Wnt7b expressed by hypertrophic chondrocytes is a stimulatory factor for endochondral ossification that is regulated by Smad4 activity Free

In vertebrates, including mice and humans, the skeleton is formed through two processes: intramembranous ossification and endochondral ossification (Berendsen and Olsen, 2015; Ono and Kronenberg, 2016). The former process contributes to the development of flattened bones, such as parietal bone and the scapula. The latter process mainly contributes to longitudinal skeletal growth of the long bones during, not only embryonic development, but also the postnatal growth period. Bone formation through the endochondral ossification process also contributes to fracture healing. During the endochondral ossification process, cartilaginous tissue templates formed by chondrocytes are gradually replaced by bone tissue formed by osteoblasts. Growth plate cartilage at the metaphysis in the long bones plays a central role in endochondral ossification (Kronenberg, 2003; Mackie et al., 2008). In the growth plate cartilage, proliferating chondrocytes gradually enlarge in size and differentiate into hypertrophic chondrocytes, which are terminally differentiated cells, and they line columnar structures. Osteoblasts appear close to hypertrophic chondrocytes at the junction between growth plate cartilage and primary bone, which is called the primary spongiosa, during the endochondral ossification process (Kronenberg, 2003; Mackie et al., 2008). Therefore, hypertrophic chondrocytes are believed to secrete factors to initiate bone formation during endochondral ossification. Indian hedgehog (Ihh) secreted by pre-hypertrophic and hypertrophic chondrocytes has been found to be the factor inducing osteogenesis in endochondral ossification, especially in bone collar formation, during embryonic development (Chung et al., 2001; Long et al., 2004; Ohba, 2020).

Many growth factors and hormones are involved in skeletal development. Transforming growth factor β (TGFβ) family members are important growth factors for the development of cartilage and bone during embryogenesis and postnatal tissue repair (Wu et al., 2016; Wang et al., 2020). Members of the TGFβ family, including TGFβs, BMPs, GDFs and activins, directly regulate the proliferation and differentiation of skeletal cells, such as chondrocytes, osteoblasts and osteoclasts (Salazar et al., 2016a; Wu et al., 2016). Ligands in the TGFβ family can be classified into osteogenic and nonosteogenic subfamilies according to their activity in a heterotopic bone induction assay of skeletal muscle in vivo. Osteogenic ligands, such as BMP2, BMP4 and BMP7, can induce heterotopic ossification in soft tissues through the endochondral ossification process (Katagiri and Watabe, 2016; Katagiri et al., 2018a,b). Intracellular signals downstream of TGFβ family ligands are activated by two types of transmembrane serine/threonine kinase receptors. Type II receptors phosphorylate type I receptors, ALK1 to ALK7, and then type I receptors activate downstream transcription factors, the Smad proteins. The osteogenic ligands in the TGFβ family activate Smad1, Smad5 and Smad9 (also called Smad8), and the non-osteogenic ligands activate Smad2 and Smad3 (Katagiri and Watabe, 2016; Salazar et al., 2016a; Katagiri et al., 2018a,b). Smad4 is a co-activator of these receptor-activated Smad proteins; therefore, it is essential for intracellular signaling downstream of both osteogenic and non-osteogenic ligands of the TGFβ family (Shi and Massagué, 2003; Katagiri and Tsukamoto, 2013). The roles of Smad4-dependent signals in skeletal growth, especially in adults, are still unclear because global knockout of Smad4 in mice is lethal during early embryonic development at a stage before skeletal development (Yang et al., 1998). We recently established a mouse line with Smad4 conditional knockout (cKO) induced by tamoxifen injection (Machiya et al., 2020).

