Amizona Scientific LLC (Birmingham, AL, United States) provided recombinant murine RANKL and macrophage colony stimulating factor (M-CSF), fetal bovine serum (FBS), and minimal essential medium α (α-MEM). Ssa was purchased from MedChemExpress (Shanghai, China). The Tartrate-resistant acid phosphatase (TRAP) Staining kit was obtained from Sigma-Aldrich (St. Louis, MO, United States). The Cell Counting Kit-8 (CCK-8), Lipid Peroxidation Probe -BDP 581/591 C11, Iron Assay Kit-Colorimetric kit, and FerroOrange were obtained from Dojindo Molecular Technologies (Kumamoto, Japan). Thermo Fisher Scientific (Waltham, MA, United States) provided the MitoSOX Red mitochondrial superoxide indicator. Primary antibodies against nuclear factor erythroid 2-related factor 2 (Nrf2), solute carrier family seven member 11 (SCL7a11), cathepsin K and glutathione peroxidase 4 (GPX4) were purchased from HUABIO (Hangzhou, China). Antibodies against Fos proto-oncogene, AP-1 transcription factor subunit (c-Fos), TNF receptor associated factor 6 (TRAF6), and β-actin were obtained from Abcam (Cambridge, United Kingdom. Bioworld (Nanjing, China) provided the goat-anti-rabbit and secondary antibodies. The enzyme-linked immunosorbent assay (ELISA) kits for osteoprotegerin (OPG), ACP5, and the alkaline phosphatase (ALP) biochemical kits were obtained from Nanjing Jiancheng (Nanjing, China). Beyotime (Shanghai, China) provided the remaining reagents and kits.

C57BL/6 mice (six to eight weeks old) were sacrificed, and their long bones were removed, to isolate bone marrow-derived macrophages (BMMs). Soft tissues were removed from mice’s femurs and tibias. After flushing the bone marrow cavities of the long bones using α-MEM containing 10% FBS, we isolated bone marrow suspensions. The next day, cells were separated from the suspension and then the erythrocytes were lysed. The nonadherent cells (BMMs) were cultured in α-MEM for subsequent use. M CSF(40 ng/mL) was used in all experiments to maintain BMMs survival. To stimulate osteoclast differentiation, 50 ng/mL of RANKL was administered. Next, 4% formalin was used to fix the cells for 30 min, followed by TRAP staining. We identified osteoclasts as TRAP positive cells with > three nuclei.

According to the manufacturer’s instructions, cell viability was measured using a CCK-8 kit. 5 × 10 4 cells were inoculated into a 96-well plates and grown for 24 h. Subsequently, various doses of Ssa (3.125, 6.25, 12.5, and 25 µM) were used to treat the cells for 96 h, with or without RANKL. To analyze the cause of cell death, BMMs were administered with 10 μM of ZVAD-FMK, 10 μM of necrostatin-1, and 2 μM of Ferrostatin-1 (Fer-1), for 96 h, with or without Ssa (6.25 μM). CCK-8 (10 μL) was added and incubated for 1.5 h at 37°C. A microplate reader was then used to measure the absorbance at 450 nm.

BMMs were grown in medium comprising 40 ng/mL of M-CSF, 50 ng/mL of RANKL, and different amounts of Ssa. Four days later, mature osteoclasts were identified, followed by fixation for 15 min. The cells were incubated with Actin-Tracker Red Rhodamine to visualize the osteoclast cytoskeletons and 4, 6-diamidino-2-phenylindole (DAPI) was used to identify nuclei. A fluorescence microscope was then used to image the F-actin rings.

BMMs were seeded on bone slices and stimulated with M-CSF and RANKL for 10 days, with or without Ssa (6.25 and 12.5 μM). A scanning electron microscope (ZEISS GeminiSEM 300, Oberkochen, Germany) was employed to view the effect the BMMs on the bovine bone slices.

Based on the instructions provided by the manufacturer, an iron assay kit was used to determine the intracellular iron concentration. The FerroOrange fluorescent probe enables live-cell fluorescent imaging of intracellular Fe2+. The cells were added with medium without FBS, but with 1 μmol/L of the FerroOrange reagent. Following incubation for 30 min, a fluorescence microscope was used to visualize the intracellular iron ions.

Dichloro-dihydro-fluorescein diacetate (DCFH-DA) and MitoSOX™ Red were used to detect the intracellular and mitochondrial ROS, respectively. Cells were incubated with DCFH‐DA in a 37°C incubator for 30 min or with MitoSOX™ at 37°C for 10 min. Following three washes with 1 × phosphate-buffered saline (PBS), a microscope was used to view the cells.

