Reactive Oxygen Species Generated by NADPH Oxidase 2 and 4 Are Required for Chondrogenic Differentiation*

Although generation of reactive oxygen species (ROS) by NADPH oxidases (Nox) is thought to be important for signal transduction in nonphagocytic cells, little is known of the role ROS plays in chondrogenesis. We therefore examined the possible contribution of ROS generation to chondrogenesis using both ATDC5 cells and primary chondrocytes derived from mouse embryos. The intracellular level of ROS was increased during the differentiation process, which was then blocked by treatment with the ROS scavenger N-acetylcysteine. Expression of Nox1 and Nox2 was increased upon differentiation of ATDC5 cells and primary mouse chondrocytes, whereas that of Nox4, which was relatively high initially, was decreased gradually during chondrogenesis. In developing limb, Nox1 and Nox2 were highly expressed in prehypertrophic and hypertrophic chondrocytes. However, Nox4 was highly expressed in proliferating chondrocytes and prehypertrophic chondrocytes. Depletion of Nox2 or Nox4 expression by RNA interference blocked both ROS generation and differentiation of ATDC5 cells, whereas depletion of Nox1 had no such effect. We also found that ATDC5 cells depleted of Nox2 or Nox4 underwent apoptosis. Further, inhibition of Akt phosphorylation along with subsequent activation of ERK was observed in the cells. Finally, depletion of Nox2 or Nox4 inhibited the accumulation of proteoglycan in primary chondrocytes. Taken together, our data suggest that ROS generated by Nox2 or Nox4 are essential for survival and differentiation in the early stage of chondrogenesis.

Although generation of reactive oxygen species (ROS) by NADPH oxidases (Nox) is thought to be important for signal transduction in nonphagocytic cells, little is known of the role ROS plays in chondrogenesis. We therefore examined the possible contribution of ROS generation to chondrogenesis using both ATDC5 cells and primary chondrocytes derived from mouse embryos. The intracellular level of ROS was increased during the differentiation process, which was then blocked by treatment with the ROS scavenger N-acetylcysteine. Expression of Nox1 and Nox2 was increased upon differentiation of ATDC5 cells and primary mouse chondrocytes, whereas that of Nox4, which was relatively high initially, was decreased gradually during chondrogenesis. In developing limb, Nox1 and Nox2 were highly expressed in prehypertrophic and hypertrophic chondrocytes. However, Nox4 was highly expressed in proliferating chondrocytes and prehypertrophic chondrocytes. Depletion of Nox2 or Nox4 expression by RNA interference blocked both ROS generation and differentiation of ATDC5 cells, whereas depletion of Nox1 had no such effect. We also found that ATDC5 cells depleted of Nox2 or Nox4 underwent apoptosis. Further, inhibition of Akt phosphorylation along with subsequent activation of ERK was observed in the cells. Finally, depletion of Nox2 or Nox4 inhibited the accumulation of proteoglycan in primary chondrocytes. Taken together, our data suggest that ROS generated by Nox2 or Nox4 are essential for survival and differentiation in the early stage of chondrogenesis.
The formation of cartilage is characterized by the condensation of mesenchymal cells and the subsequent differentiation of chondrocytes, which includes cell proliferation, maturation, and hypertrophy (1). Chondrocyte differentiation is accompanied by the secretion of a matrix rich in collagen type II and aggrecan, and this process is regulated by transforming growth factor-␤, fibroblast growth factors, bone morphogenetic proteins, and the transcription factor Sox9 (2,3). The chondrogenic cell line ATDC5 has been studied as an in vitro model of chondrocyte differentiation (4). Specifically, exposure of ATDC5 cells to insulin induces a process reminiscent of chondrocyte differentiation (5). Insulin stimulation elicits the rapid production of H 2 O 2 in various cell types, resulting in oxidative inhibition of protein-tyrosine phosphatases, followed by the enhancement of protein tyrosine phosphorylation (6,7). ROS 3 such as H 2 O 2 , superoxide anion (O 2 . ), and hydroxyl radical ( ⅐ OH) are classically considered to be harmful to cells and tissues. However, more recent studies have revealed that ROS function as important signaling molecules in the regulation of cell growth and differentiation (8).
