Decreased trabecular, but increased cortical, bone-mass in Atp6ap2 Ocn-Cre mice

To investigate ATP6AP2’s function in bone, we generated OB-selective ATP6AP2 conditional knockout (CKO) mice, Atp6ap2Ocn-Cre, by crossing Atp6ap2flox/X with osteocalcin (Ocn)-Cre mice (Fig. S1a), whose cre expression is under the control of the human bone gamma carboxyglutamate (BGLAP) promoter/enhancer, and thus largely in the OB-lineage cells, including bone marrow stromal cells (BMSCs), OBs, and osteocytes.2,10,32,33 As expected, ATP6AP2 protein levels were selectively reduced in the OB-lineage, but not the OC-lineage, cells in Atp6ap2Ocn-cre mice, as compared with those of littermate control mice (Ocn-Cre mice) (Fig. S1b–e). Atp6ap2Ocn-cre mice showed reduced body size and body weight, started as early as P15 (postnatal day 15) (Fig. S1f, g), implicating a deficit in mouse skeleton development.

We further examined femur structures in control and Atp6ap2Ocn-cre mice by microcomputer tomographic (μCT) analysis. Decreases in trabecular bone volumes over total volumes (BV/TV), trabecular bone number (Tb. N) and trabecular bone thickness (Tb.TH), and increases in trabecular bone space (Tb. Sp), and cortical bone volumes/total volumes (Cb, BV/TV) were all detected in Atp6ap2Ocn-Cre mice (3-month), as compared with those of control littermates (Ocn-Cre mice) (Fig. 1a–f). The trabecular bone deficit was further verified by H&E staining analysis (Fig. 1g–i). H&E staining analysis of the femur bone sections showed a marked reduction in the trabecular bone-mass and an increase in bone marrow fat as early as 1-month in Atp6ap2Ocn-cre mice (Fig. 1g–i). These results demonstrated a trabecular, but not cortical, bone-loss, and suggest the necessity of osteoblastic ATP6AP2 in trabecular bone homeostasis.

Fig. 1

Reduced trabecular, but increased cortical, bone-mass in Atp6ap2Ocn-Cre mice. a Representative μCT 3D images of femurs from 3-month male ctrl and Atp6ap2Ocn-Cre mice littermates. b–f Quantification analyses of trabecular bone (Tb) volumes over total volumes (BV/TV), Tb number (Tb.N), Tb thickness (Tb.TH), Tb space (Tb. SP) and cortical bone (Cb) volumes over total volumes (BV/TV) by direct model of μCT analysis. g Representative images of H&E stained femur sections from male ctrl and Atp6ap2Ocn-Cre mice at ages of 1-month and 3-month. Bar, 150 μm. h, i Quantification analyses of trabecular bone (Tb) volumes over total volumes (BV/TV) and bone marrow fat number in (g). Data in (b–f), (h) and (i) are shown as box plots together with individual data points, and whiskers indicate minimum to maximum (n = 4 or 6 male mice of each genotype). P values obtained by unpaired two-tailed t test. *P < 0.05. **P < 0.01. ***P < 0.001

Enhanced OC formation and bone resorption in Atp6ap2 Ocn-Cre trabecular and cortical bone regions

The bone-mass is controlled by both bone formation and bone resorption.34,35 We thus examined osteoclastic bone resorption in control and Atp6ap2Ocn-Cre mice. Measuring serum levels of deoxy-pyridinoline (PYD, a marker for bone resorption) demonstrated elevated PYD levels, suggesting an increased bone resorptive activity in the mutant mice at the ages of 1-month and 3-month (Fig. S2a). We then carried out TRAP staining analysis of OCs, which showed an increase in the number per unit of both trabecular and cortical bone surface in both 1-month and 3-month Atp6ap2Ocn-Cre mice (Fig. S2b–d). We next asked whether ATP6AP2 regulates OC formation and resorptive activity in vitro by examining M-CSF- and RANKL-induced OC differentiation from BMMs derived from Atp6ap2Ocn-Cre and control mice. More TRAP+ MNCs (multi-nuclei cells) were detected in BMM cultures from Atp6ap2Ocn-Cre mice than that of the control BMMs on day 5 of RANKL treatment (Fig. S2e, f), implicating that more OC progenitors were present in the cultured BMMs from Atp6ap2Ocn-Cre mice. To determine whether these MNCs have resorptive function, we cultured these cells on coverslips coated with inorganic hydroxyapatite matrix. An increase in the resorptive pit-like area was detected in the mutant MNCs or OCs, as compared with that of control cells (Fig. S2e, g). Together, these results uncover a role of osteoblastic ATP6AP2 in preventing hyper-active OC formation and activation in both trabecular and cortical bone regions, unlike the selective trabecular bone loss, but increase of cortical bone mass, in Atp6ap2Ocn-Cre mice, and implicating additional pathological mechanism(s) to underlie the bone phenotypes in the mutant mice.

