Nkx3.2-mediated Repression of Runx2 Promotes Chondrogenic Differentiation*

Runx2, a transcription factor known to be essential for osteoblast maturation and skeletogenesis, is also expressed in pre-cartilaginous mesenchymal condensations in the developing embryo. It is therefore necessary to understand the control and consequential regulatory activity of the Runx2 gene within the context of chondrogenic differentiation of a mesenchymal progenitor cell. We identify the homeodomain protein Nkx3.2 as a potent sequence-specific repressor of the Runx2 promoter that acts through a regulatory element 0.1 kb upstream from the site of transcriptional initiation. The biological significance of this repression is established by utilizing bone morphogenic protein 2 (BMP-2)-induced chondrogenic differentiation of pluripotent C3H10T1/2 cells as a model for the initial events of mesenchymal chondrogenesis. We demonstrate that induction of the chondrogenic phenotype and endogenous Nkx3.2 expression is accompanied by a repression of Runx2 gene activity. Bypassing Runx2 repression by adenoviral-mediated introduction of Runx2 into C3H10T1/2 cells can prevent the induction of chondrogenesis, but cannot reverse the chondrogenic phenotype once it has been initiated, as evidenced by Sox9 and type II collagen expression and extracellular matrix deposition. Our results demonstrate that Runx2 is a direct transcriptional target of Nkx3.2, and that repression of Runx2 at the onset of chondrogenesis is a prerequisite for the activation of a chondrocyte-specific program of gene expression. We postulate that Runx2 is a critical link in BMP-2-mediated initiation of mesenchymal chondrogenesis that results in activation of Sox9 at least in part through the Nkx3.2-dependent repression of Runx2.

Runx2, a transcription factor known to be essential for osteoblast maturation and skeletogenesis, is also expressed in pre-cartilaginous mesenchymal condensations in the developing embryo. It is therefore necessary to understand the control and consequential regulatory activity of the Runx2 gene within the context of chondrogenic differentiation of a mesenchymal progenitor cell. We identify the homeodomain protein Nkx3.2 as a potent sequence-specific repressor of the Runx2 promoter that acts through a regulatory element 0.1 kb upstream from the site of transcriptional initiation. The biological significance of this repression is established by utilizing bone morphogenic protein 2 (BMP-2)-induced chondrogenic differentiation of pluripotent C3H10T1/2 cells as a model for the initial events of mesenchymal chondrogenesis. We demonstrate that induction of the chondrogenic phenotype and endogenous Nkx3.2 expression is accompanied by a repression of Runx2 gene activity. Bypassing Runx2 repression by adenoviral-mediated introduction of Runx2 into C3H10T1/2 cells can prevent the induction of chondrogenesis, but cannot reverse the chondrogenic phenotype once it has been initiated, as evidenced by Sox9 and type II collagen expression and extracellular matrix deposition. Our results demonstrate that Runx2 is a direct transcriptional target of Nkx3.2, and that repression of Runx2 at the onset of chondrogenesis is a prerequisite for the activation of a chondrocyte-specific program of gene expression. We postulate that Runx2 is a critical link in BMP-2-mediated initiation of mesenchymal chondrogenesis that results in activation of Sox9 at least in part through the Nkx3.2-dependent repression of Runx2.
During mammalian embryogenesis, the axial skeleton is initially formed as a cartilaginous template prior to conversion into mineralized bone through the coordinated actions of osteoclasts and osteoblasts. The mesenchymal progenitor cells responsible for cartilage formation aggregate around the notochord and are induced to proliferate and subsequently to differentiate in response to the secreted signaling molecules Sonic hedgehog (Shh) 1 and bone morphogenic protein 2 (BMP-2) (1)(2)(3)(4). The sequential and cooperative action of these molecules results in the induction of the pro-chondrogenic NKrelated homeodomain protein Nkx3.2 in mesenchymal progenitor cells. Nkx3.2 mediates transcriptional repression of target genes through its interactions with BMP-responsive SMAD proteins and histone deacetylase HDAC1 (5). Activation of Nkx3.2 is required for and closely followed by activation of the master chondrogenic transcriptional regulator Sox9. The activation of Sox9 through Nkx3.2 ultimately leads to activation of chondrocyte phenotypic genes, including type II collagen and aggrecan, and formation of a cartilaginous extracellular matrix (1,6). The importance of Nkx3.2 in this process is evident from Nkx3.2 null mice, in which mesenchymal progenitor cells of the sclerotome fail to differentiate resulting in malformation or absence of axial skeletal elements and perinatal lethality (7)(8)(9).
