All animal experiments in this study were approved by the Animal Ethics Committee of Fudan University Pudong Medical Center (No.20211201-1) and were conducted in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals.

Berbamine (purity>99%) was purchased from MedChemExpress (New Jersey, USA). We dissolved berbamine in dimethyl sulfoxide (DMSO) to prepare a stock solution (4 mM) and stored the stock solution at -80°C. The stock solution was further diluted with complete culture medium for in vitro experiments. For animal experiments, the stock solution was further diluted with DMSO and plant oil. A Cell Counting Kit-8 (CCK-8) was purchased from Dojindo (Tokyo, Japan). A TRAP staining kit was purchased from Sigma–Aldrich (MO, USA). Recombinant m-RANKL and recombinant M-CSF were purchased from R&D Systems (Minneapolis, MN, USA). Alpha-modified minimal essential medium (α-MEM), penicillin–streptomycin (P/S) and fetal bovine serum (FBS) were purchased from Thermo Fisher Scientific (Scoresby, Vic., Australia). Rhodamine-conjugated phalloidin and DAPI were obtained from Solarbio Co., Ltd. (Beijing, China). Universal RNA extraction kits and Evo M-MLV RT kits were purchased from Accurate Biotechnology Co., Ltd. (Hunan, China). Primary antibodies against CTSK, TRAP, MMP-9, NFATc1 (nuclear factor of activated T cells 1), CD44, DC-STAMP (dendritic cell specific transmembrane protein) and GAPDH were purchased from Proteintech (Wuhan, Hubei, China).

BMMs were isolated and differentiated as described in a previous article (13). We isolated BMMs from the femoral and tibial bone marrow of 6-week-old C57BL/6 mice. We euthanized the mice according to procedures approved by the Animal Ethics Committee of Fudan University Pudong Medical Centre. Sterile complete culture medium (containing 10% FBS and 100 U/ml P/S in α-MEM) was used to flush the femoral medullary cavity, and the BMMs were collected in a sterile dish. Then, the BMMs were cultured in complete culture medium with 20 ng/mL M-CSF for approximately 72 hours. The adherent cells were collected and detected by flow cytometry (BD FACSCaliburTM, USA), and the proportion of BMMs was more than 90% ( Figure S1 ). For osteoclast differentiation, BMMs were cultured in complete culture medium containing 50 ng/ml RANKL and 20 ng/ml M-CSF.

Cell viability was detected by CCK-8 assay according to the kit manufacturer’s protocol. BMMs were cultured in 96-well plates (8×10³/well) and incubated with different concentrations of berbamine (12.5, 6.25, 3.125, 1.5625 and 0 μM, 3 duplicate wells for each concentration) for 24, 48, 72 and 120 hours. At each time point, we washed the 96-well plates three times with 1× PBS solution, added 100 μL of CCK-8 working solution to each well, and incubated the plates at 37°C for 1 hour. Then, the absorbance was detected at 450 nm with a microplate reader (Thermo Fisher, Waltham, MA, USA). The half-maximal inhibitory concentration (IC50) of berbamine was detected via CCK-8 assay according to the kit manufacturer’s protocol. BMMs were cultured in 96-well plates (8×10³/well) and incubated with different concentrations of berbamine (200, 100, 50, 25, 12.5, 6.25, 3.125, 1.5625, 0.78125, 0.390625 and 0 μM, with 3 duplicate wells for each concentration) for 48 hours. We washed the 96-well plates 3 times with 1× PBS solution, added 100 μL of CCK-8 working solution to each well and incubated the plate at 37°C for 1 hour. Then, the absorbance was detected at 450 nm with a microplate reader (Thermo Fisher, Waltham, MA, USA). GraphPad Prism (version 7, GraphPad Software, San Diego, CA, USA) was used to calculate the IC50 of berbamine on BMMs.