Reduced trabecular, but elevated cortical, bone formation in Atp6ap2 Ocn-Cre mice

We then examined bone formation in the control and Atp6ap2Ocn-Cre mice. First, measuring serum levels of osteocalcin, a marker for bone formation, by ELISA showed decreases in both 1-month and 3-month Atp6ap2Ocn-Cre mice, as compared with those of littermate control mice (Fig. 2a). Second, examining bone formation in both trabecular and cortical bones of the control and mutant mice by injecting fluorochrome-labeled calcein green into 1-month mice twice at a 5-days interval showed decreases in mineral apposition rate (MAR), mineral surface/bone surface (MS/BS), and bone formation rate (BFR) selective in trabecular bone region (Fig. 2b, c). In contrast, in the mutant periosteum and endosteum bone regions, the MAR, MS/BS, and BFR were all increased, as compared with those of control mice (Fig. 2b, d, e). These results thus demonstrate an association of the trabecular/cortical bone phenotypes with the altered bone formation rate in Atp6ap2Ocn-Cre mice. Third, using in vitro OB differentiation assay, BMSCs from Atp6ap2Ocn-cre mice formed less ALP+ OBs and lower calcified bone matrix viewed by Alizarin Red S staining than those of OBs derived from control BMSCs (Fig. S3a–c). In addition, the transcript levels of Runx2 (Runt-related transcription factor 2) and Osx (Osterix), both transcription factors that play important roles in OB differentiation and bone formation,36 were also significantly reduced in Atp6ap2-KO BMSCs during osteogenic differentiation (Fig. S3d). These results demonstrate a reduced OB-genesis and function in the mutant mice, suggest a cell autonomous role of ATP6AP2 in this event, and implicate the bone region selective alteration of the bone formation rate to underlie the regional specific bone phenotypes in Atp6ap2Ocn-Cre mice.

Fig. 2

Decreased trabecular, but increased cortical, bone formation in Atp6ap2Ocn-Cre mice. a The serum osteocalcin levels in 1-month and 3-month male ctrl and Atp6ap2Ocn-Cre mice were measured by ELISA. b Representative images of histologic sections showing calcein labeling of trabecular, periosteum and endosteum bone regions (diaphysis region at the same distance from the growth plate, close to metaphysis) in femur of male ctrl and Atp6ap2Ocn-Cre mice at age of 1-month. c–e MAR (mineral apposition rate), MS (mineral surface) / BS (bone surface), and BFR (bone formation rate) are presented. Data in (a) and (c–e) are shown as box plots together with individual data points, and whiskers indicate minimum to maximum (n = 4 to 6 animals per genotype). P values obtained by unpaired two-tailed t test. *P < 0.05. **P < 0.01

