Focal Adhesion Kinase/Src Suppresses Early Chondrogenesis

Adhesive signaling plays a key role in cellular differentiation, including in chondrogenesis. Herein, we probe the contribution to early chondrogenesis of two key modulators of adhesion, namely focal adhesion kinase (FAK)/Src and CCN2 (connective tissue growth factor, CTGF). We use the micromass model of chondrogenesis to show that FAK/Src signaling, which mediates cell/matrix attachment, suppresses early chondrogenesis, including the induction of Ccn2, Agc, and Sox6. The FAK/Src inhibitor PP2 elevates Ccn2, Agc, and Sox6 expression in wild-type mesenchymal cells in micromass culture, but not in cells lacking CCN2. Our results suggest a reduction in FAK/Src signaling is a critical feature permitting chondrogenic differentiation and that CCN2 operates downstream of this loss to promote chondrogenesis.

Adhesive signaling plays a key role in cellular differentiation, including in chondrogenesis. Herein, we probe the contribution to early chondrogenesis of two key modulators of adhesion, namely focal adhesion kinase (FAK)/Src and CCN2 (connective tissue growth factor, CTGF). We use the micromass model of chondrogenesis to show that FAK/Src signaling, which mediates cell/matrix attachment, suppresses early chondrogenesis, including the induction of Ccn2, Agc, and Sox6. The FAK/Src inhibitor PP2 elevates Ccn2, Agc, and Sox6 expression in wildtype mesenchymal cells in micromass culture, but not in cells lacking CCN2. Our results suggest a reduction in FAK/Src signaling is a critical feature permitting chondrogenic differentiation and that CCN2 operates downstream of this loss to promote chondrogenesis.
Cartilage is a connective tissue that possesses a wide array of functions, including establishing the skeletal framework during embryogenesis and cushioning joints in adulthood. Chondrocytes, the cell type found in cartilage, generate and maintain the cartilaginous extracellular matrix (ECM), 8 which is crucial for chondrogenic development and homeostasis (1). Formation of the adult skeleton is achieved by intramembranous and endochondral ossification (2,3). The endochondral skeleton consti-tutes most skeletal elements of the body and is formed in two main steps. First, cartilage is formed by chondrogenesis. Chondrogenesis is initiated when mesenchymal cells aggregate to form condensations that ultimately determine the shape and location of future bones (2,3,4). During this process, a high cell density is achieved that promotes cell-cell interactions resulting in the propagation of signal transduction events necessary for the initiation of chondrogenesis (5). Cells within these condensations begin to express markers typical of early chondrogenic cells. These markers include such proteins as the "Sox trio" of transcription factors (Sox9 and the related factors L-Sox5 and Sox6), type II collagen, and aggrecan (6,7). Moreover, cells change in morphology from a fibroblast-like appearance to a spheroidal shape (8). Endochondral ossification, a process involving the creation of bone tissue utilizing the cartilage as template, follows (4).
In developing limbs in vivo, cells originating from the lateral plate mesoderm condense and form aggregations (5). In mouse development, formation of these aggregations occurs 10.5-12.5 days post-coitum (dpc) (9). The level and efficiency of overall chondrogenic differentiation, and hence subsequent bone development, is directly correlated to the density of the initial condensation (10). The Sox trio appear to cooperate with each other to co-regulate the expression of chondrogenic markers such as aggrecan and type II collagen. However, much remains unknown about the additional requirements for chondrogenesis. Recently, it was shown that the formation of condensations and cartilage nodules requires adhesive signaling and remodeling of the actin cytoskeleton, via Rho and Rac (11)(12)(13)(14). At later stages, adhesive signaling through integrins is involved (15). However, the contribution of cell-ECM interactions in early chondrogenesis is poorly understood. For example, the role of focal adhesion kinase (FAK)/Src, which mediates cell-ECM interactions, in chondrogenesis has not been thoroughly investigated. Moreover, the role of CCN2, a member of the CCN (CYR61, CTGF, NOV) family of pro-adhesive matricellular signaling modulators (16), in early chondrogenesis is unclear.
Culture systems have been developed that promote chondrogenesis. The three-dimensional micromass system is an excellent culture model to promote chondrogenic differentiation of a wide range of mesenchymal cells by providing sufficiently high cell density to promote the cell-cell interactions required for chondrogenesis (10). In this report we use the micromass * This work was supported in part by the Canadian Foundation for Innovation, the Arthritis Society, the Canadian Institute of Health Research (CIHR), and the Scleroderma Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  cell culture system to probe the role of FAK/Src and CCN2 in early chondrogenesis in vitro.
