The U251, TG905, and A172 cell lines were purchased from the Chinese Academy of Sciences Cell Board and Keygen Biotech.

RXLD was purchased from the Bozhou Medicine Market (Anhui, China)and was identified by Dr. YuYun Fan from the Department of Traditional Chinese Medicine in Linyi People’s Hospital. All RXLD samples were powdered and passed through a 100-mesh sieve before extraction. The RXLD powders (20 g) were extracted with 100 ml of 75% ethanol overnight at room temperature. The extracts were centrifuged at 3,500 rpm for 10 min, and the supernatants were collected. Then we applied a set of device named rotary evaporation, which is consists of a low temperature coolant circulation pump (DLSB-5L/10, Gongyi, China), a heating bath (BUCHI, B-100, Switzerland), a rotary evaporator (BUCHI, R-100, Switzerland), and a vacuum pump (BUCHI, I-100, Switzerland) , to separate the ethanol from RXLD extracts. Undergoing the rotary evaporation at a low vacuum condition, the ethanol was evaporated efficiently and the ingredients in the drug were not damaged. The extracts were dried using a vacuum drier (YIHENG17, Shanghai, China) overnight, and frozen by a lyophilizer (BIOCOOL, Beijing, China) afterwards. Finally, we obtained the extracts as powder, and the yield was 30%. The final extracts were dissolved in dimethyl sulfoxide (DMSO) in order to formulate a 100 mg/ml stock solution for use in subsequent experiments.

Cell proliferation assays were completed by using a CCK-8 assay (BestBio, Shanghai, China). Human GBM cell lines A172, TG905, and U251 were seeded into 96-well plates at a density of 3,500 cells/well and incubated overnight before treatment with different concentrations of the RXLD samples. After 24−96 h, the CCK-8 reagent was added to each well, and the absorbance values were measured by using an M5 microplate reader (Molecular Devices, USA) at the wavelength of 450 nm. Finally, we used GraphPad Prism 8.0 software to construct the drug-efficient relationship of the tested drugs and calculate the IC₅₀ value.

Apoptosis assays were conducted by using the annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) double staining apoptosis detection kit (BestBio, Shanghai, China) according to the manufacturer’s instructions. Firstly, the cells were seeded into 6-well plates at a density of 2× 10⁵ cells/well and incubated overnight before treatment with different concentrations of the RXLD samples. After incubating the RXLD samples for 48 h, the cells and their supernatant were harvested and washed twice with ice-cold phosphate-buffered saline (PBS). After centrifugation allowed us to discard the supernatant, the cells were resuspended in PBS and incubated with FITC at 4°C for 15 min, and then incubated with PI at 4°C for 5 min, centrifuged once again in order to discard the supernatant, and then finally resuspended in PBS prior to the analysis via flow cytometry (within 1 h).

Cells were seeded into 6-well plates at a density of 1× 10⁶ cells/well and were incubated overnight before treatment with different concentrations of the RXLD samples. After being treated with RXLD samples for 48 h, the cells were harvested and subjected to the following assays. According to the manufacturer’s instructions, cell cycle assays were conducted using a cell cycle detection kit (BestBio, Shanghai, China). The cells were washed twice with ice-cold PBS and were fixed in 70% ethanol at 4°C overnight before being centrifuged to discard the supernatant. After having their precipitation washed twice with ice-cold PBS, the cells were resuspended in PBS and incubated with RNase A at 37°C for 30 min, and then incubated with PI at 4°C for 1 h (while avoiding exposure to light). The cell cycle distribution was assessed by flow cytometry (BD).

Cells were seeded into 6-well plates at a density of 5× 10⁵ cells/well. When the cell confluence reached 80-90% overnight, we used a 200-µL pipette tip to draw a line in the middle of each well. We washed the cells with PBS and incubated them with different concentrations of the RXLD samples and with serum-free medium for 48 h. The cell healing results were observed by microscopy at 0, 6, 12, 24, and 48 h after the addition of the RXLD samples, while ImageJ and GraphPad were used for analysis.

Cells were seeded into 6-well plates at a density of 2 × 10⁵ cells/well and were incubated overnight before their treatment with different concentrations of the RXLD samples for 48 h. The cells were suspended by serum-free medium and then transferred to the upper layer of the transwell chamber (while a complete medium was added to the lower chamber) and cultured 24 h. Finally, the cells were fixed using a multi-formaldehyde solution, stained with crystal purple, and then observed and pictured under the microscope.

