FOXO1 transcription factor regulates chondrogenic differentiation through transforming growth factor β1 signaling

The forkhead box O (FOXO) proteins are transcription factors involved in the differentiation of many cell types. Type II collagen (Col2) Cre-Foxo1-knockout and Col2-Cre-Foxo1,3,4 triple-knockout mice exhibit growth plate malformation. Moreover, recent studies have reported that in some cells, the expressions and activities of FOXOs are promoted by transforming growth factor β1 (TGFβ1), a growth factor playing a key role in chondrogenic differentiation. Here, using a murine chondrogenic cell line (ATDC5), mouse embryos, and human mesenchymal stem cells, we report the mechanisms by which FOXOs affect chondrogenic differentiation. FOXO1 expression increased along with chondrogenic differentiation, and FOXO1 inhibition suppressed chondrogenic differentiation. TGFβ1/SMAD signaling promoted expression and activity of FOXO1. In ATDC5, FOXO1 knockdown suppressed expression of sex-determining region Y box 9 (Sox9), a master regulator of chondrogenic differentiation, resulting in decreased collagen type II α1 (Col2a1) and aggrecan (Acan) expression after TGFβ1 treatment. On the other hand, chemical FOXO1 inhibition suppressed Col2a1 and Acan expression without suppressing Sox9. To investigate the effects of FOXO1 on chondrogenic differentiation independently of SOX9, we examined FOXO1's effects on the cell cycle. FOXO1 inhibition suppressed expression of p21 and cell-cycle arrest in G0/G1 phase. Conversely, FOXO1 overexpression promoted expression of p21 and cell-cycle arrest. FOXO1 inhibition suppressed expression of nascent p21 RNA by TGFβ1, and FOXO1 bound the p21 promoter. p21 inhibition suppressed expression of Col2a1 and Acan during chondrogenic differentiation. These results suggest that FOXO1 is necessary for not only SOX9 expression, but also cell-cycle arrest during chondrogenic differentiation via TGFβ1 signaling.

The forkhead box O (FOXO) proteins are a family of transcription factors that play a wide range of roles in lifespan (8,9), apoptosis (10,11), and cell differentiation (12)(13)(14)(15). In mammals, the FOXO family has four members: FOXO1, FOXO3, FOXO4, and FOXO6 (16). FOXO1, FOXO3, and FOXO4 are expressed in nearly all tissues, whereas FOXO6 expression is largely restricted to neural cells (17). Recently, several reports described the roles of FOXOs in articular cartilage and mature chondrocytes. Expression and activity of FOXO1 and FOXO3 decrease with aging, resulting in osteoarthritis due to the consequent reduction in the expression of antioxidant and autophagy-related proteins (18,19). In addition, investigations of the roles of FOXOs in cartilage and bone using Foxo-knock-out mice revealed that Col2-Cre-Foxo1-knockout and Col2-Cre-Foxo1,3,4 triple-knockout mice exhibit growth plate malformation (20,21). These findings indicate that FOXOs can regulate chondrogenic differentiation, but the specific contribution of FOXOs to this process remains to be clarified.
TGF␤1 is one of the most important factors involved in chondrogenic differentiation. TGF␤1 binds its type I and II receptors on the cell surface; the receptors then phosphorylate SMAD2 and SMAD3, which form a complex with SMAD4. The complex translocates to the nuclei, where it regulates a variety of target genes (22). TGF␤1/SMAD signaling promotes the gene expression of SOX9, collagen type II ␣1 (COL2A1), and ACAN (4) (23)(24)(25). Importantly, TGF␤1 regulates the expression and nucleus localization of FOXOs (14,18,26). TGF␤1 may also regulate the expression and activity of FOXOs during chondrogenic differentiation.
FOXOs are also cell-cycle regulators. Cell-cycle arrest in the G 0 /G 1 phase is required for differentiation of many cell types (27). Previous studies reported that p21, a cyclin-dependent kinase inhibitor, is involved in chondrogenic differentiation (28,29), but the mechanism of regulation of the cell cycle during chondrogenic differentiation remains unclear. FOXOs promote the expression of some cyclin-dependent kinase inhibitors and induce cell-cycle arrest (30 -34). Therefore, we hypothesized FOXOs regulate the cell cycle during chondrogenic differentiation.
In this study, we investigated the expression and roles of FOXOs during chondrogenic differentiation. We also verified the effects of TGF␤1 as a regulator of FOXOs. Finally, we investigated the influence of FOXOs on the cell cycle during chondrogenic differentiation.

FOXO1 expression increases along with chondrogenic differentiation in ATDC5
First, we confirmed the gene expression patterns of Sox9, Col2a1, and Acan (as chondrogenic differentiation markers) and collagen type X ␣1 (Col10a1) (as a hypertrophic differentiation marker) during chondrogenic differentiation of ATDC5 cells, an in vitro model of chondrogenic differentiation (35,36). To induce chondrogenic differentiation, ATDC5 cells were incubated in medium containing 1% insulin-transferrinselenium (ITS). Expression of Sox9 increased from day 4 in a time-dependent manner, and that of Col2a1 and Acan increased from day 7 (Fig. 1A). Subsequently, expression of Col10a1 increased on day 14. We then evaluated the gene expression of Foxo1, Foxo3, and Foxo4 in these cells. Expression of Foxo1 started to increase on day 4 in the same manner as Sox9 over the course of chondrogenic differentiation, whereas expression of Foxo3 and Foxo4 was nearly unchanged (Fig. 1B). In addition, ITS treatment significantly induced FOXO1 protein expression in ATDC5 in a time-dependent manner (Fig.  1C).

