Runx3 negatively regulates Osterix expression in dental pulp cells

Abstract

Osterix, a zinc-finger-containing transcription factor, is required for osteoblast differentiation and bone formation. Osterix is also expressed in dental mesenchymal cells of the tooth germ. However, transcriptional regulation by Osterix in tooth development is not clear. Genetic studies in osteogenesis place Osterix downstream of Runx2 (Runt-related 2). The expression of Osterix in odontoblasts overlaps with Runx3 during terminal differentiation in vivo. Runx3 down-regulates Osterix expression in mouse DPCs (dental pulp cells). Therefore the regulatory role of Runx3 on Osterix expression in tooth development was investigated. Enforced expression of Runx3 down-regulated the activity of the Osterix promoter in the human embryonic kidney 293 cell line. When the Runx3 responsive element on the Osterix promoter, located at −713 to −707 bp (site 3, AGTGGTT) relative to the cap site, was mutated, this down-regulation was abrogated. Furthermore, electrophoretic mobility-shift assay and chromatin immunoprecipitation assays in mouse DPCs demonstrated direct functional binding of Runx3 to the Osterix promoter. These results demonstrate the transcriptional regulation of Osterix expression by Runx3 during differentiation of dental pulp cells into odontoblasts during tooth development.

INTRODUCTION

The transcriptional regulation of cell proliferation and differentiation by the Runx (Runt-related) family of DNA-binding transcription factors is critical for both morphogenesis and regeneration. The regulatory function of the Runx family on the promoters and enhancers of target genes, where they associate with cofactors and other DNA-binding transcription factors to modulate gene expression, is well known [1]. The Runx family is composed of three members designated Runx1 (AML1/Cbfα2), Runx2 (AML2/Cbfα1) and Runx3 (AML3/Cbfα3) [2,3]. Although the Runx members share highly conserved DNA-binding domains, they regulate distinct functions [4–7]. Runx1 is involved in the regulation of haematopoiesis [8], Runx2 is essential for bone and tooth development [9–11], and Runx3 is critical for gastric epithelial differentiation, neurogenesis of the dorsal root ganglia and T cell differentiation [8–10,12–16].

Stringent control of gene activation and suppression is required for tooth development. The optimal gene expression during dentin formation is dependent on integration and regulation of signals that govern the commitment of stem/progenitor cells into the pulp cell lineage, and their subsequent proliferation and differentiation into odontoblasts. Runx2 is essential for tooth formation. Molar development is arrested at the late bud stage in Runx2 homozygous mice [11], correlating with the intense expression of Runx2 in the dental mesenchyme during the bud and cap stages [17]. Runx3 is co-expressed in dental papillae at the cap and early bell stages, along with Runx2. Later Runx3 is restricted to the odontoblastic layer at the late bell stage, while Runx2 is no longer detected [17]. Runx proteins might play a pivotal role in governing the control of the physiological response of dental genes.

Osterix, a zinc-finger-containing transcription factor, is required for osteoblast differentiation and bone formation [18]. In Osterix null mice, no bone formation occurs, similar to the phenotypes in Runx2 null mice [9,18]. However, Runx2 is expressed without major alterations in Osterix null mice. In contrast, Osterix is not expressed in Runx2 null mice, demonstrating that Osterix acts downstream of Runx2 [18]. Transcriptional regulation of Osterix by Runx2 in cartilage has been suggested recently [19]. In addition, Osterix expression has been observed in mesenchymal cells of the tooth germ [18]. The expression of Osterix and its transcriptional regulation by Runx proteins during tooth development have not been investigated.

In the present study, we investigated the expression of Osterix during tooth development, and demonstrate that Osterix is expressed strictly in the odontoblastic layer at the bell and the differentiation stages, overlapping with Runx3. Therefore the regulation of the expression of Osterix by Runx3 was examined further. Our results demonstrate that Runx3 binds directly to the Osterix promoter and down-regulates its expression in DPCs (dental pulp cells).

