3313233140, September 30, 2005 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.
Canonical WNT Signaling Promotes Osteogenesis by Directly Stimulating Runx2 Gene Expression*
Received for publication, January 18, 2005, and in revised form, June 3, 2005 Published, JBC Papers in Press, July 25, 2005, DOI 10.1074/jbc.M500608200
Tripti Gaur, Christopher J. Lengner, Hayk Hovhannisyan, Ramesh A. Bhat, Peter V. N. Bodine, Barry S. Komm, Amjad Javed, Andre J. van Wijnen, Janet L. Stein, Gary S. Stein, and Jane B. Lian1 From the Department of Cell Biology and the Cancer Center, University of Massachusetts Medical School, Worcester, Massachusetts 01655-0106 and the Womens Health Research Institute, Wyeth Research, Collegeville, Pennsylvania 19426
Both activating and null mutations of proteins required for canonical WNT signaling have revealed the importance of this pathway for normal skeletal development. However, tissue-specific transcriptional mechanisms through which WNT signaling promotes the differentiation of bone-forming cells have yet to be identified. Here, we address the hypothesis that canonical WNT signaling and the bone-related transcription factor RUNX2/CBFA1/ AML3 are functionally linked components of a pathway required for the onset of osteoblast differentiation. Our findings show that, in bone of the SFRP1 (secreted frizzled-related protein-1)-null mouse, which exhibits activated WNT signaling and a high bone mass phenotype, there is a significant increase in expression of T-cell factor (TCF)-1, Runx2, and the RUNX2 target gene osteocalcin. We demonstrate by mutational analysis that a functional TCF regulatory element responsive to canonical WNT signaling resides in the promoter of the Runx2 gene (97 to 93). By chromatin immunoprecipitation, recruitment of -catenin and TCF1 to the endogenous Runx2 gene is shown. Coexpression of TCF1 with canonical WNT proteins resulted in a 25-fold activation of Runx2 promoter activity and a 7 8-fold induction of endogenous mRNA in mouse pluripotent mesenchymal and osteoprogenitor cells. This enhancement was abrogated by SFRP1. Taken together, our results provide evidence for direct regulation of Runx2 by canonical WNT signaling and suggest that Runx2 is a target of -catenin/TCF1 for the stimulation of bone formation. We propose that WNT/TCF1 signaling, like bone morphogenetic protein/transforming growth factor- signaling, activates Runx2 gene expression in mesenchymal cells for the control of osteoblast differentiation and skeletal development. have functions related to cell specification, formation of the body plan, cell growth, differentiation and apoptosis (7, 8). WNT proteins function through Frizzled receptors, which transduce the signal through either the canonical -catenin pathway or non-canonical pathway (7, 8). Activation of the Frizzled receptor complex results in inhibition of a phosphorylation cascade that stabilizes intracellular -catenin levels. -Catenin is subsequently translocated to the nucleus to form a transcriptionally active heterodimeric -catenin TCF2 lymphoid enhancer factor (LEF) DNA-binding complex. Both gain- and loss-of-function mutations in canonical and non-canonical WNT signaling components have revealed critical requirements of WNT proteins for normal skeletogenesis (3, 4). Altered expression of several WNT proteins (WNT3a, WNT4, WNT5a, WNT7a, and WNT14/9a) causes defects in somite formation, chondrogenesis, limb development, and endochondral bone formation (9 13). Recently, a rare human genetic disorder (tetra-amaelia) characterized by absence of all limbs has been linked to mutation in Wnt3 (14). More direct effects mediated through canonical WNT signaling on formation and turnover of the mature skeleton have been revealed. An activating mutation in the WNT coreceptor LRP5 (low density lipoprotein receptor-related protein-5) results in the high bone