Recent findings in our laboratory and others have demonstrated that pre-chondrogenic mesenchymal progenitor cells exhibit Runx2 gene activity as early as 9.5 days postcoitum, 3 days prior to overt chondrogenesis in the murine embryo (10 -13). The Runx family of DNA-binding transcription factors governs cell fate determination in a variety of tissues. Runx factors are essential for hematopoiesis, skeletal development, and development of the digestive and nervous systems (14 -18). Runx2 is indispensable for the differentiation of the boneforming osteoblast, and Runx target genes in this cell lineage have been well characterized (10, 11, 19 -22). The observation that the Runx2 gene is active in progenitor cells that are destined to undergo chondrogenesis rather than osteogenesis necessitates determining whether Runx2 plays a role in BMP-2-induced chondrogenic differentiation of mesenchymal progenitor cells in addition to the well characterized function of Runx2 in promoting the osteogenic differentiation of osteoblast precursors later during skeletogenesis.
In this study, we identify the Runx2 gene as a target of Nkx3.2-mediated transcriptional repression and demonstrate that Runx2 is a critical link in the BMP-2 induced pathway of chondrogenesis. We show that Nkx3.2 represses Runx2 activity through an interaction with a regulatory element in the Runx2 promoter. Furthermore, our results suggest that at the onset of BMP-2-induced chondrogenesis of C3H10T1/2 mesenchymal progenitor cells, an increase in Nkx3.2 activity suppresses Runx2 gene expression. We demonstrate that bypassing the observed suppression of Runx2 at the onset of chondrogenesis inhibits chondrocytic differentiation. Our findings establish that Runx2 is a critical modulator of the commitment of mesenchymal progenitor cells to the chondrogenic lineage.

EXPERIMENTAL PROCEDURES
Cell Culture and Transient Transfection-C3H10T1/2 and NIH3T3 cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (Atlanta Biologicals, GA). MC3T3 cells were maintained in ␣-minimal essential media supplemented with 10% fetal bovine serum. ROS17/2.8 cells were maintained in F-12 media supplemented with 5% fetal bovine serum. Transient transfections were performed in 6-well plates at 70% confluence using 5 l of FuGENE 6 transfection reagent (Roche Diagnostics) and 4 g of total DNA per well in accordance with the manufacturers protocol. For Nkx3.2 expression, 100 ng of an Nkx3.2 expression vector in a PCS2 plasmid backbone (a kind gift from Dr. Andrew Lassar, Harvard Medical School) was transfected into each well unless otherwise noted. As a control, 100 ng of PCS2 expression vector (empty vector) was transfected into each well. For Sox9 expression, 100 ng of a Sox9 expression construct was introduced into each well of a 6-well plate. Shh (R&D systems, Minneapolis, MN) and BMP-2 (a kind gift from Dr. John Wozney, Wyeth Ayerst, Cambridge, MA) were added exogenously at concentrations of 200 and 100 ng/ml, respectively. To observe repression of endogenous Runx2 protein in the presence of Nkx3.2, C3H10T1/2 cells were co-transfected with either Nkx3.2 and cytomegalovirus-driven enhanced green fluorescence protein (EGFP) or empty vector (PCS2) and EGFP. After 24 h, cells were trypsinized and FACS sorted to collect cells positive for EGFP fluorescence. EGFP positive cells were replated and harvested 12 h later for Western analysis. To monitor transfection efficiency, transfections included 0.5 g of a cytomegalovirus-driven LacZ expression vector per well. For Runx2 promoter-reporter assays, transfections included 2 g of either a Runx2 0.6-kb promoter-luciferase construct, or an empty PGL3 luciferase construct. All results were normalized to activity of the PGL3 empty vector (Promega, Madison, WI).
Luciferase Reporter Assays-Cells transfected with Runx2 promoter/luciferase reporter constructs were harvested 36 -48 h after transfection and each well was lysed at room temperature for 20 min in the presence of 0.5 ml of reporter lysis buffer (Promega, Madison, WI). Firefly luciferase activity was quantitated in a luminometer using a 12-s read time immediately after addition of 20 l of cell lysate to 100 l of substrate (Promega Luciferase Assay System). With the exception of Fig. 1A, all results were normalized to the luciferase activity resulting from transfection of the promoterless PGL3 luciferase construct (Promega).
Chondrogenic Induction and Adenoviral Infection of C3H10T1/2 Cells-Induction of chondrogenesis was carried out by plating C3H10T1/2 cells (between passages 19 and 25) in high-density micromass cultures (10 5 cells in a 10-l drop of media) (23-25), followed by a 3-h incubation period in which cells were allowed to adhere. Following adhesion, micromass cultures were fed with F-12 media containing 5% fetal bovine serum and 100 ng/ml recombinant hBMP-2 (kindly provided by Dr. John Wozney, Wyeth-Ayerst, MA). For adenoviral transduction experiments, C3H10T1/2 cells were infected with viruses containing either the Runx2 cDNA or the ␤-galactosidase cDNA as a control at a multiplicity of infection of 100. To introduce adenovirally expressed proteins prior to chondrogenic induction, C3H10T1/2 cells were infected while proliferating in a monolayer at ϳ80% confluence. At 12 h after infection, cells were trypsinized and plated in high-density micromass cultures for chondrogenic differentiation as described above. Cultures were harvested 24 h after induction of chondrogenesis for analysis of gene expression. To deliver adenovirally expressed proteins after induction of chondrogenesis, C3H10T1/2 cells were plated in highdensity cultures, allowed to adhere, then infected with viruses containing either the Runx2 cDNA or the ␤-galactosidase cDNA. Immediately after infection, cultures were fed with F-12 media containing 5% fetal bovine serum and 100 ng/ml recombinant hBMP-2. Cultures were harvested 24 h after the induction of chondrogenesis for analysis of gene expression.
