1. XAF1 was increased in both osteoporotic patients and OVX mice
To explore the specific gene in OCgenesis and osteoporosis process, RNA-seq was performed between OVX and sham mice. Notably, pathway enrichment analysis of these genes revealed significant enrichment for the OC differentiation, IFN signaling and apoptosis, which are essential for OC function and survival (Fig. 1A). Meanwhile, among all the differentially expressed genes (DEGs) in the femurs treated with OVX, IFN and apoptosis signaling were highly and significantly enriched (Fig. 1B). To further explore the common target for osteoporosis both in human and mice, data from GSE230665 was utilized to overlap with data from mice, demonstrating 489 DEGs in common (Fig. 1C). In the view of the significance of apoptosis and IFN, we next overlap DEGs with interferon-stimulated genes and apoptosis genes, XAF1, IFIT2 and IRF1 were screened out (Fig. 1D). The intimate connection between XAF1, IFIT2 and IRF1 expression and OC function prompted us to explore the pathological involvement of them in human osteoporosis. We obtained bone specimens from patients with osteoporosis (OP) and normal bone mineral density. The source of normal BMD bone specimens was lamina bone of young patients with lumbar disc herniation for laminectomy surgery. RT-PCR was performed, indicating that XAF1 increased with TRAP, CTSK and MMP9 in OP patients instead of IRF1 and IFIT2 (Fig. 1E, S1A). The expression of the XAF1 protein in the OP group was higher than that in the control group (Fig. 1F). These data suggested that XAF1 be closely associated with osteoporosis. Hence, Xaf1 global knockout mice was generated as displayed (Fig. 1G) and knockout efficiency was presented in Supplementary Fig. 1B-C.
2. Global deletion of Xaf1 promoted RANKL-induced osteoclastogenesis
To determine the influence of Xaf1 targeting OCs, BMMs were isolated from WT and Xaf1-/- mice and stimulated with RANKL for 7 d. Mature OCs were assessed by TRAP staining and bone resorption assay. Xaf1-/- BMMs formed more TRAP+ multinucleated OCs (Fig. 2A-B) and exhibited increased bone resorption capacity than BMMs derived from WT mice, manifested as increased number and area of OCs as well as more resorption pits and trails on the bone slices (Fig. 2C-D). Consistent with results above, a striking size increase in actin ring structure was observed (Fig. 2E) and both the number of nuclei per osteoclast and actin rings per osteoclast were increased via phalloidin staining in Xaf1-/- OCs (Fig. 2F). Moreover, a significant increase in the OC-specific gene expression, such as Acp5, Oscar, Atp6v0d2, Ctsk, Dcstamp, and Mmp9 were observed in Xaf1−/− BMMs with RANKL stimulation (Fig. 2G). Subsequent western blot analyses of the key OC-specific makers NFATc1, c-Fos, MMP9, TRAP and cathepsin K (CTSK) showed that RANKL induced the expression of all these markers, and Xaf1 deletion augmented RANKL-induced expression of all markers compared with WT during OCgenesis (Fig. 2H).
ALP (Fig. S2A-B) and alizarin red staining (Fig. S2C-D) were used to determine osteogenic activity, which demonstrated no difference between WT and Xaf1−/− BMSCs. Likewise, gene expression levels of a series of osteoblast markers, including Runx2, Osterix, Ocn, and Osteopontin were indistinguishable between WT and Xaf1−/− BMSCs (Fig. S2E). In summary, depletion of XAF1 did not significantly affect osteoblast differentiation in vitro. Together, these results indicated that Xaf1 deletion selectively enhanced OC generation without affecting osteogenesis in vitro.
3. Xaf1 deletion exacerbated OVX-induced bone loss in vivo
After establishing the cellular effects of Xaf1 on OCs, we sought to define its role in vivo. OVX or sham surgeries were performed in 8-week old WT and Xaf1−/− mice. Reconstructed three-dimensional (3D) imaging was conducted using micro-computed tomography (μCT) (Fig.3A). OVX treated Xaf1−/− mice revealed decreased bone mass, BMD, BV/TV, Tb.N, as well as increased Tb.Sp compared with OVX treated WT mice (Fig.2B).