mass trait in humans (15, 16), a phenotype that is reproduced in the mouse model (17, 18). Consistent with this phenotype, the LRP5 loss-of-function mutation exhibits osteopenia and decreased bone mass in humans (19) and in a mouse model (20). Targeted disruption of axin-2 (a negative regulator of the canonical WNT pathway) in mice leads to increased osteoblast differentiation and matrix mineralization (21). Embryonic lethality of the -catenin-null mouse (22), lack of skeletal structures derived from the cranial neural crest (23), and arrest of osteoblast differentiation (24 26) in conditional -catenin mutants as well as transgenic mouse models (27) demonstrate the importance of the canonical WNT pathway during early developmental stages and for bone formation. The WNT pathway is negatively regulated by several modulators, including secreted frizzled-related proteins (SFRPs), which contain a cysteine-rich domain for interaction with WNT proteins, thereby preventing them from interacting with the membrane-bound receptor Frizzled and/or coreceptor LRP5/6 (28). A null mutation of the WNT antagonist SFRP1 in mouse also results in high bone mass between the ages of 7 and 9 months (29), suggesting very selective effects exhibited by either SFRP1 or specific WNT proteins in maintaining the bone mass of the adult skeleton. RUNX2, a member of the runt homology domain transcription factor family, is essential for osteoblast differentiation (5, 6, 30, 31). Mutations
During development of the skeleton and formation of bone tissue, several morphogenic growth factor and hormone signaling pathways impinge upon transcriptional regulators to induce the osteogenic phenotype (1, 2). The challenge is to identify how developmental cues and regulatory factors are integrated to accommodate the requirements for biological control of cell differentiation and tissue formation. Here, we have addressed the interaction of two key signals for osteogenesis: the WNT pathway, which contributes to the development of skeletal structures (3, 4), and the transcription factor RUNX2 (CBFA1/AML3), which is required for embryonic bone formation (5, 6). WNT signaling comprises a family of 19 secreted glycoproteins that
* This
work was supported by National Institutes of Health Grants AR39588, P01 AR48818, and P30 DK32520. 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 To whom correspondence should be addressed: Dept. of Cell Biology and the Cancer Center, University of Massachusetts Medical School, 55 Lake Ave. North, Worcester, MA 01655-0106. Tel.: 508-856-5625; Fax: 508-856-6800; E-mail: jane.lian@ umassmed.edu.
The abbreviations used are: TCF, T-cell factor; LEF, lymphoid enhancer factor; BMP, bone morphogenetic protein; MEF, mouse embryonic fibroblast; WT, wild-type; RT, reverse transcription; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.
EXPERIMENTAL PROCEDURES
MiceTransgenic mice containing the lacZ gene under the control of the 3-kb Runx2 P1 promoter (42) and SFRP1 knockout (sfrp1/) and wild-type (sfrp1/) mice (29) were used for in vivo analysis. Animals were maintained at the University of Massachusetts following procedures approved by the Institutional Animal Care and Use Committees. Genotyping was carried out as described previously for both mouse models (29, 42). Cell CultureMouse embryonic fibroblasts (MEFs) were isolated from postcoitus day 12.5 mice as described previously (43). Briefly, each embryo was sheared in an 18-gauge syringe in the presence of 1 ml 0.25% trypsin and 1 mM EDTA (Invitrogen). After a 15-min incubation at 37 C, Dulbeccos modified Eagles medium with 15% fetal bovine serum was added to inactivate trypsin. The cells were plated and incubated for 24 h at 37 C. The adherent cells were used as MEF cells. The experiments were initiated after replating the adherent cells in 6-well plates, followed by treatment for 48 h with