ROS are generated in cells by several pathways. The major sources of intracellular ROS production include mitochondria, various metabolic and detoxifying enzymes, and Nox. Nox isoforms generate ROS in a variety of cells and tissues in response to stimulation with various growth factors or cytokines (9). To date, Nox isoforms include Nox1, Nox2, Nox3, Nox4, Nox5, Duox1, and Duox2 (10). Although they share structural features, each member of the Nox family is thought to play a specific biological role depending on the type of cell or tissue (11,12). The classic Nox isoform Nox2 (also known as gp91 phox ) was first identified in phagocytes (8,13). Nox isoforms including Nox2 have been shown to be expressed in a wide variety of nonphagocytic as well as phagocytic cells and to function in signal transduction (8,11).
The receptor activator of nuclear factor-B ligand (RANKL) induces osteoclast differentiation in a manner dependent on Nox-derived ROS (14,15). The production of superoxide by Nox4 in osteoclasts also contributes to bone resorption (16). Furthermore, intracellular ROS were found to inhibit the proliferation of immature chondrocytes and promote the induction of chondrocyte hypertrophy in cells derived from ataxia telangiectasia mutated knock-out mice, whereas ROS produced by the up-regulation of Nox1 expression were found to be associated with chondrocyte differentiation (17).
The role of ROS generation in chondrogenesis remains to be fully characterized. We show here that intracellular ROS generated by Nox2 and Nox4 are required for differentiation of ATDC5 cells and primary chondrocytes.

EXPERIMENTAL PROCEDURES
ATDC5 Cell Culture-ATDC5 cells were obtained from the Riken Cell Bank (Tsukuba Science City, Japan) and were cultured in a 1:1 (v/v) mixture of DMEM and F-12 (Invitrogen) medium supplemented with 5% FBS (Invitrogen), human transferrin (10 g/ml; Sigma), and 3 ϫ 10 Ϫ8 M sodium selenite (Sigma), as described previously (18). The initial cell density was 6 ϫ 10 4 cells/well in six-well plates (Nunc, Rockilde, Denmark). For induction of differentiation, confluent cells (culture day 0) were cultured in medium supplemented with bovine insulin (10 g/ml, Sigma). The cells were maintained at 37°C under a humidified atmosphere of 5% CO 2 , and the medium was replaced every other day.
Staining with Alcian Blue-The cells were stained with Alcian Blue 8GX (Sigma) as described previously (19). Images of stained cells were recorded using Olympus CKX41. Stained cells were extracted with 6 M guanidine hydrochloride. The absorbance of the extracted dye was measured at 655 nm.
Isolation of RNA and RT-PCR Analysis-Total RNA was extracted from cells with TRIzol reagent (Invitrogen), and 1-g portions of RNA were subjected to RT-PCR using oligo(dT) primer and SuperScript III reverse transcriptase (Invitrogen). RT-PCR analysis of the resulting cDNA preparation was performed using a PCR Premix, Sapphire (Super Bio Co., Seoul, Korea), and the following primers (forward and reverse, respectively): The PCR products were resolved on a 1.5 or 2% agarose gel and were visualized by staining with ethidium bromide.
Real time qPCR analysis was performed using a LightCycler 480 SYBR Green I Master (Roche Applied Science). Primer sequences (forward and reverse, respectively) were as follows: RNA Interference-Stealth siRNAs specific to mouse Nox1, Nox2, or Nox4 were obtained from Invitrogen; their sequences were: siRNA-1, GCU AUG GAG UUU AUC CGC AGG AAU U, and siRNA-2, GCA ACA UUG CUG GUC AUG CAG CAU U for Nox1; siRNA-1, GAA ACU ACC UAA GAU AGC AGU UGA U, and siRNA-2, GGC CUA AGU AUA AUU AUA CUC GAA for Nox2; and siRNA-1, AAG CAA CAU UUG GUG UCC ACU UUA A, and siRNA-2, GGA GAA CAA GAA GAU UGU UGG AUA A for Nox4. A Stealth negative control siRNA was also obtained from Invitrogen. ATDC5 cells were plated in 60-mm dishes and maintained in a 1:1 (v/v) mixture of DMEM/F-12 supplemented with 5% FBS and antibiotics until ϳ60% confluence was achieved. The cells were then transfected with siRNAs (100 nM) for 6 h using Lipofectamine 2000 (Invitrogen) in OPTI-MEM medium (Invitrogen). The medium was then replaced with normal culture medium.