Wnt family ligands are also regulators of bone formation. The importance of Wnt signaling in bone formation was originally discovered in the context of a rare autosomal recessive skeletal disorder, sclerosteosis (OMIM #269500), which is characterized by progressive skeletal overgrowth. A homozygous mutation was found in the SOST gene, which encodes an antagonist of Wnt ligands, in individuals with the disorder (Brunkow et al., 2001; van Bezooijen et al., 2007). Osteoporosis-pseudoglioma syndrome (OMIM #259770) is another rare skeletal disorder characterized by osteoporosis, fractures, osteogenesis imperfecta and low mineral density in the bones and retinoblastoma in the eyes. Loss-of-function mutations were identified in the LRP5 gene, which encodes a membrane receptor for Wnt ligands (Gong et al., 2001; Kato et al., 2002). A gain-of-function mutation in LRP5 was also found in familial individuals with high bone mass (Boyden et al., 2002; Little et al., 2002). There are 19 different ligands in the Wnt family, and they activate canonical and/or non-canonical intracellular signaling pathways. Canonical Wnt ligands, such as Wnt1, Wnt3a, Wnt10b and Wnt16, have been shown to regulate bone formation by binding to membrane receptors, such as Lrp5/Lrp6 and Frizzled proteins (Fzd1 to Fzd10) (MacDonald and He, 2012). They inhibit GSK3β in the cytoplasm, which is a serine/threonine kinase that phosphorylates β-catenin to induce proteasome-dependent degradation. β-Catenin stabilized by the canonical Wnt signal accumulates in nuclei as a complex with the TCF/LEF family and induces transcription of target genes, such as Axin2 (MacDonald et al., 2009).

In the present study, we found a marked increase in bone volume in the metaphysis in Smad4 cKO mice by stimulation of endochondral ossification. Osteoblasts forming the primary spongiosa accumulated with β-catenin. We identified Wnt7b in growth plate cartilage as the factor responsible for the stimulation of endochondral ossification, not only in Smad4 cKO mice, but also during physiological skeletal development in juvenile mice. We conclude that Wnt7b is one of the essential factors secreted from hypertrophic chondrocytes in growth plate cartilage to initiate endochondral ossification.

To examine the role of Smad4-dependent signaling in skeletal tissues in adult mice, we established mice with cKO of Smad4 by injecting tamoxifen into 10-week-old Smad4^f/f;CAG-Cre-ERt^Tg/− mice (Fig. S1A) (Machiya et al., 2020). Metaphyseal trabeculae, but not the cortex, increased aberrantly in the proximal tibiae and distal femur in Smad4 cKO mice compared with littermate controls on day 23 (Fig. 1A-E). Bone histomorphometric analysis showed that parameters related to bone formation, such as both the bone formation rate and the trabecular thickness, were increased in Smad4 cKO mice (Fig. 1F; Fig. S1B,C). In contrast, parameters of bone resorption, such as the osteoclast surface per bone surface and the number of osteoclasts per bone surface were not changed in Smad4 cKO mice (Fig. S1D,E). Stimulation of bone formation in Smad4 cKO mice was further supported by an increase in the number of osteocalcin-positive cuboidal cells in the metaphyseal trabeculae, especially in the primary spongiosa (Fig. 1G; Fig. S1F). The length of the primary spongiosa in the tibia increased in Smad4 cKO mice compared with controls (Fig. 1H). To distinguish bones that formed before and after Smad4 ablation, we injected calcein once to label pre-existing bones 1 day before tamoxifen injection (Fig. S1G). The length of the unlabeled spongiosa, which was formed after tamoxifen injection, increased in Smad4 cKO mice (Fig. 1I,J). Taken together, these results suggest that Smad4 ablation increases bone volume in the trabeculae by activating osteoblast-dependent bone formation.

