Madecassoside inhibits estrogen deficiency‐induced osteoporosis by suppressing RANKL‐induced osteoclastogenesis

Abstract Osteoporosis is the most common osteolytic disease characterized by excessive osteoclast formation and resultant bone loss, which afflicts millions of patients around the world. Madecassoside (MA), isolated from Centella asiatica, was reported to have anti‐inflammatory and antioxidant activities, but its role in osteoporosis treatment has not yet been confirmed. In our study, MA was found to have an inhibitory effect on the RANKL‐induced formation and function of OCs in a dose‐dependent manner without cytotoxicity. These effects were attributed to its ability to suppress the activity of two transcription factors (NFATc1 and c‐Fos) indispensable for osteoclast formation, followed by inhibition of the expression of bone resorption‐related genes and proteins (Acp5/TRAcP, CTSK, ATP6V0D2/V‐ATPase‐d2, and integrin β3). Furthermore, we examined the underlying mechanisms and found that MA represses osteoclastogenesis by blocking Ca2+ oscillations and the NF‐κB and MAPK pathways. In addition, the therapeutic effect of MA on preventing bone loss in vivo was further confirmed in an ovariectomized mouse model. Therefore, considering its ability to inhibit RANKL‐mediated osteoclastogenesis and the underlying mechanisms, MA might be a potential candidate for treating osteolytic bone diseases.


| INTRODUCTION
Bone undergoes continuous remodelling to maintain its homeostasis by regulating the balance between osteoblasts and osteoclasts (OCs). 1 The disequilibration of this homeostasis, in favour of sthenic bone resorption caused by excess number or activity of mature OCs, can result in a myriad of skeletal lytic diseases, such as osteopetrosis, 2 Paget's disease 3 and osteoporosis. 4 Amongst these diseases, the high morbidity of osteoporosis makes it a serious public health concern, as approximately 40% of women older than 50 will suffer from osteoporosis, which eventually leads to increased risk of pathological fractures. 5 Current drugs for clinical use in treating osteoporosis, such as bisphosphonates, calcitonin, and estrogen, help maintain bone protein and mineral content and reduce fractures. 6 However, serious side effects such as breast and endometrial cancer and hypercalcemia limit the use of these pharmacological drugs. 7,8 Natural compounds with fewer side effects are more suitable than synthetic drugs for treating chronic diseases that require long-term treatment. [9][10][11] Therefore, exploration of natural compounds that have therapeutic effects targeting the differentiation and function of OCs is urgently needed.
OCs are multinucleated cells that originate from bone marrow monocytes (BMMs). 12 The process of this differentiation is mainly regulated by two osteoblast-derived factors: receptor activator of nuclear factor κ-B ligand (RANKL) and macrophage colony-stimulating factor (M-CSF). 13 M-CSF functions both to maintain the proliferation of osteoclast precursor (OCP) cells and to stimulate RANK expression. 14 RANKL plays a major role in the differentiation of OCP cells. Once RANKL combines with RANK on the membrane of a preosteoclast, adaptor molecules, such as TNF receptor-associated factor, will be recruited to stimulate the nuclear factor κ-B (NF-κB), mitogen-activated protein kinase (MAPK) and calcium signaling pathways for OC differentiation and formation. [15][16][17] Then, the two main transcription factors in osteoclast differentiation, activator protein 1 (AP-1) and nuclear factor of activated T cells, cytoplasmic 1 (NFATc1), are triggered to promote preosteoclast differentiation and improve the expression of OC function-related genes and proteins, including tartrate-resistant acid phosphatase (TRAcP/acp5), cathepsin K (CTSK), V-ATPase d2 (ATP6V0D2) and integrin β3). [17][18][19] Therefore, if steps in RANKL-induced osteoclastogenesis can be suppressed, skeletal lytic diseases may have the potential to be cured.
Madecassoside (MA), a pentacyclic triterpenoid saponin isolated from Centella asiatica, 20 exhibits various bioactivities, including antioxidative and anti-inflammatory. 21 Treatment with MA can significantly reduce myocardial ischemia-reperfusion injury and lipid peroxidation. 22,23 Furthermore, MA was also found to have inhibitory effects on the NF-κB and ERK/p38 pathways, leading to a reduction in oxidative stress and suppression of inflammatory responses. 24 However, the effects of MA on osteoclast differentiation and function have not yet been reported. Since the NF-κB and ERK/p38 pathways are also instrumental in RANKL-induced osteoclastogenesis, we studied MA-mediated regulation of RANKL-stimulated osteoclastogenesis from BMMs to mature OCs and its underlying mechanism.
In this study, to examine the role of MA in osteoclast differentiation and function, multimodal measurements (including the number and function of OCs, the expression of OC-related genes and proteins, etc.) were performed. Particularly, we focused on the effects of MA on the activities of two main transcription factors (NFATc1 and c-Fos), which control the terminal differentiation of OCs, and their underlying regulatory mechanisms, such as Ca 2+ oscillations and the NF-κB and MAPK signaling pathways. Furthermore, an estrogen deficiency-induced osteoporosis mouse model was used to confirm the therapeutic effect of MA on preventing bone loss in vivo. Our data provide exciting results indicating that MA might be a potential treatment option for osteolytic bone diseases.

