Lymphoid Enhancer Factor-1 and β-Catenin Inhibit Runx2-dependent Transcriptional Activation of the Osteocalcin Promoter*

Functional control of the transcription factor Runx2 is crucial for normal bone formation. Runx2 is detectable throughout osteoblast development and maturation and temporally regulates several bone-specific genes. In this study, we identified a novel post-translational mechanism regulating Runx2-dependent activation of the osteocalcin promoter. A functional binding site for the high mobility group protein lymphoid enhancer-binding factor 1 (LEF1) was found adjacent to the proximal Runx2-binding site in the osteocalcin promoter. In transcription assays, LEF1 repressed Runx2-induced activation of the mouse osteocalcin 2 promoter in several osteoblast lineage cell lines. Mutations in the LEF1-binding site increased the basal activity of the osteocalcin promoter; however, the LEF1 recognition site in the osteocalcin promoter was surprisingly not required for LEF1 repression. A novel interaction between the DNA-binding domains of Runx2 and LEF1 was identified and found crucial for LEF1-mediated repression of Runx2. LEF1 is a nuclear effector of the Wnt/LRP5/β-catenin signaling pathway, which is also essential for osteoblast proliferation and normal skeletal development. A constitutively active β-catenin enhanced LEF1-dependent repression of Runx2. These data identify a novel mechanism of regulating Runx2 activity in osteoblasts and link Runx2 transcriptional activity to β-catenin signaling.

These genetic studies demonstrate that cellular control of Runx2 expression levels and function is crucial for skeletal development.
At the molecular level, Runx2 activity is regulated by multiple transcriptional and post-translational mechanisms (9). Runx proteins activate or repress gene expression by binding the DNA sequence, TGPuGGTPu (10). Runx-binding sites are often necessary but are not sufficient for transcriptional regulation of tissue-specific genes; thus, it has been hypothesized that Runx proteins are organizers that facilitate the assembly of transcriptional regulatory complexes on gene regulatory elements (11,12). Runx factors, in fact, interact with numerous transcription factors (e.g. AP1, C/EBP, Ets1, and SMADs) and transcriptional co-factors (e.g. p300, ALY, mSin3A, TLE, and HDAC6) to regulate tissue-specific gene expression (13)(14)(15)(16). Runx1-dependent trans-activation of the T cell receptor enhancer is also increased indirectly by DNA-binding proteins, such as LEF1, 1 which bends DNA to facilitate long distance cooperative interactions between Runx1 and Ets1 (17,18).
LEF1 is a high mobility group (HMG) protein and nuclear effector of the canonical Wnt signaling pathway (19,20). LEF1 binds the consensus DNA sequence C/TCTTTGAA in the minor groove, creates a 130°bend in the double helix, and alters of binding of other transcription factors to neighboring sites (17,21). LEF1 transcriptional activity is regulated by interactions with transcriptional co-activators and co-repressors. In the absence of Wnt signals, LEF1 binds to transcriptional co-repressors TLE, CtBP, and HDACs to inhibit gene expression (22). Wnts convert LEF1 into a transcriptional activator by stimulating the disassembly of the GSK-3-Axin-APC multi-protein complex to prevent ubiquitin-mediated degradation of ␤-catenin (22, 23). Increased ␤-catenin expression facilitates its translocation to the nucleus where it displaces co-repressors from LEF1, binds LEF1, and recruits co-activators to induce gene transcription (22,23). LEF1, c-myc, and cyclin D1 are among many genes whose promoters are activated by the Wnt/ ␤-catenin/LEF1 pathway (24 -26). It was recently demonstrated that this canonical pathway also directly represses gene transcription in a manner dependent upon Wnt concentrations. In Drosophila, Wnt/wingless (Wg) induces Armadillo (␤-catenin homologue) and dTCF1/Pangolin (LEF1 homologues) to repress transcription of shavenbaby, stripe, decapen-taplegic, and daughterless (27)(28)(29)(30). The mechanism of Wntinduced transcriptional repression through LEF1-binding sites has not yet been defined, but these studies demonstrate that Wnts stimulate multiple, complex nuclear events.