Apoptosis Assay-For quantification of apoptosis, a FITCconjugated annexin V and propidium iodide assay (BD Biosciences) was performed according to the manufacturer's instructions. Analysis of FITC and propidium iodide fluorescence intensities was performed with a FACSCalibur flow cytometer. A total of 10,000 events/sample was acquired.
Immunohistochemistry-For histological examination, the samples were obtained from the forelimbs of CD1 mouse (Orient Bio, Sung-Nam, Korea) embryos at 16.5 days postcoitum. The samples were immediately fixed in 4% (w/v) paraformaldehyde, dehydrated with graded ethanol, and embedded in paraffin. Tissue samples were cut into 5-m-thick slices. According to the manufacturer's protocol (CHEMICON IHC Select immunoperoxidase secondary detection system), the sections were treated with sodium citrate buffer for antigen retrieval, followed by protein blocking to reduce nonspecific binding. The sections were then immunostained with antibodies (Nox1, Nox2, Nox4, and collagen type II in 2% BSA). Control was immunostained with control rabbit IgG in 2% BSA. After immunostaining, the sections were counterstained with hematoxylin. The sections were stained with 0.1% safranin O and 0.03% Fast Green as previously described (22). After mounting with Clarion TM mounting medium (Sigma), the images were recorded using an Olympus BX51 microscope.
Preparation, Electroporation Transfection, and Micromass Culture of Mouse Limb Primary Chondrocytes-Primary chondrocytes were isolated from the forelimbs of CD1 mouse embryos at 11.5 days postcoitum, as described previously (23) with minor modifications. siRNA transfection was performed using a Neon TM transfection system device (Invitrogen), according to the manufacturer's instructions. Briefly, the isolated cells were suspended in R buffer containing siRNAs (siRNA control, Nox2, and Nox4; final concentration, 100 nM), followed by electroporation using one pulse at 1600 V for 20 ms. After incubation for 1 h at 37°C to allow cell attachment, the micromasses were supplied with growth medium (DMEM/F-12 (2:3,v/v) supplemented with 10% FBS). Six hours after transfection, the cells were supplied with ascorbic acid and ␤-glycerophosphate. The medium was replaced daily.
Statistical Analysis-All of the results are expressed as the means Ϯ S.E. from at least three independent experiments. Statistical significance was determined using Student's t test. p values of Ͻ0.05 were considered significant.

ROS Generation Is Accompanied by Insulin-induced ATDC5
Cell Differentiation-To induce chondrogenesis, confluent ATDC5 cells (day 0) were exposed to insulin. The cells were cultured up to 10 days and then stained with Alcian Blue to detect the accumulation of sulfated proteoglycans as a marker of chondrogenic differentiation (24). Exposure of cells to insulin resulted in a time-dependent increase in the intensity of Alcian Blue staining. However, this effect of insulin was blocked in the presence of N-acetylcysteine (NAC), a ROS scavenger (Fig. 1A). RT-PCR analysis revealed that insulin increased the mRNA expression of collagen type II (Col2a1), collagen type X (Col10a1), and aggrecan (Agc1), which are major chondrogenic marker genes, in a time-dependent manner (Fig. 1B). The effects of insulin on these genes were also blocked by NAC. Together, these results suggest that ROS generated in response to insulin stimulation were required for chondrogenic differentiation of ATDC5 cells. When ATDC5 cells were treated with H 2 O 2 , the expression of Sox9, Col2a1, and Col10a1 was also increased in a time-dependent manner for 48 h (supplemental Fig. S1). This phenomenon shows that additional ROS indeed accelerated chondrogenesis. This result is consistent with that of a previous report (17). We next measured the intracellular level of ROS during insulin-induced chondrogenesis in ATDC5 cells. Flow cytometric analysis of cells loaded with the ROS-sensitive indicator DCFH-DA revealed that the intracellular ROS concentration increased within 2 min of cell exposure to insulin (data not shown) and remained high throughout chondrogenesis, even after 10 days (Fig. 1C). To identify the source of ROS produced in response to insulin, we examined the effects of various inhibitors. Similar to NAC, diphenyleneiodonium chloride (DPI), a flavoprotein inhibitor, and apocynin, a Nox inhibitor (9), blocked the insulin-induced accumulation of ROS. In contrast, rotenone and antimycin, which are inhibitors of mitochondrial complexes I and III, respectively (25), had no such effect (Fig. 1D). Although rotenone reduced the basal level of ROS, the generation of ROS by insulin treatment was not affected. These results suggest that Nox was the source of ROS generated in association with insulin-induced chondrogenesis in ATDC5 cells.
