|Home | About | Journals | Submit | Contact Us | Français|
Wnt ligands activate β-catenin-dependent canonical and –independent noncanonical signaling pathways. Wnt regulates many physiological events such as the development of organs and bone metabolism. In contrast, failed signaling leads to pathological conditions including cancer and osteoporosis. Analyses of loss-of-function mutations in the low-density lipoprotein receptor-related protein (Lrp) 5 gene revealed that Lrp5 acted as a co-receptor of Wnt/β-catenin signals and positively regulated bone mass in humans and mice. Many players in Wnt signals including sclerostin, an osteocyte-derived Wnt antagonist, also have since been found to influence bone mass. Bone mass is regulated by the activities of bone-forming osteoblasts, -resorbing osteoclasts and matrix-embedded osteocytes. The roles of Wnt/β-catenin signals in osteoblastogenesis and osteoclastogenesis have been established by the findings of a large number of in vitro and in vivo studies. In contrast, the roles of noncanonical Wnt signals in bone metabolism are only now being examined. In this review, we introduced and discussed recent information on the roles of Wnt signals in bone resorption.
Osteoclasts, multinucleated giant cells, are differentiated from monocyte-macrophage lineage cells.1 This differentiation is regulated by osteoblast-lineage cells including osteocytes. Osteoblasts express two cytokines that are essential for osteoclastogenesis, colony-stimulating factor 1 (Csf1) and receptor activator of nuclear factor-κB (NF-κB) ligand (Rankl, encoded by the Tnfsf11 gene).2,3,4 Rankl is inducibly expressed by osteoblasts5,6 and osteocytes7,8 in response to bone resorption-stimulating factors such as 1α,25(OH)2D3 and parathyroid hormone. Csf1 is constitutively expressed in osteoblasts. In contrast, osteoblasts and osteocytes also express osteoprotegerin (Opg, encoded by the Tnfrsf11b gene),9 which has been shown to inhibit the binding of Rankl to its receptor Rank (encoded by the Tnfrsf11a gene).10 Opg negatively regulates osteoclast differentiation.
The binding of Rankl to Rank in osteoclast precursors is known to activate several signaling pathways such as mitogen-activated protein kinases (Mapks), NF-κB, c-Fos and nuclear factor-activated T-cell cytoplasmic 1 (Nfatc1).11 Consequently, osteoclasts develop and express osteoclast-related genes such as tartrate-resistant acid phosphatase, cathepsin K and calcitonin receptors.
Glass et al.12 previously reported the importance of Wnt/β-catenin signals in osteoclast differentiation in vivo. They demonstrated that mice expressing a dominant active form of β-catenin in osteoblasts (DA-β-catenin mutant mice) exhibited a high bone mass phenotype with impaired osteoclast differentiation. The expression of Opg was also higher in DA-β-catenin mutant mice. These findings indicated that Wnt/β-catenin signals in osteoblasts induced the expression of Opg, which, in turn, inhibited osteoclast differentiation.
Wnt/β-catenin signals also regulate osteoblast differentiation to increase bone mass.13 Song et al.14 generated preosteoblast-specific β-catenin conditional knockout mice (β-catenin cKO), in which the β-catenin gene was postnatally deleted, and analyzed their bone phenotypes. β-catenin cKO exhibited a low bone mass phenotype with increased bone marrow adiposity. Ex vivo studies showed that the osteoblast differentiation was suppressed in bone marrow stromal cells isolated from β-catenin cKO. In contrast, the adipocyte differentiation was enhanced in those cells. These findings suggest that Wnt/β-catenin signals determine the cell fate of preosteoblasts from osteoblasts to adipocytes.
Subsequent studies found that β-catenin-independent noncanonical signals were also involved in osteoclastogenesis. In this review, we introduced recent information on the roles of β-catenin-dependent and β-catenin-independent Wnt signals in osteoclast differentiation.
The canonical Wnt pathway: in the absence of canonical Wnt ligands including Wnt3a, β-catenin is effectively degraded by a β-catenin destruction complex consisting of axin, adenomatous polyposis coli, glycogen synthase kinase 3 and casein kinase 1. Phosphorylated β-catenin is degraded by the destruction complex through the ubiquitin-proteasome pathway (Figure 1a). The binding of Wnt ligands to the receptor complex of frizzled and Lrp5/6 inactivates the β-catenin destruction complex, which, in turn, induces the cytosolic accumulation and nuclear translocation of β-catenin in target cells. The nuclear β-catenin with members of the T-cell factor (Tcf)/lymphoid enhancer factor family initiates the transcription of target genes (Figure 1b).
