We have previously shown that targeted expression of a dominant-negative truncated form of N-cadherin (Cdh2) delays acquisition of peak bone mass in mice and retards osteoblast differentiation; whereas deletion of cadherin 11 (Cdh11), another osteoblast cadherin, leads to only modest osteopenia. To determine the specific roles of these two cadherins in the adult skeleton, we generated mice with an osteoblast/osteocyte specific Cdh2 ablation (cKO) and double Cdh2+/−;Cdh11−/− germline mutant mice. Age-dependent osteopenia and smaller diaphyses with decreased bone strength characterize cKO bones. By contrast, Cdh2+/−;Cdh11−/− exhibit severely reduced trabecular bone mass, decreased in vivo bone formation rate, smaller diaphyses and impaired bone strength relative to single Cdh11 null mice. The number of bone marrow immature precursors and osteoprogenitor cells is reduced in both cKO and Cdh2+/−;Cdh11−/− mice, suggesting that N-cadherin is involved in maintenance of the stromal cell precursor pool via the osteoblast. Although Cdh11 is dispensable for postnatal skeletal growth, it favors osteogenesis over adipogenesis. Deletion of either cadherin reduces β-catenin abundance and β-catenin-dependent gene expression, whereas N-cadherin loss disrupts cell-cell adhesion more severely than loss of cadherin 11. Thus, Cdh2 and Cdh11 are crucial regulators of postnatal skeletal growth and bone mass maintenance, serving overlapping, yet distinct, functions in the osteogenic lineage.
Cadherin-mediated cell-cell adhesion is a fundamental mechanism involved in cell fate specification, tissue organization and morphogenesis during embryonic development (Takeichi, 1995). Evidence from different tissue systems indicates that expression of different repertoires of cadherins is required for commitment of progenitor cells to specific lineages during cell fate specification and tissue differentiation (Gumbiner, 1996; Vleminckx and Kemler, 1999). We and others have previously demonstrated that uncommitted mesenchymal embryonic cells express several cadherins, and that commitment and differentiation into osteogenic, myogenic or adipogenic lineages is associated with regulated expression of specific cadherin repertoires (Kawaguchi et al., 2001b; Shin et al., 2000; Stains and Civitelli, 2005a). More specifically, cells of the osteoblast lineage are defined by the presence of two cadherins, N-cadherin and cadherin 11; and earlier in vitro data had suggested that cadherin 11 was specifically involved in osteoblast differentiation (Cheng et al., 1998; Kawaguchi et al., 2001b; Okazaki et al., 1994). In fact, overexpression of cadherin 11 in uncommitted embryonic cells gives rise to teratomas containing bone and cartilage (Kii et al., 2004). Cadherin 11 has a multi-functional role in the skeletal system, and has also been reported to be crucial for the development of the synovium (Lee et al., 2007). However, targeted ablation of the cadherin 11 gene (Cdh11) does not affect skeletal development, and Cdh11-deficient mice exhibit only a modest osteopenia and minor calcification defects of the cranial sutures (Kawaguchi et al., 2001a), raising the possibility that lack of cadherin 11 might be compensated for by another member of this superfamily – the most likely candidate being N-cadherin.
The notion that osteoblast cadherins may serve overlapping functions in postnatal osteogenic differentiation and bone remodeling is borne out of a previous in vivo study from our group demonstrating that disruption of cadherin function by expression of a dominant-negative mutant of the N-cadherin gene (Cdh2) in differentiated osteoblasts results in delayed peak bone mass acquisition, impaired osteogenic differentiation and an osteogenic to adipogenic shift in bone marrow stromal cell precursors (Castro et al., 2004). The skeletal phenotype produced by in vivo expression of this dominant negative cadherin is more severe than the phenotype reported in Cdh11-null mice, strengthening the hypothesis that another cadherin may compensate for lack of cadherin 11. Although such a dominant-negative approach demonstrated the biological role of osteoblast cadherins in postnatal bone homeostasis, the relative role of specific cadherins could not be distinguished. Targeted gene ablation is a more suitable method for investigating these roles. In fact, a compound gene ablation strategy has previously shown more severe abnormalities of somite structure in double Cdh2;Cdh11-null embryos relative to single cadherin gene deletion (Horikawa et al., 1999).
N-cadherin is more widely expressed than cadherin 11, but its function in the skeletal system remains to be fully clarified (Marie, 2002; Stains and Civitelli, 2005b). Several in vitro studies have demonstrated impairment of osteoblast differentiation by disruption of N-cadherin function using inhibitory peptides (Cheng et al., 1998; Ferrari et al., 2000), N-cadherin function blocking antibodies (Hay et al., 2000), antisense RNA (Lemonnier et al., 2001) or a dominant-negative mutant (Cheng et al., 2000; Ferrari et al., 2000). However, emerging in vivo data from cadherin mutant mouse models suggest a more complex function of N-cadherin in the osteogenic lineage. Ex vivo limb bud cultures from Cdh2-null embryos partially rescued by transgenic expression of E-cadherin were able to undergo cartilage condensation and develop into structured limbs in the absence of Cdh2 (Luo et al., 2005), even though earlier in vitro studies had shown that N-cadherin was involved in this process (Haas and Tuan, 1999; Tuli et al., 2003). Furthermore, we have recently reported that Cdh2 haploinsufficiency in mice does not alter postnatal skeletal growth, but it accentuates ovariectomy induced bone loss, the result of an attenuated activation of bone formation following estrogen deprivation (Fang et al., 2006). Defective bone formation response to ovariectomy was associated with reduced osteoblast recruitment from stromal cell precursors in Cdh2 haploinsufficient mice, whereas full osteoblast differentiation was actually facilitated by partial loss of N-cadherin. These data raise the intriguing possibility that N-cadherin may in fact inhibit late steps of osteoblast differentiation, whereas its major biological effect may be exerted at the stage of osteogenic commitment (Fang et al., 2006). In fact, an inhibitory action of N-cadherin on full osteoblast differentiation is supported by the recent demonstration that in vivo overexpression of Cdh2 in osteoblasts leads to osteopenia, via inhibition of Wnt signaling (Hay et al., 2009).