Inhibition of FOXO1 suppresses chondrogenic differentiation in ATDC5
To investigate the effects of FOXO1 on the gene expression of chondrogenic differentiation markers, we differentiated ATDC5 cells transfected with siRNA targeting Foxo1. Gene and protein expression of FOXO1 were significantly decreased by siRNA relative to the control for up to 7 days ( Fig. 2A). FOXO1 knockdown significantly decreased the expression of Col2a1 from day 4, and the expression of Sox9 and Acan on day 7 ( Fig. 2A).
For longer inhibition of FOXO1 up to 28 days, we used AS1842856, a chemical compound that inhibits FOXO1 activity by binding to FOXO1 (37). The gene expression of Foxo1 was not affected by AS1842856 treatment. Inhibition of FOXO1 activity significantly decreased the expression of Col2a1 and Acan on days 4, 7, and 14 (Fig. 2B). Expression of Sox9 was slightly decreased on day 14. Expression of Col10a1 was significantly decreased on days 7 and 14 (Fig. 2B). Production of proteoglycan stained by Alcian blue was significantly suppressed by AS1842856 at all points in time (Fig. 2, C and D).

FOXO1 regulates chondrogenic differentiation in mesenchymal cells from mouse limb buds
To validate the effects of FOXO1 on chondrogenic differentiation in primary cells, we undertook micromass culture of mesenchymal cells from limb buds derived from mouse embryos at embryonic day 11.5 (E11.5), which have been used as an in vitro model of chondrogenic differentiation (38). Expression of Foxo1 mRNA and FOXO1 protein significantly increased from day 3 (Fig. 3, A and B). Mesenchymal cells formed nodules of differentiated chondrocytes, which were positively stained by Alcian blue on day 5 (Fig. 3C). In immunocytochemistry, FOXO1 protein was strongly expressed in differentiated nodules (Fig. 3D). FOXO1 inhibition by AS1842856 considerably suppressed proteoglycan production stained by Alcian blue in micromass culture on day 5 (Fig. 3E). Consistently, gene expression of Sox9, Col2a1, and Acan on day 5 were significantly decreased by AS1842856 treatment (Fig. 3F).

FOXO1 regulates chondrogenic differentiation in human mesenchymal stem cells
As an additional chondrogenic differentiation model using primary cells, we applied pellet culture of human mesenchymal stem cells (hMSCs) in the presence or absence of TGF␤1. Incubation in the presence of TGF␤1 for 21 days induced proteoglycan production stained by Alcian blue (Fig. 4A). In immunohistochemistry, FOXO1 protein was strongly expressed in the pellets incubated in chondrogenic medium including TGF␤1 (Fig. 4A). Gene and protein expression of FOXO1 was significantly increased during the culture in chondrogenic medium in a time-dependent manner (Fig. 4, B and C). FOXO1 inhibition by AS1842856 significantly decreased the production of proteoglycan in the pellets with chondrogenic medium on day 21 (Fig. 4D). Expression of SOX9, COL2A1, and ACAN on day 21 was significantly decreased by AS1842856 (Fig. 4E).

TGF␤1/SMAD signaling promotes expression and activity of FOXO1
For mechanistic analyses, we investigated the effects of TGF␤1 on the expression and activity of FOXO1 during chondrogenic differentiation in ATDC5 cells. Gene expression of

FOXO1 regulates chondrogenic differentiation
Tgfb1 increased from day 4 over the course of chondrogenic differentiation (Fig. 5A). Among Foxo1, Foxo3, and Foxo4, TGF␤1 treatment specifically induced expression of Foxo1 (Fig.  5B). Western blotting revealed that expression of total FOXO1 protein significantly increased upon incubation with TGF␤1 for more than 4 h (Fig. 5C). Meanwhile, the level of phosphorylated FOXO1 (pFOXO1) protein at Ser-253, a major cytoplasmic form regulated via the phosphatidylinositol-3-OH kinase (PI3K)/AKT pathway, was not affected by TGF␤1 (Fig. 5C). FOXO1 protein both in the cytoplasm and nucleus significantly increased upon incubation with TGF␤1 (Fig. 5D). By immunohistochemistry, we confirmed an increase in the level of nuclear FOXO1 after incubation with TGF␤1 in a time-dependent manner (Fig. 5E).
TGF␤1/SMAD signaling is the canonical TGF␤1 pathway. Hence, ATDC5 cells were transfected with siRNA targeting Smad3, to determine whether SMAD3 affects the expression of FOXO1 following TGF␤1 treatment. SMAD3 knockdown suppressed the induction of FOXO1 protein by TGF␤1 in Western blotting and immunocytochemistry (Fig. 5, F and G). In addi-tion, we used SIS3, a chemical inhibitor of SMAD3 (39). Inhibition of SMAD3 by SIS3 suppressed the induction of FOXO1 protein by TGF␤1 in ATDC5 (Fig. 5, H and I). SIS3 completely suppressed gene expression of Foxo1, as well as chondrogenic differentiation markers, over the course of chondrogenic differentiation (Fig. 5J). TGF␤1 treatment continued to promote FOXO1 protein expression up to 21 days, whereas the absence of TGF␤1 did not increase FOXO1 at any point in time (Fig.  5K). These results indicate that TGF␤1/SMAD signaling is essential for promoting the expression and activity of FOXO1 during chondrogenic differentiation.

