The human RCC cell lines 786-O and A498 were purchased from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences. Cells were grown in RPMI 1640 medium (Sigma-Aldrich; Merck KGaA) supplemented with 10% fetal bovine serum (HyClone; GE Healthcare Life Sciences) and 100 µg/ml penicillin/streptomycin at 37˚C in a humidified incubator with 5% CO₂. GA (C17:1; C₂₄H₃₈O₃; Tauto Biotech Co., Ltd.; Fig. 1A) was dissolved in DMSO (Sigma-Aldrich; Merck KGaA) to make a 20 mM stock solution. For cell treatment, 10 and 20 µM GA (diluted in RPMI 1640 medium) were used.

786-O and A498 cells were seeded at a density of 1x10³ cells per well in 96-well plates. After culture for 12 h at 37˚C, the medium was replaced with maintenance medium containing the desired concentrations (10 and 20 µM) of GA or DMSO. Cell viability was assessed after 72 h incubation at 37˚C using an MTT assay kit (Nanjing KeyGen Biotech Co., Ltd.) according to the manufacturer's protocols. DMSO was used to dissolve formazan crystals and absorbance was measured in each well at 450 nm using a 318C microplate reader (Shanghai Peiou Analytical Instruments Co., Ltd.).

To determine c ell invasion, invasion assays were performed using 24-well Matrigel-coated invasion chambers (8-µm pore size; Corning Inc.). RCC cells were pre-treated with GA for 24 h, seeded into the upper chamber of the Transwell insert and incubated for 24 h at 37˚C. Subsequently, cells on the upper side of the membrane were wiped off and invading cells were fixed in 100% cold methanol at 4˚C for 20 min and stained with 0.2% crystal violet at room temperature for 30 min (Sigma-Aldrich; Merck KGaA). The stained cells were quantified under a light microscope, from five randomly selected fields of view (magnification, x400).

RCC cells were pre-treated with 10 and 20 µM GA for 24 h at 37˚C and allowed to grow to 90-95% confluence without serum. The cell monolayers were scratched with a sterile 200-µl pipette tip. Subsequently, the cells were washed with RPMI 1640 medium to remove cell debris and fresh cell culture medium (RPMI 1640 medium supplemented with 10% FBS) was added. Images of the scratches were obtained at 0 and 24 h using a light microscope (magnification, x200). Migratory ability was measured as the percentage of wound closure with the initial wound width defined as 1.

Cell cycle distribution and apoptosis were analyzed using the Annexin V-FITC Apoptosis kit (Solarbio Life Sciences) according to the manufacturer's protocols. Briefly, RCC cells were cultured in RPMI 1640 medium in 6-well plates at a density of 1x10⁶ cells for 12 h at 37˚C. Then, the medium was replaced with RPMI 1640 medium containing 20 µM GA. Cells were stained with Annexin V-FITC and propidium iodide buffer at room temperature for 15 min followed by the addition of Annexin V binding buffer. Cell cycle and apoptosis assays were performed on a BD Accuri C6 (BD Biosciences) flow cytometer and BD Accuri C6 software Version 1.0 (Becton, Dickinson and Company) was used to analyze the data.

Animal experiments were approved by The Animal Care and Use Committee of the Second Military Medical University (approval no. NMU20181032). Animals had free access to autoclaved chow diet and water in a pathogen-free state, at 22±1˚C with 45±10% humidity and 12-h light/dark cycles. The health and behaviour of the mice were monitored every 12 h. Moribund mice (exhibiting anorexia, immobility or an unkempt coat) would be euthanized by CO₂ at the rate of 3.5 l/min displacing 20% of the cage volume per minute. Death was verified by monitoring cessation of breathing. To detect the effect of GA on tumorigenicity in nude mice, 6-8-week-old male BALB/c mice (n=10; weight, 20-30 g; Second Military Medical University) were injected subcutaneously with 786-O or A498 cells (1x10⁷ dissolved in PBS per site; one tumor burden per mouse) on the right flank. Mice were randomly divided into two groups (n=5 per group) that either received an intragastric administration of GA (10 mg/kg) or 0.9% saline twice per week (13). At 6 weeks after GA injection, mice were euthanized by decapitation. The tumor volumes and weight were measured. Prior to 6 weeks after GA injection, no mice were euthanized or found dead.

The Online Mendelian Inheritance in Man (OMIM) Database (version 1.0; www.ncbi.nlm.nih.gov/omim), from which the original data used in the present study was obtained, is a central-level database covering relevant information and literature on human genetic diseases and gene loci. Data were input into the Search Tool for the Retrieval of Interacting Genes/Proteins (version 11.0; string-db.org) database for further network construction. The obtained PPI networks were then visualized using Cytoscape software (version 3.6.1; https://cytoscape.org). Topology analysis (CentiScaPe plug-in; version 2.2; http://apps.cytoscape.org/download/stats/centiscape/) was used to identify key targets in the network. Degrees of betweenness were assessed using the key hubs of centrality, namely betweenness (the number between nodes) and closeness (the distance between nodes).