Complementing these findings, serum was analyzed for the concentrations of RANKL, which promoted OCgenesis, and its regulatory decoy receptor OPG. Xaf1 deletion significantly increased the RANKL/OPG ratio in blood serum (Fig. 3C). Elevated OC formation and bone resorption were often the underlying causes of low-bone mass osteoporotic phenotypes. In order to determine the number and activity of OCs, femur bone slices were stained with TRAP and H&E. As seen in Fig. 3D-E, femoral sections from Xaf1−/− mice had higher numbers of OCs than WT mice and exhibited low-bone mass osteoporotic phenotype in TRAP-stained and H&E-stained tissue samples, respectively. This observation revealed that Xaf1 may have an osteoprotective impact, which accounted for the decreased bone mass in Xaf1−/− mice.
4. Xaf1 deletion exacerbated Ti-particle induced osteolysis
Osteolysis was an inflammatory bone loss caused by excessive OC formation and activity. Given the effect of Xaf1 on noninflammatory osteoporosis, we considered if Xaf1 conferred additional protection during osteolysis. To induce osteolysis, Ti particles were injected onto the calvaria of WT and Xaf1−/− mice. μCT analysis revealed that the Xaf1−/− mice exhibited a significant decrease in osteolysis compared to WT mice (Fig. 4A). Morphological analysis revealed a significant reduction in BV/TV and an increase in the number of pores and the percentage of porosity from Xaf1−/− mice compared to WT mice (Fig. 4B). Serum levels of RANKL were upregulated from Xaf1−/−mice, while OPG levels were comparable, resulting in the increase of RANKL/OPG ratio, a parameter for assessing OCgenesis (Fig. 4C). Collectively, these results indicated that Xaf1 deletion exacerbated bone loss in Ti-particle-induced osteolysis.
Histological analysis was performed on calvarial sections to validate μCT results. H&E staining showed significant bone defects (Fig. 4D) and increased proportion of eroded surface in Xaf1−/− mice (Fig. 4E). TRAP staining confirmed that the activity of osteoclasts along the surface of trabecular bone was higher in Xaf1−/− mice than in WT mice (Fig. 4D). Osteoclast number per bone perimeter (N.Oc/B.pm) was larger in in Xaf1−/− mice than in WT mice (Fig. 4F). Together, these results suggested that depletion of Xaf1 deteriorated osteolytic bone loss by aggravating OCgenesis in vivo.
5. Xaf1 deletion chimeras resulted in a low bone mass phenotype
In fact, global Xaf1 deletion led to several nonskeletal pathologies, making it challenging to distinguish between direct effects of Xaf1 deletion on bone and secondary effects on other organs. Then, bone marrow transplantation (BMT) from the WT and Xaf1−/− donors was performed to generate WT mice containing macrophages deficient in Xaf1 (Xaf1−/−→WT) and control mice (WT→WT) (Fig. 5A). The efficiency of BMT was verified using flow cytometry (Fig. S3A). Similar with global deletion, Xaf1−/− chimeric mice illustrated bone loss phenotype in OVX model (Fig. 4B). Morphometric analyses of trabecular including BV/TV, BS/TV, Tb. N and Tb.pf confirmed low bone mass in Xaf1−/− chimeric mice (Fig. 5C). H&E and TRAP staining showed decreased bone mass, trabecular area, and increased OC number in Xaf1−/− chimeric mice (Fig. 5D-E). Moreover, the elevated serum levels of RANKL and RANKL/OPG ratio in Xaf1−/− chimeric mice compared to WT chimeric mice further confirmed higher bone resorption activity (Fig. S3B). No difference in serum OPG content was found between Xaf1−/− chimeric mice and WT chimeric mice (Fig. S3B).
Meanwhile, severe bone loss in osteolysis model was observed in Xaf1−/− chimeric mice (Fig. 5F), confirmed by morphometric (Fig. 5G) and histological analysis (Fig. 5H), wherein an increase in the number and activity of OCs was observed (Fig. 5I). Serum levels of RANKL and RANKL/OPG ratio indicated a severe osteolysis phenotype (Fig. S3C). Both Xaf1 global depletion and Xaf1−/− chimera mice generation promoted osteoclastic bone resorption in the OVX and osteolysis model. Together, our findings demonstrated that Xaf1 deficient in macrophages demonstrated a low-bone mass phenotype due to elevated OC number and activation.