A Lipid Peroxidation MDA assay kit was used to determine the MDA content in cell lysates, and the absorbance at 532 nm was measured using a microplate reader. BMM GSH contents were assayed using a GSH and GSSG Assay Kit following the supplier’s protocol. The total protein level, as assessed using a BCA Protein Assay Kit, was used to normalize the MDA and GSH assay results.

Lipid peroxidation was detected using boron-dipyrromethene (BODIPY™) 581/591 C11 (2 μmol/L), which was incubated directly with cells for 30 min at 37°C. Fluorescence was detected using a confocal microscope.

JC-1 fluorescence mitochondrial imaging was used to determine the mitochondrial membrane potential. JC-1 was incubated with the cells 20 min at 37°C. The cells were then washed twice using JC-1 buffer. The fluorescent images were captured using a laser scanning confocal microscope. The mitochondrial membrane potential was represented by the ratio of red to green fluorescence.

The TRIzol reagent was used to isolate total RNA according to the supplier’s guidelines and reverse transcribed to cDNA employing a Prime Script RT-PCR kit (TAKARA Korea, Seoul, Korea). The cDNA was used as the template for a quantitative real-time PCR reaction carried out using SYBR mix in the CFX384 real-time system (BioRad, Hercules, CA, United States). The primers utilized were synthesized by Sangon Biotech (Shanghai, China) and are shown in Table 1.

Cell lysis buffer was used to lyse the cells, whose total protein concentration was quantified using the BCA protein assay kit. 4%–12% SDS-PAGE was used to separate the proteins in the cell lysate, which were eletrotransferred onto polyvinylidene fluoride membranes. 5% non-fat milk was used to block the membranes for 2 h, followed by incubation with the primary antibodies overnight at 4°C. Next day, the membranes were reacted with anti-rabbit secondary antibody conjugated with horseradish peroxidase (HRP) (1:10,000) for 1 h at room temperature. Finally, an Amersham Imager 680 image system (Amersham United Kingdom) was used to capture the immunoreactive protein signals.

Cells were washed three time using PBS, fixed using 4% paraformaldehyde at room temperature for 30 min, blocked using 5% (w/v) BSA in PBS-Tween20 (PBST), and then immunoassayed using anti-cathepsin K (CTSK), anti-Nrf2, and anti-c-FOS antibodies at 4°C overnight. Next day, the cells were reacted with goat anti-rabbit Alexa Fluor-488-conjugated secondary antibody, washed three times with PBS, DAPI-stained, and observed under a laser scanning confocal microscope.

The Institutional Animal Ethics Committee supervised all the processes and approved the animal experiments. Twenty 10-week-old Wistar rats were divided randomly into four groups of 5: Sham group, LPS group, and LPS + Ssa at low concentration (50 mg/kg body weight) and high concentration (100 mg/kg body weight). Every other day, we injected normal saline into the Sham group, and injected LPS into the other groups, with or without Ssa administration by gavage. After 4 weeks of treatment, all rats were euthanized. Their alveolar bones were fixed for micro-computed tomography (µCT) and histological analysis, and their collected serum was used for biochemical assays and ELISA.

Three-dimensional images of the calvaria were scanned using a high resolution µCT scanner following the supplier’s guidelines.

The alveolar bones of the rats in each group were decalcified, paraffin embedded, sectioned using a microtome at 4 μm thick, and stained using hematoxylin and eosin (H&E) and TRAP for osteoclasts. Immunofluorescence was used to show protein expression in the cells.

Data from the experiments are displayed as the mean ± SD (n ≥ 3). Student’s t-test between the treatment and control groups or analysis of variance (ANOVA) for multiple groups were used to determine the statistical significance. A p-value less than 0.5 was considered to indicate statistical significance.