A more precise understanding of the biological role of Cdh2 in bone-forming cells requires selective gene ablation, as germline Cdh2 null mutation is embryonically lethal (Radice et al., 1997). Conditional gene ablation and compound haploinsufficient models have been used to assess the functional relationship between two proteins, particularly when single gene deletion is lethal, as in the case of Cdh2. In this work, we used both approaches to determine the relative roles of Cdh2 and Cdh11 in postnatal skeletal growth, and on the differentiation and function of bone forming cells. We find that germline deletion of one Cdh2 allele in a Cdh11-null background severely worsens the postnatal growth defect of Cdh11-null mutation, a phenotype that is more severe than that produced by deleting Cdh2 selectively in committed osteoblasts. At the cellular level, our results reveal that Cdh2 and Cdh11 are both crucial for osteogenesis, but they serve distinct, though partially overlapping functions: Cdh2 contributes to maintain the pool of bone marrow stromal cell precursors, whereas Cdh11 is involved in osteoblast commitment and full differentiation. These actions are associated with modulation of cadherin-dependent cell-cell adhesion and β-catenin abundance.
Decreased bone mass and microarchitectural abnormalities in cadherin deficient mice
Conditionally Cdh2 ablated Cdh2−/fl;ColCre mice are viable at birth and show no skeletal dysmorphisms, except that they are smaller than their wild-type equivalent littermate and at 6 months of age they have ~13±2.3% lower body weight. Although whole BMD by DXA was not different between Cdh2−/fl;ColCre and Cdh2+/fl or heterozygous Cdh2−/fl mice during the first 3 months of life, mild osteopenia developed with age, and at 6 months BMD was about 5% lower in conditional Cdh2-ablated mice relative to their wild-type and heterozygous littermates (Fig. 1A). Microstructural analysis of femur proximal metaphyses by μCT confirmed thinned trabecular patterning in the conditional knockout animals (Fig. 1B,C). Smaller cross-sectional areas of femoral mid diaphyses were also observed in the mutant relative to wild-type Cdh2+/fl littermates (Fig. 1B′-C′). As heterozygous Cdh2−/fl mice were phenotypically identical to Cdh2+/fl mice, they were not followed further, although cells were used for some in vitro experiments.
However, whole body BMD was significantly decreased in the double Cdh2+/−;Cdh11−/− mutants relative to Cdh11 null littermates, a difference that became more pronounced with age. At 6 months, Cdh2+/−;Cdh11−/− mice were 10.1±6.3% more osteopenic relative to Cdh11−/− littermates, whereas there was no difference between Cdh11−/− and wild-type mice (Fig. 1D). Furthermore, Cdh2+/−;Cdh11−/− double mutants appeared slightly smaller at birth, and their body weight was significantly lower than that of the other genotypes at all ages, with a difference of 19.8±2.2% at 6 months relative to Cdh11−/− mice. Tibiae of double mutants were also slightly shorter than those of Cdh11-null mice, with a difference of 3.3±0.35% at 6 months (P<0.01). Accordingly, microstructural analysis of proximal tibiae clearly showed a severely rarefied trabecular component at the tibial metaphysis (Fig. 1E-F), and significantly reduced cross-sectional area in Cdh2+/−;Cdh11−/− bones relative to Cdh11−/− bones (Fig. 1E′-F′). There were no differences in BMD or other structural parameters between Cdh11−/− and wild-type mice; thus, no further in vivo analyses were performed in the latter.
Quantitative assessment of μCT scans confirmed significantly lower trabecular bone volume in Cdh2+/−;Cdh11−/− relative to Cdh11−/− mice, with only a marginal decrease in BV/TV in Cdh2−/fl;ColCre relative to Cdh2+/fl (Fig. 2A). Cortical thickness in either Cdh2+/−;Cdh11−/− or Cdh2−/fl;ColCre mice was not different relative to their respective comparators (Fig. 2B). However, cross-sectional area of the tibial diaphysis was significantly smaller in both Cdh2+/−;Cdh11−/− and Cdh2−/fl;ColCre mutants (Fig. 2C). Accordingly, the calculated cross-sectional moment of inertia was significantly decreased in the two mutants compared with their controls (Fig. 2D), suggestive of a lower resistance to mechanical strain in cadherin-deficient mice. This conclusion was corroborated by direct assessment of resistance to load in a three-point bending test, which resulted in significantly reduced ultimate force in both Cdh2+/−;Cdh11−/− and Cdh2−/fl;ColCre mutants relative to their respective comparator littermates (Fig. 2E).
Histomorphometric analysis of trabecular bone did not reveal any differences in osteoblast or osteoclast number when comparing either cadherin deficient mutants with their controls (Fig. 2F,G). Likewise, dynamic indices of bone formation demonstrated no differences in MAR and BFR/BV between conditional Cdh2 knockout mice and their controls (Fig. 2H,I), with abundant double fluorescent labels present in either genotype (Fig. 2J,K). However, both parameters of new bone formation were significantly reduced in the double Cdh2+/−;Cdh11−/− mutants (Fig. 2H,I), relative to single Cdh11 knockout mice, with evident though sparse double labels in the latter (Fig. 2L), and very few areas with double labels were visible in the double mutant bones (Fig. 2M).