Inhibition of FOXO1 suppresses Col2a1 and Acan expression via TGF␤1 signaling
To investigate the effects of FOXO1 on the gene expression of chondrogenic differentiation markers following TGF␤1 treatment, ATDC5 cells transfected with siRNA targeting Foxo1 or treated with AS1842856 were stimulated with TGF␤1. TGF␤1-induced Sox9, Col2a1, and Acan expression were significantly decreased by knockdown of Foxo1 (Fig. 6A). On the Foxo4 (B) in differentiating ATDC5 cells cultured for 3 weeks were measured by qRT-PCR. Gene expression at each stage is shown relative to the level on day 0; n ϭ 3. C, levels of FOXO1 protein after incubation with or without 1% ITS for 3 weeks, as determined by Western blotting. The graph shows FOXO1 protein level relative to the level on day 0; n ϭ 4. Data are presented as mean Ϯ S.D. Statistical analysis was performed using Wilcoxon's rank-sum test. *, p Ͻ 0.05.

FOXO1 promotes cell-cycle arrest by regulating p21 expression during chondrogenic differentiation
For further mechanistic analyses, we investigated the effects of FOXO1 on the cell cycle during chondrogenic differentiation in ATDC5 cells. The percentage of cells in the G 0 /G 1 phase significantly increased 4 days after the initiation of chondrogenic differentiation (Fig. 7A). Inhibition of FOXO1 by AS1842856 significantly decreased the percentage of cells in G 0 /G 1 phase on day 4 (Fig. 7A). Consistently, the number of cells on days 2 and 3 in the presence of AS1842856 was significantly increased compared with control ( Fig. 7B).
In gene expression of cell-cycle inhibitors, including p21, p27, and cyclin G2, inhibition of FOXO1 by AS1842856 significantly suppressed the expression of p21 and cyclin G 2 on day 4 (Fig. 7C). Next, we transfected the cells with plasmids overexpressing constitutively active FOXO1-ADA, a mutant in which all three AKT phosphorylation sites have been replaced by nonphosphorylatable amino acids (T24A, S253D, and S316A). Western blotting confirmed that exogenous FOXO1-ADA accumulated in the nucleus (Fig. 7D). FOXO1-ADA overexpression increased the gene level of p21, but not p27 or cyclin G 2 (Fig. 7D). FOXO1-ADA overexpression increased p21 at the protein level by Western blotting (Fig. 7E). Furthermore, FOXO1-ADA overexpression increased the percentage of cells in the G 0 /G 1 phase (Fig. 7F). These results indicate that FOXO1 promotes the expression of p21 and cell-cycle arrest during chondrogenic differentiation.

FOXO1 mediates p21 expression and chondrogenic differentiation via TGF␤1 signaling
Among its many functions, TGF␤1 is a cell-cycle regulator (40). Hence, we evaluated the effect of TGF␤1 on the gene expression of p21, p27, and cyclin G 2 in ATDC5 cells. TGF␤1 promoted the expression of p21, but had no effect on p27 or cyclin G 2 (Fig. 8A), reminiscent of the effects of FOXO1-ADA overexpression.
Next, we investigated whether FOXO1 affected the expression of p21 promoted by TGF␤1. Knockdown of Foxo1 by siRNA significantly suppressed the gene expression of p21 induced by TGF␤1 (Fig. 8B). Western blotting also revealed that FOXO1 knockdown suppressed the induction of p21 protein by TGF␤1 (Fig. 8C).
To evaluate the direct effect of FOXO1 on the expression of p21 at the transcription level, nascent RNA analysis and ChIP assay were performed. After TGF␤1 treatment for 24 h, ATDC5 cells were treated with AS1842856 for 30 min, and then extracted nascent RNA was synthesized during the following 1-h period (Fig. 8D). qRT-PCR revealed that inhibition of FOXO1 significantly decreased the expression of nascent p21 RNA (Fig. 8D). In ChIP assay, there was a FOXO-binding site (TGTTTAC) in the promoter region of p21, and FOXO1 bound the promoter of p21 after TGF␤1 treatment (Fig. 8E). In Fig. 8E, a positive band of PCR product was found in the immunoprecipitation lane by anti-FOXO1 antibody in cells treated with TGF␤1.

Figure 2. FOXO1 inhibition suppresses chondrogenic differentiation.
A, ATDC5 cells were differentiated for 1 week after transfection with siControl or siFOXO1. Levels of FOXO1 protein on days 0, 4, and 7 were determined by Western blotting. Relative mRNA levels of Foxo1, Sox9, Col2a1, Acan, and Col10a1 were measured by qRT-PCR. Gene expression at each stage is expressed relative to the level on day 0 in cells transfected with siControl; n ϭ 4. B, ATDC5 cells were differentiated for 2 weeks after incubation with or without AS1842856 (0.1 M) for 24 h. Relative mRNA levels of Foxo1, Sox9, Col2a1, Acan, and Col10a1 were measured by qRT-PCR. Gene expression at each stage is expressed relative to the level on day 0 in cells incubated without AS1842856; n ϭ 3. C, ATDC5 cells were differentiated for 4 weeks after incubation with or without AS1842856 (0.1 M) for 24 h, and Alcian blue staining was performed. D, Alcian blue-stained cultures were extracted and measured at 620 nm; n ϭ 4. Data are presented as mean Ϯ S.D. Statistical analysis was performed using Wilcoxon's rank-sum test. *, p Ͻ 0.05. The graph below shows measurements of Alcian blue staining at 620 nm. Bars represent 300 m; n ϭ 4. F, micromass cultures were incubated with or without AS1842856 (0.1 M) for 5 days. Relative mRNA levels of Sox9, Col2a1, and Acan were measured by qRT-PCR. Gene expression is given relative to the level in cultures incubated without AS1842856; n ϭ 4. Data are presented as mean Ϯ S.D. Statistical analysis in A and B was performed using one-way repeated measures ANOVA with the Tukey-Kramer post hoc test. Statistical analysis in E and F was performed using Wilcoxon's rank-sum test. *, p Ͻ 0.05.

