TOPGAL Mice Show That the Canonical Wnt Signaling Pathway Is Active During Bone Development and Growth and Is Activated by Mechanical Loading In Vitro


  • Julie R Hens,

    1. Section of Endocrinology and Metabolism, Department of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut, USA
    Search for more papers by this author
  • Kimberly M Wilson,

    1. Department of Orthopedics, Yale University School of Medicine, New Haven, Connecticut, USA
    Search for more papers by this author
  • Pamela Dann,

    1. Section of Endocrinology and Metabolism, Department of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut, USA
    Search for more papers by this author
  • Xuesong Chen,

    1. Section of Endocrinology and Metabolism, Department of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut, USA
    Search for more papers by this author
  • Mark C Horowitz,

    1. Department of Orthopedics, Yale University School of Medicine, New Haven, Connecticut, USA
    Search for more papers by this author
  • John J Wysolmerski MD

    Corresponding author
    1. Section of Endocrinology and Metabolism, Department of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut, USA
    • Section of Endocrinology and Metabolism, Department of Internal Medicine, Yale University School of Medicine, TAC S131, Anlyan Center for Medical Research and Education, 300 Cedar Street, New Haven, CT 06520-8020, USA
    Search for more papers by this author

  • The authors have no conflict of interest.


We identified cellular targets of canonical Wnt signaling within the skeleton, which included chondrocytes, osteoblasts, and osteocytes in growing bone, but only osteocytes and chondrocytes in the mature skeleton. Mechanical deformation induced Wnt signaling in osteoblasts in vitro.

Introduction: Genetic evidence in mice and humans has implicated the canonical Wnt signaling pathway in the control of skeletal development and bone mass. However, little is known of the details of Wnt signaling in the skeleton in vivo. We used Wnt indicator TOPGAL mice to identify which cells activated this pathway during bone development and in the mature skeleton.

Materials and Methods: We examined canonical Wnt signaling during embryonic and neonatal bone development in TOPGAL mice. The TOPGAL transgene consists of a β-galactosidase gene driven by a T cell factor (TCF)β-catenin responsive promoter so that canonical Wnt activity can be detected by X-gal staining. Expression of Wnt signaling components was examined in primary calvarial cell cultures by RT-PCR. The effect of mechanical deformation on Wnt signaling was examined in primary calvarial cells grown on collagen I and stretched using Flexercell Tension Plus System FX-4000T. Immunohistochemistry was used to examine the localization of β-catenin in cartilage, bone, and cultured calvarial cells exposed to physical deformation.

Results and Conclusions: Canonical Wnt signaling was active in several cell types in the fetal and neonatal skeleton, including chondrocytes, osteoblasts, and osteocytes. With age, activation of Wnt signaling became less prominent but persisted in chondrocytes and osteocytes. Although osteoblasts in culture expressed many different individual Wnt's and Wnt receptors, the TOPGAL transgene was not active in these cells at baseline. However, Wnt signaling was activated in these cells by physical deformation. Together with the activation of canonical Wnt signaling in osteocytes seen in vivo, these data suggest that Wnt signaling may be involved in the coupling of mechanical force to anabolic activity in the skeleton.


Osteoporosis is a common medical problem that results from an imbalance between bone formation and bone resorption that leads to net bone loss. Most therapy for this disease is aimed at inhibiting bone resorption, but to cure osteoporosis, new bone formation must be stimulated. Therefore, there is currently great interest in understanding the regulation of osteoblast activity to guide the development of anabolic therapies. One molecular pathway that recently has been implicated as an important inducer of bone formation is the canonical Wnt signaling system. Wnt's are a family of 19 distinct cysteine-rich glycoproteins that have been extensively studied in cancer and development.(1) They signal through several different mechanisms, but the canonical pathway is the best described.(2–4) Canonical signaling is initiated when Wnt's bind to a cell-surface co-receptor complex consisting of 1 of 10 separate Frizzled receptors (FZ) and either LRP5 or LRP6. Activation of these co-receptors ultimately leads to the accumulation of β-catenin in the nucleus where it combines with a member of the lymphoid enhancing factor (Lef)/T-cell factor (TCF) family of transcription factors at specific DNA binding sites to activate Wnt-responsive genes. In the absence of Wnt signaling, cytoplasmic β-catenin undergoes sequential phosphorylation by casein kinase I (CK1) and glycogen synthase kinase-3β (GSK3), both of which act as part of a multiprotein assembly associated with the scaffolding protein, Axin, and the tumor suppressor protein, adenomatous polyposis coli (APC). Phosphorylated β-catenin is ubiquitinated and targeted for proteosomal degradation, which makes it unavailable for signaling. In the presence of Wnt's, activation of the receptor complex leads to the inactivation of GSK3, the inhibition of β-catenin degradation, and the accumulation of cytoplasmic β-catenin, which can then enter the nucleus and participate in transcriptional regulation.