recombinant human BMP2 (kindly provided by Dr. John Wozney, Wyeth Research, Cambridge, MA) in WNT3-conditioned medium (collected from WNT3a-overexpressing L-cells obtained from American Type Culture Collection). Cells were harvested either for histochemical detection of -galactosidase activity after fixation in 0.5% glutaraldehyde (29) or for biochemical assay by addition of 300 l of 1 reporter lysis buffer (Promega Corp.). A 50-l aliquot of the lysates was incubated with 100 l of 1 reporter lysis buffer and 150 l of substrate solution (80 mM phosphate-buffered saline (pH 7.3), 102 mM 2-mercaptoethanol, 9 mM MgCl2, and 8 mM o-nitrophenyl -galactopyranoside) for 30 min at 37 C. The reaction
33133
RNase-free DNase treatment. The RT reaction was performed on 1 g of total RNA using a first strand synthesis kit (Invitrogen). Relative transcript levels were measured by real-time PCR in a 50-l reaction volume on 96-well plates using an ABI PRISM 7000 sequence detection system (Applied Biosystems) following the recommended protocol for SYBR Green (Bio-Rad). Transcript levels were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels using primers from Applied Biosystems and SYBR Green Master Mix (Bio-Rad). The primers used for amplification are listed in TABLE ONE. Electrophoretic Mobility Shift AssayNuclear extracts were prepared by hypotonic cell lysis and high salt nuclei extraction as described previously (46). Oligonucleotides from the rat Runx2 promoter containing the binding site for LEF/TCF (WT and mutated at the TCF site, referred to as mTCF) were synthesized. The upper strand oligonucleotide (200 ng) was labeled with 32P for 1 h at 37 C using T4 polynucleotide kinase (New England Biolabs Inc.) following the manufacturers instructions. Annealing was performed in the presence of a 3-fold excess amount of the lower strand oligonucleotide. Labeled oligonucleotides were purified using a quick-spin Sephadex G-25 column (Roche Applied Science) following the manufacturers protocol. Reaction mixtures for gel shift assay were prepared using 10 fmol of probe, 50 mM KCl, 12 mM HEPES, 1 mM EDTA, 1 mM dithiothreitol, 12% glycerol, 12.5 g/ml salmon sperm DNA, and 5 g of nuclear extracts. After a 20-min incubation at room temperature, 100-fold unlabeled double-stranded WT competitor oligonucleotide or nonspecific AP1 oligonucleotide was added to the reaction mixtures. For immunoshift assays, nuclear extract was incubated with either anti-TCF1 antibody (2 g) or nonspecific control antibody (Sp1) at 37 C for 1 h before addition of the probe. The samples were electrophoresed on a 4% nondenaturing polyacrylamide gel, dried, and autoradiographed on Biomax MR film (Eastman Kodak Co.). Chromatin ImmunoprecipitationChromatin immunoprecipitation studies were performed as described previously (47) with the following modifications. Formaldehyde cross-linking was quenched by addition of glycine to a final concentration of 125 mM at room temperature for 10 min, followed by washing once with ice-cold 1 phosphate-buffered saline before resuspension in lysis buffer. Sonication was performed eight times at setting 3 for 15 s. The precleared cell lysate was incubated overnight at 4 C with 5 g of anti--catenin antibody (Upstate Biotechnology, Inc.), 4 g of anti-TCF1 and anti-LEF1 antibodies, or 3 g of anti-RUNX2 antibody M-70 (Santa Cruz Biotechnology, Inc.). The cross-linking reaction was reversed by overnight incubation of the solutions at 65 C, and the DNA was recovered by phenol/chloroform extractions. DNA was ethanol-precipitated and dissolved in 40 l of
Tris/EDTA. Four l of DNA was used for quantitative RT-PCR to detect the presence of specific DNA segments with the following primer pairs: Set 1, CAGTGGTAGGCAGTCCCACTTTAC (forward) and GGCTGGTAGTGACCTGCAGAGAT (reverse); and Set 2, GAGCAAGGGGGAAGCCACAGTG (forward) and GTGAGGCGAATGAAGCATTCACAC (reverse).
