Inhibition of Axin1 in osteoblast precursor cells leads to defects in postnatal

Bing Shu, Yongjian Zhao, Shitian Zhao, Haobo Pan, Rong Xie, Dan Yi, Ke Lu , Junjie Yang, Chunchun Xue,Jian Huang , Jing Wang, Dongfeng Zhao, Guozhi Xiao, Yongjun Wang and Di Chen

INTRODUCTION

β-catenin is a central molecule in the canonical wingless/integrated(Wnt)pathway.When Wnt ligands interact with Frizzled and low-density lipoprotein receptor-related protein 5 and 6 coreceptors, β-catenin is activated, accumulated in the cytoplasm,and translocated into the nucleus. In the nucleus, β-catenin activates transcription of downstream genes. β-catenin signaling plays important roles in bone development, postnatal bone growth, and bone remodeling. Animal studies, in which the β-catenin signaling was either inhibited or activated in chondrocytes, osteoblasts, or osteocytes, demonstrated that activation of β-catenin signaling promotes osteoblast differentiation and bone formation and inhibition of osteoclast formation and bone resorption.1-7However, the detailed mechanisms of how postnatal bone growth and bone remodeling are regulated by β-catenin signaling remain unclear.

The β-catenin signaling in mesenchymal stem cells (MSCs)promoted osteoblast differentiation, inhibited chondrocyte differentiation, and enhanced endochondral ossification.8-9For example,deletion of LRP6,the co-receptor of Wnts,in nestin-expressing cells caused bone mass decrease.10In addition,β-catenin signaling in mature osteoblasts inhibits osteoclast formation.11-13Activation of β-catenin signaling also prevented osteoblast apoptosis.14In contrast,the function of β-catenin signaling in osteoblast precursor cells has not been fully investigated.

Without Wnt ligands interacting with the cell surface receptors,cytoplasmic β-catenin is degraded by the ubiquitin-proteasome system,mediated by the destruction complex in a phosphorylationdependent manner.Axin1 and Axin2 are scaffolding proteins in the destruction complex and promote β-catenin phosphorylation and degradation.Upon Wnt ligand binding to Wnt receptors on the cell surface,the destruction complex is dissociated and β-catenin is then released from the destruction complex and subsequently translocated into the nucleus.

Several in vivo studies were conducted to determine the role of the β-catenin destruction complex in bone development.15-17For example, Axin2 KO mice showed craniosynostosis and significantly increased trabecular bone mass.18-19Mice lacking APC,a member in destruction complex,in osteoblasts exhibited dramatically increased bone deposition.11Axin1 is also a scaffold protein in the destruction complex and is the negative regulator of β-catenin signaling.15,17Axin1 was expressed ubiquitously, and the systematic deletion of Axin1 led to early embryonic lethality in mice.20Therefore, exact functions of Axin1 at different differentiation stages during MSC differentiation have not been investigated due to the limitation of the lethality of conventional deletion of Axin1.

To determine the potential role of Axin1 in osteoblast precursor cells at postnatal stage during bone remodeling, we generated Axin1flox/floxmice21and bred these mice with Osx-Cre mice to produce Axin1Osxconditional KO mice.We found that loss of Axin1 in osteoblast precursor cells mainly affects osteoclast formation in metaphyseal bone area.