Micromass Cell Culture-Mesenchymal cells were cultured in Dulbecco's modified Eagle's medium, 10% fetal bovine serum (Invitrogen), 1% penicillin/streptomycin/amphotericin B (Invitrogen). Cells were plated at a density of 1 ϫ 10 5 per 10-l droplet into each well of a 24-well tissue culture plate (Nunc) and left to adhere for an hour. Once adhered, micromass cultures were given 1 ml of medium. Micromass cultures were grown for a period of 6 days. In experiments involving inhibitor treatment, starting on Day 0, the day of plating, medium was supplemented with the FAK/Src inhibitor PP2 (10 M in dimethyl sulfoxide (Me 2 SO) (Calbiochem)), whereas control cultures were supplemented with the vehicle Me 2 SO only. Medium was replenished daily. The high density culture system allows for the recapitulation of the in vivo high density environment during chondrogenesis that allows for a high degree of cell-cell contact (11, 12, 20 -22). This culture system works to effectively promote chondrogenesis in a variety of mesenchymal cells, including embryonic and adult fibroblasts (22)(23)(24)(25).
RNA Isolation and Real-time RT (Reverse Transcription) PCR-RNA extraction was performed with the Qiagen RNeasy kit according to the manufacturer's protocol (Qiagen Inc.). RNA was collected from micromass cultures on days 1, 3, 6 of differentiation. RNA concentration was determined using a spectrophotometer (Beckman Coulter). 25 ng of RNA/reaction was used for real-time RT-PCR according to established protocols (20 -22). A total 15-l reaction volume was used containing the TaqMan one-step master mix kit (Applied Biosystems) and gene-specific target primers (Assays-on-demand; Applied Biosystems) and probes (FAM (6-carboxyfluorescein) dye layer) and endogenous reference primers and probes (VIC dye layer). The FAM dye layer yields quantification of the target genes, whereas VIC yields simultaneous quantification of glyceraldehyde-3-phosphate-dehydrogenase (Gapdh) as an internal control. Relative gene expression was determined by measuring Col2a1, Agc, Dcn, Hapln, L-Sox5, Sox6, Sox-9, Ccn1, Ccn2, and Ccn5, using 40 cycles on the ABI Prism 7900 HT sequence detector (PerkinElmer Life Sciences). All samples were amplified in three parallel reactions per trial, and three independent trials were performed.
RNA Quality Assessment, Probe Preparation, and Gene Chip Hybridization and Analysis-Microarrays and analyses were performed essentially as previously described (19,26). Gene Chips were processed at the London Regional Genomics Centre (Robarts Research Institute, London, ON). RNA was harvested (TRIzol; Invitrogen) and quantified. Quality was assessed using the Agilent 2100 Bioanalyzer (Agilent, Palo Alto, CA), the RNA 6000 Nano kit (Caliper Life Sciences, Mountain View, CA), and the Degradometer.
Biotinylated complementary RNA (cRNA) was prepared from 10 g of total RNA (Affymetrix, Santa Clara, CA). Double-stranded cDNA was synthesized using SuperScript II (Invitrogen) and oligo(dT) 24 primers. Biotin-labeled cRNA was prepared by in vitro transcribing cDNA (Enzo Brioche, New York, NY). Fifteen g of labeled cRNA was hybridized to Mouse Genome 430 2.0 Gene Chips for 16 h at 45°C (Affymetrix). Gene Chips were stained with streptavidinphycoerythrin, followed by an antibody solution and a second streptavidin-phycoerythrin solution (GeneChip Fluidics Station 450; Affymetrix). Gene Chips were scanned with the GeneChip Scanner 3000 (Affymetrix). Signal intensities for genes were generated using GCOS1.2 (Affymetrix) using default values for the statistical expression algorithm parameters and a target signal of 150 for all probe sets and a normalization value of 1. Normalization was performed in GeneSpring 7.2 (Agilent Technologies Inc.). The RMA preprocessor was used to import data from the .cel files. Data were transformed, (measurements Ͻ0.01 set to 0.01), normalized per chip to the 50th percentile and per gene to wildtype control samples. Experiments were performed twice, and -fold changes were identified using the GeneSpring filter. Data presented in Table 1 are an average of these independent studies. The -fold change between treated and untreated samples had to be at least 2-fold, in both sets of experiments, to identify a transcript as being altered.