RNA Isolation and Analysis-RNA was isolated from cultures of C3H10T1/2 cells using TRIzol reagent (Invitrogen, Carlsbad CA) according to the manufacturer's protocol. After purification, 5 g of total RNA was DNase treated using a DNA-free RNA column purification kit (Zymol Research, Orange, CA). RNA (1 g) was then reverse transcribed using oligo(dT) primers and the SuperScript 1 st Strand Synthesis kit (Invitrogen) according to the manufacturers protocol. Gene expression was assessed by quantitative real-time PCR (Sox9, Nkx3.2, type II collagen, type I collagen, alkaline phosphatase, and Runx2). Quantitative PCR was performed using either SYBR Green 2 ϫ master mixture (Eurogentec, Belgium) or Fam-conjugated Taqman probes and Taqman 2ϫ master mixture in the case of Nkx3.2 and Runx2 (Applied Biosciences, Foster City, CA) and a two-step cycling protocol (anneal and elongate at 60°C, denature at 94°C). Specificity of primers was verified by dissociation of amplicons when using SYBR Green as a detector. Primers used for PCR reactions are listed in Table I.
Electrophoretic Mobility Shift Analysis (EMSA)-Expression constructs containing the Nkx3.2 coding region or the empty PCS2 vector were subjected to in vitro transcription/translation using the TNTcoupled rabbit reticulocyte lysate system (Promega) and Sp6 RNA polymerase (New England Biolabs, Beverly, MA) according to the manufacturer's protocol. In vitro synthesis of recombinant Nkx3.2 was verified by Western blotting against the HA tag present in the PCS2 vector (data not shown). Wild type and mutant oligonucleotides containing the Nkx3.2 consensus site derived from the Runx2 P1 promoter was end labeled by incubating 10 M of the sense strand of each oligo with 50 Ci of [␥-32 P]ATP in the presence of 10 units of polynucleotide kinase (New England Biolabs) for 30 min at 37°C. Unlabeled antisense oligo (30 M) was then added to the reaction followed by boiling the mixture for 5 min. The reaction was then gradually allowed to cool to room temperature to allow the annealing of sense and antisense strands to occur. Double stranded oligos were then purified over a Sephadex G-25 column to remove unincorporated nucleotides. DNA binding assays were performed by incubating 10 fmol of double-stranded oligo with 2 g of nuclear proteins from C3H10T1/2 cells along with or without 0.4 g of rabbit polyclonal ␣-HA antibody (Santa Cruz, Santa Cruz, CA) or a nonspecific antibody, ␣-cMYC (Santa Cruz), at room temperature for 20 min. Complexes were visualized after separation on a 16% 40:1 acrylamide:bis-acrylamide gel in the presence of 0.5ϫ TBE buffer followed by autoradiography.
Total protein (20 g) was subjected to electrophoreses in a denaturing 10% polyacrylamide gel containing 10% SDS. Proteins were then transferred onto Immobilon-P membranes (Millipore) using a semi-dry transfer apparatus. Membranes were blocked in PBS, 0.01% Tween 20 containing 2% nonfat powdered milk (Bio-Rad). Proteins were detected by incubating with antibodies at a concentration of 50 ng/ml in blocking solution. Antibodies used in this study are as follows: Nkx3.2, ␣-HA epitope mouse monoclonal antibody (Santa Cruz, Santa Cruz, CA); Runx2 mouse monoclonal antibody was a generous gift from Drs. Yoshi Ito and Kosei Ito, National University, Singapore; and ␣-actin goat polyclonal antibody. Primary antibodies were detected with goat ␣-mouse secondary antibody conjugated to horseradish peroxidase. Secondary antibodies were detected using Western Lightning Chemiluminescence Reagent (PerkinElmer Life Sciences, Boston, MA).
