Wnt signaling is an important pathway that critically controls tissue development by regulating cell growth, differentiation and survival during the prenatal and postnatal life.1 A variety of proteins control Wnt signaling, including ligands, soluble antagonists, receptors, membranous molecules, and intracellular proteins.2 Perturbations in the Wnt/β-catenin signaling components are associated with alterations in cell signaling and diseases, such as premature aging and cancers.3, 4 Notably, attenuation of Wnt signaling is associated with osteoporosis5, 6 and coronary artery disease.7 Therefore, targeting Wnt/β-catenin signaling represents an attractive approach in a variety of diseases.8–11 Several strategies have been developed to enhance Wnt/β-catenin signaling in osteopenic disorders.12, 13 Specifically, pharmacological modulation of intracellular Wnt signaling proteins proved to be effective in increasing bone mass in vivo.14, 15 However, this approach may be limited by the unknown consequences of abruptly promoting the Wnt/β-catenin signal transduction pathway. This limitation could be bypassed by targeting the signal transduction mechanisms that control Wnt signaling. In this context, molecules targeting extracellular Wnt antagonists were found to be effective in increasing bone mass.16–18 So far, molecules targeting transmembrane Wnt partners that could be applied for therapeutic approaches are not available.
Cadherins are transmembrane cell adhesion molecules that bind β-catenin at the juxtamembrane surface and contribute to regulate Wnt signaling in part by sequestrating β-catenin at the cell membrane.19 We previously established that N-cadherin plays additional role in the control of Wnt signaling via its interaction with the Wnt coreceptors LRP5/6 in osteoblasts.20 Here, by investigating the molecular interaction between LRP5/6 and N-cadherin, we identified specific intracellular domains in N-cadherin that interact with LRP5/6 to inhibit Wnt signaling. We show that truncated N-cadherin constructs that impair N-cadherin-LRP5/6 interactions promote Wnt/β-catenin signaling and osteoblast differentiation in vitro. Based on this knowledge, we show that disruption of N-cadherin-LRP5/6 interaction with a competitor peptide-based approach stimulates Wnt/β-catenin signaling, osteoblast function in vitro, and bone formation in vivo. These findings provide a competitor peptide-based strategy to target Wnt/β-catenin signaling in osteoblasts, which could serve as a basis for promoting bone formation.
Materials and Methods
Cell cultures and constructs
Murine MC3T3-E1 osteoblastic cells or L cells (ATCC) were transiently transfected with N-cadherin Flag-tagged20 or with N-cadherin constructs that were based on database analysis (ELM Motif Search: elm.eu.org), and established by PCR and then cloned in pcDNA 3.1. The Flag-Tag was added at the N-terminal extremity.
The peptide (DSCPPSPATERSYFHLFPPPPSPCTDSS) mimicking the identified LRP5 domain interacting with N-cadherin was synthesized on an Applied Biosystems Model 433A peptide synthesizer, using standard automated continuous-flow solid-phase peptide synthesis methods. Electrospray mass spectrometric sequence analysis was used to confirm the correct sequence. The corresponding tagged peptide was also obtained after incorporation of 6 histidine (His).The final peptides were obtained after reverse-phase chromatography purification. Peptide stability was examined in vitro. No degradation was found at pH 7 for 24 hours, albeit a 20% degradation was found at 48 hours and a 50% degradation occurred at 72 hours. In vitro and in vivo bioavailability of the peptides was ensured using a protein delivery reagent consisting of a cationic amphiphile and helper lipid (SAINT-PhD; Synvolux Therapeutics, BV, Groningen, Netherlands). In all experiments, cells were treated with the peptide or the delivery substance.