FOXO1 regulates chondrogenic differentiation
Finally, we evaluated the effects of p21 on chondrogenic differentiation. To this end, we differentiated ATDC5 cells in the presence or absence of UC2288, a specific inhibitor of p21 (41). UC2288 treatment suppressed the expression of p21 on days 7 and 14. Inhibition of p21 suppressed the expression of Col2a1 and Acan on days 4, 7, and 14, although the expression of Sox9 was decreased only on day 14. Expression of Col10a1 decreased on days 7 and 14 (Fig. 8F). These results indicate that p21 is necessary for expression of COL2 and ACAN during chondrogenic differentiation in ATDC5.

Discussion
In this study, we revealed the effects of FOXOs on chondrogenic differentiation. To examine this issue, we used the ATDC5 cell line, an in vitro model of chondrogenic differentiation. ATDC5 cells can be induced to undergo chondrogenic differentiation by insulin treatment (35). We showed that of the three FOXOs expressed in these cells, gene expression of Foxo1 increased over the course of chondrogenic differentiation, whereas expression of Foxo3 and Foxo4 was nearly unchanged. A recent study showed that Col2-Cre-Foxo1-knockout and Col2-Cre-Foxo1,3,4 triple-knockout mice exhibit growth plate malformation, whereas Col2-Cre-Foxo3-knockout and Col2-Cre-Foxo4 knockout mice do not (20). These findings suggest that of these three FOXOs, FOXO1 is directly involved in chondrogenic differentiation. Indeed, inhibition of FOXO1 expression and activity suppressed gene expression of Col2a1, Acan, and Col10a1 over the course of chondrogenic differentiation.

FOXO1 regulates chondrogenic differentiation
Interestingly, FOXO1 knockdown suppressed the expression of Col2a1, but not Sox9, on day 4. Chemical inhibition of FOXO1 activity also suppressed the expression of Col2a1 and Acan, but not Sox9, on days 4 and 7. Furthermore, inhibition of FOXO1 activity suppressed the production of pro-teoglycan. In addition, we showed that FOXO1 was required for chondrogenic differentiation in mesenchymal cells from mouse limb buds. In particular, it is important that FOXO1 was expressed in differentiated nodules but not in the undifferentiated cells surrounding nodules. This strongly

FOXO1 regulates chondrogenic differentiation
indicates that FOXO1 is necessary for chondrogenic differentiation.
Insulin and insulin-like growth factor-1 (IGF-1) are the main negative regulators of FOXOs. Insulin and IGF-1 promote phosphorylation of FOXOs through the PI3K/AKT pathway (42). Phosphorylation of FOXOs causes their nuclear export and ubiquitination-mediated degradation (43). However, we showed that insulin did not suppress the expression of FOXO1 during chondrogenic differentiation. Rather, the FOXO1 protein level was significantly increased by insulin treatment. Therefore, we hypothesized that other factors might regulate the expression and activity of FOXO1 during chondrogenic differentiation. Previous studies reported the effects of TGF␤1 on chondrogenic differentiation of ATDC5 cells (44,45). In addition, several studies reported that TGF␤1 regulates the expression and nuclear localization of FOXO1. TGF␤1 promotes the expression of FOXO1 protein in human mature chondrocytes, but has no effect on the level of pFOXO1 (18). That report also demonstrated that the levels of total FOXO1 and pFOXO1 were not influenced by IGF-1 treatment (18). TGF␤1 also promotes the expression and nuclear localization of FOXO1 in cardiac myofibroblasts and regulates their differentiation (14). In this study, we also confirmed that expression of FOXO1 increased along with TGF␤1-induced chondrogenic differentiation in hMSCs, and that inhibition of FOXO1 suppressed chondrogenic differentiation. Hence, we investigated the mechanism of regulation of FOXO1 by TGF␤1 in ATDC5 cells. First, we confirmed that gene expression of Tgfb1 increased along with chondrogenic differentiation by insulin treatment. Furthermore, we showed that TGF␤1 promoted gene expression of Foxo1, but had no effect on Foxo3 or Foxo4. This is consistent with the observation that ATDC5 cells differentiate in response to insulin. TGF␤1 also promoted the expression of total FOXO1 protein, whereas expression of pFOXO1 at Ser-253 was unchanged. This indicates that TGF␤1 regulates the expression of FOXO1 without going through the PI3K/AKT pathway. We also showed that TGF␤1 promoted the expression of nuclear FOXO1 protein by Western blotting and immunocytochemistry. These results indicate that TGF␤1 promotes not only expression of FOXO1, but also its activity, by stimulating its nuclear translocation. On the other hand, TGF␤1 also promoted the expression of cytoplasmic FOXO1 protein. Con- Gene expression is shown relative to the level in cells incubated without TGF␤1; n ϭ 4. C, levels of total FOXO1 and phosphorylated FOXO1 (pFOXO1) after incubation with TGF␤1 (10 ng/ml) for 1, 4, 8, or 24 h, as determined by Western blotting. Graphs below show total FOXO1 protein level and pFOXO1 protein level, relative to that in cells incubated without TGF␤1; n ϭ 5. D, levels of cytoplasmic and nuclear FOXO1 protein after incubation with TGF␤1 (10 ng/ml) for 4, 8, or 24 h, as determined by Western blotting. Graphs below show cytoplasmic FOXO1 protein level and nuclear FOXO1 protein level, relative to the corresponding levels in cells incubated without TGF␤1. GAPDH was used as a loading control for cytoplasmic extracts. Lamin B1 was used as a loading control for nuclear extracts; n ϭ 4. E, FOXO1 localizations after incubation with TGF␤1 (10 ng/ml) for 24 h were visualized by immunocytochemistry using anti-FOXO1 antibody and Alexa Fluor 568-conjugated secondary antibody (red staining). Nuclei were detected with DAPI (blue staining). Bars represent 50 m. ATDC5 cells were incubated with TGF␤1 (10 ng/ml) for 24 h after transfection with siControl or siSMAD3. FOXO1 and SMAD3 levels were determined by Western blotting (F) and immunocytochemistry (G). In G, SMAD3 is shown in green, and FOXO1 is shown in red. Bars in G represent 50 m. ATDC5 cells were incubated with TGF␤1 (10 ng/ml) and SIS3 (3 M) for 24 h. Levels of FOXO1 protein were determined by Western blotting (H) and immunocytochemistry (I). Bars in I represent 50 m. J, ATDC5 cells were differentiated for 2 weeks after incubation with or without SIS3 (3 M) for 24 h. Relative mRNA levels of Foxo1, Sox9, Col2a1, and Acan were measured by qRT-PCR. Gene expression at each stage is shown relative to the level on day 0 in cells incubated without SIS3; n ϭ 3. K, levels of FOXO1 protein after incubation with or without TGF␤1 (10 ng/ml) for 3 weeks, as determined by Western blotting. Data are presented as mean Ϯ S.D. Statistical analysis in B and J was performed using Wilcoxon's rank-sum test. Statistical analysis in C and D was performed using one-way repeated measures ANOVA with the Tukey-Kramer post hoc test. *, p Ͻ 0.05.

