Oxygen Tension Regulates Chondrocyte Differentiation and Function during Endochondral Ossification*

Cartilage functions at a lower oxygen tension than most other tissues. To determine the role of oxygen tension in chondrocyte differentiation and function, we investigated the influence of oxygen tension in the pluripotent mesenchymal cell line C3H10T1/2 and 14.5E mice embryo forelimb organ culture. 10T1/2 cells and embryo forelimbs were cultured under normoxia (20% O2) or hypoxia (5% O2) in the presence of recombinant human bone morphogenetic protein 2. To elucidate the mechanism by which oxygen tension influences chondrocyte differentiation, the Smad pathway was examined using Smad6 overexpression adenovirus and Smad6 transgenic mice embryo forelimbs. The p38 MAPK pathway was examined using dominant-negative MKK3 and FR167653, a specific p38 MAPK inhibitor. The transcriptional activities of Sox9 and Runx2 were also investigated. Hypoxia promoted bone morphogenetic protein 2-induced glycosaminoglycan production and suppressed alkaline phosphatase activity and mineralization of C3H10T1/2. Thus, hypoxia promoted chondrocytic commitment rather than osteoblastic differentiation. In the mice embryo forelimb organ culture, hypoxia increased cartilaginous matrix synthesis. These effects were primarily mediated by p38 MAPK activation, independent of Sox9. Hypoxia inhibited Col10a1 (type X collagen α1) expression via down-regulation of Runx2 activity by Smad suppression and histone deacetylase 4 activation. In conclusion, hypoxia promotes chondrocytic differentiation and cartilage matrix synthesis and suppresses terminal chondrocyte differentiation. These hypoxia-induced phenomena may act on chondrocytes to enhance and preserve their phenotype and function during chondrocyte differentiation and endochondral ossification.

A number of pathophysiological findings suggest that a correlation exists between hypoxia and chondrogenesis. For example, articular cartilage is an avascular tissue that functions at an oxygen tension that is lower than that of most other tissues. Articular cartilage derives both its nutrition and oxygen supply by diffusion from the synovial fluid and the subchondral bone. It has been estimated that articular chondrocytes in the deepest layers may have access to no more than 1-6% O 2 (1)(2)(3)(4)(5)(6). Furthermore, although the majority of mammalian cells derive their energy by using oxygen for mitochondrial oxidative phosphorylation (7), few mitochondria are present in articular chondrocytes (8). Carbohydrate breakdown in articular cartilage is dominated by the conversion of glucose to lactate via the Embden-Meyerhof-Parnas pathway (9 -11) that consumes no O 2 . Similarly, during the endochondral ossification processes that occur in the growth plate, chondromodulin-1, an endogenous inhibitor of neovascularization, is highly expressed by chondrocytes. Of note, most of the growth plate is avascular (12). Recently, in an in vivo experiment, it was found that hypoxia-inducible factor 1, which appears to be one of the major regulators of the hypoxic response, is essential for chondrocyte growth arrest and survival (13). Therefore, hypoxia is considered to be a key factor for the growth and survival of chondrocytes. Chondrocytes are derived from undifferentiated mesenchymal cells that have the potential for multidirectional differentiation (14 -16). Bone morphogenetic protein (BMP) 2 -2 promotes the chondrocytic differentiation of undifferentiated mesenchymal cells (17)(18)(19)(20)(21). BMP-2 activates Smad1-Smad5-Smad8, which subsequently associates with Smad4, relocates to the nucleus, and regulates the expression of target genes (22,23). However, the influence of oxygen tension on BMP-Smad signaling remains to be elucidated. In addition to the Smad signaling pathway, p38 mitogen-activated protein kinase (p38 MAPK) is also activated by BMP-2 (24,25). Several other cytokines (26,27) or stress signals (28 -30) can also activate the p38 MAPK pathway. Of particular interest, hypoxia has been found to be one of the stresses that can phosphorylate and activate p38 MAPK (31). Although p38 MAPK has been shown to be implicated in the regulation of chondrogenesis (32)(33)(34), the precise role of p38 MAPK in chondrogenesis remains elusive.
During endochondral ossification, chondrocytes undergo hypertrophy and secrete an extracellular matrix that becomes mineralized and allows vascular invasion and osteoblast differentiation (35,36). To date, Runx2 and Runx3 (runt-related transcriptional factors 2 and 3) have essential roles in inducing chondrocyte hypertrophy; furthermore, these two interact closely (37)(38)(39)(40). In addition, Runx2 is probably a direct tran-scriptional activator of chondrocyte maturation. It binds in vivo to multiple recognition sites in the Col10a1 promoter and activates Col10a1 reporter constructs through these elements in vitro (41). Runx2 is also necessary for osteoblast differentiation (42,43). Runx2 transcriptional activity has been shown to be positively regulated both by the Smad pathway and by the p38 MAPK pathway (44 -46). HDAC4, a member of the class II histone deacetylases (HDACs), is expressed in prehypertrophic and hypertrophic chondrocytes and regulates chondrocyte hypertrophy and endochondral bone formation by interacting with Runx2. Recently, HDAC4 has been found to inhibit Runx2 expression by repressing its positive feedback mechanism and inhibiting Runx2 activity (47). Although hypoxia has been shown to down-regulate Runx2 expression in osteoblasts (48), it remains unclear whether Runx2 is involved in the regulation of chondrocyte differentiation and function by oxygen tension.
