RhoA/ROCK Signaling Regulates Sox9 Expression and Actin Organization during Chondrogenesis

Endochondral ossification is initiated by the differentiation of mesenchymal precursor cells to chondrocytes (chondrogenesis). This process is characterized by a strong interdependence of cell shape, cytoskeletal organization and the onset of chondrogenic gene expression, but the molecular mechanisms mediating these interactions are not known. Here we investigated the role of the RhoA/ROCK pathway, a well-characterized regulator of cytoskeletal organization, in chondrogenesis. We show that pharmacological inhibition of ROCK signaling by Y27632 resulted in increased glycosaminoglycan synthesis and elevated expression of the chondrogenic transcription factor Sox9, whereas overexpression of RhoA in the chondrogenic cell line ATDC5 had the opposite effects. Suppression of Sox9 expression by ROCK signaling was achieved through repression of Sox9 promoter activity. These molecular changes were accompanied by reorganization of the actin cytoskeleton where RhoA/ROCK signaling suppressed cortical actin organization, a hallmark of differentiated chondrocytes. This led us to analyze the regulation of Sox9 expression by drugs affecting cytoskeletal dynamics. Both inhibition of actin polymerization by cytochalasin D and stabilization of existing actin filaments by jasplakinolide resulted in increased Sox9 mRNA levels, whereas inhibition of microtubule polymerization by colchicine completely blocked Sox9 expression. In conclusion, our data suggest that RhoA/ROCK signaling suppresses chondrogenesis through the control of Sox9 expression and actin organization.


Introduction
Chondrocytes fulfill two major roles in mammals. During development, most of our bones form through endochondral ossification in which bones are first laid down as cartilage precursors (1,2). In this process, cartilage serves as a template for subsequent bone formation and controls bone growth through the coordinated proliferation and differentiation of chondrocytes in the growth plate.
In the adult, chondrocytes are the sole cell type of articular cartilage and play crucial roles in joint function (3,4). Disruption of chondrocyte function in either context results in severe consequences for affected individuals (4)(5)(6).
Deregulated proliferation or differentiation of growth plate chondrocytes (e.g. through gene mutations, hormonal disorders or medication) commonly results in skeletal deformities and growth retardation (7,8), whereas loss of articular chondrocyte function is a major contributing factor in the pathogenesis of osteoarthritis (9)(10)(11).
Despite some clear differences, growth plate and articular chondrocytes share several common features. Extracellular matrix proteins such as collagen II and aggrecan are among the cartilage markers produced by both cell types (12,13). The transcription factor Sox9 has been shown to be required for chondrocyte formation (chondrogenesis) and directly regulate transcription of the collagen II gene, in conjunction with the related Sox5 and Sox6 genes (14)(15)(16).
Another common feature of all chondrocytes is the interdependence of cell shape and differentiation status (17)(18)(19)(20). Chondrogenesis is characterized by by guest on March 23, 2020 http://www.jbc.org/ Downloaded from drastic changes in cell shape from a fibroblastoid to a round or polygonal morphology (21). This transition is accompanied by changes in gene expression and reverted when chondrocytes dedifferentiate, for example in osteoarthritis and in monolayer culture in vitro (4,22). The molecular mechanisms responsible for this interplay are largely unknown, but the actin cytoskeleton appears to play important roles in this context (23)(24)(25). Chondrocytes display mostly cortical organization of their actin filaments in vivo and in vitro, whereas precursor cells or dedifferentiated chondrocytes are characterized by a more fibrillar organization (26)(27)(28)(29). Moreover, inhibition of actin polymerization by cytochalasin B has been shown to induce redifferentiation of dedifferentiated chicken chondrocytes (30)(31)(32)(33)(34).
These data suggest that the pathways controlling the organization of the chondrocyte actin network could also play major roles in regulating chondrocyte differentiation and function. Rho GTPases are the best characterized upstream regulators of the actin cytoskeleton (35,36). Through a multitude of downstream effectors, they control not only cytoskeletal organization, but also many other cellular functions such as transcription, cell cycle progression and vesicle trafficking (37)(38)(39). The kinases ROCKI and ROCKII are among the most important effectors of the prototype GTPase RhoA (40,41

Materials
Timed pregnant CD1 mice were purchased from Charles River Laboratories. All cell culture media components were from Invitrogen or Sigma unless stated otherwise. All inhibitors were purchased from Calbiochem or Sigma. All other reagents were of analytical grade from commercial suppliers.
The pRlCMV plasmid was from Promega, and the SRF reporter plasmid was purchased from Stratagene. Antibodies against ROCK I (catalogue number Sc-5560) and ROCK II (Sc-5561) were from SantaCruz, the b-actin antibody (A-544) was from Sigma. A plasmid containing the proximal 2 kb of the mouse Sox9 promoter controlling expression of firefly luciferase was provided by Dr. M.
Underhill (University of British Columbia).