In vitro assays
Wnt3a-conditioned medium (CM) was prepared as described21 and used at 15% concentration. For luciferase reporter assays, 20 ng of pCMV β-Gal was added to the transfection mix (90 ng of TCF and 30 ng of TopFlash or FopFlash). Luciferase was determined using a Luciferase Assay Kit (Promega, Charbonierres, France), and β-Gal activity was evaluated with β-Gal Reporter Gene Assay (Roche, Indianapolis, IN, USA) at 24 hours. For analysis of cell replication, cells were plated at 2000 cells/dish in 96 wells, treated as indicated, and cell replication was determined at 24 hours using the BrdU ELISA assay (Amersham, Cell Proliferation Biotrak ELISA System, Les Ulis, France). In this assay, BrdU was added to the cells for 2 hours, cells were fixed, DNA was denatured by addition of fixative, peroxidase-labeled anti-BrdU was added, and the immune complexes were detected by spectrophotometry at 450 nm after subsequent substrate reaction. Absorbance values that correlate directly to the amount of DNA synthesis were expressed as optical density (OD) units and corrected for total cells evaluated by crystal violet. ALP activity was assayed using an alkaline phosphatase kit (Bio-Rad, Hercules, USA) at 48 hours. ALP staining was performed using Sigma Fast BCIP/NBT kit (Sigma, St Quentin Fallavier, France) at 48 hours. For in vitro mineralization, cells were cultured in the presence of 50 µg/mL ascorbic acid and 3 mmol/L inorganic phosphate (Sigma) and the mineralized extracellular matrix was monitored by alizarin red staining (Sigma) at 3 weeks. Cell apoptosis was assayed by Tunel staining (Millipore, Chemicon, Tamecula, CA, USA) at 24 hours using the Apoptose Tag Kit (Chemicon USA) according to manufacturer's recommendations. The number of Tunel-positive cells was expressed as percent of total cells. Quantitative real-time PCR analysis of total RNA was performed as described20 using reported primers and GAPDH as control.
For immunocytochemistry, cells were fixed with 3.7% formaldehyde (Sigma) for 10 min, washed twice with phosphate-buffered saline, permeabilized with 0.025% Triton X-100 (Sigma) for 5 min and blocked with phosphate-buffered saline, 3% bovine serum albumin for 15 minutes. Cells were incubated overnight at 4°C with anti-β-catenin (Santa Cruz, CA, USA), anti-N-cadherin (AbCam, Cambridge, MA, USA), or anti-HIS (AbCam, USA), used at 1:100 dilution, and incubated with a goat anti-rabbit conjugated to fluorescein isothiocyanate (Beckman Coulter, Villepinte, France), or a goat anti-mouse conjugated to rhodamin (Beckman Coulter). Cover glasses were viewed using apotome fluorescence microscopy (Carl Zeiss, Jena, Germany). For Western blot analysis, cells were frozen in liquid nitrogen and incubated at +4°C in a buffer containing 50 mM Tris (pH 7.4), 150 mM NaCl, 2 mM Na3VO4, 1% protease inhibitor cocktail (Sigma), 1% Nonidet P-40, 1% sodium deoxycholate, and 0.1% SDS for 30 minutes. Lysates were then cleared by centrifugation at 6000 g for 30 minutes. A 25 µg sample of lysate was diluted in reducing sample buffer containing 125 mM Tris buffer (pH 6.8), 4% SDS, 20% glycerol, 0.05% bromophenol blue, and 200 mM dithiothreitol. The mixture was then heated at +90°C for 5 minutes, and subjected to gel electrophoresis on 8% or 12% gels. After SDS-PAGE electrophoresis, proteins were transferred to Immobilon P membranes and immunoblotted with specific antibodies, as previously described.20 Immunoprecipitation analyses were performed as described20 using anti-Flag (Sigma), anti-axin (Zymed, San Francisco, CA, USA), anti-N-cadherin (AbCam), anti-β-catenin (Santa Cruz), anti-phospho-β-catenin (catalog no. 9561, Cell Signalling, Denver, CO, USA), anti-Glycogen Synthase Kinase 3 (GSK3) (Cell Signalling), anti-HIS (AbCam), anti-LRP5 (Zymed catalog no. 36-5400), or anti-LRP6 (Cell Signalling) used at 1:1000 and detected with a secondary horseradish peroxidase antibody (Vector or Beckman Coulter, Fullerton, USA). In some experiments, N-cadherin or axin siRNA were silenced using siRNA (2 µg/30,000 cells; Santa Cruz Biotechnology, Santa Cruz, CA, USA), as previously described.20
Immunohistochemical and immunohistofluorescent analysis was performed on paraffin-embedded histological sections of tibia metaphysis or mouse calvaria. Immunohistochemical was performed using the Vectastain Elite ABC Kit (Vector, Abcys, Paris, France). Briefly, after paraffin removal, sections were incubated in citrate buffer at 90° for 30 minutes for antigenic retrieval and treated with hyaluronidase (1 mg/mL) at 37° for 15 minutes. Endogenous peroxydase was inhibited by incubating the tissue section in 0.3% H202 for 15 minutes. Tissue sections were incubated with appropriate serum during 1 hour before primary antibody incubation (1 hour) using anti-LRP5 (Zymed), anti-N-cadherin (Santa Cruz), or anti-β-catenin (Santa Cruz) used at 1:100 dilution, or Tunel labeling (Chemicon), and revealed according to the manufacturer's instructions.