Differential Expression of Nox during Differentiation of ATDC5 Cells and Primary Chondrocytes-To determine which members of the Nox family might be responsible for insulin-induced ROS generation in ATDC5 cells, we examined the abundance of various Nox mRNAs by RT-PCR analysis. The genes for Nox1, Nox2, Nox4, and p22 phox as well as those for the cytosolic components Rac1, p40 phox , p47 phox , and p67 phox were expressed in ATDC5 cells ( Fig. 2A). In contrast, expression of Nox3 was only slightly expressed in cultures at day 10, and Nox5 was not detected at all (data not shown). NOX5 was previously shown to not be expressed in mouse (9,26).
The mRNA levels of Nox1, Nox2, p22 phox , p40 phox , p47 phox , and p67 phox were increased during insulin-induced differentiation of ATDC5 cells, with Rac1 slightly increased ( Fig. 2A). In contrast, the amount of Nox4 mRNA, which was relatively high at day 0, was decreased in association with chondrogenic differentiation. These changes in Nox1, Nox2, and Nox4 gene expression were confirmed by real time qPCR analysis (Fig.  2C). The protein expression of Nox1, Nox2, Nox4, and p22 phox was also analyzed by immunoblotting (Fig. 2B).
Using a micromass culture system, we also examined the mRNA expression of Nox during chondrogenesis. Primary chondrocytes derived from mouse embryonic limb buds were prepared as described under "Experimental Procedures." RT-PCR analysis revealed that the mRNA levels of Nox1 and Nox2 were increased in association with chondrogenesis. However, the expression of Nox4 was decreased during chondrogenesis (Fig. 2D). The protein expression of Nox1, Nox2, and Nox4 was determined by immunoblotting (Fig. 2E). These results are consistent with those obtained using ATDC5 cells.
Expression of Nox1, Nox2, and Nox4 in Developing Cartilage-To examine how Nox is expressed in chondrocytes in vivo, we carried out immunostaining of forelimbs isolated from mouse E16.5 using antibodies against Nox1, Nox2, or Nox4. As a positive control, collagen type II was also determined. To distinguish morphologically distinct chondrocyte populations, the sections were stained with safranin O/Fast Green. Consistent with previous reports (27,28), proliferating chondrocytes, prehypertrophic chondrocytes, and hypertrophic chondrocytes were detectable in the limbs at this stage (Fig. 3A). Collagen type II was detected in proliferating and prehypertrophic chondrocytes (Fig. 3B). Both Nox1 and Nox2 were detected in prehypertrophic and hypertrophic chondrocytes (Fig. 3, D and E). In the case of Nox1, it seems to have been slightly more expressed in prehypertrophic chondrocytes than in hypertrophic chondrocytes. The Nox4 signal was more strongly detected in proliferating chondrocytes and prehypertrophic chondrocytes than in hypertrophic chondrocytes (Fig. 3F). Further, expression of Nox was de-tected in the perichondrium (27), surrounded by a layer of fibroblast-like cells.