| Cell culture and cytotoxicity assays
BMMs were extracted from the long bone marrow of C57BL/6 mice at 10 weeks of age using procedures approved by the Animal Ethics Committee of the University of Western Australia (RA/3/100/1244). WANG ET AL.

| Drug screening and cytotoxicity assays
BMMs at passage 2 were used to screen the effectiveness of drugs at suppressing RANKL-induced osteoclastogenesis. Cells were seeded into 96-well plates at a concentration of 5 × 10 3 cells per well in complete α-MEM medium prepared with M-CSF (50 ng/mL) and were cultured overnight to adhere. Then, the BMMs were stimulated with GST-rRANKL (50 ng/mL). Additionally, 10 μmol L −1 of natural compounds was added to screen for effective drugs. The complete medium was changed every 2 days, accompanied by the addition of fresh GST-rRANKL and compounds until OCs formed on the sixth day. The cells were fixed with 2.5% glutaraldehyde for 15 min and stained with TRAcP staining solution. TRAcP-positive multinucleated cells that had more than three nuclei were counted as OCs and used to evaluate the effect on inhibiting formation. After that, cytotoxicity assays were performed. BMMs were seeded into 96-well plates at a

| Immunofluorescence staining
A total of 5 × 10 3 BMMs per well were cultured in a 96-well plate with M-CSF (50 ng/mL) for 24 h. As mentioned above, the cells were then treated with GST-rRANKL (50 ng/mL) for 6 days to form mature OCs with or without varying concentrations (5 μmol L −1 or 10 μmol L −1 ) of MA. After that, the cells were fixed for 10 min and blocked for 20 min using 4% paraformaldehyde and 5% bovine serum albumin, respectively. Next, the cells were probed with rhodamine-conjugated phalloidin (Thermo Fisher Scientific) for 45 min to stain for F-actin. After being washed with PBS and stained with DAPI, cells were visualized on a confocal fluorescence microscope (Nikon, A1 PLUS, Tokyo, Japan) at 100× magnification.

| Hydroxyapatite resorption assay
The hydroxyapatite resorption assay was used to measure the function of the induced OCs as described previously. 27 BMMs were first seeded into a 6-well collagen-coated plate (BD Biosciences, Sydney, Australia) with 1 × 10 5 cells in each well and then were stimulated with M-CSF (50 ng/mL) and GST-rRANKL (100 ng/mL) every 2 days to allow OCs to form. Next, the cells were dissociated from the collagen plate, and equal numbers of cells were transferred to the wells of a hydroxyapatite-coated plate (CLS3989, Corning, NY, USA).
Mature OCs were cultured in complete medium with GST-rRANKL (50 ng/mL) and M-CSF (50 ng/mL) in the presence or absence of MA (5 μmol L −1 or 10 μmol L −1 ). After 2-3 days, the wells were separated into two groups. One group was used to count the number of multinucleated cells in each well by TRAcP staining, as described above. The other group was used to measure the resorbed areas by bleaching for 10 min and removing the cells from the wells. The

