Alpha modification of Eagle’s medium (α-MEM) were purchased from HyClone (GE Healthcare Life Sciences, Chicago, IL, United States). Fetal bovine serum (FBS), penicillin, and streptomycin were provided by Gibco (Thermo Fisher Scientific, Waltham, MA, United States). GPF, provided by Chengdu Must Bio-Technology (Chengdu, China), was dissolved in dimethyl sulfoxide. Antibodies specific for ß-Actin, p38, phospho (p)-p38, ERK, p-ERK (1:2,000), JNK, p-JNK, IκBα, p-IκBα, NF-κB p65, p-NF-κB p65, and c-Fos, together with the anti-rabbit or anti-mouse fluorescence-labeled IgG secondary antibodies were provided by Cell Signaling Technology (CST, Danvers, MA, United States). Antibodies specific for NFATc1 (1:250) was provided by Santa Cruz Biotechnology (Dallas, TX, United States). All antibodies were diluted at 1:1,000 unless otherwise indicated. Phenylmethylsulfonyl fluoride, RIPA cell lysis buffer, BCIP/NBT Alkaline Phosphatase Color Development Kit, Cell Counting kit-8, phosphatase inhibitors, and protease inhibitors were provided by Beyotime Institute of Biotechnology (Shanghai, China). The recombinant molecules: RANKL and M-CSF were provided by R&D Systems (Minneapolis, MN, United States).

The non-adherent bone marrow-derived macrophages/monocytes were obtained from tibias and femurs in the male C57BL/6J rats (5 weeks old), which were then inoculated into medium A (α-MEM containing 10% FBS, 50 ng/ml M-CSF and 1% PS). After four days of culture, cells were rinsed by PBS twice; then those adherent cells were collected as BMMs for experimental purposes.

BMMs (indicated cell density, 8 × 10³ cells/well) dependent on M-CSF were inoculated into a 96-well plate and were incubated overnight. After replacing the medium, GPF was added at specific doses. Ninety six hours later, all wells were added with 10 µL CCK-8 solution, and the cells were subjected to 2 h of incubation in the dark under 37°C. Later, the Cytation five Cell Imaging Multi-Mode Reader (BioTek, Vermont, United States) was employed to detect the absorbance (OD) value at 450 nm. Besides, for each sample treated with GPF, the blank OD₄₅₀ value was taken away. Then, the OD₄₅₀ value of each sample treated with GPF was divided by the control value and then multiplied by 100 to determine the viable or proliferating cell percentage. The formula is as follows: % viable   cells = ( absorbance GPF − treated   sample − absorbance blank ) / ( absorbance control − absorbance blank ) × 100.

To generate osteoclasts, BMMs were seeded in 96-well plates at indicated cell density and were cultured in medium B (medium A plus 50 ng/ml RANKL, also called “induced medium”) with indicated concentrations of GPF for seven days. We replaced medium at intervals of two days till matured osteoclasts were formed. Thereafter, cells were subjected to 4% paraformaldehyde (PFA) fixation for 15 min and then stained to detect the enzymatic activity of TRAP by using the TRAP-staining kit. Finally, osteoclasts were determined as cells having at least three nuclei. Images were captured by a Cytation 5 Cell Imaging Multi-Mode Reader.

The same culture conditions (8 × 10³ cells/well in 96-well plates, with or without GPF) were applied to osteoclastogenesis. When mature osteoclasts were formed, they were subjected to 15 min of 4% PFA fixation, washing by PBS thrice and 15 min of permeabilization using 0.1% TritonX-100. Thereafter, cells were subjected to 1 h of 0.25% bovine serum albumin (BSA) treatment, followed by 2 h of incubation in the dark using Rhodamine-Phalloidin. Following washing with PBS three times, cells were stained with DAPI for 6 min. Images were captured by fluorescence microscopy.