The docking between GA and key targets of RCC was implemented using Sybyl X software (version 2.1.1; https://en.freedownloadmanager.org/Windows-PC/SYBYL-X.html). The conformation of GA was downloaded from the Traditional Chinese Medicine Systems Pharmacology Database (version 1.0; https://tcmspw.com/tcmsp.php). The mol2 format file was further transformed and extracted by Openbabel software (release date, 2017-02-22; https://sourceforge.net/projects/openbabel/). The X-ray crystal structures of key targets were obtained from the RCSB Protein Data Bank (version 1.0; www.rcsb.org). Prior to docking, targets were structurally optimized by adding hydrogen atoms and removing water molecules, and the representative protein activity docking pockets were subsequently generated. Molecular docking was implemented based on the Docking Suite module (release date, 2013-04-08; https://sourceforge.net/projects/wpfdocking/) and hydrogen bond stability and domain structure similarity were used to evaluate the reliability of results. The criteria of binding scoring: 0-3, Poor; 4-6, moderate; and 7-10, good.

Cells were pre-treated with 10 ng/ml epidermal growth factor (EGF; PeproTech, Inc.) for 12 h at 37˚C. Immunohistochemical assays of xenografts were conducted using Ki-67 (cat. no. ab15580; 1:50; Abcam) and EGF receptor (EGFR; cat. no. ab52894; 1:50; Abcam) monoclonal antibodies, as previously described (20). The expression level was analyzed from images obtained using a light microscope (magnification, x200) from five random fields of view. Immunoblot assays were performed as previously described (20). Primary antibodies included those targeted against phosphorylated (p)-EGFR (cat. no. 4370; 1:1,000; Cell Signaling Technology, Inc.), EGFR (cat. no. 2085; 1:1,000; Cell Signaling Technology, Inc.), p-Akt (Ser473; cat. no. 4060; 1:1,000; Cell Signaling Technology, Inc.), Akt (cat. no. 4691; 1:1,000; Cell Signaling Technology, Inc.), p-ERK (Thr202/Tyr204; cat. no. 4370; 1:1,000; Cell Signaling Technology, Inc.) and ERK (cat. no. 4370; 1:1,000; Cell Signaling Technology, Inc.). Horseradish peroxidase-conjugated secondary antibody (1:2,000; cat. no. ab6112; Abcam) was used. β-actin antibody (cat. no. ab8226; 1:5,000; Abcam) was used as a control. Relative protein expression and staining intensity was calculated based on the densitometric analysis by Image Lab software Version 3.0 (Bio-Rad Laboratories, Inc.).

All data are presented as the mean ± SD. SPSS software (version 18.0; SPSS Inc.) was used for statistical analysis. The statistical significance between two groups was analyzed using an unpaired Student's t-test. One-way ANOVA followed by a Dunnett's post hoc test was performed to compare data among ≥3 groups. P<0.05 was considered to indicate a statistically significant difference.

To investigate the role of GA in RCC, human RCC cells (786-O and A498) were incubated with different concentrations of GA. An MTT assay was performed 72 h after treatment and a dose-dependent decrease in cell viability was observed (Fig. 1B). In addition, GA increased the proportion of cells in the G₀/G₁ phase and decreased the proportion of cells in the S phase (Fig. 1C). However, GA treatment did not significantly affect RCC cell apoptosis (Fig. S1). Based on in vitro data, xenograft implantation in BALB/c nude mice was performed to clarify the effect of GA on RCC in vivo. RCC cell suspensions (1x10⁷) were subcutaneously injected into nude mice. The tumor volumes and weights of the group treated with GA were significantly lower compared with those of the group treated with 0.9% saline (P<0.05; Figs. 1D and S2A). In the present study, the maximum diameter of a single tumor was 1.8 cm. However, no significant difference in body weight loss was observed between the two groups (Fig. S2C). Additionally, immunohistochemical staining revealed that GA significantly reduced Ki-67 expression in the xenograft (P<0.05; Fig. 1E).

The effect of GA on the invasion and migration of RCC cells was evaluated using Transwell invasion and wound healing assays, respectively. Transwell invasion assays showed that GA reduced the number of invading RCC cells in a dose-dependent manner (Fig. 2A). Consistent with this result, wound healing assays also showed that GA inhibited RCC cell migration in a dose-dependent manner (Fig. 2B).

The interaction network was constructed based on OMIM original data and is depicted in Fig. 3, and included 549 nodes and 11,836 edges. Larger areas and deeper colours represent higher node connectivity, which likely indicates greater importance of the node in the network. The results displayed the scale-free property of the network, meaning the number of nodes decreases as the number of edges in the network increases. Betweenness degree was used to distinguish these highly connected key nodes from other nodes in the network (Fig. 4). Finally, 15 key targets were selected: Tumor protein p53, 284 points; albumin, 267 points; AKT1, 251 points; vascular endothelial growth factor A, 243 points; insulin, 230 points; interleukin 6, 223 points; JUN, 215 points; tumor necrosis factor, 213 points; MYC, 211 points; phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit-α, 208 points; EGFR, 198 points; mitogen-activated protein kinase 1, 186 points; transforming growth factor β1, 185 points; BCL2, 179 points; and STAT3, 173 points.

Molecular docking, a tool for studying the interaction strength between molecules, was used to explore the accuracy and effectiveness of the PPI network. To assess the reliability of the aforementioned drug-target interactions and to further investigate binding mode and affinity, molecular docking models were constructed for GA and the 15 key targets. Analysis of the binding data indicated stable conformational binding and good scoring (>7 points) of EGFR to GA (9.4558 points; Fig. 5A). Immunohistochemical staining displayed lower EGFR expression in the xenograft after GA treatment compared to before treatment (Fig. 5B). The immunoblotting assay demonstrated that EGFR, p-Akt and p-Erk (downstream of the EGFR signaling pathway) were significantly downregulated after GA treatment in xenografts, as well as in vitro (P<0.05; Fig. 5C).