In the present study, we assessed the influence of oxygen tension on chondrocytic differentiation and cartilage matrix synthesis in the C3H10T1/2 pluripotent mesenchymal cell line and the N1511 murine chondrocyte cell line. We used the mouse embryo organ culture system with wild type mice and Smad6 transgenic mice that we had previously generated. In this paper, we show that hypoxia promotes chondrocytic commitment and cartilage matrix production via the p38 MAPK pathway but that hypoxia inhibits terminal differentiation via the Smad pathway and HDAC4 activation.

EXPERIMENTAL PROCEDURES
Cell Culture and Analysis for Chondrocytic, Osteoblastic, and Adipocytic Differentiation-C3HT101/2 cells were obtained from RIKEN (Saitama, Japan) and were cultured in Dulbecco's modified Eagle's medium (Invitrogen). Since it has been estimated that articular chondrocytes in the deepest layers may have access to no more than 1-6% O 2 , we selected 5% O 2 as the hypoxic environment for the cells. The cells were incubated at 37°C under 5% CO 2 and 20% O 2 (normoxia) or 5% CO 2 and 5% O 2 (hypoxia) with or without recombinant human BMP-2. BMP-2 stimulation was added at 90% confluence. To evaluate chondrocytic differentiation, C3H10T1/2 cells were fixed with 10% formalin, washed with distilled water and 0.1 N HCl, and then stained with Alcian blue solution, Alcian blue 8GX (Sigma). We also cultured N1511 murine chondrocytes (49) and stained with Alcian blue. For the quantitative analysis of chondrocytic differentiation, the absorbance of Alcian blue dyes bound to sulfated glycosaminoglycan (GAG) was measured (50). All experiments were performed in triplicate. To determine osteoblastic differentiation, the C3H10T1/2 cells were fixed with 10% formalin, washed with PBS (pH 7.4) twice, and incubated with alkaline phosphatase (ALP) substrate solution, 0.1 mg/ml naphthol AS-MX (Sigma), and 0.6 mg/ml fast violet B salt (Sigma) in 0.1 M Tris-HCl, pH 8.5. The activity of ALP was also measured using an ALP test kit (Wako, Osaka, Japan) according to the manufacturer's instruction. To evaluate matrix mineralization, the cultures were stained with alizarin red solution (Sigma) (pH 6.0) and incubated in 100 mM cetylpyridinium chloride for 1 h to solubilize and release calcium-bound alizarin red S into solution. The absorbance of the released alizarin red S was measured (51). To determine adipocytic differentiation, the C3H10T1/2 cells were fixed in 10% formalin, washed with diluted water and 60% isopropyl alcohol, and stained with Oil red O (Sigma) solution. Quantitative analysis for oil droplet was performed as described previously (52). To measure the value of absorbance for alcian blue, alizarin red, and oil red O relative to cell density, the absorbance data were normalized by total DNA content. Total DNA content was extracted using a DNeasy tissue kit (Qiagen, Düsseldorf, Germany) and measured. Cell proliferation analysis was also performed. C3H10T1/2 and N1511 cells were cultured in flat bottomed 96-well microplates at a concentration of 1 ϫ 10 4 cells/ml in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. After 7 days, cell viability was assessed by cell proliferation assay system kit (Takara Bio Inc., Otsu, Japan), using the sulfonated tetrazolium salt 2-(4-iodophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium monosodium (WST-1).