EXPERIMENTAL

Cloning of the Osterix promoter

To clone the Osterix promoter (nucleotides 66 to 1751; GenBank accession no. DQ229136), genomic DNA was isolated from the tail of an ICR mouse. PCR was performed using two primers: Osterix promoter 5′-1, 5′-TCTGTCCCTCAGTCCTGCTT-3′ and Osterix promoter 3′-2, 5-GGGCAAGTTGTCAGAGCTTC-3′. The approx. 1.7 kbp PCR product was then subcloned into MluI and XhoI digested pGL3-promoter vector (Promega, Madison, WI, U.S.A.), and named pOsx1.7-luc. To prepare the MSCV (murine stem cell virus)-EGFP (enhanced green fluorescent protein)-FLAG-Runx3 expression vector, the following primers were used: FLAG-Runx3-5′, 5′-GGCAGATCTGCCACCATGGACTACAAGGACGATGACGACAAGGCTTCCAACAGCATCTTTG-3′ and Flag-Runx3-3′, 5′-ATATGAGCTCTCCCGCGTGGT-3′ to generate a Runx3 fragment with FLAG motif at N-terminal. The 300 bp PCR product was cloned in between the BglII and SacI sites in the pSL1180 vector (GE Healthcare, Buckinghamshire, U.K.) and named Flag-Runx3-300 bp-pSL1180. A 1.0 kbp Runx3 fragment was digested with SacI from the MSCV-EGFPRunx3 plasmid (kindly provided by Dr Taniuchi Ichiro, Laboratory of Transcriptional Regulation, RIKEN Research Centre for Allergy and Immunology, Yokohama, Japan) and subcloned into the FLAG-Runx3-300 bp-pSL1180 vector to give FLAG-Runx3-pSL1180. The N-terminally FLAG-tagged full-length 1.3 kbp Runx3 was digested with BglII from FLAG-Runx3-pSL1180 and subcloned into the MSCV-EGFP vector, named MSCV-EGFP-FLAG-Runx3. The orientation of the inserts was confirmed by sequencing.

Site-directed mutagenesis

Three putative Runx2-binding sequences at positions −1823 to −1817 bp, −1776 to −1771 bp and −713 to −707 bp relative to the Cap site [19] were mutated using the QuikChange® Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, U.S.A.) according to the manufacturer's recommendations. We generated mutants as follows; 5′-AACCACA-3′ at −1823/−1817 bp was changed into 5′-GAGCTCA-3′, 5′-ACCACT-3′ at −1776/−1771 bp was changed into 5′-GCTACT-3′ and 5′-AGTGGTT-3′ at −713/−707 bp was changed into 5′-ATAGACT-3′. The mutated nucleotides are indicated in bold. Mutations in single, double, and triple motifs were termed M1-M5 (Figure 3B). Incorporation of the mutated substitutions of all the constructs was confirmed by sequencing.

Figure 3. Down-regulation of Osterix promoter activity by Runx3 in HEK-293 cells.

(A) Determination of endogenous expression levels of Runx3, Runx2 and Osterix in mouse DPCs and HEK-293 cells. (B) Wild-type (WT) and mutant (MT) Osterix promoter plasmids were analysed 48 h after co-transfection with MSCV-EGFP-Flag-Runx3 into HEK293 cells. (C) Shortened wild-type (WT) or mutant (MT) Osterix promoter plasmids containing only site 3 (−713 to −707) were co-transfected with MSCV-EGFP-Flag-Runx3 into HEK-293 cells. The activities were determined after 48 h and normalized against the co-transfected internal control plasmid (pRL-SV40). The values represent means±S.D. for four individual samples. The experiment was repeated twice with similar results. **, P<0.01 compared with the empty MSCV plasmid. NC, negative control.