Chromatin Immunoprecipitation Assays-To cross-link proteins to DNA, C3H10T1/2 cells were incubated for 10 min at room temperature To isolate the nuclei, cells were homogenized in a Dounce homogenizer followed by centrifugation at 1,100 ϫ g at 4°C. The pelleted nuclei were resuspended in 300 l (300 l/100-mm plate) of sonication buffer (50 M HEPES/NaOH (pH 7.9), 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% Na-deoxycholate, 0.1% SDS, 25 M MG132, 1ϫ Complete protease inhibitor). Samples were sonicated to shear DNA into 0.2-0.6-kb fragments. Cellular debris was removed by centrifugation at 14,000 ϫ g for 15 min at 4°C and the resulting chromatin-containing solutions were distributed into multiple 1-ml aliquots that were used as the starting material of all subsequent steps. Chromatin aliquots were precleared with 100 l of a 25% (v/v) suspension of 2 g of single-stranded DNA-coated protein A/G and 1 mg/ml bovine serum albumin. Samples were used directly for immunoprecipitation reaction with 2 g of ␣-HA epitope or ␣-Runx2 (M-70, Santa Cruz Biotechnology) antibody and normal rabbit/mouse IgG as a control. Chromatin immunoprecipitation reactions were allowed to proceed for 2-4 h at 4°C on a rotating wheel. Immune complexes were mixed with 100 l of 25% (v/v) pre-coated protein A/G-agarose suspension followed by incubation for 1 h at 4°C on a rotating wheel. Beads were collected by brief centrifugation and the immunocomplexes were eluted twice by adding 150 l of freshly prepared elution buffer (100 mM NaHCO 3 , 1% SDS). After reversal of cross-links at 68°C overnight, the eluate was treated with 100 g/ml proteinase K followed by phenol-chloroform extraction and ethanol precipitation using 5 g of glycogen as carrier. An aliquot (2-3 l) of each sample was assayed using quantitative PCR for the presence of specific DNA fragments using primers in the proximal Runx2 promoter. This region contains both the Nkx3.2 binding motif as well as Runx2 autoregulatory motifs. The primers are: forward, 5Ј-CTCCAGTAATAGTGCTTGCAAAAAAT-3Ј and reverse, 5Ј-GC-GAATGAAGCATTCACACAA-3Ј. Quantitative real-time PCR was carried out using 2 ϫ SYBR Green mixture (Eurogentec, Belgium) and a 2-stage cycling protocol (60°C annealing and extension, 94°C denaturation, 40 cycles). Amplicon specificity was verified by analysis of melting temperature. All data were collected during the linear phase of amplification.

RESULTS
The Runx2 P1 Promoter Is Highly Active in C3H10T1/2 Cells and Is Suppressed by Nkx3.2-To gain insight into the regulation of the Runx2 gene in a pluripotent mesenchymal progen-itor cell, we transfected a vector containing 600 base pairs of the Runx2 P1 promoter (26,27) fused to the luciferase reporter gene into C3H10T1/2 cells. This promoter (P1-Luc) was transcriptionally active (ϳ10 times that of a promoterless luciferase construct, PGL3, Fig. 1A) in this cell line, comparable with luciferase activity observed in osteogenic ROS17/2.8 cells that express high levels of Runx2 protein. This observation is consistent with previous findings that the endogenous Runx2 gene is active in mesenchymal progenitor cells during development in vivo (10 -13).
We next examined the effect of various signaling molecules and transcription factors known to be critical for mesenchymal chondrogenesis in vivo on the activity of the Runx2 P1 promoter, including Shh, BMP-2, Nkx3.2, and Sox9. These molecules act sequentially during murine embryogenesis and are critical at different stages for the expansion and differentiation of mesenchymal progenitor cells into mature chondrocytes (Fig.  1B). Our results demonstrate that Shh, BMP-2, and Sox9 exhibit little to no effect on the activity of the Runx2 promoter. In contrast, expression of the transcription factor Nkx3.2 in C3H10T1/2 cells resulted in a dramatic, dose-dependent suppression of Runx2 promoter activity (Fig. 1C). These findings indicate that the robust activity of Runx2 regulatory regions in mesenchymal progenitor cells is negatively regulated by Nkx3.2, a molecule that acts to promote chondrogenic differentiation of these cells.
Nkx3.2 Represses the Runx2 Promoter via Interaction with an Nkx3.2 Consensus Binding Sequence-We examined the possibility that the Nkx3.2-mediated repression of the Runx2 promoter occurred through direct interaction of Nkx3.2 with its consensus-binding site, HRAGTG (H ϭ A, C, or T; R ϭ A or G) (28). Indeed, we identified two consensus binding sites for Nkx3.2 within the Runx2 P1 promoter (at Ϫ580 bp and Ϫ98 bp from the Runx2 transcription initiation site) ( Fig. 2A). To define the contribution of these sites to Nkx3.2-mediated repression of the Runx2 promoter, we co-transfected the Nkx3.2 expression construct into C3H10T1/2 cells along with various deletion constructs of the Runx2 promoter. Fig. 2A demonstrates that the distal Nkx3.2 binding site is dispensable for repression. However, the Ϫ108-bp Runx2 promoter construct still exhibits repression in response to Nkx3.2, and removal of this site (Ϫ92 bp deletion) results in a loss of repression, as well as an abrogation of basal promoter activity. When the data is examined as -fold repression mediated by Nkx3.2 upon the deletion constructs, it becomes evident that an additional promoter region between Ϫ458 and Ϫ351 bp may be playing a contributing role (Fig. 2B). Because there is no acceptable Nkx3.2 consensus binding sequence in the Ϫ458/Ϫ351 region, this effect may occur independently of Nkx3.2.
