Interaction of LEF1 with TAZ is necessary for the osteoblastogenic activity of Wnt3a

Canonical Wnt signalling plays an important role in osteoblast differentiation and bone formation. However, the molecular mechanisms by which canonical Wnt signalling exerts its osteoblastogenic effect remain elusive. Here, we investigated the relationship between lymphoid enhancer-binding factor 1 (LEF1) and transcriptional co-activator with PDZ-binding motif (TAZ), both of which are transcriptional regulators that mediate canonical Wnt signalling during osteoblast differentiation. Reporter assay and co-immunoprecipitation experiments revealed functional and physical interaction between LEF1 and TAZ. Overexpression of dominant-negative forms of either LEF1 or TAZ markedly inhibited Wnt3a-dependent osteoblast differentiation. Moreover, we found that LEF1 and TAZ formed a transcriptional complex with runt-related transcription factor 2 (Runx2) and that inhibition of LEF1 or TAZ by their dominant-negative forms dramatically suppressed the osteoblastogenic activity of Ruxn2. Additionally, Wnt3a enhanced osteoblast differentiation induced by bone morphogenetic protein 2 (BMP2), which stimulates osteoblast differentiation by regulating Runx2. Collectively, these findings suggest that interaction between LEF1 and TAZ is crucial for the osteoblastogenic activity of Wnt3a and that LEF1 and TAZ contribute to the cooperative effect of Wnt3a and BMP2 on osteoblast differentiation through association with Runx2.

mesenchymal cells 20 . Thus, the role of TAZ in osteoblast differentiation of mesenchymal cells is complex and remains controversial.
To further understand the molecular mechanism by which canonical Wnt signalling stimulates osteoblast differentiation of mesenchymal cells, we examined the roles of LEF1 and TAZ in osteoblast differentiation. We found that interaction between LEF1 and TAZ is necessary for the osteoblastogenic activity of Wnt3a, a well-known canonical Wnt family member 6,[21][22][23] . Moreover, LEF1 and TAZ associate with Runx2 and enhance its activity. Our findings provide a novel paradigm for the molecular mechanism of canonical Wnt signalling during osteoblast differentiation of mesenchymal cells.

Results
Roles of LEF1 and TAZ in osteoblast differentiation. To examine whether canonical Wnt signalling is involved in the regulation of TAZ expression, we first determined the effect of Wnt3a, a well-known canonical Wnt family member and widely used stimulus for osteoblast differentiation 6,21-23 , on TAZ mRNA and protein expression. Although Wnt3a had little effect on TAZ mRNA expression, TAZ protein levels were increased by Wnt3a in the mesenchymal cell line, C3H10T1/2 ( Supplementary Fig. 1a,b). As previously shown 6,[21][22][23] , Wnt3a upregulated ALP expression and Top-Flash activity ( Supplementary Fig. 1b,c). Additionally, a reporter assay using an 8xGTIIC-luciferase construct showed activation of TAZ activity by Wnt3a ( Supplementary Fig. 1c). These data are consistent with a previous report, which showed that TAZ functions as a mediator of canonical Wnt signalling 16 , but are contrary to another study that claimed that a non-canonical Wnt, Wnt5a, rather than canonical Wnt signalling controls TAZ 20 . Indeed, we observed no influence of Wnt5a on TAZ protein expression, mRNA expression and activity, or ALP activity (Supplemental 1a-c). Unexpectedly, another canonical Wnt, Wnt10b, which plays an important role in bone formation 24,25 , did not affect Top-Flash activity, TAZ expression and activity, or ALP activity (Supplemental Fig. 1a-c).