Decreased bone marrow stromal cell precursors and altered differentiation in cadherin-deficient mice
We then studied the cellular basis for the low bone mass phenotype of cadherin mutant mice. Calvaria cells isolated from Cdh2−/fl;ColCre mice did not exhibit defects of in vitro mineralization, if anything they mineralized even faster than wild-type cells (Fig. 3A). Accordingly, gene expression profiling of Cdh2−/fl;ColCre calvaria cells by real-time QPCR did not reveal significant differences in mRNA abundance for Runx2 and Osx mRNA, two factors important for osteogenic commitment, or for osteocalcin mRNA, a late osteoblast differentiation gene, relative to Cdh2+/fl cell cultures (Fig. 3B-D). We also compared in vitro osteoblast differentiation of calvaria cells from Cdh11−/− and Cdh2+/−;Cdh11−/− with cells isolated from calvaria of wild-type mice of the same mixed background. Consistent with a previous report (Kawaguchi et al., 2001a), we found reduced in vitro mineralization in Cdh11−/− calvaria cell cultures, relative to wild-type cultures, after 3 weeks in mineralization medium. Intriguingly, mineralization was not different in Cdh2+/−;Cdh11−/− double mutant calvaria cells relative to Cdh11-null cells, although the degree of mineralization was still lower than in wild-type cultures (Fig. 3E, upper panel). We also observed the presence of lipid containing cells in both Cdh11−/− and Cdh2+/−;Cdh11−/− calvaria cells cultured in the presence of adipogenic stimuli, when compared with very rare Oil red O-positive cells wild-type cultures (Fig. 3E, lower panel). The number of lipid containing cells was 9.7-fold and 13.2-fold higher in Cdh11−/− and Cdh2+/−;Cdh11−/− than in wild-type calvaria cells, respectively, in the presence of insulin and indomethacin, which stimulate adipogenic differentiation (Fig. 3E). Accordingly, we detected lower abundance of Runx2 and Osx mRNA in double mutant calvaria and Cdh11-null cells relative to wild-type cultures, although only Osx abundance was significantly lower in Cdh2+/−;Cdh11−/− relative to Cdh11−/− cells (Fig. 3F,G). The reason for the latter finding is unclear but it may reflect further impaired osteoblast commitment in the double mutants, which in the context of an already severely compromised osteogenesis may not impact on mineralization. Accordingly, expression of the late osteoblast gene, osteocalcin, was also profoundly, but equally, decreased in the two genotypes compared with wild-type cells (Fig. 3H). Conversely, PPARγ2 mRNA was increased in both mutant cells, though the difference was not statistically significant (Fig. 3I). There was no difference in cell number among the three genotypes during the proliferative, pre-confluence phase of calvaria cell cultures (not shown).
The modest differences in mineralization and osteoblast gene expression between single Cdh11−/− and double Cdh2+/−;Cdh11−/− mutant calvaria cells contrasts with the strikingly more severe osteopenia of the latter animals relative to the single mutants. However, we have previously shown that bone marrow stromal cell precursors are decreased in Cdh2 haploinsufficient mice (Fang et al., 2006). Therefore, we applied colony-forming unit assays to assess the number of stromal cell precursors in the bone marrow of mutant animals. The number of CFU-F, an index of stromal uncommitted precursors, originating from bone marrow cells of Cdh2−/fl;ColCre mice was significantly reduced compared with those isolated from wild-type mice. The number of CFU-O was also decreased, without apparent abnormalities of CFU-A number (Fig. 4A,B). This result was unexpected, considering that Cdh2 ablation is restricted to differentiated osteoblasts. Importantly, the number of CFU-F was significantly decreased in the double-mutant animals relative to both wild-type and single Cdh11−/− mice (Fig. 4C,D), as was the number of osteogenic precursors (CFU-O) in both Cdh11−/− and Cdh2+/−;Cdh11−/− mice relative to wild-type mice. Conversely, the number of adipogenic precursors (CFU-A) was significantly increased in both mutant genotypes (Fig. 4C,D). Thus, Cdh11 ablation leads to an osteogenic to adipogenic shift in stromal lineage commitment, whereas Cdh2 haploinsufficiency results in a reduced number of early precursors.
Altered junctional structures and cell-cell adhesion in cadherin-deficient cells
Cadherins interact with β-catenin and plakoglobin in the adherence junction, and β-catenin also serves as transcription factor, which is required for osteogenic commitment (Hill et al., 2005; Holmen et al., 2005). The abundance of nuclear β-catenin in calvaria cells isolated from conditionally deleted Cdh2−/fl;ColCre and haploinsufficient Cdh2−/fl mice was not appreciably different relative to wild-type equivalent cells, even though the cytosol/membrane pool was slightly decreased (Fig. 5A). However, we found a reduced abundance of β-catenin protein in both cytoplasmic/membrane and nuclear fractions in both single Cdh11−/− and double Cdh2+/−;Cdh11−/− mutant calvaria cells compared with wild-type cells, whereas the abundance of Tcf3/4 was unchanged in the nuclear fraction (Fig. 5B). Accordingly, less intense β-catenin-specific immunofluorescence was noted in Cdh11−/− and more so in Cdh2+/−;Cdh11−/− mutant calvaria cells relative to wild-type cells, with decreased signal intensity at both appositional membranes and in the nucleus (Fig. 5C).