Micromass culture and ATDC5 cell culture
Mesenchymal limb buds cells were obtained from mice at 11.5 dpc, and cultured in micromass cultures as described (45,46). Briefly, cells were suspended in 60% F12, 40% DMEM, 10% FBS (Gibco), 0.25% PenStrep and 0.25% L-glutamine, at a density of 2.5x10 7 cells/ml and plated in 10 µl droplets to simulate the high density of chondrogenic condensations. One hour after plating, media (as above, supplemented with 1 mM beta-glycerol phosphate and 50 µg/ml of ascorbic acid) was added to the cultures. As indicated, medium is supplemented with DMSO, 10 µM Y27632 (dissolved in DMSO), 3 µM cytochalasin D (dissolved in DMSO), 50 nM jasplakinolide (dissolved in DMSO), or 1 µM colchicine (dissolved in water). Media and inhibitors were changed every 24 hours until harvesting.
For confocal microscopy, micromass cultures were plated on glass coverslips and cultured as above. After one day in culture, cells were fixed in 4% paraformaldehyde at room temperature for ten minutes, followed by 2 x 10 minutes washes in phosphate buffered saline (PBS). Membranes were then permeabilized with 0.1% Triton-X/PBS solution for 5 minutes and washed twice for 5 minutes with PBS. The cells were then incubated in the dark for 50 minutes at room temperature with 2.5 U/PBS rhodamine phalloidin and mounted with Vectashield® (Vector Laboratories Inc., Burlingane, CA). Images were taken using a Zeiss LSM510 Meta confocal microscope with 40-fold magnification and analyzed using LSM-10 software.
RhoA-and pcDNA3 vector-transfected ATDC5 cells were cultured and induced to differentiate as described (42).

Reverse Transcriptase and Real-Time PCR
RNA was extracted using the QIAGEN RNeasy kit according to the manufactures instructions. 500 ng of collected RNA is reverse transcribed using both random hexamers, pd (N) 6

Western Blot Analysis
Protein was isolated from treated micromass cultures on days 1 to 4.
Cells were collected in cold PBS and centrifuged at 10,000 rpm for 5 minutes at

Alcian Blue Stain
Treated micromass cultures were fixed after 2, 3 or 4 days in culture in 100% ethanol for 20 minutes at -20ºC and then incubated with 0.1% HCl-Alcian blue for two hours (45). Excess stain was washed off with double distilled water and pictures were taken. Stain was quantified by solubilizing the stain in 6 M guanidine hydrochloride for 8 hours at room temperature. Absorbance was measured using a spectrophotometer at 620 nm. Nodule number was assessed through manual counting by an independent observer unaware of experimental conditions.

Isolation of primary chondrocytes
Primary chondrocytes were isolated from tibias of day 15.5 timed mouse embryos. Tibias were isolated and digested for 15 minutes at 37°C with trypsin, followed by a two hour digestion in 3 mg/ml Collagenase P (ROCHE) dissolved in  (47).

Statistical Analysis
Data collected from Real time PCR are represented as the average of three independent experiments (e.g. three independent isolations of primary cells) run in triplicate. Means were quantified relative to 18S rRNA and/or GAPDH and data were normalized to day 1 of control treated RNA per trial.
Alcian blue quantification and nodule counting was an average of three to four independent cell isolations of two replicates per treatment, per time point. Data of luciferase activity represent an average of four independent cell isolations performed in quadruplicate each. Statistical significance was determined with one way ANOVA or two way ANOVA followed by a post-hoc Bonferroni test using GraphPad Prism software.

ROCK I/II are expressed during chondrogenesis
While we had earlier demonstrated expression of ROCKI/II in chondrocytes (42), no data on temporal profiles of their expression during early chondrogenesis are available. We first examined the expression of both ROCK I and II in our micromass cultures. Transcripts were detected with the expected amplicon sizes of 480 and 275 base pairs for ROCKI and II, respectively, throughout the micromass culture period from days 1 to 4 (Fig. 1a). Expression of both kinases was also demonstrated at the protein level by Western blot analyses (Fig. 1b). No obvious changes in ROCKI and II mRNA or protein levels were observed during chondrogenesis.

ROCK suppresses glycosaminoglycan production
We next asked whether inhibition of ROCK signaling would interfere with chondrogenesis by analyzing the effects of the ROCK inhibitor Y27632 (10 mM) on Alcian blue staining. Alcian blue stains for glycosaminoglycans and is therefore an established maker of chondrogenesis. Alcian blue staining increased over time in micromass culture, indicating advanced chondrogenic differentiation. ROCK inhibition did not affect glycosaminoglycan production at days 2 or 3 or micromass culture, but resulted in a visible increase in Alcian blue staining by day 4 of micromass culture (Fig. 2a). Stimulation of glycosaminoglycan production by Y27632 at day 4 was confirmed quantitatively by dye extraction and measurement of absorbance (Fig. 2b) (Fig. 2c). The size of these nodules also does not appear to change between the control and treated cultures. These data suggest that ROCK inhibition does not affect cell condensation, but results in increased chondrogenic differentiation and chondrocyte-specific extracellular matrix synthesis of mesenchymal precursor cells within nodules.

