α-asarone (ASA) was purchased from MedChemExpress. Dimethyl sulfoxide (DMSO), TRAP staining kit were purchased from Sigma-Aldrich. CCK8 kit was from Dojindo Molecular Technology. ɑ-modified eagle’s medium (ɑ-MEM), fetal bovine serum (FBS), penicillin and streptomycin (PS) were from Thermo Fisher Scientific. M-CSF and RANKL were obtained from Novoprotein Scientific Inc. TRIZOL was purchased from Invitrogen, reverse transcript reagents and SYBR Green PCR Master Mix were from Takara Biotechnology. Primers were synthesized from Invitrogen. Phalloidin-iFlour™ 555 Conjugate was from AAT bioquest. Inc. Antibodies for c-fos, NFATc1, JNK1/2, p-JNK1/2, ERK1/2, p-ERK1/2, p-p38, p38, p-p65, p65, p-IκB-α, AKT, p-AKT, CSTK, TRAP, MMP9 and β-actin were from Cell Signal Technology.

The procedure of obtaining BMMs for OCs differentiation was a standard method which was described previously (Xu et al., 2016b). To obtain BMMs, bone marrow cells were collected from the femur and tibiae of four-week-old male C57/BL6 mice. Briefly, the bone marrow cells in the femur and tibiae were flushed out and washed. After removing the red blood cells, cells were incubated in full medium containing ɑ-MEM, 10% FBS, 1% penicillin/streptomycin in an incubator at 37°C with 5% CO₂. After 16 h, the supernatant cells were collected and maintained in full medium with 30 ng/mL M-CSF for another 2 days. Then the cells adhering to the bottom of the dish were classified as BMMs. BMMs were then harvested and cultured in full medium plus 30 ng/mL M-CSF and 50 ng/ml RANKL for additional 5-7 days to obtain mature OCs. Since cells obtained from different animals may respond differently, we mixed the cells from two different mice each time for BMM cultures to minimize such differences in our replication. Besides, the in vitro experiments were carried out independently at least three times. ASA were dissolved in DMSO, and 0.1% DMSO (v/v) as a final concentration was used as a vehicle control for all experiments in vitro.

The MC3T3-E1 preosteoblast cell line was purchased from ATCC (Manassas, VA, United States) and maintained in ɑ-MEM, 10% FBS, 1% penicillin/streptomycin in an incubator at 37°C with 5% CO₂. The cells were cultured in full ɑ-MEM for proliferation and in full ɑ-MEM plus 10 mM β-glycerophosphate (β-GP) and 50 mg/ml ascorbic acid (AA) for osteoblastogenesis.

We evaluated the effect of ASA on BMMs cell proliferation using CCK-8 assay. Cells were seeded in 96-well plates at a density of 3 × 10³ cells/well in full medium with M-CSF (30 ng/ml) and incubated overnight to adhere. Then, the cells were cultured in medium with indicated concentrations of ASA in triplicate for different days. After that, the cells were incubated with CCK8 solution (10 μL/well) at 37°C for an additional 2 h. The optical density (OD) at a wavelength of 450 nm was quantitatively measured with an Infinite F200 PRO absorbance microplate reader (Tecan). Cell proliferation (%) was analyzed relative to the control.

BMMs were seeded in 96-well plates at a density of 3 × 10³ cells/well and exposed to 30 ng/mL M-CSF and 50 ng/ml RANKL for 5-7 days to obtain mature OCs which were classified as multinucleated cells in tartrate-resistant acid phosphatase (TRAP) staining assay. For TRAP staining assay, multinucleated OCs from BMMs were fixed with 4% paraformaldehyde. Then we used a leukocyte acid phosphatase kit (Sigma) to visualize multinucleated OCs which was performed according to the manufacturer’s instructions. Cells were incubated at room temperature for 30 min. Images of multinucleated cells were obtained and quantified under an Olympus microscope. TRAP-positive cells in different divisions such as mononuclear, multinuclear (2–5 nuclei; 5-10 nuclei) and giant cells (>10 nuclei) were scored and then analyzed respectively. The average number of TRAP-positive in each division in the control group was defined as 100%.