FOXO1 regulates chondrogenic differentiation
sidering the result that TGF␤1 promoted the expression of Foxo1 mRNA, our results indicate that TGF␤1 could promote the expression of FOXO1 at the transcriptional level. In addition, we investigated the effects of SMAD signaling on the expression and activity of FOXO1. SMADs are co-factors of FOXOs; these proteins form a transcriptional complex that regulates their target genes (31, 32). However, little is known about the effect of SMADs on the expression and activity levels of FOXOs. In this study, we showed that inhibition of SMAD3 suppressed the expression of FOXO1 protein following TGF␤1 treatment. Inhibition of SMAD3 also suppressed the gene expression of Foxo1, as well as chondrogenic differentiation markers, over the course of chondrogenic differentiation. Thus, we showed that TGF␤1 promoted expression and activity of FOXO1 through SMAD signaling during chondrogenic differentiation.  1 g). Relative mRNA levels of p21, p27, and cyclin G 2 in FOXO1-ADA-transduced ATDC5 cells were measured by qRT-PCR. Gene expression at each stage is expressed relative to the level in pEGFP-transfected cells; n ϭ 4. E, levels of p21 protein in ATDC5 cells transfected with empty pEGFP or FOXO1-ADA-GFP (0.5 g), as determined by Western blotting. F, cell-cycle profiles in ATDC5 cells transfected with empty pEGFP or FOXO1-ADA-GFP (0.5 g) were analyzed by flow cytometry. Percentages of cells in G 0 /G 1 , S, and G 2 were calculated; n ϭ 4. Data are presented as mean Ϯ S.D. Statistical analysis in A was performed using the Tukey-Kramer test. Statistical analysis in B-D, and F was performed using Wilcoxon's rank-sum test. *, p Ͻ 0.05. Figure 8. FOXO1 mediates p21 expression and chondrogenic differentiation via TGF␤1 signaling. A, relative mRNA levels of p21, p27, and cyclin G 2 after incubation with TGF␤1 (10 ng/ml) for 24 h. Gene expression at each stage is shown relative to the level in cells incubated without TGF␤1; n ϭ 4. B, ATDC5 cells were incubated with TGF␤1 (10 ng/ml) for 24 h after transfection with siControl or siFOXO1. Relative mRNA levels of p21 were measured by qRT-PCR. Gene expression is given relative to the level in cells transfected with siControl; n ϭ 4. C, ATDC5 cells were incubated with or without TGF␤1 (10 ng/ml) for 24 h after transfection with siControl or siFOXO1. Levels of FOXO1 and p21 protein were determined by Western blotting. D, nascent RNA was extracted from ATDC5 cells incubated with EU (0.5 mM) for 1 h after incubation with AS18428568 (1 M) for 30 min following TGF␤1 treatment (10 ng/ml) for 24 h. Expression of nascent p21 RNA was measured by qRT-PCR. Gene expression is shown relative to the level in cells incubated without AS18428568; n ϭ 3. E, scheme of a FOXO-binding site in the p21 promoter region. ChIP assay was performed using anti-FOXO1 antibody or control IgG in ATDC5 cells incubated with or without TGF␤1 (5 ng/ml) for 24 h. Semi-quantitative RT-PCR was performed using primers for the p21 promoter (from Ϫ2304 to Ϫ2020 bp) and negative control (from Ϫ2968 to Ϫ2636 bp). F, ATDC5 cells were differentiated for 2 weeks after incubation with or without UC2288 (2.5 M) for 24 h. Relative mRNA levels of Foxo1, p21, Sox9, Col2a1, Acan, and Col10a1 were measured by qRT-PCR. Gene expression at each stage is shown relative to the level on day 0 in cells incubated without UC2288; n ϭ 4. Data are presented as mean Ϯ S.D. Statistical analysis was performed using Wilcoxon's rank-sum test. *, p Ͻ 0.05.