The recent elucidation of the molecular lesions underlying several genetic bone disorders has highlighted the important function(s) of canonical Wnt signaling in bone. Individuals with osteoporosis-pseudoglioma syndrome (OPPG) suffer from blindness and osteoporotic fractures caused by low bone mass. This disorder is caused by loss-of-function mutations in the Wnt co-receptor gene LRP5.(5,6) Gain-of-function mutations in the LRP5 gene cause the opposite: an autosomal dominant disorder characterized by high BMD, torus palatinus, and an altered mandible shape.(7,8) Transgenic mouse models have corroborated these studies in humans. Mice lacking the LRP5 gene mimic OPPG and suffer from osteoporosis,(9) whereas expression of gain-of-function LRP5 mutations in mice has been reported to increase bone mass.(5) In addition, inhibition of Wnt signaling through the transgenic production of a secreted Wnt antagonist, Frzb-1, inhibits trabecular bone formation in adult mice by disrupting osteoblast and osteocyte function. Finally, disruption of the secreted Frizzled-related protein 1 gene (sFRP-1) in mice results in high bone mass caused by the postnatal activation of bone formation.(10) Thus, genetic models in both humans and rodents indicate that Wnt signaling is involved in the regulation of bone mass.

In addition to bone, Wnt's seem to be important in the development and function of cartilage and the growth plate. Wnt's 3a, 4, 5a, 5b, 7a, 11, and 14 have all been found to be differentially expressed during specific stages of mesenchymal condensation and/or chondrocyte differentiation.(11–21) Wnt's 4, 5a, and 14 are expressed within developing joints, and genetic evidence suggests that Wnt 14 may participate in synovial joint formation.(16) Finally, as with bone cells, it seems that inhibitors of Wnt signaling, such as sFRP-1 and frizzled motif associated with bone development-1 protein (Frzb-1), are required for proper chondrocyte maturation.(22–24)

Wnt signaling is complex, and although genetic evidence points to Wnt's as important in the regulation of skeletal growth and turnover, little is known of the details by which these pathways operate in the skeleton. In this study, we sought to clarify which cells in the developing and adult mouse skeleton showed evidence for activation of the canonical Wnt signaling pathway. To approach this question we used Wnt indicator, or TOPGAL, transgenic mice. These mice have a transgene consisting of the lacZ reporter gene under the regulation of a Wnt responsive element consisting of three concateramized T cell factor (TCF)β-catenin sites upstream of a minimal c-fos promoter. In these mice, activation of the canonical pathway leads to the production of β-galactosidase, which can be detected histologically.(25) We found that canonical Wnt signaling was active in chondrocytes at all stages of differentiation and in osteoblasts in growing bone. We found that osteocytes were the most prominent site of canonical Wnt signaling in bone and importantly that the canonical pathway was activated by mechanical deformation in cultured osteoblast-like cells. These data suggest that Wnt signaling has multiple functions in skeletal development and growth, including a potential role in mediating the anabolic activity induced by mechanical force.



TOPGAL mice were a gift from Dr Fuch's laboratory, and the derivation of TOPGAL mice has been described in detail previously.(25) They were maintained on an outbred (CD-1) background and were identified using PCR analysis of tail DNA. The primer set used in this analysis was as follows: forward 5′-TTGGAGTGACGGCAGTTATCTGGA-3′ and reverse 5′-TCAACCACCGCACGATAGAGATTC-3′. Embryos were harvested from timed pregnancies in which the morning of the appearance of a vaginal plug was considered to be day 0 of gestation. Adult bones were collected at various time-points as specified in the Results section. All animal use was approved and reviewed by IACUC (protocol 2001-07834).

Detection of β-galactosidase in skeletons

β-Galactosidase was detected as previously described,(26) but with the following modifications. Adult bones were fixed in 2% paraformaldehyde and 0.02% glutaraldehyde in PBS for 1 h at room temperature and washed twice in PBS. Bones were first decalcified in 4% EDTA for 17 days and then washed in PBS for 3 h before being incubated in 0.1% 4-chloro-5-bromo-3-indolyl β-d-galactopyranoside (X-gal), 2 mM MgCl2, 5 mM EGTA, 0.02% Nonidet P-40, 5 mM K3Fe(CN)6, and 5 mM K4Fe(CN)6·3 H2O at 30°C overnight. Bones were subsequently washed once with PBS and then postfixed in 4% paraformaldehyde at 4°C overnight. Individual bones were rinsed in 70% ethanol, embedded in paraffin wax, sectioned, and counterstained with eosin.(27) Neonatal and embryonic skeletons were not decalcified before staining for β-galactosidase. The brains and internal organs were removed from otherwise intact neonates or embryos, after which they were washed in PBS, fixed in 2% paraformaldehyde with 0.02% glutaraldehyde for 1 h at room temperature, washed again in PBS, and incubated overnight in the X-gal staining solution.(26) Neonates/embryos were postfixed in 4% paraformaldehyde for 45 minutes and stored in PBS with 0.02% sodium azide for a duration of up to 2 weeks before tissue was cleared with KOH as described.(28)