| Quantitative real-time PCR analysis
A total of 5 × 10 3 BMMs per well were plated in a 6-well plate in the presence of GST-rRANKL (50 ng/mL) and M-CSF (50 ng/mL) with or without various densities of MA for 5 days. As described in our previous study, 32 TRIzol (Qiagen, Hilden, Germany) was used to extract total RNA from cells. With an oligo-dT primer, singlestranded cDNA was reverse transcribed from 2 μg total RNA. The resulting cDNA was then used for real-time PCR based on SYBR Green (Imgenex, Littleton, CO, USA) with the specific primers displayed in Table 1. The expression level was normalized to Hmbs expression. The fold change was determined using the Livak equation, and the ratios compared to the vehicle group were calculated.

| Western blot
To examine the expression of bone resorption-related proteins or the NFATc1 signaling pathway, BMMs were seeded (1 × 10 5 cells/well) into 6-well plates and incubated with or without MA (10 μmol L −1 ) in the presence of GST-rRANKL (50 ng/mL) for 5 days. Cells were then lysed, and total protein was harvested using radioimmune precipitation assay (RIPA) lysis buffer (containing 100 g/mL PMSF, 500 g/mL DNase I and phosphatase inhibitors) at the following time points: 0, 1,

| Mouse ovariectomy (OVX) procedures
Female C57BL/6 mice (10 weeks, n = 30) were provided by the Animal Center of the Chinese Academy of Science (Shanghai, China) and were randomly divided into three groups: a sham group (n = 10), OVX group (n = 10) and OVX+MA group (n = 10) (10 mg/kg). All the mice were kept in individual ventilated cages (IVC, five rats per cage) in a specific-pathogen-free (SPF) room. After a week of adjustable feeding, ovariectomy based on a previously described method 27 was performed for the OVX group and OVX+MA group, whereas a sham operation was performed for the sham group as a control.
Seven days later after the surgery, intraperitoneal injections of MA (10 mg/kg) for the OVX+MA group and PBS for the sham and OVX groups were given every 2 days for a total of 6 weeks. 27 Then, the mice were all sacrificed, and the femurs were removed for histological and micro-CT (μCT) analysis as previously described. 27,29 2.11 | Micro-CT scanning The femur samples were fixed with 4% paraformaldehyde for 24 h and analysed by a Skyscan 1176 micro-CT instrument (Skyscan, Bruker, Belgium). Images were acquired using a 50-kV X-ray tube voltage, a 500-μA current, an isotropic pixel size of 9 μmol L −1 (1600 × 2672-pixel image matrix) and a 0.5-mm-thick aluminum filter for beam hardening. The images were reconstructed using NRecon Reconstruction software (Bruker microCT, Kontich, Belgium). After that, a refined volume of the 0.5 mm below the growth plate and

| Statistical analysis
The results are presented as the means ± standard deviation (SD).
The significance of difference between two groups was determined by Student's t test. One-way ANOVA plus Tukey's test or Kruskal-Wallis analysis (non-parametric ANOVA) plus Dunn's multiple comparison (when the data failed the assumptions of the one-way ANOVA) were used to test differences between more than two groups. A two-way ANOVA was conducted to examine the effects of time and different treatment groups. P < 0.05 was considered statistically significant.

| MA inhibits RANKL-induced osteoclast formation in a dose-dependent manner
Numerous traditional Chinese medicines were added to the osteoclastogenesis assay as candidates to screen their ability to inhibit the RANKL-induced formation of OCs from BMMs at a concentration of 10 μmol L −1 (Table 2). Moreover, an MTS assay was performed to measure the cell cytotoxicity of these natural compounds, which might be confused with their effect on the suppression of OC