RESULTS
Runx2 Expression Is Induced by Enhanced WNT Signaling in VivoA biological linkage of Runx2 and WNT signaling was first established by examining the bone tissue of WT and sfrp1/ mice. Inactivation of SFRP1 results in a highly significant increase in the mineral apposition rate and decreased apoptosis in 79-month-old sfrp1/ mice, leading to higher bone density (29). Total RNA from two independent litters of 7-month-old WT and sfrp1/ mice was isolated from the cortical bone of the diaphysis. We observed that Runx2 mRNA levels at this age were 6 9-fold higher in sfrp1/ versus WT mice (Fig. 1A, upper panel). This induction in transcript levels was accompanied by a 23-fold increase in RUNX2 protein levels in sfrp1/ versus WT mice (Fig. 1A, lower panel). This observation suggests that RUNX2-mediated osteoblast differentiation is a contributing factor to the high bone mass phenotype at 7 months. To ascertain the significance of elevated Runx2 expression in reflecting enhanced WNT signaling and osteoblast activity, we examined bone samples from growing 1-month-old mice. We found that Runx2 mRNA levels were elevated from 10- to 20-fold in sfrp1/ mice (Fig. 1B). The 27-fold increase in TCF1 (a known target of canonical WNT signaling) in the sfrp1/ bones suggests that the WNT pathway functions at a higher level than in the wild-type bones. The induction of Runx2 levels was paralleled by 1.33-fold higher levels of osteocalcin (a RUNX2 target gene) and a 2 8-fold suppression of alkaline phosphatase in sfrp1/ mice (Fig. 1B). These modifications in gene expression (high osteocalcin and low alkaline phosphatase levels) reflect the stage of a mature osteoblast/osteocyte in a mineralized matrix (37). Furthermore, histone H4 gene expression, which is coupled to DNA synthesis (48), was significantly lower in sfrp1/ cells (Fig. 1B), indicating a more rapid exit from the cell cycle, consistent with increased osteoblast differentiation. We observed no difference in expression of the apoptotic marker bcl2 (Fig. 1B). Considering the small sample size (three mice/ group), the significance of the differences in gene expression between WT and sfrp1/ mice reached p 0.05. Normalization of these markers to GAPDH expression (or 18 S RNA) (data not shown) provided evidence that the increase in Runx2 occurred on a per cell basis, there-
FIGURE. 1. In vivo regulation of osteogenic differentiation by the WNT pathway. A, analysis of RUNX2 expression in the long bones (marrow-free cortical bones) of 7-month-old WT and sfrp1/ mice by quantitative RT-PCR (three mice/group) (upper panel) or by Western blot analysis (two mice/group) (lower panel). The transcript levels were normalized to GAPDH values. See Experimental Procedures for RNA and protein lysate preparations and antibody dilution. B, quantitative RT-PCR analysis of Runx2, TCF1, osteocalcin (OC), alkaline phosphatase (AP), histone H4, and bcl2 mRNA levels in the long bones (marrow-free cortical bones) of 1-month-old WT (E) and sfrp1/ () mice (three mice/group). The transcript levels of each gene were normalized to GAPDH values.
fore suggesting an increased representation of more mature osteoblasts in sfrp1/ bone. Therefore, the changes observed in osteoblasts that reflect stimulated osteogenesis appear to be the result of increased Runx2 expression. The increase in Runx2 expression and osteoblast maturation in mouse sfrp1/ bone is consistent with our earlier report of accelerated osteoblast differentiation of bone marrow stromal cells from sfrp1/ mice (29). Thus, our findings show that, in the absence of SFRP1-mediated antagonism of WNT signaling, Runx2 expression is elevated and promotes osteoblast differentiation. We conclude that, in vivo, SFRP1 regulates bone formation by inhibiting WNT-mediated increases in Runx2 gene expression. Canonical WNT Signaling Increases Runx2 Promoter ActivityDirect regulation of Runx2 transcription by WNT signaling was initially addressed in MEFs isolated from a transgenic mouse harboring the lacZ gene under the control of the 3-kb Runx2 P1 promoter. The transgene is expressed in mesenchyme early in development prior to formation of mineralized tissue (42). Because the WNT/-catenin pathway is linked to bone formation (1520, 49, 50), we focused on regulation of the Runx2 gene by canonical WNT signaling. Runx2 