To observe the bone-resorptive function of osteoclasts in vitro, a resorption pit assay was performed. BMMs were then inoculated into 6-well plates, followed by five days of incubation using the induced medium. After the formation of small cells similar to osteoclasts, they were taken out of the 6-well plates, and an equivalent number of cells were inoculated in the 96-well hydroxyapatite plates at the indicated cell density and were cultured in an induced medium with indicated concentrations of GPF for about four days. After fixing 1/2 wells, cells were subjected to staining to detect the TRAP activity. In the remaining 1/2 wells, cells were bleached and removed with hypochlorous acid. Wells were washed thrice with sterile water, air-dried, and then captured using a Cytation 5 Cell Imaging Multi-Mode Reader. Resorbed area per well was measured by Image J (NIH, Bethesda, Maryland, United States) and was used to evaluate the osteoclast function.

BMMs were plated in 6-well plates (2 × 10⁵ cells/well) and incubated in an induced medium with or without GPF. When osteoclasts matured, the RNAiso Plus was used to extract cellular RNA in accordance with specific protocols. Then, 2 µg total RNA was used to prepare cDNA through reverse transcription, with the use of 4 µL of 5 × primescript RT Enzyme MIX in a total volume of 20 µL by supplementing RNase free distilled water. Afterward, the quantitative real-time PCR assay was performed on the qTOWER³ Real-time PCR Thermal Cycler by using TB Green Premix Ex Taq. The PCR conditions were as follows: 5 min under 95°C; 10 s under 95°C, 20 s under 60°C, and 20 s under 72°C for 30 cycles; and finally 10 min of elongation under 72°C. To normalize the relative gene expression, GAPDH was used as a reference gene. Primer sequences are listed in Table 1.

BMMs (5 × 10⁵ cells/well) were inoculated into the 6-well plates for overnight incubation. After 2 h of serum starving, cells were incubated with a-MEM with or without GPF for an additional 2 h. Then, RANKL stimulation was performed at specific times (0, 5, 10, 20, 30, and 60 min). After washing with PBS, the RIPA cell lysis buffer that contained PMSF, a phosphatase inhibitor, and protease inhibitor was used to lyse cells. After 15 min of centrifugation of the lysates at 12,000 rpm, the supernatants were harvested for further analysis. SDS-PAGE gels were used to separate the proteins. After migration, the Trans-Blot Turbo Transfer System (Bio-Rad Laboratories, Hercules, CA, United States) was used to transfer the separated proteins onto nitrocellulose membranes. Later, 5% (w/v) skimmed milk powder within TBST that contained 0.05% Tween-20 was used to block the membranes for 2 h under room temperature (RT), followed by overnight incubation using specific antibodies under 4°C. Then, membranes were rinsed by TBST thrice, followed by another 2 h of incubation with fluorescence-labeled secondary antibody. Finally, the ChemiDoc^TM MP Imaging System (Bio-Rad Laboratories, Hercules, CA, United States) was used to obtain images.

Reactive Oxygen Species Assay Kit was used to measure the intracellular production of ROS. The fluorescence-free DCFH-DA can freely penetrate the cell membrane. It is hydrolyzed by the intracellular esterase after it enters the cells, which generates DCFH that is unable to penetrate the cell membrane. The ROS produced in cells is involved in oxidizing the non-fluorescent DCFH for the production of fluorescent 2′,7′-dichlorofluorescein (DCF). Typically, the DCF fluorescence may be detected to assess the intracellular ROS contents. Upon 100 ng/ml RANKL stimulation with or without GPF at the 10, 20, and 40 µM doses, a-MEM that contained 10 µM DCFH-DA was further used to culture BMMs for 1 h in dark. The fluorescence microscope was also used to capture cell images. Then, the Image J software was employed to analyze the average fluorescence intensity in every cell as well as the ROS-positive cell percentage in each field.

To perform flow cytometry, the above procedures were done, and then cells were harvested, centrifuged, and resuspended. Next, we detected the fluorescence intensity at the excitation and emission wavelengths of 488 and 525 nm, respectively, through flow cytometric analysis. A total of 10,000 events were acquired for analysis. Finally, the average value was quantitatively determined.