The canonical Wnt signaling pathway regulates bone density by regulating postnatal osteoblast proliferation (31). Inactivating mutations in Wnt receptor, LRP5, decrease bone mass accrual during growth and cause the autosomal-recessive disorder osteoporosis-pseudoglioma syndrome in humans and mice by decreasing osteoblast proliferation and without altering Runx2 mRNA levels (32,33). Conversely, activating mutations in LRP5 are linked to autosomal-dominant high bone mass traits and increased osteoblast proliferation (34,35). The downstream effects of Wnt signaling on osteoblast-specific gene transcription have not yet been described. In this report, we demonstrate that Runx2-dependent activation of the most osteoblast-specific gene, osteocalcin, is repressed by LEF1. A constitutively activated ␤-catenin protein enhances LEF1-mediated repression of Runx2. These data suggest that in addition to stimulating osteoblast proliferation, the Wnt signaling pathway may slow differentiation by inhibiting Runx2 activity.
Immunofluorescence-MC3T3 cells grown on coverslips were washed twice with PBS and fixed at room temperature for 30 min in 4% paraformaldehyde. Fixed cells were lysed for 5 min at room temperature in 0.3% Triton X-100 in PBS and blocked for 30 min at room temperature in immunofluorescence solution (3% bovine serum albumin, 20 mM MgCl 2 , 0.3% Tween 20, PBS). The cells were then incubated with either anti-LEF-1 (REMB1, Exalpha Biologicals) or anti-␤-catenin (BD Biosciences) diluted to 1 g/ml in immunofluorescence solution for 1 h at 37°C in a humidified chamber. Excess primary antibody was removed by washing three times for 5 min in PBS with 0.1% Triton X-100. The cells were incubated for 30 min with goat anti-mouse IgG conjugated to Alexa 546 (1:50 dilution in immunofluorescence solution) at 37°C in a humidified chamber. Excess secondary antibody was removed by washing three times for 5 min in PBS with 0.1% Triton X-100. The nuclei were stained with 5 g/ml Hoescht 3258 for 5 min. The coverslips were mounted on glass microscope slides and viewed with fluorescence microscopy.
Transcription Assays-The cells were harvested 2 days post-transfection, and luciferase expression was measured using a luciferase assay kit or a dual luciferase assay kit (Promega). Firefly luciferase activities were normalized to either human placental secreted alkaline phosphatase (SEAP) or Renilla luciferase. SEAP values were determined as previously described (38).
In Vitro Protein Interaction Studies-LEF1 proteins were in vitro transcribed and translated from pPCR-Script-LEF1 in the presence of 35 S-labeled amino acids using rabbit reticulocyte lysates (Promega). GST and GST-Runx fusion proteins were produced in Escherichia coli and extracted in lysis buffer A (10 mM MES, pH 6.5, 150 mM NaCl, 2 mM MgCl 2 , 0.5 mM EDTA, 0.5% Triton X-100, 5 mM dithiothreitol, and protease inhibitors (aprotinin, leupeptin, phenylmethylsulfonyl fluoride), and 50 g/ml ethidium bromide as indicated) with sonication. GST proteins were purified with glutathione-Sepharose beads and incubated with radiolabeled LEF1 proteins in the presence or absence of 50 g/ml ethidium bromide. The beads were then washed three times with buffer A, resuspended in Laemmli buffer, boiled, and pelleted by centrifugation. The proteins in the supernatant were resolved by SDS-8% PAGE. The gel was fixed with 40% methanol and 10% acetic acid, incubated with Amplify (Amersham Biosciences), dried, and exposed to film.

