BTP samples were prepared and provided by Neijiang traditional Chinese medicine hospital (Neijiang, China). The powder of BTP (10 g) was ultrasonically dissolved with 300 ml, 75% ethanol, followed by concentrating to 100 ml using a rotary evaporator. Then, the supernatant was filtered by microporous membranes with a pore size of 0.22 μm. Then, the obtained 10 μl filtrates of BTP were analyzed by UPLC-MS/MS.

For qualitative analysis, a Thermo Scientific Q Exactive Orbitrap HRMS (Thermo Fisher Scientific, Massachusetts, United States) was connected to a Thermo Scientific Vanquish UPLC (Thermo Fisher Scientific, Massachusetts, United States). Chromatographic separation was achieved on a Thermo Scientific Accucore C₁₈ (3 × 100 mm, 2.6 μm) in a thermostatically controlled column compartment (30 °C). The aqueous and organic mobile phases used were acetonitrile (A) and 0.1% phosphoric acid aqueous solution (B), respectively. The gradient elution program was set up as follows: 0–20 min, 12–25% A; 20.1–30 min, 25–28% A; 30.1–32 min, 28–40% A; 32.1–40 min, 40–50% A; 40.1–60 min, 50% A; 60.1–65 min, 12% A. The flow rate was set as 0.3 ml/min, and 2 μl of the extraction was injected into the UPLC system. The instrument was operated in the positive ion mode to perform full-scan analysis over an m/z range of 100–1,500. The optimized parameters were set as follows: a sheath gas flow rate of 35 L/min; a spray voltage of 3000 V; a capillary temperature of 320 V; an aux gas flow rate of 10.00 L/min; a max spray current of 100 A; a probe heater temperature of 350 C; and an S-lens RF level of 50.00%.

Compounds of BTP were analyzed by UPLC-MS/MS. Subsequently, targets regarding these identified compounds via UPLC-MS/MS were obtained from the online database of SwissTargetPrediction (http://www.swisstargetprediction.ch/) (Daina et al., 2019).

As a common worldwide resource for gene expression studies, the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo) allows open access to high-throughput gene expression and other functional genomics datasets (Clough and Barrett, 2016). The GEO database was searched using “osteonecrosis of femoral head” as the keyword. We aimed to find a comparison of gene expression data sets between femoral head samples from patients with ONFH and normal femoral head samples. In order to ensure the stability and credibility of the analysis results, only data sets with a sample size larger than 10 were selected. Finally, only one suitable data set of ONFH was downloaded from the GEO database, namely, dataset GSE123568. In this dataset, 10 normal femoral head samples and 30 ONFH samples were included.

In order to obtain DEGs, dataset GSE123568 was analyzed with GEO2R, which was the official differential expression analysis tool of the GEO database. In this study, the difference with logFC (fold change) ≥1.3 and p-values <0.01 was statistically significant, and the corresponding genes were identified as DEGs.

In this study, the protein–protein interaction (PPI) of BTP and DEGs from ONFH were constructed by Cytoscape software (3.7.2). Then, the overlapping PPI network was obtained by using BisoGenet in Cytoscape. The median of three important attribute values, namely, degree, betweenness, and closeness, was employed for screening to obtain the core gene. After two screenings, a PPI network containing 17 nodes and 114 edges was obtained and used for enrichment analysis. The detailed attribute values and screening process of these targets are presented in Table 3 and Figure 3, respectively.

The Metascape platform (http://metascape.org/gp/index.html) has a comprehensive annotation function and data of gene annotation monthly updated (Zhou et al., 2019). In this study, functional annotation of gene ontology (GO) and KEGG pathways was performed with the Metascape platform. The main biological processes and involved metabolic pathways were analyzed by submitting the core targets of BTP-DEGs from ONFH to Metascape with p < 0.01. Then, the results were saved and visualized by Bioinformatics (http://www.bioinformatics.com.cn/) (Huang, 2009). Afterward, as shown in Figure 4, the network of the BTP-DEGs-Pathway was constructed with Cytoscape (3.7.2).

Fetal bovine serum (FBS), phosphate-buffered saline (PBS), penicillin–streptomycin, 0.25% trypsin–EDTA (1x), and Dulbecco’s modified eagle’s medium (DMEM) were purchased from GIBCO (Grand Island, NY, United States). A cell counting kit-8 (CCK8) detection kit was purchased from the Beijing 4A Biotech Co., Ltd. (Beijing, China). The assay kits for DCFH-DA, horseradish peroxidase (HPR)-conjugated secondary antibodies, β-actin, BCA protein assay reagents, and primary antibodies for cleaved (C) caspase-3 were purchased from Boster Biol. Tech. (Wuhan, China). Primary antibodies for phosphorylation- (p-) Akt, Akt, PI3K, and P-PI3K were obtained from ImmunoWay Biotechnology Co. (Suzhou, China). Annexin V-FITC/PI apoptosis kits were obtained from US Everbright^® Inc (Suzhou, China). Polyvinylidene fluoride (PVDF) membranes were purchased from Sigma-Aldrich (Shanghai, China). 5,5′,6,6′-Tetrachloro-1.1′,3.3′-tetraethyl-imidacarbocyanine iodide (JC-1) was obtained from Jiangsu KeyGen Biotech (Nanjing, China).