promoter responsiveness to WNT proteins was evaluated by monitoring -galactosidase activity in MEF cells treated with WNT3a-conditioned medium. Low basal Runx2 promoter activity was detected by lacZ staining in a subpopulation of the cells at confluence. Quantitation of Runx2 promoter activity by spectrophotometric determination of -galactosidase activity showed a 3 4-fold increase in response to canonical WNT signaling (Fig. 2A). In MEF cells, alkaline phosphatase and osteocalcin (markers of the osteoblast phenotype) were not induced after 48 h of treatment (data not shown). We used the pre-osteoblastic MC3T3 cell model to examine Runx2 promoter activity in response to transfected WNT1 in MC3T3 cells. We observed a 1.8-fold induction of 3-kb mouse pro-
moter activity by WNT1. Although the basal activity of the 0.6-kb Runx2 promoter was 2-fold lower compared with that of the 3-kb promoter, WNT1 treatment resulted in the same -fold induction (1.7fold) (Fig. 2B). These findings indicate that WNT signaling can up-regulate Runx2 transcription in both committed osteoprogenitor cells and embryonic mesenchymal cells prior to the induction of osteoblast phenotypic genes. On the basis of the similar responses of the 3- and 0.6-kb promoter fragments to WNT1 and several studies demonstrating that key regulatory elements are confined to the 0.6-kb fragment (44, 51), we focused on the 0.6-kb promoter for further characterization of WNT responsiveness. As LEF/TCF family proteins interact with -catenin to regulate expression of target genes for the canonical WNT signaling pathway (52), we examined the effect of coexpressing TCF1 with a panel of WNT proteins on the Runx2 promoter in MC3T3 cells (Fig. 2C). In the absence of TCF1, WNT1 stimulated Runx2 promoter activity by 1.7fold. However, the presence of TCF1 and either WNT1, WNT3, WNT3a, or WNT6 increased Runx2 promoter activity from 3- to 4-fold over basal activity (i.e. no treatment control). WNT7a and WNT7b with TCF1 resulted in a modest but statistically significant 1.5-fold (p 0.005) stimulation of Runx2 transcription. Thus, TCF1 potentiates activation of the Runx2 promoter by canonical WNT proteins in osteoprogenitor cells. We next addressed whether the observed WNT/TCF1 activation of Runx2 is regulated by the secreted antagonist for WNT proteins, SFRP1. We found that SFRP1 modestly decreased (20%, p 0.05) the basal activity of the Runx2 promoter in the absence of exogenous WNT proteins. This result indicated that SFRP1 can antagonize endogenous WNT signaling in MC3T3 cells (53), which regulates the Runx2 promoter (Fig. 2D). Activation of the Runx2 promoter by WNT1 and TCF1
33135
FIGURE. 2. Induction of the Runx2 P1 promoter by canonical WNT signaling. A, activation of the lacZ transgene under the control of the 3-kb Runx2 P1 promoter in MEFs in response to WNT3a. MEFs were isolated from postcoitus day 12.5 mouse embryos and cultured as described under Experimental Procedures. Because MEFs are not transfected with significant efficiency, cells at confluence were treated for 48 h with 10% conditioned medium containing WNT3a. A representative area of a well is shown (magnification 2.5) (upper panels). -Galactosidase activity in cell lysates after the indicated treatment was quantified (lower panel). A representative experiment shows means S.D. of 2 wells in a 6-well plate. B, activation of the Runx2 P1 promoter by WNT proteins. MC3T3 cells were transiently transfected with the 3- or 0.6-kb Runx2 promoter reporter (1 g/well) and WNT1 or the empty vector pUSEAmp (50 ng/well). The means S.D. of 6 wells are shown. C, synergistic activation of the Runx2 promoter by canonical WNT proteins and TCF1. MC3T3 cells were transiently transfected with the 0.6-kb Runx2 promoter reporter (1 g/well) along with the expression construct for TCF1 (250 ng/well) and the indicated WNT proteins (50 ng/well). pGL3 is the empty vector control for the Runx2 P1 promoter-luciferase reporter. The means S.D. of 6 wells are shown. The straight line represents basal Runx2 promoter activity. D, SFRP1 suppresses Runx2 promoter activity induced by several WNT proteins. Transient transfection studies were performed in MC3T3 cells with 1 g of the 0.6-kb Runx2 promoter, 50 ng of the indicated WNT proteins, and 300 ng of SFRP1. Error bars indicate the means S.D. (n 6).