To evaluate the effects of different concentrations of berbamine on the osteoclastogenesis of BMMs, BMMs (8×10³/well) were cultured in 96-well plates and divided into a control group (cultured in complete culture medium containing 20 ng/ml M-CSF), a RANKL group (cultured in complete culture medium containing 20 ng/ml M-CSF and 50 ng/ml RANKL) and 2 drug intervention groups (cultured in complete culture medium containing 3.125 or 6.25 μM berbamine, 20 ng/ml M-CSF and 50 ng/ml RANKL).

To evaluate the effects of berbamine (6.25 μM) on BMMs at the different phases of osteoclastogenesis, BMMs (8×10³/well) were cultured in 96-well plates and divided into a control group (cultured in complete culture medium containing 20 ng/ml M-CSF), a RANKL group (cultured in complete culture medium containing 20 ng/ml M-CSF and 50 ng/ml RANKL), an early-phase group (cultured in complete culture medium containing 20 ng/ml M-CSF and 50 ng/ml RANKL; berbamine was added to the medium only on days 1-2 of culture), a middle-phase group (cultured in complete culture medium containing 20 ng/ml M-CSF and 50 ng/ml RANKL; berbamine was added to the medium only on days 3-4 of culture) and a late-phase group (cultured in complete culture medium containing 50 ng/ml RANKL and 20 ng/ml M-CSF; and berbamine was added to the medium only on days 5-6 of culture). After 6 days of culture in an incubator with 5% CO2 at 37°C, the BMMs were stained with a TRAP staining kit according to the manufacturer’s protocol. The staining was photographed under a microscope (Nikon Corporation, Tokyo, Japan), and BMMs were regarded as osteoclasts when they contained more than 3 nuclei (11). All experiments were repeated at least 3 times.

To evaluate the effects of different concentrations of berbamine on the osteoclastogenesis of BMMs, BMMs (8×10³/well) were cultured in 96-well plates and treated with 3.125 or 6.25 μM berbamine.

To evaluate the effects of berbamine (6.25 μM) on BMMs at the different phases of osteoclastogenesis, BMMs (8×10³/well) were cultured in 96-well plates and treated with 6.25 μM berbamine at different phases of osteoclastogenesis. The cells were divided into 5 groups (a control group, a RANKL group, an early-phase group, a middle-phase group and a late-phase group; the details are described above). After 6 days of culture in an incubator with 5% CO2 at 37°C, the cells were fixed with paraformaldehyde (4%) for 20 minutes at 37°C and washed with 1× PBS solution 3 times, and then 0.1% Triton-X 100 was used to increase cell permeability. The cells were incubated in the dark with rhodamine-conjugated phalloidin for 20 minutes and DAPI for 10 minutes. Then, a fluorescence microscope (Nikon Corporation, Tokyo, Japan) was used to detect the podosome belts and nuclei of the cells. BMMs were regarded as osteoclasts when they had more than 3 nuclei (13).

A pit resorption assay was used to evaluate the inhibitory effect of berbamine on the bone resorption function of osteoclasts. BMMs (8×10³/well) were plated on the surfaces of bone slices placed in 96-well plates. The cells were divided into 4 groups: a control group (cultured in complete culture medium containing 20 ng/ml M-CSF), a RANKL group (cultured in complete culture medium containing 20 ng/ml M-CSF and 50 ng/ml RANKL) and 2 drug intervention groups (cultured in complete medium containing 3.125 or 6.25 μM berbamine, 20 ng/ml M-CSF and 50 ng/ml RANKL). After 6 days of culture in an incubator with 5% CO2 at 37°C, the bone slices were removed from the 96-well plates and sonicated to remove the cell debris. A scanning electron microscope (GeminiSEM 500, Gemini, Germany) was used to detect the surface topography of the bone slices in the different groups. ImageJ software (NIH, Bethesda, MD, USA) was used to detect the proportion of the pit resorption area on the surfaces of the bone slices in each group.