The RAW 264.7 cells were purchased from Wuhan Pu-nuo-sai Life Technology Co., Ltd. (Wuhan, China) and used throughout the study. The RAW264.7 cells were cultured in DMEM containing 10% FBS (v/v) and 1% penicillin–streptomycin (v/v) at 37°C in a humidified atmosphere of 5% CO₂.

The RAW 264.7 cells were incubated in 96-well plates (1×10⁴ cells/well) with different concentrations (0, 5, 10, 20, 40, 60, 80, 100, 150, and 200 μg/ml) of BTP, followed by induction with RANKL (100 ng/ml). After incubation for 24 and 48 h, the supernatant was discarded, and the cells were washed three times with PBS. Then, 10 µl CCK-8 was added to each well and incubated for 0.5 h. The optical density (OD) values were measured at 450 nm by the microtablet reader (Bio-RAD, United States). Each experiment was repeated three times. The concentration- and time-dependent effects of BTP on raw 264.7 cells are presented in Figure 5.

The RAW 264.7 cells were seeded in 24-well plates (1×10⁵ cells/well) and then pretreated with BTP (0, 50, 75, and 100 μg/ml) for 24 h, followed by induction with RANKL (100 ng/ml) for 7 days. Untreated cells were used as a negative control, while only RANKL-induced cells were used as a positive control group. During the induction period, the medium was changed every other day. After 7 days, the cells were washed with PBS, followed by fixing with 4% paraformaldehyde for 30 min, and then stained to detect TRAP activity. Finally, trap-positive cells were enumerated via a microscope. Mature osteoclasts with more than three nuclei were counted in a random area in each replicate sample.

The RAW 264.7 cells were seeded in 96-well plates (1×10⁴ cells/well) containing bovine bone slices and then pretreated with different concentrations of BTP (0, 50, 75, and 100 μg/ml) for 24 h, followed by induction with RANKL (100 ng/ml) for 9 days. Then, bovine bone slices seeded with cells were washed with PBS, fixed in 2.5% glutaraldehyde over 10 min, ultrasonically cleaned with 0.25 mol/l ammonia hydroxide, and stained with 1% toluidine blue over 10 min. After washing with ddH2O, the osteolysis pits were photographed with a light microscope. Bone resorption capacity was defined as the bone resorption relative area (resorption area/total bone area).

Cell apoptosis was detected by flow cytometry (CytoFLEX FCM, Beckman Coulter Inc., Atlanta, Georgia, United States). The RAW 264.7 cells were incubated in 6-well plates with RANKL (100 ng/ml) for 5 days and then treated with BTP (0, 50, 75, and 100 μg/ml) for 24 h. Subsequently, cell apoptosis was detected with the manufacturer’s instructions of the Annexin V/PI apoptosis kit and cell cycle kit; the cells were stained with Annexin V-FITC and PI, respectively, and then detected by flow cytometer analysis. Each experiment was repeated three times.

The change of intracellular mitochondrial membrane potential (MMP, ΔΨm) is generally recognized as an important indicator of mitochondrial dysfunction, and JC-1 is commonly considered as an ideal probe to evaluate ΔΨm. In this study, the cells were inoculated in laser confocal dishes (1×10⁴/well), treated with different concentrations of BTP for 24 h, and subsequently induced with RANKL (100 ng/ml) for 5 days. After the induction, the cells were incubated with the JC-1 probe (10 μg/ml) in the dark for 30 min. Confocal laser microscopy was used to observe the green and the red fluorescence (Leica, SP8 SR, Wetzlar, Germany) of cells at 488 and 647 nm, respectively, and the images were recorded.

The RAW 264.7 cells were seeded in the confocal dish (1×10⁵) and processed as described above. Then, they were washed three times with PBS, fixed with 4% paraformaldehyde for 30 min, and then washed three times with PBS. Afterward, the cells were treated with 0.5% Triton-X-100 for 30 min, followed by blocking with goat serum for 1 h at room temperature, and then, PI3K and P-PI3K antibodies (1:500) were added separately and incubated overnight at 4°C. Soon afterward, antirabbit IgG (H&L) Alexa Fluor^® 488 and antimouse IgG (H + L) Alexa Fluor^® 647 were added and incubated at room temperature for 1 h in the dark. Then, PBS was used to wash out the unbound secondary antibody, and the nuclei were stained with DAPI. Then, AKT/P-AKT and their fluorescence intensity were observed using a confocal laser microscope (Leica, SP8 SR, Wetzlar, Germany).