In vivo studies
Experiments were approved by the local institutional ethical board Lariboisière-Villemin (no. CEEALV/2010-04-03). Tibias from 1.5-month-old female N-cadherin transgenic and wild-type (WT) mice were obtained, as previously described.20, 22 For in vivo injection experiments, SWISS mice (Charles River) aged 6 weeks (21 +/−1.8 g b.w.) were injected with the 28 AA peptide (10 µg) dissolved in a protein delivery reagent (SAINT-PhD, Synvolux Therapeutics, Groningen, The Netherlands) or with the protein delivery reagent in a small volume (250 µl), twice a week for 3 weeks, in order to allow transient instead of permanent stimulation of Wnt signaling. BMP2 (150 ng) was administered 5 times a week for 1 week to allow early stimulation of osteoblast precursor cell differentiation and to avoid possible induction of apoptosis in more mature differentiated osteoblasts.23 These mice were used as positive control animals. All mice were injected calcein (10 mg/Kg) and tetracycline (20 mg/Kg) at 4 and 1 days, respectively, before sacrifice to ensure double labeling of the new bone matrix formed.20 In vivo new bone formation was determined by histomorphometric analysis on 5-µm-micron-thick sections of sagittal sections of the parietal bone embedded in methyl methacrylate or in paraffin after demineralization for immunohistochemistery. Sections obtained with a Leica microtome (SM2500S) (Wetzlar, Germany) were stained with toluidine blue or left unstained for fluorochrome evaluation. The mineral apposition rate (MAR) was measured using image analyzer (Biocom, Les Ulis, France) on double-labeled surfaces. The mineralizing surfaces (MS) were measured in the same area using the objective eyepiece Leitz integrate plate II, and the bone formation rate (BFR) was derived from the product of MAR by MS. All readings were performed without knowledge of the treatment.
The triplicate experiments refer to three independent experiments with each experiment performed in triplicate wherever applicable. Data are the mean +/– SD and were analyzed by Student's test with a minimal level of P < 0.05 considered to be significant.