Organ Culture of Embryonic Limb Explants-The forelimbs from embryonic day 14.5 embryos of ICR wild type mice (Charles River, Osaka, Japan) and Smad6-overexpressing transgenic mice (53) were stripped of skin and muscles. They were then cultured for 5 days in BGJ-B medium (Invitrogen) with 1% penicillin/streptomycin (Invitrogen) and 0.1% fetal bovine serum in organ culture dishes under humidified conditions as previously reported (54 -56). Cultures were supplemented with 500 ng/ml BMP-2 under normoxia or hypoxia. After 5 days, limb explants were fixed overnight in 4% paraformaldehyde at 4°C and embedded in paraffin. Serial 3-m-thick sections from the wild type mice (n ϭ 6) and the transgenic mice (n ϭ 5) were processed for safranin O staining and in situ hybridization. Briefly, sections were stained with hematoxylin (Sigma) and fast green (Merck) to identify cells and with safranin-O (Sigma) to identify GAGs. (n ϭ 12). *, p Ͻ 0.05; N.S., not significant. G, C3H10T1/2 cells were cultured with BMP-2 for 7 days, total RNA was extracted, and Northern blotting analysis was done for Col2a1 genes (10 g of total RNA/lane). Expression of 18 S was used to control the amount of RNA. H, C3H10T1/2 cells were cultured with 500 ng/ml BMP-2 for 7 days. Total RNA was extracted and subjected to Northern blotting analysis for the Col10a1 gene. OCTOBER 13, 2006 • VOLUME 281 • NUMBER 41

JOURNAL OF BIOLOGICAL CHEMISTRY 31081
Computer-assisted histological analysis for the proportion of the GAG production area per whole radius was performed using a Nikon ECLIPSE E1000 microscope with a Plan Apo objective, combined with a Nikon DXM 1200 Digital Camera (Tokyo, Japan) and WinRoof image processing software (Mitani Corp.) for Windows. Digitized pictures taken for each radius were analyzed to calculate the ratio of the area stained with safranin-O per whole radius. In situ hybridization for Col10a1 was done as described previously (53,57). The proportional length of the hypertrophic positive signal of Col10a1 with respect to the whole hypertrophic chondrocyte zone along the midline was calculated. The care and handling of the animals and the procedures used in this study were in accordance with the guidelines of and were approved by the Osaka University Medical School Animal Care and Use Committee.
Antibodies and Reagents-Anti-phospho-Smad 1/5/8 monoclonal antibody, anti-phospho-p38 MAPK polyclonal antibody, and anti-p38 MAPK monoclonal antibody were purchased from Cell Signaling Technology (Beverly, MA). Anti-Smad 1/5 polyclonal antibody was purchased from Calbiochem. Anti-Sox9 antibody was purchased from Chemicon (Temecula, CA). Anti-HDAC4, anti-␤-actin, and antiproliferating cell nuclear antigen polyclonal or monoclonal antibody was purchased from Santa Cruz Biotechnology, Inc. Reporter Constructs and Luciferase Reporter Assay-Four tandems of 48-bp chondrocyte-specific enhancer segments of type II collagen ␣1 (Col2a1) were synthesized as previously reported (58) and inserted into the PGL3 promoter vector (Promega), 4Col2E-Luc. Six tandems of the Runx2 binding site were also inserted into the PGL3 promoter vector (Promega), 6Runx2E-Luc. Reporter assays were performed by transient transfection of 0.4 g of the PGL3-promoter vector (4Col2E-Luc or 6Runx2E-Luc) and 0.01 g of the TK-Renilla luciferase construct (TK Renilla) (Promega). Luciferase activity was measured using a Dual Luciferase assay kit (Promega) and luminometer (Berthold Technologies, Bad Wildbad, Germany) and normalized by determining the activity of Renilla luciferase. All experiments were performed in triplicate.
RNA Interference-RNA interference was done using commercially synthesized siRNA (Qiagen, Düsseldorf, Germany) and used as described in the protocols provided by the manufacturer. Cells were treated with siRNA to a final concentration of 10 M. The siRNA duplex sequence targeting the HDAC4 protein was aauguacgacgccaaagautt (sense strands), as previously described (59). Control siRNA consisted of siRNA targeted against luciferase (Dharmacon, Lafayette, CO).
Phosphorylation of Sox9-Serine/threonine phosphorylation of Sox9 was analyzed by affinity chromatography using a phosphoprotein purification kit (Qiagen). Cells were lysed using the lysis buffer provided in the kit. The extracted protein was applied to a phosphorylation purification column, and phosphorylated protein was eluted. Unphosphorylated protein was obtained from the flow-through fraction. These samples were blotted with anti-Sox9 antibody.
Western Blot Analysis-Western blot analyses were performed using whole cell lysates. To detect HDAC4 protein, nuclear extracts were obtained as previously reported (60). The MC3T3-E1 cell (mouse osteoblastic cell) was used as a control. The blots were first incubated with appropriate antibodies and then with horseradish peroxidase-coupled anti-mouse or rabbit IgG antibodies (Amersham Biosciences). For the blots, 20 g of each sample was applied.

Oxygen Tension Regulates Chondrocytes
Reverse Transcription PCR Analysis-First-strand cDNA was synthesized using SuperScript II RNase H Ϫ reverse transcriptase (Invitrogen). The PCR was performed using Ex Taq (Takara Bio Inc., Otsu, Japan). The primers for the Runx2 gene were the same as those used for Northern blotting. The GAPDH primers included the forward primer (5Ј-TGAACGG-GAAGCTCACTGG-3Ј) and the reverse primer (5Ј-TCC-ACCACCCTGTTGCTGTA-3Ј).