Diminished LRP6 in ATP6AP2-KO OBs

To investigate how osteoblastic ATP6AP2 regulates trabecular bone formation, we carried out proteomic analysis using liquid chromatography-tandem mass spectrometry to screen for altered plasma membrane proteins in primary cultured ATP6AP2-KO Ocn-Cre+ OBs (Fig. 3a). Both Ctrl (Ocn-Cre) and ATP6AP2-KO OBs were incubated with NHS-biotin for 45 min at 4 °C to label cell surface membrane proteins. The biotin labeled surface proteins were pulled down by streptavidin-agarose beads and subjected to proteomic analysis and Western blot analysis. Among a total of 542 proteins identified, 27 proteins were down-regulated, and 2 proteins was up-regulated in ATP6AP2-KO OBs, as compared with those of ctrl (Fig. 3b, c). Gene ontology (GO) biological processes analysis showed several biological processes, including tube morphogenesis, cell-cell adhesion, and regulation of bone remodeling that were altered by ATP6AP2-KO (Fig. 3b). Notice that the proteins in the regulation of bone remodeling pathway include the Wnt/β-catenin pathway proteins, such as Lrp6, Cdh2 (N-Cadherin) and Ptk7 (Fig. 3d). Given the reports that LRP6 is a receptor for Wnt/β-catenin signaling,1,37 an essential pathway for osteoblastic bone formation, and N-cadherin/β-catenin not only controls cell-cell adhesion, but also proliferation, differentiation, and survival of mesenchymal cells,38 we further examined LRP6, its homolog LRP5, as well as N-cadherin’s cell surface levels in control and ATP6AP2-KO OB-lineage cells. Western blot analysis of lysates of subcellular fractions of OBs showed that the cell surface levels of LRP6, LRP5 and N-cadherin were lower in ATP6AP2-KO OB-progenitors than those of controls (Fig. 3e, f), suggesting ATP6AP2’s function in promoting their surface targeting. The cytoplastic and total levels of LRP6 and N-cadherin, but not LRP5, were lower in the mutant OBs (Fig. 3e, f). The reductions of these membrane proteins appeared to be selective, as the LRP1 level was un-changed. The mRNA level of Lrp6 in the mutant OBs was comparable to that of control OBs (Fig. S4a), suggesting that ATP6AP2 increases LRP6 at the post-transcriptional level.

Fig. 3

Decreased LRP6 and N-cadherin surface distribution and total protein levels in ATP6AP2-KO OBs. a Schematic diagram of quantitative proteomic analysis. Proteomic data were obtained from surface proteins of ctrl and ATP6AP2-KO OBs. b The gene ontology (GO) Biological Processes analysis shows the top 20 significant enrichment terms represented by biological processes, with longer column representing more significant enrichment. Many proteins that regulate bone remodeling were involved. c Volcano plots of differentially expressed proteins in the cell surface of ctrl and ATP6AP2-KO OBs. The up-regulated proteins were marked in red, and the down-regulated proteins were indicated in blue (P < 0.05). d Heat map protein expression z-scores computed for the 10 proteins that were involved in regulation of bone remodeling. e, f Western blot analyses of lysates of total, cytoplasm and cell surface fractions of ctrl and ATP6AP2-KO OBs. Quantification of the data was shown in (f). Data in (f) are presented as mean ± SD (n = 3). P values obtained by unpaired two-tailed t test. *P < 0.05

Attenuation of LRP6/β-catenin signaling in ATP6AP2-KO OBs, but not osteocytes, in culture and in vivo

The OB-lineage cells include BMSCs, OBs, and osteocytes.39,40,41 To address whether ATP6AP2 regulates LRP6 and LRP5 proteins in these OB-lineage cells, we further examined LRP6 and LRP5 levels in lysates from BMSCs and osteocytes (Ocys) derived from control and Atp6ap2Ocn-cre mice by Western blot. Notice that LRP6 appeared to be less in Ocys than that of BMSCs (Fig. 4a, b); and β-catenin, but not ATP6AP2, was also lower in Ocys than that of BMSCs (Fig. 4a, b), implicating a more critical role of LRP6/β-catenin signaling in the trabecular bone region. Interestingly, both LRP6 and β-catenin were decreased in the mutant BMSCs, but not Ocys, as compared with those of control cells (Fig. 4a, b). In contrast, the total level of LRP5 was selectively reduced in the mutant Ocys, but not BMSCs, implicate a differential regulation of LRP5 and LRP6 by ATP6AP2 in a cell subtype selective manner. Together, these results suggest that ATP6AP2 selectively regulate LRP6 and β-catenin levels in the early stage of OB-lineage cells (e.g., BMSCs and OBs) but regulates LRP5 in the later stage (e.g., Ocys), providing additional evidence for ATP6AP2’s function in regulating Wnt/β-catenin signaling in a cell type or bone region selective manner.