Identification of N-cadherin/LRP5/6 domains of interactions
We used in vitro and in vivo strategies to identify targetable domains of N-cadherin that interact with LRP5 to negatively control Wnt signaling. Using immunohistochemistry, we showed that N-cadherin colocalizes with LRP5 in long bone from WT mice (Supplementary Fig. S1). This was observed even more clearly in long bone from mice overexpressing N-cadherin in osteoblasts under the control of the type-1 collagen promoter24 (Supplementary Fig. S1). These results indicate that N-cadherin interacts with LRP5 in vivo. Based on these results, we developed a molecular strategy to identify the domains of interaction between N-cadherin and LRP5 that could be targeted to promote Wnt/β-catenin signaling. We designed N-cadherin constructs (Supplementary Fig. S2A) to investigate the role of putative domains of interaction with axin, casein knase I and GSK3 (Δ62 construct), p53-MDM2 (Δ114 construct), or MAPK and cyclin D1 (Δ153 construct) (Supplementary Fig. S2B). We showed that osteoblasts transfected with these constructs express the corresponding truncated N-cadherin domain, as revealed by Western blot analysis (Supplementary Fig. S2C) and immunocytochemistry (Supplementary Fig. S2D). We next tested the ability of these constructs to disrupt N-cadherin-LRP5 interaction. Immunoprecipitation analysis showed that deletion of the minimum sequence of 62 AA in the cytoplasmic tail of N-cadherin led to disruption of the N-cadherin-LRP5 interaction in MC3T3-E1 osteoblastic cells (Fig. 1A). We also investigated whether the identified N-cadherin domain that interacts with LRP5 may interact with LRP6, another functional Wnt coreceptor that is widely expressed.2, 25 We found that the Δ62 N-cadherin construct cannot bind to LRP6, confirming the results with LRP5 (Fig. 1A). These results indicate that the N-cadherin-LRP5/6 interaction requires the 62 last AA in the cytoplasmic tail of N-cadherin.
Functional effects of truncated N-cadherin domains in osteoblasts
Based on the above data, we hypothesized that targeting the 62 AA in the cytoplasmic tail of N-cadherin may lead to inhibit its interaction with LRP5/6 and to subsequent activation of Wnt/β-catenin signaling. We first investigated the biochemical and functional effects of truncated N-cadherin domains in MC3T3-E1 cells. Activation of Wnt signaling leads to inhibition of β-catenin phosphorylation, its stabilization and translocation into the nucleus where it activates gene expression by T cell factor (TCF) lymphoid-enhancing factor (LEF) transcription factors.3 As expected, Wnt3a increased TCF/TOP transcriptional activity, whereas overexpression of full-length N-cadherin had a negative effect (Fig. 1B). We found that deletion of the 62 AA cytoplasmic domain of N-cadherin fully abrogated the negative effect of N-cadherin on β-catenin transcriptional activity (Fig. 1B). These results indicate that the 62 AA cytoplasmic domain in N-cadherin that interacts with LRP5/6 contributes to the negative effect of N-cadherin on canonical Wnt signaling in osteoblasts.
Based on this finding, we analyzed whether targeting the 62 AA cytoplasmic domain in N-cadherin may have functional consequences on osteoblast proliferation and function. As expected from its negative interaction with LRP5,22 overexpression of full-length N-cadherin decreased cell replication. We found that this negative effect was fully abrogated by deletion of the 62AA N-cadherin domain (Fig. 1C). We also found that the negative effect of full-length N-cadherin on alkaline phosphatase (ALP) activity, an early marker of osteoblast differentiation, was no longer observed in cells transfected with the Δ62 N-cadherin construct (Fig. 1D). Consistent with this finding, quantitative RT-PCR analysis showed that deletion of the 62AA domain markedly increased expression of Runx2, ALP. and type I collagen that are phenotypic osteoblast genes26 (Fig. 1E). This is in contrast with full-length N-cadherin, which was shown to decrease osteoblast gene expression in these cells.20 These results show that the inhibitory effect of full-length N-cadherin on Wnt signaling and osteoblast gene expression is abrogated by deletion of the 62 AA domain in N-cadherin. This identifies the 62 AA cytoplasmic N-cadherin intracellular domain as an important determinant involved in the N-cadherin-LRP5/6 interaction that inhibits Wnt/β-catenin signaling and osteoblast gene expression, providing a target that could be used for modulating Wnt signaling.