FOXO1 regulates chondrogenic differentiation
TGF␤/SMAD signaling promotes expression (23) and stabilization (46,47) of SOX9, which is a master regulator of chondrogenic differentiation. SOX9 directly binds the promoter and enhancer elements of Col2a1 and Acan (4,24,25). SMAD2/3 activated by TGF␤ signaling also forms a transcriptional complex with SOX9 and stimulates expression of Col2a1 (23). In this study, we showed that FOXO1 knockdown suppressed TGF␤1-induced Sox9 expression in ATDC5 cells. Chemical inhibition of FOXO1 activity also decreased the expression of Sox9 in mouse limb bud micromass cultures and pellets of hMSCs after induction of chondrogenic differentiation. These results indicate that FOXO1 is required for the expression of SOX9 via TGF␤1 signaling during chondrogenic differentiation. On the other hand, our results also showed that inhibition of FOXO1 by a chemical inhibitor suppressed the induction of Col2a1 and Acan by TGF␤1, although Sox9 was slightly increased in ATDC5 cells. This is consistent with the observation that inhibition of FOXO1 suppressed gene expression of Col2a1 and Acan, but not Sox9, during chondrogenic differentiation in response to insulin treatment. These findings suggest that FOXO1 mediates the expression of COL2 and ACAN via TGF␤1 signaling, not only by regulating the expression of SOX9, but also by other mechanisms.
FOXOs are also important regulators of the cell cycle. In particular, FOXOs induce cell-cycle arrest by promoting expression of p21, p27, and cyclin G 2 (30 -34). Generally, cell differentiation requires cell-cycle arrest in the G 0 /G 1 phase (27). In this study, we observed cell-cycle arrest in the G 0 / G 1 phase during chondrogenic differentiation; inhibition of FOXO1 suppressed the arrest. In addition, we showed that inhibition of FOXO1 suppressed the gene expression of p21, on the other hand, FOXO1 overexpression promoted the expression of p21, and induced cell-cycle arrest in the G 0 /G 1 phase. These results indicate that FOXO1 promotes cell-cycle arrest in the G 0 /G 1 phase by regulating p21 expression during chondrogenic differentiation. A previous study reported that the height of the proliferative zone is higher in Col2-Cre-Foxo1knockout and Col2-Cre-Foxo1,3,4 triple-knockout mice than in their WT counterparts (20). These in vivo observations support our findings.
TGF␤/SMAD signaling also induces cell-cycle arrest (40). In glioblastoma cells, SMAD2 and SMAD3 form a complex with FOXOs and induce expression of p21 (31). We found that FOXO1 knockdown suppressed the induction of p21 by TGF␤1. Although this finding indicates that FOXO1 mediates the expression of p21 via TGF␤1 signaling, there are some problems with accurately evaluating the effect of FOXO1 on the expression of p21 via TGF␤1 signaling. First, previous studies reported that both of p21 mRNA and p21 protein have short half-lives (48,49). In addition, FOXO1 has so many target genes that inhibition of FOXO1 for a long time may affect many genes other than p21, resulting in indirect effects on the expression of p21. Therefore, we needed to examine the effects of short-term inhibition of FOXO1. To resolve these problems, we extracted nascent RNA synthesized during the following 1-h period only 30 min after AS1842856 treatment. Nascent RNA is useful for distinguishing transcriptional from post-transcriptional gene regulatory effects (50). As a result, expression of nascent p21 RNA in response to TGF␤1 signaling was decreased by inhibition of FOXO1. This indicates that FOXO1 regulates the expression of p21 at the transcriptional level. In other cells, p21 is a direct target of FOXOs (31,51). In this study, we also found a FOXO-binding site (TGTTTAC) in the promoter region of p21, and FOXO1 bound the promoter of p21 in the ChIP assay, suggesting that FOXO1 directly regulates p21 expression in ATDC5 cells. p21 is necessary for chondrogenic differentiation (28,29). Here, we also showed that inhibition of p21 suppressed expression of Col2a1 and Acan without suppressing expression of Sox9, reminiscent of the effects of FOXO1 inhibition. These data suggest that FOXO1 mediates the expression of p21 via TGF␤1 signaling, and that p21 is required for expression of COL2 and ACAN during chondrogenic differentiation.
In summary, we demonstrated that expression of FOXO1 increases over the course of chondrogenic differentiation. TGF␤1 increases the expression and activity of FOXO1 through SMAD signaling, and FOXO1 is required for the expression of COL2 and ACAN in response to TGF␤1 signaling. We also showed that FOXO1 promotes cell-cycle arrest in the G 0 /G 1 phase by mediating the expression of p21 via TGF␤1 signaling and regulates the expression of COL2 and ACAN during chondrogenic differentiation. We suggest that TGF␤1 promotes not only SOX9 expression, but also cell-cycle arrest, by regulating FOXO1, which is necessary for the expression of COL2 and ACAN.
E11.5 fore-and hindlimb buds were collected in PBS. Limb buds were incubated in PBS containing 0.1% trypsin, EDTA (0.4 mM), and 0.1% collagenase at 37°C for 20 min. Mesenchymal cells were resuspended in DMEM/Ham's F-12 medium with 10% FBS at 2 ϫ 10 7 cells/ml. Ten microliters of this suspension was placed in the center of a well in 24-well plates, which was then incubated for 3 h at 37°C and 5% CO 2 to allow cell attachment. Subsequently, 500 l of DMEM/Ham's F-12 medium with 10% FBS and L-ascorbic acid (50 g/ml) (Sigma) was gently added to the plate. All animal experiments were approved by the Animal Experiment Committee of Kyushu University. The experiments were performed under the guidelines of our institution.