Immunofluorescence-Ccn2 ϩ/ϩ and Ccn2 Ϫ/Ϫ cells were plated in monolayer at a density of 12,000 cells/well in a 24-well dish (Falcon) on glass coverslips. Cells were har-vested and fixed in 4% paraformaldehyde for 30 min at 4°C. Cells were washed in phosphate-buffered saline (PBS), incubated for 5 min with 0.1% Triton-X in PBS, and rinsed again in PBS. Cells were then incubated in blocking solution containing goat serum (Sigma) in PBS, 1:20, for 30 min at room temperature. Primary antibodies directed to Sox6 were diluted in blocking solution at a concentration of 1:200 and incubated with coverslips for 1 h at room temperature. Coverslips were rinsed in PBS and then incubated with an Alexa-Fluor 488-conjugated secondary antibody, diluted 1:300 in PBS, for 1 h at room temperature in the dark. Following another wash in PBS, coverslips were mounted in Vecta-Shield anti-fade mounting medium containing 4Ј,6-diamidino-2-phenylindole. Images were taken with a Zeiss Axiophot microscope using Northern Eclipse software (Empix) and exported into Adobe Photoshop.
Peanut Agglutinin (PNA) Staining-Micromass cultures were performed as above and were fixed on day 6 of culture in 4% paraformaldehyde at 4°C for 30 min. Cells were rinsed with PBS and then incubated for 2 h in 50 g/ml PNA diluted in PBS. Cultures were washed again with PBS, and PNA was detected colorimetrically by diaminobenzidine (Dako Cytomation). Images were captured with a Nikon SMZ1500 microscope at ϫ5.6 magnification using a Photometrics coolSNAP-cf color digital camera (Photon Technology International).
Alcian Blue Staining-Alcian Blue staining was carried out on day 6 of micromass culture as follows. Cultures were washed twice with cold PBS, fixed in 100% ethanol for 20 min at Ϫ20°C, and incubated with 0.1% HCl-Alcian blue for 2 h (20). Excess stain was washed off with double distilled water and pictures were taken as described above. Stain was quantified by solubilizing the stain in 6.0 M guanidine hydrochloride for 8 h at room temperature. Absorbance was measured using a spectrophotometer at 620 nm.
Statistical Analysis-Data collected from real-time RT-PCR are an average of three trials of samples, from completely independent experiments, run in triplicate. Means were quantified relative to Gapdh, and then data were normalized to day 1 of control per trial or in the case where there is only 1 day being examined, data were normalized to the control sample. Statistical significance was determined by Student's paired t test or two-way analysis of variance, with a level of significance defined as p Ͻ 0.05 using the Bonferroni post-test and GraphPad Prism version 4.00 for Windows.

Loss of FAK Promotes Features of Chondrogenesis, Including Nodule Formation and CCN2 Expression-
The earliest stage of chondrogenesis, namely condensation, involves the migration of mesenchymal cells via cell/ECM interactions and is associated with increased tyrosine phosphorylation of FAK (12). However, the role of FAK in subsequent phases of chondrogenesis, including nodule formation and the induction of chondrogenic genes, is wholly unknown. To specifically assess the effect of loss of FAK on early chondrogenesis, we used the micromass culture system. Micromass cell culture is an established method of inducing chondrogenesis in vitro in a variety of mesenchymal cell types, including embryonic and adult fibroblasts (11,(22)(23)(24)(25). To test the contribution of FAK to this process, we subjected Fak ϩ/ϩ and Fak Ϫ/Ϫ fibroblasts (Fig. 1a) to high density micromass-induced chondrogenesis. Fak Ϫ/Ϫ animals die at embryonic day 8.5 (17,27), precluding use of this type at later embryonic stages. Fibroblasts at 8.5 dpc represent "naïve" cells that have not begun to differentiate (9). Fak ϩ/ϩ and Ϫ/Ϫ fibroblasts were plated in micromass cultures for 6 days. On day 6 of culture, PNA-stained images (Fig. 1b) were taken of the cultures. To our surprise, loss of FAK expression, even in naïve E8.5 fibroblasts, resulted in increased PNA staining, a hallmark of chondrogenic differentiation, in micromass culture as compared with wild-type cells (Fig. 1b). In the Fak Ϫ/Ϫ cells, L-Sox5 and Col2a1 mRNA expression was significantly higher than in the wild-type controls (Fig. 1c). Conversely, Sox9 expression was unaltered (Fig. 1c). Intriguingly, Agc and Hapln mRNAs were not detected either in Fak ϩ/ϩ and Ϫ/Ϫ fibroblasts (data not shown). Similarly, Sox6 was not induced in Fak Ϫ/Ϫ E8.5 cells (Fig. 1c). These results surprisingly suggest that the loss of FAK in these naïve cells was sufficient to induce some, but not all, features of chondrogenesis.