Further experiments also revealed that serine/threonine phosphorylated β-catenin was actually increased in both cadherin mutant calvaria cells, relative to wild-type cells (Fig. 5D). As β-catenin phosphorylation at serine and threonine residues is the first step for ubiquitylation and degradation (Sadot et al., 2002), these results might suggest that cadherin deficiency leads to reduced β-catenin levels in part because of increased β-catenin degradation. Although to a lesser extent, we also found reduced abundance of plakoglobin, another component of adherens junctions, particularly in Cdh2+/−;Cdh11−/− mutant cells (Fig. 5D). To determine possible functional consequences of such decrease of β-catenin abundance, we assessed the expression of Cyclin D1, a β-catenin target gene. Consistent with a decrease in β-catenin transcriptional activity, the level of Cyclin D1 mRNA was significantly reduced in Cdh2+/−;Cdh11−/− cells relative to both wild-type and Cdh11-null cells, in which Cyclin-D1 mRNA was only marginally decreased (Fig. 5E).
Decreased β-catenin and plakoglobin is indicative of decreased adherens junctions, possibly leading to reduced cell-cell adhesion. To directly test this hypothesis, we assessed cell-cell adhesion between cells derived from the bone marrow and calvaria cells isolated from the same animal. Relative to Cdh2+/fl cells, we detected a lower number of Cdh2−/fl;ColCre bone marrow cells adherent to calvaria cells of the same genotype (Fig. 5F). The number of cells remaining adherent to the monolayer was lower in the Cdh11−/− background, but not significantly. However, decreased cell-cell adhesion was more pronounced and statistically significant when Cdh2+/−;Cdh11−/− bone marrow stromal cells were added to double mutant calvaria cells, with ~50% difference relative to control wild-type cells (Fig. 5G).
These studies demonstrate that both Cdh2 and Cdh11 contribute to postnatal skeletal growth and acquisition of normal bone mass, microarchitecture and strength. These two cadherins serve partially overlapping yet distinct functions; whereas Cdh11 is pro-osteogenic and anti-adipogenic, Cdh2 is involved in maintaining bone marrow stromal cell precursors. They both contribute to maintaining the organization of adherens junctions and β-catenin abundance.
The substantially more severe skeletal phenotype present in double mutant mice, which is characterized by smaller bones, osteopenia and reduced bone strength, relative to Cdh11-null mice demonstrates that N-cadherin partially compensates for loss of cadherin-11. Likewise, as Cdh2 haploinsufficiency does not cause major skeletal abnormalities, except in conditions of increased bone turnover (Fang et al., 2006), and as the phenotype of conditionally Cdh2 ablated mice is modest relative to Cdh2+/−;Cdh11−/− double mutants, the present results can also be interpreted to suggest that Cdh11 can provide compensatory action for Cdh2 deficiency. The conclusion that these two cadherins are partially redundant in regulating adult bone homeostasis is also in keeping with our previous data demonstrating that transgenic expression of a dominant-negative cadherin, which interferes with both N-cadherin and cadherin-11, results in low peak bone mass (Castro et al., 2004).
Only modest trabecular osteopenia has been reported in Cdh11−/− mice (Kawaguchi et al., 2001a), and we did not detect significant defects in whole body BMD over time. Lack of an in vivo phenotype contrasts with the obvious defective mineralization and decreased gene expression in Cdh11 null calvaria cells, a finding also reported by others (Kawaguchi et al., 2001a). However, neonatal calvaria cells represent bone cells during development, and such abnormalities may reflect a delay in mesenchymal transition into osteoblasts during ossification. In fact, the calvarium of Cdh11−/− mice is hypomineralized in young animals (Kawaguchi et al., 2001a); but as the skeleton of Cdh11−/− adult mice is essentially normal, such defects are compensated for postnatally. Accordingly, we find no abnormalities in the number of undifferentiated and osteogenic precursors among bone marrow stromal cells isolated from Cdh11−/−. Thus, cadherin 11 is dispensable for adult skeletal growth, even though it promotes osteogenic differentiation, as suggested by the differentiation defects of Cdh11-null calvaria cells, by enhanced precursor commitment to osteogenesis upon Cdh11 upregulation (Cheng et al., 2000; Kawaguchi et al., 2001b) and by the observation that Cdh2 null embryonic stem cells transfected with Cdh11 form teratomas that contain primarily cartilage and bone (Kii et al., 2004). However, lack of cadherin 11 enhances adipogenesis, as indicated by spontaneous emergence of adipocytic cells in calvaria cells, by the increased number of adipogenic precursors in the bone marrow of Cdh11-deficient mice and by the associated trend towards increased expression of PPARγ in Cdh11-null cells. This potential anti-adipogenic role of cadherin-11 deserves further study.
In vivo, the function of cadherin 11 may be in part compensated by N-cadherin. Selective Cdh2 ablation in osteoblasts/osteocytes results in modest trabecular osteopenia, without detectable abnormalities in static or dynamic indices of bone formation. It also leads to development of smaller diaphyses and decreased bone strength. More importantly, removal of only one Cdh2 allele in a Cdh11-null background produces an even worse phenotype, with a drastic reduction of bone formation rate and significant trabecular osteopenia in addition to smaller cortical bone and lower resistance to fractures. These genetic studies indicate that Cdh2 and Cdh11 interact and can partially compensate for each other in support of skeletal growth and maintenance in the adult animal.