In situ hybridization

ICR mouse embryos at 15.0 days post coitum, 17.0 days post coitum and P1 (postnatal day 1) were fixed in 4% (w/v) paraformaldehyde at 4 °C overnight. In situ hybridization was carried out as described previously [20]. The following primers were used to amplify the mouse Osterix cDNA: Osterix-5′-1, 5′-GGTCCAGGCAACACACCTAC-3′ and Osterix-3′-2, 5′-GGTAGGGAGCTGGGTTAAGG-3′. The PCR product was ligated into the pBluescript II SK (−) vector (Stratagene). Mouse Runx3 cDNA was removed from the MSCV-EGFP-Runx3 plasmid by digestion with EcoRI and then subcloned into the pBluescript II SK (−) vector. All inserts were confirmed by sequencing. The following cDNAs were used to generate sense (see Supplementary Figure 1 at http://www.BiochemJ.org/bj/405/bj4050069add.htm) and antisense riboprobes using either T3 or T7 RNA polymerase: a 184 bp murine Osterix fragment, a 1.2 kb Runx3 fragment and a 1.2 kb Bmp2 fragment. In situ hybridization was performed as described previously [21].

Cell culture and transfection studies

Mouse DPCs were isolated from tooth germ at 17.0 days post coitum. Mouse DPCs and HEK-293 (human embryonic kidney 293) cells were maintained in DMEM (Dulbecco's modified Eagle's medium) (Sigma, St. Louis, MO, U.S.A.) supplemented with 100 units/ml penicillin G, 100 μg/ml streptomycin (Invitrogen, Carlsbad, CA, U.S.A.) and 10% (v/v) FBS (foetal bovine serum; SAFC Biosciences, Lenexa, KS, U.S.A.). Experiments assessing promoter activity by luciferase reporter gene expression were performed as follows. HEK-293 cells (1×10⁵) were plated in 24-well plates in serum-free DMEM without antibiotics 1 day before use, and transiently transfected with 2 μg of the promoter–luciferase reporter gene plasmids, 3 μg of expression plasmid and 0.2 μg of SV40 (simian virus 40) promoter construct (Promega) as an internal standardized control for transfection efficiency. Transfections were performed using 2 μl/well of Lipofectamine™ 2000 (Invitrogen) following the manufacturer's instructions. The MSCV-EGFP plasmid was also transfected as a control. After 4 h, the medium was replaced with DMEM supplemented with 10% (v/v) FBS and cultured for an additional 44 h. Cells were then lysed, and luciferase activity was determined using a Dual Luciferase Reporter Assay kit as instructed by the manufacturer (Promega). All activities were normalized against the co-transfected internal control plasmid pRL-SV40 (Promega). For overexpression experiments, 4×10⁶ DPCs were transfected with 8 μg of expression plasmid using an ECM 830 Electroporator (BTX, San Diego, CA, U.S.A.) following the manufacturer's instructions, then plated on to a collagen type I-coated 35-mm-diameter dish (Iwaki, Chiba, Japan). After 4 h, the medium was replaced with DMEM and 10% (v/v) FBS. Cells were harvested at 0, 24 and 48 h after transfection. The cell viability was determined with Trypan Blue soon after transfection, and the efficiency was estimated by fluorescent microscopy 24 h after transfection with the plasmid vector AFP (kindly provided by Dr Hidesato Ogawa, Graduate School of Biological Sciences, Nara Institute of Science and Technology, Japan).

Real time RT (reverse transcriptase)-PCR analysis

Total RNA was extracted by using Trizol (Invitrogen), and 2 μg of freshly isolated RNA was reverse transcribed with SuperScript II RT (Invitrogen) following the manufacturer's recommendations. The resulting cDNA was then amplified by real time RT-PCR with a Light Cycler-FastStart DNA master SYBR Green I (Roche Diagnostics, Mannheim, Germany). The primers used in the RT-PCR analysis are presented in Table 1.

Table 1. Primers for RT-PCR.