CCN2 is a key modulator of adhesive signaling and promotes chondrogenic gene expression in vivo and in vitro (16,18). To probe a possible connection between loss of FAK and CCN2 expression, real-time RT-PCR and Western blot analysis of Fak ϩ/ϩ and Ϫ/Ϫ fibroblasts in micromass culture for 6 days were used. This analysis revealed that CCN2 protein and mRNA were significantly elevated in the absence of FAK, whereas CCN2 protein and mRNA were not induced in Fak ϩ/ϩ cells (Fig. 1, a and c). Collectively, these data suggested the intriguing notion that, in cells competent to undergo bone fide chondrogenic differentiation, FAK signaling may suppress chondrogenesis by preventing the induction of CCN2.
Loss of Ccn2 Disrupts Expression of Early Chondrogenic Genes in Monolayer Culture-To test the hypothesis that CCN2 acted downstream of the loss of FAK to promote chondrogenesis, it was necessary to employ cells capable of undergoing full chondrogenic differentiation. To investigate this question, we therefore used E13.5 mouse mesenchymal cells, as chondrogenesis occurs in vivo at this stage (9). Microarray analysis of mRNAs isolated from Ccn2 ϩ/ϩ and Ϫ/Ϫ mesenchymal cells growing in monolayer culture was performed. This analysis revealed that many early chondrogenic genes, such as Agc, Dcn, Hapln (the gene encoding link protein), and Sox6, showed a Ͼ2-fold reduction in expression in the absence of CCN2 (Table 1). Moreover, loss of Ccn2 resulted in reduced expression of the CCN family members Ccn1 and Ccn5 (Table 1), both of which have been implicated in bone formation (16). Differences in gene expression between wild-type and knock-out cells revealed by microarray analysis were verified using real-time RT-PCR analysis. As expected, real-time PCR analysis revealed that Ccn2 mRNA expression was observed in wild-type, but not in Ccn2 Ϫ/Ϫ , cells (Fig. 2). Confirming our microarray data, realtime PCR analysis also showed that Ccn2 Ϫ/Ϫ mesenchymal cells possessed a significant decrease in the mRNAs encoded by Ccn1 and Ccn5, Sox6, Agc, Dcn, and the link protein gene Hapln APRIL 4, 2008 • VOLUME 283 • NUMBER 14 JOURNAL OF BIOLOGICAL CHEMISTRY 9241 (Fig. 2). We used Western blot analysis to confirm reduced expression of Sox6 and aggrecan protein in the absence of CCN2 (Fig. 3A). Moreover, down-regulation of Sox6 in the Ccn2 Ϫ/Ϫ cells was confirmed by immunofluorescence analyses of Ccn2 ϩ/ϩ and Ϫ/Ϫ mesenchymal cells. Sox6 was localized around the nuclei of Ccn2 ϩ/ϩ cells but was markedly reduced in Ccn2 Ϫ/Ϫ cells (Fig. 3B). In contrast to these results, here was a significant increase in L-Sox5 gene expression and no change in Sox9 (Fig. 2). Furthermore, no change in mRNA transcript levels for Col2a1 or Col10a1 (Fig. 2) was observed in the CCN2deficient mesenchymal cells. Western blot analysis confirmed that loss of CCN2 did not result in reduced L-Sox5 and type II collagen protein expression (Fig. 3A). These data demonstrate that CCN2 is required for the expression of some, but not all, chondrogenic genes in monolayer cell culture.

Role of CCN2 in Early Chondrogenesis
Loss of Ccn2 Results in Impaired Chondrogenesis in Micromass Culture-Based on our mRNA and protein expression data indicating a role for CCN2 in early chondrogenesis, we assessed the effect of loss of CCN2 in the micromass culture system (11). Ccn2 ϩ/ϩ and Ϫ/Ϫ mesenchymal cells were plated in high density micromass cultures and cultured for up to 6 days. PNA staining of the resultant cultures revealed that loss of CCN2 expression impaired the overall ability of mesenchymal cells to undergo condensation (Fig. 4a). Similarly, Alcian Blue staining revealed that loss of CCN2 expression resulted in reduced glycosaminoglycan production (Fig. 4b). RNAs were harvested on days 1, 3, and 6 of micromass culture, and transcription of chondrogenic genes was analyzed by real-time RT-PCR. In wild-type mesenchymal cells, CCN2 was expressed at constant levels throughout the time course, and, as expected, CCN2 was undetectable in Ccn2 Ϫ/Ϫ cells (Fig. 5). Confirming our experiments using monolayer culture, loss of CCN2 expression resulted in decreased expression of Agc and Hapln (encoding link protein) mRNAs throughout the 6 days of culture (Fig. 5). In the absence of CCn2, Col2a1 mRNA expression was significantly increased throughout the 6 days of culture (Fig. 5). These data suggest that the absence of Protein isolated from mouse mesenchymal cells on day 6 was examined for FAK and CCN2 by Western blot analysis. FAK protein expression was absent in Fak Ϫ/Ϫ mouse embryo fibroblasts. CCN2 expression was upregulated in the Fak Ϫ/Ϫ mouse embryo fibroblasts. ␤-actin was used as a loading control. b, PNA staining was performed on day 6. Fak Ϫ/Ϫ mouse embryo fibroblasts show