Quantitative Real Time PCR Analysis-We obtained cDNA by reverse transcription as mentioned above and proceeded with real time PCR using the Roche Applied Science Light Cycler system. The SYBR Green assay, in which each cDNA sample was evaluated in triplicate 20-l reactions, was used for all target transcripts. Expression values were normalized to GAPDH. The primers for the Runx2 and GAPDH genes were the same as above. The Sox9 primers were as follows: forward, 5Ј-ATGAATCTCCTGGACCCC-TT-3Ј; reverse, 5Ј-TTGGGGAAGGTGTTCTCCT-3Ј.
Statistical Analysis-All data are expressed as mean Ϯ S.E. Differences between groups were assessed using Student's t test, and differences among three or more groups were assessed by post hoc testing. A p value of Ͻ0.05 was considered statistically significant.

Hypoxia Promotes Chondrocytic Differentiation and GAG Production, whereas It Suppresses Osteoblastic Differentiation and Chondrocyte Terminal Differentiation in C3H10T1/2 Cell
Culture-In C3H10T1/2 cell culture, BMP-2 induced GAG production in a dose-dependent manner (Fig. 1A). At every BMP-2 concentration tested, hypoxia clearly increased GAG content additively with BMP-2. Hypoxia-induced GAG synthesis was also found in the N1511 murine chondrocyte culture (Fig. 1B). On the other hand, low oxygen tension clearly suppressed ALP activity and alizarin red staining of C3H10T1/2 cells (Fig. 1, C and D), suggesting that hypoxia suppressed BMP-2-induced osteoblastic differentiation. Adipocytic differentiation, as identified by oil red O staining, was not affected by oxygen tension (Fig. 1E). After cell density became confluent, there was no difference of the cell proliferation between normoxia and hypoxia at 7 days in C3H10T1/2 cells and N1511 cells (Fig. 1F). The type II collagen ␣1 (Col2a1) gene, a well characterized, specific marker of commitment to the chondrogenic lineage, was up-regulated by BMP-2 in a dose-dependent manner (Fig. 1G) as well as by hypoxia at each BMP-2 concentration tested. The mRNA expression of the type X collagen ␣1 (Col10a1) gene, a well characterized, specific marker for chondrocyte terminal differentiation (61,62), was suppressed by hypoxia (Fig. 1H), in contrast to Col2a1 mRNA expression.
Hypoxia-induced GAG Production Is Mediated by the p38 MAPK Pathway rather than the Smad Pathway-To investigate the intracellular signal transduction mechanisms responsible for hypoxia-induced chondrocyte differentiation, we first examined the activities of the Smad and the p38 MAPK pathways. As shown in Fig. 2, A and B, phosphorylation of p38 MAPK is up-regulated by hypoxia, whereas phosphorylation of Smad is down-regulated. When Smad signaling was inhibited by the overexpression of WT-Smad6, which inhibits phosphorylation of the Smad1-Smad5-Smad8 complex, BMP-2-induced GAG production was markedly reduced under both oxygen levels. However, hypoxia was still able to promote GAG production despite Smad inhibition (Fig. 2C). In contrast, hypoxia-induced GAG production was abolished by the overexpression of DN-MKK3 that specifically inhibits p38 MAPK phosphorylation (Fig. 2D) and also by FR167653, a specific p38 MAPK inhibitor (Fig. 2E). When both the Smad and the p38 MAPK pathways were blocked by Smad6 overexpression and FR167653, hypoxia did not influence the GAG production level (Fig. 2F).
Regulatory Mechanisms of Col2a1 and Col10a1 Gene Expression by Oxygen Tension-Next, we assessed the role of the p38 MAPK and Smad pathways in the regulation of Col2a1 and Col10a1 gene expression. In the presence of FR167653, hypoxia-induced Col2a1 expression was suppressed (Fig. 2, G-1). However, when WT-Smad6 was overexpressed, although Col2a1 gene expression was suppressed under both oxygen conditions, hypoxia strongly induced the Col2a1 gene. When both Smad and p38 MAPK signaling was blocked by WT-Smad6 and FR167653, Col2a1 gene induction caused by hypoxia was again strongly attenuated (Fig. 2, G-2). Our findings suggest that hypoxic regulation of Col2a1 gene expression is mediated by the p38 MAPK pathway rather than the Smad pathway, which is very similar to the mechanism of hypoxic regulation involved in GAG production. On the other hand,  OCTOBER 13, 2006 • VOLUME 281 • NUMBER 41

JOURNAL OF BIOLOGICAL CHEMISTRY 31085
Col10a1 gene expression was reduced by hypoxia as shown in Fig. 1G, and it seemed to be regulated in a complicated manner that is different from Col2a1 gene expression. The blockade of the p38 MAPK pathway did not alter the hypoxia-related reduction of Col10a1 gene expression, but the blockade of the Smad pathway abolished it. However, when both pathways were blocked, hypoxia reduced Col10a1 gene expression (Fig. 2G).