RESULTS

Increased expression of β-catenin in Axin1OsxKO mice To delete Axin1 in osteoblasts, primary osteoblasts isolated from calvariae of Axin1flox/floxmice were infected with adenovirus-Cre recombinase.We found that Axin1 mRNA and protein expressions were significantly decreased in calvarial osteoblasts isolated from Axin1flox/floxmice infected with Adeno-Cre (Fig. 1a, b). In contrast,β-catenin expression was increased in these cells (Fig. 1b). We then bred Axin1flox/floxmice with Osx-Cre mice and generated Axin1OsxKO mice. We found that Axin1 expression was decreased while β-catenin expression was significantly increased in trabecular bone of tibiae of Axin1OsxKO mice (Fig. 1c). In addition, we detected Axin1 and β-catenin expression in adipocytes and perivascular cells in the bone marrow. We found that Axin1 expression was decreased while β-catenin expression was significantly increased in adipocytes (Fig. 1d). However, we did not observe obvious changes in Axin1 and β-catenin expression in perivascular cells (Fig. 1d). We also examined changes in Wnt inhibitors by determining the expression of Dkk1 and sclerostin in trabecular bone and found that expression of both Dkk1 and sclerostin was upregulated in trabecular bone area below the growth plate in Axin1OsxKO mice(Fig.1e,f).To determine changes in skeletal structure, newborn Axin1OsxKO mice and their littermate controls were collected and performed whole body Alizarin red/Alcian blue staining. We did not observe significant changes in skeletal structure in Axin1OsxKO mice (Fig. 1g). In contrast, slightly delayed mineralization of calvarial bone was observed in newborn and 4-week-old Axin1OsxKO mice (Fig. S1).

Delayed endochondral bone growth in Axin1Osx KO mice

We performed histological analyses and observed an expanded hypertrophic zone in tibial growth plates of newborn and 1-weekold Axin1OsxKO mice(Fig.2a,d).In newborn Axin1OsxKO mice,the length of the hypertrophic zone is almost three times longer than that of Cre-negative mice (Fig. 2a). New bone formation in the primary ossification center was delayed in newborn Axin1OsxKO mice (Fig. 2b). In 1-week-old Axin1OsxKO mice, the length of the hypertrophic zone of Axin1OsxKO mice was significantly longer compared with that of Cre-negative controls (Fig. 2d). Large amounts of uncalcified bone with accumulated osteoid were found in the trabecular bone area below the growth plate(Fig.2e),suggesting defects in bone remodeling in Axin1OsxKO mice. To determine if these changes are due to activation of β-catenin signaling and to compare the difference between Axin1OsxKO mice and β-catenin conditional activation mice,we generated and analyzed newborn and 1-week-old β-catenin(ex3)Osxactivation mice. Compared with Axin1OsxKO mice, we did not observe obvious expansion of hypertrophic cartilage in β-catenin(ex3)Osxactivation mice (Fig. 2c, f). These findings suggest that Axin1 may also act through a β-catenin-independent mechanism to regulate postnatal bone growth. In 2-week-old Axin1OsxKO mice, we observed a slightly delayed formation of a secondary ossification center(Fig.2g).In contrast,the formation of secondary ossification centers were significantly delayed in β-catenin(ex3)Osxactivation mice (Fig. 2h). In 4-week-old mice, it seems that growth plate cartilage development and the formation of a secondary ossification center were normal in Axin1OsxKO mice (Fig. 2i) or in β-catenin(ex3)Osxactivation mice(Fig.2j).Results of IHC analyses showed that extensively increased Col-X-positive hypertrophic chondrocytes were found in the metaphyseal bone area of newborn Axin1OsxKO tibiae (Fig. 2k). Similarly, extensively increased MMP13-positive cells were also found in the expanded hypertrophic zone (Fig. 2l). This phenotype was 100% penetrated in Axin1OsxKO mice.

