Reactive oxygen species induce chondrocyte hypertrophy in endochondral ossification

We began by asking whether ROS levels were elevated in vivo during chondrogenesis by undertaking dihydroethidium staining of wild-type E17.5 mouse embryos. Dihydroethidium is intracellularly oxidized by ROS and converted to fluorescent ethidium (Eth). Eth can bind to chromosomal DNA, yielding Eth-DNA; therefore, ROS can be detected by nuclear Eth-DNA accumulation. Higher levels of intracellular ROS were detected in prehypertrophic and hypertrophic chondrocytes compared with proliferating chondrocytes or surrounding tissues (Fig. 1 A). To further analyze ROS levels during chondrogenesis, the chondrogenic cell line ATDC5 was cultured in differentiation medium containing insulin, transferrin, and sodium selenite, as previously described (16). Under this culture condition, ATDC5 cells efficiently undergo a chondrogenic differentiation program similar to that observed in vivo (17). Hypertrophic chondrocyte markers such as type X collagen and mmp13, as well as increased intracellular levels of ROS detected by dichlorodihydrofluorescein diacetate (DCF-DA), were induced in ATDC5 cells cultured in differentiation medium for 2 wk (Fig. 1, B and C). We examined the effect of NAC treatment on intracellular ROS levels during chondrocyte hypertrophy. After NAC treatment, the higher intracellular ROS levels in ATDC5 cells cultured in differentiation medium were effectively reduced to levels comparable to those seen in cells cultured in control medium (Fig. 1 C). Basal ROS levels in cells cultured even in control medium gradually increased and were NAC resistant (Fig. 1 C), suggesting the existence of DCF-sensitive oxidants other than H₂O₂. Because NAC is a glutathione precursor, which reduces intracellular H₂O₂ levels, differentiation medium may specifically elevate intracellular H₂O₂. Nicotinamide adenine dinucleotide phosphate–oxidase–1 generates superoxide and is thus responsible for ROS production (10). H₂O₂ is generated via dismutation of superoxide. We found that nox1 expression was highly induced in ATDC5 cells cultured in differentiation medium compared with control medium (Fig. S1), suggesting that differentiation signals induce expression of an ROS-generating enzyme. Interestingly, nox1 expression in ATDC5 cells cultured in differentiation medium was reduced by antioxidant treatment (Fig. S1). Thus, H₂O₂ signals may induce expression of ROS-generating enzymes, which in turn stimulate chondrocyte hypertrophy by increasing ROS levels.

We next examined the effect of increased ROS on chondrocyte hypertrophy by manipulating intracellular ROS with NAC. Hypertrophic chondrocytes were detected in cartilage nodules, the cellular condensations formed in cultures of ATDC5 cells (17). To determine whether hypertrophic changes in chondrocytes are inhibited by reducing ROS, chondrocyte nodules were analyzed in the presence or absence of NAC (Fig. 2 A). Hypertrophic chondrocytes detected as type X collagen–positive cells were more effectively induced in the nodules by differentiation medium compared with control medium, and such hypertrophic changes were inhibited by NAC treatment. However, there was a recovery in the formation of type X collagen–positive cell–containing nodules even after NAC treatment at 3 and 4 wk of culture. This is possibly caused by an elevation in the levels of basal oxidants resistant to NAC, as shown in Fig. 1 C.

It has been reported that mmp13 and type X collagen are late hypertrophic chondrocyte differentiation markers, and their expression is not detected in proliferating chondrocyte zones in vivo (2, 3). Runx2 is a transcription factor essential for osteoblast differentiation, and Runx2-deficient mice show defective hypertrophic chondrocyte maturation, whereas forced Runx2 expression induces chondrocyte hypertrophy (5, 18). To evaluate chondrocyte hypertrophy in ATDC5 cells cultured in various conditions with or without NAC, we examined the expression of hypertrophic chondrocyte markers such as mmp13, type X collagen, and runx2 by RT-PCR. Expression of mmp13, type X collagen, and runx2 was reduced by NAC treatment, and the hypertrophic changes in chondrocytes were inhibited by reducing ROS (Fig. 2 B). Again, there was a recovery in expression of these genes at late culture stages after NAC treatment.