Hypoxia Enlarges the Cartilaginous Matrix Area Associated with Endochondral Ossification via the p38 MAPK Pathway in
Organ Culture-Our data show that hypoxia promoted the commitment of C3H10T1/2 cells to a chondrocytic lineage and enhanced cartilage matrix production via the p38 MAPK pathway. To further confirm the influence of oxygen tension on cartilage biology in a setting similar to that found in vivo, we cultured embryonic day 14.5 embryo forelimbs obtained from wild type and Smad6 transgenic mice. In ex vivo embryo limb cultures in the presence of BMP-2, explants grew quickly both in length and in width and wound greatly as they grow. Therefore, to evaluate cartilaginous matrix production, we calculated the proportion of the area of each radius that was stained with safranin-O compared with the whole radius area. The embryo forelimbs that were cultured under hypoxic conditions showed a significantly greater enlargement in the matrix area, and hypoxia-induced cartilage enlargement was abolished by FR167653 (Fig. 3, A-1). In explants from Smad6 transgenic mice, hypoxia enlarged the cartilage matrix area, and FR167653 abolished hypoxia-induced matrix enlargement (Fig. 3, B-1). These findings are consistent with our in vitro observation that hypoxia promotes cartilaginous matrix synthesis not via the Smad pathway but via p38 MAPK signaling (Fig. 2G).
Terminal Differentiation of Chondrocyte Is Suppressed by Hypoxia in Mouse Embryo Organ Cultures-In C3H10T1/2 cell cultures, Col10a1 gene expression was repressed by hypoxia. In accordance with this, when using in situ hybridization, we found that, in wild type embryo forelimb cultures, hypoxia reduced the Col10a1 gene expression level estimated by the proportional length of the Col10a1-positive site per whole hypertrophic chondrocyte zone (Fig. 4, bidirectional arrows) along the midline. When p38 MAPK signaling was suppressed using FR167653, the Col10a1-positive area was markedly reduced with both oxygen levels. In addition, hypoxia apparently reduced the Col10a1-positive area regardless of the use of p38 MAPK inhibitor (Fig. 4A). On the other hand, hypoxia alone did not repress Col10a1 gene expression in Smad6 trans-genic mice embryo forelimb organ culture, although hypoxia did suppress Col10a1 gene expression when the p38 MAPK pathway was inhibited by FR167653 (Fig. 4B). These findings were in agreement with the in vitro results and suggest that the Smad pathway mediates Col10a1 gene down-regulation and that the p38 MAPK pathway mediates its up-regulation. In addition, hypoxia reduced Col10a1 expression even when both Smad and p38 MAPK signaling were blocked, which would suggest that unknown factors that down-regulate Col10a1 gene expression are involved.
Sox9 Gene Expression and Its Transcriptional Activity Are Not Up-regulated by Hypoxia-Sox9 is a key transcriptional factor for chondrocytic differentiation and regulates transcription of the Col2a1 gene. Thus, we evaluated the influence of oxygen tension on Sox9 gene expression. Interestingly, Sox9 mRNA expression was not altered by hypoxia (Fig. 5A). As shown in Fig. 5B, Sox9 transcriptional activity was clearly upregulated by BMP-2. However, it was not up-regulated but rather was down-regulated by hypoxia. Sox9 transcriptional activity has been reported to be enhanced by cAMP-dependent protein kinase A serine/threonine phosphorylation (63). We found that phosphorylation of Sox9 was not promoted but was suppressed by hypoxia (Fig. 5C), which was consistent with the results of the reporter assay.