To detect β-galactosidase in primary calvarial osteoblasts from TOPGAL mice, cells on collagen I-coated Flexcell membranes were washed once in PBS, fixed for 3 minutes in 2% paraformaldehyde, washed twice with PBS, and stained overnight in X-gal solution at 30°C. After this, cells were washed twice with PBS, counterstained with Contrast red (Kirkegaard and Perry Laboratories, Gaithersburg, MD, USA), and coverslipped in anhydrous glycerin.

Cell isolation and culture

Murine calvarial osteoblasts were prepared as previously described.(29) In brief, calvaria from CD1 mice less than 48 h old were pretreated with EDTA in PBS for 30 minutes and subjected to sequential collagenase digestion (Worthington Biomedical Corp., Lakewood, NJ, USA), and cells collected from fractions 3-5 were used as the starting population. Cells were seeded on either plastic or collagen I-coated plates and cultured in α-MEM supplemented with 10% FCS. For the physical deformation experiments, calvarial cells were seeded on collagen I-coated plates (Flexcell Corp, Hillsborough, NC, USA) and were grown for 7-10 days before being subjected to repetitive deformation using a Flexercell Tension Plus System FX-4000T. We used a regimen consisting of cycles of 2 h stretch at 2.5-3% elongation at 0.3 Hz, alternating with 1 h rest for a total of 3, 6, 12, and 24 h. Each experiment was performed in triplicate. Statistical analysis for stretch versus not-stretched cells was with a one-sided Student's t-test, and all other comparisons were analyzed using the Tukey's multiple comparison test.

Isolation of RNA and RT-PCR

RNA was isolated from calvarial osteoblasts by the addition of Trizol to the culture dish per the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA) and was treated with DNase I before use (GenHunter, Nashville, TN, USA). First-strand cDNA synthesis and PCR amplification was performed using the Superscript One-Step RT-PCR with Platinum Taq kit from Invitrogen. The initial cDNA synthesis was performed at 50°C for 30 minutes. using 100 ng of RNA per reaction. The PCR amplification conditions varied depending on the primer set used (Table 1). RT-PCR analysis of calvarial osteoblast RNA was performed in three independent experiments, all of which gave the same results.

Table Table 1.. Expression of Wnt Pathway Genes in Primary Calvarial Cells, Assayed by RT-PCR Using Primers That Amplify the Nucleotides Indicated
original image

For quantitative real-time PCR, the initial cDNA synthesis was performed at 50°C for 30 minutes with 100 ng of RNA per reaction with 50 nM primers and an annealing temperature of 60°C using the 1-step Brilliant SYBR green QRT-PCR master mix kit (Stratagene, La Jolla, CA, USA). Cyclin D1 primers amplified a 110-bp region spanning nts 181-290 of BC044841 and were 5′-GCGTACCCTGACACCAATCT-3′ and 5′-ATCTCCTTCTGCACGCACTT-3′. We used MGAPH as our internal control. MGAPH amplified a 258-bp region spanning nts 325-583 of M32599 and were 5′-CGTGGAGTCTACTGGTGT-3′ and 5′-GATGGCATGGACTGTGGTCATGAGC-3′.


Primary calvarial cells were grown on collagen I-coated Flexercell plates. The membranes were excised from the plate and cut into fourths. Cells were fixed for 5 minutes with 4% paraformaldehyde/0.1% Triton-X100 in PBS and washed three times with PBS. Cells were blocked with 3% BSA and 10% goat serum for 1 h at room temperature. Rabbit-anti-mouse β-catenin (a gift from Dr D Rimm) was used at 1/500 in 0.1% BSA/PBS overnight at 4°C. Cells were washed three times in PBS and incubated with goat-anti-rabbit-Alexa546 at 1/500 in 0.1% BSA/PBS at room temperature for 30 minutes. Cells were washed again three times with PBS and incubated with 1 μg/ml mouse anti-SM antigen (Biomeda, Foster City, CA, USA) in 0.1% BSA/PBS overnight at 4°C. Cells were washed three times with PBS and incubated with 1/500 goat anti- mouse-Alexa488 (Molecular Probes, Eugene, OR, USA) in 0.1% BSA/PBS for 30 minutes. Cells were washed three times with PBS and one time in dH2O, dried slightly, and mounted in Prolong Anti-fade (Molecular Probes) per the manufacturer's instructions.