Canonical Wnt signaling through β-catenin is an important regulator of bone formation (Baron and Kneissel, 2013; Maeda et al., 2019). In fact, abundant β-catenin accumulated in osteoblasts on the surface of metaphyseal trabeculae in the proximal tibiae in Smad4 cKO mice (Fig. 2A). Furthermore, simultaneous injection of a small chemical inhibitor of Wnt secretion, LGK-974, suppressed the increase in trabeculae in Smad4 cKO mice (Fig. 2B; Fig. S2A), suggesting that Wnt signaling is involved in increased bone formation in Smad4 cKO mice. We directly dissected metaphyseal growth plate cartilage and quantified the expression levels of mRNAs related to Wnt signaling. Neither Wnt antagonists, such as sclerostin, Dkk1 and sFRP, nor Wnt receptors, such as Lrp5, Lrp6 and Frizzled 1-10, exhibited altered mRNA expression in Smad4 cKO mice (Fig. S2B-D). Among the Wnt ligands examined, only Wnt7b mRNA expression increased more than fivefold in Smad4 cKO mice (Fig. 2C). Therefore, we established Smad4 and Wnt7b double cKO mice (Fig. S2E). Ablation of Wnt7b with Smad4 blunted the increase in metaphyseal trabeculae observed in Smad4 cKO mice (Fig. 2D,E), indicating that Wnt7b is essential for the increase in bone formation in Smad4 cKO mice.

To examine the role of Wnt7b in physiological bone development, we further established Wnt7b cKO and injected tamoxifen into 10-day-old Wnt7b^f/f;CAG-Cre-ERt^Tg/− juvenile mice (Fig. S3A). Wnt7b cKO mice were smaller and had shorter bones, such as the femurs and tibiae, than control mice (Fig. 3A,B; Fig. S3B-E). The bone area in metaphyseal tibiae, but not epiphysial tibiae, decreased in juvenile Wnt7b cKO mice compared with controls owing to the marked reduction in trabeculae (Fig. 3C,D). In contrast to bone, growth plate cartilage in the tibiae was unchanged even in Wnt7b cKO mice (Fig. S3F,G). Together, these results suggest that Wnt7b is important for bone formation through endochondral ossification.

Wnt7b protein was detected in hypertrophic chondrocytes in the growth plate in wild-type mice (Fig. 4A). Expression of Wnt7b mRNA was detected in primary chondrocyte cultures prepared from wild-type mice and increased during maturation to hypertrophic chondrocytes (Fig. 4B). Furthermore, Wnt7b mRNA expression was increased by Smad4 ablation in cultured chondrocytes prepared from Smad4^f/f;CAG-CreERt^Tg/− mice (Fig. 4C). Treatment with BMP4 inhibited Wnt7b mRNA expression in chondrocytes (Fig. 4C). Smad4 ablation in cultured chondrocytes increased Wnt7b mRNA expression and blunted the inhibition induced by BMP4 (Fig. 4C). Treatment with BMP4, but not TGFβ1, decreased Wnt7b mRNA expression within 3 h, and it was almost completely inhibited at 24 h in C2C12 cells (Fig. S4A).

Lrp5/6 and Reck function as receptors of Wnt7b in endothelial cells (Cho et al., 2017). Overexpression of Lrp5 or Reck alone induced a minimal response to Wnt7b in human embryonic kidney 293A (HEK293A) cells, which do not respond to Wnt7b under normal conditions (Fig. 4D). Co-expression of Lrp5 and Reck significantly increased the reporter activity of β-catenin signaling in response to Wnt7b stimulation (Fig. 4D). Furthermore, Wnt7b-containing conditioned medium stimulated osteogenic differentiation of bone marrow stromal cells and primary osteoblasts, with an increase in Axin2 mRNA expression (Fig. 4E; Fig. S4B,C). Cells expressing both Lrp5/6 and Reck were detected among bone marrow stromal cells from 6-week-old wild-type mice by flow cytometry analysis (Fig. 4F).