To validate the interaction between Nkx3.2 and its consensus binding site at Ϫ98 bp in the Runx2 promoter, EMSAs were performed using a 24-bp oligo containing the Nkx3.2 site (Fig.  3A). This site lies downstream of a purine-rich region of the promoter in a highly conserved area that contains binding sequences for a number of transcriptional regulators including HLH, ATF, vitamin D, as well as Runx2 itself. Nkx3.2 binds to this site in a sequence-specific manner as mutation of this site completely abrogates binding activity (Fig. 3B compare WT to MT). The specificity of this complex was confirmed using supershift analysis of the Nkx3.2 interaction with the wild type oligo. Incubation of the radiolabeled wild type oligo results in the formation of a specific protein-DNA complex only in the presence of Nkx3.2 expression vector (Fig. 3C). The addition of mouse polyclonal ␣-HA antibody completely supershifts this complex. To further confirm that Nkx3.2 is occupying this site in the native promoter, CHiP assays were performed 24 h after transfection of Nkx3.2 into C3H10T1/2 cells using antibodies against endogenous Runx2 or HA-tagged Nkx3.2. We were able to immunoprecipitate the Runx2 promoter using an ␣-HA antibody, but not using an ␣-Runx2 or nonspecific (IgG) antibody, demonstrating that Nkx3.2 is indeed occupying the Runx2 promoter (Fig. 3D). These results demonstrate that the Nkx3.2 transcription factor interacts with a functional consensus binding site in the Runx2 P1 promoter ϳ100 bp upstream from the site of initiation of transcription.
The Proximal Nkx3.2 Binding Site in the Runx2 Promoter Is Required for Maximal Repressive Effects of Nkx3.2-To determine whether the proximal Nkx3.2 binding site is responsible for mediating the observed repression of the Runx2 promoter in the presence of Nkx3.2, we introduced the mutation used for EMSA into both the Ϫ600-bp and Ϫ108-bp Runx2 promoter constructs. Mutation of this site in the Ϫ600-bp promoter resulted in significant abrogation of the Nkx3.2-mediated repres-  (26,27,55). The ATF and HLH sites are putative. The putative Nkx3.2 binding sequence was used to create wild type (WT) or mutated (MT) oligos for EMSA. EMSA demonstrates that Nkx3.2 can bind the wild type, but not mutant oligos (B). The specificity of the observed protein-DNA interaction was verified by supershift assay using an anti-HA antibody against the tagged Nkx3.2 protein (C). Chromatin immunoprecipitation assay in C3H10T1/2 cells transfected with HA-tagged Nkx3.2 confirms the physical interaction between Nkx3.2 and the endogenous Runx2 promoter (D). sion of the Ϫ600-bp Runx2 promoter (Fig. 4A, from 5-to 2-fold). The persistence of repression of the mutated Ϫ600-bp promoter in the presence of Nkx3.2 may be due to the Nkx3.2-responsive region between Ϫ458 and Ϫ351 bp observed in Fig. 2B. To minimize the effects of other potential promoter regions through which Nkx3.2 may act, we introduced the mutation of the Nkx3.2 binding sequence into the 108-bp Runx2 promoter construct (Fig. 4B). Whereas this construct has less basal promoter activity than the 600-bp construct (see Fig. 2A), Nkx3.2 still exhibits strong repressive activity upon this promoter fragment (Fig. 4B). Mutation of the Nkx3.2 binding site in this context abolishes Nkx3.2-mediated repression of the Runx2 promoter confirming that Nkx3.2 is exerting repressive effects through this site.
We next addressed whether the observed Nkx3.2-mediated repression of the Runx2 promoter affects physiologic Runx2 levels in C3H10T1/2 cells. These cells were co-transfected with the Nkx3.2 expression vector and an EGFP expression vector followed by FACS sorting to purify EGFP positive cells. Cells positive for EGFP were then lysed and endoge-nous Runx2 protein levels were examined. Expression of Nkx3.2 was verified by Western blotting against the HA epitope tag fused to the Nkx3.2 protein (Fig. 4C). In the absence of Nkx3.2, Runx2 is strongly expressed in C3H10T1/2 cells, whereas cells overexpressing Nkx3.2 exhibit a dramatic loss of Runx2 protein. Taken together, these results demonstrate that Nkx3.2 suppresses the activity of the Runx2 promoter in pluripotent mesenchymal progenitor cells through its interaction with a functional Nkx3.2 binding element in the proximal Runx2 promoter.