Cartilage from 1- to 2-day-old CD-1 mice and bones from 4-week-old CD-1 mice were fixed in phosphate-buffered 4% paraformaldehyde at 4°C for 12 h. Bones were held in 111 mM EDTA (Baker Co. 8993-01) containing 5% polyvinylpyrrolidone (PVP-40; Acros Organics 9003-39-8) at pH 7.4 for 1 week at 4°C. Routine immunohistochemistry was performed on dewaxed, paraffin-embedded cartilage and bone sections after using 7 mM citrate buffer (pH 6.0) heated under pressure for antigen retrieval. Nuclear β-catenin was detected with mouse anti-β-catenin (1-2 μg/ml) from Transduction Laboratories (Lexington, KY, USA) at 4°C overnight followed by the Vectastain Elite ABC (BA9200) and DAB (SK4100) kits according to Vector Laboratories (Burlingame, CA, USA).


Embryonic skeletal development

TOPGAL embryos were used to identify cells that were targets of canonical Wnt signaling in developing bone and cartilage. Embryos were harvested and stained for β-galactosidase at embryonic day 16 (e16) and e18. On e16, extensive β-gal staining was noted throughout the entire cartilaginous skeleton, including the base of the skull, the vertebral column, the ribs, and long bones (Fig. 1A). Wildtype littermates showed little endogenous β-galactosidase (Fig. 1B). Examination of e16 long bones at higher magnification showed that Wnt signaling was activated primarily within the cartilaginous ends of the long bones; β-gal staining was largely absent from the developing bone collar in the midportion of the bone (Fig. 1C).

Figure FIG. 1..

X-gal staining of β-galactosidase activity in cartilage and bones of e16 and neonatal TOPGAL mice. (A) TOPGAL embryo at e16. (B) Control embryo at e16. (C) TOPGAL femur e16. Note the relative lack of X-gal staining in the central portion of the bone where the bone collar is forming. (D) Neonate TOPGAL mouse skeleton. (E) β-galactosidase activity in the knee joint of a neonate TOPGAL mouse. (F) β-galactosidase in costal chondrocytes of a neonate TOPGAL mouse. (G) Spinal column of a neonate TOPGAL mouse showing intense β-galactosidase activity in the developing intervertebral discs. Scale bar = 1 mm.

By e18, β-gal was detected at both the base and dome of the skull. In the vertebral column, canonical Wnt signaling became restricted to the intervertebral discs of the spine. As before, there was intense β-gal staining at the ends of the long bones in the cartilaginous growth plates and within the developing joints between bones. In addition, TOPGAL expression was now noted in a speckled pattern within newly forming bone comprising the developing diaphysis. The cartilaginous ends of the ribs continued to stain intensely for β-gal (data not shown).

Postnatal bone development

In the neonate (Figs. 1D-1G), the pattern of β-gal staining was similar to that of an e18 embryo. Chondrocytes within the growth plates of the long bones (Fig. 1E) and within the costal cartilage (Fig. 1F) continued to show activation of Wnt signaling. There was a prominent speckled pattern of β-gal staining in the diaphysis of the long bones (Fig. 1E), the bony portion of the ribs, and in the dome of the skull (Fig. 1D). In the neonatal spine, the developing intervertebral discs stained intensely for β-gal (Fig. 1G).

We examined the histology of neonatal bones to discern which cell types within the skeleton were targets of canonical Wnt signaling. Clusters of chondrocytes were positive for β-gal throughout the growth plate of the long bones (Fig. 2A). β-gal+ chondrocytes were present in the resting, proliferative, prehypertrophic, and hypertrophic zones of the growth plates of long bones and costal cartilage (Figs. 2A and 2B). Thus, it did not seem that activation of Wnt signaling was specific for any one stage of chondrocyte differentiation. The growth plates of neonatal vertebrae showed a similar pattern of TOPGAL transgene activation. In addition, cells found in the fibrous annulus surrounding the intervertebral discs were also positive for Wnt signaling (Fig. 2G). Control, wildtype littermates showed no X-gal staining at these sites (Fig. 2I and data not shown).

Figure FIG. 2..

Histological sections of neonatal and 3-week-old TOPGAL mice assayed for β-galactosidase. (A-D, G, and I) Neonatal bones. (E, F, H, and J) 3-week-old bones. (A) Proximal radius. Note the prominent X-gal staining in the growth plate and osteocytes in the bone cortex. (B) Costal cartilage. Note the X-gal staining in clusters of chondrocytes. (C) Endosteal surface of the radius. (D) Periosteal surface of the clavicle. Note the β-galactosidase activity in the osteoblasts and osteocytes. (E) Vertebra. Note that osteocytes within the trabeculae show β-galactosidase activity but osteoblasts on the surface of the trabeculae do not. (F) β-galactosidase staining of osteocytes located in the calvarium. (G and H) Intervertebral discs. (G) In neonates, the growth plate and entire fibrous annulus are positive for β-galactosidase. (H) By 3 weeks of age, β-galactosidase activity becomes concentrated in cells within the lateral aspect of the fibrous annulus. (I) β-galactosidase staining of control neonate radius. (J) β-galactosidase staining of calvarium at 4 weeks of age in control mouse. Scale bar = 100 μm in A, B, E, G, H, and J. Scale bar = 25 μm in C, D, F, and I.