The noncanonical Wnt pathway: several Wnt ligands including Wnt5a activate the Wnt/Ca2+ and Wnt/planar cell polarity pathways. An increase in cytosolic Ca2+ has been shown to activate Ca2+-sensitive enzymes such as Ca2+-calmodulin-dependent protein kinase II and protein kinase C.17,18,19 Small G proteins including Rho and Rac and c-Jun N-terminal kinase (Jnk) were previously reported to have roles in the planar cell polarity pathway.20,21,22 Receptor tyrosine kinase-like orphan receptor (Ror) 1 and Ror2 act as alternative receptors or co-receptors for Wnt5a.23,24,25
Wnt/β-catenin signaling in osteoblasts is known to influence osteoclast differentiation in vivo.12 DA-β-catenin mutant mice had a high bone mass phenotype. A histomorphometric analysis revealed that osteoclast differentiation was impaired in DA-β-catenin mutant mice. In contrast, the number of osteoblasts was normal in these mutant mice. Using a micro-array analysis, Opg was found to be downregulated in Lrp5−/− calvariae and upregulated in DA-β-catenin mutant mouse calvariae. This study also demonstrated that β-catenin with TCF1 was recruited to the Opg promoter to induce the transcription of Opg (Figure 1b). However, the identity of the Wnt ligands that induced the expression of Opg remained unclear until recently.
Genome-wide association studies (GWAS) demonstrated that the Wnt16 locus reproducibly associated with bone mineral density, cortical bone thickness and non-vertebral fractures.26,27,28,29,30,31 These findings suggested that Wnt16 regulated bone formation and resorption to maintain bone mass in humans. Movérare-Skrtic et al.32 recently discussed the roles of Wnt16 in bone metabolism. Wnt16 was shown to act on osteoblasts and induce the expression of Opg expression (Figure 2). The expression level of Wnt16 mRNA in cortical bone was the highest among various tissues including trabecular bone. In cell cultures, primary osteoblasts strongly expressed Wnt16, whereas bone marrow-derived macrophages and osteoclasts did not. The treatment of MC3T3E1 cells, an osteoblastic cell line, with recombinant Wnt16 induced the expression of Opg mRNA due to the activation of Wnt/β-catenin signaling. The bone mass of cortical bone in Wnt16−/− mice was reduced, whereas that of trabecular bone was not. This finding supported previous data from the GWAS analysis and indicated that Wnt16 is a candidate ligand that induces the expression of Opg in bone tissues. Although Wnt16 was shown to activate Wnt/β-catenin signaling in osteoblasts in cultures, the mineralization of Wnt16−/− osteoblasts was similar to that of wild-type osteoblasts.32 In addition, the expression of osteoblast marker genes, such as Runx2 and Alpl (encoding alkaline phosphatase), and histomorphometric parameters for bone formation remained unchanged in the cortical bone of Wnt16−/− mice. In contrast, the expression of Opg mRNA was significantly lower in the cortical bone of Wnt16−/− mice. These findings indicated that Wnt16, secreted from osteoblasts, induced the expression of Opg in cortical bone, thereby predominantly regulating bone resorption in vivo. This study also showed the direct effects of Wnt16 on RANKL-induced osteoclast formation. This topic is discussed in a later part of this review.
An analysis of single nucleotide polymorphisms identified the WNT4 gene as one of the five genes associated with the cross-sectional area of vertebrae in caucasian men aged more than 65 years in the osteoporotic fractures in men study.33 GWAS subsequently confirmed that the WNT4 gene was associated with bone mineral density.26 Prior to these human studies being conducted, the role of Wnt4 in the differentiation of mesenchymal stem cells (MSCs) into osteoblasts was examined.34 The overexpression of Wnt4 in MSCs promoted their osteoblastic differentiation in cultures. Wnt4 was shown to enhance bone morphogenetic protein signals via p38 Mapk and promoted osteoblastogenesis. The transplantation of MSCs expressing Wnt4 enhanced the restoration of alveolar bone and calvarial bone defects. Recently, the overexpression of Wnt4 in osteoblasts is reported to impact osteoclast differentiation.35 This topic is discussed in a later part of this review.