markedly more condensation formations than Fak ϩ/ϩ mouse embryo fibroblasts in the PNA-stained micromass cultures, as assessed by diameter of individual micromass cultures. Six different micromass cultures for each condition were examined. Representative cultures are shown. c, RNAs were harvested on day 6, and transcripts were analyzed by real-time RT-PCR. Ccn2, Col2a1, and L-Sox5 mRNA expression was significantly increased in Fak Ϫ/Ϫ mouse fibroblasts. Expression of Sox6 and Sox9 mRNAs did not change. Data shown represent means ϩ S.E. from three independent experiments and are normalized to control GAPDH levels (each performed in triplicate). *, p Ͻ 0.05 using Student's paired t test. Note that Agc and Hapln mRNAs were not detected in any samples tested. CCN2 alters the balance of collagen and proteoglycan gene expression. Confirming our prior data using a monolayer culture system, loss of CCN2 resulted in reduced Sox6 expression in micromass culture (Fig. 5). Conversely, Sox9 and L-Sox5 were unaltered in Ccn2 Ϫ/Ϫ micromass cultures on days 1 and 3 and displayed a significant increase in expression in knock-out cells on day 6 ( Fig. 5). When regulation of the CCN family was examined, we found that Ccn5 mRNA expression increased significantly during differentiation of wild-type cells but was significantly lower in knock-out cells throughout the culture period (Fig. 5). It is interesting to note that Ccn1 mRNA expression was significantly lower in Ccn2 Ϫ/Ϫ cells on days 1 and 3 of culture but by 6 days of culture similar expression levels of Ccn1 mRNA were observed both in the presence and absence of CCN2 (Fig. 5). Thus, expression of CCN1, a protein possessing functions similar to CCN2 in vitro (16), occurs via a CCN2-dependent fashion in early chondrogenesis but via a CCN2-independent mechanism in later chondrogenesis. Collectively, these data suggest that CCN2 mediates the expression of a subset of chondrogenic genes.
CCN2 Acts Independently of BMP2-Members of the transforming growth factor ␤ family, including the BMP proteins that induce chondrogenesis (31), activate CCN2 expression (32,33). To assess whether CCN2 acted downstream of BMPs to promote chondrogenesis, we subjected Ccn2 ϩ/ϩ and Ccn2 Ϫ/Ϫ mesenchymal cells to micromass-induced chondrogenesis in the presence and absence of added BMP2 (150 ng/ml). We found that BMP2 potentiated chondrogenesis in both Ccn2 ϩ/ϩ and Ccn2 Ϫ/Ϫ mesenchymal cells (Fig. 6), suggesting that BMPs promote chondrogenesis via a CCN2-independent pathway. In fact, the responsiveness of cells to BMP2 was enhanced in the absence of CCN2 (Fig. 6). These results are consistent with previous data using 10T1/2 cells that showed CCN2 blocked BMP action by binding to BMPs, thereby preventing BMPs from interaction with their receptors (34). These results emphasize that, although CCN2 is required for certain features of early chondrogenesis, CCN2-independent pathways play roles during this process.

Inhibition of FAK/Src Signaling Increases Proteoglycan Gene Expression in a Ccn2-dependent Fashion-
Having shown that FAK suppressed and CCN2 enhanced chondrogenesis, we sought to test the interrelationship between these two processes. To probe the possible connection between FAK, CCN2, and early chondrogenesis, Ccn2 ϩ/ϩ and Ccn2 Ϫ/Ϫ mesenchymal cells were plated in micromass cultures and grown in the presence of Me 2 SO (as a control) or 10 M PP2 (a FAK/Src inhibitor) for 6 days. RNAs were harvested and subjected to real-time RT-PCR analysis. Consistent with our observations using E8.5 mesenchymal cells, FAK/Src inhibition induced Ccn2 mRNA (Fig. 7). To assess whether CCN2 operated downstream of the loss of FAK to induce the expression of chondrogenic genes, we assessed whether PP2 altered the mRNA levels of genes previously shown to be CCN2-dependent. We found that Agc, L-Sox6, and HapIn mRNAs were significantly increased in wild-type cells treated with PP2 compared with control cultures treated with Me 2 SO (Fig. 7). In contrast, these transcripts were not induced in Ccn2 Ϫ/Ϫ cells treated with PP2 (Fig. 7). These results confirm that CCN2 is required for the induction of Agc, L-Sox6, and HapIn and show that CCN2 acts downstream of loss of FAK signaling in this process (Fig. 7). Finally, we showed that full-length recombinant CCN2 induced Agc and L-Sox6 expression in Ccn2 Ϫ/Ϫ cells, indicating that these genes are direct targets of CCN2 (Fig. 8). (Expression of HapIn mRNA was undetectable in Ccn2 Ϫ/Ϫ cells with or without CCN2 treatment, indicating that additional proteins work with CCN2 to induce expression of this gene (not shown).) Collectively, these data indicate that a reduction in FAK signaling enhances chondrogenesis in a fashion that is at least partly mediated by CCN2.