The Institutional Animal Ethics Committee of the Second Affiliated Hospital, Shantou University Medical College, China, approved all the in vivo experiments, which conformed to all relevant regulatory standards. The mice were maintained under the pathogen-free controlled environment, and at most, five mice were kept in each cage. All animals were allowed to freely drink water and eat a standard rodent chow diet. After acclimation for one week, eighteen C57BL/6J female mice (19.5 ± 1.4 g, 11 weeks old) were randomized into three groups (n = 6 each): sham group, OVX group, and OVX + 10 mg/kg GPF group. According to standard procedures, under anesthesia with 3% chloral hydrate via intraperitoneal injection, surgical bilateral ovariectomy was performed on the mice in the OVX + GPF group and OVX group. Ovaries were exteriorized from mice in the sham group rather than their removal. After surgery, every mouse was injected intraperitoneally with 80,000 units of penicillin to prevent infection. One week after postoperative recovery, intraperitoneal injection of the indicated concentration of GPF was delivered to the OVX + GPF group every two days for six weeks. As a vehicle control, normal saline (NS) was injected into mice in both OVX and sham groups. Six weeks later, all mice were killed through cervical dislocation. 4% formaldehyde was used to fix the excised left femurs for histological and Micro-CT analysis.

The distal region of the left femur was used for micro-computed tomography (micro-CT) analysis. Femurs were resected from mice immediately after their sacrifice and were subjected to 24 h of 4% PFA fixation, followed by scanning by the Skyscan 1272 micro-CT scanner (Bruker micro-CT, Kontich, Belgium). The following scanning parameters were set: source current, 142 μA; source voltage, 70 V; rotation step, 0.20 degree; pixel size, 9 µm; and AI, 0.5 mm filter. A region of interest (ROI) 0.5 mm over the dismal femur growth plate; height, 2 mm was selected to analyze the trabecular bone. Then, some bone-associated parameters—bone volume/tissue volume (BV/TV), trabecular thickness (Tb. Th), trabecular number (Tb. N), connectivity density (Conn. Dn), Bone surface (BS), as well as trabecular separation (Tb. Sp)—were examined.

After micro-CT scanning, the EDTA decalcified solution (pH 7.4) was used to decalcify the left femurs for three weeks, followed by dehydration and paraffin embedding. Then, all specimens were sliced into the 4-µm sections by the microtome for H&E staining to observe the bone microstructure of the femur and TRAP for the visualization of osteoclasts. The Nikon microscope (Nikon Corporation, Minato, Tokyo, Japan) was used to capture images.

All experiments were repeated at least three times. SPSS 19.0 (IBM, United States) was utilized for statistical analysis. The quantitative data were expressed in the manner of mean ± standard deviation (SD). Student’s t-test was used to determine the difference between two groups, and one-way ANOVA was used to assess the difference of more than two groups. A difference of p < 0.05 indicated statistical significance.

The chemical structure of GPF is demonstrated in Figure 1A. We first tested whether GPF exhibits cytotoxicity against primary BMMs by CCK-8. We found that GPF had an insignificant effect on the viability of BMMs at concentrations below 80 µM (Figure 1C). Therefore, the GPF level was used in later analysis. For assessing how GPF affected the formation of osteoclasts, the medium supplemented with M-CSF and RANKL with or without increasing concentrations of GPF as indicated was used to culture BMMs. Then, TRAP staining suggested that the differentiation of osteoclasts was significantly inhibited (Figures 1B, D,E). 40 µM GPF was added to the culture medium at different stages (early period: day 1–3, middle period: day 3–5, later period: day 5–7, whole period: day 1–7) during the osteoclast differentiation, that facilitated a better understanding of the osteoclast formation degree with the highest sensitivity to GPF exposure (Figure 1F). As a result, inhibition of osteoclastogenesis occurred in all three stages, but more obviously in the early stage (Figures 1G–I).

For better identifying the molecular impact of GPF on osteoclasts in vitro, osteoclast-specific gene expressions, including c-Fos, NFATc1, and Acp5, were tested by quantitative real-time PCR during osteoclast formation with or without GPF treatment. As shown, osteoclast-specific gene expressions of the control group were remarkably upregulated during osteoclast differentiation, but in the GPF group, compared to the control group, these gene expressions were inhibited by GPF in a dose-dependent manner (Figures 1J–L).