Cells were seeded into 12-well plates at a density of 8 × 10⁴ cells/well and were incubated overnight before their treatment with different concentrations of the RXLD samples. After incubating the RXLD samples for 48 h, we harvested the cells and stained them with a 2’,7’-dichlorofluorescein (DCF) probe according to the ROS kit (BestBio, Shanghai, China) instruction manual. We finally employed flow cytometry for the assessment of the active oxygen species assay.

Cells were seeded into 6-well plates at a density of 2 × 10⁵ cells/well and were incubated overnight before their treatment with different concentrations of the RXLD samples. After incubation of the RXLD samples for 48 h, we harvested the cells and employed the JC-1 probe to stain the cells according to the JC-1 mitochondrial membrane potential detection kit (BestBio, Shanghai, China) instructions. The stained cells were then observed under a fluorescence microscope.

The cells were seeded into 6-well plates at a density of 5 × 10⁵ cells/well and were incubated overnight prior to their treatment with the RXLD samples. After 48 h, the cells were harvested and lysed on ice, and their proteins were extracted according to the instructions of the protein extraction kit (BestBio, Shanghai, China). Protein concentrations were determined using the bicinchoninic acid (BCA) protein assay kit. After determining the protein concentrations, the extracted lysates were denatured by a sodium dodecyl sulfate (SDS) loading buffer. Equal amounts of protein from the lysates were separated by the SDS-PAGE and transferred onto polyvinylidene fluoride (PVDF) membranes. Membranes were blocked with 5% non-fat milk in TBST (TBS+Tween) and were incubated with primary antibodies at 4°C, overnight. Subsequently, the membranes were washed and incubated with secondary antibodies for 1 h, then washed and visualized by using chemiluminescence (ECL, Amersham-Pharmacia, Uppsala, Sweden). Finally, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal loading control.

All experiments were performed in triplicates and repeated at least twice for each experiment. Two-group comparisons were analyzed for variation and significance using the Student’s t-test or the Pearson χ² test. All data shown are mean ± standard deviation (s.d.), while Pearson’s correlation coefficient was also used to measure the correlation of the gene co-expression.

By putting 154 chemical ingredients of RXLD into the PharmMapper database, we obtained 97 kinds of pharmacophore models, of which 39, 17, 14, 20, and 7 corresponded to flavonoids, coumarins, diterpenoids, lignans, and volatile oils and other ingredients, respectively (Supplementary 1). In total, 734 targets were obtained by the chemical structures in the pharmacophore models, which might represent the drug targets.

Firstly, we merged the GBM-related disease targets obtained from the various databases (3,068 targets) (Supplementary 2) and constructed a Venn diagram for GBM targets ( Figure 1A ). Secondly, we combined the pharmacophore models of the RXLD active ingredients with the GBM targets to build the RXLD-related GBM targets’ Venn diagram, a process that returned 216 intersection targets ( Figure 1B ). The RXLD-related GBM targets’ network ( Figure 2 ) (Supplementary 3, 4) consisted of 78 nodes and 111 edges, wherein the nodes represented 26 compounds ( Table 1 ) and 52 genes. Among them, the degrees of sphondin (COMP14), neochamaejasmin A (COMP5), wikstrol A (COMP12),subtoxin A (COMP22), isochamaejasmenin (COMP2), and chamaejasmine (COMP1) were greater than or equal to 5; as a result, these compounds were identified as the most important compounds in this network. RGS18, Calml3, E2F1, MDM2, and RAC1 were the top 5 targets in gene nodes. The above results showed the above-mentioned active ingredients and targets were in the hub position in the network.

We have put 216 RXLD-related GBM target genes into the String database ( Figure 3 ), while the screened initial PPI network consisted of 167 nodes and 496 edges. After employing “BC > 50.9, CC > 0.53, DD > 4, EC > 0.02, the LAC > 1.5, and NC > 2” for the screening conditions (all median) criterion, the generated sub-network consisted of 51 nodes and 209 edges. The second screening was performed after employing “BC > 19.88, CC > 0.45, DD > 7, EC > 0.099, LAC > 2.85, and NC > 3.7” for the screening conditions criterion and obtained the final network that consisted of 13 nodes and 39 edges. The top 5 node degrees in the final network were TP53, EP300, VEGFA, CTNNB1, and CCND1, which we speculated as representing core targets for RXLD in treating GBM (Supplementary 5).