To analyze the role of ROS in chondrocyte hypertrophy in vivo, we analyzed chondrogenesis in mice treated with NAC. To do so, we injected NAC into mice every other day from the day of birth until 2 or 3 wk of age, when animals were analyzed. ROS levels in growth-plate chondrocytes were effectively reduced by this treatment (Fig. 3 A). To analyze chondrocyte hypertrophy in growth plates treated with NAC, proteoglycans were stained by Alcian blue or Safranin-O, as previously described (19, 20). Proteoglycans are mainly produced in proliferating chondrocytes and down-regulated in hypertrophic chondrocytes. NAC treatment reduced the size of the hypertrophic chondrocyte zone in growth plates (Fig. 3 B). Interestingly, the proteoglycan synthesis indicated by Alcian blue staining was up-regulated by NAC treatment (Fig. 3 B), and the reduced hypertrophy and up-regulated proteoglycan production were confirmed by Safranin-O staining (Fig. S2).

The number of BrdU-positive proliferating cells also increased in growth-plate chondrocytes after NAC treatment (Fig. 3 C). Chondrocyte hypertrophy induces vascular endothelial growth factor expression, resulting in angiogenesis, and the hypertrophic chondrocytes then undergo apoptosis. NAC administration every other day suppressed growth of endothelial cells, as detected by platelet-endothelial cell adhesion molecule PECAM staining (Fig. 3 D). The number of apoptotic cells was also reduced by this treatment (Fig. S3 A).

Treatment of ATDC5 cells with H₂O₂ significantly reduced the number of cells, a change abolished by addition of NAC to the culture medium (Fig. S3 B). H₂O₂ treatment also induced apoptosis, as judged by the appearance of Annexin V–positive and propidium iodide–negative ATDC5 cells, which was reversed by NAC (Fig. S3 C). Thus, increased ROS inhibit proliferation and induce both chondrocyte hypertrophy and apoptosis. Such changes mediated by H₂O₂ were reversed by addition of NAC to the culture. It is assumed that physiologically increasing ROS inhibits proliferation and stimulates the differentiation of proliferating chondrocytes into hypertrophic chondrocytes, which eventually undergo apoptosis.

During endochondral ossification, hypertrophic chondrocytes induce angiogenesis and mineralization below the hypertrophic chondrocyte zone. Mineralization below the growth plates was weakly reduced by NAC treatment (unpublished data). Collectively, these results indicate that increasing ROS in chondrocytes physiologically induce hypertrophy and further regulate the apoptosis of chondrocytes, which are followed by angiogenesis and mineralization in growth plates.

To analyze ROS signaling in chondrocyte hypertrophy directly, we elevated ROS levels by treating ATDC5 cells with H₂O₂ in control medium for 48 h and assayed expression of differentiation markers by RT-PCR (Fig. 4 A). H₂O₂ treatment effectively induced expression of these markers after 48 h of culture (Fig. 4 A). Induction of mmp13 by 48 h of treatment with H₂O₂ was dose dependent (Fig. 4 B). Runx2- deficient mice show defective hypertrophic chondrocyte maturation, and forced Runx2 expression induces chondrocyte hypertrophy (4, 5, 18, 21). Induction of both mmp13 and runx2 after treatment with H₂O₂ for 48 h was inhibited by NAC (Fig. 4 C).

