The radix of R. uniflorum (L.) DC. samples were collected in Henan Province, China. The original specimen was deposited for future use, and the RUEA extracts were dissolved in dimethyl sulfoxide (DMSO) and diluted to obtain two different storage concentrations (50 μg/mL and 50 mg/mL). The extracts were stored at 4°C and diluted to the required concentration before use.

Chromatographic analysis was performed using an Acquity UPLC system (Waters, Milford, MA, United States) with a 2 μL injection volume. Next, Tandem MS was performed using a Q-TOF mass spectrometer (Waters) with an electrospray ionization interface. The results were analyzed using MassLynx v4.1 software (Waters).

Pharmacology-based network prediction and analysis were performed using systemsDock¹. The required processes were divided into three main steps: (i) Selection of ROS-dependent signaling proteins (in SBML format) in OSCC from literature surveys (Tang, 2009, unpublished data); (ii) Uploading structure files based on the results of UPLC-Q/TOF-MS analysis in PubChem; (iii) Obtaining the prediction results and screening out the most interesting proteins.

Human OSCC cell line SCC15 (American Type Culture Collection) was cultured in Dulbecco’s modified Eagle medium/Nutrient Mixture F-12 medium (Gibco, United States) containing 12.5% fetal bovine serum (Gibco) and cultured at 37°C incubator containing 5% CO₂.

SCC15 cells were transfected with shRNA Prx1 plasmid (Santa Cruz Biotechnology, Santa Cruz, CA, United States) using Lipofectamine 2000 (Invitrogen, Life Technologies Corp., Carlsbad, CA, United States). The plasmid was constructed according to standard techniques, and the target sequence for the Prx1 shRNA was: 5′-CGAAGCGCACCAATTGCTCA-3′. The shRNA Plasmid-A (Santa Cruz Biotechnology) was used as a vector control. The efficiency of Prx1 knockdown was determined by reverse transcription polymerase chain reaction (RT-PCR) and western blotting after selection with puromycin.

SCC15 cells were treated with RUEA extracts at the concentrations of 0 (vehicle control), 12.5, 25, 50, and 100 μg/mL for 24, 48, and 72 h. Cell viability was detected by using Cell Counting kit-8 (CCK-8; Dojindo, Tokyo, Japan). Cell viability (%) was presented as [OD] test wells/[OD] control wells × 100%.

SCC15 cells treated with 0, 12.5, 25, and 50 μg/mL RUEA extracts for 24 and 48 h were seeded into 6-well plates (5 × 10⁵ cells/well). Scratch wounds were made through the cell monolayer with a 200 μL pipette tip in each well. Wound distance was measured in three locations for each well at 0, 24, and 48 h.

Cell invasion assay was performed using 24-well Transwells (8 μm pore size; Corning, NY, United States) coated with Matrigel (100 μL per well, thickness: 3 mm; Becton Dickinson, San Jose, CA, United States) in triplicate. SCC15 cells were treated with 0, 12.5, 25, and 50 μg/mL RUEA for 24 and 48 h. Cells (10⁵ well) were resuspended in serum-free medium, plated into the upper chamber, and then incubated for 24 h with complete culture medium added to the lower chamber. Following incubation, cells inside the chamber were wiped off with a cotton swab. The invading cells which stuck to the lower side of the filter membrane were stained with Hematoxylin (Salarbio, Beijing, China) and examined using a microscope (Olympus, Tokyo, Japan).

SCC15 cells were treated with 0, 12.5, 25, 50, and 100 μg/mL RUEA extracts for 24 and 48 h. Cells were then resuspended and incubated with 5 μL of Annexin V-fluorescein isothiocyanate (FITC) for 15 min, and then incubated with 5 μL propidium iodide (PI) for 5 min in the dark. Apoptosis rates were analyzed by flow cytometry (FACSVerse, BD, San Jose, CA, United States).