In neonatal bones, both osteoblasts and osteocytes appeared to be targets of canonical Wnt signaling. β-gal+ osteoblasts could be found on both endosteal and periosteal surfaces of the long bones, calvarium, and clavicle (Figs. 2C and 2D, data not shown), but only a minority of osteoblasts actually expressed β-gal. Interestingly, in all bones assessed, osteocytes were more prominently positive for β-gal; both a higher proportion of osteocytes stained and the staining was darker for each individual cell (Figs. 2C and 2D). We did not detect TOPGAL transgene expression within the bone marrow.

At 3 weeks of age, TOPGAL transgene expression could still be detected as speckled β-gal staining seen on whole mount preparations of long bones and the skull (data not shown). At this age, β-gal staining appeared more prominent at the metaphysis of long bones than in the diaphysis. On histological examination, TOPGAL expression was present in cartilage and growth plates as before. However, it now became difficult to find any transgene expression within osteoblasts. In contrast, osteocytes continued to show prominent TOPGAL transgene expression (Figs. 2E and 2F). Osteocytes in control mice did not stain for β-gal activity (Fig. 2J). By 9 weeks of age, there was a general decrease in the speckled pattern of β-gal staining within long bones and the skull. As before, no positive osteoblasts were noted in the bones on histological exam, and the overall decrease in staining corresponded to a decrease in the number of positive osteocytes (data not shown).

The lack of X-gal staining in control mice showed that the β-gal activity that we detected in skeletal cells from TOPGAL mice was the result of transgene expression and not the result of endogenous β-gal expression. To verify that transgene expression in these cells reflected activation of Wnt signaling, we also stained sections of cartilage and bone for β-catenin by immunohistochemistry. Canonical Wnt signaling should result in the translocation of β-catenin to the nucleus and nuclear staining for β-catenin has generally correlated with X-gal staining in TOPGAL mice in other tissues.(25,30) This was also the case for the skeleton (Fig. 3 and data not shown). As shown in Fig. 3A, although there was widespread cytoplasmic expression of β-catenin in chondrocytes, similar to the pattern of X-gal staining, only a subset of chondrocytes showed concentrated nuclear staining. Likewise, we could detect β-catenin in the nuclei of many osteocytes in bone tissue (Fig. 3C). These data support the notion that chondrocytes and osteocytes are targets of canonical Wnt signaling and show the specificity of TOPGAL expression for activation of this pathway in vivo.

Figure FIG. 3..

β-catenin staining in chondrocytes and osteocytes. (A and B) Sections through the growth plates of neonatal long bones. (A) Staining with a specific β-catenin antibody. (B) Staining with nonspecific IgG as a control. Note that there is widespread cytoplasmic staining of chondrocytes. In addition, a subset of chondrocytes (arrows) shows concentrated nuclear staining for β-catenin. (C and D) Sections through 3-week-old long bones. (C) Staining with a specific β-catenin antibody. (D) Staining with nonspecific IgG as a control. Note that a subset of osteocyte nuclei is positive for β-catenin staining. Scale bar = 10 μM.

TOPGAL expression in cultured osteoblasts

Given the previous literature suggesting that the canonical Wnt signaling pathway is an important stimulator of bone formation, we were surprised that osteoblasts were not a more prominent site of Wnt-inducible transgene expression in TOPGAL mice. To examine activation of this pathway in osteoblasts further, we cultured primary calvarial osteoblasts from 2-day-old TOPGAL mice. Mirroring our results in older mice in vivo, cultured osteoblasts did not express β-galactosidase during any stage of their growth in culture, including preconfluent, confluent, and postconfluent states (data not shown). Recently, discoid domain receptors (DDRs) and integrin-linked kinases (ILKs) have been shown to interact with components of the Wnt signaling pathway,(31,32,33) suggesting that extracellular matrix attachments may provide contextual clues for Wnt signaling. Therefore, we next examined calvarial osteoblasts grown on collagen I-coated plates for β-gal expression. As before, we examined preconfluent, confluent and postconfluent cultures for β-gal activity, and only rare cells were positive. To confirm that the cultured cells were competent to activate canonical Wnt signaling, we used RT-PCR to examine the expression of Wnt pathway genes. As shown in Table 1, these cells expressed all TCFs, FZ1-FZ9, LRP5, LRP6, DKK1 and 2, and SFRP1-SFRP5. All Wnt genes analyzed were expressed by the cultured cells, with the exception of Wnt1, 3, 3a, 7a, 8a, and 8b.