Lrp4 has been identified as a receptor of agrin, a motor neuron-derived ligand.36 Agrin binds to Lrp4 and forms a complex with muscle-specific kinase (Musk) to develop neuromuscular synapses. Lrp4 has also been shown to bind Wise (Wnt modulator in surface ectoderm, also called sost domain-containing protein 1, Sostdc1).37 Choi et al.38 previously reported that Lrp4-bound dickkopf 1 (Dkk1) and sclerostin in vitro. They generated Lrp4-deficient mice (Lrp4-ECD), in which the transmembrane domain and intracellular domain of Lrp4 were deleted, and analyzed their bone phenotypes. Lrp4-ECD mice grew and exhibited a low bone mass at 10 weeks of age. The phenotype of Lrp4-ECD mice was largely different from that of Lrp4-null mice (described below). This discrepancy was attributed to the interaction between the mutant form of Lrp4 and its ligands in Lrp4-ECD mice. Leupin et al.39 searched for binding partners of sclerostin using tandem affinity purification proteomic approaches and identified Lrp4 as a new binding partner. The expression of Lrp4 was shown to enhance the inhibitory effects of sclerostin on Wnt/β-catenin signaling in HEK293 cells.39 An immunohistochemical study demonstrated that Lrp4 was expressed in the osteoblasts and osteocytes of human bone tissues. Notably, two mutations in the LRP4 gene (R1170W and W1186S) were identified in sclerosteosis patients. Both mutations were detected in the third propeller domain of LRP4. These mutant forms of LRP4 reduced the binding capacity and inhibitory effects of sclerostin. These findings indicated that Lrp4 bound sclerostin through the third propeller domain (Figure 1b).
Two recent studies demonstrated the important roles of Lrp4 in bone metabolism using cKO mice.40,41 Chang et al.40 generated mice lacking Lrp4 in their osteoblasts and osteocytes (Lrp4flox/flox; osteocalcin-cre mice) and osteocytes only (Lrp4flox/flox; Dmp1-cre mice). These cKO mice showed a high bone mass phenotype with a larger number of osteoblasts and a smaller number of osteoclasts. Serum levels of sclerostin in these mutant mice were higher compared with those in control mice, suggesting that Lrp4 retained sclerostin in bone tissues. Chang et al.40 also developed anti-Lrp4 antibodies that bound to the third propeller domain of Lrp4. These antibodies inhibited the facilitative effects of Lrp4 on sclerostin. As discussed above, Lrp4 is involved in Musk signaling to form neuromuscular synapses. The anti-Lrp4 antibodies did not inhibit the binding of Lrp4 to Agrin in C2C12 cell cultures. Furthermore, treatments with anti-Lrp4 antibodies increased bone mass in 7-month-old rats. These findings suggested the applicability of anti-Lrp4 antibodies to the treatment of osteoporosis, similar to anti-sclerostin antibodies.
Xiong et al.41 analyzed bone phenotypes in two Lrp4 mutant mice, osteoblast-specific Lrp4 knockout mice (Lrp4flox/flox; osteocalcin-cre mice) and muscle-rescued Lrp4-null mice (mr-Lrp4mitt). As Lrp4-null mice die at birth because of defects in neuromuscular junctions, Lrp4-null mice were rescued by crossing with transgenic mice expressing human skeletal α-actin promoter-driven Lrp4. The bone mass of trabecular and cortical bones was high in these mutant mice. Histomorphometric analysis revealed that parameters for bone formation were increased. Serum levels of sclerostin were higher in these mutant mice. The inhibitory effects of sclerostin and Wise, but not Dkk1, on alkaline phosphatase expression disappeared in bone marrow stromal cells isolated from mr-Lrp4mitt mice. This finding confirmed the roles of Lrp4 in the inhibitory actions of sclerostin and Wise. Furthremore, the number of osteoclasts and serum levels of deoxypyridinoline, a marker for bone resorption, were lower in both mutant mice. The treatment of bone marrow stromal cells from control mice, but not from mr-Lrp4mitt mice, with sclerostin increased the expression of Rankl and decreased that of Opg mRNA. This finding suggested that the expression of Rankl as well as Opg is regulated by the Wnt/β-catenin signaling pathway; however, the underlying mechanism has not yet been elucidated in detail. Thus, Lrp4 represents a new therapeutic target for increasing bone mass. We anticipate that anti-Lrp4 antibodies will be used in the future treatment of osteoporosis.