Total RNA was extracted from SCC15 cells using TRIzol (Invitrogen). cDNA was synthesized using a HiFi-MMLV cDNA Kit (Cwbiotech, Beijing, China). Real-time PCR reaction was performed using UltraSYBR Mixture (Cwbiotech) according to the manufacturer’s protocol. Primers were as follows: Prx1, 5′-GGGTATTCTTCGGCAGATCA-3′ and 5′-TCCCCATGTTTGTCAGTGAA-3′; E-cadherin, 5′-TTGCTACTGGAACAGGGACA-3′ and 5′-GTATTGGGAGGAAGGTCTGC-3′; vimentin, 5′-GAAGAGAACTTTGCCGTTGA-3′ and 5′-CGAAGGTGACGAGCCATT-3′; Snail, 5′-TTACCTTCCAGCAGCCCTAC-3′ and 5′-GACAGAGTCCCAGATGAGCA-3′, and GAPDH, 5′-AGGTCGGTGTGAACGGATTTG-3′ and 5′-TGTAGACCATGTAGTTGAGGTCA-3′. PCR was performed in triplicate, and fold enrichment was calculated with the – ΔΔCt method relative to the expression of GAPDH.

Proteins from SCC15 cells and xenograft tumor tissues were extracted using immunoprecipitation assay buffer. The concentration of total protein was determined using the Lowry method. Equal amounts of protein were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis on 12% gels and transferred to nitrocellulose membranes. The membranes were then blocked and incubated with primary antibodies against Prx1 (1:1000; Abcam, Cambridge, MA, United States), Snail (1:500; Abcam), E-cadherin (1:1000; Cell Signaling Technology, Beverly, MA, United States), vimentin (Bioss, Beijing, China), and GAPDH (1:2000; Sigma–Aldrich, United States). The protein bands were detected with horseradish peroxidase (HRP)-conjugated secondary antibodies and visualized using an enhanced chemiluminescence detection system.

All procedures performed in studies involving animals were carried out in accordance with the ethical standards of the Ethics Committee of Capital Medical University School of Stomatology (Approval No. KQYY-201604-010). Female BALB/c-NU mice (4–5 weeks old, 16–20 g) were purchased from Beijing Vital River Laboratory Animal Technology Co. (Beijing, China). SCC15 cells (5 × 10⁶/100 μL/mouse) were subcutaneously injected into the backs of mice, and xenografts were successfully formed after 1 week. All mice with xenografts were randomly separated into four groups: blank control group, vehicle control group, 25 mg/kg RUEA group and 250 mg/kg RUEA group. Mice in the blank control group were untreated, whereas mice in other groups were treated with 100% DMSO (5 mL/kg), 25 mg/kg RUEA extracts and 250 mg/kg RUEA extracts, respectively. Treatments were given via intratumoral injection every 4 days in the morning. During this period, tumor length, tumor width and body weight were recorded every 4 days. The tumor volume (TV) was calculated using the formula: TV = π/6 × length × (width)². At the end of the study (5 weeks after inoculation), all mice were sacrificed by cervical dislocation. Then, all tumors were removed and each single tumor was cut in two parts; one part was fixed with 10% neutral formalin for immunohistochemical staining, and the other was stored in liquid nitrogen for western blot analysis.

The tumors were collected, fixed with 10% neutral formalin for 24 h, embedded in paraffin, serially sectioned at 4 μm and processed for immunohistochemical staining. After antigen retrieval conducted with citrate buffer (pH = 6.0), the sections were treated with protein block solution (Fuzhou Maixin Biotech, China) for 15 min at 37°C and incubated with anti-Ki67 antibody (Fuzhou Maixin Biotech, China) overnight at 4°C. Positive cells were detected with HRP-conjugated secondary antibody and visualized using 3,3′-diaminobenzidine (DAB) staining. Hematoxylin was used as the counterstain. The proliferation rate of each tumor was evaluated by estimating the average percentage of Ki67 positive cells among total epithelial cells in five random areas of a section under 200× magnification.

The apoptotic cells in tumor sections were detected using TUNEL assay. Sections were dewaxed, hydrated, and incubated with proteinase K at 37°C for 15 min, and washed with phosphate-buffered saline three times. Sections were then incubated with 50 μL of TUNEL reaction mixture and 50 μL of converter-POD at 37°C for 60 and 30 min, respectively. Freshly prepared DAB solution was incubated with sections for 20 min. The apoptotic cells were photographed and counted by Image Pro.