We next examined the role of LEF1 and TAZ in osteoblast differentiation. After we confirmed transcriptional upregulation of target genes by LEF1 and TAZ (Fig. 1a), we generated LEF1 and TAZ adenoviruses (Fig. 1b) and determined the effects of LEF1 and TAZ overexpression on osteoblast differentiation of C3H10T1/2 and ST2 cells. TAZ overexpression moderately increased alkaline phosphatase (ALP) activity in ST2 cells (Fig. 1c,d), whereas LEF1 had little effect on either cell line (Fig. 1c,d). These data indicate that LEF1 overexpression is not sufficient to induce osteoblast differentiation of mesenchymal cells. TAZ inhibits adipocyte differentiation 19 ; therefore, we examined the effect of TAZ and LEF1 on adipocyte differentiation of the preadipocyte cell line, 3T3-L1, as determined by expression of adipogenic markers, PPARγ and aP2. Both TAZ and LEF1 suppressed adipocyte differentiation of 3T3-L1 cells ( Supplementary Fig. 2). LEF1 and TAZ are necessary for Wnt3a-dependent osteoblast differentiation. We next investigated whether LEF1 and TAZ are involved in the osteoblastogenic activity of Wnt3a. LEF1 and TAZ belong to multi-member families; therefore, we expected that LEF1 or TAZ knockdown would be difficult to achieve because of redundancy among the family members. We therefore generated dominant-negative (DN) forms of LEF1 and TAZ 18,26 . As expected, DN-LEF1 and DN-TAZ dramatically inhibited β-catenin-dependent Top-Flash and TAZ-dependent 8xGTIIC luciferase activity, respectively (Fig. 2a,b). Overexpression of either DN-LEF1 or DN-TAZ markedly suppressed Wnt3a-induced ALP activity in C3H10T1/2 and ST2 cells (Fig. 2c-e), which indicates that both LEF1 and TAZ play essential roles in the osteoblastogenic activity of Wnt3a. LEF1 and TAZ interact physically and functionally. Because we identified that LEF1 and TAZ are necessary for Wnt3a-dependent osteoblast differentiation, we next examined whether LEF1 and TAZ interact. We found that LEF1 and TAZ synergistically stimulated both Top-Flash and 8xGTIIC luciferase activities (Fig. 3a). Moreover, co-immunoprecipitation experiments indicated a physical association between LEF1 and TAZ (Fig. 3b). Association of endogenous LEF1 and TAZ was also demonstrated in primary osteoblasts (Fig. 3c). Taken together, these results indicate that Wnt3a promotes differentiation of mesenchymal cells into osteoblasts through formation of a complex between LEF1 and TAZ.
Role of LEF1 and TAZ in Runx2-dependent osteoblast differentiation. BMP2 and Wnt3a signalling interact during osteoblast differentiation 8,27,28 . We therefore sought to understand the molecular mechanism by which Wnt3a and BMP2 regulate osteoblast differentiation. As expected, BMP2 stimulated expression of ALP, Runx2, osteocalcin, Osterix, Msh homeobox 2 (Msx2) and bone sialoprotein (Bsp) ( Supplementary Fig. 3), whereas Wnt3a only increased the expression of ALP ( Supplementary Fig. 3). Interestingly, Wnt3a increased the osteoblastogenic effect of BMP2 on other osteoblast markers ( Supplementary Fig. 3), although Wnt10b had little effect on osteoblast markers, even in the presence of BMP2 ( Supplementary Fig. 4). These data indicate that Wnt3a and BMP2 cooperatively control osteoblast differentiation. Because Runx2 associates with TAZ 18 , we hypothesised that Runx2 might link Wnt3a signalling to BMP2 signalling during osteoblast differentiation. TAZ markedly stimulated the transcriptional activity of Runx2 (Fig. 4a). Although LEF1 had little effect on Runx2 in the absence of TAZ, LEF1 further enhanced the transcriptional activity of Runx2 in the presence of TAZ (Fig. 4a). These results suggest a cooperative effect of Runx2, LEF1 and TAZ. We then investigated interactions among Runx2, LEF1 and TAZ. Co-immunoprecipitation experiments indicated that Runx2, LEF1 and TAZ form a complex (Fig. 4b), which indicates an important role of TAZ and LEF1 in Runx2 function. Notably, we found that either DN-LEF1 or DN-TAZ dramatically inhibited Runx2-induced osteoblast differentiation (Fig. 4c,d). Collectively, these findings indicate that a complex of Runx2, LEF1, and TAZ might be critical for the cooperative osteoblastogenic effect of Wnt3a and BMP2.