was completely blocked by SFRP1. We also found that SFRP1 abolished activation of Runx2 by the other canonical WNT proteins (WNT3, WNT3a, and WNT6, for example, are shown). Taken together, these findings demonstrate that canonical WNT/TCF signaling positively regulates Runx2 promoter activity and that this regulation can be antagonized by SFRP1. The Proximal Runx2 Promoter Contains a Functional TCF DNAbinding Site and a WNT-responsive ElementWithin the 0.6-kb Runx2 promoter, a single putative TCF site, characterized by the core CTTTG, is located at 97 to 93 (Fig. 3A). We examined this element for the formation of a specific TCF1 DNA-binding complex. For electrophoretic mobility shift assays, we used nuclear extracts from MC3T3 cells in which TCF1 protein was expressed at detectable levels as determined by western blot analysis (data not shown). One major complex was formed with the WT Runx2 oligonucleotide, but not with the oligonucleotide containing a mutation in the TCF consensus sequence (mTCF) (Fig. 3B, left panel). This proteinDNA complex specifically competed with the unlabeled WT oligonucleotide, whereas a nonspecific oligonucleotide (AP1) had no effect on binding (Fig. 3B, middle panel). Furthermore, addition of anti-TCF1 antibody to the nuclear extract prevented formation of the complex, whereas a nonspecific antibody showed no effect (Fig. 3B, right panel). We did not observe RUNX2 binding in the electrophoretic mobility shift assay, even though a RUNX consensus site is located in the probe. Thus, these results suggest that TCF1 interacts with the consensus TCF1-responsive element in the proximal Runx2 promoter that is conserved between mouse and rat.
To directly address the role of the TCF-binding site in WNT-mediated regulation of the Runx2 promoter, the TCF element was mutated (Fig. 3A). Transient transfection studies revealed that mutation of the TCF-binding site resulted in loss of activation by WNT and TCF1 (Fig. 3C). These results together confirm that the WNT-mediated induction of the 0.6-kb Runx2 promoter occurs through the TCF-binding site at 97 to 93. We next determined whether the enhanced promoter activity is translated into induced gene expression under the same conditions in MC3T3 cells. We transfected these cells with WNT1 and/or TCF1 and found that endogenous Runx2 mRNA was induced by 7 8-fold by WNT1 and TCF1 together (Fig. 4A). Because these cells are transfected with 30% efficiency, modifications in Runx2 levels by either WNT1 or TCF1 alone were not detected by significant changes in mRNA, as we observed in promoter assays (see Fig. 2C). However, the observation of endogenous Runx2 gene expression increased directly by WNT1 and TCF1 was further validated by the in vivo occupancy of the TCF1 regulatory site in the Runx2 gene promoter (Fig. 4B). Chromatin immunoprecipitation studies were performed in MC3T3 cells expressing basal levels of TCF1 and LEF1 as determined by quantitative RT-PCR (Fig. 4B, left panel). Two sets of primer pairs were designed to encompass the TCF element within the Runx2 promoter (Fig. 4B, right panel). The results show that TCF1 and -catenin are associated with the promoter. The LEF1 transcription factor, although expressed in these cells, was not detected in the Runx2 promoter. Notably, RUNX2 has also been found associated with the promoter region of the Runx2 gene, which
FIGURE. 3. A functional TCF site resides in the Runx2 P1 promoter. A, schematic illustration of the Runx2 promoter showing the relative positions of binding sites for the indicated transcription factors. The LEF/TCF consensus sequence is presented, showing the proximity to a RUNX regulatory site. The sequences of oligonucleotides (oligo) for the WT TCF-binding site and the mTCF site are shown. B, binding of TCF1 to the Runx2 promoter DNA is demonstrated by electrophoretic mobility shift assay using nuclear extracts from MC3T3 cells. Five g of nuclear extracts was incubated with 10 fmol of labeled WT probe or mTCF. A proteinDNA complex formed with the WT oligonucleotide, but not with the mutant oligonucleotide (left panel). 100-Fold unlabeled WT oligonucleotide and nonspecific (n/s) AP1 oligonucleotide were used as competitors, demonstrating the specificity of the complex indicated by the arrow (middle panel). Anti-TCF1 antibody or nonspecific control antibody (Sp1) was incubated with the nuclear extract at 37 C for 1 h before addition of probe (right panel). C, loss of WNT/TCF-mediated activation in the mutant Runx2 P1 promoter. MC3T3 cells were transiently transfected with either the 0.6-kb WT Runx2 promoter reporter or the mutant TCF1 promoter (mTCF) (1 g/well) along with the expression construct for TCF1 (250 ng/well) and WNT1 (50 ng/well). The means S.D. of 6 wells are shown.