It has been reported that the MAPK/ERK activating kinase (MEK)–ERK pathway plays a role in hypertrophic differentiation in chondrocytes, whereas p38 signaling is required for the transition from a prehypertrophic to a completely hypertrophic chondrocyte phenotype (for review see reference 22). To determine whether elevated ROS directly transduce a signal through MAPK pathways, we performed a Western blot analysis to analyze phosphorylation of ERK, c-Jun N-terminal kinase (JNK), and p38 MAPK in H₂O₂-treated ATDC5 cells. H₂O₂ treatment induced ERK and p38 MAPK phosphorylation but not that of JNK (Fig. 4 D). ERK and p38 MAPK but not JNK phosphorylation were also induced in ATDC5 cells treated with insulin (Fig. S4). Because insulin was added to control medium to yield differentiation medium, H₂O₂ transduced signals similar to those mediated by differentiation medium. U016 (an MEK inhibitor that thereby inhibits the ERK pathway) and SB203580 (a p38 inhibitor) but not SP600125 (a JNK inhibitor) suppressed mmp13 up-regulation by H₂O₂, indicating that ROS transduce a differentiation signal via the ERK and p38 MAPK pathways (Fig. 4 E). These results show that elevation of ROS in chondrocytes during differentiation activates ERK and p38 MAPK, and that activation of these MAPKs is required for chondrocyte hypertrophy.

Next, we performed metatarsal culture, an ex vivo explant culture, to examine how ROS elevation induces hypertrophy. To do so, we dissected three middle metatarsals from the hindlimb of E17.5 mouse embryos and cultured them in the presence or absence of H₂O₂ with or without NAC for 5 d. Treatment of metatarsal bones by H₂O₂ disrupted formation of a columnar structure in the proliferating chondrocyte zone (Fig. 4 F). Mineralization and apoptosis of the metatarsal bones were also stimulated by H₂O₂ treatment (Fig. S5).

Increased ROS levels are detected in various cells in ataxia-telangiectasia patients (13). Because chondrogenesis in growth plates plays a pivotal role in longitudinal growth, we analyzed the expression and activation of ATM in chondrocytes from growth plates of Atm^−/− mice. Phosphorylated ATM, the active form of the protein, was detected in growth-plate chondrocytes in wild-type mice (Fig. 5 A). Next, we assayed ROS concentration in growth-plate chondrocytes. Intracellular ROS concentrations, evaluated by dihydroethidium staining, were higher in chondrocytes from Atm^−/− mice than from wild-type controls (Fig. 5 B). To analyze whether ROS elevation affects chondrocyte proliferation, we undertook BrdU analysis in Atm^−/− and wild-type mice. The number of proliferating, BrdU-positive chondrocytes was significantly reduced in Atm^−/− mice compared with controls (Fig. 5 C), suggesting that ROS levels are regulated by ATM in growth-plate chondrocytes and that such ROS accumulation may affect chondrocyte proliferation in growth plates.

In wild-type mice, proliferating chondrocytes line up along the long axis of the bone, which shows a columnar structure (Fig. 5 D, highlighted in the small boxes). However, in the growth plates of Atm^−/− mice, the formation of such columnar structures characteristic of proliferating chondrocyte zones is disrupted (Fig. 5 D). Proteoglycan synthesis was also reduced, and hypertrophic zones were enlarged in Atm^−/− mice compared with controls (Fig. 5 D). Induction of vascular endothelial cells under the hypertrophic chondrocyte zone and induction of apoptosis in hypertrophic chondrocytes were stimulated in Atm^−/− mice compared with wild-type mice (Fig. S6; and not depicted). Furthermore, mineralization under the growth-plate region, as evaluated by alizarin red staining, increased in Atm^−/− compared with control mice (Fig. 5 E). Thus, ATM may regulate the process whereby proliferating chondrocytes differentiate into hypertrophic chondrocytes.

To determine whether altered chondrogenesis in Atm^−/− mice is related to oxidative stress, we injected NAC into Atm^−/− mice every other day from the day of birth until 2 wk of age, when animals were analyzed. The high ROS levels seen in growth-plate chondrocytes were effectively reduced by such treatment in Atm^−/− mice (Fig. 6 A). Inhibited proliferation was rescued by NAC treatment in growth-plate chondrocytes of Atm^−/− mice (Fig. 6 B). The elongated hypertrophic chondrocyte zone seen in the growth plates of Atm^−/− mice and reduced proteoglycan synthesis were rescued by NAC treatment (Fig. 6 C). The increased number of apoptotic cells in Atm^−/− hypertrophic chondrocytes was also significantly rescued by NAC treatment (Fig. 6 D), as was angiogenesis stimulated below the hypertrophic chondrocyte zone in Atm^−/− mice (Fig. 6 E). Collectively, our findings suggest that elevated intracellular ROS transduce a signal suppressing proliferation and induce a hypertrophic state in chondrocytes, and that control of ROS levels by ATM is required for appropriate chondrocyte proliferation and differentiation in the growth plate.