Chemical inhibitor treatment
AS1842856 (Merck Millipore, Billerica, MA) was used to inhibit the activity of FOXO1. SIS3 (Sigma) was used to inhibit the activity of SMAD3. UC2288 (Merck Millipore) was used to inhibit the expression of p21. All inhibitors were dissolved in dimethyl sulfoxide (DMSO) (Wako Pure Chemical Industries, Osaka, Japan), and control groups were incubated with an equivalent amount of DMSO lacking inhibitor. The cytotoxicity of these inhibitors was evaluated using CellTiter-Glo (Promega, Madison, WI). In brief, ATDC5 cells, mesenchymal cells from mouse limb buds, and hMSCs were seeded in 96-well plates at a density of 0.5 ϫ 10 4 cells/well. After 2 days, each inhibitor was added to each well, and cell viabilities were measured. None of the inhibitors decreased cell viability. The results are shown in Fig. S1.

siRNA transfection
ATDC5 cells were seeded in 12-well plates at a density of 0.5 ϫ 10 5 cells/well. After the cells reached subconfluence, they were transfected with siRNAs (5 nM) against Foxo1 or Smad3 (Santa Cruz Biotechnology, Dallas, TX) for 6 h using Lipofectamine RNAiMAX (Invitrogen). After 2 days, the cells were used in experiments.

Plasmid transfection
ATDC5 cells were seeded in 12-well plates at a density of 0.5 ϫ 10 5 cells/well. After the cells reached subconfluence, they were transfected with a plasmid (0.1 or 0.5 g) encoding constitutively active FOXO1 (FOXO1-ADA-GFP; Addgene plasmid number 35640) (52) or a control pEGFP-N1-FLAG plasmid. Transfections were performed using Lipofectamine 3000 (Invitrogen). One day after transfection, culture medium was replaced with DMEM/Ham's F-12 medium supplemented with 5% FBS. After 1 day, the cells were used in experiments.

Total RNA extraction and quantitative real-time RT-PCR
Total RNA was extracted from ATDC5 cells and limb bud micromass cultures using the TRIzol reagent (Invitrogen). Total RNA from pellets of hMSCs was extracted from three pellets of each replicate for each condition. Briefly, pellets were pooled in TRIzol reagent, crushed with a mini pestle, and lysed in TRIzol reagent. Total RNA was reverse-transcribed to cDNA using the PrimeScript RT reagent kit (Takara Bio, Kusatsu, Japan). Quantitative real-time RT-PCR was performed on a Light Cycler 2.0 System (F. Hoffmann-La Roche AG, Basel, Switzerland) using SYBR Premix EX TaqII (Takara Bio). The sequences of the forward and reverse primers were shown in Table S1. Data were normalized against the corresponding levels of mouse 18S rRNA or human GAPDH, a housekeeping gene.

Nascent RNA analysis
ATDC5 cells were in pre-treated with mouse TGF␤1 (R&D Systems) treatment (10 ng/ml) for 24 h, incubated with AS18428568 (1 M) for 30 min, and then treated with the uridine analog 5-ethynyl uridine (EU) (0.5 mM) for 1 h, at which time nascent RNA was extracted using the Click-iT Nascent RNA Capture kit (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer's instructions. Briefly, after total RNA was extracted with TRIzol reagent, EU-labeled nascent RNA was biotinylated with biotin azide (0.5 mM). EU-labeled nascent RNA was purified using MyOne Streptavidin T1 magnetic beads. Nascent RNA was reverse-transcribed to cDNA using the PrimeScript RT reagent kit and quantitative real-time RT-PCR was performed.

Western blotting
Whole-cell lysates were extracted from ATDC5 cells and limb bud micromass cultures using the Cell Lytic M (Sigma) with protease inhibitor (cOMplete TM Mini; Sigma) and phosphatase inhibitor (PhosSTOP; Roche Diagnostics, Mannheim, Germany). Whole-cell lysates from pellets of hMSCs were extracted from 2 pellets of each replicate for each condition. Briefly, pellets were pooled in Cell Lytic M, crushed with a mini pestle, and lysed in Cell Lytic M. Nuclear and cytoplasmic extracts were isolated using nuclear and cytoplasmic extraction reagents (Thermo Fisher Scientific). Cell lysates were electrophoresed in 4 -12% gradient polyacrylamide gels (Invitrogen), and the resolved proteins were transferred to nitrocellulose membranes (Amersham Biosciences). Membranes were blocked with blocking buffer (Takara Bio), washed in TBS with Tween (TBST), and incubated with primary antibodies against FOXO1 (number 2880; Cell Signaling Technology, Danvers, MA), pFOXO1 (number 9461; Cell Signaling Technology), SMAD3 (number 9523; Cell Signaling Technology), p21 (ab109199; Abcam, Cambridge, UK), Lamin B1 (ab16048; Abcam), and GAPDH (number 5174; Cell Signaling Technology) diluted 1:1000 in Can Get Signal Immunoreaction Enhancer Solution 1 (TOYOBO, Osaka, Japan). After washing in TBST, secondary anti-rabbit IgG antibodies (number 7074; Cell Signaling Technology) were added. Immunoreactivity was detected with ECL Prime (Amersham Biosciences) and photographed on an Ez Capture MG (ATTO, Tokyo, Japan). Band densities were calculated using CS Analyzer 3.0 (ATTO).