FIGURE. 4. In vivo regulation of the Runx2 gene by components of WNT signaling and RUNX2. A, relative levels of endogenous Runx2 transcripts in MC3T3 cells transfected with the WNT1 (50 ng), TCF1 (250 ng), and/or SFRP1 (300 ng) expression construct. Cells were harvested after 24 h for RNA isolation and subsequent analysis by quantitative RT-PCR. The Runx2 levels were normalized to GAPDH transcript levels. B, in vivo occupancy of the proximal Runx2 promoter by RUNX2 and WNT regulatory factors -catenin, TCF, and LEF1. The relative expression levels of TCF1 and LEF1 as detected by quantitative RT-PCR analysis are shown (left panel). Formaldehyde-cross-linked chromatin samples from MC3T3 cells were used for immunoprecipitation reaction with antibodies against TCF1, LEF1, RUNX2, and -catenin. The cross-linking was reversed overnight at 65 C, and DNA was purified and quantified by quantitative RT-PCR using two different sets of primers as indicated. Two primer pairs (as indicated) were used to amplify the immunoprecipitated DNA. Normal rabbit IgG was used as a control. The values are plotted as a fraction of the input precipitated (right panel). C, WNT and RUNX2 effects on Runx2 promoter activity. Osteoprogenitor MC3T3, fibroblastic NIH3T3, and pluripotent mesenchymal C3H10T1/2 cells were transfected at 50 60% confluence with the Runx2 P1 promoter (1 g) and the WNT1 (50 ng), TCF1 (250 ng for MC3T3 cells and 100 ng for other cell lines), and/or RUNX2 (250 ng) expression construct. The means S.D. for promoter activities (n 6 wells) are shown.
33137
DISCUSSION
In this study, using both in vitro and in vivo approaches, we found that WNT signaling results in activation of the Runx2 promoter and induces endogenous Runx2 gene expression in pluripotent mesenchymal and osteoprogenitor cells. Our study has identified a critical TCF1-responsive element in the 0.6-kb promoter that mediates WNT activation, which is abrogated upon mutagenesis of this regulatory element located at 97 to 93. We have shown direct activation of Runx2 mediated by multiple canonical WNT proteins together with TCF1 in transient promoter assays and by in vivo chromatin immunoprecipitation studies. Our chromatin immunoprecipitation studies also revealed selective association of TCF1 (but not LEF1) with the Runx2 promoter. Furthermore, we found that the WNT antagonist SFRP1 blocks canonical WNT activation of the Runx2 promoter as well as endogenous gene expression. The in vivo importance of WNT-induced bone formation mediated in part through RUNX2 is further demonstrated by a 10 20fold activation of TCF1 and Runx2 in the SFRP1-null mouse. Our finding of WNT activation of endogenous gene expression and Runx2 promoter activity in MEFs and osteoprogenitor cells suggests that WNT signaling in mesenchymal progenitor cells may be an important activator of Runx2 gene expression in the embryo prior to BMP2-induced osteogenesis and formation of a mineralized skeleton. Our study provides compelling evidence for a mechanism by which the canonical WNT pathway promotes bone formation through -catenin/TCF1-mediated activation of the master osteogenic transcription factor RUNX2, which drives mesenchymal cells to the osteogenic lineage. This mechanism accounts for in vivo observations in mouse models that the extent of bone formation, as a result of either increased or inhibited -catenin signaling, correlates with modifications in Runx2 expression (24, 25, 54). Higher levels of -catenin enhance bone formation with concomitant increases in expression of osteoblast-specific genes (24, 54), whereas conditional knockdown of