In the present study, by analyzing Smad4 cKO mice, we identified Wnt7b as the factor responsible for the initiation of bone formation during endochondral ossification. After Smad4 ablation in 10-week-old mice, bone volume in the metaphysis increased markedly as a result of increased bone formation via activation of the canonical Wnt–β-catenin pathway. In the growth plate cartilage of Smad4 cKO mice, only Wnt7b mRNA showed increased expression among the Wnt signaling-related molecules examined, such as antagonists, receptors and ligands. Wnt7b was abundantly expressed in hypertrophic chondrocytes in vitro and in vivo. Ablation of Wnt7b blunted the increase in bone in adult Smad4 cKO mice and suppressed physiological growth of the long bones in juvenile mice. Ihh secreted by pre-hypertrophic and hypertrophic chondrocytes has been found to be a crucial factor for osteogenesis in endochondral ossification during embryonic development (Chung et al., 2001; Long et al., 2004). Interestingly, expression of Wnt7b mRNA was induced by stimulation with Ihh in the undifferentiated fibroblast cell line C3H10T1/2 cells (Hu et al., 2005). Furthermore, artificial expression of Wnt7b under the control of the Col2a promoter rescued osteogenesis in the bone marrow cavity but not in the perichondrium in the Ihh and Gli3 double-deficient mice (Joeng and Long, 2014). Interestingly, the authors noticed that the expression of osteoblast markers coincided with vascularization, suggesting that osteogenesis induced by Wnt7b is caused by increased vascularization (Joeng and Long, 2014). These findings suggest that Ihh–Gli3 signaling cooperates with Wnt7b in osteogenesis during endochondral ossification. Indeed, transgenic mice overexpressing Wnt7b in osteoblast lineage cells under the promoter of osteocalcin (mature osteoblasts), osterix or type I collagen (osteoprogenitor cells) exhibited increased bone formation (Chen et al., 2014, 2021; Song et al., 2020). Furthermore, mechanical loading induced the expression of Wnt7b in mouse bone (Lawson et al., 2022). Reck and Gpr124 (Adgra2) receptors have been reported to be essential co-factors for Wnt7b-specific signaling in the central nervous system (Vanhollebeke et al., 2015; Cho et al., 2017). Recently, it was reported that the Wnt7a/Wnt7b/Gpr124/Reck signaling plays an essential role in limb development in mammals (Wang et al., 2022). We also identified double-positive Lrp5/6 and Reck cells in bone marrow stromal cell populations. Taken together, these findings suggest that Wnt7b is secreted by hypertrophic chondrocytes in the growth plate and induces osteogenesis of bone marrow stromal cells. We do not rule out the possibility that Wnt7b regulates body weight through other tissues, including the brain.

In contrast to the metaphysis, the epiphysis and cortex showed nearly unchanged bone volume in, not only Smad4 cKO, but also Wnt7b cKO mice, although they also form through the endochondral ossification process. One of the possible explanations for the different effects of Wnt7b in each part of bone is the timing of Smad4 and Wnt7b ablation during endochondral ossification. Although the metaphysis contains growth plate cartilage and continues endochondral ossification, the epiphysis and cortex do not contain hypertrophic chondrocytes or growth plates at 10 weeks of age in mice, when Smad4 ablation was initiated in our study. Another possibility is that another factor(s), rather than Wnt7b, regulates endochondral ossification in the epiphysis and cortex. Loss-of-function mutations in WNT7A have been found in individuals with Fuhrmann syndrome (OMIM #228930) and Awadi/Raas-Rothschild/Schinzel phocomel (OMIM #276820), which are rare genetic skeletal disorders characterized by deformation of the limbs (Woods et al., 2006). Indeed, Wang et al. reported that Wnt7a/Wnt7b signals are essential for limb development (Wang et al., 2022). WNT7B shows high amino acid sequence homology with WNT7A, suggesting that WNT7A is also a potential stimulatory factor during endochondral ossification of the long bones in humans. Additional studies are needed to clarify the molecular mechanisms of endochondral ossification in each part of skeletal tissues.