A competitor peptide mimicking the LRP5/6 domain that interacts with N-cadherin promotes Wnt signaling
Based on the above knowledge, we investigated whether targeting the 28 AA LRP5/6 domain that interacts with N-cadherin20 could abrogate its interaction with the 62 AA cytoplasmic N-cadherin intracellular domain, and subsequently promote Wnt/β-catenin signaling and osteoblast function. To investigate this concept, we developed a peptide-based strategy aimed at mimicking the LRP5 domain that interacts with N-cadherin using MC3T3-E1 cells that express high levels of N-cadherin,20 as do normal osteoblasts in vivo (Supplementary Fig. S1). We produced a peptide encompassing the 28 last AA in LRP5 (DSCPPSPATERSYFHLFPPPPSPCTDSS) that could, therefore, be used as a competitor. The sequence is closed to the 28 last AA in LRP6 (ESCPPSPYTERSYSHHLYPPPPSPCTDSS). We observed that, in the presence of a cationic amphiphile and helper lipid (SAINT-PhD) used as efficient protein delivery reagent, the 28 AA peptide dose-dependently entered cells within the cytosol in MC3T3-E1 osteoblastic cells with a maximal effect at 10 µg/mL based on intensity of immunofluorescent staining (Fig. 2A). We, therefore, used this dosage in all subsequent experiments. Immunoprecipitation analysis showed that the 28 AA peptide physically interacts with N-cadherin in MC3T3-E1 cells (Fig. 2B). In cells cultured in the presence of the 28 AA, deletion of the intracellular 62 AA cytoplasmic N-cadherin domain did not allow binding to the 28 AA peptide (Fig. 2C), confirming that the 28 AA peptide interacts with the 62 last AA in N-cadherin. The interaction between the 62 AA N-cadherin intracellular domain and the 28 AA peptide was functional, because the peptide specifically reduced N-cadherin/LRP5 and N-cadherin/LRP6 interactions in these cells (Fig. 2D). Thus, the designed 28 AA peptide that interacts with the 62 AA cytoplasmic N-cadherin intracellular domain specifically competes with the last 28 AA in LRP5/6 binding in osteoblasts. This competitor-mediated interaction results in abrogation of N-cadherin/LRP5/6 interaction, a finding that could be used to modulate Wnt/β-catenin signaling for therapeutic intervention.
Based on the evidence that the 28 AA peptide functionally targets N-cadherin/LRP5/6 interactions in osteoblasts, we examined the biochemical effect of this peptide on components of the Wnt pathway that act downstream of activated LRP5/6. We found that the 28 AA peptide dramatically reduced phosphorylation of β-catenin, which is required for its proteasomal degradation, in MC3T3-E1 cells. Consequently, we observed that the 28 AA peptide increased β-catenin protein levels in osteoblasts at 24 hours (Fig. 3A), which resulted in increased β-catenin staining in the nucleus (Fig. 3B). Quantification showed that the 28 AA peptide increased by 2-fold the number of cells with a positive nuclear β-catenin staining (32.4 versus 15.6% in control cells). Consistent with this effect, the 28 AA peptide-induced β-catenin nuclear translocation was associated with increased TCF transcriptional activity in the absence or presence of Wnt3a ligand (Fig. 3C). The competitor peptide had no effect on TCF/LEF transcriptional activity in L cells that do not express N-cadherin, whereas it was active in L cells transfected with full-length N-cadherin (Fig. 3D), indicating that activation of Wnt/β-catenin signaling by the 28 AA peptide is dependent on its interaction with N-cadherin. LRP5 interacts with axin via axin-binding sites in the 1–28 AA domain of LRP5.27 Accordingly, we observed that the 28 AA peptide interacts with axin, but not with GSK3, as shown by immunoprecipitation analysis (Fig. 3E, F, G), suggesting that a tripartite molecular complex between the 28 AA, N-cadherin, and axin concurs to the observed peptide-mediated increase in Wnt/β-catenin signaling. To confirm this concept, we used axin siRNA to reduce axin levels in MC3T3-E1 cells. Efficient reduction of axin protein levels using siRNA (Supplementary Fig. S3A, single arrow), markedly decreased the amount of peptide associated with N-cadherin (Supplementary Fig. S3A, double arrow). Quantification of the scanned blots obtained from three separate experiments showed that reducing axin level by about 50% significantly decreased the amount of 28 AA peptide associated with N-cadherin, further indicating that axin is involved in the molecular interaction between the 28AA peptide and N-cadherin (Supplementary Fig. S3B). Because the β-catenin and GSK3 binding sites are located within the 62 AA C-terminal region of N-cadherin, it appears that the 28 AA peptide binding to N-cadherin blunts its interaction with β-catenin, resulting in decreased β-catenin sequestration, increased β-catenin phosphorylation, and degradation. This does not rule out, however, the possibility that activation of β-catenin signaling by the peptide may also be due to effects independent of its association with N-cadherin.