the -catenin gene at early developmental stage causes ectopic chondrogenesis and abnormal osteoblast differentiation (24 26). Furthermore, it has been shown that WNT10b shifts a mesenchymal cell toward the osteoblast lineage with concomitant induction of the bone-related transcription factors (including a 5-fold increase in Runx2) and suppression of adipogenic transcription factors (CCAAT/enhancer-binding protein- and peroxisome proliferator-activated receptor-) (54). A similar induction of the Runx2 gene was also observed in WNT14 transgenic mice (24). We have identified a critical regulatory element in the Runx2 gene promoter that
FIGURE. 5. Illustration of WNT/TCF1 activation and SFRP1 attenuation of Runx2 gene expression for control of bone formation. A, summary of our findings of WNT/ TCF1 activation of Runx2 and integration of known BMP2 stimulation of Runx2 for osteogenesis. Activation of the canonical WNT pathway results in multiple signaling events that lead to induction of TCF1 expression as well as translocation of -catenin into the nucleus to form a complex with TCF1 on the Runx2 promoter for its induction. The induced levels of TCF1 in the sfrp1/ mouse are also indicated. In parallel, BMP2 also leads to Runx2 expression and formation of a RUNX2Smad coregulatory complex essential for osteogenesis. Cross-talk between WNT and BMP2 signaling for skeletal development is suggested from the literature, but mechanisms are not clearly defined. B, canonical WNT signaling regulates lineage commitment of undifferentiated mesenchymal cells to osteoblasts through RUNX2. The extent of canonical WNT signaling can determine the lineage progression of mesenchymal progenitor cells expressing RUNX2 toward chondrogenesis or osteogenesis (24, 25). Induced canonical WNT signaling can up-regulate Runx2 expression, leading to osteoblast differentiation. However, lower levels of canonical WNT signaling result in chondrogenesis. Further progression to chondrocyte differentiation occurs by suppression of RUNX2 by Nkx3.2 (69).
selectively interacts with the TCF1 transcription factor in osteogenic cells. Our finding of a striking increase in TCF1 mRNA levels upon enhanced WNT signaling in vivo in the SFRP1-null mouse supports the requirement of TCF1 for induction of Runx2 gene expression and osteogenesis. The mouse sfrp1/ model has uncovered a positive feed-forward regulation of canonical WNT signaling through the TCF1 transcription factor that mediates -catenin activity in target genes. The regulation of LEF/TCF by WNT/-catenin signaling is not unprecedented (5557). Constitutive activation of WNT signaling in colon cancers resulting from stabilization of -catenin that precedes colon cancer results in activation of full-length LEF1 (55, 56). Inappropriate activation resulting from mutations in adenomatous polyposis coli or -catenin activates the TCF4 transcription factor, and one of its target genes is TCF1 in epithelial cells (57). Our finding that canonical WNT proteins increase both the TCF1 and Runx2 genes in osteoblasts supports the hypothesis that RUNX2 is the molecular switch by which WNT promotes osteogenesis instead of chondrogenesis (Fig. 5). The expression of various secreted WNT antagonists (Wif1 and SFRPs) has been proposed for regulation of osteoblast differentiation (53). Our observation of elevated Runx2 and osteocalcin expression in bone of the SFRP1 knockout mouse provides evidence for SFRP1 in regulating WNT/TCF1/RUNX2-mediated osteoblast maturation. The effect of increased RUNX2 is to promote the osteogenic differentiation
18.
19.
20.
27.
33.
34.
39.
33139