The toxicity of Ssa toward primary BMMs was assessed. CCK-8 assays suggested that there was no detectable toxicity of Ssa toward BMMs until the concentration reached 12.5 μM (Figures 1A,B). To determine the effect of Ssa on the differentiation of RANKL-induced osteoclasts, BMMs incubated with RANKL and M-CSF were treated with different Ssa concentrations (0, 3.125, 6.25, and 12.5 μM). TRAP staining showed that Ssa markedly inhibited multinucleated osteoclast differentiation in a dose-dependent manner (Figure 1C). As the concentration of Ssa increased, the density and coverage of multinucleated TRAP-positive osteoclasts gradually decreased (Figures 1D,E). Next, the effect of Ssa on bone-resorbing activity was examined, which showed that that Ssa dose-dependently inhibited bone resorption (Figure 1F). Moreover, since F-actin ring formation is neceSsary for osteoclast function, we tested this process using phalloidin. Immunofluorescent staining demonstrated that Ssa markedly inhibited typical F-actin ring formation relative to that in the control group (Figure 1G). The findings clearly demonstrated that Ssa inhibits osteoclast function and bone-resorbing activity in vitro.

We further assessed whether Ssa inhibited the expression of osteoclast‐related genes and proteins. Western blotting demonstrated that the levels of proteins associated with osteoclast differentiation (cathepsin K (CTSK) and c-Fos) were in a dose-dependent manner after Ssa intervention (Figures 2A,B). qRT‐PCR analysis displayed that the mRNA expression levels of genes related to the functions of mature osteoclast, including Acp5 (encoding acid phosphatase 5, tartrate resistant), Nfatc1 (encoding nuclear factor of activated T cells 1), Ctsk and Fos increased markedly at 4 days after RANKL intervention relative to those in the undifferentiated cells, which were decreased in the Ssa-treated groups, especially when the dose is higher (Figure 2C). Immunofluorescence verified that Ssa markedly decreased the expression of c-Fos and CTSK (Figure 2D).

To investigate the impact of Ssa on osteoclast differentiation, BMMs were induced using RANKL and treated with or without Ssa (6.25 μM) for 4 days. Ssa treatment suppressed the viability of RANKL-treated cells. Ferrostatin-1 (a selective ferroptosis inhibitor), Necrotatin-1 (a specific necroptosis inhibitor), and zVAD-fmk (a specific apoptosis inhibitor) were employed to study the mechanism by which Ssa inhibits osteoclast differentiation. Impressively, the CCK-8 assay results showed that ferrostatin-1, but not Necrostatin-1 or zVAD-fmk, markedly inhibited osteoclast death induced by Ssa (Figure 3A). The FerroOrange fluorescent probe was employed to detect the level of ferroptosis. Ssa markedly increased the red fluorescence density (representing the Fe2+ content) compared with that in Control group (Figure 3B). The levels of Fe2+ in cells were also assessed using iron assay kits, which confirmed the results above (Figure 3C). GPX4 is an important ferroptosis regulator, the upstream mediator of which is SLC7A11 (Zhang et al., 2021). Moreover, Nrf2 regulates SLC7A11 to inhibit ferroptosis (Xie et al., 2020; Yuan et al., 2021). Therefore, to determine whether ferroptosis is associated with Ssa-mediated inhibition of RANKL-induced osteoclastogenesis in BMMs, the protein levels of SLC7A11, GPX4, and Nrf2 were determined. Ssa (6.25 μM) markedly reduced Nrf2, SLC7A11, and GPX4 levels in BMMs (Figures 3D,E). Simultaneously, Ssa treatment strikingly reduced Nrf2 nuclear translocation (Figure 3F). Therefore, we concluded that Ssa regulates the Nrf2/SLC7A11/GPX4 axis to promote the ferroptosis of osteoclasts.

Ferroptosis and ROS accumulation are closely related (Lin and Beal, 2006; Dixon et al., 2012); therefore, we assessed total ROS production using DCFH‐DA. RANKL significantly increased ROS levels in BMMs, and further Ssa treatment increased the ROS levels to a greater extent than RANKL treatment alone (Figure 4A). MitoSox‐Red staining showed that the mitochondrial ROS levels were consistent with the changes in intracellular ROS abundance. Collectively, these findings indicated that ROS are crucial for Ssa‐induced ferroptosis (Figure 4B). Next, the levels of oxidative stress-related factors MDA and GSH were detected. Ssa intervention could increase MDA levels (Figure 4C), and decrease GSH levels (Figure 4D). Next, we used BODIPY 581/591 C11 (Figure 4E) and the JC‐1 probe (Figure 4F) to detect the levels of lipid peroxidation and mitochondrial membrane potential (MMP), respectively. The analysis indicated that lipid peroxides accumulated and the MMP increased after the addition of RANKL. The addition of Ssa caused a further increase. Ssa-treated cells exhibited the characteristic morphological features of ferroptosis, including a reduction in mitochondrial membrane density and corresponding volume, diminished or vanished mitochondrial cristae, and a rupturing of the outer membrane (Figure 4G).