Name	Direction	Sequence	Product size (bp)	Accession number
β-Actin	Forward	5′-AAATCGTGCGTGACATCAAA-3′
	Reverse	5′-AAGGAAGGCTGGAAAAGAGC-3′	178	X03765
Runx3	Forward	5′-GGTTCAACGACCTTCGATTC-3′
	Reverse	5′-AGGCCTTGGTCTGGTCTTCT-3′	180	NM-019732
Runx2	Forward	5′-CAGACCAGCAGCACTCCATA-3′
	Reverse	5′-CAGCGTCAACACCATCATTC-3′	178	NM-009820
Osterix	Forward	5′-GGTCCAGGCAACACACCTAC-3′
	Reverse	5′-GGTAGGGAGCTGGGTTAAGG-3′	178	AF184902
Dspp	Forward	5′-GGAACTGCAGCACAGAATGA-3′
	Reverse	5′-CAGTGTTCCCCTGTTCGTTT-3′	199	NM-010080
Enamelysin	Forward	5′-CGACAATGCTGAGAAGTGGA-3′
	Reverse	5′-CCCTTTCACATCATCCTTGG-3′	180	NM-013903
Klk4	Forward	5′-TTGCAAACGATCTCATGCTC-3′
	Reverse	5′-TGAGGTGGTACACAGGGTCA-3′	228	NM-019928

Preparation of nuclear extracts

Nuclear extract was isolated as described previously [22]. Briefly, mouse DPCs were washed with 10 ml of PBS, scraped into 1.5 ml of ice-cold PBS, and centrifuged at 100 g for 5 min. The pellet was suspended in 1 ml of PBS and centrifuged again at 660 g for 15 s. After resuspension in cold buffer A [10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1mM DTT (dithiothreitol) and 0.5 mM PMSF] on ice for 15 min, the cell membranes were lysed by 0.5% Nonidet P40 and then centrifuged at 660 g for 30 s. The pelleted nuclei were resuspended in cold buffer C (20 mM Hepes, pH 7.9, 0.4 M NaCl, 1mM EDTA, 1 mM EGTA, 1 mM DTT and 1 mM PMSF). The nuclear protein was extracted by shaking at 4 °C for 15 min, followed by centrifugation at 15000 g for 5 min and the supernatant fractions were collected. The protein content of the nuclear extracts was determined using the Bradford method [23].

EMSA (electrophoretic mobility-shift assay)

Individual oligonucleotides were annealed to equimolar amounts of their complementary strands (wild-type, Osterix-gel-WT5′-1: 5′-CAGATCTCTAATTAGTGGTTTGGGGTTTGTTCCTTTTC-3′ and Osterix-gel-WT-3′-2: 5′-GAAAAGGAACAAACCCCAAACCACTAATTAGAGATCTG-3′; mutant, Osterix-gel-MT-5′-1: 5′-CAGATCTCTAATTATAGACTTGGGGTTTGTTCCTTTTC-3′ and Osterix-gel-MT-3′-2: 5′-GAAAAGGAACAAACCCCAAGTCTATAATTAGAGATCTG-3′) by heating to 95 °C for 5 min and cooling slowly to room temperature (25 °C). DIG (digoxigenin) Gel Shift Kit, 2nd generation (Roche Diagnostics) was used in the EMSA according to the manufacturer's protocol. Briefly, wild-type double-stranded oligonucleotide probes were labelled with DIG-11-ddUTP at the 3′-ends. The labelled probes (20 fmol) were added to 10 μg of nuclear extract in a binding buffer [20 mM Hepes, pH 7.6, 1 mM EDTA, 10 mM (NH₄)₂SO₄, 1 mM DTT, 0.2% (w/v) Tween 20, 30 mM KCl, 25 ng/μl poly(dI-dC)·(dI-dC), 25 ng/μl poly(dA-dT)·(dA-dT) and 50 ng/μl poly L-lysine] at room temperature for 30 min. For competition experiments, 125-fold unlabelled oligonucleotides were added to the mixture. After incubation, the protein–DNA complexes were separated by native PAGE (6% gels), transferred on to a nylon membrane (Whatman, New Jersey, NJ, U.S.A.) by contact-blotting, and detected by the DIG-detection kit. An anti-Runx3 antibody (Active Motif, Carlsbad, CA, U.S.A.) was used to examine the specificity of the protein–DNA complexes.