BMMs were seeded in 6-well plates at a density of 2 × 10⁵ cells/well and exposed to 30 ng/mL M-CSF and 50 ng/ml RANKL with or without ASA for 5-7 days to obtain mature OCs. For actin cytoskeleton staining, cells were fixed with 4% paraformaldehyde. Then we used iFluor™ 555-Phalloidin working solution to visualize multinucleated OCs which was performed according to the manufacturer’s instructions. Cells were incubated at room temperature for 30 min. Then cells were washed in PBS for three times and incubated with DAPI (Sigma-Aldrich) for 5 min. Images of multinucleated cells were acquired under a ZEISS fluorescence microscope.

The procedure of bone resorption assay was described previously (Suda et al., 1999; Binder et al., 2009; Liu et al., 2019). Flat bone slices at 100 μM thick from bovine femur were used in bone resorption assay. BMMs were seeded on bone slices in 96-well plates overnight. The next day BMMs were induced by M-CSF and RANKL for 5-6 days, followed by M-CSF, RANKL and ASA treatment for another 2 days. Bone slices were then fixed with 2.5% glutaraldehyde for 7 min, after which cells were removed with vigorous sonication in 0.25 M ammonium hydroxide for 3 times. The bone resorption lacunae or pits were then photographed using a confocal laser microscope with ×200 magnification since it was good enough to characterize the surface roughness of the bone slices (Boyde and Jones 1991). Pit area was analyzed by Image-Pro Plus. Three view fields of each bone slice were randomly selected for further analysis.

qRT-PCR was performed using an ABI 7500 Sequencing detection system. Total mRNA was isolated using TRIzol reagent (Life Technologies, Carlsbad, CA) and then subjected to reverse transcription. Complementary DNA (cDNA) was obtained and used as a template targeting TRAP, CTSK, ATPasev0d2 (ATPase, H+ transporting, lysosomal v0 subunit d2), MMP-9, RANKL, ALP, OCN and β-actin using specific primers. All primer sequences were summarized below (Table 1). β-actin was used as an internal control. Indicated mRNA levels were normalized to β-actin mRNA.

Whole-cell lysates of BMMs were prepared in SDS lysis buffer, followed by incubation at 95°C for 5 min. Cell lysates were thus subjected to SDS-PAGE. Proteins were separated and transferred onto NC membranes by electroblotting, which were incubated with blocking buffer for 1 h. Then the membranes were incubated with targeted primary antibodies at 4°C overnight, washed three times with Tris‐buffered saline plus Tween (TBST) for 10 min each time, and incubated with secondary antibodies at room temperature for 2 h. Finally, the bands were detected via analysis of immunoreactivity using an Immobilon Western kit (MILLIPORE) and photographed by ImageQuant LAS 500 (GE Healthcare). The expressions of indicated proteins were qualified by Image J software.

To study the effects of ASA on oestrogen deficiency induced bone loss in mice, female C57BL/6 mice at 8-weeks were ovariectomized (OVX) or subjected to a sham operation (SHAM) which was described previously (Liu et al., 2019). After recovery, SHAM-operated with vehicle treatment (group SHAM, control, n = 6), ovariectomized-operated with vehicle treatment (group OVX, control, n = 6), ovariectomized-operated with ASA treatment (group ASA, 30 mg/kg, n = 6) were administered intraperitoneally once a day. ASA was dissolved in the final concentration of 1.25% DMSO (v/v) and 2.5% Tween80 in saline (v/v) as a vehicle. The group SHAM and OVX mice also received a similar injection with 1.25% DMSO (v/v) and 2.5% Tween80 (v/v) in saline as control. The body weight was collected once a week since the body weight increased steadily after OVX operation as reported (Ghayor et al., 2015; Nishio et al., 2019). Animals were sacrificed after 4 weeks’ treatment. The femurs and tibias were dissected and fixed with 4% paraformaldehyde for 24 h for subsequent inspections, such as pQCT scanning, μCT scanning, TRAP staining and H&E staining. The livers, lungs, hearts, kidneys and spleens were also dissected and fixed with 4% paraformaldehyde for H&E staining.

Three-dimensional reconstructions of the tibias were obtained from images acquired using a high-resolution micro-computed tomography scanner (Skyscan 1172; Skyscan; Aartselaar, Belgium). The tibias image acquisition was carried out with same parameters as follows: pixel size, 10 μm; X-ray voltage, 50 kV; electric current, 500 μA; rotation step, 0.7°, in accordance with the recommendations of the American Society for Bone and Mineral Research (ASBMR) (Dempster et al., 2013). Then the scans were integrated into 3D voxel image and analyzed to quantitatively assess bone parameters using SkyScan CT Analyser software (version 1.15.4.0) according to standardized protocols.