The expression levels of RANKL-induced osteoclast-related genes, including CTSK, NFATc1, CD44, MMP-9, DC-STAMP and TRAP, were detected via quantitative real-time polymerase chain reaction (qRT–PCR). To evaluate the effects of different concentrations of berbamine on the expression levels of RANKL-induced osteoclast-related genes, BMMs (4×10⁴/well) were cultured in 6-well plates and divided into 4 groups (a control group, a RANKL group and 2 drug intervention groups). In the RANKL group, the BMMs were treated with M-CSF (20 ng/ml) and RANKL (50 ng/ml). In the first drug intervention group, the BMMs were treated with M-CSF (20 ng/ml), RANKL (50 ng/ml) and 3.125 μM berbamine. In the second drug intervention group, the BMMs were treated with M-CSF (20 ng/ml), RANKL (50 ng/ml) and 6.25 μM berbamine.

To evaluate the effects of berbamine (6.25 μM) on BMMs at the different phases of osteoclastogenesis, BMMs (4×10⁴/well) were cultured in 6-well plates and treated with 6.25 μM berbamine at different phases of osteoclastogenesis. The cells were divided into 5 groups (a control group, a RANKL group, an early-phase group, a middle-phase group and a late-phase group; the details are described above). After 6 days of culture in an incubator with 5% CO2 at 37°C, the total RNA of the cells was extracted with a Universal RNA Extraction Kit according to the manufacturer’s protocol. Then, the total RNA was reverse-transcribed into cDNA with Evo M-MLV RT kits. The qRT–PCR procedures have been described in detail in our previously published articles. The data was analyzed by the 2^-△△CT method. The sequences of the primers are listed in detail in Table 1 .

Western blotting was performed to evaluate the effects of different concentrations of berbamine on the expression levels of RANKL-induced osteoclast-related proteins, including CTSK, NFATc1, CD44, DC-STAMP, MMP-9 and TRAP, in BMMs. BMMs (4×10⁴/well) were cultured in 6-well plates and divided into 3 groups (a RANKL group and 2 drug intervention groups). In the RANKL group, BMMs were treated with M-CSF (20 ng/ml) and RANKL (50 ng/ml). In the first drug intervention group, BMMs were treated with M-CSF (20 ng/ml), RANKL (50 ng/ml) and 3.125 μM berbamine. In the second drug intervention group, BMMs were treated with M-CSF (20 ng/ml), RANKL (50 ng/ml) and 6.25 μM berbamine.

To evaluate the effects of 6.25 μM berbamine on BMMs at the different phases of osteoclastogenesis via Western blotting, BMMs (4×10⁴/well) were cultured in 6-well plates and divided into 5 groups (a control group, a RANKL group, an early-phase group, a middle-phase group and a late-phase group; the details are described above).

To detect whether berbamine can suppress RANKL-induced activation of several common signaling pathways, such as the MAPK, NF-κB and PI3K-AKT-NFATc1 signaling pathways, in the early stage of osteoclastogenesis, we treated BMMs with RANKL (50 ng/ml) and berbamine (6.25 μM) and examined AKT, p65, ERK, IκBα and JNK phosphorylation at four time points (0, 15, 30, and 60 minutes).

The treated BMMs were lysed on ice in radioimmunoprecipitation assay lysis buffer (Teye, Bioteke Corp., Beijing, China). Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) was used to separate equal amounts of osteoclast-related proteins, and the separated proteins were transferred to polyvinylidene fluoride (PVDF) membranes (GE Healthcare, Silverwater, Australia). The membranes were blocked with 5% fat-free milk solution for 60 minutes at room temperature and then incubated at 4°C overnight with primary antibodies against CTSK, NFATc1, CD44, MMP-9, DC-STAMP, TRAP and GAPDH. Subsequently, the membranes were incubated with corresponding secondary antibodies for 60 minutes at room temperature. The protein bands were visualized with enhanced chemiluminescence (ECL) liquid (Teye, Bioteke Cor. Beijing, China) and an image development device (Amersham Pharmacia Biotech, Sydney, NSW, Australia).