Discussion
Genetic and clinical evidence clearly indicate important roles of canonical Wnt signalling in bone formation and osteoblast differentiation 8,10,29 . The mechanisms by which canonical Wnt mediates intracellular signalling and regulates transcription of target genes are becoming clearer 30,31 . Identification of TAZ, a well-known mediator of Hippo signalling 32 , as a novel mediator of canonical Wnt signalling is also a striking finding 16 . However, the relationship between the β-catenin/LEF1 and TAZ pathways was unclear, while the role of LEF1 in osteoblast differentiation and bone formation was also controversial. In this study, we found that interaction between LEF1 and TAZ is required for the osteoblastogenic activity of Wnt3a. Supporting this finding, physical and functional interaction of LEF1 and TAZ was demonstrated by co-immunoprecipitation and reporter assays. Additionally, our data indicate that both LEF1 and TAZ are crucial for Wnt3a to exert its osteoblastogenic effect.
One recent study proposed that non-canonical Wnt signalling rather than canonical Wnt signalling activated TAZ and stimulated osteoblast differentiation, contrary to previous findings 20 . We confirmed that Wnt3a but not Wnt5a is involved in TAZ expression and function (Supplementary Fig. 1). In addition, we did not observe the effect of Wnt5a on osteoblast differentiation of mesenchymal cells (Supplementary Fig. 1). Our results support a model in which TAZ functions as a mediator of canonical Wnt signalling. Our results are consistent with Wnt5a inhibiting ALP activity induced by Wnt3a 33 . Additionally, Wnt5a inhibits osteoblast differentiation 34 . Considering the inhibitory effect of non-canonical Wnt signalling on β-catenin signalling, which strongly stimulates bone formation and osteoblast differentiation 35 , it is unlikely that non-canonical Wnt signalling activates TAZ. On the other hand, bone volume was clearly decreased in Wnt5a-deficient mice 36 . Bone resorption was proposed to be decreased by suppression of RANK expression in the Wnt5a deficient mice 36 ; however, it is still possible that Wnt5a is implicated in osteoblast function in a different manner from Wnt3a. Indeed, Wnt5a appears to upregulate expression of Lrp5 and Lrp6 37 . Further experiments are necessary to understand the relationship between non-canonical Wnt signalling and TAZ during osteoblast differentiation and bone formation.
It is surprising that Wnt3a had little effect on the expression of osteoblast markers other than ALP. These data appear inconsistent with the strong osteogenic effects of canonical Wnt signalling reported in animal models and in humans 10,29 . One potential explanation is that canonical Wnt signalling exerts an osteogenic effect in cooperation with other cytokines and growth factors in vivo. Indeed, we showed that Wnt3a and BMP2 cooperatively stimulated osteoblast differentiation of mesenchymal cells. Because BMP2 stimulates the expression and activity of Runx2 38,39 , and Runx2 interacts with TAZ 18 , we investigated interactions among LEF1, TAZ and Runx2. Our data indicate that Wnt3a and BMP2 signalling undergo crosstalk via a complex formed by Runx2, LEF1 and TAZ, consequently exerting a stronger osteoblastogenic effect than BMP2 alone. BMP2 regulates osteoblast differentiation though a Wnt autocrine loop 6 . Wnt3a also cooperatively stimulates BMP2 signalling through inhibitor of DNA binding 1 (Id-1) expression 27 and glycogen synthase kinase 3 β (GSK-3β) 28 . It is, therefore, likely that crosstalk between Wnt3a and BMP2 signalling exists at several different points, and consequently Wnt3a and BMP2 exhibit cooperative osteoblastogenic activity.
