Col10a1 promoter. These results suggest that GADD45␤ interaction with MTK1 promotes downstream phosphorylation events via MKK6 or MKK3 that activate C/EBP␤. In confirmation of this interpretation, the dominant negative phosphorylation site mutant, p38␣apf (p38m), but not the JNKapf mutant (Jnkm), reduced the GADD45␤-dependent enhancement of C/EBP␤ activity on the Col10a1 promoter (Fig. 5C). Finally, overexpression of MKP1, a specific MAPK phosphatase (47), diminished C/EBP␤ transactivation of the Col10a1 promoter induced by GADD45␤ or MKK3 (M3) (Fig. 5D). Together, these results provide strong evidence that phosphorylation events in the MTK1/MKK3/6/p38␣ cascade are critical for GADD45␤ signaling to C/EBP␤-and GADD45␤-dependent transactivation of Col10a1 by C/EBP␤.

C/EBP␤ Threonine 235 Mutation Disrupts Its Activation by GADD45␤, as Assessed with a C/EBP Consensus-Luc Reporter but Not in the Context of the Col10a1
Promoter-p38 is known to phosphorylate C/EBP␤ at threonine 235 (60). Unexpectedly, cotransfection of the C/EBP␤ T235A mutant did not significantly affect Col10a1-(Ϫ60/ϩ52) reporter activity (Fig. 6A), whereas it did disrupt the GADD45␤-enhanced transactivation of the pCEBP-Luc reporter (Fig. 6B). In contrast, the C/EBP␤ mutants, Y274F and L320R, which affect DNA binding and protein dimerization, respectively (55,61), decreased the GADD45␤-dependent activities of both Col10a1 and C/EBP reporters (Fig. 6, A and B). These results indicate that the promoter context is important for the permissive effects of GADD45␤ on transactivation of the Col10a1 promoter by C/EBP␤ and that the protein dimerization domains of C/EBP␤ are important for driving its activation by GADD45␤. A, siGFP and si45␤-ATDC5 cells were differentiated for 1 week (1w) and 2 weeks (2w). Chromatin was cross-linked and enzymatically sheared, and after reverse cross-linking of the DNA-protein complexes, the precleared lysates were incubated with antibodies against C/EBP␤ (ϩ) or normal rabbit IgG (Ϫ) overnight at 4°C. The mouse Col10a1 promoter region was PCR-amplified using primers spanning from Ϫ114 to ϩ91, and the PCR products were resolved on a 2.5% agarose gel. Data are representative of two independent experiments performed in duplicate. B, total RNA isolated from 2-week differentiated siGFP-and si45␤-ATDC5 was analyzed by real-time PCR. Each value was normalized to ␤-actin in the same sample and shown as mean Ϯ S.D. *, p Ͻ 0.05. C, the siGFP-and si45␤-ATDC5 cells were incubated with 100 ng/ml BMP-2 for the indicated times, and total lysates were analyzed on Western blots using antibodies against phospho-p38 (P-p38), total p38 (p38), and ␤-actin. MARCH 12, 2010 • VOLUME 285 • NUMBER 11

Col10a1 Activation by GADD45␤ via C/EBP␤-TAD4
SY, blunted the GADD45␤-dependent enhancement of C/EBP␤driven Col10a1 promoter activity (Fig. 7D). Co-transfection of mLAP (Fig. 7B) or hLAP (Fig. 7D) together with GADD45␤ significantly enhanced the activity of the Col10a1 reporter above that observed with either species of C/EBP␤ alone. Wild type murine C/EBP␤ had a stronger transactivating effect than the human homologue in these experiments due to the latter's reduced expression in ATDC5 murine cells (data not shown). Collectively, these experiments indicate that GADD45␤ enhances C/EBP␤ transactivation of the Col10a1 promoter via the evolutionarily conserved C/EBP␤-TAD4 domain.