ChIP (chromatin immunoprecipitation) assay

Mouse DPCs were treated for 10 min with 1% formaldehyde and washed three times with ice-cold PBS. The cells were harvested and centrifuged at 100 g for 5 min. The pellet was suspended in 200 μl of SDS lysis buffer [50 mM Tris/HCl, pH 8.0, 10 mM EDTA, 1% (w/v) SDS, 1 mM PMSF, 1 μg/ml aprotinin and 1 μg/ml leupeptin] and incubated on ice for 20 min. The sample was sonicated for 7.5 min (high power, on 30 s, off 1 min) using a Bioruptor (Cosmo Bio, Tokyo, Japan) to produce soluble chromatin with an average size of 500 bp. The chromatin sample was then diluted 9-fold in ice-cold ChIP dilution buffer [50 mM Tris/HCl, pH 8.0, 167 mM NaCl, 1.1% (v/v) Triton X-100, 0.11% sodium deoxycholate, 1 mM PMSF, 1 μg/ml aprotinin and 1 μg/ml leupeptin]. From the diluted sample, 200 μl was removed as the input fraction and kept at 4 °C. The rest of the sample was pre-cleared for 6 h at 4 °C by incubation with 60 μl of protein G–Sepharose beads pre-blocked with salmon sperm DNA. The beads were removed by centrifugation at 10000 g for 10 s and the supernatant was collected. Rabbit anti-Runx3 polyclonal antibody (20 μg; Active Motif, Carlsbad, CA, U.S.A.) or 10 μg of goat anti-mouse Runx2 polyclonal antibody (Santa Cruz, CA, U.S.A.) was added and incubated overnight at 4 °C. To collect the immunocomplex, 60 μl of protein G–Sepharose beads pre-blocked with salmon sperm DNA were added to the samples for 3 h at 4 °C. The beads were washed once in each of the following buffers, in order: low salt, high salt and LiCl wash solution, and were then washed twice in TE buffer. The bound protein–DNA immunocomplexes were eluted twice with 200 μl of ChIP direct elution buffer (10 mM Tris/HCl, pH 8.0, 300 mM NaCl, 5 mM EDTA and 0.5% SDS) and subjected to reverse cross-linking at 65 °C for 6 h. The reverse cross-linked chromatin DNA was further purified by 50 μg/ml proteinase K digestion at 55 °C for 1 h and phenol/chloroform extraction. DNA was then precipitated in ethanol and dissolved in 20 μl of TE buffer. DNA (2 μl) was used for each PCR with primers Osx-ChIP-F: 5′-GAGTGTCGTCCCCAATCC-3′ and Osx-ChIP-R: 5′-CTGCTACCACCGAGGCTG-3′, yielding a 120 bp product. As a negative control, another 1×10⁷ mouse DPCs was treated as above, except 20 μg rabbit IgG or 10 μg goat IgG antibodies were used instead of specific antibodies. Input (diluted 1:20) was used as the positive control for PCR.

Statistics

Statistical analyses were performed using Student's unpaired t test. Each experiment was performed at least twice, and the representative data are presented as means±S.D. for at least three independent replicates.

RESULTS

Expression of Runx3, Runx2, Osterix and Bmp2 during tooth development

In the developing tooth, Runx3 was detected in the dental papillae at the late cap stage (15.0 days post coitum). Runx3 was progressively restricted to the odontoblastic layer of tooth germ from the bell stage (17.0 days post coitum) until the differentiation stage, P1 during terminal differentiation of odontoblasts (Figures 1A–1D). In contrast, Osterix was first detected in the odontoblastic layer at 17.0 days post coitum, and was more pronounced at P1 and P4 (Figures 1E–1H) and had overlapping expression with Runx3. In the odontoblasts, Bmp2 also was strongly expressed at P1 (Figure 1O), but not Runx2 (Figure 1K). No positive signal was detected when using sense probes.