F I G U R E 4 Madecassoside (MA) blocks osteoclast-specific gene expression. (A) NFATc1, (B) c-Fos, (C) V-ATPase-d2 (ATP6V0D2), and (D)
Acp5 (TRAcP). Gene expression levels were standardized to Hprt expression. Data are presented as the means ± SD; *P < 0.05, **P < 0.01, and ***P < 0.001 relative to RANKL-induced controls formation. As can be seen from Table 2, MA was found to significantly inhibit osteoclast formation at a concentration of 10 μmol L −1 , although most of the other compounds had no effect on suppressing osteoclastogenesis. Furthermore, MA did not decrease the number of BMMs compared to that of the control group, an extremely encouraging result that confirmed that the inhibitory effect of MA on the generation of OCs from BMMs was not caused by cytotoxicity ( Figure 1A   The statistical significance of differences in protein expression between the MA-treated group and control group was analysed. The expression of all the proteins mentioned above was determined relative to β-actin expression. The data represent the means ± SD. Significant differences between the treatment and control groups are indicated as *P < 0.05, **P < 0.01, and ***P < 0.001 well exhibited no significant differences among all the groups, which is consistent with the above conclusion that MA inhibits osteoclastogenesis mainly in the early and middle stages. However, the resorption area decreased with increasing drug concentrations, especially at 10 μmol L −1 , which indicates that MA can also attenuate the resorption activity of OCs apart from osteoclast formation.

| MA suppresses the expression of genes related to RANKL-induced OC formation and function
Next, a qRT-PCR assay of various key genes (e.g, Acp5 (TRAcP), ATP6V0D2, NFATc1, and c-Fos) that play a significant role in RANKL-induced osteoclast formation and function 18 was used to provide deeper insight into the inhibitory effect that MA exerts on osteoclastogenesis. Figure 4 shows that the expression levels of these osteoclast-related genes were remarkably increased after stimulation with RANKL (50 ng/mL) for 5 days. However, both osteoclast formation-regulating genes (NFATc1 and c-Fos) and osteoclast function-related genes (Acp5 and CTSK) were significantly downregulated by MA at the concentration of 10 μmol L −1 in the presence of RANKL. These results confirm the inhibitory effects of MA on RANKL-induced OC formation and function, as described above.

| MA represses NFATc1 activity and downstream protein expression
To provide a complete, detailed picture of the role of MA in regulating osteoclast formation and function, a luciferase reporter assay and western blotting were chosen to detect the activity of NFATc1 and downstream protein expression. As presented in Figure 5A, the activity of NFATc1 was significantly downregulated by MA at con-