Nox2 and Nox4 Are Required for ROS Generation in ATDC5 Cells-To investigate the relative roles of Nox1, Nox2, and Nox4 during the chondrogenic differentiation of ATDC5 cells, we transiently introduced two different types of siRNA for each enzyme into the cells. The expression of siRNA-1 and siRNA-2 was reduced by over 50% as demonstrated by RT-PCR and immunoblotting. (Fig. 4, A and B). However, siRNA-2 of each enzyme was more effective, which prompted us to use siRNA-2 in the following experiments. We then performed FACS analysis after knockdown of Nox genes. Whereas cells transfected with control or Nox1 siRNAs experienced a marked increase in intracellular ROS levels 2 days after insulin exposure, those transfected with Nox2 or Nox4 siRNAs did not (Fig. 4C). Furthermore, the basal ROS level in cells transfected with Nox4 siRNA was reduced by 75% compared with that in cells transfected with control siRNA. However, the basal ROS level was not affected by knockdown of Nox2, probably because of the presence of Nox4. As shown in Fig. 2, Nox4 was highly expressed in nonstimulated cells, but Nox2 was not. This suggests that Nox4 was largely responsible for basal ROS production in ATDC5 cells. In Nox4 knockdown cells at day 2, the ROS level was increased over 2-fold compared with that at day 0 because of the presence of Nox2. However, the ROS level was not sufficient for differentiation. Further, the ROS level was not increased in Nox2 knockdown cells at day 2 compared with that at day 0. At day 2, the level of Nox4 had decreased by over 50% (Fig. 2C). Therefore, Nox4 might not be able to overcome FIGURE 1. Effects of ROS scavenger on ATDC5 cell differentiation. A, ATDC5 cells were cultured in the presence of insulin (10 g/ml) with or without NAC (5 mM) for the indicated times, after which the cells were stained with Alcian Blue to detect proteoglycan. The dye was then extracted from stained cells, and the absorbance of the extract was measured at 655 nm; the data are the means Ϯ S.E. from three independent experiments. Significant differences are compared with insulin (**, p Ͻ 0.01; ***, p Ͻ 0.001; insulin versus insulin ϩ NAC). B, total RNA isolated from ATDC5 cells cultured as in A was subjected to RT-PCR analysis for Agc1, Col2a1, Col10a1, and 18S mRNA. The data are representative of three independent experiments. C, ATDC5 cells were cultured in the presence of insulin for up to 10 days. The cells were loaded with DCFH-DA for measurement of intracellular ROS using DCF fluorescence. Significant differences are compared with time 0. *, p Ͻ 0.05; **, p Ͻ 0.01). D, ATDC5 cells were deprived of serum for 12 h and then incubated first for 30 min with apocynin (100 M), DPI (100 nM), NAC (5 mM), antimycin (20 M), or rotenone (5 M) in a serum-free 1:1 (v/v) mixture of DMEM/F-12 and then for 5 min in the presence of additional insulin (10 g/ml). The level of intracellular ROS was then measured as in C. Significant differences are compared with control versus inhibitor as well as insulin versus insulin ϩ inhibitors (*, p Ͻ 0.05; control versus NAC, apocynin and rotenone, ***, p Ͻ 0.001; control versus DPI; *, p Ͻ 0.05; insulin versus insulin ϩ NAC, ϩ DPI, **, p Ͻ 0.01; insulin versus insulin ϩ apocynin). All of the data are the means Ϯ S.E. from three independent experiments. the decrease in ROS by Nox2 knockdown. These results indicate that Nox2 and Nox4, but not Nox1, worked cooperatively for ROS generation by changing their expression during the early phase of insulin-induced differentiation of ATDC5 cells.
Depletion of Either Nox2 or Nox4 Induces ATDC5 Cell Apoptosis-Nox2-or Nox4-depleted ATDC5 cells appeared to be dying during chondrogenesis (data not shown). To examine whether or not the cells underwent apoptosis, the cells were stained with FITC-annexin V and propidium iodide. Upon insulin exposure for 2 days, the frequency of apoptosis was markedly increased in ATDC5 cells depleted of Nox2 or Nox4 compared with cells transfected with control or Nox1 siRNAs (Fig. 5A). Western blot analysis revealed that expression of cleaved caspase 3 was increased by Nox2 or Nox4 knockdown in differentiated ATDC5 cells (Fig. 5B). We next tested whether or not activation of the protein kinases Akt and ERK is related to the apoptosis observed in cells depleted of Nox2 or Nox4. Intracellular MAPKs and Akt are the major oxidative stress-sensitive signal transduction pathways (29). Akt is a key regulator of cell survival in response to growth factor stimulation (30) and promotes matrix synthesis and chondrocyte survival (31). We thus examined the phosphorylation of Akt on Ser 374 . Immunoblot analysis revealed that phosphorylation of Akt was greatly diminished in insulintreated cells depleted of Nox2 or Nox4 compared with that in insulin-treated cells transfected with control or Nox1 siRNAs FIGURE 2. Differential expression of Nox during differentiation of ATDC5 cells and mouse primary chondrocytes. A and C, ATDC5 cells were cultured in the presence of insulin for up to 10 days, after which the expression of Nox1, Nox2, Nox4, Rac1, p22 phox , p40 phox , p47 phox , p67 phox , and 18S was examined by RT-PCR (A) or real time qPCR analysis (C). Quantitative data are the means Ϯ S.E. from three independent experiments. Significant differences are compared with day 0 (***, p Ͻ 0.001; *, p Ͻ 0.05). B, lysates of cells cultured for the indicated times were subjected to immunoblot analysis with antibodies against Nox1, Nox2, Nox4, p22 phox , and ␣-tubulin. D, primary chondrocytes were obtained from mouse embryos at 11.5 days postcoitum and cultured for up to 9 days. Total RNA isolated from primary chondrocytes was subjected to RT-PCR analysis for Nox1, Nox2, Nox4, and ␤-actin mRNA. The data are representative of three independent experiments. E, lysates of cells cultured for the indicated times were subjected to immunoblot analysis with antibodies against Nox1, Nox2, Nox4, and ␤-actin.  (Fig. 5B). According to previous reports (31)(32)(33), the balance between phosphatidylinositol 3-kinase-Akt and MEK-ERK activity plays a vital role in chondrocytes. Interestingly, phosphorylation of ERK in cells transfected with Nox2 or Nox4 siRNA was increased compared with that in cells transfected with control or Nox1 siRNA. These results suggest that ROS generation by Nox2 or Nox4 plays a role in the survival of ATDC5 cells.