The reduced bone mass and cortical phenotype of Cdh2 deficient mice apparently contrasts with recent findings demonstrating that overexpression of Cdh2 in osteoblasts in vivo results in osteopenia (Hay et al., 2009). In fact, the different genetic models help better clarify N-cadherin biology in osteogenic cells. The osteopenic phenotype of Cdh2-overexpressing transgenic mice was attributed to a persistent inhibitory effect of N-cadherin on osteoblast differentiation via interference with the Lrp5/β-catenin signaling pathway (Hay et al., 2009). However, the low bone mass observed in our Cdh2 conditionally deleted mice can be explained by the decreased availability of stromal cell precursors, which leads to a global deficiency of all stromal cell lineages. Notably, bone marrow stromal cell precursor numbers are decreased in Cdh2−/fl;ColCre mice and in double mutant mice relative to Cdh11−/− littermates, thus suggesting that this is a function specific to N-cadherin. Accordingly, we also reported a reduced osteoprogenitor number in Cdh2 haploinsufficient mice (Fang et al., 2006). Thus, one biological function of N-cadherin could be to maintain a normal pool of bone marrow stromal precursors in the adult bone marrow and a deficiency – even though partial – of osteogenic precursors may explain the slow but progressive age-dependent osteopenia we observed in all models of Cdh2 deficiency. At later stages of differentiation, downregulation of Cdh2 expression may be required to allow precursors to exit the progenitor pool and differentiate along the osteogenic lineage. Indeed, we have demonstrated that N-cadherin protein is downregulated as calvaria cells progress through their differentiation process (Fang et al., 2006); and others have shown that overexpression of N-cadherin in micromass cultures of embryonic cartilage prevents terminal chondrocyte differentiation, although it favors cell aggregation (Cho et al., 2003; DeLise and Tuan, 2002). Together, these data further strengthen the notion that N-cadherin must be downregulated to allow full chondro-ostegenic differentiation.
As Cdh2 deletion in Cdh2−/fl;ColCre mice occurs only in differentiated osteoblasts, the reduced number of stromal cell precursors must be the consequence of a non-cell-autonomous defect. This conclusion is in full agreement with our previous findings in transgenic mice expressing a dominant-negative form of N-cadherin, which resulted in an osteogenic to adipogenic shift in the numbers of bone marrow stromal cell precursors (Castro et al., 2004). Although the exact nature of such an indirect mechanism remains to be determined, one possibility is that osteoblasts may interact with stromal cell precursors via direct N-cadherin-mediated contact. Such a hypothesis is consistent with the findings that Cdh2 is necessary to maintain stem cells in Drosophila (Song and Xie, 2002); and cardiac stem cells in the adult mouse heart are in direct contact with myocytes via cadherins and connexins (Urbanek et al., 2006). Furthermore, downregulation of Cdh2 by c-myc results in loss of hematopoietic stem cells in mice, whereas upregulation of N-cadherin by c-myc deletion causes retention of stem cells in an undifferentiated state, resulting in defective hematopoiesis (Wilson et al., 2004). Our results suggest that osteoblasts can modulate the stromal cell compartment of the bone marrow, most likely via N-cadherin mediated ‘orthotypic’ cell-cell interactions (Mbalaviele et al., 2006), as suggested by our findings of decreased cell-cell adhesion between BMSC and calvaria cells of both double Cdh2+/−;Cdh11−/− mutant and conditionally Cdh2-ablated cells.
The decreased abundance of β-catenin and plakoglobin, crucial components of adherens junctions, and the decreased localization of β-catenin at cell-cell contact sites further support the notion that cell-cell adhesion is disrupted by cadherin deficiency in osteoblasts. Whether the decrease in nuclear β-catenin we found in Cdh2+/−;Cdh11−/− cells also interferes with canonical Wnt signaling remains an unresolved issue. Our results seem to point in this direction, as the changes in β-catenin phosphorylation in Cdh2+/−;Cdh11−/− cells, and to a lesser degree, in Cdh11−/− cells are consistent with β-catenin destabilization and increased degradation with a potential negative effect on signaling. More to the point, expression of Cyclin-D1, a known target of β-catenin transcriptional regulation (Tetsu and McCormick, 1999) is decreased in Cdh2+/−;Cdh11−/− cells. Finally, the increase in CFU-A and the presence of adipocytes in calvaria cells of cadherin-deficient mice is consistent with altered β-catenin activity, as canonical Wnt signaling is inhibitory of adipogenesis (Ross et al., 2000). It is now established that β-catenin is crucial for bone development and for the differentiation of bone-forming cells (Day et al., 2005; Hill et al., 2005; Hu et al., 2005), although in differentiated osteoblasts its main function seems to be primarily to regulate osteoclastogenesis (Glass et al., 2005). In addition to modulating β-catenin abundance and possibly signaling, cell-cell adhesion itself also contributes to the action of osteoblast cadherins. This is particularly true for N-cadherin, as we detected ~40% loss of cell-cell adhesive properties among Cdh2−/fl;ColCre cells, whereas nuclear β-catenin abundance was only marginally decreased. Our results might be interpreted to suggest that N-cadherin-mediated adhesion between osteoblastic cells and BMSCs contribute to maintain a normal pool of uncommitted bone marrow precursors. Differentiation cues converging on the osteoblast result in N-cadherin downregulation, with consequent loss of adhesion and release of undifferentiated cells to the osteogenic program. At the same time, β-catenin is released in the cytoplasm where it might be available for activation resulting in osteogenic and anti-adipogenic signals (Mbalaviele et al., 2006).
In summary, we have found that N-cadherin and cadherin 11 serve different, though partly overlapping, functions in the osteoblast differentiation program. Osteoblast N-cadherin is important in maintaining the pool of bone marrow progenitor cells, perhaps via cell-cell adhesion. Cadherin 11 is involved in osteoblast commitment and differentiation, and is antagonistic to adipogenesis, possibly by modulating β-catenin abundance and stabilization. However, our mouse genetic studies also suggest that these two cadherins can partially compensate for each other in modulating osteoblast differentiation in the adult skeleton. As the role of cadherins in the interactions between osteoblasts and bone marrow stromal cell precursors are elucidated, these molecules might become a potential therapeutic target to achieve bone anabolism.