Molecular Basis of GADD45␤ Enhancement of C/EBP␤-dependent Col10a1
Promoter Transactivation-In this study, we showed that GADD45␤ regulates the expression of genes involved in terminal chondrocyte differentiation, Col10a1 and MMP13, by modulating C/EBP␤ activity in a manner dependent on the functional binding sites of each promoter and the molecular interplay among signaling molecules and transcription factors in ATDC5 cells (Tables 1 and  2). Although GADD45␤ modulates transactivation of consensus promoters by a number of bZIP class transcription factors, including AP-1, CREB/ATF, and C/EBP family members (Table 1), the Col10a1 promoter responded specifically to C/EBP family members in synergism with GADD45␤ ( Table 2). Site-directed mutagenesis of the conserved sites in the Col10a1 promoter showed that the proximal C/EBP␤ site was functional for the GADD45␤/C/EBP␤ response, and ChIP assays revealed C/EBP␤ binding to this site in the endogenous Col10a1 promoter. As we expected and consistent with ourdataindicatingthatGADD45␤dependent signaling modulates C/EBP␤ activity by phosphorylation, the binding of C/EBP␤ was not affected by siRNA-Gadd45␤ knockdown, whereas Col10a1 mRNA levels were decreased significantly in those cells (Fig. 4, B and C).

GADD45␤ Regulation of Genes Involved in Terminal Chondrocyte Differentiation and the Role of C/EBP␤-GADD45␤
modulates stress-responsive MAPKs, such as p38 and JNK, by binding to MTK1, which activates MAP2Ks (10,11,63,64). Our results indicate that GADD45␤ acts via the MTK1/ MKK3/6/p38 cascade to enhance C/EBP␤ activation of the Col10a1 promoter (Fig. 5A), further demonstrating the important role of this pathway in chondrocyte terminal differentiation. Interestingly, mutant mice harboring the MTK1 kinase-inactive mutant allele, which is incompetent for p38 signaling, display abnormalities in axial skeletogenesis, although the role of MTK1 in terminal chondrocyte differentiation was not addressed (23). In addition, transgenic mice expressing a Col2a1 promoter/enhancerdriven, constitutively active MKK6 mutant have shorter limbs associated with delayed chondrocyte terminal differentiation and reduced Col10a1 expression (24), suggesting that excessive MKK6 signaling inhibits chondrocyte proliferation and subsequent differentiation. In contrast, enforced MKK6 activation of the p38 pathway in synovial fibroblasts induces terminal chondrocyte differentiation markers, including type X collagen, and mineralization (65). Moreover, in vitro studies with specific p38 inhibitors also substantiate the importance of p38 signaling for chondrocyte differentiation to hypertrophy (27,28). Due to these contradictory observations, the role of p38 signaling in chondrocyte terminal differentiation has not been definitively established. Several variables may influence p38-dependent phenotypic changes because excessively high or low p38 activity could differentially modulate hypertrophic processes in the

Col10a1 Activation by GADD45␤ via C/EBP␤-TAD4
8404 JOURNAL OF BIOLOGICAL CHEMISTRY growth plate depending on the availability of transcription factors that could activate or repress target genes (21). Indeed, after treatment with BMP-2, a positive activator of both GADD45␤ and p38, si45␤-ATDC5 cells exhibited defective p38 phosphorylation, indicating that p38 participates in the signaling pathway that goes from GADD45␤ to Col10a1.
In this study, we showed that bZIP family members ATF2, Fra2, and C/EBP␤, known to be involved in endochondral ossification processes (53,66), are downstream GADD45␤ targets (Tables 1 and 2). ATF2 mutant mice present with skeletal abnormalities in proliferating growth plate chondrocytes (67). In contrast, Fra2 knock-out mice display delayed hypertrophic chondrocyte differentiation characterized by decreased Col10a1 gene expression (36). However, in data not shown in that study, Fra2 failed to bind to the distal AP-1 sites (36), which may serve as silencers in vivo (51). Here, we demonstrate that, unlike the MMP13 promoter, the Col10a1 promoter does not respond to Fra2, either with or without GADD45␤, further confirming that the 4.6-kb Col10a1 promoter is not a Fra2 target ( Table 2, top). Interestingly, the C/EBP␤ promoter is regulated directly by Fra2 in other cell types (68), and our finding that C/EBP␤ activates the Col10a1 promoter via the proximal GADD45␤-C/EBP responsive element (GCRE)1 (Fig. 3B) suggests that Fra2-induced C/EBP␤ protein expression, together with GADD45␤, may account for Fra2-regulated Col10a1 expression in the embryonic growth plate (36).