Impaired osteoclast formation in Axin1Osx KO mice

Osteoclasts in growth plate can phagocytose dying hypertrophic chondrocytes and absorb the mineralized cartilage matrix. Therefore, we analyzed changes in osteoclast formation in Axin1OsxKO mice. In newborn Cre-negative tibiae, large numbers of TRAPpositive osteoclasts were found on the metaphyseal bone area and the inner side of cortical bone(Fig.3a).In addition,osteoclasts were also found in the ossification front, where osteoclasts invaded hypertrophic chondrocytes (Fig. 3a, lower left panel). In newborn Axin1OsxKO tibiae,TRAP-positive osteoclasts in the bone marrow cavity and ossification front were significantly decreased(Fig. 3a, right panel). Osteoclast formation in the subchondral bone of 2- and 4-week-old Axin1OsxKO tibiae was also decreased compared with that of Cre-negative tibiae(Fig.3b).Quantification of the ratios of osteoclast numbers and trabecular bone perimeters showed reduced TRAP-positive osteoclast numbers in 4-week-old Axin1OsxKO mice (Fig. 3c). Real-time PCR and IHC results also showed decreased expression of osteoclast markers,MMP9(Fig.3d,e)and cathpesin K (Fig.3f,g)in both subchondral bone and metaphyseal bone of 4-week-old Axin1OsxKO tibiae. By contrast,no obvious changes in osteoblast numbers were found in the metaphyseal bone of same-aged Axin1OsxKO mice(Fig.4a,b).Results of calcein labeling and quantification of mineral appositional rates (MAR) showed that bone formation of cortical bone was not significantly changed in Axin1OsxKO mice (Fig. 4c, d).

Increased Opg expression in Axin1Osx KO mice

The ratio of Opg/Rankl is an important index reflecting changes in osteoclast formation. In the tibiae of Cre-negative mice, osteoprotegerin (OPG)-positive staining cells were observed on the trabecular bone surface(Fig.5a).In Axin1OsxKO mice,the number and staining intensity of OPG-positive cells were significantly increased in Axin1OsxKO mice, especially on the trabecular bone surface (Fig. 5a). Osteoblasts are the important cell resource for OPG production and osteoblast precursor cells are the Osterixexpressing cells. Therefore, we examined changes in Opg and Rankl expression in primary calvarial osteoblasts of newborn Axin1OsxKO mice and Cre-negative littermate controls. In calvarial osteoblasts of newborn Axin1OsxKO mice, though both Opg and Rankl expressions were increased (Fig. 5b, c), the Opg:Rankl ratio was still significantly higher in Axin1-deficient osteoblasts than that of control osteoblasts (Fig. 5d). We then cultured wild-type(WT)bone marrow derived macrophages(BMM)cells treated with the conditioned media (CM) collected from primary osteoblasts isolated from Axin1OsxKO mice and Cre-negative mice, respectively. Osteoclast formation in the BMM cells treated with CM collected from osteoblasts of Axin1OsxKO mice was obviously decreased compared with the cells treated with CM collected from control osteoblasts (Fig. 5e). These findings were further confirmed by the results of osteoclast quantification (Fig. 5f).Nuclear factor of activated T-cells, cytoplasmic 1 (NFATc-1) and c-Fos are two key regulators of osteoclast differentiation upon receptor activator of nuclear factor kappa-Β ligand (RANKL)induction. Real-time PCR assay revealed that expression of Nfatc-1 and c-Fos was decreased in trabecular bone of Axin1OsxKO mice(Fig.5g, h). IHC assays further demonstrated that NFATc-1- and c-Fos-expressing cells were detected in trabecular bone, and that NFATc-1 was especially highly expressed in the ossification front of the tibiae in control mice (Fig. 5i, j). The numbers of NFATc-1 and c-Fos positive cells were significantly decreased in the tibiae of Axin1OsxKO mice (Fig. 5i, j).

DISCUSSION

Osterix is the osteoblast specific transcription factor that regulates osteoblast precursor cell differentiation and inhibits cell proliferation.22-23In Osx-expressing osteoblast precursor cells,Wnt/β-catenin signaling activation is at a low level. Osx-Cre targeting cells were established by breeding Osx-Cre mice with ROSAmT/mGreporter mice,followed by fluorescence microscopic analysis.Osx-Cre targeting cells were detected in trabecular bone and on the endosteal region of cortical bone; they were especially highly expressed in the area of metaphyseal bone that is close to the growth plate hypertrophic cartilage area.22Consistent with these findings, when Axin1 expression was inhibited in osteoblast precursor cells using Osx-Cre transgenic mice,β-catenin-positive staining cells were also detected in trabecular bone and on the endosteal region of the cortical bone.