DISCUSSION
In this report, we investigated the contribution of FAK and CCN2 to early chondrogenesis. Perhaps our most intriguing findings were that FAK suppressed chondrogenic gene expres-sion and that loss of FAK increased CCN2 expression. It is well established that FAK is necessary for focal adhesion turnover and to form connections between the cell and ECM (17,27). During chondrogenesis, however, cell-cell contacts are favored (35). (Indeed, the micromass culture system is designed to promote chondrogenesis by achieving heightened cell-cell contacts in an environment non-permissive to cell motility.) In our current study, we found that FAK signaling suppressed the expression of chondrogenic markers, including CCN2, aggrecan, and link protein. As an example, the FAK/Src inhibitor PP2 induced expression of CCN2, aggrecan, and link protein in E13.5 mesenchymal cells. Intriguingly, PP2 was unable to induce aggrecan and link protein mRNAs in Ccn2 Ϫ/Ϫ mesenchymal cells. Overall, our data suggest the novel idea that FAK inhibits chondrogenesis. Moreover, our data suggest that inhibition of CCN2 expression by FAK mediates some of the antichondrogenic effects of FAK. A possible explanation for the effect of FAK on chondrogenesis may be that the absence of FAK promotes a shift from an environment favoring cell-ECM interactions to one promoting cell-cell interactions, the latter being a key feature required for chondrogenesis. Another potential explanation for the increase in chondrogenic capability of Fak Ϫ/Ϫ cells or PP2-treated cells is that FAK is known to be required for migration (16,36). Cells undergoing chondrogenesis in vivo and in vitro are not motile, thus loss of FAK may promote chondrogenesis by impairing cell migration. Intriguingly, it is believed that whereas cell-cell interactions drive cell condensation and the consequent commitment of mesenchy-FIGURE 3. In the absence of Ccn2, levels of early chondrogenic matrixassociated proteins Sox6 and aggrecan are decreased. A, protein isolated from Ccn2 ϩ/ϩ (WT) and Ϫ/Ϫ (KO) mouse mesenchymal cells was cultured in monolayer for 48 h. Conditioned medium and protein extracts were examined for aggrecan and Sox6, respectively, by Western blot analysis. Both Sox6 and aggrecan protein expression were decreased in Ccn2 Ϫ/Ϫ mouse mesenchymal cells. ␤-actin was used as a loading control. Densitometry analysis of aggrecan and Sox6, relative to ␤-actin, shows a significant difference between samples. No difference was seen in the expression of type II collagen and Sox5. Data shown represent means ϩ S.E. from three trials. *, p Ͻ0.05 using Student's paired t test. B, for immunofluorescence analysis, cells were fixed in paraformaldehyde and stained with fluorescein isothiocyanate-labeled (FITC) antibody for Sox6 and with 4Ј,6-diamidino-2-phenylindole (DAPI) for nuclei. Sox6 was localized around the nuclei in the Ccn2 ϩ/ϩ mouse mesenchymal cells, whereas expression was markedly decreased in the Ccn2 Ϫ/Ϫ mouse mesenchymal cells. Six different fields were examined. A representative field is shown. mal cells to a chondrogenic lineage, cell-ECM interactions have been shown mediate subsequent stages of cartilage differentiation; thus it is possible that FAK/Src may promote later stages of chondrogenesis (14,37).