Runx2 Gene Expression and Its Transcriptional Activity Are Suppressed by Hypoxia-Since hypoxia suppressed chondrocyte hypertrophy and osteoblastic differentiation, we studied the role of Runx2 in the regulation of hypoxia-induced phenomena. Hypoxia suppressed Runx2 gene expression and its transcriptional activity in C3H10T1/2 cells stimulated with BMP-2 (Fig. 5, D and E). Blockade of the p38 MAPK pathway did not alter the suppression of Runx2 caused by hypoxia, but blockade of the Smad pathway abolished the effect of hypoxia. However, when both pathways were blocked, hypoxia could again suppress Runx2 gene expression (Fig. 5D). This pattern is very similar to the regulatory pattern we found for the hypoxic regulation of Col10a1 gene expression. To examine how Runx2 contributes to chondrocyte terminal differentiation in hypoxia, we transfected C3H10T1/2 cells with vector expressing wild type Runx2. When Runx2 was overexpressed, the Runx2 transcriptional activity suppressed by hypoxia was remarkably upregulated to a level comparable with that with normoxia (Fig.  5E). In addition, Col10a1 gene expression that had been reduced by hypoxia recovered to a level that matched the level expressed under normoxic conditions with Runx2 overexpres-  (n ϭ 3) of the proportion ratio for Runx2 gene expression as compared with normoxia. *, p Ͻ 0.001; N.S., not significant. Ⅺ, normoxia; f, hypoxia. E, C3H10T1/2 cells were co-transfected with 6Runx2E-Luc and TK-Renilla reporter constructs. As a control, the mouse osteoblast cell line (MC3T3-E1) was used. 24 h after transfection, cells were incubated with BMP-2 (500 ng/ml) for 2 days, and relative luciferase activity was measured and normalized by determining Renilla luciferase activity. Data are shown as mean Ϯ S. E. (n ϭ 3). *, p Ͻ 0.05; **, p Ͻ 0.0002; N.S., not significant. F, 24 h after transfection with wild type Runx2, the cells were stimulated with BMP-2 (500 ng/ml) and incubated under normoxia or hypoxia for 5 days. As a control, the cells were transfected with pGL3 promoter vector. Then Northern blotting for Col10a1 genes was performed. OCTOBER 13, 2006 • VOLUME 281 • NUMBER 41 JOURNAL OF BIOLOGICAL CHEMISTRY 31087 sion (Fig. 5F). These data suggest that Runx2 mediates Col10a1 gene suppression by hypoxia and that a third regulatory mechanism, independent of the Smad and the p38 MAPK pathways, is involved in the regulation by oxygen tension of Runx2 gene expression.

Oxygen Tension Regulates Chondrocytes
HDAC4 Is Expressed by Chondrocytes and Accumulates in the Nucleus under Hypoxia-Recently, it was reported that HDAC4 inhibits chondrocyte hypertrophy by suppressing Runx2 gene expression. HDAC4 has been reported to be expressed by chondrocytes in the prehypertrophic and hypertrophic zones (47). We could detect HDAC4 only in chondrocytic lineage cells (N1511 or C3H10T1/2) and almost not in osteoblastic cells (MC3T3-E1) (Fig. 6A). We furthermore confirmed that HDAC4 protein was expressed in the hypertrophic chondrocytes of the growth plate but not in proliferative chondrocytes (Fig. 6B). To elucidate whether HDAC4-Runx2 regulation is involved in the hypoxic suppression of terminal differentiation, we first examined the effect of oxygen tension on HDAC4 activation. Our immunofluorescence study revealed that the expression pattern of HDAC4 changed from being located in the cytoplasm to being located in the nucleus when cells were incubated under hypoxic conditions (Fig. 6, C-1). Accordingly, HDAC4 protein expression in the nuclear extract by Western blotting was up-regulated after 24 h of hypoxic stimulation (Fig. 6, C-2).

Inhibition of HDAC4 Up-regulates Runx2 Gene Expression and Its Transcriptional Activity and Restores Col10a1 Gene Expression That Is Decreased by Hypoxia-To evaluate whether HDAC4 is involved in the hypoxic regulation of Runx2 and
Col10a1 gene expression, we inhibited HDAC4 by RNA interference (Fig. 6D). The inhibition of HDAC4 partly restored the reduced Runx2 gene expression that resulted from 2 days of hypoxia. On day 5, Runx2 gene expression was clearly promoted by HDAC4 silencing regardless of oxygen level, and the hypoxia-induced suppression of Runx2 expression was completely restored to that seen with normoxia (Fig. 6E). Consistent with these results, HDAC4 inhibition completely abolished the reduction of Runx2 transcriptional activity caused by hypoxia (Fig. 6F). As a consequence, after 5 days of hypoxic stimulation, Col10a1 gene expression that had been suppressed by hypoxia was also up-regulated by HDAC4 inhibition and recovered to the level seen with normoxia. These data strongly suggested that hypoxia activates HDAC4, which in turn suppresses Runx2 activity and subsequently leads to the reduction of Col10a1 gene expression.