Briefly, the proximal end of the tibias corresponding to 0–5 cm region below the growth plate was scanned. The images of the trabecular region of interest (ROI) extended from the 100th layer proximally to the end of the distal growth plate over 200 layers toward the diaphysis were contoured for trabecular bone analysis. Bone parameters such as bone volume/tissue volume (BV/TV), bone surface/tissue volume (BS/TV), bone surface/bone volume (BS/BV), trabecular bone number (Tb.N), trabecular thickness (Tb.Th) and the 3D images were obtained.

The femurs were decalcified in 10% EDTA for 3  weeks, paraffin-embedded, sectioned and stained by TRAP or HE solution. The histomorphometric examination was imaged using a Zeiss microscope with ×200 magnification. The percentage of the bone surface and osteoclast numbers were calculated according to standardized protocols.

C57BL/6 mice were purchased from SLAC laboratory and were maintained under standard animal housing conditions (55–60% humidity, 22-24°C with a 12/12-h light/dark cycle, free access to food and water). All animal experiments were conducted in accordance to the guidelines of the human use and care of laboratory animals and were approved by the Shanghai Jiao Tong University School of Medicine Animal Study Committee.

The results are expressed as the mean ± standard deviations. One-way ANOVA and two-tailed non-paired Student’s t test were used to compare differences, and statistical significance was displayed as *p < 0.05 **p < 0.01 or ***p < 0.001.

ASA structure was shown in Figure 1A. Murine BMMs were used to test the anti-osteoclastogenic activity of ASA. Two major events dominate the complex multi-step process of OCs formation: the proliferation of BMMs under the induction of M-CSF, and the subsequent differentiation of pre-OCs into mature OCs induced by RANKL. Prior to anti-osteoclastogenic activity, we investigate the effect of ASA in murine BMMs proliferation to find a safe dose in vitro. ASA in different doses were incubated with murine BMMs for different days, cell proliferation was then detected by CCK8 assay. As shown in Figure 1B, even at 80 μM, ASA did not exert any inhibitory effect in murine BMMs proliferation.

TRAP is an important and reliable indicator to identify OCs, therefor TRAP staining assay was applied to identify these osteoclast-like giant and TRAP-positive cells. BMMs were incubated with 50 ng/ml RANKL and 30 ng/mL M-CSF with different doses of ASA for 5-7 days. The TRAP staining results in Figure 1C indicated that ASA inhibited the formation of TRAP-positive multinucleated cells as well as mononuclear cells in a dose-dependent manner, which was confirmed by quantitative analysis of OCs numbers (Figure 1D) as well.

In Figure 1B, we have shown that even at 80 μM, ASA did not exert any inhibitory effect on M-CSF induced proliferation. To further determine which stage did ASA inhibit osteoclastogenesis, ASA (80 µM) was added to OCs differentiation medium beginning from different days shown in Figure 1E. The results showed that ASA could inhibit osteoclast differentiation at a time-dependent manner. ASA inhibited OCs osteoclastogenesis by 98.8, 94.4, 62.3% at early-stage, mid-stage and late-stage, respectively (Figure 1F and Figure 1G). Therefore, ASA was more efficient to prevent OCs osteoclastogenesis from early-stage (day1) and mid-stage (day3), rather than late-stage (day5).

F-Actin ring is a characteristic actin structure that is essential for mature OCs bone resorption. Hence, we further explored the effect of ASA on the formation of F-actin ring. The results showed pre-OCs could differentiate into mature OCs upon RANKL stimulation and cytoskeleton was then reorganized to create an F-actin-rich ring upon adhesion on the dish surface (Figure 1H). As shown in Figure 1H, the numbers of F-Actin ring were significantly decreased by 80 μM ASA, suggesting ASA has the potential to suppress actin-rings formation in OCs.

The formation of resorption pits on mineralized surface is a standard assay for assessing OCs function. Therefore, we detected the effects of ASA on the formation of bone resorption pits. Pre-OCs were placed on bone slices and were differentiated into mature OCs for 5-7 days, then ASA was added for two more days. The results in Figure 2A and the quantitative analysis in Figure 2B indicated that ASA (80 µM) reduced OCs bone resorption by 70.2%, compared with control group.