Figure 1. Expression of Runx3, Osterix, Runx2 and Bmp2 by in situ hybridization during tooth development in mouse.

(A–D) Runx3 was progressively restricted to the odontoblastic layer of tooth germ starting from the bell stage (17.0 days post coitum) to the differentiation stage (P1) during terminal differentiation of odontoblasts. (E–H) Osterix was first detected weakly in the odontoblastic layer at 17.0 days post coitum, and was a more pronounced at P1, overlapping with Runx3 expression. (I–L) Runx2 was not expressed in odontoblast layer after P1. (M–P) Bmp2 was strongly expressed in the odontoblasts at P1. Arrowheads indicate the positive signals in the odontoblastic layer. dp, dental papillae; ol, odontoblast layer. Scale bar=200 μm.

Expression of Runx3 and Osterix during differentiation of the dental pulp cells into odontoblasts in vitro

We next determined whether the mouse DPCs have in vitro expression patterns of Runx3 and Osterix similar to those observed in vivo. RT-PCR was performed to examine gene expression of Runx3, Osterix and the odontoblast markers Dspp (dentin sialophosphoprotein), enamelysin and KLK4 (kallikrein 4) in cell culture (Figure 2A). Dspp and KLK4 were first detected clearly on day 21 and enamelysin on day 28, showing spontaneous differentiation of the DPCs into odontoblasts. Runx3 expression was weakly detected on day 1, and increased further on day 21. Osterix expression was first detected on day 21 (Figure 2A). These results correlated with the in vivo expression during tooth development, suggesting that the DPCs might be useful for the study of the regulation of expression of Osterix by Runx3 at the stage before terminal differentiation of odontoblasts.

Figure 2. Down-regulation of Osterix expression by Runx3 in mouse DPCs in vitro.

(A) mRNA expression levels for Runx3, Osterix, and for the differentiation markers of odontoblasts, Dspp, enamelysin and KLK4 in mouse DPCs in cell culture. (B) Osterix expression was down-regulated in mouse DPCs at 48 h after Runx3 transfection. The experiment was repeated twice with similar results.

Runx3 down-regulates Osterix expression in mouse DPCs

To examine whether Osterix expression was regulated by Runx3, MSCV-EGFP-Flag-Runx3 was transfected by electroporation into the mouse DPCs. Electroporation, at three square-wave pulses at a frequency of 1 Hz with a pulse length of 99 μs and 1350 V, provided an optimal method for gene transfer in vitro. The cell viability was nearly 70%, as determined with Trypan Blue staining, and the efficiency was nearly 35%, as estimated by fluorescence microscopy. Real-time RT-PCR showed that the expression of Runx3 mRNA was enhanced approx. 3-fold in DPCs transfected with MSCV-EGFP-Flag-Runx3 compared with control DPCs transfected with MSCV-EGFP after 24 h (results not shown). Runx3 mRNA expression levels, however, were reduced to almost the same level as the control 48 h after transfection. On the other hand, Osterix expression was reduced by 25% 48 h after transfection with MSCV-EGFP-Flag-Runx3 compared with control transfections (Figure 2B). These results suggest that Runx3 negatively regulates Osterix expression in the DPCs.

Runx3 down-regulates the Osterix promoter activity in HEK-293 cells

A recent study has shown that Runx2 specifically up-regulated Osterix promoter activity in C3H10T1/2 and ATDC5 cells, which are mesenchymal cell lines from bone and cartilage respectively [19]. There has, so far, been no report concerning Osterix regulation by Runx3. Runx3 shares highly conserved DNA-binding domains with Runx2. Both the Runx2 and the Runx3 promoters have putative Runx-binding sites that are fully conserved in sequence and location [24], therefore cross-regulation between Runx2 and Runx3 might be plausible. To avoid this possible endogenous effect, HEK-293 cells, in which neither Runx2 nor Runx3 are expressed (Figure 3A), were used to examine transcriptional activity induced by Runx3.