LEF1 Binds the Osteocalcin Promoter-
In an effort to discover novel mechanisms that regulate Runx2-dependent transactivity, we identified a sequence adjacent to the crucial proximal Runx2-binding site on the murine osteocalcin promoter, mOG2 (37), that resembles the consensus LEF1 recognition site (Fig. 1A). The LEF1 recognition sequence was separated from the proximal Runx-binding site by just four base pairs. The LEF1 binding sequence and its relative location to the Runx site are conserved in mouse, rat, and human osteocalcin promoters. To determine whether LEF1 binds the mOG2 promoter, electrophoretic mobility shift assays (EMSAs) were performed with radiolabeled DNA probes containing mOG2 residues, Ϫ144 to Ϫ110 (Fig. 1A). COS lysates expressing LEF1-HA produced two complexes that hindered the migration of the probe (Fig. 1B). These bands were specifically competed with either unlabeled double-stranded oligonucleotides containing a wild type LEF (L) consensus binding site or HAspecific antibodies. Oligonucleotides containing a mutant LEFbinding site (mtL) or a Runx2 antibody did not alter the LEF complexes (Fig. 1B). Runx2 lysates produced a single band when incubated with the mOG2 probe. This band was specifically competed by the addition of 100-fold excess unlabeled wild type Runx (R) consensus binding site, and its migration was hindered by Runx2-specific antisera. The Runx2 complex was not altered by anti-HA antibodies or by unlabeled oligonucleotides containing a scrambled Runx-binding site or the wild type LEF1-binding site. These data demonstrate that LEF1 specifically interacts with the osteocalcin promoter.
LEF1 Inhibits Runx2-dependent Activation of the mOG2 Promoter-Because the LEF1-and Runx2-binding sites are separated by only four base pairs, we examined the effects of LEF1 on the transcriptional activity of Runx2. MC3T3 (murine preosteoblasts), C3H10T1/2 (murine mesenchymal precursors), or UMR-106 (rat osteosarcoma) cells were transiently transfected with the mOG2-luciferase reporter in the presence or absence of Runx2 and/or LEF1 expression plasmids. Consistent with data previously reported by others (37,40), Runx2 proteins with amino-terminal sequences of either MRIPV or MASNSL activated the mOG2 promoter ϳ4-fold in MC3T3 cells (Fig. 2). LEF1 did not significantly alter the basal activity of the wild type mOG2 reporter. However, LEF1 blocked the activation induced by either Runx2 isoform ( Fig. 2A). LEF1 similarly repressed Runx2-dependent activation of mOG2 in osteoblast precursor C3H10T1/2 cells (Fig. 2B) and in UMR-106 osteosarcoma cells (Fig. 2C). The inhibition was dependent on the amount of LEF1 expression plasmids added to UMR-106 and MC3T3 cells (Fig. 2, C and D). Complete inhibition was observed when equal amounts of LEF1 and Runx2 expression plasmids were co-transfected into these cells. Importantly, coexpression of LEF1 did not suppress Runx2 protein levels (Fig.  2E). These data indicate that LEF1 represses Runx2 activity in osteoblasts, thereby defining a novel intermolecular mechanism that may regulate Runx2 activity during differentiation and a functional role for LEF1 in osteoblasts.
The LEF1-binding Site in mOG2 Is a Repressive Element but Is Not Required for Repression of Runx2 Activity-To determine the requirement of the LEF1-binding site in osteocalcin regulation, we made point mutations in the mOG2 EMSA probes and luciferase reporter constructs that disrupt the binding sites for LEF1, Runx2, or both factors (Fig. 3A). The mutations completely eliminated DNA binding of the respective transcription factor(s) without affecting the DNA binding ability of the other (Fig. 3B). Mutations in the Runx2-binding site decreased basal activity of mOG2 (Fig. 3C). This result is consistent with reports from other laboratories (37,40). Interestingly, mutations in the LEF1-binding site increased the basal activity of the mOG2-luciferase reporter in MC3T3 cells by 2-3-fold (Fig. 3C). Immunoblot and in situ immunofluorescence analyses indicated that MC3T3 cells express LEF1 in their nuclei (Fig. 3, D and E). These data indicate that endog- enous LEF1 suppresses the basal activity of mOG2 in osteoblasts. This is consistent with the role of LEF1 as a transcriptional repressor in the absence of Wnt signals (22).