Functional effects of the competitor peptide on osteoblast differentiation
To determine whether the 28 AA peptide-mediated activation of Wnt/β-catenin signaling may translate into functional cellular activity in vitro, we tested the activity of the peptide on osteoblast differentiation controlled by the Wnt/β-catenin pathway.28 As expected, Wnt3a increased ALP staining (Fig. 4A) and activity (Fig. 4B). The 28 AA peptide also increased ALP staining and activity in the presence or absence of the ligand Wnt3a (Fig. 4A, B). Furthermore, the 28 AA peptide increased in vitro matrix mineralization in the presence or absence of Wnt3a, as revealed by alizarin red staining and quantification of calcium content, which typifies matrix mineralization (Fig. 4C, D). Consistent with an increased osteoblast function, the 28 AA peptide increased the expression of the phenotypic osteoblast marker genes Runx2, ALP, and type 1 collagen, and the late marker osteocalcin. Remarkably, with the exception of osteocalcin, the stimulatory effect of the 28 AA peptide was similar to the effect of Wnt3a (Fig. 4E). We also observed that reducing N-cadherin expression using siRNA decreased the positive effect of the 28AA peptide on osteoblast gene expression (Supplementary Fig. S3C). These results indicate that N-cadherin/peptide interaction is necessary for peptide activity.
Because osteoblast differentiation is known to be controlled by cadherin-mediated cell-cell adhesion,29, 30 we examined whether the positive effect of the 28 AA peptide on cell differentiation may be related to changes in cell-cell adhesion. We observed that the 28 AA peptide did not affect cell-cell adhesion in a standard cell aggregation in vitro assay (Supplementary Fig. S4A). In contrast, overexpression of full length or the Δ62 AA truncated N-cadherin intracellular domain increased cell-cell adhesion (Supplementary Fig. S4B). This indicates that the 28 AA competitor peptide increases osteoblast differentiation marker expression independently of N-cadherin-mediated cell-cell adhesion, which is consistent with its interaction with the intracellular N-cadherin domain.
Effects of the competitor peptide on osteoblast proliferation and survival
Constant activation of Wnt signaling is often associated with activated cell replication and cancer onset and progression.31 It is, therefore, critical to test whether the 28 AA peptide that we found to activate Wnt/β-catenin signaling in osteoblastic cells does not impact cell replication. We observed that the addition of Wnt3a increased cell replication in MC3T3-E1 osteoblastic cells, as shown by bromodeoxyuridine (BrdU) incorporation at 24 hours. In contrast, the 28 AA peptide had no effect on cell replication (Fig. 5A), indicating that the positive effect of the 28 AA peptide on bone cell function is not hindered by exaggerated cell proliferation. Another typical effect of Wnt proteins is reduction of cell apoptosis, which is mediated in part by activation of noncanonical Wnt signaling.32 Because we previously found that N-cadherin/LRP5 interaction controls osteoblast survival via attenuation of ERK and PI3K/Akt signaling,22 we tested the effect of the 28 AA peptide on these signaling pathways. The addition of the peptide slightly increased phosphorylated p-PI3K and p-ERK MAPK levels but had no effect on p-Akt levels at 30 minutes, indicating absence of significant activation of this pathway by the competitor peptide (Fig. 5B). Cell survival is in part under the control of ERK1/2 signaling, and we observed that the addition of the 28 AA peptide reduced osteoblast apoptosis induced by serum deprivation, as indicated by Tunel analysis and quantification at 24 hours (Fig. 5C, D). These data indicate that the peptide has no significant effect on cell replication and that the effectiveness of the compound on osteoblastogenesis in vitro results mainly from increased cell differentiation and survival.