In this project we discovered that deletion of Axin1 in osteoblast precursor cells led to β-catenin upregulation, which further resulted in increased OPG expression. Increased OPG expression could inhibit osteoclast formation, which may be responsible for the reduction of apoptosis in hypertrophic chondrocytes and decreased resorption of mineralized cartilage matrix. Wnt/β-catenin signaling controls osteoblast and osteoclast differentiation and regulates bone mass.24-25In our study,osteoblast numbers and bone mass were not significantly changed in Axin1OsxKO mice. In contrast, osteoclast formation was dramatically decreased when Axin1 was deleted in Osxexpressing osteoblast precursor cells. In Axin1OsxKO mice,β-catenin levels were increased in osteoblast precursor cells. It is known that inhibition of β-catenin signaling causes an increase in chondrocyte apoptosis during postnatal cartilage development.26β-catenin, together with TCF proteins, regulates Opg expression in osteoblasts.12Loss of β-catenin in mature osteoblasts resulted in a decreased ratio of OPG/RANKL, while accumulation of β-catenin signaling led to an increased ratio of OPG/RANKL.11-13OPG is a decoy receptor that prevents RANKL/RANK interaction and inhibits osteoclast differentiation.27-29Therefore, activation of β-catenin signaling in mature osteoblasts resulted in a decrease in osteoclast formation, while inhibition of β-catenin signaling caused an increase in osteoclast formation leading to severe osteopenia.11-13In addition,β-catenin signaling in chondrocytes could also regulate osteoclast formation and bone resorption through regulation of OPG expression.30In this report, we have demonstrated that deletion of Axin1 in osteoblast precursor cells led to upregulation of β-catenin signaling, increased the ratio of OPG/RANKL and prevented osteoclast formation. We also found that Rankl expression in primary calvarial osteoblasts of Axin1Osxmice was also upregulated. This is not consistent with previous reports that β-catenin signaling downregulated Rankl expression in a glucocorticoid receptor-dependent manner.27This discrepancy may be due to the possibility that Axin1 may also affect other signaling molecules in addition to that of β-catenin. In fact, comparing differences in the histological results of newborn,1-week-old and 2-week-old Axin1OsxKO mice with those of β-catenin(ex3)Osxactivation mice, the phenotypes of the expanded hypertrophic zone and formation of the secondary ossification center are very different between Axin1 KO mice and β-catenin activation mice, suggesting that Axin1 may also act through a β-catenin-independent mechanism to regulate postnatal bone growth.

During the endochondral bone formation process,hypertrophic chondrocytes in the growth plate undergo apoptosis and the mineralized cartilage matrix was absorbed, followed by bone formation. At the ossification front, hypertrophic chondrocytes directly contact with osteoclasts.Findings with annexin-V labeling confirmed that osteoclasts could remove dying chondrocytes by phagocytosis.31Meanwhile, osteoclasts are also capable of absorbing the mineralized cartilage matrix.32-33Mice lacking TRAP resulted in an expanded hypertrophic zone and disordered hypertrophic chondrocyte columns that extended into the trabecular bone region.34Mice lacking RANK showed an even more severe phenotype of an expanded hypertrophic zone. The bone marrow cavity was almost filled with unabsorbed cartilage.35-36In Axin1OsxKO mice, we also found an expanded hypertrophic zone and unabsorbed cartilage in the bone marrow cavity, which may be caused by decreased osteoclast formation.Meanwhile, large numbers of Col-X-positive prehypertrophic and hypertrophic chondrocytes and MMP13-expressing terminally differentiated hypertrophic chondrocytes were found in the expanded hypertrophic zone of the growth plate as well as in the bone marrow cavity.This suggests that the terminal apoptosis of chondrocytes was inhibited or delayed due to the loss of Axin1 in osteoblast precursor cells. Although Osterix was also detected in a subset of chondrocytes,37-38in Axin1OsxKO mice, β-catenin upregulation was found in the chondrocytes located in the upper part of the proliferative zone, but not in the hypertrophic zone.However,the phenotypes of the expanded hypertrophic zone and unabsorbed cartilage matrix observed in Axin1OsxKO mice may not be primarily caused by the alteration of β-catenin signaling in hypertrophic chondrocytes. The evidence that the phenotype of the expanded hypertrophic zone was not observed in β-catenin activation mice further suggest that the phenotype observed in early postnatal Axin1 KO mice was not caused by the activation of β-catenin signaling. The detailed mechanism of Axin1 regulation of postnatal bone growth needs to be further investigated.

