Sfrp4 repression of the Ror2/Jnk cascade in osteoclasts protects cortical bone from excessive endosteal resorption

Sfrp4 Cell-Autonomously Regulates Osteoclast Differentiation and Activity.

Sfrp4 is a major determinant of both cortical and trabecular bone (13) and is equally expressed in these two bone compartments (SI Appendix, Fig. S1A). In cortical bone, Sfrp4^−/− mice display an increased number of endosteal OCs (Table 1) (13) and the increase in Rankl/Opg ratio seen in the cortical bone of these mice (13) suggests an Sfrp4 osteoblast-dependent effect on OCgenesis. As shown in Fig. 1A, Sfrp4 is also expressed by bone marrow macrophages (BMMs) and its expression increases significantly during Rankl-induced OCgenesis. We found that OC formation was significantly enhanced in Sfrp4^−/− cultures compared with wild-type (wt) cultures, as indicated by the increase in tartrate-resistant acid phosphatase (TRAP⁺) multinucleated cells (MNCs) and the expression of OC-specific genes including Nfatc1, c-fos, Rank, Dc-Stamp, Trap, and Cathepsin K (CtsK) (Fig. 1 B–D). Moreover, in the absence of Sfrp4, OC activity was also significantly increased as shown by the area of resorption pits on the dentin slides seeded with Sfrp4^−/− cultures (Fig. 1 E and F). Thus, these results indicate that OC-expressed Sfrp4 cell-autonomously regulates OC differentiation and activity. Of note, Sfrp1 and Sfrp2 gene expression, but not that of Sfrp3 and Sfrp5, was increased in Sfrp4^−/− cultures compared with wt cultures (SI Appendix, Fig. S1B). Supporting a role for Sfrp4 in OCgenesis, we found that sFrp4 treatment dose-dependently suppresses the ability of BMMs to respond to Rankl-induced OC differentiation (Fig. 2). To confirm these paracrine and autocrine effects of Sfrp4, we performed mixed-and-matched cocultures of calvarial OBs (cOBs) and BMMs isolated from wt or Sfrp4^−/− mice. We first asked whether Sfrp4 expression is mainly expressed by cOBs or OCs and found that its expression was similar in these cell types (Fig. 3A). A significantly higher number of TRAP⁺ MNCs were formed in Sfrp4^−/− cOB/wt BMM cocultures and wt cOB/Sfrp4^−/− BMM cocultures than in wt cOB/wt BMM cocultures. Importantly, an additive effect was found when Sfrp4^−/− cOBs were cocultured with Sfrp4^−/− BMMs (Fig. 3 B and C). We then treated BMMs with or without sFrp4 2 d after inducing OCgenesis and found that the number of TRAP⁺ MNCs was significantly decreased in the sFrp4-treated cells compared with untreated cells (SI Appendix, Fig. S1 C and D), suggesting that Sfrp4-regulated signaling can modulate early and late OC precursor differentiation. Importantly, as shown in SI Appendix, Fig. S2, TUNEL assay clearly showed that Sfrp4 suppresses OCgenesis without affecting cell viability. Collectively, these findings reveal combined OC- and OB-expressed Sfrp4 effects on OCgenesis, likely reflecting the in vivo phenotype seen on the endosteal surface of Sfrp4^−/− mice (13).

Sfrp4 Mediates Osteoclastogenesis via the Noncanonical Wnt/Ror2/Jnk Cascade.