Nox2 and Nox4 Are Required for Differentiation in ATDC5 Cells and Primary Chondrocytes-The roles of Nox2 and Nox4 in chondrogenesis were further confirmed by determining the expression of chondrogenic marker proteins in Nox1-, Nox2-, or Nox4-depleted cells. First, ATDC5 cells were transfected with control, Nox1, Nox2, or Nox4 siRNA and then exposed to insulin for 2 days for differentiation. The specific depletion of each Nox enzyme by its corresponding siRNA was verified by RT-PCR and Western blot analysis. Further, the expression of marker proteins such as Sox9, collagen type II, and N-cadherin was decreased in cells depleted of Nox2 or Nox4. In contrast, depletion of Nox1 did not inhibit differentiation but rather appeared to promote it (Fig. 6, A and B).
As described above, Nox2 and Nox4 play an important role in the differentiation of ATDC5 cells. To further confirm the role of Nox2 and Nox4 in chondrogenesis, we also introduced the same siRNAs used in ATDC5 cells into primary chondrocytes by RNAi, as described under "Experimental Procedures." The depletion of Nox2 or Nox4 was analyzed by RT-PCR and immunoblotting. Expression of marker proteins, Sox9, and collagen type II was diminished in Nox2-or Nox4depleted cells (Fig. 6, C and D).
To exclude off-target effects, we investigated the effects of differentiation using siRNA-1 and siRNA-2 of both Nox2 and Nox4 in ATDC5 cells and primary cells. We performed real time qPCR analysis for expression of Col2a1 mRNA. Both siRNA-1 and -2 inhibited ϳ50% of the expression of Col2a1 (Fig. 6E). As shown in Fig. 6F, knockdown of two different types of Nox2 or Nox4 reduced the accumulation of proteoglycan by over 50% (Fig. 6F, left panel) in a micromass culture of primary chondrocytes at day 3. The accumulated proteoglycan was stained with Alcian Blue, which was then extracted and measured for its absorbance at 655 nm for quantification (Fig. 6F, right panel).
Sox9 is a master regulator of cartilage formation (1,34). Adhesion molecules such as N-cadherin are important for the condensation of chondroprogenitor mesenchymal cells (35,36). These results show that Nox2 and Nox4 were important  for the condensation and differentiation of ATDC5 cells and primary chondrocytes.

DISCUSSION
We investigated the association of ROS generated by Nox and chondrogenesis in ATDC5 cells. Among Nox family members, Nox1, Nox2, and Nox4 were found to be expressed in ATDC5 cells during chondrogenesis (Fig. 2). Depletion of either Nox2 or Nox4, but not that of Nox1, inhibited ROS generation, suppressed differentiation, and induced apoptosis in ATDC5 cells (Figs. 5 and 6). These results suggest that ROS generation by Nox2 and Nox4 was required for chondrogenic differentiation of ATDC5 cells.