Materials and Methods
Antibodies against N-cadherin and cadherin 11 have been described previously (Fang et al., 2006). All the other chemicals, including the tissue culture reagents were from Sigma (St Louis, MO, USA), unless otherwise indicated.
Cdh11-null mice have been described previously (Horikawa et al., 1999; Kawaguchi et al., 2001a), and were kindly provided by Dr Akira Kudo, Tokyo Institute of Technology, Japan. Cdh2-null mice were generated by one of the co-authors (Radice et al., 1997). To generate double Cdh2+/−;Cdh11−/− mice, we first mated Cdh11−/− mice with Cdh2+/− mice, and then crossed the resulting double heterozygous Cdh2+/−;Cdh11+/− with Cdh2+/+;Cdh11−/− mice, thus obtaining the desired genotype, Cdh2+/+;Cdh11−/− and Cdh2+/−;Cdh11−/−, in addition to Cdh2+/+;Cdh11+/− and Cdh2+/−;Cdh11+/−. The colony was maintained by mating Cdh2+/+;Cdh11−/− with Cdh2+/−;Cdh11−/− mice. Litter sizes were small (5-7 pups per litter), and we obtained ~27% Cdh2+/−;Cdh11−/− double mutants. This may reflect a lower viability of Cdh2+/−;Cdh11−/− embryos; and pups were slightly smaller than Cdh2+/+;Cdh11−/− littermates at birth.
Genotyping was performed by PCR in extracts of tail DNA using the following primer sets. For Cdh2, (forward) 5′-CAG CCA ATT GAC TTT GAA ACG and (reverse) 5′-GGC GAA TGA TTT TAG GAT TTG GGG; Neo, (forward) 5′-TCC TCG TGC TTT ACG GTA TC; and a cycling protocol previously detailed (Fang et al., 2006). This primer set generates 180 and 450 bp bands corresponding to wild-type and Cdh2 null alleles, respectively (Radice et al., 1997). For Cdh11, (forward) 5′-TTC AGT CGG CAG AAG CAG GAC and (reverse) 5′-GTG TAT TGG TTG CAC CAT GGG G; Neo, (forward) 5′-TCT ATC GCC TTC TTG ACG AGT TC; and a cycling protocol previously detailed (Kawaguchi et al., 2001a). This primer set generates 270 and 420 bp bands corresponding to wild-type and Cdh11-null alleles, respectively.
To produce osteoblast-specific Cdh2 ablation in vivo, we used mice harboring a Cdh2 ‘floxed’ allele (Cdh2fl), generated by one of the co-authors (GLR), and previously described (Kostetskii et al., 2005). Mice homozygous for this allele (Cdh2fl/fl) were mated with transgenic mice expressing bacterial Cre under the control of a 2.3 kb fragment of α1(I) collagen promoter (abbreviated as ColCre) (Dacquin et al., 2002; Kostetskii et al., 2005), also carrying one Cdh2-null allele (ColCre;Cdh2−/+), so that Cre-mediated recombination results in deletion of Cdh2 exon 1 (Kostetskii et al., 2005). This mating strategy avoids potential activation of Cre in the parental germ lines and generates approximately equal numbers of Cdh2 conditionally deleted mice (ColCre;Cdh2−/fl), wild-type equivalent (Cdh2+/fl), heterozygous equivalent (Cdh2−/fl) and conditional heterozygous (ColCre;Cdh2+/fl). The following primers were used to detect the Cdh2fl allele: L06, 5′-CCAAAGCTGAGTGTGACTTG-3′ and L08, 5′-TACAAGTTTGGGTGACAAGC-3′. They yielded one PCR product corresponding to Cdh2fl (290 bp) and a second product (250 bp) corresponding to either the wild-type or null allele. The latter two alleles were identified in a separate PRC reaction as described above. The ColCre transgene was detected by using the following primers: Cre 1123-1104, 5′-AAG TGC CTT CTC TAC ACC TG-3′; Cre 982-1002, 5′-TGC TTA TAA CAC CCT GTT ACG-3′; MS1, 5′-GCT CAG CAA GCT CAC AGC AA-3′; and LM6,5′-GAG CTT ACA CAT TTC GTC-3′. These primers generate 141 bp Cre-specific amplicon and a 448 bp Cre-negative amplicon (Chung et al., 2006). Each genotype was obtained at the expected Mendelian frequency.
Bone mineral density
Total body bone mineral density (BMD) was monitored by dual-energy X-ray absorptiometry (DXA) using a PIXImus scanner (GE/Lunar, Madison, WI), using procedures detailed elsewhere (Castro et al., 2004; Chung et al., 2006). Mice were injected with an anesthetic cocktail (ketamine 100 mg/kg and zylaxine 10 mg/kg i.p.), and after induction of anesthesia they were placed on the imaging-positioning tray in a prostrate position. Heads were excluded from the analysis by masking. The precision of this technique in our laboratory, assessed by the root mean square method is 1.34% for whole body BMD.