CCN molecules bind to the cell surface via integrin and heparin sulfate proteoglycans. As a consequence, CCN family members are known to be potent pro-adhesive molecules directly promoting adhesive signaling and modulating adhesive signaling in response to other stimuli (16,17,28,29,30,38,39). As matricellular proteins, CCN family members appear not to be themselves growth or differentiation signals but rather seem to modify signaling of bone fide growth factors and cytokines, the identity of which varies depending on the particular cellular context. Indeed, CCN2 appears to create an environment permissive for other stimuli to induce potent signaling responses (30). Consistent with these ideas, we found that although induction of CCN2 was observed in E8.5 Fak Ϫ/Ϫ mesenchymal cells subjected to micromass culture, induction was not observed of genes shown to be CCN2-dependent using differentiation-competent E13.5 cells. Moreover, recombinant CCN2, which partially rescued the reduced expression of mRNAs encoding Sox6 and aggrecan mRNAs in Ccn2 Ϫ/Ϫ cells, was not able to restore mRNA encoding link protein. Collectively, these observations are consistent with the notion that the mere presence of CCN family members is insufficient to recapitulate the entire range of CCNmodulated activity (29,30,39). It is possible therefore that, in E13.5 cells, CCN2 may potentiate the action of other chondrogenic signals. The identity of these putative signals is not known (indeed the mechanisms underlying chondrogenesis are poorly understood) and is beyond the scope of the current study. However, it is interesting to note that we found that CCN2 was not required for cells to respond to the potent chondrogenic protein BMP2. Indeed, we found that Ccn2 Ϫ/Ϫ cells were more responsive to BMP2 than Ccn2 ϩ/ϩ cells. These results are fully consistent with previous observations using 10T1/2 cells that CCN2 blocks BMP action (34).
Our results indicating a role for CCN2 in early chondrogenesis are fully consistent with previous data examining the role of CCN2 in bone formation. CCN2 is expressed in mesenchymal cells during conditions of tissue remodeling and repair, including development and wound healing, and in fibroproliferative disorders, including cancer and fibrosis (30,40). More specifically, CCN2 is expressed in hypertrophic chondrocytes in the growth plate of cartilage (30,40). A direct role for CCN2 in cartilage and bone formation in vivo has been recently demonstrated. Mice homozygous for a deletion in the Ccn2 gene have expanded hypertrophic zones in their long bones and possess an underdeveloped rib cage (18). CCN2-deficient mice die soon after birth, presumably due to an inability to breathe properly (18). Moreover, Ccn2 Ϫ/Ϫ mice showed reduced aggrecan and link protein expression (18). These latter observations are completely consistent with our results obtained using cultured mesenchymal cells that CCN2 was required for both aggrecan and link protein expression. Type II collagen expression was not affected by loss of CCN2 in our studies; however, it is interesting to note that CCN2 induces expression of type II and X FIGURE 5. Loss of Ccn2 affects proteoglycan mRNA expression. Ccn2 ϩ/ϩ (WT) and Ϫ/Ϫ (KO) mouse mesenchymal cells were subjected to micromass cultures for 6 days. For real-time RT-PCR, RNA was harvested on days 1, 3, and 6. Agc, Sox6, and Hapln mRNA expression was significantly reduced in Ccn2 Ϫ/Ϫ mouse mesenchymal cells on all 6 days examined. Conversely, Col2a1, L-Sox5, and Sox9 expression was significantly increased in Ccn2 Ϫ/Ϫ mouse mesenchymal cells. Data shown are relative to GAPDH and represent means ϩ S.E. from three independent experiments (each performed in triplicate). *, p Ͻ 0.05 using a two-way analysis of variance. Expression in WT cells on day 1 was taken to represent 1. FIGURE 6. BMP-2 treatment of Ccn2 ؊/؊ micromass cultures increases expression of chondrogenic extracellular matrix-associated genes. Ccn2 ϩ/ϩ (WT) and Ϫ/Ϫ (KO) mouse mesenchymal cells were subjected to micromass culture for 6 days. RNA was harvested on day 3, and transcripts were analyzed by real-time RT-PCR. a-c, induction of Agc, Sox6, and Col2a1 mRNA is greater in Ccn2 Ϫ/Ϫ mouse mesenchymal cells compared with Ccn2 ϩ/ϩ mouse mesenchymal cells. Data shown are relative to GAPDH and represent means ϩ S.E. from three independent experiments (each performed in triplicate). *, p Ͻ 0.05 using Student's paired t test. -Fold increase in response to BMP-2 is shown. collagen in chondrosarcoma cells in vitro (28). It should be pointed out that the current in vitro studies of CCN2 function pertaining to bone formation are primarily based on evidence obtained from the chondrosarcoma-derived chondrocytic cell line HCS-2/8 or from mature osteoblasts (28,(41)(42)(43). These models are perhaps more relevant in suggesting roles for CCN2 in cancer or later stages of development, respectively, rather than chondrogenesis or osteogenesis per se.