Nkx3.2-mediated Repression of Runx2 Is Abrogated in Cells
Committed to the Osseous Lineage-To gain insight into the biological significance of Nkx3.2 repression of the Runx2 gene, we asked whether the observed repression was cell type specific. We found that Nkx3.2-mediated repression of the Runx2 promoter was much greater in undifferentiated mesenchymal cells (C3H10T1/2 and NIH3T3) in comparison to cells already committed to the osteoblast lineage (MC3T3 and ROS17/2.8) (Fig. 5). This phenomenon may reflect a biological role for Runx2 repression by Nkx3.2 in undifferentiated mesenchyme, as Runx2 activity in committed osteoblasts is critical for their ability to differentiate and these cells may therefore have passed a differentiation checkpoint after which Nkx3.2 can no longer exert repressive effects upon the Runx2 gene.
Nkx3.2 and Runx2 Genes Exhibit Reciprocal Expression Patterns during Chondrogenic Differentiation-To assess the significance of the specificity of Nkx3.2-mediated repression of the Runx2 gene to undifferentiated mesenchymal cells, we examined the expression of patterns of endogenous Runx2, Nkx3.2, Sox9, and type II collagen genes during chondrogenic differentiation of C3H10T1/2 cells. These cells proliferate as undifferentiated fibroblasts, and, like primary mesenchymal progenitor cells, are capable of differentiating into several cell types including adipocytes (29,30), osteoblasts (30,31), myoblasts (32), and chondrocytes (33,34), depending on the environment in which they are cultured. Here C3H10T1/2 cells were induced to undergo chondrogenesis by plating them in high-density micromass cultures in the presence of BMP-2 (33).
Gene expression was monitored over a 4-day period after induction of chondrogenesis (Fig. 6). Strong induction of Nkx3.2 expression was observed by 1 day of culture and was closely followed by activation of the expression of the chondrogenic transcriptional regulator Sox9 and the cartilage-specific extracellular matrix protein, type II collagen, demonstrating that these cultures have entered the pathway of chondrocytic differentiation. In contrast, Runx2 expression was strongly repressed by 1 day of culture, consistent with our findings that Nkx3.2 activity suppresses expression of the Runx2 gene. Taken together these findings demonstrate that a loss of Runx2 gene activity is associated with the transition of the C3H10T1/2 mesenchymal progenitor cell into the chondrogenic lineage.
Runx2 Acts to Prevent Mesenchymal Progenitor Cells from Undergoing Chondrogenesis-Based on our observations of Runx2 gene expression during chondrogenesis of C3H10T1/2 cells, we tested the hypothesis that there is a requirement for Runx2 suppression for induction of chondrogenesis. We therefore introduced exogenous Runx2 into C3H10T1/2 cells by adenoviral infection prior to induction of chondrogenesis to circumvent the suppression of Runx2 observed at the onset of chondrogenic differentiation. Proliferating C3H10T1/2 cells were infected with adenovirus containing either Runx2 cDNA or LacZ cDNA as a control, and gene expression was analyzed 24 h after the induction of chondrogenesis with BMP-2 and high-density culture. Adenoviral infection maintains high levels of Runx2 expression 24 h after induction of chondrogenesis in Runx2-infected cells when compared with LacZ-infected control cultures (Fig. 7A). Remarkably, BMP-2-treated micromass cultures infected with Runx2 fail to significantly activate Sox9 or type II collagen genes (Fig. 7A) and are unable to produce significant cartilaginous extracellular matrix, as assessed by Alcian blue staining of sulfonated proteoglycans present in cartilage (Fig. 7B, upper panel). Surprisingly, when analyzing expression of Nkx3.2 in these cultures we found that it was significantly increased in the presence of Runx2, supporting the hypothesis that Nkx3.2 acts upstream of Runx2 and suggesting that a feedback loop exists in which the Nkx3.2 gene becomes induced to suppress Runx2 activity (Fig. 7A). Our results clearly demonstrate that in an undifferentiated mesenchymal progenitor cell, Runx2 is able to inhibit chondrogenic differentiation, and Runx2 repression may be a pre-requisite for entry into the chondrocytic lineage.
Given the role of Runx2 in promoting osteogenesis, we examined whether the inhibition of chondrogenic differentiation in adenoviral Runx2-infected C3H10T1/2 cultures was associated with increased osteogenesis. We examined adenoviral cultures for expression of the osteogenic genes alkaline phosphatase and type I collagen and observed that mRNA levels of these genes were indeed increased in the presence of Runx2 adenovirus (Fig. 7A). When examining the activity of the alkaline phosphatase enzyme in these cultures by colorimetric reaction we observe that alkaline phosphatase-positive cells are local-ized primarily to cells in monolayer at the edges and top surface of the nodule (Fig. 7B, lower panel). These observations suggest that while sustained Runx2 activity under pro-chondrogenic culture conditions prevents the majority of these cells from undergoing chondrogenesis, its expression is capable of inducing an osteogenic response in a subset of these cells. Interestingly, we find that if Runx2 is introduced by adenoviral infection after the onset of chondrogenic differentiation it is unable to suppress chondrocyte phenotypic genes (Fig. 7C). This phenomenon may reflect the passing of a critical threshold during the chondrogenic differentiation of mesenchymal cells after which the activity of Runx2 cannot reverse the differentiation process once the program of chondrogenesis has begun. Taken together, these findings indicate that suppression of Runx2 gene expression by Nkx3.2 is a requirement for the progression of chondrogenic differentiation of mesenchymal progenitor cells.