Bone histology and histomorphometry
Mice were euthanized under light anaesthesia by exsanguination through dorsal aortic puncture. Blood was collected and the serum stored at −70°C for later assays. Bone samples were prepared according to previously described methods, with some modifications (Chung et al., 2006). Briefly, dissected tibiae were fixed in 10% formalin and either decalcified in 7% EDTA for 14 days and embedded in paraffin, or left undecalcified and embedded in methyl methacrylate. Plastic sections were stained using the Masson trichrome technique, and tartrate resistant acid phosphatase activity stain was used for paraffin sections using the Acid Phosphatase leukocyte kit (Sigma). Calcein (15 μg/kg body weight i.p.) was injected 7 and 2 days before sacrifice for ColCre;Chd2−/+ and their control littermates. Because of the very low bone turnover rate of Cdh2+/−;Cdh11−/− mice, calcein labelling was performed 2 weeks apart in these animals, and three consecutive daily doses were given for each labelling course (days 13-11 and 4-2 before sacrifice). For all samples, 8 μm sections were left unstained for dynamic bone histomorphometry. Quantitative histomorphometry was performed in an area 175-875 μm distal to the growth plate using the OsteoMeasure software program (Osteometrix, Atlanta, GA) in an epifluorescence microscopic system. The following parameters of bone remodeling were estimated (Parfitt et al., 1987): trabecular bone volume as a percentage of total tissue volume (BV/TV), trabecular thickness (Tb.Th, in μm), trabecular number (Tb.N., per μm), trabecular separation (Tb.Sp. in μm), osteoblast perimeter per bone perimeter (Ob.Pm/B.Pm. in percent) and osteoclast perimeter per bone perimeter (Oc.Pm/B.Pm. in percent).
Before embedding, some samples (femur) were scanned using a μCT system (μCT-40, Scanco Medical, Zurich, Switzerland) in standard resolution function with an isotropic spatial resolution and slice increment of 16 μm. Femora were stabilized in 2.0% agarose gel and an area of interest was manually delineated on the selected sections to encompass as much trabecular bone as possible, excluding any cortical bone (as for the optical histomorphometry) and proximal to the growth plate. The bone volume fraction (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th) and trabecular separation (Tb.Sp) were measured for each specimen in 3D reconstructed image. For cortical bone analysis, four transverse scans were taken on the diaphysis at ±0.05 mm and ±1.5 mm from the midpoint of the femur at standard resolution, as described previously (Grimston et al., 2008). Radiation energy was 55 kVp. A total of six slices per scan were analyzed from each location and diaphyseal BMD was determined. Total tissue cross sectional area (Tt.Ar.) and marrow area (Ma.Ar.) were measured and cortical area (Ct.Ar.) calculated as Tt.Ar–Ma.Ar. Mean cortical thickness (Ct.Th.) was determined by measuring the cortical width at four different locations and taking the average value.
Mechanical testing in three-point bending to failure was conducted on the previously scanned femora. As a first step, the resistance to bending about the neutral axis was estimated as the second area moment of inertia (Ixx), as previously described (Silva et al., 2005). Briefly, the μCT images corresponding to each femur were imported into the ImageJ image analysis program (http://rsb:info.nih.gov/ij) and the x-y coordinates corresponding to the cross-sections of the diaphyses were used to calculate Ixx at an area corresponding to the axis of bending. For assessment of bone strength, specimens were stabilized over two supports placed 7 mm apart in an Instron 8500R apparatus (Instron, Canton, MA). A loading force was applied in the anterior/posterior direction midway between the two supports by displacement ramp at a rate of 0.03 mm/s. Force and displacement data were collected at 100 Hz (Labview 5.0, National Instruments, Austin, TX) and test curves were analyzed to determine ultimate force to failure (Silva and Ulrich, 2000).
Calvaria cell cultures
Osteoblast-enriched calvaria cultures were prepared from newborn mice by sequential collagenase digestion as described (Castro et al., 2004; Lecanda et al., 2000), and grown in α-modified essential medium (αMEM; Mediatech, Herndon, VA), supplemented with 10% fetal bovine serum (FBS; Atlanta Biologicals, Norcross, GA) and 100 IU/ml penicillin and 100 μg/ml streptomycin (Sigma Chemicals, St Louis, MO). Osteoblast-enriched calvaria cells were seeded at the same density for each genotype in 12-well culture plates, grown to confluence and incubated in 10% FBS α-MEM in presence of 50 μM ascorbic acid and 10 μM β-glycerophosphate for 2 or 3 weeks and subjected to Alizarin Red staining or Von Kossa staining as previously described (Lecanda et al., 2000). In some experiments, the number of viable cells in the culture was assessed after incubation in 10% FBS α-MEM (3-72 hours) by adding 3-4, 5-dimethylthiazol-2y-2,5 diphenyltetrazolium bromide (MTT) to 100 μl culture medium (0.5 mg/dl) at 37°C for 4 hours. After fixation in isopropanol, MTT absorbance was determined at 570 nm.