Wnt signaling is induced in cultured osteoblasts by mechanical strain

Given the prominent activation of TOPGAL expression in osteocytes in vivo and the ability of osteocytes to sense mechanical forces, we next asked if physical deformation might activate TOPGAL expression in cultured osteoblasts. Calvarial cells from 2-day-old TOPGAL mice were cultured on untreated or type I collagen-coated Flexcell plates until ∼80-90% confluent. At time 0, the medium was changed, and the cells were subjected to cycles of physical deformation for 3, 6, 12, or 24 h and subsequently stained for β-gal. Under these conditions, we noted a modest increase in β-gal activity (5.3 ± 0.9% of cells positive at baseline) in response to fresh media in cultures that were grown on type I collagen-coated plates, but not in cells on noncoated plates (Figs. 4A and 4C, data not shown). When the cells were exposed to mechanical deformation for 12 h, there was a marked induction in the number of β-gal+ (12.5 ± 0.5% of cells positive), but only in the cultures grown on collagen-coated plates (Figs. 4B and 4C). Activation of Wnt signaling was transient, with the number of β-gal+ cells peaking at 12 h of treatment. By 24 h, β-gal staining had diminished, despite the continued presence of the mechanical stimulus (data not shown). To ensure that the activation of TOPGAL expression in cultured osteoblast-like cells was a specific indicator of Wnt signaling, we performed two additional experiments. First, we treated the cultures with a soluble Wnt inhibitor. We used a commercially available hybrid molecule consisting of the extracellular domain of Frizzled 4 fused to the Fc portion of human IgG1 (FZ4/Fc; R&D Systems, Minneapolis, MN, USA) that has been shown to mimic the actions of the frizzled-related proteins and to inhibit canonical Wnt signaling.(34,35) As can be seen in Fig. 4C, addition of FZ4/Fc to cultured osteoblasts from TOPGAL mice lowered the percentage of cells staining for β-galactosidase activity at baseline and completely abolished the response to mechanical deformation. We also examined the expression of cyclin D1, an established Wnt target gene, in cultured wildtype osteoblasts exposed to mechanical deformation in the presence or absence of FZ4/Fc. As seen in Fig. 4D, mechanical deformation almost doubled the level of cyclin D1 mRNA as measured by real-time RT-PCR. As with β-galactosidase activity, inhibition of Wnt signaling with FZ4/Fc both lowered the basal level of cyclin D1 expression and abolished the response to mechanical deformation. Finally, we performed immunohistochemistry to determine if physical deformation was associated with translocation of β-catenin into the nucleus. As seen in Fig. 5, in cells grown on type I collagen and on Flexercell plates at baseline, the majority of β-catenin was located outside the nucleus in the cytoplasm and at the cell surface. However, exposure to physical deformation led to a concentration of β-catenin staining within the nuclei of the cells. These combined data are all consistent with the activation of Wnt signaling in response to mechanical deformation in cultured osteoblast-like cells.

Figure FIG. 4..

Mechanical deformation induces β-galactosidase in primary calvarial osteoblast-like cells grown on collagen I-coated Flexercell plates. (A) Representative example of β-galactosidase expression in nonstretched primary calvarial cells. (B) Representative example of β-galactosidase expression in primary calvarial cells after physical deformation for 12 h. (C) Quantification β-galactosidase induction on exposure to physical deformation in the presence or absence of the soluble Wnt inhibitor FZ4/Fc. Bars represent the mean and SE of two to five replicates. Comparison of nonstretched calvarial cells vs. stretched calvarial cells was p = 0.005. Likewise, a comparison of nonstretched vs. nonstretched cells treated with FZ4/Fc and stretched vs. stretched treated cells treated with FZ4/Fc were both statistically significant (p < 0.01). (D) Relative levels of expression of cyclin D1 mRNA by real-time RT-PCR. Scale bar = 50 μm.

Figure FIG. 5..

Selected fields of β-catenin staining in primary calvarial osteoblast-like cells in the presence or absence of physical deformation. (A, C, and E) Cells not exposed to physical deformation. (B, D, and F) Cells exposed to cyclic deformation for 12 h. (A and B) Confocal images of β-catenin staining. (C and D) Confocal images of nuclear SM-antigen staining. (E and F) Merged and colorized (β-catenin in red, nuclear SM-antigen in green) images showing co-localization (yellow staining) of β-catenin and the SM-antigen. (A and E) Note that in the absence of physical deformation, β-catenin is located throughout the cells and is not concentrated within nuclei. (B and F) However, after exposure to physical deformation, β-catenin becomes concentrated in the nuclei and co-localizes with the nuclear marker in some cells. See Fig. 4 for a quantitative assessment of activation of Wnt signaling by physical deformation.