The competitor peptide promotes bone formation in vivo
In order to test the activity of the competitor peptide on bone tissue formation, we used an established in vivo assay allowing rapid evaluation of anabolic agent effectiveness in mice.33 Consistent with our finding that the 28 AA peptide activates Wnt/β-catenin in vitro, we found that subcutaneous injection of the peptide (40 µg/mL) increased β-catenin nuclear staining in cells on the surface of mouse cranial bone compared with the injected delivery substance (Fig. 6A), indicating that the 28 AA peptide was effective in inducing β-catenin translocation in vivo. We next examined whether peptide-mediated activation of Wnt/β-catenin signaling translated into changes in cellular function and de novo tissue formation. We initially observed that the 28 AA peptide had no effect on osteoblast replication, as indicated by Ki67 staining (data not shown), supporting the lack of effect of the peptide in vitro. Similarly to our in vitro findings (Fig. 5C), we found that the 28 AA peptide decreased Tunel staining in vivo, which was confirmed by quantitative analysis (Fig. 6B, C), indicating reduced osteoblast apoptosis. Remarkably, the local injection of the 28 AA peptide increased bone matrix apposition that typifies osteoblast activity, as shown by the mean distance between double-labeled surfaces documenting active bone formation (Fig. 6D). Quantification of de novo bone formation revealed that the competitor peptide increased the matrix appositional rate (MAR), mineralizing surface, and bone formation rate (Fig. 6E). As a result, we observed that the local injection of the 28 AA peptide increased de novo bone tissue formation compared with the injected delivery substance (Fig. 6F). Quantitative analysis of the average distance between the new bone formed and the bone surface showed that the anabolic effect of the 28 AA peptide on de novo tissue formation was closed to that induced by the local injection of BMP2, a standard anabolic agent promoting bone tissue formation34 (Fig. 6G). The effectiveness of the 28 AA peptide in vivo supports the concept that peptide-mediated disruption of N-cadherin-LRP5/6 interaction by a competitor peptide can promote Wnt/β-catenin signaling, osteoblast function and de novo bone tissue formation (Fig. 6H).
The identification of molecules that can promote the Wnt/β-catenin signaling pathway is an important issue for enhancing bone formation in osteopenic disorders. We previously showed that that N-cadherin inhibits Wnt signaling in osteoblasts, in part via its interaction with the Wnt coreceptors LRP5/6. In this study, we identified the molecular and functional interactions between N-cadherin and LRP5/6, and we report a strategy targeting the intracellular domain of N-cadherin that results in increased Wnt signaling, osteoblast function and bone formation. We first identified the N-cadherin intracellular domain that interacts with LRP5/6 and thereby inhibits Wnt/β-catenin signaling. We found that the N-cadherin-LRP5/6 interaction requires the 62 last AA in the cytoplasmic tail of N-cadherin. Using this knowledge, we hypothesized that deletion of this domain can abrogate the negative interaction between N-caherin and LRP5/6 in osteoblasts. Our finding that a Δ62 AA truncated N-cadherin construct promotes Wnt/β-catenin signaling and osteoblast differentiation supports the concept that the 62 last AA in the cytoplasmic tail of N-cadherin is involved in the negative impact of N-cadherin on Wnt signaling via its interaction with LRP5/6. This finding therefore provides a possible target for molecular intervention to promote Wnt/β-catenin signaling and osteoblast function.