Bone marrow stromal progenitor cells
As previously described (Castro et al., 2004), the entire marrow cavity of the shafts (femora, tibia and humeri) of 3-month-old mice was flushed with 20 ml of αMEM (Mediatech, Herndon, VA) with penicillin and streptomycin using a 25-gauge needle. The material was then filtered through a 70 mm cell strainer (Falcon). After hemolysis in 0.8% NH4Cl, 0.017 M Tris-HCl buffer (pH 7.5), the resulting bone marrow cell suspension was rinsed in PBS and plated at 1-2.5×105 cells/cm2 for determination of the colony-forming unit: fibroblast (CFU-F), osteoblast (CFU-O) and adipocyte (CFU-A). At this seeding density, individual cells can be seen 24 hours after plating; and non-adherent cells were removed by repeated washing. For CFU-F cultures, cells were incubated in αMEM, containing 15% heat inactivated fetal calf serum (HI-FBS) and antibiotics. The number of CFU-F was determined after 2 weeks of culture, by fixation in absolute ethanol and staining with Giemsa. Colonies containing more than 50 cells were counted. Each experiment was performed in quadruplicate and the average was normalized to the number of cells seeded. For determination of CFU-A, cells were cultured in α-MEM supplemented with 15% HI-FBS, 50 nM insulin, 60 μM indomethacin for 14 days. The cultured media was refreshed every 3 days. After fixation in paraformaldehyde (PFA), the cultures were stained in a filtered solution of 0.3% Oil Red O in 60% isopropanol for 30 minutes. As for CFU-F, colonies containing more than 50 cells were counted. For CFU-O assessment, bone marrow cells were cultured in α-MEM supplemented with 15% HI-FBS, and containing 50 μM ascorbic acid and 10 μM β-glycerophosphate, with medium replacement every 3 days. Cells were cultured for 25-28 days, then fixed in 50% ethanol, 18% formaldehyde and stained for mineralized matrix using either Von Kossa or Alizarin Red stains, as described (Castro et al., 2004; Fang et al., 2006). Each experiment was performed in quadruplicate and the average was plotted as a function of number of cells seeded.
According to a previously described method (Cheng et al., 2000; Fang et al., 2006), single cell suspensions were obtained by trypsin/EDTA digestion, and labeled with the membrane-permanent fluorescent dye PKH26 (2 mM). These cells were added on top of a confluent unlabeled monolayer of the same genotype in the presence of 1 mM CaCl2, and allowed to settle for 60 minutes. The non-adherent cells were gently washed away with PBS and the number of fluorescent cells adherent to the cell substratum counted as an index of cell-cell adhesion.
Cells were counted and seeded on glass coverslips, grown at confluence in α-MEM with 10 % FBS, and then fixed in 4% (PFA)/PBS. The fixed cells were washed once with PBS, permeabilized with 0.3% Triton-X-100/PBS for 30 minutes, and blocked in 1% BSA, 0.3% Triton-X-100, 0.3% deoxycolate and 5% normal goat serum in PBS for 1 hour. The samples were incubated with β-catenin antibody (BD biosciences; 10 μg/ml in blocking buffer). After washing, bound antibodies were detected using 2 μg/ml of fluorescently labeled goat anti-mouse or anti-rabbit secondary antibody (Alexa fluor 488, Alexa fluor 568, Invitrogen); nuclei were counterstained with DAPI. The cells were then visualized and photographed using a Nikon Eclipse E600 microscope (Melville, NY).
Cytosolic and nuclear cell extracts were prepared according to procedures already described (Castro et al., 2004; Stains et al., 2003). For cytosolic fractions, cells were lyzed with a buffer containing 20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% NP-40 and protease inhibitors. For subcellular fractionation of β-catenin, cells were rinsed with cold PBS, scraped from the plate and centrifuged at 10,000 g for 5 minutes at 4°C. The pellets were gently resuspended in a hypotonic buffer [10 mM HEPES-KOH (pH 7.9); 10 mM KCl; 1.5 mM MgCl2; 1.5% NP-40 and 1 mM PMSF) and centrifuged again at 10,000 g for 5 minutes at 4°C to generate supernatants containing the cytosolic and membrane proteins. The pellets were used for nuclear extraction using a high-salt lysis buffer [(20 mM HEPES-KOH (pH 7.9); 1.22 mM MgCl2; 420 mM NaCl; 0.2 mM EDTA; 25% glycerol; 1 mM NaF; 1 mM Na2VO3; 0.5 mM DTT and 1 mM PMSF], and spun at 10,000 g for 5 minutes at 4°C to generate supernatants containing nuclear extracts. Total protein concentration was measured using Bio-Rad protein assay kit, and both fractions were separated by SDS-PAGE before transfer onto nitrocellulose membranes (Invitrogen, Carlsbad, CA). After immunoblotting with the appropriate antibodies, immune complexes were visualized by incubation with horseradish peroxidase-conjugated antibody and the ECL detection system (Amersham, Piscataway, NJ).
As previously described (Stains et al., 2003), total RNA was isolated from cells using the TRI Reagent (Sigma) and reverse transcribed (2 μg) using Superscript II reverse transcriptase and oligo(dT)15 primers. Real-time PCR analysis of mRNA was performed in a GeneAmp 5700 sequence detector system using the SYBR green PCR method according to manufacturer's instruction (Applied Biosystems, Foster City, CA). Primers used are detailed elsewhere (Mbalaviele et al., 2005; Stains et al., 2003). The mean cycle threshold value (Ct) from triplicate samples was used to calculate gene expression, and PCR products were normalized to GAPDH levels for each reaction. Relative gene expression levels were determined as described in User's Bulletin (P/N 4303859) from Applied Biosystems.
Group means were compared by t-test for unpaired samples. Data on repeated measured were analyzed by ANOVA. Data were analyzed using SPSS version 12.0.0 (SPSS, Chicago, IL). All data are expressed as the mean±s.d. (unless otherwise indicated).
This work was supported by National Institutes of Health grants R01 AR043470 and AR055913 (R.C.), and by Bridge Grants by the Endocrine Society and the American Society for Bone and Mineral Research. The authors thank the Clinical Nutrition Research Unit, Washington University (supported by P30 DK056341) for allowing us to use the PIXImus unit in the early phases of this project, and the Research Center for Auditory and Vestibular Studies Histology Core, Washington University (supported by P30 DC004665). Part of these results was reported, in abstract form, at the 27th and 28th annual meetings of the American Society for Bone and Mineral Research, Nashville, TN, September 23-27, 2005 and Philadelphia, PA, September 15-19, 2006. Deposited in PMC for release after 12 months.