We simulated the molecular docking of berbamine with DC-STAMP via AutoDock 4 software (version 4.2.6, Scripps Research Institute, La Jolla, CA, USA). The PDB-format file of the three-dimensional (3D) structure of DC-STAMP was downloaded from UniProt (https://www.uniprot.org/; UniProt ID of DC-STAMP: Q7TNJ0). The 2D structure of berbamine was sketched in ChemDraw (version 20.0, CambridgeSoft, USA), converted to a 3D structure via the Chem3D Ultra program (version 10.0, CambridgeSoft, USA) and saved as a PDB-format file. The charges were calculated, and hydrogen atoms were added to the 3D structures of both DC-STAMP and berbamine with AutoDockTools (version 1.5.7, Scripps Research Institute, La Jolla, CA, USA) to minimize the system’s energy. After completing the above steps, the binding pocket of DC-STAMP and berbamine was predicted via sitemap, and then molecular docking of DC-STAMP and berbamine was performed using AutoDock 4 software (version 4.2.6, Scripps Research Institute, La Jolla, CA, USA).

The animal experiments were conducted in a specific pathogen-free (SPF) animal laboratory. Female six-week-old C57BL/6 mice were purchased from Shanghai Laboratory Animal Research Centre (Shanghai, China). All mice were housed in an SPF animal laboratory with clean water and fodder. All mice were randomly divided into 3 groups [a sham group, an ovariectomy (OVX) group and a berbamine group, n=5]. The mice were intraperitoneally injected with 2% (w/v) pentobarbital (40 mg/kg) for anesthesia. The mice in both the OVX group and the berbamine group underwent bilateral OVX via a dorsal approach, while the mice in the sham group only underwent removal of a small volume of fatty tissue from around their ovaries via the same approach. After the incisions were properly sutured and the animals recovered from anesthesia, the mice were returned to their own cages and allowed to recover for 3 days before intraperitoneal injection. The mice in the berbamine group received berbamine (20 mg/kg) dissolved in DMSO (CDMSO<0.2%) and plant oil every two days for eight weeks. The mice in both the sham group and the OVX group received the same solvent only through intraperitoneal administration every two days for eight weeks. Calcein (10 μg/g) was injected into the mice 10 days and 3 days before the animals were sacrificed via intraperitoneal injection. After eight weeks, all mice were euthanized, and the bilateral femurs and serum of the mice were collected for the follow-up experiments.

The right femurs of the mice were scanned by microcomputed tomography (Skyscan1172, Bruker, Kontich, Belgium). The tube voltage was set to 50 kV with an electric current of 500 μA, and the voxel size of the reconstructed image was 9×9×9 μm³. We analyzed the images for the following histological parameters: the BMD, trabecular number (Tb.N), trabecular pattern factor (Tb.Pf), bone surface (BS)/bone volume (BV) ratio, BS density (BS/total volume [TV]), and percent BV (BV/TV). We reconstructed two-dimensional (2D) and 3D images of the distal femurs of the mice with Mimics 18.0 software (Materialise, Leuven, Belgium).

We fixed the right femurs of the mice in 4% paraformaldehyde at room temperature. A Leica SM2500 microtome (Germany) was used to prepare unstained 3-μm-thick sections, and a fluorescence microscope was used to detect the calcein labels in the distal femurs at 515 nm.

We fixed the left femurs of the mice and decalcified them in 10% tetrasodium-EDTA for 14 days. Then, the left femurs were paraffin-embedded, and 4-μm-thick sections were prepared with a microtome (Jung, Heidelberg, Germany). We used hematoxylin and eosin to stain trabecular bone. TRAP immunohistochemical staining was performed to detect TRAP-positive osteoclasts in the distal femurs of the mice in each group. DC-STAMP immunohistochemical staining was performed to detect DC-STAMP-positive osteoclasts in the distal femurs of the mice in each group. The mineral apposition rate (MAR) was used to measure osteoblastic activity by measuring the average distance between two labels. All data were calculated with ImageJ software (NIH, Bethesda, MD, USA).