Based on the above results, we developed a competitor peptide-based approach to target N-cadherin-LRP5/6 interaction. Our previous data revealed that N-cadherin binds to the 28 last amino acids of LRP5.20 Based on this finding, we tested whether a small competitor peptide containing the 28 last amino acids of LRP5 may functionally interact with the 62 last AA in N-cadherin in osteoblasts. We found that the peptide efficiently binds to the 62 AA N-cadherin domain that interacts with LRP5. We also show that the peptide displaced the natural interaction of N-cadherin with LRP5/LRP6 in osteoblastic cells, suggesting that the peptide acts as an effective competitor molecule. Having shown that the peptide functionally abrogates the interaction between N-cadherin and LRP5/6 receptors, we analyzed its functional impact on Wnt signaling. Our results show that the competitor peptide increased β-catenin translocation and transcriptional activity and abolished the negative impact of N-cadherin on Wnt signaling, indicating that disruption of the LRP5-N-cadherin interaction using this peptide is sufficient to promote Wnt/β-catenin signaling in osteoblastic cells. We further analyzed the mechanisms by which the peptide may activate Wnt signaling in these cells. Cadherins are known to interact with β-catenin and to affect its signaling activity.19, 35 Additionally, we showed that N-cadherin may modulate Wnt signaling independently of β-catenin sequestration.20 Our finding that the 28 AA peptide interacts with axin, but not with GSK3, suggests that the formation of a tripartite molecular complex between the peptide, N-cadherin, and axin leads to abrogate the negative impact of N-cadherin-LRP5/6 interaction on Wnt signaling. We cannot rule out, however, the possibility that the efffect of the peptide could also be indirect and mediated through the axin protein. The mechanism by which the competitor peptide promoted Wnt signaling in osteoblastic cells differs from previous approaches using neutralizing antibodies or inhibitors of endogenous Wnt antagonists.16–18 The peptide-based strategy described here may thus offer an alternative mode of Wnt pathway stimulation by a small peptide competitor for promoting Wnt/β-catenin signaling.
Having shown that disrupting N-cadherin/LRP5/6 interactions through the use of the designed competitor results in increased Wnt signaling in osteoblastic cells, we examined whether this effect may be functional in promoting osteoblast replication or function. Our finding that the peptide increased osteoblast gene expression and matrix mineralization and induced cell protection against apoptosis in vitro indicates that the peptide is efficient in promoting osteoblastogenesis as a result of increased cell differentiation and survival. One important question was to determine whether the competitor peptide may activate bone cell function without altering cadherin-dependent cell-cell adhesion. We showed that the 28 AA peptide activates osteoblast differentiation marker expression independently of N-cadherin-mediated cell-cell adhesion, which is consistent with the concept that cadherins may induce signaling in the absence of cell-cell contacts.36 Another important issue was to determine whether the competitor peptide-based strategy may be efficient to promote bone formation in vivo. We found that the local injection of the 28 AA peptide increased β-catenin nuclear translocation and promoted bone formation rate in vivo. This anabolic effect appears related to increased bone-forming activity of osteoblasts and reduced osteoblast apoptosis, which is in accordance with the observed effect of the peptide on osteoblastic cells in vitro. Aberrant activation of the Wnt/β-catenin pathway is often associated with abnormal proliferation in cancer and other diseases.31, 37, 38 Here we found that the 28 AA peptide-mediated activation of Wnt/β-catenin and bone formation was not associated with deregulation of cell replication in vitro or in vivo. Overall, the results support the concept that disruption of the N-cadherin-LRP5/6 interaction by the competitor peptide results in activation of Wnt/β-catenin signaling, resulting in increased osteoblast function and survival and bone tissue formation in vivo (Fig. 6H). Given the high level of N-cadherin in osteoblasts,29, 30 these results strengthen the importance of targeting N-cadherin/LRP5/6 interaction to promote bone formation. Interestingly, we found that the peptide competitor promoted osteoblast function and bone tissue formation in vivo almost as efficiently as BMP2, a molecule that is widely used to promote bone tissue repair.34 We, therefore, propose that this competitor peptide-based approach could be of therapeutic interest for promoting bone formation in conditions such as aging, where intrinsic osteoblast function is compromised39 and Wnt signaling is altered.40 Additionally, this approach may provide a more general basis for the development of small molecules that could be therapeutically used in severe chronic diseases characterized by impaired Wnt signaling.7
All authors state that there are no conflicts of interest.
This work was supported in part by grants from the European Commission FP6 and FP7 research funding programs Anabonos (LSHM-CT-2003-503020) and Talos (HEALTH-F2-2008-201099). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Authors' roles: EH and PJM were involved in conception and design; TB, CM, SDaN, and PS did data acquisition; and EH, PS, and PJM drafted and revised the manuscript.