Smad4 is an essential co-activator for intracellular signaling downstream of both osteogenic and non-osteogenic members of the TGFβ family through Smad1/5 and Smad2/3, respectively (Katagiri and Watabe, 2016; Katagiri et al., 2018a,b). Although TGFβ1, a non-osteogenic ligand, stimulates early chondrogenesis in vitro, BMP4, an osteogenic ligand, stimulates terminal differentiation into hypertrophic chondrocytes (Fujimoto et al., 2014). Smad4 cKO mice showed an increase in Wnt7b mRNA expression with an increase in hypertrophic chondrocytes, suggesting that Smad4-dependent signaling suppresses both Wnt7b expression and chondrocyte maturation in growth plate cartilage. Ablation of a BMP type I receptor, ALK3 (Bmpr1a), in osteoblastic cells increased bone formation, with activation of canonical Wnt signaling, by suppressing the expression of Wnt antagonists, such as sclerostin and Dkk1 (Kamiya et al., 2008, 2010). These results suggest that Smad4-dependent intracellular signaling pathways involving the TGFβ family regulate various effectors of endochondral ossification, including Wnt7b, in growth plate chondrocytes. There is a possibility that signaling pathways other than Smad4, are involved in the expression of other Wnt ligands in bone metabolism. Additional studies are needed to identify the ligand(s) and receptor(s) upstream of Smad4-dependent signaling in chondrocytes during endochondral ossification.

Smad4^f/f mice were a generous gift from Dr Chuxia Deng (Yang et al., 1998). Wnt7b^f/f and CAG-Cre-ERt^Tg/− mice were purchased from The Jackson Laboratory (#008467, #004682) (Hayashi and McMahon, 2002; Rajagopal et al., 2008). Wild-type mice were purchased from CLEA Japan. The conditional Smad4 and Wnt7b alleles undergo Cre-mediated excision of exon 8 of the Smad4 gene and exon 2 of the Wnt7b gene, respectively. CAG-Cre-ERt^Tg/− mice express the Cre:ERα protein under the control of the CAG promoter. Smad4^f/f mice and Wnt7b^f/f mice were crossed with CAG-Cre-ERt^Tg/− mice to develop Smad4^f/f;CAG-CreERt^Tg/− mice, Wnt7b^f/f;CAG-CreERt^Tg/− mice and Smad4^f/f;Wnt7b^f/f CAG-CreERt^Tg/− mice. These mice were injected with tamoxifen (40 mg/kg/day or 100 mg/kg/day) intraperitoneally (Machiya et al., 2020). In the experiment involving injection of a Wnt secretion inhibitor, daily administration of LGK-974 (Adooq Bioscience) (3 mg/kg/day) to Smad4^f/f;CAG-CreERt^Tg/− mice was performed. Mice were randomly assigned to different experimental groups, and maintained on a 12-h light/dark cycle. This study was approved by the Institutional Animal Care and Use Committee and was conducted according to the Saitama Medical University Animal Experimentation Regulations.

Tissues were fixed in 4% paraformaldehyde (Nacalai Tesque) at 4°C overnight. Representative μCT images from a μCT35 system (Scanco Medical) and CosmoScan GX are shown. The spongiosa was evaluated by measuring 464 slices (2.78 mm) immediately below the growth plate. The key parameters for μCT scan using μCT35 were as follows: voxel size: 6 μm; X-ray tube potential: 70 kVp; X-ray intensity: 114 μA; integration time: 400 ms. Quantitative analysis of the cortical bone was performed using CosmoScan GX and Analyze 12.0 software (AnalyzeDirect, Inc.). X-energy was set at 90 kV and 88 μA and 4 min exposure time.

Tissues were dissected and fixed with 4% paraformaldehyde (Nacalai Tesque) in PBS at 4°C overnight. Subsequently, they were decalcified with Osteosoft (Merck Millipore) for 1 week at room temperature. Sections were embedded in paraffin and sliced at a thickness of 4 μm (for Hematoxylin & Eosin staining and immmunohistochemistry) or 10 μm (for Alcian Blue and Nuclear Fast Red staining). For immunohistochemical analysis, the paraffin sections were deparaffinized with xylene. Endogenous peroxidase activity was quenched for 30 min with 0.3% H₂O₂. Sections were treated with blocking reagent (Nacalai Tesque) at room temperature for 10 min and then sequentially incubated with primary antibodies against osteocalcin (rabbit polyclonal antibody; Takara Bio, M173; 2 μg/ml), β-catenin (rabbit polyclonal antibody; Merck Millipore, 06-734; 2 μg/ml) and Wnt7b (rabbit polyclonal antibody; Gene Tex; GTX65897; 1:1000) at 4°C overnight. The sections were reacted with a horseradish peroxidase-conjugated secondary antibody (Vector Laboratories, MP-6401) at room temperature for 30 min. Visualization was performed using a horseradish peroxidase substrate kit (Vector Laboratories, SK-4100), and the nucleus was counterstained with Hematoxylin.