As expected, Sfrp4 deletion in OCs led to activation of both canonical Wnt/β-catenin and noncanonical Wnt/Jnk cascades (Fig. 4A and SI Appendix, Fig. S3A). Accordingly, sFrp4 treatment significantly suppressed both cascades in wt BMMs (SI Appendix, Fig. S3B). Given that canonical Wnt/β-catenin signaling suppresses OC differentiation while the noncanonical Wnt/Ror2/Jnk cascade promotes it (26, 28, 31), we hypothesized that Sfrp4 relies on the latter cascade to modulate OCgenesis. To test this hypothesis, we first blocked canonical Wnt/β-catenin signaling in Sfrp4^−/− cells using increasing doses of the tankyrase inhibitor XAV939, which stimulates β-catenin degradation by stabilizing axin in the destruction complex (32) but not Jnk phosphorylation (Fig. 5A). As shown in Fig. 5 B and C, blocking canonical Wnt/β-catenin signaling had an additive effect on the increase in OC differentiation seen in Sfrp4^−/− cultures. We then used Cre-ER^T2;Lrp5/6^fl/fl mice and generated Lrp5/6^fl/fl deficient BMMs upon treatment with tamoxifen (Tam) in vitro (Fig. 5D). As shown in Fig. 5 E and F, knockout of Lrp5/6 coreceptors did not affect the ability of sFrp4 to suppress OCgenesis. In contrast, blocking the noncanonical Wnt/Ror2/Jnk cascade pharmacologically, via the selective Jnk inhibitor (Sp600125) (28) (Fig. 6A), significantly impaired the Sfrp4 deficiency-dependent increased OCgenesis (Fig. 6 B and C). Confirming these data, knockout of the Ror2/Jnk axis, via in vitro excision of Ror2 in Cre-ER^T2;Ror2^fl/fl BMMs, significantly reduced the ability of sFrp4 to block OCgenesis (Fig. 6 D–F). Moreover, sFrp4 significantly suppressed the Wnt5a-dependent induction of OCgenesis by blocking Wnt5a downstream activation of the Ror2/Jnk signaling cascade (Fig. 7) (28, 29). Excluding an overall change in Wnt ligand expression in the absence of Sfrp4, Wnt5a expression, as well as that of other key Wnts, was not changed in either Sfrp4^−/− OC cultures or in cortical bone (SI Appendix, Fig. S4 A and B).

Recent studies have reported that Wnt3a can inhibit OC differentiation via a β-catenin–independent cAMP/PKA/pCreb pathway (31); we therefore assessed whether Sfrp4 might also utilize these alternative cascades. As shown in SI Appendix, Fig. S5, this pathway is not affected in Sfrp4^−/− cultures. Likewise, the NF-κB/Tak1 and MAPK signaling pathways (that include ERK and P38 MAPKs), also involved in OCgenesis (31, 33–35), are unchanged in Sfrp4^−/− cultures (SI Appendix, Fig. S5), therefore excluding a potential role for this signaling cascade downstream of Sfrp4 in BMMs. Similarly, and in contrast with OB-lineage cells (13), deletion of Sfrp4 in OCs does not affect BMP signaling (SI Appendix, Fig. S4C). Altogether, these data demonstrate a key role for the Sfrp4/Ror2/Jnk axis in OCgenesis and support the hypothesis that in the absence of Sfrp4, noncanonical Wnt/Ror2/Jnk signaling activation in OCs overrides the negative effect of canonical Wnt/β-catenin signaling activation on OCgenesis.

Ror2 Ablation in OCs In Vivo in Sfrp4^−/− Mice Prevents the Increase in Endosteal Resorption and Partially Rescues Cortical Bone Mass.