Three putative Runx-binding sites were identified at positions −1823 to −1817 bp (site 1, ACCACA), −1776 to −1771 bp (site 2, ACCACT) and −713 to −707 bp (site 3, AGTGGTT) relative to the cap site by computer analysis of the Osterix promoter (Figure 3B). A wild-type luciferase reporter plasmid containing all the three putative Runx-binding sites was compared with the Osterix promoter containing five different mutations (M1–M5), in which some of the three putative sites were mutated (Figure 3B). Cells in which the wild-type reporter plasmid was co-transfected with Runx3 reduced the Osterix promoter activity to approx. 55%. Transfection of the mutant reporters in which site 1 and/or site 2 were mutated resulted in almost the same reduced activity as that of the wild-type promoter. In contrast, in the mutant reporters in which site 3 was mutated (M4 and M5), only a weak repression (approx. 90% activity) was detected (Figure 3B). These results suggest that site 3 is essential for Osterix promoter activity. To confirm this, shorter wild-type and mutant plasmids containing only site 3 were used. The Osterix promoter activity was significantly reduced in cells transfected with the wild-type construct, whereas the cells transfected with the mutant construct were unaffected (Figure 3C). These results suggest that site 3 is essential for Osterix promoter activity.

Characterization of Runx3 binding to site 3

To determine whether transcriptional repression of Osterix is due to direct binding of Runx3 to site 3, EMSAs using nuclear extracts from the mouse DPCs were performed. As shown in Figure 4(A), 38 bp end-labelled oligonucleotide containing the site 3 (−713 to −707 bp) of the Osterix promoter formed a DNA–protein complex (Figure 4A, lane 2, arrowhead). The complex was competed out completely by a 125-fold excess of unlabelled wild-type oligonucleotide (Figure 4A, lane 3). An oligonucleotide in which site 3 was mutated did not affect this binding (Figure 4A, lane 4). Furthermore, an anti-Runx3 antibody bound to the DNA–protein complex (Figure 4A, lane 5, arrow), indicating the specificity of the DNA–protein complex. No band could be detected when only nuclear extract was loaded (Figure 4A, lane 6).

Figure 4. Characterization of Runx3 binding to the Osterix promoter by EMSA and ChIP assay.

(A) Interaction of nuclear extract (NE) of mouse DPCs with Osterix promoter −713 to −707 bp sequence (site 3) in EMSA. The arrowhead indicates the retarded protein–DNA complexes. The arrow indicates the super-shift band in the reaction with the addition of an anti-Runx3 antibody (lane 5). Lane 1 shows a binding reaction with no protein. Lane 2 is a control reaction with no competing oligonucleotides. Lanes 3 and 4 represent competition reactions with wild-type (WT) or mutated (MT) unlabelled oligonucleotides. Lane 6, nuclear extract only. (B) ChIP assays were performed to investigate whether both Runx3 and Runx2 could bind to the Osterix promoter in vivo. A 120-bp band could be detected by PCR in both anti-Runx3 and anti-Runx2 antibody treated samples, but not in anti-rabbit or anti-goat IgG treated samples. Ab, antibody; NC, negative control.

Next, we performed ChIP assays to test if Runx3 binds specifically to the putative response element in vivo. Using mouse DPCs, proteins were cross-linked on to chromatin and immunoprecipitated with an anti-Runx3 antibody. The presence of the Osterix promoter DNA was detected by PCR using primers flanking site 3 (−713 to −707 bp) (Figure 4B), indicating that Runx3 binds to site 3 of Osterix promoter both specifically and functionally. The use of an anti-Runx2 antibody resulted in a similar result (Figure 4B), suggesting that both Runx3 and Runx2 are able to bind site 3 in vivo.