ROCK inhibition induces cortical actin morphology
We next asked whether the effects on glycosaminoglycan synthesis are accompanied by chondrocyte-specific changes in cellular morphology and actin organization. Primary chondrocytes in monolayer culture rapidly lost their rounded morphology and developed a fibroblastoid cell shape with extensive stress fibers (Fig. 3a). Treatment with Y27632 caused reorganization of the actin cytoskeleton to a cortical pattern with parallel rounding of cells, suggesting that ROCK inhibition supports the establishment of a chondrocyte-specific cell shape and actin organization. Similar mechanisms were observed in three-dimensional micromass cultures by confocal microscopy; cultures treated with Y27632 displayed increased cell rounding and a reduced number of actin fibers (Fig. 3b

RhoA overexpression suppresses glycosaminoglycan synthesis and induces stress fiber formation in chondrogenic cells
RhoA is an upstream activator of ROCKI/II and requires ROCK activity for its effects in later stages of chondrogenic differentiation (42). We therefore asked whether RhoA regulates chondrogenesis in a similar fashion as ROCK kinases.
Overexpression of RhoA in the chondrogenic cell line ATDC5 resulted in reduced Alcian blue staining (Fig. 4a). These effects were reversed by treatment with Y27632. RhoA overexpression in ATDC5 cells also caused cell elongation and formation of stress fibers, when compared to vector-transfected control cells (Fig.   4b). ROCK inhibition with Y27632 rescued this effect. These data demonstrate that RhoA suppresses chondrogenic differentiation through a ROCKI/IIdependent mechanism.

RhoA/ROCK signaling inhibits Sox9 expression during chondrogenesis
We next asked whether changes in actin organization and glycosaminoglycan expression are accompanied by alteration of chondrogenic gene expression by investigating Sox9 mRNA expression. Similar to Alcian blue staining, Sox9 mRNA levels increased markedly after two days of micromass culture. While Sox9 mRNA levels were similar in Y27632-treated and control cultures until day 2, real-time PCR analysis showed that ROCK inhibition greatly reduces the increase in Sox9 expression on days 3 and 4 (Fig. 5a). RhoA overexpression in the ATDC5 cell line caused a 50% reduction in Sox9 mRNA levels. This effect is rescued by the addition of Y27632 (Fig. 5b). We examined the effects of Y27632 on the activity of a 2 kb fragment of the mouse Sox9 promoter. ROCK inhibition caused a twofold induction of this promoter fragment in micromass cultures, suggesting that ROCK signaling controls Sox9 mRNA levels through transcriptional mechanisms (Fig. 5c).

Manipulation of actin polymerization regulates Sox9 mRNA levels
Our data had shown regulation of chondrocyte actin organization and  (Fig. 6B), and also completely inhibited Alican blue staining (data not shown). These data show that microtubule polymerization is absolutely required for chondrogenesis to occur, in agreement with earlier studies that have shown reduced glycosaminoglycan and proteoglycan production in colchicine-treated chondrocytes (49,50).

Effects of cytoskeletal modifications on Serum Response Factor activity
Previous publications have described a role of the transcription factor serum response factor (SRF) 1 in transcriptional response to cytoskeletal modifications and Rho signaling (51)(52)(53)(54). We therefore asked whether the diverse drugs used in this study would signal through SRF. We transiently transfected micromass cultures with an SRF responsive promoter to examine regulation of SRF activity. A significant increase of activity of the SRF is observed upon inhibition of actin or microtubule polymerization (Fig. 7A, B). While Y27632 and Jasplakinolide appeared to activate SRF to some degree, these effects were not statistically significant. The only transcription factor known to date to be absolutely required for chondrogenesis is Sox9 (14,56,57). This led us to study whether Sox9 RhoA/ROCK signaling exerts its cellular effects through cytoskeletondependent and -independent mechanisms. We therefore asked whether effects of ROCK inhibition could be mimicked by drugs affecting actin remodeling.
Cytochalasin D is an inhibitor of actin polymerization that has been shown promote the chondrogenic phenotype (23,24,34). However, the molecular mechanisms of how disruption of actin polymerization promotes chondrogenesis have not been identified. We show here that cytochalasin D treatment enhances Sox9 expression during chondrogenesis, thus providing a molecular explanation for its chondrogenic activities. Jasplakinolide stabilizes actin filaments and promotes actin polymerization (58) and was therefore expected to have opposing biological activities to cytochalasin D. However, jasplakinolide treatment induced Sox9 mRNA levels even more than cytochalasin D.
While these similar effects of two drugs with apparently opposing biological activities are puzzling, they are not without precedence.      All data shown are the mean relative gene expression ± standard error from three independent trials run in triplicate each normalized to 18s rRNA and/or GAPDH RNA levels (*: p< 0.05).
c) Micromass cultures were transiently transfected with a 2 kb minimal Sox9 promoter construct and pRICMV (for normalization) and treated daily with vehicle or 10 mM Y27632. After 3 days, cells were harvested and relative luciferase activity was determined by normalizing firefly luciferase activity to renilla luciferase activity. The Sox9 promoter conferred a two-fold increase in luciferase activity when the RhoA/ROCK pathway was inhibited. Data shown are an average of relative luciferase activity of four independent experiments run in quadruplicate ± standard error (*: p< 0.05).