All experiments were repeated 3 times, and the results are presented as the χ ¯ ± SD for data that were normally distributed. In this study, GraphPad Prism (version 7, GraphPad Software, San Diego, USA) was used for statistical analysis. Comparisons between two groups were performed by Student’s t test for normally distributed variables or Mann–Whitney U test for nonnormally distributed variables. One-way ANOVA with Tukey’s post-hoc test was used to determine whether the differences among several groups were statistically significant. Values of P<0.05 were considered to indicate statistically significant differences.

BMMs were treated with different concentrations of berbamine (0, 1.5625, 3.125, 6.25 and 12.5 μM), and the absorbance at 450 nm was tested after 24, 48, 72 and 120 hours of treatment to determine the cytotoxicity of berbamine. The data revealed that berbamine had no significant cytotoxicity toward BMMs at concentrations less than or equal to 6.25 μM ( Figure 1B ). BMMs were also cultured in 96-well plates (8×10³/well) and incubated with different concentrations of berbamine (200, 100, 50, 25, 12, 5, 6.25, 3.125, 1.5625, 0.78125, 0.390625 and 0 μM) for 48 hours. The IC50 of berbamine for BMMs was calculated to be 10.48 μM ( Figure 1C ).

To determine the effect of berbamine on the osteoclastogenesis of BMMs, we plated BMMs into 96-well plates and divided them into four groups. We observed that the cells in the two drug intervention groups were dyed by the TRAP staining kit. However, the results revealed that berbamine (both 3.125 μM and 6.25 μM) inhibited RANKL-induced osteoclastogenesis of BMMs in vitro compared with that in the RANKL group. We found that the numbers of mature, polynuclear osteoclasts were reduced significantly in a dose-dependent manner after treatment with berbamine ( Figures 2A, B ).

The progression of osteoclastogenesis consists of multiple phases. Thus, it was important for us to determine the phase in which berbamine could inhibit RANKL-induced osteoclastogenesis of BMMs. BMMs were plated into 96-well plates and treated with berbamine (6.25 μM) at different phases of osteoclastogenesis. The results revealed that the inhibitory effect of berbamine on osteoclastogenesis of BMMs mainly occurred in the middle and late phases. However, the inhibitory effect was relatively weak in the early phase ( Figures 2C, D ).

We plated BMMs into 96-well plates and treated them with berbamine to determine the inhibitory effect of berbamine on RANKL-induced podosome belt formation during the progression of osteoclastogenesis. Berbamine (both 3.125 μM and 6.25 μM) inhibited RANKL-induced podosome belt formation of BMMs in vitro, and the numbers of osteoclasts were reduced in a dose-dependent manner after treatment with berbamine ( Figures 2E, F ).

Similar to previous results, the results revealed that berbamine inhibited RANKL-induced podosome belt formation in BMMs mainly in the middle and late phases of osteoclastogenesis. However, the inhibitory effect of berbamine was relatively weak in the early phase of osteoclastogenesis ( Figures 2G, H ).

We placed bone slices into 96-well plates, plated BMMs on the surfaces of the bone slices and divided them into 4 groups (a control group, a RANKL group and 2 drug intervention groups) to determine the inhibitory effect of berbamine on RANKL-induced formation of osteoclasts with bone resorption function in vitro. After 6 days, the bone slices were fixed with paraformaldehyde (4%) and sonicated to remove the cell debris from the surfaces. Then, the surface morphology of the bone slices was investigated by scanning electron microscopy ( Figure 2I ), and the data showed that the percent area of pit resorption on the surfaces of bone slices was significantly higher in the RANKL group than in the berbamine treatment groups ( Figure 2J ), which revealed that berbamine (both 3.125 μM and 6.25 μM) could inhibit RANKL-induced formation of osteoclasts with bone resorption function in vitro.