DISCUSSION
In this study we identify the Runx2 gene as a direct target for repression by the NK-family homeodomain factor Nkx3.2. We establish the biological significance of this repression by demonstrating that inhibition of Runx2 expression at the onset of BMP-2-induced chondrogenesis of C3H10T1/2 mesenchymal progenitor cells is a pre-requisite for cartilage formation. Whereas prior studies have shown that osteogenic differentiation of progenitor cells in response to BMP-2 treatment requires Runx2 activity (35)(36)(37) and that terminal hypertrophy and mineralization of cartilage is driven by Runx2 (38 -48), our results demonstrate for the first time that BMP-2-induced chondrogenic differentiation of mesenchymal progenitor cells requires suppression of Runx2. This finding provides direct evidence that Runx2 is also a critical regulator of BMP-2induced chondrogenic fate determination in pluripotent mesenchymal cells.
We identified Nkx3.2 as a potential regulator of the Runx2 gene based on the pattern of Runx2 promoter activity in precartilaginous mesenchymal condensations of the axial skeleton as early as E9.5 in the murine embryo (12). Cells comprising the mesenchymal condensations first proliferate, then undergo chondrogenic differentiation in response to the coordinated and sequential actions of soluble signaling molecules sonic hedgehog and BMP-2 (1,6,49). The expression of the Runx2 gene in these condensations precedes the activation of Nkx3.2 and overt chondrogenesis. Chondrogenic differentiation of these cells is dependent upon activation of Nkx3.2 by Shh and BMP-2, which ultimately leads to induction of Sox9 (1,6,50). Runx2 expression is absent from mature, Sox9 expressing chondrocytes: thus the repression of Runx2 by Nkx3.2 at the onset of chondrogenesis is a key regulatory event. Interestingly, recent findings by Eames et al. (48) have suggested that interplay between Runx2 and Sox9 creates a binary switch in which Runx2 activity in chondrocytes drives hypertrophy and mineralization, whereas Sox9 activity can maintain chondrocytes in a mature, but not hypertrophic state, thereby inhibiting the pro-hypertrophic effects of Runx2 (44). Whereas these findings and others have demonstrated a role for Runx2 in promoting terminal chondrocyte hypertrophy, our study suggests that Runx2 plays an additional, anti-chondrogenic role in the pluripotent mesenchymal progenitor cell prior to the onset of differentiation and that repression of Runx2 in these progenitors is a prerequisite for the progression of chondrogenesis.
Significantly, we find that the Nkx3.2 regulatory element (CACTT) at Ϫ108 bp in the Runx2 P1 promoter is located in a region that is essential for basal transcriptional activity. This element lies in close proximity to Runx and other regulatory sites and marks the 5Ј boundary of the minimal promoter region required for transcriptional activity. This Nkx3.2 site confers potent, dose-dependent transcriptional repression in the presence of Nkx3.2. Our findings show that Nkx3.2 physically interacts with the Nkx3.2 binding element within the Runx2 promoter, and that this interaction is required for effective repression of Runx2 promoter activity. These results make Runx2 the first bona fide target of Nkx3.2 to be identified to date.
Based on our prior observations of Runx2 promoter activity in vivo and our findings in this study demonstrating that the Runx2 gene is a target of Nkx3.2-mediated transcriptional repression, we investigated the biological significance of Runx2 gene activity in pre-chondrogenic mesenchymal progenitors as they undergo BMP-2-induced chondrogenesis. Although the mechanisms that support Runx2 gene activity in these progenitor cells are unknown, several lines of evidence suggest that Runx2 is responsive to sonic hedgehog signaling. In this study we observe a minor (10%) activation of Runx2 promoter activity in response to Shh treatment of C3H10T1/2 cells. The lack of a robust response to Shh may reflect the fact that these cells have endogenously active Runx2. In support of this, other studies have observed an induction of Runx2 as a result of Shh treatment (51) and have observed that the Runx2 gene is active in the notochord (the source of soluble Shh during embryogenesis) in vivo (11). Thus, the Runx2 gene appears to be responsive to signaling pathways that are essential for early skeletal development.