Fig. 4

Impairment of Wnt/β-catenin signaling selectively in the OBs of trabecular bone, but not osteocytes of cortical bone, of Atp6ap2Ocn-Cre mice. a, b Decreased LRP6 and β-catenin expression in BMSCs, buy not Ocys, derived from 3-month Atp6ap2Ocn-Cre mice by Western blots. a Representative blots; and (b). Quantification analysis. c–f Selectively decreased β-catenin signaling in the trabecular bone of Atp6ap2Ocn-Cre mice at 1-month, by Axin2LacZ reporter activity. Ctrl and Atp6ap2Ocn-Cre mice are crossed with Axin2LacZ reporter line and β-galactosidase activity is revealed by X-gal staining. Representative images were shown in (c–e). Bar, 50 μm. Quantification analysis as mean ± SD was shown in (f). Data in (b) and (f) are presented as mean ± SD (n = 3 or 9). P values obtained by unpaired two-tailed t test. *P < 0.05. **P < 0.01. ***P < 0.001, significant difference

To test above view in vivo, we crossed Axin2LacZ mice, a Wnt/β-catenin signaling LacZ reporter mouse line, with Atp6ap2Ocn-cre to generate Axin2LacZ; Atp6ap2Ocn-cre mice. X-gal staining analysis revealed abundant β-galactosidase activity in the trabecular region of the ctrl mice (Axin2LacZ; Ocn-Cre), which was marked reduced in the mutant mice (Fig. 4c, d, f). In contract, β-galactosidase activity in the skeletal muscle, where ATP6AP2 has not been knocked out, showed no significant changes in the mutant mice compared with the ctrl mice, supporting the view for ATP6AP2 as a positive regulator of Wnt/β-catenin signaling in vivo.16,17 Notice that in the cortical bone diaphysis regions of both control and mutant mice, the β-galactosidase activity was weakly detectable, but comparable between control and mutant mice (Fig. 4c, e, f). These results thus provide in vivo evidence that ATP6AP2 is required for Wnt/β-catenin signaling in the OB-lineage cells (OBs) at the trabecular bone region, but not osteocytes in the cortical region.

Impairments in Wnt3a induced β-catenin signaling and OB genesis in cultured BMSCs from Atp6ap2 Ocn-Cre mice

LRP6, one of receptors for Wnts, mediates accumulation and nuclear translocation of β-catenin, and regulates a series of target genes expression and multiple cellular processes, including OB genesis.1,2,3 We thus asked whether ATP6AP2 is necessary for Wnt3a induced β-catenin signaling and OB formation. Primary cultured BMSCs derived from ctrl and Atp6ap2Ocn-Cre mice were treated with recombinant Wnt3a protein for overnight (Fig. 5a). Western blot results showed that Wnt3a treatment increased the β-catenin level in the control BMSCs (Fig. 5b, c). However, in the ATP6AP2-KO BMSCs, Wnt3a failed to induce the β-catenin level, in addition to the lower β-catenin and LRP6 in the mutant cells under basal condition (Fig. 5b, c). Notice that wnt3a treatment enhanced ATP6AP2 but reduced Lrp6/Lrp5 protein levels and mRNA levels (Fig. 5b, c, Fig. S4a–c). This may be a kind of negative feedback on Wnt3a treatment. These results suggest a necessity of osteoblastic ATP6AP2 in maintaining both basal and Wnt3a induced β-catenin levels. Moreover, immunofluorescence staining analysis showed that Wnt3a induced the nuclear localization of β-catenin in control, but not ATP6AP2-KO, BMSCs (Fig. 5d–g), compared with control, ATP6AP2-KO BMSCs also showed decreased surface and cytoplasm levels of β-catenin (Fig. 5d–g), providing evidence for a requirement of ATP6AP2 in Wnt3a induced β-catenin nuclear distribution.