Rhodamine-phalloidin was used to stain cells for the visualization of morphological alterations and the formation of podosome belts within cells subjected to GPF treatment or not. In line with the suppressed osteoclast differentiation, as mentioned above, the GPF treatment group showed markedly decreased podosome belt area in each field and showed reduced nucleus number in each osteoclast (Figures 2A–C).

We next performed a resorption pit assay to observe the bone-resorptive function of osteoclasts in vitro. The osteoclast number and the resorbed area per osteoclast decreased significantly with GPF application (Figures 2D–F). Therefore, the compound exhibited an inhibitory effect on osteoclast bone-resorptive function as the bone resorption assay showed, and possibly inhibited the terminal process when pre-osteoclast-like cells turned into mature osteoclasts. The above findings indicate that GPF suppressed osteoclastogenesis and osteoclast activity.

To examine the interfering effect of GPF on the RANKL-mediated ROS generation in cells during osteoclast differentiation, a Reactive Oxygen Species Assay kit was used. As cell images showed, stimulation of BMMs increased the intensity of DCF fluorescence. GPF dose-dependently inhibited RANKL-stimulated ROS production in BMMs (Figures 3A,B). Furthermore, flow cytometry analysis also indicated that treatment with GPF inhibited the RANKL-mediated increase in intracellular ROS (Figure 3C). Thus, current findings suggested that GPF suppressed osteoclastogenesis probably by scavenging intracellular RANKL-generated ROS in osteoclast precursors.

This study determined how GPF affected the MAPK as well as NF-κB pathways to better illustrate the mechanism underlying inhibition of the GPF on osteoclastogenesis. With regard to the MAPK pathways, RANKL down-regulation increased the phosphorylation levels of JNK, p38, and ERK (Figure 4A). It was observed that GPF inhibited phosphorylated ERK at 10–20 min (Figure 4B) and JNK at 5–30 min (Figure 4C) but had little effect on phosphorylated p38 (Figure 4D). We also found that GPF made no difference to RANKL-activated p65 and IκBα (Figures 4A,E,F). Collectively, our data suggested that GPF influence osteoclastogenesis probably by inhibiting RANKL-induced activation of ERK and JNK, rather than p38 and NF-κB pathways.

We next examined whether GPF regulates activation of c-Fos and NFATc1. According to Figures 4G–I, RANKL-induced NFATc1 and c-Fos protein expression levels were significantly suppressed by GPF. Overall, the findings in this study suggested that GPF might scavenge the RANKL-triggered ROS production and attenuate the RANKL-regulated JNK and ERK signaling activation to suppress the activation of the downstream c-Fos and NFATc1. Thus, this affected osteoclast-specific gene expression, finally suppressing osteoclast formation and bone resorption activity.

Having demonstrated that GPF inhibits osteoclast formation and bone resorption, we hypothesized that GPF might ameliorate ovariectomy-induced osteoporosis. As shown in Figure 5A, the mice underwent OVX- or sham-surgery and were given an intraperitoneal injection of 10 mg/kg GPF or normal saline (NS) every two days for six weeks after postoperative recovery. GPF injections relieve ovariectomy-induced bone loss as demonstrated by micro-CT analysis (Figure 5B). Relevant bone parameters were measured, and quantitative analysis showed the increased BV/TV, BS, Tb. N, Tb. Th and Conn. Dn levels, whereas decreased Tb. Sp level of OVX + GPF group in comparison with the OVX + NS group (Figures 5E–J). Also, GPF could obviously relieve ovariectomy-induced bone loss, as was exhibited by HE staining (Figures 5C, K). Then, TRAP staining (Figure 5D) was done to investigate the role of GPF on osteoclast formation during osteoporosis. The mean TRAP-positive cell percentage per bone surface were respectively measured. It showed that, when compared with the NS-treatment group, GPF treatment significantly decreased osteoclastogenesis (Figure 5L).