These results prompted us to analyze whether Ror2 signaling is required for the action of Wnts on OCs in the absence of Sfrp4 in vivo. For this purpose, we used male mice since Sfrp4 deletion results in the same phenotype in males and females in both humans and mice (13, 21). We deleted Ror2 in OCs using the CtskCre mice (Ror2_OC), and crossed them with Sfrp4^+/− mice to obtain Ror2^fl/fSfrp4^+/+, Ror2_OCSfrp4^+/+, Ror2^fl/flSfrp4^−/−, and Ror2_OCSfrp4^−/− mice. Ror2_OC mice develop high trabecular bone mass as a consequence of a significant decrease in OC number and function (28) but the cortical phenotype of these mice has not been described. Microcomputed tomography (μCT) analysis of the cortical midshaft demonstrated that, at 5 wk of age, deletion of Ror2 in OCs does not significantly affect cortical bone. However, blocking the Ror2/Jnk axis in OCs in growing Sfrp4^−/− mice significantly rescued the thinning of their cortical bone as well as their cortical area, although these parameters remain lower than in control and in Ror2_OC mice (Fig. 8 A and B and SI Appendix, Table S1). In contrast, the marrow area and the total bone area were not rescued in Ror2_OCSfrp4^−/− mice (SI Appendix, Table S1). Bone histomorphometry analysis of the cortical midshaft revealed that deletion of Ror2 in OCs in Sfrp4^−/− mice prevents the increase in endosteal OC number and surface and significantly increases cortical thickness and cortical bone volume, although they remain significantly lower than in Ror2^fl/fSfrp4^+/+ mice (Fig. 8 C and D, Table 1, and SI Appendix, Fig. S6A). No significant difference in these parameters was seen in Ror2_OCSfrp4^+/+ mice; however, a decrease in both the OC number and surface was observed (Fig. 8D, Table 1, and SI Appendix, Fig. S6A). The increase in marrow area was not rescued by Ror2 OC-specific deletion in Sfrp4^−/− mice and, as expected since the Ror2/Jnk cascade is deleted only in OCs, the decreased endosteal mineral apposition rate formation and bone formation rate were not rescued in Ror2_OCSfrp4^−/− mice (Table 1), explaining the only partial rescue of cortical thickness. μCT analysis of trabecular bone confirmed that at 5 wk of age, deletion of Ror2 in OCs leads to increased trabecular bone volume and number as well as connectivity and decreased trabecular spacing and structure model index (SMI) (SI Appendix, Fig. S6B and Table S2). Consistent with our previous findings (13), Ror2^fl/flSfrp4^−/− mice display increased trabecular bone mass and Ror2 OC-specific deletion in Sfrp4^−/− mice has an additive effect on trabecular bone volume, connectivity, and SMI compared with Ror2^fl/fSfrp4^+/+ mice (SI Appendix, Fig. S6B and Table S2). These findings therefore support our hypothesis that Sfrp4 secreted in cortical bone by OB-lineage cells and/or by OCs themselves suppresses endosteal OCgenesis, at least in part, by blocking the noncanonical Wnt/Ror2/Jnk signaling cascade in OCs, an effect that in turn regulates cortical bone homeostasis.

Animals.

Sfrp4 null mice were previously described (13). CtsKCre mice were kindly provided by S. Kato, University of Tokyo, Tokyo, Japan, and Lrp5/6^fl/fl mice were kindly provided by B. Williams, Van Andel Research Institute, Grand Rapids, MI. All animal studies are described in SI Appendix.

Cell Culture.

Bone marrow macrophages were isolated from 6- to 8-wk-old wt and Sfrp4^−/− mice as previously described (46). OC generation, treatment, and mixed-and-matched experiments are detailed in SI Appendix.

Tartrate-Resistant Acid Phosphatase Staining.

TRAP staining was performed according to the manufacturer’s protocol (Sigma-Aldrich). The number of TRAP⁺ cells with 2 or more nuclei per well was counted. BMMs from 5 to 9 distinct mice per genotype were used.

Bone Resorption Pit Assay.

BMMs isolated from wt and Sfrp4^−/− mice were treated with 30 ng/mL M-CSF and 10 ng/mL Rankl (both from R&D Systems) for 4 d to induce OCgenesis. Pit assay was performed as detailed in SI Appendix.

TUNEL Assay.

Wt BMMs were cultured with 30 ng/mL M-CSF and 10 ng/mL Rankl w/wo sFrp4 (5 to 20 μg/mL) (R&D Systems) for 4 d. TUNEL assay was performed using an In Situ Cell Death Detection Kit, TMR red (Roche; 12156792910) according to the manufacturer’s protocols.

Canonical Wnt/β-Catenin and Noncanonical Wnt/Ror2/Jnk Signaling Cascade Inhibition.

For pharmacological inhibition and in vitro excision, BMMs were isolated from 6- to 8-wk-old mice and treated as detailed in SI Appendix.

Real-Time PCR.

Total RNA was isolated from cells and cortical and trabecular bone and gene expression was determined as detailed in SI Appendix.

Western Analysis.

Total proteins (10 μg) were resolved by SDS/PAGE under reducing conditions. Antibodies used and methods are detailed in SI Appendix.

Skeletal Phenotype.

For microcomputed tomography scanning, a high-resolution desktop microtomographic imaging system (μCT35; Scanco Medical) was used to assess cortical and trabecular bone morphology as detailed in SI Appendix. Quantitative bone histomorphometric measurements were performed using the OsteoMeasure System as detailed in SI Appendix.

Statistical Analysis.

Data are expressed as the mean ± SEM. Statistical analysis was conducted using unpaired two-tailed Student’s t test. For comparison of three or more groups, two-way ANOVA followed by Tukey’s multiple comparisons test for all groups was used. GraphPad PRISM 7 was used for statistical analysis. P < 0.05 was considered statistically significant.