Fig. 5

Attenuation of Wnt/β-catenin signaling in Atp6ap2Ocn-Cre BMSCs. a Experimental strategy. Cells were treated with 100 ng/mL Wnt3a for 12 h as indicated. b, c Western blot analysis of ATP6AP2, total-β-catenin, LRP5 and LRP6 expression in BMSCs from indicated mice. Quantitative data were shown in (c). d–g Primary cultured control and ATP6AP2-KO BMSCs were treated with 100 ng/mL Wnt3a for 6 h. Fluorescence microscopy was undertaken to monitor β-catenin. Bar, 10 µm. Quantification analysis was shown in (e–g). h, i ALP staining analysis of cultured BMSCs in the presence of veh or Wnt3a. Bar, 20 µm. Quantitative analysis of the average ALP+ cell number was shown in (i). j Real-time PCR analysis of gene expressions in BMSCs derived from control and ATP6AP2-KO mice after wnt3a treatment (12 h). Data in (c), (e–g) and (i–j) are presented as mean ± SD (n = 3 or 6). P values obtained by two-way ANOVA followed by Bonferroni post hoc test. *P < 0.05. **P < 0.01. ***P < 0.001, significant difference

We then asked whether ATP6AP2 is necessary for Wnt3a induced OB-genesis. BMSCs from control and Atp6ap2Ocn-Cre mice were induced for OB-differentiation in the presence or absence of Wnt3a. As shown in Fig. 5h, i, Wnt3a significantly enhanced ALP+ OBs in control BMSCs, but little to no induction in ATP6AP2-KO BMSCs (Fig. 5h, i), supporting the view for the requirement of ATP6AP2 in this event. We further tested this view by examining the expression of Wnt/β-catenin target genes, including Axin2, Tcf7, Lef1 and Myc. As shown in Fig. 5j, RT-PCR analysis showed increased expression of all of these β-catenin target genes in the control cells; but in ATP6AP2-KO cells, the inductions in Tcf7 and Myc, but not Axin2 or Lef1, were abolished (Fig. 5j), suggesting a selective role of ATP6AP2 in regulating Wnt3a/β-catenin target gene expression for OB-differentiation.

Notice that ATP6AP2-deficiency in BMSCs also impaired the expression of osteoprotegerin (OPG), thus increased the ratio of receptor activator of nuclear kappa B ligand (Rankl) over OPG, which may underlie the increased OC formation in ATP6AP2-KO mice (Fig. S4d). Wnt3a could up-regulate the expression of OPG thus reduce the ratio of Rankl/OPG in the ctrl, but not in the ATP6AP2-KO group (Fig. S4d). These results suggested that the impairment of Wnt/β-catenin signaling in Atp6ap2Ocn-Cre BMSCs may be responsible for not only the reduced OB genesis and OB mediated bone formation, but also the OC-genesis and OC mediated bone resorption.

Independent on ATP6AP2 for Wnt3a suppression of β-catenin phosphorylation

To investigate how ATP6AP2 increases β-catenin expression in both basal and Wnt3a stimulated conditions, we first examined β-catenin’s mRNA levels in control and ATP6AP2 KO BMSCs. The mRNA levels of β-catenin in ATP6AP2-KO cells showed no significantly change compared with the ctrl, indicating that ATP6AP2 mainly regulates the protein level of β-catenin (Fig. S4e). Second, we examined β-catenin phosphorylation, a process mediated by GSK3β, suppressed by Wnt3a, and critical for β-catenin degradation,1,3 in control and ATP6AP2 KO BMSCs treated with or without Wnt3a. Western blot analysis showed that the phosphorylations of β-catenin, recognized by the phosphor-Ser33/37/Thr41 antibodies, were indeed decreased by Wnt3a in control cells (Fig. 6a–d). Unexpectedly, such a Wnt3a suppression of phosphor-β-catenin was also detectable in ATP6AP2-KO cells (Fig. 6a–d). Similar to these results were the observations by co-immunostaining analyses, which showed marked decrease of phosphor-β-catenin in both control and ATP6AP2-KD MCT3T3 cells in response to Wnt3a, while the total β-catenin, in particular the nuclear and cytoplasm β-catenin levels were induced by Wnt3a in control, but not ATP6AP2-KD cells (Fig. 6e–k). We further examined the ubiquitination of β-catenin and found that Wnt3a decreased ubiquitin conjugated β-catenin in ctrl, but not in ATP6AP2-KD MC3T3 cells (Fig. S5), suggesting a role of ATP6AP2 in Wnt3a suppression of β-catenin’s ubiquitination or ubiquitin mediated proteosome degradation. These results thus suggest an independence of ATP6AP2 in Wnt3a suppression of β-catenin phosphorylation, implicating β-catenin phosphorylation/dephosphorylation independent mechanism(s) to be involved in ATP6AP2 stabilizing β-catenin.