A previous study on ATDC5 cells revealed a role for ROS in chondrocyte hypertrophy and showed that induction of chondrogenic differentiation is associated with the up-regulation of Nox1 gene expression (17). We now have shown the expression of Nox1 along with that of Nox2 and Nox4 in ATDC5 cells as well as in primary chondrocytes isolated from mouse embryonic limb buds (Fig. 2). Using RNA interference, we also found that Nox2 and Nox4 were required in the early phase of chondrogenic differentiation in ATDC5 cells. The level of intracellular ROS was increased early and maintained during ATDC5 cell differentiation. The expression of the Nox1 and Nox2 genes was increased as differentiation progressed, whereas that of the Nox4 gene, which was initially high, was gradually decreased during differentiation (Fig. 2). By immunohistochemical analysis of developing limbs obtained from mouse embryos, we found that both Nox1 and Nox2 were strongly expressed in prehypertrophic and hypertrophic chondrocytes, whereas the Nox4 signal was observed in proliferating chondrocytes and prehypertrophic chondrocytes (Fig. 3). We also found that Nox2 and Nox4 were important for the condensation and differentiation of primary chondrocytes derived from mouse embryos (Fig. 6).
Nox1 did not appear to contribute to ROS generation during the early phase of ATDC5 cell differentiation, suggesting that the level of intracellular ROS was maintained by Nox2 and Nox4 and that the relative importance of these enzymes increases and decreases, respectively, with changes in their relative levels of expression. It was recently shown that Nox2 and Nox4 each compensate for the deficiency of the other in lung endothelial cells (37).
Down-regulation of Nox4 expression is associated with adipocyte differentiation (38), and Nox4 has been proposed to act as a switch that controls the transition between proliferation and differentiation of preadipocytes (39). Our results now suggest that increasing the expression of Nox1 and Nox2 and decreasing the expression of Nox4 are characteristics of chondrogenesis. Depletion of Nox2 or Nox4 resulted in the downregulation of Sox9 and collagen type II expression in ATDC5 and primary chondrocytes (Fig. 6). Sox9 is an essential chondrogenic transcription factor, and N-cadherin is an important downstream target of Sox9 that modulates cell adhesion (40). Signaling by Rac1, a member of the Rho family of small GT-Pases, is also thought to regulate N-cadherin expression during chondrogenesis (20). Rac1 is also an important cytosolic activator of Nox enzymes (9,12). Although our results suggest that Nox1 does not contribute to the early phase of differentiation in ATDC5 cells, observations that Nox1 expression gradually increased as differentiation progressed suggest that Nox1 may play a role in the later stages of differentiation.
We found that apoptosis was induced in ATDC5 cells depleted of Nox2 or Nox4 (Fig. 5). Apoptosis, a redox-regulated event, is triggered by disruption of the extracellular matrix in adherent cells. Matrix-cell contacts result in the generation of ROS as well as activation of pro-survival signaling (41). Nox2or Nox4-depleted cells also did not undergo condensation or differentiation. Suppression of ROS generation by depletion of Nox2 or Nox4 inhibited the insulin-induced phosphorylation of Akt. Further, ERK was phosphorylated in Nox2 or Nox4 knockdown cells. It is reported that a balance between phosphatidylinositol 3-kinase-Akt and MEK-ERK activity regulates chondrocyte matrix synthesis, and this balance is modulated by ROS (33). ROS generated by Nox have been shown to be important for the survival, proliferation, and differentiation of various cell types, including pancreatic adenocarcinoma, glioma, endothelial, vascular smooth muscle, and embryonic stem cells (37,(42)(43)(44)(45)(46).
Although no bone abnormalities have been reported in Nox1 Ϫ/Ϫ or Nox2 Ϫ/Ϫ mice (14,47,48), it is clear that Noxderived ROS are important for chondrogenesis. Nox2 and Nox4 were expressed in converse fashion (Fig. 2), but knockdown of each exerted a dramatic reductive effect on ROS generation even after 2 days of chondrogenesis (Fig. 4). This critical effect of Nox2 and Nox4 was also observed in the suppression of apoptosis (Fig. 5). However, expression of marker proteins was more reduced in Nox4 knockdown cells than in Nox2 knockdown cells (Fig. 6). This implies that Nox4 was important for the basal state, whereas Nox2 was important for differentiation. Thus, both Nox2 and Nox4 worked cooperatively for ROS generation by changing their expression. In summary, our findings suggest that ROS generated by both Nox2 and Nox4 played key roles in the early stage of chondrogenic differentiation in ATDC5 cells and primary chondrocytes.