To further confirm the inhibitory effects of ASA on OCs differentiation and bone resorption, BMMs were induced into osteoclastogenesis and expressions of bone resorption-related gene in different stages were measured by RT-PCR and WB. As shown in Figures 2C–F, mRNA expressions of the bone resorption-related genes such as TRAP, ATPv0d2, MMP9 and CTSK were inhibited by ASA in different stages. We also detected the protein expression of CTSK, TRAP and MMP9, since they were largely responsible for cleaving and removing the bone matrix. As indicated in Figure 2G, protein level of CTSK, TRAP and MMP9 increased in a time-dependent manner upon RANKL stimulation, while ASA decreased its protein level, especially at day 5.

Phosphorylation of ERK1/2, JNK1/2, p38 and AKT are known to play a role in the early stage of RANKL induced OCs differentiation. As shown in Figures 3A,B, BMMs were treated by 50 ng/ml RANKL for indicated time, with or without pre-treatment 80 μM ASA for 2 h, then WB was applied to investigate phosphorylation of these signal molecules. The blots and quantification in Figures 3A,B indicated all signaling molecules were activated within 5-10 min, and peaked at 5 min, while the phosphorylation of AKT and p38 were attenuated upon ASA treatment. There was no difference in ERK1/2 and JNK1/2 phosphorylation. NF-κB signaling is also important for OCs differentiation. The phosphorylation IκB-α and p65 were activated by RANKL and then p65/p50 heterodimers would translocate to the nucleus to induce expressions of c-Fos and NFATc1. Thus, the phosphorylation of IκB-α and p65 in NF-κB signaling pathway were also detected. As show in Figures 3A,B, the phosphorylation of IκB-α was promoted by RANKL within 5 min subsequently by p65 phosphorylation, while p-IκB-α and p-p65 were both attenuated by ASA treatment.

NFATc1/c-fos, the most important transcription factors in osteoclastogenesis, controlled the expression of OCs differentiation and resorption related genes expressions such as TRAP, CTSK, MMP9, ATPv0d2 and so on. Thus, the inhibitory effect of ASA on the expression of c-Fos and NFATc1 was also evaluated by WB analysis. As shown in Figure 3C, RANKL induced the expression of c-Fos and NFATc1 both at the early-stage (day1) and mid-stage (day 3) of osteoclastogenesis. In particular, ASA strongly blocked RANKL-induced protein expression of c-fos at day1 and day3, with mild inhibitory effects on NFATc1 at day3.

All above results concluded that the anti-osteoclastogenesis effects of ASA in OCs differentiation and resorption were related with AKT, p38 phosphorylation as well as NF-κB signaling pathway, followed by NFATc1/c-fos signal pathway, then TRAP, CTSK, ATPv0d2 and MMP9 expressions.

Oestrogen-deficiency is associated with bone loss and primary OP, resulting in an increased risk of fracture. We further investigated the effect of ASA on oestrogen-deficiency induced OP in vivo. OVX-operated mice were intraperitoneally administrated with 30 mg/kg ASA (1.25% DMSO+2.5% Tween80 in saline as vehicle control) for 4 weeks and then the tibias were isolated and scanned by microCT. As shown in Figure 4A, OVX-operated mice gained an average of 13% body weight during the study period, while ASA group did not show any significant change on body weight at week 4 compared to OVX-operated. As shown in the bone quality parameters analysis of ROI (Figures 4B–F), ASA increased BV/TV, BS/BV and Tb. n compared with OVX-group which indicated that ASA might improve the quality of bone microarchitecture. This protective effect was also indicated in 3D reconstructed images (Figure 4G), compared with SHAM operated mice, OVX-mice exhibited extensive OP while ASA improved it.

In addition, the femurs were cut into slices and were subjected to TRAP and H&E staining (Figure 5A). The TRAP staining and statistics suggested that BV/TV were decreased by OVX and increased upon ASA treatment which were in accordance with microCT analysis (Figure 5B). Besides, OCs surface/BS in vivo were decreased upon ASA treatment, while OBs surface/BS were increased upon ASA treatment with no significance (Figures 5C,D).