In conclusion, loss of Axin1 in osteoblast precursor cells led to activation of β-catenin signaling, which in turn upregulated OPG expression and increased the ratio of OPG/RANKL. Osteoclast formation was then inhibited by OPG, which is responsible for decreased apoptosis of hypertrophic chondrocytes and reduced resorption of the mineralized cartilage matrix. However, it is not totally clear whether only a small portion of osteoblast precursor cells was affected leading to the changes in β-catenin signaling in those cells. This possibility needs to be further investigated.

MATERIALS AND METHODS

Animals

The use of animals was approved by Shanghai Laboratory Animal Use Committee. The generation of Axin1flox/floxmice has been reported in previous studies.21Osx-Cre mice were obtained from the Jackson Laboratory. Axin1Osxconditional KO mice were generated by crossing Axin1flox/floxmice with Osx-Cre transgenic mice(Osx-Cre;Axin1flox/flox).β-catenin(ex3)flox/-mice have been used in our previous studies.30,39

Isolation and culture of calvarial osteoblasts

The periosteal layers on both sides of the skull of 3-day-old mice were removed. The calvariae were transferred into a 50 mL conical tube and digested in 1 mg·mL−1collagenase A (Roche, Basel,Switzerland) in serum-free αMEM in a 37°C water bath for 15 min.After two repeated digestions in fresh collagenase A solution, the retained mixtures of collagenase A and cells were filtered into a new tube. The cell suspensions were mixed, centrifuged, and resuspended in αMEM with 10% FBS and 50 μg·mL−1ascorbic acid. The suspension was finally plated in six-well culture plates at a density of 2×105cells per well. The media were changed every other day.

Adenovirus mediated deletion of Axin1 in primary calvarial osteoblasts

Primary osteoblasts were isolated from calvariae of Axin1flox/floxmice and cultured in αMEM with 10% FBS in six-well plates. When cells reach 40% density, Adeno-Cre or Adeno-GFP (Hanbio, Shanghai,China)was added to the culture medium at a concentration of 2×108plaque-forming unit (PFU) per mL. The culture medium was changed 24 h after infection. The cells were collected 40 h after virus infection for real-time PCR and western blot assays.

Real-time PCR assay

Total RNA was extracted from primary calvarial osteoblasts using a RNeasy mini kit (Qiagen, Hilden, Germany). DNAse I-treated total RNA was reverse transcripted using a RT reagent kit (Takara Bio,Tokyo,Japan).The cDNA was amplified by PCR in a total volume of 20 μL reaction solution containing 10 pmol·L−1primers (primer names and sequences are listed in Table 1).

Whole skeleton Alizarin red and Alcian blue staining

After removing skin and adipose tissues,the skeletons were fixed in 95%ethanol for 2 days followed by fixation in acetone for another day. The skeletons were then stained with 0.015% Alcian blue and 0.005%Alizarin red for 3 days.Pictures were taken after most of the soft tissue was digested in 1% potassium chloride solution.

Histological analysis

Tibial samples were fixed in 4% paraformaldehyde, decalcified,dehydrated, and embedded in paraffin. 4-μm thick serial midsagittal sections of tibias were cut and stained with Alcian blue/HEG (ABH) and Safranin O/Fast green (SOF). Histomorphometricanalysis was performed using an Olympus BX50 microscope(Olympus, Tokyo, Japan) and Image-Pro Express software (Media Cybernetics,Rockville,MD,USA). TRAP-positive osteoclast numbers were quantified.