6. Xaf1 deletion inhibited the apoptosis of OCs
Previous studies had demonstrated that mature OCs undergo spontaneous apoptosis. However, multinucleated OCs derived from Xaf1−/− mice exhibited resistance to apoptosis due to the removal of cytokines compared with OCs derived from WT mice (Fig. 6A-B). These findings led us to examine induction of apoptosis in Xaf1−/− OCs. Therefore, we performed RT-PCR to determine the involvement of Xaf1 in OC apoptosis, which confirmed downregulation of pro-apoptotic Bax and upregulation of anti-apoptotic Bcl2 and Xiap in Xaf1−/− cells during OCgenesis (Fig. 6C). Meanwhile, inactivation of caspase-3/7, as indicated by cleaved caspase-3/7, was also observed in BMMs derived from Xaf1−/− mice throughout the differentiation process (Fig. 6D). When compared to the WT BMMs, Xaf1−/− BMMs had significantly fewer cells that were positive for DNA fragmentation (TUNEL+) after 3 days following RANKL stimulation, which suggested that Xaf1−/− BMMs suppressed spontaneous apoptosis during OCgenesis (Fig. 6E-F). Flow cytometry was performed to determine OC apoptosis, indicating a decreased percentage of OC apoptosis in Xaf1−/− BMMs treated with RANKL (Fig. 6G-H). These findings collectively suggested that Xaf1 deficiency causes excessive OC generation via inhibiting apoptosis.
7. BV6 (XIAP inhibitor) suppressed RANKL-induced OC formation
Xaf1 was identified as an XIAP inhibitor gene, which was verified by STRING analysis (Fig. 7A). RT-PCR (Fig. 7B) and western blot (Fig. 7C) analysis demonstrated that Xaf1 experienced a complementary change with XIAP during OCgenesis. BV6, a XIAP inhibitor, was used to determine if Xaf1 regulated OCgenesis via XIAP in vitro. CCK-8 assays were conducted to determine the viability of BMMs treated with different concentrations of BV6 for 24 and 48 hours, indicating that BV6 at concentrations up to 5 μM had no discernible effect on the viability of BMMs at all time points analyzed (Fig. S4A).
To demonstrate how BV6 stimulates OCgenesis, we treated BMMs with BV6 at non-toxic concentrations for 7 days. However, a dose-dependent reduction in the number and size of OCs was observed in the BV6-treated groups (Fig. 7D), which was validated by number and area quantification of TRAP+ OCs (Fig. 7E). Simultaneously, RT-PCR analyses showed that BV6 treatments dose-dependently induced the transcription of OC-specific genes in BMMs, including Oscar, Dcstamp, Ctsk, Atp6v0d2, Mmp9, and Acp5 (Fig. 7F). By using total cellular proteins extracted from BMMs stimulated with BV6, we observed the dose-dependent inhibition of NFATc1, c-Fos, MMP9, CTSK and TRAP (Fig. 7G). Thus, BV6 suppressed RANKL-induced OCgenesis and inhibited OC-specific gene and protein expression in a dose-dependent manner, implying XIAP functioned as XAF1 downstream signaling to mediate OC generation and function.
8. XAF1 interacted with XIAP to promote the caspase-3 activity
After establishing the cellular effects of Xaf1, we aimed to elucidate the underlying molecular mechanism. It’s been known that XIAP directly interacted with caspase and inhibits cell apoptosis. Based on these results, we hypothesized that XAF1 antagonize interaction between XIAP and caspase, thereby promoting OC apoptosis. Co-immunoprecipitation assay was performed to validate the endogenous interaction between XIAP and caspase3/7 (Fig. 8A, B). Furthermore, in situ proximity ligation assay (PLA) revealed a direct interaction between endogenous XIAP and caspase3 in BMMs (Fig. 8C), which confirmed by increased number of XIAP-caspase3 interaction complex (Fig. 8D). Moreover, caspase3 activity was decreased in Xaf1−/− BMMs treated with RANKL compared to WT BMMs (Fig. 8E). These results were consistent with the effect of Xaf1 in antagonizing XIAP-caspase interaction.