Chondrogenesis occurs via three steps: (a) commitment to chondrocyte differentiation by mesenchymal cells, seen as mesenchymal condensation; (b) chondrocyte proliferation in the growth plate to facilitate longitudinal development; and (c) differentiation of proliferating chondrocytes to hypertrophic chondrocytes (for review see references 1, 2). During endochondral ossification, chondrocyte hypertrophy is critical, because cells alter the extracellular matrix and induce vascular invasion. Extrinsic factors such as bone morphogenetic proteins, Indian hedgehog, and modulators such as Sox9, Runx2, Smads, and histone deacetylase 4 are reportedly essential for chondrogenesis (1, 2). Loss of histone deacetylase 4 causes early onset of hypertrophy, followed by growth retardation (6). Overexpression of Runx2 under the control of a type II collagen promoter/enhancer induces dwarfism because of precocious endochondral ossification and accelerated chondrocyte differentiation (4, 5). Thus, regulation of hypertrophy is critical for normal growth (4–6), and abnormalities in chondrocyte hypertrophy cause skeletal growth retardation.

We propose that ROS play an essential role as signal transducers for chondrocyte hypertrophy. We demonstrate that ROS mediated inhibition of proliferation in proliferating chondrocytes and induction of hypertrophic chondrocytes. Because chondrocyte hypertrophy is associated with apoptosis, angiogenesis, and mineralization, regulation of ROS levels in chondrocytes is crucial for normal longitudinal bone growth. We show that ROS was specifically elevated in chondrocytes during the hypertrophic changes. Chondrocyte hypertrophy in ATDC5 cells, a chondrogenic cell line, was inhibited by antioxidant NAC treatment, whereas H₂O₂ treatment stimulated hypertrophic activity. Inhibition of chondrocyte hypertrophy was observed in vivo after administration of NAC to normal mice, indicating that physiological levels of ROS determine chondrocyte hypertrophy. We also show that ROS transduced a differentiation signal via ERK and p38 MAPK pathways, suggesting that ROS act as a biological messenger in chondrocyte hypertrophy. Chondrocyte apoptosis, angiogenesis, and mineralization were inhibited by NAC treatment, suggesting that these events result from chondrocyte hypertrophy induced by ROS. To further clarify the in vivo role of ROS in chondrogenesis, we analyzed the chondrogenesis of Atm^−/− mice, which show higher intracellular accumulation of ROS compared with wild-type mice in growth-plate chondrocytes. ATM likely functions in oxidative defense by inducing major antioxidative systems, and this study shows that ATM loss increased ROS levels in chondrocytes and induced premature hypertrophy in chondrocytes. Because apoptosis is stimulated in the chondrocytes of ATM-deficient mice, elongation of the hypertrophic chondrocyte zone in ATM-deficient mice is not particularly marked compared with other mutant animals such as MMP13-deficient mice (19). We demonstrate that in vivo administration of NAC to Atm^−/− mice down-regulated intracellular H₂O₂ levels and up-regulated proliferation and proteoglycan synthesis in chondrocytes, thereby rescuing premature hypertrophy. ROS-induced tissue damage is implicated in diseases such as osteoarthritis (23, 24), Alzheimer's disease (25, 26), and ataxia-telangiectasia (13) and in aging (27–29) and stem cell dysfunction (14). We underscored a physiological role for ROS in chondrocyte hypertrophy and observed abnormalities of chondrocyte hypertrophy after an increase in ROS levels.