We used qRT–PCR analysis to determine the inhibitory effects of berbamine on the expression of RANKL-induced osteoclast-related genes, such as CTSK, DC-STAMP, MMP-9, NFATc1, CD44 and TRAP. BMMs were plated in 6-well plates and treated with berbamine (0, 3.125, and 6.25 μM). Berbamine (both 3.125 μM and 6.25 μM) significantly downregulated the expression levels of the osteoclast-related genes CTSK, DC-STAMP, MMP-9, NFATc1 and CD44; however, it did not inhibit the gene expression of TRAP ( Figure 3A ). BMMs were also plated into 6-well plates and treated with berbamine (6.25 μM) at different phases of osteoclastogenesis. The results revealed that berbamine significantly downregulated the expression levels of the osteoclastogenic genes CTSK, DC-STAMP, MMP-9, NFATc1 and CD44 mainly in the middle and late phases of osteoclastogenesis. However, treatment of BMMs with berbamine did not significantly downregulate the expression level of the TRAP gene ( Figure 3B ).

Similar to the previous qRT–PCR results, the protein expression levels of CTSK, NFATc1, CD44, MMP-9 and DC-STAMP were downregulated by berbamine (both 3.125 μM and 6.25 μM) in a dose-dependent manner. However, TRAP protein expression was not significantly downregulated by berbamine ( Figures 3C, D ). Berbamine significantly downregulated CTSK, NFATc1, CD44, MMP-9 and DC-STAMP protein expression mainly in the middle and late phases of osteoclastogenesis. However, it did not significantly downregulate the protein expression level of TRAP ( Figures 3E, F ).

RANKL induced the activation of multiple common signaling pathways, such as the MAPK signaling pathway, the NF-κB signaling pathway and the PI3K-AKT-NFATc1 signaling pathway, which play important roles in the process of osteoclastogenesis. We used Western blot analysis to examine the levels of phosphorylated AKT, p65, ERK, IκBα and JNK in order to detect whether berbamine could inhibit RANKL-induced activation of the corresponding common signaling pathways in the early phase of osteoclastogenesis and found that it could not ( Figures 4A, B ).

We used AutoDock 4 to simulate the molecular docking of berbamine with DC-STAMP. The results predicted that berbamine can bind with DC-STAMP by forming a stable hydrophobic interaction with a binding pocket consisting of MET-66, CYS-69, LEU-61, LYS-445, GLN-431, ARG-428 and THR-103. In addition, berbamine can form hydrogen bonds with SER-63 of DC-STAMP. The free energy of the binding of berbamine with DC-STAMP was calculated to be -5.808 kcal/mol; thus, berbamine can bind tightly to DC-STAMP ( Figure 4C ).

According to the above results, we speculated that berbamine can alleviate the extent of PMOP by inhibiting osteoclast differentiation and osteoclast bone resorption function in vivo. Thus, we established a PMOP animal model via OVX. Berbamine (20 mg/kg) was administered by intraperitoneal injection. We used microcomputed tomography to scan the distal femurs of the mice, and we reconstructed 2D and 3D images of the distal femurs. Related parameters were analyzed, including the BMD, BV/TV, BS/TV, Tb.N and Tb.Pf. The extent of bone loss was significantly greater in the OVX group than in the sham group and the berbamine group, and there were significant differences in the parameters between the mice in the OVX group and those in the berbamine group ( Figures 5A, B ).

The results of hematoxylin-eosin staining of the distal femurs of the mice showed that the bone trabeculae were significantly more abundant in the berbamine group than in the OVX group ( Figures 5C, D ). In addition, there were significantly fewer TRAP-positive multinuclear cells in the distal femurs of the mice in the berbamine group than in the OVX group ( Figures 5E, F ). The results of DC-STAMP immunohistochemical staining showed that there were significantly fewer DC-STAMP-positive multinuclear cells in the distal femurs of mice in the berbamine group than in the OVX group ( Figures 5G, H ). The results of calcein double labeling for measurement of femoral bone formation showed that the MAR was significantly higher in the OVX group than in both the sham group and the berbamine group ( Figures 5I, J ).