Canonical Wnt signaling has long been recognized as important to the regulation of cell proliferation and differentiation during development, and dysregulation of Wnt signaling has been shown to contribute to carcinogenesis. More recently, the canonical Wnt signaling pathway also has been implicated in the regulation of chondrogenesis and osteogenesis in the skeleton.(5–7,17,36–39) To help clarify the normal functions of this signaling pathway in the skeleton, we used Wnt indicator (TOPGAL) mice to identify cells that were targets of Wnt signaling during normal bone development and maturation. We found that in the developing and growing skeleton, canonical Wnt signaling was active in chondrocytes, the fibrous annulus of intervertebral disks, developing joints, osteoblasts, and osteocytes. In more mature bone, TOPGAL expression seemed to be restricted to chondrocytes and osteocytes. Finally, consistent with the apparent activation of Wnt signaling in osteocytes, we found that physical deformation was able to activate Wnt signaling in cultured osteoblast-like cells.

Canonical Wnt signaling was active in chondrocytes throughout the cartilaginous portions of the skeleton, and TOPGAL expression within chondrocytes persisted throughout life. During embryonic development, canonical Wnt signaling was active in the cartilaginous condensations marking the beginnings of endochondral bone development. Once bone formation commenced, TOPGAL transgene expression was prominent in chondrocytes located in the growth plates of endochondral bones. In adult mice, Wnt signaling remained active in chondrocytes within the growth plates, but TOPGAL expression was less prominent and was detected in a smaller proportion of cells. However, Wnt signaling remained active in the articular and costal cartilage that persists in the adult. These patterns suggest that Wnt signaling could serve one or more of several different functions in cartilage development: (1) the specification of a chondrocyte fate from mesenchymal precursors, (2) the regulation of chondrocyte proliferation and/or differentiation within the growth plate, and (3) the maintenance of the chondrocyte phenotype in parts of the skeleton that remain cartilaginous. These results also are consistent with a growing literature showing that Wnt signaling is important to the regulation of chondrocyte biology. Several studies have shown that a variety of different Wnt's, including 2b, 3a, 5a, 5b, 7a, 7b, 8, and 10b, are expressed by chondrocytes in culture as well as in cartilage in vivo. Likewise, several Wnt receptors, such as Frizzleds 1, 2, 3, 5, 7, 8, and 9, as well as LRP5 and 6, are expressed in chondrocytes. In cell culture experiments, individual Wnt's have been shown either to promote or inhibit the proliferation and differentiation of chondrocytes.(12,17,40–43) Transgenic studies have suggested that Wnt5a and Wnt5b might regulate the transition between different stages of chondrocyte differentiation within the growth plate.(15,17,44) We found TOPGAL expression within all types of chondrocytes and in all types of cartilage, so that apart from confirming that Wnt signaling is active in these cells in vivo, our results do not help to sort out the biological functions of individual Wnt's in cartilage. However, when bred to other transgenic models of altered Wnt signaling in chondrocytes, TOPGAL and other Wnt indicator mice may prove useful in assigning specific functions to individual Wnt signaling components in the growth plate and other cartilaginous structures.

We also observed strong TOPGAL expression within developing joints in the appendicular skeleton and within the intervertebral discs within the spine. Canonical Wnt signaling was active in articular cartilage as well as within the cells forming the fibrous annulus. Within the intervertebral discs, Wnt activation persisted in cells at the periphery of the fibrous annulus in older mice. These findings are consistent with prior reports and likely represent the actions of several different Wnt's during joint formation. Hartmann and Tabin(16) showed that Wnt14 is expressed at sites of joint formation in the appendicular skeleton and that misexpression of Wnt14 could lead to the formation of ectopic joints. Additionally, Wnt4 and Wnt5a have been localized to the fibrous connective tissue of the joint capsule as well as the synovial membrane, where they may participate in the formation of joint structures.(15,17) No reports have specifically addressed the functions of Wnt signaling in the intervertebral discs but the patterns of TOPGAL expression suggest that Wnt signaling might play a role both in the formation and in the maintenance of this structure.

A series of recent reports has documented that Wnt signaling is involved in the regulation of bone mass in humans and mice. Gain-of-function mutations in the Wnt co-receptor, LRP5, result in high bone mass in humans(7,8) and increased trabecular bone volume, cortical thickness, and bone formation in mice.(45,46) Loss-of-function mutations in LRP5 cause osteoporosis pseudoglioma (OPPG) syndrome in humans.(5) This syndrome results in low bone mass as a result of a defect in bone accrual. In mice lacking LRP5, osteoblast proliferation and differentiation are suppressed, and bone formation and trabecular volume are diminished.(9) In addition, deletion of secreted frizzled-related protein 1 (sFRP1) has been shown to increase bone mass in mice. Alterations in bone mass in these genetic models seem to result from alterations in osteoblast function. Studies in both cell culture and in mice have suggested that canonical Wnt signaling is able to increase osteoblast precursor proliferation, promote osteoblast differentiation, and prolong osteoblast survival.(5,3,46) Our studies with TOPGAL mice confirm that the canonical pathway is active in osteoblasts in situ, especially in the developing and growing skeleton. However, we did not find evidence of TOPGAL expression in osteoblasts in the mature skeleton, suggesting that canonical Wnt signaling may play a more prominent role in bone development and modeling than in normal bone remodeling. The lack of TOPGAL expression in osteoblasts in mature bone might also be explained by a lack of sensitivity of transgene expression in response to low levels of Wnt signaling. It will be interesting to see if other studies that use Wnt-indicator mice reproduce our data. However, these findings are consistent with the early onset of high bone mass in both humans and mice carrying activating mutations in LRP5. Furthermore, sFRP1−/− mice develop increases in bone mass only with aging because of an apparent derepression of Wnt signaling. Therefore, it is possible that the level of canonical Wnt signaling seen in fetal and neonatal osteoblasts is normally somewhat dampened by sFRP1 and/or other Wnt inhibitors with age, resulting in the relative loss of TOPGAL expression that we observed.