β-catenin signals in osteoclast precursors have been reported to have roles in osteoclast differentiation.42 DA-β-catenin mice and β-catenin cKO mice were generated using mice expressing Cre recombinases under the Ppar-γ promoter. Bone mass was higher in osteoclast precursor-specific DA-β-catenin and β-catenin cKO mice due to impaired osteoclast formation. Furthermore, the Rankl and Csf1-induced formation of osteoclasts was diminished in bone marrow-derived macrophage cultures isolated from these mice. These findings indicated that β-catenin signals in osteoclast precursors were required for their proliferation, whereas the sustained activation of β-catenin signals inhibited osteoclast formation. In contrast, we and others showed that the treatment of osteoclast precursors with Wnt3a had no effect on the Rankl-induced formation of osteoclasts; however, Wnt3a induced the cytosolic accumulation of β-catenin in these cells.43,44 Our finding was different from those in a former study using DA-β-catenin. This difference was attributed to the different methods used to activate Wnt/β-catenin signals. The expression of DA-β-catenin activated Wnt/β-catenin signals in cells more strongly than treatment with recombinant Wnt3a. Thus, Wei et al.42 reported that finely balanced β-catenin signals in osteoclast precursors may be required for proper osteoclastogenesis.
Wnt5a, a typical noncanonical Wnt ligand, activates β-catenin-independent signals in various types of cells. Ror1 and Ror2 have been shown to function as receptors or co-receptors of Wnt5a.23 In in vitro experiments, we found that osteoblasts strongly expressed Wnt5a, whereas bone marrow macrophages expressed receptor components of Wnt5a such as Frizzleds 1, 2 and 5, and Ror2.43 Immunofluorescence studies confirmed that osteoblasts expressing Runx2 and Osterix had positive signals for anti-Wnt5a antibodies on the bone surfaces of bone tissues.
Femurs isolated from 8-week-old Wnt5a+/− and Ror2+/− mice were analyzed to clarify the roles of Ror2-mediated signals in bone formation and resorption, because Wnt5a-deficient mice and Ror2-deficient mice die during the perinatal period. The findings of micro-computed tomography and a histomorphometric analysis revealed that Wnt5a+/− mice had a lower bone mass with impaired bone formation and resorption, whereas Ror2+/− mice had a higher bone mass with decreased bone resorption. The number of osteoclasts in both heterozygous mice was significantly lower than that in wild-type mice. To further clarify the roles of Ror2 signals in osteoclast formation, osteoclast precursor-specific Ror2 cKO mice were generated by crossing RankCre/+ mice with Ror2flox/flox mice. Ror2 cKO showed a high bone mass with impaired osteoclast differentiation. The number of osteoclasts was also lower in osteoblast-specific Wnt5a cKO mice, which were generated by crossing Wnt5aflox/flox with osterix-Cre mice. Taken together, these findings indicated that Wnt5a-Ror2 signals between osteoblasts and osteoclast precursors were critical for osteoclast differentiation.
We then attempted to elucidate the mechanism by which Ror2 signals positively regulate osteoclast differentiation. Osteoclasts were formed in bone marrow macrophage cultures treated with Rankl and Csf1. The addition of recombinant Wnt5a into the cultures enhanced the formation of osteoclasts. In contrast, the addition of Wnt3a, a typical canonical Wnt ligand, had no effect. The treatment with Wnt5a did not enhance Rankl-induced osteoclast formation in cultures of osteoclast precursors isolated from Ror2−/− mice.
During efforts to clarify how Ror2 signals enhance Rankl-induced osteoclast formation, we found that the expression of Rank mRNA as well as Ctsk (encoding cathepsin K), Calcr (encoding the calcitonin receptor) and Acp5 (encoding tartrate-resistant acid phosphatase) mRNAs was lower in tibiae from Ror2 cKO mice. The expression of Csf1r (encoding c-fms) mRNA remained unchanged. Furthermore, recombinant Wnt5a increased the expression of Rank mRNA in cultures of osteoclast precursors in the presence of Csf1. We also generated Rank-enhanced green fluorescence protein (EGFP) mice, in which EGFP was expressed by the Rank promoter-driven Cre recombinase. Using bone marrow cells isolated from these mice, the number of Rank-expressing cells was higher in bone marrow cell cultures treated with Csf1 plus Wnt5a than in those treated with Csf1 only.
A reporter assay using the Rank promoter revealed that the cluster of Sp1 sites on the proximal Rank promoter was important for Wnt5a-induced Rank expression.43 Chromatin immunoprecipitation assays further showed that the recruitment of c-Jun to the cluster region of Sp1 sites in the Rank promoter was critical for the Wnt5a-induced expression of Rank. Therefore, Wnt5a secreted from osteoblasts enhanced the expression of Rank in osteoclast precursors through the c-Jun–Sp1 interaction, thereby enhancing Rankl-induced osteoclast formation (Figure 3). We previously reported that osteoclast precursors highly expressing Rank were present in the vicinity of alkaline phosphatase-positive osteoblasts, even in bone tissues of Rankl-deficient mice.45 Wnt5a together with Csf1 secreted from osteoblasts may have roles in the localization of osteoclast precursors in the vicinity of osteoblasts.