Fig. 6

Wnt3a inhibition of phosphor-β-catenin in ATP6AP2 independent manner. a–d Primary cultured control and ATP6AP2-KO BMSCs were treated with 100 ng/mL Wnt3a for 12 h. Western blotting analysis of ATP6AP2, β-catenin and phosphor-β-catenin (Ser33/37, Thr41) were shown in (a). β-actin was used as the loading controls. Quantification analyses were shown in (b-d). e–k Immunostaining analysis of β-catenin (e) and phosphor-β-catenin (j) in ctrl and ATP6AP2-KD MC3T3 cells transfected with the LRP6-eGFP. Cells were treated with 100 ng/mL Wnt3a for 6 h. Bar, 10 µm. Images marked with yellow squares were amplified and shown below. Nuclei were marked with white dashed lines. Quantification analyses were shown in (f, g, h, i, k). Data in (b–d) are presented as mean ± SD (n = 3). Data in (f–i) and (k) are shown as box plots together with individual data points, and whiskers indicate minimum to maximum (n = 10). P values obtained by two-way ANOVA followed by Bonferroni post hoc test. *P < 0.05. **P < 0.01. ***P < 0.001

Requirement of ATP6AP2 for LRP6/β-catenin targeting to the cell surface and preventing LRP6 distribution in the lysosomes

Notice that both LRP6 and β-catenin were lower in ATP6AP2-KO cells (Fig. 5b, c); and LRP6 interacts with ATP6AP2.17 In light of these observations, we speculate that ATP6AP2 may stabilize β-catenin via LRP6. To test this view, we first examined whether the distribution of β-catenin and LRP6 were altered in ATP6AP2-KD MC3T3 cells. LRP6-GFP was transfected into ctrl and ATP6AP2-KD MC3T3 cell lines. As shown in Fig. 6e, f, the β-catenin was co-localized with LRP6 largely on the cell membrane. However, in the ATP6AP2-KD cells, the β-catenin was significantly decreased in the cell membrane, and it was mainly co-localized with LRP6 in the cytoplasm (Fig. 6e–i). Additionally, ATP6AP2-KD significantly reduced the nucleus distribution of β-catenin, as compared with that in ctrls (Fig. 6e–i).

To understand how ATP6AP2 promotes LRP6 surface targeting, we examined ATP6AP2 and LRP6 interaction and mapped the binding domain in ATP6AP2. ATP6AP2 and its deletion mutants were co-transfected with LRP6-GFP into MC3T3. As shown in Fig. S6a–c, ATP6AP2 appeared to form a complex with LRP6 via its ECD domain. Interestingly, the ICD domain, also called M8.9, which binds to the V-ATPase complex proteins and promotes V-ATPase complex assemble and activation,42 was not co-localized with LRP6 (Fig. S6a–c), implicating that ATP6AP2 may interact with LRP6 independent on its association with V-ATPase.

In addition to the reduced LRP6 surface targeting in ATP6AP2-KD MC3T3 cells, LRP6 in EEA1+ endosomes were decreased, but in Lamp1+ lysosomes were increased (Fig. S6d, e). We then further examined LRP6’s half-life. Indeed, a shorter half-life of LRP6 was detected in ATP6AP2-KD MC3T3 cells (Fig. S6f, g). These results suggest that ATP6AP2 may increase LRP6 stability by promoting its recycle from endosomes to the cell surface; and thus prevent its degradation (Fig. S6h).

Requirement of ATP6AP2 for N-cadherin/β-catenin protein complex distribution in the cell surface

In addition to LRP6, our proteomics analysis and western blot analysis showed reduced cell surface level of N-cadherin (Cdh2) in ATP6AP2-KO OBs (Fig. 3c, d). Given the reports that N-cadherin-KO impairs OB differentiation, and N-cadherin could interact with LRP6,38 we further examined their distribution in control and ATP6AP2-KD MC3T3 cells by co-immunofluorescence staining. Interestingly, LRP6 was found to form a complex with N-cadherin and β-catenin at non-adhesive regions of the cell surface (Fig. 7a, b), but not the cell-cell adhesion sites, in OBs (Fig. 7a, c). These cell surface associated LRP6/N-cadherin/β-catenin complexes were un-detectable in ATP6AP2-KD OB cells (Fig. 7a, d). Similar with LRP6, N-cadherin’s distribution at the cell surface was reduced, but its localization in LAMP1+ lysosomes was increased in ATP6AP2-KD MC3T3 cells (Fig. S7a, b). These results suggest that ATP6AP2 may promote both LRP6 and N-cadherin surface targeting, where they interacted with β-catenin, thus form a “β-catenin pool” at the plasma membrane.