The ferroptosis inhibitor ferrostatin-1 was administered to further verify above conclusion. BMMs were classified into four groups: The RANKL group, the RANKL + Ssa group, the RANK + Fer-1 group, and the RANKL + Ssa + Fer-1 group. The RANKL + Ssa + Fer-1 group showed markedly increased Fe2 + levels compared with those in the RANK + Ssa group (Figure 5A). The levels of ferroptosis-associated proteins GPX4, SLC7A11, and Nrf2 increased markedly and the levels of osteoclast differentiation proteins CTSK and c-Fos decreased significantly (Figures 5B,C). ROS levels in the RANKL + Ssa + Fer-1 group increased markedly relative to those in the RANKL + Ssa group (Figure 5D). The MDA level increased and the GSH level decreased in the RANK + Ssa + Fer-1 group compared with those in the RANKL + Ssa group (Figures 5E,F). Subsequently, the amount of lipid peroxides in the RANKL + Ssa + Fer-1 group was increased markedly significantly increased relative to that in the RANKL + Ssa group (Figure 5G). These results showed that ferrostatin-1 reversed the inhibitory effect of Ssa on RANKL-induced osteoclastogenesis, suggesting that Ssa inhibits RANKL-induced osteoclastogenesis by promoting ferroptosis.

The Nrf2 activator TBHQ was used to confirm the role of the Nrf2 signaling pathway in the ferroptosis induced by Ssa in osteoclasts. Immunofluorescence assay of Nrf2 expression confirmed that TBHQ significantly increased the fluorescence signal of Nrf2 (Figure 6A). TRAP staining demonstrated markedly increased osteoclastogenesis in the RANKL + Ssa + TBHQ relative to that in the RANKL + Ssa group (Figures 6B,C). Compared with that in the RANKL + Ssa group, the iron and ROS contents decreased in the RANKL + Ssa + TBHQ group (Figures 6D,E). Fluorescence detection of the lipid peroxide products showed that the RANKL + Ssa group had lower levels of lipid peroxide products after TBHQ treatment (Figure 6F). Collectively, these results revealed that Ssa could promote osteoclast ferroptosis by inhibiting the Nrf2/SCL7A11/GPX4 signaling pathway, which attenuated alveolar bone resorption in the rat model of periodontitis.

To determine whether Ssa could alleviate LPS-mediated bone resorption, a rat model of periodontitis was induced using LPS (illustrated diagrammatically in Figure 7A). The reconstructed 3D model overall morphology and palatal linear bone loss are shown in Figure 7B. The LPS group had the greatest extent of alveolar bone loss. Compared with that in the Sham group, the distance between the cementoenamel junction (CEJ) and the alveolar bone crest (ABC) for the first molar increased markedly relative to that in the Sham group, thus verifying the successful construction of the experimental model of periodontitis. Ssa decreased the CEJ–ABC distances. CT analysis showed that model group had marked bone loss, accompanied by decreased thickness (Tb.Th) and the trabecular number (Tb.N), together with increased trabecular spaces (Tb.Sp) relative to that in the LPS group. Compared with the model group, Ssa treatment reduced alveolar erosion and reversed the trabecular parameters. The treatment effect in the high-dose Ssa group was better than that of low-dose Ssa group (Figure 7C). H&E staining revealed that compared with the Sham group, the attachment of collagenous fibers was destroyed and the inflammatory cell infiltration into the epithelial layer and connective tissue was significantly reduced in the groups treated with intragastric administration of Ssa (Figure 7D). In addition, TRAP staining revealed an increase in the number of large osteoclasts in the Sham group compared with the reduced osteoclast formation in the LPS + Ssa group (Figure 7E). Serum indicators showed significantly increased ACP5 and ALP levels in the LPS group, suggesting enhanced osteoclastogenesis and repair of osteogenesis, meanwhile the levels of OPG, an osteoclast suppressive cytokine, were decreased. Ssa treatment effectively reversed the levels of these biomarkers compared with those in the LPS group (Figure 8A). Moreover, Ssa caused a decrease in GPX4 on the inflammatory alveolar bone surfaces (Figure 8B). Thus, Ssa could decrease osteoclastogenesis to attenuate alveolar bone resorption in the rat model of periodontitis and attenuated bone loss by inhibiting the Nrf2/SCL7A11/GPX4 signaling pathway without osteoblast toxicity.