The RAW 264.7 cells were inoculated in 6-well plates (2×10⁶/well) at 37°C in a humidified atmosphere of 5% CO₂ and incubated for 24 h. Then, the RAW 264.7 cells were treated with RANKL (100 ng/ml) for 5 days and then treated with BTP (50, 75, and 100 μg/ml) for 24 h. At the end of the treatment, 100 μl of RIPA cell lysates was used to extract the total cell protein. The protein concentrations were determined using BCA protein assay reagents. Then, the total protein was separated by 12% SDS-PAGE and then transferred to the PVDF membrane. After blocking by 5% skimmed milk, the PVDF membrane was incubated overnight with diluted primary antibodies of PI3K, P-PI3K, C-caspase-3, Akt, and p-Akt (dilution 1:1,000) at 4°C. Subsequently, the PVDF membrane was incubated for 1 h with the diluted secondary antibody. Finally, the protein bands were stained with ECL detection kits, and β-actin was used as the internal reference. Image analysis software ImageJ (version 1.51, National Institutes of Health, MD, United States) was utilized for gray analysis.

All the data were presented as the mean ± standard deviations (SDs), statistical comparisons were evaluated using one-way analysis of variance (ANOVA), and significant differences between the mean values were measured using Duncan’s multiple range test. p < 0.01 was deemed statistically significant.

A total of 34 components were identified from BTP, of which six compounds belonged to Angelica sinensis (Oliv.) Diels, seven to Astragalus membranaceus (Fisch.) Bge, six to Glycyrrhiza uralensis Fisch., six to Panax notoginseng (Burk.) F. H. Chen, and nine to Pheretima aspergillum (E.Perrier). Their specific information and total ion current diagram are shown in Table 1, Table 2, and Figure 2, respectively. Moreover, a total of 1,232 related targets were obtained from the online databases.

A total of 178 DEGs were screened out, including 143 upregulated genes and 35 downregulated genes, and were visualized with a volcano plot, as shown in Figure 2. In the volcano plot, the vertical axis represents log₂ FC (fold change), and the horizontal axis represents log₁₀ (p-value). The DEGs with log₂ FC > 0 were defined as upregulated genes, and the others were downregulated genes. Additionally, the expression situation regarding all the DEGs was visualized with a heatmap, as shown in Figure 3A. In the heatmap, the vertical axis represents DEGs, while the horizontal axis represents the sample. DEGs were effectively divided into normal and ONFH groups. Red indicates the upregulated gene, and green represents the downregulated genes.

Centrality expresses the degree of a node in a certain network in the center of the whole network. The degree is defined as the sum of direct connections between nodes and represents the cohesion of a given node in a certain network. Furthermore, betweenness indicates the number of shortest paths passing through a node. These three factors are the indispensable basis for assessing the importance of a specific target. Through screening, 17 core targets with 114 vital interactions were obtained, and the detailed information and identification process of core targets are shown in Table 3 and Figure 3C, respectively.

The signal pathway of core targets was analyzed using the Metascape platform, and the results were visualized by bioinformatics. The results showed that the functions of core targets were closely related to the occurrence of ONFH. KEGG and GO enrichment are presented in Table 4 and Figure 4, respectively.

Based on the results of KEGG enrichment, as shown in Table 4, multiple pathways jointly contributed to the antiONFH effect of BTP, including the PI3k-Akt signaling pathway, prostate cancer, pathways in cancer, the cell cycle, the estrogen signaling pathway, protein processing in the endoplasmic reticulum, hepatitis c, proteoglycans in cancer, the central carbon metabolism in cancer, small-cell lung cancer, progesterone-mediated oocyte maturation, endocrine resistance, oocyte meiosis, fluid shear stress and atherosclerosis, and breast cancer. Thus, it suggested that the antiONFH effect of BTP might associate with a synergistic effect of multiple pathways.

In conformity with the results of GO enrichment, the biological processes (BPs) in which BTP chiefly participated involved DNA repair, cell cycle phase transition, regulation of cell cycle phase transition, regulation of the cell cycle process, mitotic cell cycle G1/S transition, cell cycle G1/S phase transition, mitotic cell cycle phase transition, positive regulation of the cell cycle, the DNA biosynthetic process, response to topologically incorrect protein, protein insertion into the membrane, the proteasome-mediated ubiquitin-dependent protein catabolic process, regulation of mitotic cell cycle phase transition, regulation of cell cycle G2/M phase transition, and developmental growth involved in morphogenesis.