As discussed above, Wnt16 activated Wnt/β-catenin signals, which, in turn, induced the expression of Opg in osteoblasts. In contrast, the treatment of osteoclast precursors with Wnt16 failed to induce the cytosolic accumulation of β-catenin or the expression of Axin2, a target gene of Wnt/β-catenin signals.32 Alternatively, Wnt16 markedly phosphorylated Jnk and c-Jun proteins, both of which were activated by noncanonical Wnt signals in osteoclast precursors.45 Wnt16 inhibited Rankl-induced osteoclast formation in human and mouse osteoclast precursors in culture. Wnt16 also inhibited the Rankl-induced activation of NF-κB signals and suppressed the Rankl-induced expression of Nfatc1. These findings suggested that Wnt16 also directly acted on osteoclast precursors, thereby inhibiting osteoclastogenesis (Figure 3). Although Wnt16 activated c-Jun in osteoclast precursors, the molecular relationship between c-Jun and the inhibitory effects of Wnt16 has not yet been clarified. We previously reported that Wnt16 failed to bind GST-soluble Ror2 containing the extracellular domain of Ror2.45 Movérare-Skrtic et al.32 showed that Wnt16 failed to activate Ror2 signals using reporter assays; however, Wnt16 induced the phosphorylation of Jnk in osteoclast precursors. A previous study demonstrated that the activation of c-Jun was important for the Rankl-induced formation of osteoclasts.46 Taken together, these findings indicated that the inhibitory effects of Wnt16 on osteoclast formation were not mediated by the Ror2-Jnk signaling axis in osteoclast precursors. Further studies are needed to elucidate the mechanism by which Wnt16 inhibits Rankl-induced signals.
Yu et al.35 generated Wnt4 transgenic (Wnt4-Tg) mice, in which Wnt4 was expressed in osteoblasts, and analyzed their bone phenotype. Bone mass was higher in Wnt4-Tg mice than in control mice. A histomorphometric analysis showed that the number of osteoblasts in bone tissues was moderately higher in 3-month-old Wnt4-Tg mice. Wnt4-Tg mice were protected from ovariectomy-induced and tumor necrosis factor (TNF)-α-induced bone loss. The number of osteoclasts was significantly increased in ovariectomized and Tnf-α transgenic mice. The overexpression of Wnt4 in the osteoblasts of these mouse models decreased the number of osteoclasts. These findings suggested that Wnt4 secreted from osteoblasts inhibited osteoclast formation.
The inhibitory action of Wnt4 was then examined. The treatment of osteoclast precursors with Wnt4 inhibited the Rankl-induced phosphorylation of transforming growth factor-activated kinase 1, thereby suppressing the activation of NF-κB signals. Wnt4 also suppressed Rankl-induced Nfatc1 expression in osteoclast precursors. Furthermore, Wnt4 failed to activate Wnt/β-catenin signals in these cells. These findings indicated that Wnt4 activated noncanonical signals, which, in turn, inhibited NF-κB and Nfatc1 signals in osteoclast precursors (Figure 3). The administration of Wnt4 proteins to mice also prevented ovariectomy-induced bone loss.
Wnt ligands and antagonists constitute a signal network in bone tissues and regulate osteoclast differentiation. Wnt5a activates the Ror2-Jnk-c-Jun signaling axis and enhances Rankl-induced osteoclast formation. In contrast, Wnt16 and Wnt4 inhibit osteoclast formation by inhibiting Rankl–Rank signals. In addition to the expression of Opg by Wnt/β-catenin signals in osteoblasts, Wnt ligands secreted from osteoblasts tightly regulate osteoclast differentiation through noncanonical Wnt signals. The relationship between Wnt ligands and their receptors and the signaling pathways in osteoclasts need to be clarified in more detail in future studies. Thus, an analysis of noncanonical Wnt signals as well as the canonical signals may lead to the development of new therapeutic drugs for osteoporosis.
This work was supported in part by Grants-in-Aid for Scientific Research (KAKEN) (25221310 (NT) and 25293423 (YK)) from the Ministry of Education, Cultures, Sports, Science and Technology of Japan.
The authors declare no conflict of interest.