We next utilized the mOG2 mutant reporter constructs to begin to determine the mechanism by which LEF1 represses Runx2. MC3T3 cells were transiently transfected with wild type or mutant mOG2 reporters in the presence or absence of Runx2 and LEF1 expression plasmids. Because basal activities of the mutant reporter constructs differed (Fig. 3), the data were normalized to the activity of the indicated mOG2 reporter in the absence of Runx2 and LEF1 expression plasmids (Fig. 4). As previously demonstrated, Runx2 activates the wild type mOG2 promoter ϳ3-fold (Fig. 4A, top panel). The Runx2-binding site is necessary for Runx2 activation (Fig. 4A, second and  fourth panels from top). However, mutations in the LEF1binding site did not significantly affect Runx2-dependent activation (Fig. 4A, third panel from top). Surprisingly, the LEF1binding site was also not required for LEF1-mediated repression of Runx2 (Fig. 4A, third panel from top). To confirm that LEF1-binding sites are not necessary for repression, we tested the effects of LEF1 on Runx2-driven expression of an artificial promoter, p6OSE2 (37), which contains six Runxbinding elements upstream of the minimal osteocalcin promoter (Ϫ34/ϩ13) and luciferase, and has no recognizable LEF1-binding sites. Runx2 activated the p6OSE2 reporter by more than 10-fold (Fig. 4B). LEF1 did not affect the basal activity of the reporter but repressed Runx2-induced activation ϳ4-fold. In contrast, LEF1 did not inhibit GAL or GAL-VP16 activity on a GAL-responsive promoter, pAH205 (Fig. 4C). Thus, LEF1 specifically inhibits Runx2, and the LEF1 DNA binding is not required for repression of Runx2-dependent activation of the mOG2 promoter.
LEF1 Interacts with the RHD-To determine how LEF1 is suppressing Runx2, we considered five potential mechanisms of transcriptional repression: 1) competition for DNA binding sites, 2) squelching of the basal transcriptional machinery, 3) quenching of the activation properties of DNA-bound Runx2, 4) indirect transcriptional repression by complexing with Runx2 and preventing DNA binding, and 5) co-repressor recruitment (41)(42)(43). Competition is unlikely because a DNA probe containing just the Runx-binding element does not compete with a mOG2 probe for LEF1 (Fig. 1B), and LEF1 does not bind to a Runx2-specific probe (data not shown). Squelching also does not appear to be the mechanism because ectopic LEF1 did not repress the basal activity of the wild type mOG2 promoter (Fig.  2). To further explore the possibility that LEF1 quenches or indirectly represses Runx2, we asked whether LEF1 physically interacts with Runx2. Using a panel of Runx2 and Runx1 GST

FIG. 3. The LEF1 site is not required for LEF1-mediated transcriptional repression.