Fig. 7

Requirement of ATP6AP2 for N-cadherin/β-catenin protein complex distribution in the cell surface. a Immunostaining analysis of N-cadherin and β-catenin in ctrl and ATP6AP2-KD MC3T3 cells transfected with the Lrp6-eGFP. Cells were treated with 100 ng/mL Wnt3a for 6 h. Bar, 10 µm. Images marked with yellow squares were amplified and shown below. The blue dotted box marked surface adhesive regions, and the purple dotted box marked surface non-adhesive regions. b, c Quantification analyses of LRP6, N-cadherin, and β-catenin fluorescence distribution in the marked regions. d Quantification analyses of the LRP6-N-cadherin-β-catenin complex fluorescence intensity in the surface of cells. Data in (d) are shown as box plots together with individual data points, and whiskers indicate minimum to maximum (n = 10). P values obtained by two-way ANOVA followed by Bonferroni post hoc test. ***P < 0.001

Diminished OB differentiation deficit in Atp6ap2 Ocn-Cre BMSCs by expression of active β-catenin

To determine whether the impaired Wnt/β-catenin signaling in Atp6ap2Ocn-Cre BMSCs is responsible for the reduced OB genesis and increased OC-genesis, we expressed the active β-catenin into ATP6AP2-KO BMSCs, and examined its effect on OB differentiation. As illustrated in Fig. 8a, a lentivirus encoding β-catenin-DeltaN90 (constitutively active β-catenin, or CA-β-catenin) was generated, which efficiently express CA-β-catenin in BMSCs (Fig. 8b–d). Expression of the CA-β-catenin indeed diminished the OB differentiation deficit in Atp6ap2Ocn-Cre BMSCs by ALP staining analysis (Fig. 6e, f), and increased expression levels of the -catenin target genes in both ctrl and ATP6AP2-KO cells by RT-PCR analysis (Fig. 8g). Note that similar to Wht3a treatment, the CA-β-catenin also significantly increased the mRNA and protein levels of ATP6AP2 (Fig. 8b–d, Fig. S8), supporting the view for ATP6AP2 as a downstream target gene of Wnt3a/β-catenin signaling. Additionally, expression of the CA-β-catenin induced the expression of OPG thus reduced the ratio of RANKl/OPG in both ctrl and ATP6AP2-KO BMSCs (Fig. S8d). These results thus support the view for the impaired Wnt3a/β-catenin signaling in ATP6AP2-KO BMSCs to be responsible for the OB-differentiation deficit, and implicating the increased RANKL/OPG ratio in ATP6AP2-KO BMSCs to be involved in the elevated osteoclast genesis and bone resorption.

Fig. 8

Diminished OB differentiation deficit in Atp6ap2Ocn-Cre BMSCs by expression of active β-catenin. a Experimental strategy. BMSCs derived from 3-month male ctrl or Atp6ap2Ocn-Cre mice infected with active β-catenin (DeltaN90) lentivirus followed by G418 selection. The cells were then used for western blot, Q-PCR and osteoblast differentiation experiments. b–d Western blot of ATP6AP2 and β-catenin in primary cultured BMSCs. β-actin was used as the loading controls. Quantification analyses were shown in (c, d). e, f BMSCs were cultured in osteogenic differentiation medium for 14 days (ALP staining). Osteoblast-like cells indicated by ALP (alkaline phosphatase) staining were shown in (e). Bar, 20 μm. Quantitative analysis of the ALP cell number was shown in (f). g Real-time PCR analysis of gene expressions in BMSCs. Data in (c), (d), (f), (g) are presented as mean ± SD (n = 3 or 6). P values obtained by two-way ANOVA followed by Bonferroni post hoc test. *P < 0.05. **P < 0.01. ***P < 0.001, significant difference