Alcian blue staining
ATDC5 cells were seeded in 12-well plates at a density of 0.5 ϫ 10 5 cells/well. After 2 days, AS1842856 was added to each well to a final concentration of 0 or 0.1 M. After 1 day, the cells were differentiated in medium containing 1% ITS for 4, 7, 14, 21, or 28 days. Limb bud micromass cultures were incubated with or without AS1842856 (0.1 M) for 5 days. The differentiated cells were washed three times with PBS, and then fixed with 4% paraformaldehyde (PFA) (Wako Pure Chemical Industries) at room temperature for 10 min. The fixed cells were stained overnight with 0.3% Alcian blue 8GX (Sigma) in 0.1 N HCl (Wako Pure Chemical Industries). The stained cells were washed three times with 0.1 N HCl. For quantitative analyses, the amount of extracted dye was measured. Alcian blue-stained FOXO1 regulates chondrogenic differentiation cultures were extracted with 200 l of 6 M guanidine HCl (Wako Pure Chemical Industries) for 6 h at room temperature. Optical densities were measured at 620 nm using an iMark microplate reader (Bio-Rad Laboratories).
Pellets of hMSCs were differentiated in chondrogenic medium with or without AS1842856 (0.1 M) for 21 days, fixed overnight with 4% PFA, and embedded in paraffin. Sections were deparaffinized, rehydrated, and stained with 0.3% Alcian blue 8GX for 30 min. The percentages of positively stained areas were measured using the BZ-II Analyzer software program (Keyence, Osaka, Japan).

Immunocytochemistry
ATDC5 cells were seeded in 4-well chamber slides at a density of 1 ϫ 10 4 cells/well. After 1 day, the cells were incubated with TGF␤1 (10 ng/ml) and SIS3 (3 M) for 24 h. ATDC5 cells transfected with siSMAD3 were also incubated with TGF␤1 (10 ng/ml) for 24 h. Limb bud micromass cultures were incubated in 8-well chamber slides for 5 days. The cells were fixed with 4% PFA (Wako Pure Chemical Industries) for 10 min at room temperature, and then blocked with 5% goat serum (Wako Pure Chemical Industries) and 0.3% Triton X-100 (Sigma) for 1 h at room temperature. Subsequently, the cells were incubated with antibodies against FOXO1 (number 2880, number 14952; Cell Signaling Technology) and SMAD3 (number 9523; Cell Signaling Technology) diluted 1:100 in Can Get Signal Immunostain Solution B (TOYOBO) for 1 h at room temperature. The cells were washed three times with PBS and incubated with Alexa Fluor 488 (A11008; Invitrogen) and 568 (A11011, A11004; Invitrogen) for 1 h at room temperature. SlowFade diamond antifade mountant with DAPI (Invitrogen) was used for nuclear staining. Immunostaining was visualized by fluorescence microscopy (BZ-X700; Keyence).

Immunohistochemistry
Pellets of hMSCs were differentiated in chondrogenic medium or normal medium for 21 days, fixed overnight with 4% PFA, and embedded in paraffin. Sections were deparaffinized and rehydrated, and antigen retrieval was performed by incubation overnight with EDTA (1 mM) at pH 8.0. Endogenous peroxidase activity was blocked by incubation with 3% hydrogen peroxidase in methanol for 30 min. After blocking with normal horse serum (Vectastain Universal Elite ABC kit; Vector Laboratories, Burlingame, CA) for 20 min, sections were incubated with antibodies against FOXO1 (number 2880; Cell Signaling Technology) for 2 h at room temperature. Sections were incubated with biotinylated secondary antibodies for 30 min at room temperature, followed by incubation with streptavidin-peroxidase complex (Vectastain Universal Elite ABC kit) for 30 min at room temperature. Antibody complexes were visualized using the diaminobenzidine substrate system (Wako Pure Chemical Industries), and counterstained with hematoxylin.

Cell cycle analysis
ATDC5 cells were seeded in 6-well plates at a density of 1 ϫ 10 5 cells/well. After 2 days, AS1842856 was added to each well to a final concentration of 0 or 0.1 M. After 1 day, the cells were differentiated in medium containing 1% ITS. After 4 days, the cells were centrifuged at 4°C, washed with PBS, and fixed with 70% ethanol (Wako Pure Chemical Industries) at 4°C for 30 min. After the cells were centrifuged and washed with PBS, propidium iodide solution (PI/RNase; Cosmo Bio Company, Tokyo, Japan) was added, and the cells were incubated at room temperature for 15 min. The cell cycle was analyzed on a BD Accuri TM C6 flow cytometer (BD Biosciences). For each sample, 60,000 events were recorded.
The cell cycle of ATDC5 cells transfected with plasmids overexpressing FOXO1-ADA was also analyzed 2 days after transfection. For each sample, 60,000 events were recorded.

Cell proliferation assay
ATDC5 cells were seeded in 96-well plates at a density of 0.6 ϫ 10 4 cells/well. After 2 days, AS1842856 was added to each well to a final concentration of 0 or 0.1 M. After 1 day, the cells were differentiated in medium containing 1% ITS. After 1, 2, or 3 days, cell proliferation was evaluated using CellTiter-Glo (Promega).

Chromatin immunoprecipitation (ChIP) assay
ChIP assay was performed using the SimpleChIP Plus Sonication Chromatin IP kit (Cell Signaling Technology). ATDC5 cells were seeded in 15-cm dishes at a density of 1 ϫ 10 6 cells/dish. After 2 days, TGF␤1 (5 ng/ml) was added. After incubation with TGF␤1 for 24 h, ATDC5 cells were fixed with 37% formaldehyde and sonicated. For immunoprecipitation experiments, antibodies against FOXO1 (ab39670; Abcam) and normal rabbit IgG (number 2729; Cell Signaling Technology) were used. PCR products were amplified for 34 cycles. The sequences of primers used in PCR are shown in Table S2.

Statistical analysis
All experiments were repeated at least three times. Data are presented as mean Ϯ S.D. Wilcoxon's rank-sum test was used for two-group comparisons. Multiple comparisons in Western blot were evaluated by one-way repeated measures ANOVA with the Tukey-Kramer post hoc test. For multiple comparisons in qRT-PCR and flow cytometry, the Tukey-Kramer test was performed. All data analyses were performed using the JMP 13 statistical software (SAS Institute, Cary, NC). p Ͻ 0.05 was considered statistically significant.