The Gadd45␤ and C/ebp␤ knock-out mice display similar phenotypic changes in the embryonic growth plate, characterized by a delay in terminal chondrocyte differentiation along with decreased expression of genes linked to this process, including Col10a1 and Mmp13 (8,37,69). In addition, the embryonic limbs of C/ebp␤ knock-out mice display impaired osteogenesis (53). GADD45␤ acts early in the MAPK pathway as a positive switch to facilitate the up-regulation of downstream target genes, such as Col10a1 and MMP13, by activation of Fra2/JunD, ATF2, or C/EBP (Table 2). In addition, GADD45␣ and GADD45␥ also activate the p38 pathway to induce Col10a1 and MMP13 promoter activities in cooperation with C/EBP family members in ATDC5 cells (data not shown). Further, GADD45␥ is highly expressed in hypertrophic growth plate chondrocytes (data not shown), suggesting the possibility of partially compensatory or redundant systems for GADD45␤ signaling to C/EBP proteins and their regulation of genes related to chondrocyte terminal differentiation.
We identified the highly conserved TAD4 of all C/EBP family members (70 -72) as the C/EBP␤ domain required for the GADD45␤-dependent enhancement (Fig. 7). TAD4 is known to be essential for direct binding of C/EBP-TAD to the TAF2B and TBP components of the general RNA polymerase II transcriptional apparatus (70) and also for interaction of C/EBP␤ with CBP and p300 transcriptional coactivators and SWI/SNF chromatin remodeling factors (73)(74)(75)(76)(77). Thus, GADD45␤ activation of C/EBP␤-TAD4 function via MTK1/MKK3/6/p38 signaling suggests a general mechanism for facilitating processes driven by transcription factors containing TAD4-like domains, such as c-Jun (72). C/EBP␤-TAD4 activation by GADD45␤ via p38 kinase or downstream kinases, such as CK2 (78,79), could conceivably involve more than one mechanism, including (i) phosphorylation of C/EBP␤ to recruit TAF2B and TBP, (ii) phosphorylation of TAF2B or TBP to induce interactions between C/EBP␤-TAD and the polymerase II transcription complexes, (iii) C/EBP␤-TAD4 phosphorylation to enhance its recruitment of coactivator or chromatin remodeling complexes, or (iv) interactions of C/EBP␤-TAD and TAF2B or TBP to produce a conformational change making this complex a better kinase substrate. Each of these possibilities could explain our observations showing that the conserved C/EBP-TAD4 FIGURE 7. GADD45␤ enhances C/EBP␤ activity via TAD4. A, mouse C/EBP␤ deletion constructs fused with the GAL4-DBD (black bar) are shown. Transactivation domains (TAD1, -2, -3, -4), also known as activation domain modules (ADM1, -2, and -3 for TAD2, -3, and -4, respectively), and repression domains (RD1 and -2) are represented as gray boxes and dashed boxes, respectively. Basic region (BR) and leucine zipper region (LR) are indicated (57). Reporter assays were carried out in ATDC5 cells co-transfected with 300 ng of pRL-TK containing the Gal4 binding site and 50 ng of the GAL4-fused mC/EBP␤ constructs, alone or in combination with 25 ng of GADD45␤ expression vector. Luciferase activities were normalized to pRL-SV40 reporter activities and shown as -fold induction compared with control (empty vector). B, wild-type mouse C/EBP␤ (mLAP); mLAP⌬SacII-(53-120), and mLIP constructs are represented. Col10a1-(Ϫ60/ϩ52) transactivation was analyzed in co-transfection experiments with 25 ng of mLAP, mLIP, or mLAPDSacII, alone or in combination with GADD45␤ expression plasmid. Luciferase activities were normalized to pRL-TK reporter activities and shown as -fold induction compared with control (mLAP). C, human and