ROS accumulation is caused by two mechanisms: increased ROS generation and decreased ROS degradation. Intracellular ROS may accumulate as a by-product of a high cellular metabolism, because proliferating chondrocytes show high proliferation activity. In addition, differentiation signals induced the expression of nox1, which is responsible for ROS production, including H₂O₂. This suggests that chondrocyte hypertrophy was stimulated by the expression of an ROS-producing gene. Further studies are needed to clarify the signals responsible for the increased generation of ROS during chondrocyte hypertrophy. Meanwhile, it has been shown that chondrocyte maturation is accompanied by a progressive decrease in catalase activity in the growth cartilage (30). Catalase plays key roles in protecting cells from oxidative stress by destruction of H₂O₂. Thus, chondrocyte hypertrophy is mediated by intracellular ROS levels, which are regulated by ATM. In fact, intracellular ROS were down-regulated by NAC, which reduced ROS levels in the chondrocytes of wild-type and ATM-deficient mice, and NAC treatment was accompanied by inhibition of chondrocyte hypertrophy. Collectively, we conclude that ROS accumulation beyond threshold levels may initiate inhibition of proliferation in late-proliferating chondrocytes and stimulate hypertrophic differentiation.

Animals were purchased from CLEA Japan, kept under pathogen-free conditions, and cared for in accordance with the guidelines of Keio University School of Medicine. Pregnant C57BL/6 mice were purchased, and embryos between E15.5–17.5 were isolated from the uteri of pregnant females. Atm^+/− mice were a gift from P.J. Mckinnon (St. Jude Children's Research Hospital, Memphis, TN). Genotyping was performed by PCR-based assays of mouse-tail DNA. Littermates served as controls in all experiments.

ATDC5 cells were purchased from the Institute of Physical and Chemical Research Cell Bank. Cells were plated at 6 × 10⁴ cells/well in six-well plates and cultured in DMEM/Ham's F12 (1:1; DMEM/F12) medium (Invitrogen) supplemented with 5% FBS (Equitech-Bio, Inc.), 10 μg/ml of human transferrin (Roche Molecular Biochemicals), and 3 × 10⁻⁸ M sodium selenite (Sigma-Aldrich) as control medium, or in 10 μg/ml of human transferrin, 3 × 10⁻⁸ M sodium selenite, and 10 mg/ml of bovine insulin as differentiation medium. Cells were also cultured in DMEM/F12 containing 5% FBS with various concentrations of H₂O₂ (Wako) instead of differentiation medium for 48 h. For experiments with NAC and MAPK inhibitors, ATDC5 cells were cultured in the presence of 100 μM H₂O₂ with 500 μM NAC (Sigma-Aldrich), 10 μM U0126 (MEK1/2 inhibitor), 10 μM SB203580 (p38 MAPK inhibitor), or 10 μM SP600125 (JNK inhibitor) for 48 h. To analyze apoptosis, ATDC5 cells cultured in the presence or absence of 100 μM H₂O₂ with or without 500 μM NAC in control medium for 24 h were stained with FITC-conjugated Annexin V and propidium iodide (BD Biosciences) and analyzed by FACSCalibur (Beckton Dickinson).

Three middle metatarsals of each hindlimb of E17.5 mouse embryos were dissected and cultured for 5 d on a culture plate insert (Millicell; Millipore) in 1 ml BGJb medium (Invitrogen) supplemented with 10% FCS. 500 μM H₂O₂ was added with or without 100 μM NAC. After cultivation, metatarsals were fixed in 4% paraformaldehyde/PBS, embedded in paraffin, sectioned to 4 μm, and stained with hematoxylin and eosin (H&E).