DISCUSSION
Murine mesenchymal C3H10T1/2 cells are pluripotent and differentiate into several lineages (19). BMP-2 and BMP-7 induce C3H10T1/2 cells to differentiate into osteoblasts, chondrocytes, and adipocytes; a low concentration favors adipocytes, and a high concentration favors chondrocytes and osteoblasts (18,19,64,65). Thus, C3H10T1/2 cells are an appropriate model for studying the mechanisms of pluripotent mesenchymal cell commitment into a particular lineage. In the present study, we cultivated C3H10T1/2 cells in the presence or absence of recombinant human BMP-2 under normoxia or hypoxia. BMP-2 treatment and low oxygen tension synergistically induced GAG production and Col2a1 gene expression and profoundly suppressed ALP activity and mineralization; they did not alter fat droplet production. These data indicate that hypoxia promoted chondrocytic commitment of the pluripotent C3H10T1/2 mesenchymal cells and inhibited osteoblastic differentiation. Our results are in accordance with previous reports, which describe that hypoxia enhances Col2a1 and Aggrecan gene expression in C3H10T1/2 cells (66). To further investigate the role of oxygen tension in chondrocytic differentiation, we used a mouse embryo forelimb explant culture system. This system enables one to cultivate cartilage tissue under hypoxic conditions for up to 2 weeks without any interference from systemic, hormonal, or neuronal responses to the hypoxia that could affect cartilage metabolism while at the same time examining the effects of both environmental factors and cytokines on the endochondral ossification process. In organ cultures of wild type mice forelimbs at 14.5 days postcoitum, we found that hypoxia clearly induced the enlargement of the cartilage matrix area stained by Safranin-O, suggesting that hypoxia promoted cartilage matrix synthesis by chondrocytes during endochondral ossification. To confirm this, we cultured N1511 chondrocytes under hypoxic conditions and found that hypoxia induced N1511 cells to produce GAG. Our data indicate that hypoxia not only induced the commitment of pluripotent mesenchymal progenitors into a chondrocyte lineage but also activated the production by chondrocytes of cartilage matrix.
In the present study, hypoxia in the presence of BMP-2 clearly suppressed Col10a1 mRNA expression in C3H10T1/2 cell culture and also in organ culture studies using tissues from wild type animals. This suggests that hypoxia suppresses the terminal differentiation of chondrocytes during endochondral FIGURE 6. With hypoxia, HDAC4 accumulates in the nucleus and down-regulates Runx2 expression, Runx2 transcriptional activity, and Col10a1 gene expression. A, nuclear and cytoplasmic extracts from C3H10T1/2, N1511, and MC3T3-E1 cells (mouse osteoblastic cells) were examined by Western blotting with anti-HDAC4 antibodies. Mixed lysate (nuclear extract (10 g) plus cytoplasmic extracts (10 g)) was blotted for ␤-actin for the control. B, immunohistochemical analysis for HDAC4 protein in normal mouse embryo humerus (embryonic day 14.5). Bars, 30 m. B-1, HDAC4 protein expression in hypertrophic chondrocytes, not in proliferative chondrocytes. B-2, negative control without primary antibody. C, nuclear translocation of HDAC4 protein by hypoxia. C-1, immunofluorescence staining for HDAC4 protein. After starvation for 24 h, C3H10T1/2 cells were stimulated with BMP-2 (500 ng/ml) and further incubated under normoxia or hypoxia for 1 h. Immunofluorescence staining was performed using anti-HDAC4 antibody (1:100) for 2 h followed by Alexa Fluor 488 for 30 min. Hoechst 33342 was used for nucleus staining. Bar, 12.5 m. C-2, nuclear and cytoplasmic extracts from C3H10T1/2 cells were harvested and examined by Western blotting with anti-HDAC4 antibodies. Proliferating cell nuclear antigen and ␤-actin acted as the internal loading control for nuclear and cytoplasmic fractions, respectively. N, normoxia; H, hypoxia. D, 24 h after transfection with siRNA for HDAC4, nuclear extracts from the cells were harvested and analyzed by Western blotting with anti-HDAC4 antibodies. Proliferating cell nuclear antigen acted as the internal loading control for nuclear fractions. After transfection with siRNA for HDAC4, the cells were incubated for 5 days, and reverse transcription-PCR analysis for Runx2 was done. E, 24 h after transfection with siRNA for HDAC4, the cells were stimulated by 500 ng/ml BMP-2 for either 48 h or 5 days. Northern blotting for Runx2 and Col10a1 were done. F, C3H10T1/2 cells were co-transfected with siRNA, 6Runx2E-Luc, and TK-Renilla reporter constructs. 24 h after transfection, cells were incubated with BMP-2 (500 ng/ml) for 48 h under normoxia or hypoxia. At the end of the culture, relative luciferase activity was determined. Data are shown as mean Ϯ S.E. (n ϭ 3). *, p Ͻ 0.05; N.S., not significant. ossification. Thus, hypoxia seems to act on chondrocytes to preserve their chondrocyte phenotype by preventing hypertrophy and, consequently, terminal differentiation.
To confirm the mechanisms by which these hypoxia-induced phenomena operate, we studied the signaling pathways of BMP-2 that are well known to induce chondrogenesis of mesenchymal cells (17)(18)(19)(20)(21) via the induction of Sox9 gene expression (67). Basically, BMP-2 signals propagate through the Smad pathway and bind directly or via other DNA-binding proteins to the promoters of BMP-2-responsive genes to stimulate or repress their transcription. Over the past few years, evidence has accumulated that suggests that BMP-2 may also stimulate other downstream pathways involving p38 MAPK. In the present study, we found that hypoxia enhanced BMP-2-induced activation of the p38 MAPK pathway, whereas hypoxia did not promote and could even suppress the Smad pathway. Furthermore, our Smad-and p38 MAPK-suppressing studies using WT-Smad6, DN-MKK3, and FR167653 revealed that it is not the Smad pathway but the p38 MAPK pathway that is indispensable for hypoxia-induced cartilaginous matrix synthesis. Overall, our data indicate that hypoxia promotes cartilaginous matrix synthesis via the p38 MAPK pathway.