Such effects were primarily reflected in several aspects, including ubiquitin-protein ligase binding, ubiquitin-like protein ligase binding, nitric-oxide synthase regulator activity, disordered domain specific binding, kinase binding, protein kinase binding, protein domain specific binding, protease binding, cell adhesion molecule binding, histone deacetylase binding, unfolded protein binding, integrin binding, protein phosphatase binding, phosphatase binding, and chromatin binding. The targets of this action were chiefly concentrated in multiple cell sites, including vesicle lumen, melanosomes, pigment granules, secretory granule lumen, cytoplasmic vesicle lumen, the perinuclear region of the cytoplasm, ficolin-1-rich granule lumen, the intracellular protein-containing complex, focal adhesion, the cell–substrate junction, growth cones, neuronal cell bodies, sites of polarized growth, ficolin-1-rich granules, and the protein-DNA complex.

Based on the above bioinformatics analysis, previous reports, and our preliminary experiments, we deduced that the molecular mechanisms of this TCM formula might attribute to PI3K/Akt-related cell apoptosis.

The network of key targets-pathway was constructed with Cytoscape (3.7.2), as presented in Figure 3B. V represents the pathway, and the diamond represents the DEGs. The greater the significance, the larger the volume of the graph was. The results showed that the targets of TP53, EGFR, and CDK2 and the pathways of the PI3K-AKT signaling pathway and cell cycle were more critical in the entire network. Therefore, subsequent experiments chose these two pathways for verification.

The cytotoxicity of BTP on normal RAW 264.7 cells was assessed by CCK-8 assays. As shown in Figure 5, BTP at concentrations up to 200 mg/ml did not exhibit obvious cytotoxicity on RAW264.7 cells within 24 and 48 h of incubation.

At 7 days after inducting RAW 264.7 cells with RANKL (100 ng/ml), TRAP-positive multinucleated cells were observed under a light microscope, as shown in Figure 6. Interestingly, BTP at 20, 40, and 80 μg/ml markedly decreased the number of osteoclasts in a dose-dependent manner (p < 0.05 at 100 μg/ml, p < 0.005 at 200 μg/ml, p < 0.0001 at 400 μg/ml). In addition, the significant inhibitory effect of BTP on bone resorption in vitro has also been confirmed. As shown in Figure 8, compared with the control group, BTP dose-dependently reduced the area of bone resorption.

In this study, two comprehensively recognized cell apoptosis detection methods, including AO-EB double staining and FITC-conjugated Annexin V/PI staining, were carried out using a laser confocal microscope and flow cytometry analysis. Therefore, we further explored the apoptosis induced by BTP treatments. As shown in Figure 7A, our present results confirmed our hypothesis. After 5 days of RANKL treatment, the RAW 264.7 cells were induced to osteoclasts. As shown in Figure 7A, BTP at the concentrations of 50, 75, and 100 μg/ml could induce apoptosis in osteoclasts, compared to the control osteoclasts. Moreover, the results of immunofluorescence also presented the same trend. As far as we know, JC-1 aggregates in the mitochondrial matrix to form polymers that emit red fluorescence under normal physiological conditions and green fluorescence when MMP is reduced. Thus, the changes of ΔΨm could be directly reflected by the fluorescence transformation. In Figure 7B, compared with the control group, the green fluorescence of the experimental group was getting stronger, while the red fluorescence was getting weaker, indicating that MMP breakdown happened.

The results suggested that BTP (50, 75, and 100 μg/ml) could induce remarkable apoptosis in osteoclasts (RANKL induced RAW 264.7 cells). Furthermore, the apoptosis markers of C-caspase-3, bcl-2, and bax were determined. As shown in Figure 8, BTP could upregulate the expression of C-caspase three and bax, downregulating bcl-2 expression in osteoclasts in a dose-dependent manner.

The PI3K/Akt pathway is a classical signaling pathway responsible for regulating cell proliferation, differentiation, apoptosis, and a series of critical physiological activities. Nevertheless, only the phosphorylation of PI3K and Akt mainly participate in these physiological activities. Besides, growing numbers of studies have found that cell apoptosis is closely related to the PI3K/Akt pathway (Xue and Feng, 2018; Zhan and Yan, 2020). Therefore, we evaluated the effect of BTP on the PI3K/Akt pathway in osteoclasts. In the present study, both immunofluorescence and western blot assays were performed to determine the effect of BTP on the PI3K/Akt signaling pathway in osteoclasts. The results revealed that after pretreatment with BTP, the expression of p-Akt and p-PI3K was reduced, while no noticeable change was evident in the expression of AKT and PI3K compared to the control group (Figure 8B). Likewise, as presented in Figure 8C, the result of the immunofluorescence revealed that green fluorescence becomes weaker, indicating that the expression of p-Akt was reduced, and the red fluorescence (the expression of Akt) had no apparent change.