Later during development of the axial skeleton, BMP-2 acts in concert with Shh to activate Nkx3.2 and induce cartilage formation (50). This sequence of events is recapitulated in this study using the C3H10T1/2 model of mesenchymal chondrogenesis. Upon BMP-2 treatment and high-density culture, Nkx3.2 is induced and Runx2 is concomitantly repressed, supporting our findings that Nkx3.2 is an inhibitor of Runx2 gene expression, and suggesting that the activity of Runx2 must be suppressed prior to activation of the chondrogenic program of gene expression. The observation that Runx2 gene activity is suppressed by Nkx3.2 at the onset of chondrogenesis appears to contrast our previous in vivo observations on the activity of a Runx2 promoter-LacZ reporter transgene. Consistent with our current findings, this transgene is active in mesenchymal progenitor cells in the somites and sclerotome, however, ␤-galactosidase activity persists into the mature chondrocyte in these mice. Perhaps this is because of stability of ␤-galactosidase, which may retain enzymatic activity for several days after inactivation of the Runx2 promoter. Our conclusion that Runx2 suppression is a prerequisite for the initiation of chondrogenesis is further supported by the finding that adenoviral overexpression of Runx2 prior to chondrogenesis inhibits induction of Sox9, type II collagen, and the chondrogenic phenotype. The observation that elevating Runx2 levels after induction of chondrogenesis does not suppress expression of chondrocyte phenotypic genes indicates that a developmental "checkpoint" has been passed. Subsequently, the ability of Runx2 to maintain an undifferentiated progenitor is lost.
Whereas these findings underscore the importance of Runx2 repression at the onset of mesenchymal chondrogenesis, the function of Runx2 in undifferentiated mesenchyme remains unclear. These progenitor cells express Runx2 during proliferative expansion both in vivo and in vitro. This observation raises the question whether Runx2 maintains an undifferentiated phenotype during expansion of the progenitor population. In this study, circumvention of Runx2 repression at the onset of BMP-2-induced cartilage formation efficiently inhibits activation of chondrocyte phenotypic genes and extracellular matrix deposition. It is therefore possible that these cells either remain undifferentiated in the presence of exogenous Runx2 and BMP-2 or, alternatively become osteogenic. However, we did not observe induction of the bone phenotypic marker osteocalcin (not shown) and observed only a minor induction of alkaline phosphatase (an early osteogenic marker) that was restricted to cells at the periphery and on the surface of the micromass cultures. Whereas other studies have observed induction of Runx2 and osteogenic gene expression upon BMP-2 treatment of C3H10T1/2 cells in monolayer cultures, we find that the osteogenic phenotype cannot be induced in high-density micromass cultures despite the high levels of exogenous Runx2 and BMP-2. These findings suggest that cell-cell contact in highdensity cultures predisposes cells to undergo chondrogenesis in response to BMP-2 signaling.
Our findings further suggest that a feedback loop may exist in which high levels of Runx2 positively affect the Nkx3.2 gene in order for Nkx3.2 to repress Runx2 and thereby promote chondrogenesis. This conclusion is based on the observation that BMP-2-treated micromass cultures infected with Runx2 adenovirus exhibit increased Nkx3.2 expression in comparison to lacZ-infected controls. These findings are of particular interest in light of the observation that the ability of Nkx3.2 to repress Runx2 gene expression is severely abrogated in committed osteoprogenitor cells, suggesting that Runx2 may act as a switching mechanism for differentiation induced by BMP signaling. In an undifferentiated mesenchymal progenitor cell, Runx2 may maintain a pluripotent state until receipt of a BMP-2 signal. Depending on the microenvironment of the progenitor cell, BMP-2 can induce Nkx3.2 to suppress Runx2 and promote chondrogenic differentiation of the pluripotent progenitor (Fig. 8). Once chondrogenesis has begun and Sox9 is activated, Runx2 cannot reverse the differentiation process. If the BMP-2 signal is osteogenic, the progenitor cell will further induce Runx2 and activate Osterix, another essential pro-osteogenic transcription factor whose activities, along with those of Runx2, are responsible for osteoblast formation (52)(53)(54). In this context, the committed pre-osteoblast has passed another checkpoint in differentiation where the activity of Nkx3.2 is no longer competent to repress Runx2 gene expression and therefore no longer able to promote entry into the chondrogenic lineage.
In conclusion, we have identified a novel role for Runx2 as a critical component of the differentiation program that is initiated by activation of the sonic hedgehog and BMP signaling pathways and ultimately results in the chondrogenic differentiation of mesenchymal progenitor cells and formation of the cartilaginous template of the vertebrate skeleton. Previous findings have shown in vivo and in monolayer cultures of mesenchymal progenitor cells that BMP-2 induces the osteogenic phenotype via activation of Runx2. Runx2 is necessary for activation of Osterix and other osteoblast phenotypic genes. Runx2 and BMP-2 activities are coordinated for expression of Osterix and other osteoblast phenotype genes (52,53,56). In condensing mesenchyme or high-density micromass cultures, BMP-2 activity activates Nkx3.2, which in turn suppresses Runx2 transcription. This repression of Runx2 allows for activation of Sox9 and expression of chondrocyte phenotypic genes. Activation of Nkx3.2 by BMP-2 therefore indirectly leads to Sox9 expression and chondrogenesis.