Heterozygous LEF1-deficient mice showed decreased bone volume, presumably because of haploinsufficiency 14 . Consistent with this finding, LEF1 stimulates osteoblast differentiation through interaction with Runx2, which regulates FGF18 expression 40 . Conversely, LEF1 inhibits osteoblast differentiation through Runx2 15 . The role of LEF1 in osteoblast differentiation and bone formation was thus elusive. On the basis of our results, we believe that LEF1 plays important roles in osteoblast differentiation through interaction with TAZ. Further investigation, especially in vivo experiments using osteoblast-specific conditional knockout mice, would provide more precise details of the mechanism by which LEF1 regulates bone formation.
Mouse genetic studies clearly demonstrate that Wnt10b is critical for bone formation 24,25 . Moreover, has been proposed that PTH exhibits osteogenic action through Wnt10b 41,42 . Wnt10b also stimulates osteoblast differentiation 43 . Therefore, we examined the effect of Wnt10b on osteoblast differentiation of mesenchymal cells. Surprisingly, Wnt10b failed to induce osteoblast differentiation of mesenchymal cells ( Supplementary Figs 1 and 4). Furthermore, Wnt10b did not enhance osteoblastogenic activity of BMP2 ( Supplementary Fig. 4) and had no cooperative effect with BMP4 to stimulate mineralisation of MEF cells 43 . Moreover, Wnt10b had little effect on TAZ and LEF1 ( Supplementary Fig. 1), at least in our hands. Several studies have shown that Wnt10b stimulates mineralisation in osteoblasts 43 . We, therefore, speculate that Wnt10b has different functional roles in osteoblast differentiation, presumably in the late stage of osteoblast maturation or mineralisation. Further investigation will shed light on the osteogenic action of Wnt10b.
In this study, we showed that two different canonical Wnt signalling pathways exhibit crosstalk and regulate osteoblast differentiation. Additionally, our data reveal a molecular mechanism by which Wnt3a and BMP2 cooperatively control osteoblast differentiation. We believe that our findings contribute to the understanding of Wnt3a osteogenic activity.
Reverse transcriptase quantitative polymerase chain reaction (RT-qPCR). Total RNA was isolated using a total RNA isolation kit (Macherey-Nagel, Duren, Germany) and denatured at 70 °C for 10 min. cDNA was synthesised with oligo dT primers and a reverse transcriptase kit (Takara). RT-qPCR amplification was performed using TaqMan PCR probe sets (Supplementary Table) and a Step-One Plus real-time PCR system (Applied Biosystems, Carlsbad, CA, USA). Target gene mRNA expression levels were normalised to β-actin mRNA expression. Data are represented as the mean ± standard deviation (n = 3).
Alkaline phosphatase staining assay. Cells were washed with phosphate-buffered saline, fixed in 3.7% formaldehyde and stained with a mixture of 330 µg/ml nitro blue tetrazolium, 165 µg/ml bromochloroindoyl phosphate, 100 mM NaCl, 5 mM MgCl 2 and 100 mM Tris (pH 9.5). Quantification of alkaline phosphatase staining intensity was measured by ImageJ image software (the National Institutes of Health, Bethesda, MA, USA).
Isolation of primary osteoblasts. Calvariae were isolated from 2-or 3-day-old neonatal ICR mice (SLC, Shizuoka, Japan), and then sequentially digested four times with 0.1% collagenase and 0.2% dispase for 7 min. The cells were then collected by centrifugation and the last three groups of fractionated cells were collected and used as primary osteoblasts.
Reporter assays. Forty-eight hours after luciferase construct transfection, the cells were lysed. Luciferase substrate (Promega, Fitchburg, WI, USA) was added to the lysates, and the luciferase activity of the lysates was measured using a luminometer (Promega).