We have obtained 2,035 biological processes through the GO enrichment analysis (Supplementary 6). Figure 4 shows the top 10 functional categories identified, mainly involving biological processes such as oxidative stress, cell proliferation, cell cycle, cell invasion, and migration. A total of 114 signaling pathways were obtained through the undertaken KEGG enrichment analysis (Supplementary 7). The top 5 signaling pathways were the MAPK signaling pathway, the PI3K-AKT signaling pathway, the RAS signaling pathway, and the cell cycle, and the glioma signaling pathway ( Table 2 ).

The target-signaling pathway network was composed of 67 nodes and 147 edges, and larger nodes represented more important genes and pathways in the network. Among them, HRAS, PRKCB, MAPK9, CCND1, and TP53 were core targets in this network ( Figure 5 ) (Supplementary 8, 9).

We chose the active pharmacological compounds that demonstrated the higher degree of nodes in the RXLD-related GBM targets network as ligands and chose the core proteins that were screened in the PPI network as receptors for docking. Finally, we got 12 groups of ligand-receptor docking results ( Table 3 ; Figure 6 ). The docking binding energy values of all groups were < 0 kcal/mol, indicating a spontaneously ability between ligands and receptors. Among them, 11 groups of docking binding energy values were < -7.0 kcal/mol, suggesting a strong combining ability between ligands and receptors.

Through the cell proliferation assay, we found that the RXLD-induced cell proliferation inhibition was time- and dose-dependent ( Figure 7A ) in the A172 cell line, while the IC50 was 0.8 ± 0.08 mg/ml ( Figures 7B, C ). When we treated the GBM cell lines (A172, TG905, and U251) with 0.8 mg/ml of RXLD extracts, we found that the cell viability of the U251 and A172 cell lines were significantly decreased (P < 0.01) ( Figure 7D ); among them, the decline of the cell viability of the A172 cell line was the most evident, and this is why the A172 cells were used in the subsequent experiments. The annexin V-FITC/PI apoptosis experiments showed that the apoptosis rate was time- and dose-dependent ( Figure 7E ), while the levels of the apoptosis-related genes p53, cleaved-caspase3, and Bax were found to be gradually increased, and the Bcl-2 levels were gradually decreased (P< 0.01) ( Figure 7F ). As a result, the RXLD extracts were shown to inhibit the proliferation of GBM cells and to induce apoptosis.

When we treated the A172 cell line with RXLD extracts, we found that as the drug concentration increased, the number of GBM cells in the G0/G1 phase decreased in the cell cycle assay, while the number of GBM cells in the S and G2/M phases increased. The difference was statistically significant (P< 0.001; Figures 8A, B ). Western blotting experiments suggested that the protein levels of β-catenin, proliferating cell nuclear antigen (PCNA), and cyclin E were gradually declining. The cyclin-dependent kinase 2 (CDK2) and cyclin A2 protein levels gradually increased ( Figures 8C, D ). The above results indicate that the RXLD extracts block the cell cycle in the G2/M and the S phases, thereby affecting GBM cells’ proliferation through the Wnt/β-catenin pathway.

The results of the cell wound healing and the cell migration experiments revealed that RXLD extracts (0.3 mg/ml) could inhibit the migration of A172 cells ( Figures 9A, B ). Western blotting experiments indicated that as the drug concentration increased, the protein expression levels of matrix metalloproteinase 2 (MMP2) and MMP9 gradually declined (P< 0.01; Figure 9C ). These results indicate that the RXLD extracts can inhibit GBM cell migration.

The ROS experiments revealed that the DCF peak was significantly moving toward the right after the A172 cells were treated with RXLD extracts (compared to the control group), suggesting that the intracellular reactive oxygen levels increased ( Figure 10A ). JC-1 is a fluorescent probe used to detect mitochondrial membrane potential (ΔΨm). The decrease of ΔΨm is a hallmark event in the early stages of apoptosis and related to ferroptosis. Compared with the control group, the green fluorescence levels increased significantly after the A172 cells were treated with 0.3 mg/ml of RXLD extracts ( Figure 10B ), suggesting that the mitochondrial membrane potential decreased and that the cells were going through an early apoptosis stage. Western blotting experiments showed that the protein expression levels of glutathione peroxidase 4 (GPX4), p62, and nuclear factor-erythroid factor 2-related factor 2 (NRF2) genes gradually decreased with the increase of the drug concentrations used (P < 0.001; Figure 10C ). These findings show that the RXLD extracts can regulate the oxidative stress of A172 cells through the p62/Nrf2 pathway and can induce ferroptosis in GBM cells.