Table 1. Names and sequences of PCR primers

Calcein labeling assay

1 mg·mL−1calcein in saline was injected into 3-week-old mice(i.p. injection, 10 mg·kg−1) followed by a second injection 4 days later. Tibiae of 4-week-old mice were fixed in 4% paraformaldehyde and dehydrated in a series of ethanol(75%-100%).Samples were embedded in methyl methacrylate and 100-μm thick serial midsagittal sections were cut. The fluorescence signal of the cortical bone was observed using an Olympus BX50 microscope(Olympus). The distances of two calcein labels were measured with DP Manager software (Olympus) to determine the MAR.

TRAP staining

Paraffin sections of newborn, 2- and 4-week-old tibias were rehydrated and incubated with a 0.4 mg·mL−1Napthol-Ether solution/basic stock incubation solution at 37°C for 30 min. A 0.04 g·mL−1sodium nitrite solution and a 0.05 g·mL−1pararosaniline dye/2 N hydrochloric acid solution was mixed and added to the basic stock incubation solution and the paraffin sections were incubated in this mixture for 10 min.

Immunohistochemistry (IHC) staining

Paraffin sections of 4-week-old tibiae were rehydrated and digested in 0.1% trypsin for 10 min at the room temperature,and then treated with 3% H2O2for 20 min. Sections were incubated with primary antibodies in PBS overnight at 4°C. Col-X, MMP13, MMP9, cathepsin K, OPG, DKK1, and sclerostin antibodies were obtained from Abcam (Cambridge, MA, USA).Axin1 and β-catenin antibodies were obtained from Sigma (St.Louis,MO,USA).NFATc-1 and c-Fos antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Negative control sections were incubated with IgG (Beyotime Biotechnology,Shanghai, China). A Polink-2 plus polymer HRP detection kit (PV-9001, ZSGB-BIO, Shanghai, China) was used for incubation with secondary antibody and horseradish peroxidase(HRP)-streptavidin.

In vitro osteoclast differentiation assay

BMM cells were isolated from long bone of 1-month-old WT mice and plated into 24-well culture plates at a density of 4×105cells per well and cultured in αMEM with 10% FBS. Cells were treated with 20 ng·mL−1macrophage colony-stimulating factor (M-CSF)for 3 days,and then switched to the medium with 10 ng·mL−1MCSF and 50 ng·mL−1RANKL for 7 days. To test Axin1OsxCM, BMM cells were treated with CM collected from cultured calvarial osteoblasts isolated from Axin1OsxKO mice and Cre-negative controls for 7 days. The culture medium was changed every 3 days. TRAP staining was performed using a TRAP assay kit(Sigma, St. Louis, MO, USA).

Statistical analysis

An unpaired Student's t test was performed for experiments involving two groups.P ＜0.05 was considered statistically significant.

DATA AVAILABILITY

All data and materials used in the analysis are available to any researcher for purposes of reproducing or extending the analysis.

ACKNOWLEDGEMENTS

This work was supported by the following funding agencies. (1) National Natural Science Foundation of China (NSFC) (81973876, 81673991 to BS, 81730107 to YJW and 81603643 to YJZ).(2)The National Key R&D Program of China(2018YFC1704302 to YJW). The Program for Innovative Research Team in University, Ministry of Education of China (IRT1270 to YJW). The Program for Innovative Research Team,Ministry of Science and Technology of China (2015RA4002 to YJW). The Three Years Action to Accelerate the Development of Traditional Chinese Medicine Plan (ZY(2018-2020)-CCCX-3003 to YJW). (3) National Natural Science Foundation of China(NSFC) (81672227) and a Frontier Science of CAS grant (QYZDB-SSW-JSC030) to HP.National Natural Science Foundation of China (NSFC) (81991513) to GX.

ADDITIONAL INFORMATION

The online version of this article (https://doi.org/10.1038/s41413-020-0104-5)contains supplementary material, which is available to authorized users.

Competing interests:The authors declare no competing interests.