| MA suppresses NF-κB activation and calcium oscillation
The transcription regulator NFATc1 plays a critical role in osteoclast differentiation. 34 To gain a detailed understanding of the molecular mechanisms underlying the regulation and activation of NFATc1, the NF-κB and Ca 2+ signaling pathways were measured using a luciferase assay, western blotting, and calcium oscillation. 35,36 IκB is a major signaling molecule related to the activation of NF-κB. As shown in Figure 6A and B, the degradation of IκB was inhibited 20-30 min after treatment with MA compared to the RANKL-only group. Furthermore, the results shown in Figure 6C Figure 8D). In addition, no statistically significant difference in Tb.Th was found in all groups, indicating that the trabecular thickness was not susceptible to the estrogen deficiency ( Figure 8E). Taken together, these results indicated that MA exerts a protective effect on preventing estrogen deficiency-induced bone loss.
Histomorphometric assessment was further applied to increase the credibility of the results (Figure 9). The BV/TV value of the OVX+MA group was obviously higher than that of the OVX group. Significant differences between the treatment and control groups are indicated as *P < 0.05, **P < 0.01, and ***P < 0.001 compounds that have been reported to possess one or more bioactivities, such as antioxidative, anti-inflammatory, and antibacterial activities, for osteoporosis treatment at clinically acceptable concentrations. 23,41,42 We found that MA, which is characterized as a triterpenoid derivative extracted from Centella asiatica, showed a remarkable effect on inhibiting RANKL-induced osteoclastogenesis.
MA has been widely used in Chinese traditional medicine and it was reported to exert numerous bioactivities, such as attenuating the inflammatory response in collagen-induced arthritis in DBA/1 mice 23 and reducing ischemia-reperfusion injury in regional ischemia-induced heart infarction. 43 Furthermore, a clinical trial showed that MA had a long-term effect on photooxidative ageing of human skin by modulating inflammatory mediators, which provided practical proof of its safety for clinical use. 42 In this study, as shown in Figure 10 OCs are multinucleated cells derived from the fusion of precursor cells; this fusion process is impaired in the absence of V-ATPase-d2. 19 Bone absorption is an important function of OCs in the body. Once The bone resorption-related genes and proteins mentioned above are regulated by NFATc1. NFATc1 is an indispensable transcription factor that acts in the formation of OCs from BMMs. 17,48 Our results displayed a significant reduction of both the gene and protein expression and the activity of NFATc1, which confirmed that this crucial factor was remarkably suppressed by MA treatment.
NF-κB acts as an initiator of NFATc1 induction during RANKLinduced osteoclastogenesis. It was reported that after treatment with dehydroxymethylepoxyquinomicin (DHMEQ) (an NF-κB inhibitor), the activity of NFATc1 was significantly suppressed. 49 Another experiment demonstrated that p50 and p65 (two components of NF-κB) activated the NFATc1 promoter 1 hour after RANKL interacted with RANK, illustrating the close relationship between NF-κB and NFATc1. 50,51 In the cytoplasm of non-stimulated cells, NF-κB exists in a complex with IκB. IκB is then degraded by inhibitor of κB kinase (IKK) after RANKL stimulation, whereas NF-κB enters the nucleus and stimulates the transcription of key genes. 52 Therefore, IκB can be regarded as a signaling molecule that reflects NF-κB activation. As is shown in our study, both the activation of NF-κB and the degradation of IκB were repressed by MA. Previous studies found that the activation and auto-amplification of NFATc1 were mediated by Ca 2+ oscillations induced by RANKL, which demonstrated that NFATc1 was also regulated by the Ca/calcineurin pathway. 35 Combined with the Ca 2+ oscillation results, the strong effect of MA on inhibiting NFATc1 and thus preventing the formation of OCs suggests a dual role in blocking the NF-κB and Ca/calcineurin pathways.
c-Fos (an AP-1 component) is an indispensable factor that triggers a transcriptional regulatory cascade by producing and cooperating with NFATc1, thereby activating a number of target genes involved in osteoclast differentiation and function. 17 This finding was confirmed by a study that found the expression of RANKLinduced NFATc1 was abrogated in c-Fos knockout mice. 37 In addition, another study demonstrated that mice developed osteopetrosis because of c-Fos deficiency, which also showed the importance of c-Fos in RANKL-induced osteoclastogenesis. 53 In our study, the gene and protein expression of c-Fos was significantly inhibited by MA.
MAPK, which consists of JNK, ERK, and p38, is part of the RANKLinduced signaling pathway that regulates the expression of AP-1. 54 It has been reported that ERK contributes to the protection of OCs from apoptosis and the stimulation of osteoclast differentiation, 55 which explains the increase in cell number after RANKL treatment.
The blockade of JNK also leads to the failure of osteoclast formation. Although p38 is not involved in osteoclast function, this protein does participate in osteoclast differentiation. 56 Our western blot results showed that MA first inhibited the phosphorylation of ERK1/ 2 (at 10 and 20 minutes) and then suppressed the phosphorylation of JNK1/2 (at 30 and 60 minutes) after stimulation by RANKL.
Therefore, we propose that MA can regulate the activation of AP-1 F I G U R E 1 0 A schematic diagram helps in understanding the role of Madecassoside in suppressing RANKLinduced osteoclastogenesis by interfering with MAPK signaling and inhibiting the subsequent induction of NFATc1.
In conclusion, our study indicated that MA can inhibit RANKLinduced osteoclastogenesis via NFATc1 by targeting the NF-κB, Ca/ calcineurin and MAPK signaling pathways. Furthermore, with the satisfactory therapeutic effect of MA on estrogen deficiency-induced osteoporosis, which was first described and confirmed in this study, MA, a natural compound extracted from a traditional Chinese herb, may be a potential therapeutic candidate for preventing and treating osteoporosis.