Little is known about how oxygen tension regulates the signal transduction systems that regulate Col10a1 gene expression. In the present study, we found that hypoxia promoted BMP-2induced activation of the p38 MAPK pathway but did not promote, and could even suppress, the Smad pathway. Previous reports suggested that Smad1-Smad5 signaling positively regulates type X collagen expression by potentiating the transcriptional activity of Runx2 (44,45) and that p38 MAPK signaling also promotes its expression (34,45,68) via Runx2 activation. These observations are consistent with the results of our experiments that blocked these pathways by Smad6, DN-MKK3, or the p38 inhibitor, FR167653. This suggests that hypoxia positively regulates Col10a1 gene expression via the up-regulation of p38 MAPK signaling and negatively regulates Col10a1 gene expression via down-regulation of the Smad pathway. In addition to this, we found that hypoxia up-regulated HDAC4 activity, which recently was reported to control chondrocyte hypertrophy by suppressing Runx2 gene expression through inhibition of the gene's positive feedback mechanism and the suppression of its transcriptional activity (47). We were able to confirm that hypoxic suppression of Runx2 gene expression, transcriptional activity, and Col10a1 gene expression were blocked by silencing HDAC4. This indicates that hypoxia activates HDAC4, thereby suppressing Runx2 activity, which in turn down-regulates Col10a1 gene expression. Our data reveal that hypoxia inhibits the hypertrophy of chondrocytes by down-regulating Runx2 activity based on the sum of the positive regulation that occurs via p38 MAPK activation and the negative regulation that is caused by Smad signaling suppression and HDAC4 activation.
Transcriptional factors Sox9, L-Sox5, and Sox6, have been reported to be essential and sufficient for regulating the expression of Col2a1 and the other genes involved in the chondrocytic program (69). This would suggest that Sox transcriptional factors should be up-regulated by hypoxia and thus propagate the signals for chondrocytic differentiation and cartilage matrix production. However, very interestingly, our results indicate that hypoxia-mediated chondrocytic differentiation did not involve the up-regulation of Sox9 gene expression, its phosphorylation, or its transcriptional activity, suggesting that hypoxia-induced Col2a1 induction was independent of Sox9. However, in the previous report, it was described that Sox9 gene expression was up-regulated by hypoxia in association with transactivation of the Sox9 promoter in ST2 stromal cells (66). Although the difference in the results might depend on the difference in cell type, further confirmation about the Sox9 promoter may be necessary. The results of the reporter assay lead us to speculate that transcription factors other than Sox9 directly regulate Col2a1 gene expression via other cis elements located at a site other than the chondrocyte-specific enhancer fragment of the Col2a1 gene (69,70). The Sox trio shares a cis element in the Col2a1 chondrocyte-specific enhancer region; however, to exclude the involvement of L-Sox5 and Sox6 in hypoxiainduced Col2a1 gene expression, further confirmation, including gene expression of these transcription factors, is needed. It is known that hypoxic stress reduces protein kinase A activity (71). Recently, protein kinase A was reported to phosphorylate Sox9 and to enhance its transcriptional activity (62). Taking these data into account, it is plausible that Sox9 transcriptional activity could have been down-regulated by hypoxia. However, the actual transcription factors responsible for hypoxia-induced Col2a1 induction remain to be elucidated (Fig. 7).
In conclusion, we demonstrated that hypoxia clearly promoted chondrocytic commitment of cells in the mesenchymal lineage, as well as cartilaginous matrix synthesis, and inhibited terminal differentiation both in cell culture and organ culture. These effects were primarily mediated by p38 MAPK activation independent of Sox9. On the other hand, hypoxia inhibited the hypertrophy of chondrocytes via the down-regulation of Runx2 activity through Smad signaling suppression and HDAC4 activation. These hypoxia-induced phenomena affect mesenchymal cell differentiation and endochondral ossification by enhancing and preserving the chondrocytic phenotype and cell function, as well as preventing chondrocytes from terminal differentiation that subsequently leads to matrix degeneration and chondrocyte apoptosis. The hypoxia-associated regulation of chondrocytes that has been outlined in our study may be of fundamental importance in the biology and pathology of cartilage tissues and may be involved in endochondral ossification in the growth plate, in joint cartilage homeostasis, and in the etiology of osteoarthritis.