Smad4 cKO mice and control mice were subcutaneously injected with calcein (Dojindo Laboratories) (10 mg/kg/day) 5 and 2 days before sacrifice. After the tibiae were fixed in 4% paraformaldehyde (Nacalai Tesque), they were embedded in compound designed for use in frozen sectioning [Super Cryoembedding Medium (SCEM), Section-Lab]. Sections of bones were prepared using the Kawamoto method (Kawamoto, 2003). Undecalcified sections of the tibia were subjected to Toluidine Blue and tartrate-resistant acid phosphatase (TRACP) staining. Dynamic histomorphometric analyses were performed using Histometry-RT (System Supply). The bone area of the metaphysis and epiphysis in paraffin sections sliced along the sagittal plane of the tibia was quantified using a Keyence BZ-X800 microscope and BZ-X800 Analyzer software (Keyence). Quantitative data of the bone area was determined using three to five sections per mouse prepared from four or five mice in each group. The abbreviations, symbols and units for histomorphometric parameters were derived from the recommendations of the American Society of Bone and Mineral Research Histomorphometry Nomenclature Committee (Dempster et al., 2013).

Murine C2C12 and L cells were obtained from ATCC, and cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 15% and 10% fetal bovine serum (FBS), respectively. HEK293A cells were purchased from Thermo Fisher Scientific, and cultured in DMEM supplemented with 10% FBS. Cell lines were maintained at 37°C in humidified CO₂ (5%) incubators. All cell lines tested negative for Mycoplasma.

Primary chondrocytes were prepared from the costa, which was collected and digested with 1 mg/ml collagenase A (Roche) at 37°C for 60 min. Fresh collagenase solution was added after removing the supernatant, and the sample was digested at 37°C overnight. Suspended cells were collected, and the filtrate was passed through a cell strainer and cultured in DMEM/F12 containing 10% FBS. Primary chondrocytes were cultured as pellets in serum-free chondrogenic medium in the absence of ligands for 1 day or 1 week (Fujimoto et al., 2014). Paraffin sections of the pellets were stained with Alcian Blue. Primary costal chondrocytes from Smad4^f/f;CAG-CreERt^Tg/− mice were treated with 4-hydroxy tamoxifen to eliminate Smad4 in vitro.

Primary osteoblasts were prepared from the calvaria of 1-day-old mice (Sone et al., 2019). Calvaria were shaken at 37°C for 10 min in α-MEM containing 0.1% collagenase Type 2 (Worthington Biochemical) and 0.2% dispase (Fujifilm Wako Pure Chemical Corporation). Digestion was repeated four times, and the second to fourth cell fractions were combined and passed through cell strainers. The cells were plated in α-MEM containing 10% FBS.

Bone marrow stromal cells were prepared from femurs from 4-week-old mice. The bones were placed in 0.6 ml tubes, which had a small hole at the bottom, and centrifuged in 1.5 ml tubes. The cell pellets were resuspended, hemolysis was performed, and the samples were filtered through a cell strainer. Adherent cells were cultured in α-MEM containing 20% FBS (Salazar et al., 2013, 2016b). To induce osteoblastic differentiation, cells were cultured with a 1:1 mixture of Wnt7b-conditioned medium (CM) and MesenCult osteogenic differentiation medium (Stem Cell Technologies). Subsequently, the cells were incubated for the indicated period of time, and the medium was changed every 3 days.