A, schematic of mutations introduced into the mOG2 reporter plasmids and EMSA probes by site-directed mutagenesis. The indicated changes were made in either the Runx recognition sequence or LEF1 recognition sequence by site-directed mutagenesis. Dashes represent nonmutated nucleotides. Electrophoretic mobility shift assay probes were synthesized to correspond to the mutations made in the reporter plasmids. B, mutations in the Runx2-and LEF1binding sites specifically eliminate binding of the respective transcription factor. The indicated wild type or mutant radiolabeled probes were incubated with COS lysates expressing Runx2 or LEF1-HA. Runx2 complexes are marked with arrowheads; LEF1-HA complexes are marked with arrows. C, deletion of the LEF1-binding site increases the basal activity of the mOG2 promoter. MC3T3 cells were transfected as described for Fig. 2A using the indicated mOG2-luc mutants. The normalized luciferase values are represented as fold activity over wild type mOG2-luc. D, LEF1 is detectable as a doublet in MC3T3 cells. MC3T3 cell lysates were resolved by SDS-PAGE and immunoblotted with anti-LEF1 or anti-␤-catenin antibodies. E, LEF1 resides in the nucleus of MC3T3 cells. MC3T3 were grown on coverslips and fixed in paraformaldehyde. The cells were lysed with 0.3% Triton X-100 in PBS. LEF1 and ␤-catenin were detected with their respective monoclonal primary antibody followed by a goat anti-mouse secondary antibody conjugated to Alexa-546. The proteins were visualized with fluorescence microscopy. WT, wild type. fusion proteins (Fig. 5A), we found that LEF1 interacts with the amino termini of both Runt domain proteins. The Runt homology domain (RHD; Runx1 residues 50 -179), which mediates DNA binding, is necessary and sufficient to interact with full-length LEF1 in vitro (Fig. 5B). A single point mutation, L148D, in the Runx1 RHD that eliminates DNA binding (44) also prevented LEF1 interactions. We then mapped the Runx-binding site in LEF1 with a panel of LEF1 truncation mutants (Fig. 5A). Neither the ␤-catenin-binding domain nor the extreme carboxyl terminus was necessary for LEF1 to interact with the RHD (residues 50 -179) (Fig. 5C). The HMG domain, however, was required and sufficient to interact with the RHD. This interaction persisted in the presence of ethidium bromide, indicating that intermolecular association between respective DNA-binding domains of LEF1 and Runx2 is not mediated by DNA present in the reactions (Fig.  5C, right panel). These data demonstrate a novel interaction between the nuclear LEF1 and Runx2 and suggest that LEF1 indirectly represses Runx2 activity by binding its DNAbinding domain.
The HMG and ␤-Catenin-interacting Domains of LEF1 Contribute to Repression of Runx2-To determine the regions of LEF1 required for repressing Runx2 activity and to test the fifth potential mechanism of repression (i.e. co-repressor recruitment), we assayed the LEF1 and Runx2 truncation mutants in the mOG2 transcription assay. The LEF1 and Runx2 proteins used in these experiments are diagrammed in Fig. 6A. All LEF1 and Runx2 proteins localized to cell nuclei (data not shown). The LEF1-HMG domain, which interacts with Runx2, was necessary for repression of Runx2 transcriptional activity (Fig. 6B). The HMG and a LEF1 protein lacking the extreme carboxyl terminus (⌬C) also partially blocked Runx2-dependent activation but was not as effective as full-length LEF1 (Fig.  6C). These results suggest that additional regions of LEF1 are required for maximal repression. This hypothesis was supported by experiments with a LEF1 protein lacking the ␤-catenin binding domain (⌬␤-cat), which did not repress Runx2-dependent transcription (Fig. 6B). Thus, the HMG is required for LEF1 to repress transcription, but the ␤-catenin-binding domain also contributes to repression.
Both Runx2 and LEF1 bind the TLE co-repressors (45). To determine whether LEF1 interacts with TLE proteins that are recruited to the promoter by the Runx2 carboxyl terminus to repress transcription, we determined the effects of LEF1 proteins on a truncated Runx2 protein (1-498) that lacks the TLE-binding domain. Runx2 (1-498) retains the DNA binding and activation domains and thus activated mOG2 by 5-fold (Fig. 6D). LEF1 and LEF1-HMG had similar effects on Runx2 (1-498) as they did on wild type Runx2. Thus, LEF1 completely blocked Runx2 (1-498) activity, and the LEF1-HMG, which lacks the TLE-binding domain, partially repressed it (Fig. 6D). These data indicate that TLE recruitment is not required for LEF1-mediated repression of Runx2; however, the possibility that TLE may contribute to LEF1-mediated repression of Runx2 cannot be excluded.