It is perhaps surprising that our current results reveal a role for CCN2 in early chondrogenesis yet the phenotype of the Ccn2 Ϫ/Ϫ mice appear to have defects in later stages of bone development, namely endochondral ossification (18). However, as discussed above, failure of cell to undergo proper condensation is believed to have profound downstream effects on the overall quality of both initial chondrogenesis and later bone formation. Moreover, our current study showed that Ccn1 mRNA expression was down-regulated in monolayer culture and on days 1 and 3 of micromass culture. However, Ccn1 expression was similar in Ccn ϩ/ϩ and Ccn2 Ϫ/Ϫ cells on day 6 of micromass culture, indicating that CCN2-independent mechanisms controlled Ccn1 expression during early, but not later, stages of chondrogenic differentiation. Previous work has shown similarity in function and expression of CCN1 and CCN2 (29,44,45). CCN1 is also known to be expressed in chondrocytes and involved in the synthesis of collagen and other ECM components in vitro (46). Therefore, a potential reason why the skeletal malformations seen in the Ccn2 Ϫ/Ϫ animals are not more severe could be due to a functional compensation by CCN1 (and potentially CCN5) during later stages of chondrogenesis. Testing this notion awaits the generation of mice lacking both CCN1 and CCN2.
Several previous studies have examined the relationship between the Sox trio and their cooperative action during the initiation of chondrogenesis (47)(48)(49). It is widely accepted that L-Sox5 and Sox6 are necessary for maximal activity of Sox9 (47)(48)(49). Although it is believed that the transcription of chondrocyte-specific genes is coordinately regulated by the Sox trio of transcription factors (48,50), our current results reveal that CCN2 was required for Sox6 and proteoglycan, but not Sox5, Sox9, and Col2a1, transcription. That is, our results reveal that chondrogenic genes need not necessarily be expressed in parallel. Supporting this notion, L-Sox5 and Sox6, but not Sox9, are down-regulated upon depolymerization of the actin cytoskeleton (13). Furthermore, L-Sox5, Sox9, and Col2a1, but not Sox6 and Agc, expression is induced upon Cdc42 overexpression in ATDC5 cells (14). Our current data therefore support a novel concept that Sox genes may show differential regulation during chondrogenesis. It should be also pointed out that, whereas animals deficient in three or more Sox5 or Sox6 alleles have substantial cartilage defects, mice in which either Sox5 or Sox6 have been deleted have a minimal phenotype (49). Moreover, overexpression of Sox6 in Ccn2 Ϫ/Ϫ cells was insufficient to rescue the gene expression defects in this cell type (not shown). Collectively, these observations suggest that alterations in Sox6 expression are not likely to contribute significantly to the chondrogenic defects observed in the CCN2-deficient cells.
In summary, we have shown for the first time that a loss of FAK/Src activity promotes early chondrogenesis, occurring, at FIGURE 7. Inhibition of FAK/Src signaling increases mRNA expression of Ccn2 and chondrogenic matrixassociated genes. Ccn2 ϩ/ϩ (WT) and Ϫ/Ϫ (KO) mouse mesenchymal cells were subjected to micromass cultures. Each day for 6 days Me 2 SO or 10 M FAK/Src inhibitor PP2 was added. RNA was harvested on day 6. mRNA levels of Ccn2, Agc, Hapln, and Sox6 were significantly increased compared with control cultures in Ccn2 ϩ/ϩ mouse mesenchymal cells treated with PP2, as determined by real-time RT-PCR. Such induction did not occur in Ccn2 Ϫ/Ϫ mouse mesenchymal cells. Data shown are relative to GAPDH and represent means ϩ S.E. from three independent experiments and are normalized to day 6 WT control sample and day 6 KO control sample (each performed in triplicate). *, p Ͻ 0.05 using a two-way analysis of variance. Note that we were not able to detect expression of CCN2 or Hapln RNA in KO cells. FIGURE 8. Recombinant CCN2 induces Agc and Sox6 mRNA expression in Ccn2 ؊/؊ mouse mesenchymal cells. Ccn2 Ϫ/Ϫ (KO) mouse mesenchymal cells were cultured in monolayer and serum-starved for 24 h prior to treatment with or without full-length 38-kDa CCN2 (100 ng/ml) for 6 h. mRNA levels of Agc and Sox6 were significantly increased by treatment with CCN2. Hapln mRNA was undetectable in both untreated and treated Ccn2 Ϫ/Ϫ mouse mesenchymal cells (not shown). Data shown are relative to GAPDH and represent means ϩ S.E. from three independent experiments (each performed in triplicate). *, p Ͻ 0.05 using two-way analysis of variance. Expression in untreated KO sample was taken to represent 1. least in part, by promoting CCN2 expression. In turn, CCN2 is required for the production of proteoglycans. Specifically, our results showed that Sox6 and aggrecan expression are CCN2dependent whereas type II collagen expression is CCN2-independent. Overall, our data provide new and valuable insights into the complex interplay of signaling cascades during early chondrogenesis.