Under a microscope, metaphyseal bone was removed from the tibia. Growth plate cartilage was dissected from the epiphyseal bone with a surgical scalpel. Total RNA from tissues or cells was prepared using a NucleoSpin RNA kit (Macherey-Nagel) and RNA was reverse-transcribed to cDNA EcoDry Premix (Takara Bio). Quantitative RT-PCR was performed using Premix Ex Taq (Takara Bio). The primers used are listed in Table S1. Data are expressed as relative expression levels compared with those of Gapdh.

Wnt7b-conditioned medium was prepared from L cells transfected with a pcDEF3 expression vector carrying mouse Wnt7b cDNA and the blasticidin-resistance gene using Lipofectamine 2000 (Thermo Fisher Scientific). One million L cells were suspended in 10 ml of culture medium in a 10 cm dish and cultured for 3 days. The culture medium collected was used as Wnt7b-conditioned medium. Control conditioned medium were prepared from L cells transfected with a pcDEF3 vector without Wnt7b cDNA.

Bone marrow cells were separated from the femurs of 6- to 8-week-old wild-type mice. Cells were resuspended, and hemolysis was performed by shaking at 37°C for 30 min with α-MEM containing 0.1% collagenase (Worthington Biochemical) and 0.2% dispase (Fujifilm Wako Pure Chemical Corporation). The solution was passed through a cell strainer to obtain bone marrow cells. The cells were stained with a primary antibody, such as anti-CD31-FITC (rat IgG; BD Biosciences, clone 390; 1:100), anti-CD45-FITC (rat IgG; BD Biosciences, clone 30-F11; 1:100), anti-Ter119-FITC (rat IgG; Miltenyi Biotec, 120-002-040; 1:100), biotin-conjugated anti-Reck (mouse IgG2a; MBL, W050-3; 1:100) or anti-Lrp5/6 (mouse IgG2b; Santa Cruz Biotechnology, sc-57354; 1:100), at 4°C for 30 min and a streptavidin-PerCP-Cy5.5- (BD Biosciences, 551419; 1:100) or Alexa Fluor 647-conjugated anti-mouse-IgG2b (Thermo Fisher Scientific, A-21242; 1:500) secondary antibody at 4°C for an additional 30 min. Stained cells were analyzed with a BD Accuri C6 Flow Cytometer (BD Biosciences) according to the manufacturer's instructions.

HEK293A cells were seeded in 96-well plates at 1×10³ cells/well and transfected with plasmids [Wnt7b, Lrp5, Reck, Super8xTOPflash (kindly provided by Dr Randall Moon) and phRL-SV40 Renilla luciferase] using Lipofectamine 2000 transfection reagent according to the manufacturer's instructions (Veeman et al., 2003; Fukuda et al., 2010). After a 24-h treatment, luciferase activity was measured using the Dual-Glo Luciferase assay system (Promega). Murine Wnt7b, Lrp5 and Reck cDNAs were obtained using a standard RT-PCR technique, and each cDNA was subcloned into the pcDEF3 expression plasmid. The DNA sequences were confirmed using an ABI3500 genetic analyzer (Applied Biosciences).

Comparisons were performed using unpaired one-way ANOVA and an unpaired t-test using GraphPad Prism 9 software (GraphPad Software, Inc.). The results are shown as mean±s.d. Statistical significance is indicated as follows: *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

We thank Noriko Sekine, Misato Okubo, Chihiro Nakatomi, Miyoko Sekikawa and Eriko Fukuda for their excellent technical assistance and helpful discussions. We are grateful to Drs Chuxia Deng and Randall Moon for kindly providing Smad4^f/f mice and the Super8xToplash plasmid, respectively. We are grateful to the staff of the Division of Analytical Science, Biomedical Research Center and Experimental Animal Laboratory of Saitama Medical University for their excellent technical assistance in animal experiments.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201734.reviewer-comments.pdf.