Gain-of-function ␤-Catenin Enhances LEF1-mediated Repression of Runx2-Because LEF1 mutants lacking the ␤-catenin-binding domain were less effective at inhibiting Runx2 than wild type LEF1, we hypothesized that ␤-catenin may participate in LEF1-mediated repression of Runx2. To test this hypothesis, we co-transfected an expression plasmid for a gain- of-function (GOF) ␤-catenin. This cDNA contains point mutations that are found in cancer cells and that prevent ␤-catenin from being degraded, thereby increasing its levels and facilitating its transport into the nucleus (46,47). Interestingly, GOF ␤-catenin augmented the basal activity of the mOG2 promoter in a concentration-dependent manner (Fig. 7). In contrast, GOF ␤-catenin did not affect Runx2-dependent activation of the mOG2 promoter, but it increased LEF1-mediated repression of Runx2 in a concentration-dependent manner. These data suggest that LEF1 and ␤-catenin cooperatively repress Runx2-dependent activation. In addition, the observation that GOF ␤-catenin activates mOG2 suggests that ␤-catenin activation may stimulate this promoter via multiple mechanisms.

DISCUSSION
Runx2 is a crucial transcription factor for osteoblast development. Runx2 expression and activity are controlled by transcriptional, translational, and post-translational mechanisms (9). Runx2 activity is further modulated by interactions with other proteins, including co-activators, corepressors, and other transcription factors (13)(14)(15)(16). Genetic alterations that decrease Runx2 expression or cause structural changes demonstrate that Runx2 is required for osteoblast development from mesenchymal stem cells, the terminal differentiation of committed osteoblasts to osteocytes, and the expression of late osteoblast genes like osteocalcin (2,3,6,48,49). Runx2 is expressed in proliferating osteoblasts; however, it does not appear to play a crucial role in regulating osteoblast number (2,3,5). This suggests that Runx2 activity may be suppressed or controlled in proliferating osteoblasts. Transgenic models support this hypothesis because increased expression of Runx2 in proliferative osteoblasts causes osteopenia and bone fractures (7,8). The mechanisms regulating Runx2 activity during osteoblast growth phases are not understood.
In this report we describe novel functional and physical interactions between Runx2 and LEF1 that may contribute to the regulation of Runx2 activity in osteoblasts. LEF1 is a nuclear effector of the canonical Wnt signaling pathway, which is commonly thought to convert LEF1 to an activator by interactions with ␤-catenin (22, 23). We identified a functional LEF1 recognition sequence in close proximity to the crucial Runx-binding site in the osteocalcin promoter. The LEF1 site is a repressive element, and ectopic expression of LEF1 proteins inhibited Runx2-dependent activation of the osteocalcin promoter in osteoblast lineage cells. This result is consistent with LEF1 being a repressor of transcription. We initially hypothesized that activation of the canonical Wnt signaling pathway would reverse LEF1-mediated repression. Surprisingly, a constitutively active ␤-catenin mutant that mimics Wnt activation did not prevent LEF1-mediated repression and in fact enhanced suppression. These data suggest that activation of LEF1 by Wnt signaling represses Runx2-dependent activation of the osteocalcin promoter. Recent data from developmental models indicate that the Wnt signaling pathway indeed directly represses transcription of some genes (27-29, 50 -52). Interestingly, LEF1-mediated repression of Runx2 on the osteocalcin promoter is remarkably similar to Wg (Wnt homologue)-and dTCF (LEF1)-mediated repression of Drosophila stripe. Thus, Wg and dTCF repress Cubitus interruptus-mediated activation of the stripe promoter where the dTCF-binding site is only four base pairs from a crucial Cubitus interruptus recognition sequence (29). These data indicate that repression mechanisms used by LEF1/TCF factors may be conserved.