Cell samples were either stained with DCF-DA (Sigma-Aldrich) at 4°C for 30 min and analyzed by FACSCalibur or loaded with 5 μM DCF-DA and incubated on a shaker at 37°C for 30 min. The peak excitation and emission wavelengths for oxidized DCF were 488 and 525 nm, respectively. For dihydroethidium staining, the forelimb of E17.5 mouse embryos was dissected, embedded in minced rat liver, and frozen in 2-methylbutane (Sigma-Aldrich) plus liquid nitrogen. Frozen sections of forelimb that had not been decalcified were cut on a cryostat (model HM505; MICROME), stained with dihydroethidium (Invitrogen) at room temperature for 30 min, and observed under a confocal microscope (FV1000; Olympus). ROS production in chondrocytes was quantified by signal intensity using a confocal image analyzer (FV10-ASW; Olympus).

Tibias were dissected from 2–3-wk-old Atm^−/− or control mice treated with or without PBS or NAC (1 g/kg, every 2 d) for 2–3 wk. Frozen sections that had not been decalcified were prepared as previously described (31). Specimens were stained with antiphosphorylated ATM antibody (Rockland), followed by Alexa Fluor 488–conjugated goat anti–rabbit IgG antibody (Invitrogen) and TOTO3 (Invitrogen) for nuclear staining. For BrdU staining, 200 μl BrdU solution was injected into the peritoneal cavity of 2-wk-old Atm^−/− or control mice. Mice were killed 2.5 h after injection, and the tibias were extracted. Frozen sections were prepared as previously described (31) and subjected to BrdU immunohistochemistry analysis (Calbiochem). Fluorescent images were obtained using a confocal laser scanning microscope. For H&E and Alcian blue staining, tibias were dissected from 3-wk-old Atm^−/− or control mice treated with or without PBS or NAC (1 g/kg, every 2 d) for 3 wk, fixed in 10% formalin, decalcified, embedded in paraffin, and cut into 4-μm sections. Deparaffinized sections of paraffin-embedded samples were stained with H&E (Muto Pure Chemical Co. Ltd.) or Alcian blue (Sigma-Aldrich).

ATDC5 cells were starved in control medium without FBS for 12 h and stimulated with 100 μM H₂O₂ for various minutes. Cell lysates were collected, and Western blot analysis was performed using polyclonal antibodies to detect ERK, phosphorylated ERK, p38 MAPK, phosphorylated p38 MAPK, JNK, and phosphorylated JNK. All antibodies were obtained from Cell Signaling Technology.

Total RNA was extracted from cultured ATDC5 cells, and cDNA was synthesized by RT. RT-PCR analysis was performed using the following primer sets: mmp13-sense, 5′-AGAAGTCTACAGTGACCTCCACAGTT-3′; mmp13-antisense, 5′-GACTCTCACAATGCGATTACTCC-3′; type X collagen–sense, 5′-TTCATGGGATGTTTTATGCTGAACG-3′; type X collagen–antisense, 5′-TTTAGGTCCTTGGGGTCCCATATTC-3′; runx2- sense, 5′-AGAAGGCTCTGGCGTTTAAATGGTT-3′; runx2-antisense, 5′-AAAAGGACTTGGTGCAGAGTTCAGG-3′; β-actin–sense, 5′-TCGTGCGTGACATCAAAGAG-3′; and β-actin–antisense, 5′-TGGACAGTGAGGCCAGGATG-3′.

p-values were calculated by the unpaired Student's t test.

Fig. S1 shows the expression of nox1 in chondrocytes. Fig. S2 shows the Safranin-O staining of chondrocytes in mice treated with PBS or NAC. Fig. S3 shows the induction of chondrocyte apoptosis by ROS. Fig. S4 shows the activation of MAPKs by insulin in chondrocytes. Fig. S5 shows the induction of mineralization and apoptosis in chondrocytes by ROS. Fig. S6 shows the chondrocyte apoptosis in ATM-deficient mice.

We thank P.J. McKinnon for providing Atm^+/− mice, and Y. Sato and A. Kumakubo for technical support.

T. Miyamoto was supported by a grant-in-aid for Young Scientists (B), Japan. T. Suda was supported by a grant-in-aid from Specially Promoted Research of the Ministry of Education, Culture, Sports, Science and Technology, Japan.

The authors have no conflicting financial interests.