DISCUSSION

During a systematic in situ hybridization study of tooth development, Osterix mRNA was first detected in terminally differentiating odontoblasts and showed co-localization with Runx3, suggesting a potential role for both genes in odontoblast differentiation. Runx3 overexpression resulted in down-regulation of Osterix in the mouse DPCs. This suggests that Osterix might be a downstream target of Runx3 in tooth development. Osterix null mice [18] have a similar phenotype to the Runx2 null mice [9,10], in which both intramembranous and endochondral bone are not formed due to the lack of osteoblast differentiation. Whereas Osterix is not expressed in the Runx2 null mutants, Runx2 expression is not changed in the Osterix null mutants [18]. These genetic studies have placed Osterix downstream of Runx2 [18]. The precise regulatory role of Runx2 in Osterix expression is not clear. A recent study has shown that a 737 bp fragment of the Osterix promoter is up-regulated upon Runx2 overexpression in ATDC5 chondroprogenitor cells, and the function of the 737 bp fragment was confirmed by site-directed mutagenesis experiments [19]. Furthermore, this functional binding site is conserved among mouse, rat and human, showing conservation of the DNA-binding site [19]. However, no information is available on the regulation of Osterix expression by Runx3. Therefore, we have performed transient co-transfection, EMSAs and ChIP assays to investigate the relationship between Runx3 and Osterix in DPCs. Structural dissection of the proximal promoter of the Osterix gene revealed the presence of three putative Runx-binding sites. Only site 3 (−713 to −707 bp) of the identified sites was preferentially and functionally occupied by Runx3. The disruption of site 3 leads to increased Osterix promoter activity in HEK-293 cells, in which both Runx2 and Runx3 are not expressed endogenously. These results indicate that Osterix expression is negatively regulated by Runx3. Furthermore, our EMSAs and ChIP assays confirmed that Runx3 directly down-regulates Osterix expression in DPCs prior to terminal differentiation into odontoblasts. It is noteworthy that Runx3 negatively regulates CD36 expression in myeloid cells [25] and suppresses gastric epithelial cell growth [26], implying a general role for Runx3 in transcriptional repression.

The distinct roles of Runx2 and Runx3 in odontoblast differentiation are not clear. Previous research indicated that tooth development was disrupted in the cap/early bell stages in the Runx2 null mice and no overt differentiation of odontoblasts was observed [11,27]. There was no conspicuous phenotype in teeth of Runx3 null mice [17]. In Runx2 null mice, Runx3 expression was enhanced dramatically in the mesenchyme of upper molars, and they differentiated into odontoblasts [27]. Our EMSAs and ChIP assays have shown that not only Runx2, but also Runx3 binds to site 3 of the Osterix promoter. Runx3 overexpression resulted in down-regulation of Osterix in DPCs. The Osterix promoter activity was down-regulated by Runx3 transfection in HEK-293 cells. These results suggest that Osterix expression is co-operatively regulated by Runx2 and that Runx3 is sharing the same binding site on the Osterix promoter. Thus, Runx3 might co-operate with Runx2 to regulate Osterix expression during odontoblast differentiation. The role of Osterix in tooth development is not clear. In skeletal development, Runx2, Runx3 and Osterix play pivotal roles in osteoblast differentiation and hypertrophic chondrocyte maturation [28,29]. Osterix may play a role in segregation of osteoblast and chondrocyte lineages [29,30]. Runx2 and Runx3 are co-expressed in the early stages of tooth development. There is overlapping expression of Osterix with Runx3, but not with Runx2 in terminal differentiation of odontoblasts. Therefore Osterix may play a role in lineage commitment of odontoblasts in tooth development. The diverse transcriptional outcomes of Runx activity are dependent on context [1]. The Runx family members act as organizing factors on the promoters of target genes, where they associate with co-activators and other DNA-binding transcription factors, including Smads [1]. Repression of Osterix by Runx3 in DPCs is an example of context-dependent regulation of lineage commitment. Thus, there might be co-operative interactions among BMPs, Smads, Runx2 and Runx3 in the regulation of Osterix expression during DPC differentiation into odontoblasts.