mouse TAD1, -2, -3, and -4 homologies among the C/EBP family. SY and FL amino acids are highlighted in boldface type. D, full-length human C/EBP␤ (hLAP*), hLAP*⌬SacII , and point SY and LF mutants are depicted. The Col10a1-(Ϫ60/ϩ52) transactivation was analyzed in co-transfection experiments with 25 ng of hLAP*, hLAP*⌬SacII, or SY or LF mutant, alone or in combination with GADD45␤ expression plasmid. Luciferase activities were normalized to pRL-TK reporter activities and shown as -fold induction compared with control (hLAP*). Means Ϯ S.D. of triplicates from three independent experiments are shown for A, B, and D. through the MTK1/MKK3/6/p38 axis may promote chondrocyte terminal differentiation via p38-dependent phosphorylation of the C/EBP␤-TAD4 to promote protein-protein interactions with CBP-p300 or TAF2B/TBP to recruit the polymerase II complex and enhance transcriptional activation of Col10a1. GADD45␤-C/EBP-responsive element 1 (GCRE) is the binding site for C/EBP␤ that we identified in the Col10a1 promoter (TCATGCAATA). TF, transcription factor.

Col10a1 Activation by GADD45␤ via C/EBP␤-TAD4
domain is required for the GADD45␤-enhanced transactivation of Col10a1 by C/EBP␤ in ATDC5 cells.
We also found that the repressor domain, RD1, blocked both the GADD45␤-enhancing effects and the activity of the fulllength C/EBP␤ in the Gal4 system (Fig. 7), in accord with a prior report showing that RD1 inhibits the TAD activity of C/EBP␤ (62). The latter authors also hypothesized that DNA binding through the bZIP domain could promote a structural change that interrupts the putative RD1-TAD interaction to unmask the TAD domain, whereas the TAD domain in a non-DNA bound protein would remain inaccessible (62). Accordingly, DNA binding by C/EBP␤ and/or its homo-or heterodimerization may be required to expose its TAD, allowing for its activation by GADD45␤ and leading to Col10a1 promoter induction. On the other hand, RD2 is also important for C/EBP-mediated promoter responses to a number of signaling pathways, including RAS, ERK, GSK3␤, and p38 (38,39,80), which disrupt RD2 interactions with either the C/EBP DNA-binding or TAD domain, leading to derepression of C/EBP␤ function (62,81). However, although the RD2 threonine 235 is a p38 substrate, mutating that residue did not affect Col10a1 promoter activity, whereas it did disrupt C/EBP-Luc induction (Fig. 6). These results highlight the importance of dynamic intra-and intermolecular interplay in the context of the Col10a1 promoter. This interpretation is supported by our finding that GADD45␤ knockdown in differentiating ATDC5 cells did not affect the binding of C/EBP␤ to the endogenous Col10a1 promoter. Armed with these new findings, future work will involve elucidating the mechanism of action of TAD4 in regulating the transcription of hypertrophic target genes in response to GADD45␤ signaling.
Taken together, we have shown co-localization of GADD45␤ and C/EBP␤ in hypertrophic chondrocytes and GADD45␤-dependent activation of C/EBP␤-TAD via the MTK1/MKK3/6/ p38 signaling pathway, leading to Col10a1 promoter activation at the level of its proximal GCRE1 site (Fig. 8). We speculate that GADD45␤ promotes or maintains chondrocyte hypertrophy by enhancing C/EBP␤-TAD function to coordinately induce the expression of terminal differentiation-related genes, including Col10a1 and MMP13. Thus, our results provide new insights into the molecular mechanisms underlying the regulation of the Col10a1 promoter that are applicable to other genes associated with chondrocyte terminal differentiation in the embryonic growth plate.