The mechanisms by which Wnt signaling pathways repress transcription are not as clearly defined. Our data suggest that LEF1 indirectly represses transcription by masking the DNAbinding domain of Runx2. We found that LEF1 interacts with Runt domain proteins and that LEF1-binding sites in DNA are not necessary for repression. LEF1 did not repress GAL-VP16dependent activation of the GAL-responsive promoter. Thus, LEF1 specifically inhibits the action of Runx2 (indirect repression) but does not repress transcription by interacting with the basal transcriptional complex (direct repression). This is a potentially powerful means of repression because it predicts that a LEF1-binding site is not necessary to suppress Runx2 transactivity. Alterations in LEF1 expression levels could therefore regulate Runx2 activity in osteoblasts. Interestingly, LEF1 mRNA levels are decreased in osteoblasts lacking the Wnt receptor LRP5 (33). These data suggest that an autofeedback loop may regulate LEF1 expression in osteoblasts. Additional data in support of this autofeedback mechanism are that the canonical Wnt signaling pathway stimulates LEF1 expression (53). High levels of the Wg (Wnt) receptor Fzd2 stabilize Wg in Drosophila which, in turn decrease Fzd2 expression (54,55). Thus, local concentrations of Wnts and/or soluble Wnt antagonists may autoregulate intracellular signals and gene transcription in vivo.
The physical and functional interactions between LEF1 and Runx2 may link two pathways that are crucial for osteoblast maturation. As described above, Runx2 is required for terminal osteoblast differentiation. The role of LEF1 in osteoblasts has not yet been directly studied; however, LEF1 is expressed in these cells and is thus a potential nuclear effector of the Wnt signaling pathway in osteoblasts. Mutations in the Wnt receptor LRP5 have significant effects on bone density. Autosomal recessive and inactivating mutations in LRP5 decrease bone density in mice and humans (32,33,56). In contrast, activating mutations in LRP5 cause autosomal dominant high bone mass traits (34,35,59). The effects of these mutations on Runx2 activity have not yet been determined. Because LRP5 activates the canonical Wnt pathway and stabilizes ␤-catenin (60), which enhances LEF1-mediated repression of Runx2, it is possible that Wnt signals transmitted through LRP5 may regulate Runx2 activity.
activity of a transcription factor crucial for bone development. Previous studies have identified numerous proteins that either enhance Runx2 function at the post-transcriptional level (13)(14)(15)(16) or modulate Runx2 activity by regulating the expression/ transcription of the Runx2 promoter (61)(62)(63)(64)(65). Our data indicate that additional mechanisms that negatively regulate Runx2 at the post-transcriptional level also exist.
Repression of Runx2-dependent trans-activation of the osteocalcin is likely only one consequence of LEF1 expression and/or ␤-catenin activation in osteoblasts. In the absence of ectopic LEF1 or Runx2, GOF ␤-catenin activated the osteocalcin promoter. These data indicate that ␤-catenin may regulate osteocalcin expression via multiple mechanisms. It remains to be determined whether the LEF1-independent ␤-catenin activation mimics LRP5 activation or whether other stimuli regulate ␤-catenin function. In addition to the osteocalcin promoter, we have identified potential LEF1 recognition sites in other Runx2-regulated osteoblast promoters (e.g. bone sialoprotein and collagenase 3) (data not shown). Studies that focus on how Wnt, LRP5, ␤-catenin, and LEF1 regulate Runx2-dependent and -independent transcription of these genes are necessary to gain a full understanding of the effects of these proteins on gene expression and the maturation of osteoblasts. It is also important to consider the possibility that LEF1 may respond to Wnt-independent stimuli (e.g. Notch) (66) in osteoblasts. LEF1 was previously shown to facilitate Runx1-dependent activation of the T cell receptor enhancer in a ␤-cateninindependent manner (67,68). Finally, the activities of Runx factors and LEF1 are regulated by interactions with the nuclear matrix (57,58). Thus, in addition to extracellular stimuli, gene structure and tissue-specific factors are also likely to influence how and when LEF1 and Runx factors interact in vivo.