At all time-points and in all bones that we examined, osteocytes expressed the TOPGAL transgene, suggesting that Wnt signaling was active in these cells. As in osteoblasts, TOPGAL expression was more prevalent in osteocytes in rapidly growing bones. However, even in mature bones, there were always some osteocytes that were positive for β-gal expression. Because osteocytes are thought to be involved in sensing mechanical forces in the skeleton, we examined the effects of mechanical deformation on activation of canonical Wnt signaling in cultured osteoblast-like cells from TOPGAL mice. As is the case in vivo, we found little TOPGAL transgene expression in osteoblasts cultured under baseline conditions on plastic. However, when these cells were grown on type I collagen and were subjected to cyclical mechanical deformation, 12.5 ± 0.5% of the cells activated canonical Wnt signaling pathway as evidenced by the induction of TOPGAL/β-gal expression and the upregulation of expression of the Wnt target gene, cyclin D, as well as the ability of the soluble Wnt inhibitor, FZ4/Fc, to block these responses. Furthermore, mechanical deformation was associated with translocation of β-catenin into the cell nuclei, another indicator of canonical Wnt pathway activation. These data suggest that Wnt activity in osteocytes may be dependent on specific interactions with the extracellular matrix as well as pathways involved in sensing mechanical forces.

The dependence of Wnt signaling on matrix interactions is not surprising because osteocytes are completely surrounded by bone matrix and because integrin-linked kinase (ILK) can phosphorylate GSK3 and enhance β-catenin/TCF-mediated signaling.(31,33,47,48) ILK also can decrease cadherin expression (freeing β-catenin from the cell surface) and increase LEF1 expression, both of which would be expected to augment canonical Wnt signaling.(49) However, although matrix interactions were permissive in our experiments, activation of Wnt signaling required mechanical deformation. It is not clear how mechanical forces induce activation of canonical Wnt signaling. It is tempting to speculate that this signal may be transmitted through integrins and/or the actin cytoskeleton given the known interactions of β-catenin with the cytoskeleton at the cell surface. However, activation might also be mediated by signaling initiated by mechano-sensitive ion channels or through an autocrine-signaling loop involving other secreted molecules. Because canonical Wnt signaling clearly can stimulate bone formation, we hypothesize that Wnt signaling in osteocytes may be involved in the coupling of mechanical deformation to anabolic activity in the skeleton.

The important limitations of this study are 2-fold. First, the histological detection of TOPGAL transgene expression in bone tissue may not be fully sensitive and we may have underestimated the numbers of cells activating Wnt signaling in vivo. However, by several criteria, we are confident that X-gal staining in TOPGAL mice was specific; that is, positive cells truly had activated Wnt signaling. Second, calvarial osteoblast-like cells in culture are not osteocytes. Therefore, it is not clear how closely they mimic the mechanotransduction pathways found in osteocytes in situ. Nevertheless, our data show that mechanical force can activate Wnt signaling in cells of the osteoblast lineage.

In summary, we used mice bearing a Wnt-responsive β-galactosidase transgene to identify cells within the skeleton that were targets of canonical Wnt signaling. We found that several cell types, including chondrocytes, osteoblasts, and osteocytes, activated this pathway in vivo. In general, canonical Wnt signaling seemed to be more active during bone development and growth than during bone remodeling in the mature skeleton. Interestingly, osteocytes were prominent targets of Wnt signaling in vivo and osteoblast-like cells activated the canonical pathway in response to mechanical deformation in vitro. Future studies should focus on Wnt signaling in osteocytes to clarify the mechanisms by which this signaling pathway might mediate the effects of physical force on the regulation of bone mass.


This work was facilitated by the Yale Core Center for Musculoskeletal Disorders (NIH AR 46032). These studies were supported by NIH grants DK55501 and CA94175 to JJW and AR47342 and AR49190 to MCH. We would like to thank Nancy Troiano and Christiane Coady for their advice and histological assistance.