Experiments were conducted in triplicates. Differences with P-values of less than 0.05, analyzed by one-way analysis of variance (SPSS v17.0), were considered statistically significant.

Figure 1 shows the base peak ion (BPI) chromatogram of RUEA extracts. A total of 14 compounds were tentatively identified based on retention time, molecular ions, major fragment ions, and previously published articles and online databases. The identified compounds were mainly classified as phytoecdysteroids and triterpenoids. The details of the identified compounds are listed in Table 1.

SystemsDock is a web server for network pharmacology-based prediction and analysis. It has an elaborately designed scoring function for molecular docking to evaluate protein–ligand binding potential. Figure 2 shows docking scores and predicted binding affinities for target proteins. Compounds of RUEA extracts searched on SystemsDock are listed in Table 2. Among the interesting ROS-related proteins, Prx1 (PDB: 4XCS; score avg.: 7.07) was identified as a potential binder.

Compared with the control group, SCC15 cells treated with 50 μg/mL RUEA extracts grew slowly and turned from polygonal to round (Figure 3A). To determine the effects of RUEA extracts on cell proliferation and apoptosis in SCC15 cells, cells were treated with 0, 12.5, 25, 50, and 100 μg/mL of RUEA extracts for 24, 48, and 72 h. Cell viability was measured by CCK8 assay and cell apoptosis rates were analyzed by flow cytometry. As shown in Figure 3B, cell viability was significantly decreased after RUEA extract treatment in a concentration-dependent manner. Annexin V-FITC/PI double-staining was used for the detection of different phases of apoptotic cells. The results showed that the proportions of the early and terminal phase of apoptotic cells increased after RUEA extract treatment in a concentration-dependent manner at 24 and 48 h, respectively (Figure 3C).

SCC15 cells were treated with 0, 12.5, 25, and 50 μg/mL RUEA extracts for 24 and 48 h, wound healing and Matrigel invasion assays were performed. Wound healing assays showed that the open wound area decreased significantly in RUEA extract treated cells in a concentration-dependent manner compared with that in control cells (Figure 4A). Similarly, RUEA extracts significantly decreased the number of invasive cells after 24 and 48 h (Figure 4B).

The EMT process accelerates cancer metastasis by increasing cell migration and invasion. To determine whether the EMT process was modulated by RUEA extracts, we evaluted the expression of E-cadherin, vimentin, and Snail in SCC15 cells treated with 50 μg/mL RUEA extracts for 48 h. We also evaluated the expression of Prx1, which was identified as a potential binder of RUEA extracts. Our data showed that at the mRNA level, RUEA extracts significantly increased the expression of E-cadherin and decreased the expression of Prx1 and Snail compared with the control group (Figure 4C). At the protein level, RUEA extracts significantly increased the expression of E-cadherin and decreased the expression of Prx1, vimentin, and Snail compared with the control group (Figure 4D).

To further explore whether RUEA extracts suppressed cell migration and invasion via Prx1 in oral cancer, Prx1 knockdown SCC15 cells were established. RUEA extracts suppressed cell invasion and migration in Prx1 knockdown SCC15 cells treated with 0, 12.5, 25, and 50 μg/mL RUEA extracts for 24 and 48 h. Moreover, RUEA extracts inhibited the expression of E-cadherin, vimentin, and Snail in Prx1 knockdown SCC15 cells for 48 h (Supplementary Figure S1), indicating that RUEA extracts may target other molecules in addition to Prx1 to inhibit cell migration and invasion in oral cancer.

To further confirm the anti-tumor effects of RUEA extracts in vivo, we established an OSCC xenograft model. Compared with the control group, mice in 25 and 250 mg/kg of RUEA extracts groups showed significantly decreased tumor weights (Figure 5A) and TVs (Figure 5B). However, there were no significant difference in tumor weights or TVs between the 25 and 250 mg/kg groups. Notably, the body weights of all mice remained stable, and there were no significant differences among groups (Figure 5C).

Hematoxylin and eosin staining revealed that necrotic tissues, abnormal mitoses, and shrinking nuclei were increased after RUEA extract treatment (Figure 6A).

As shown in Figures 6A,B, RUEA extract treatment significantly suppressed cell proliferation. The positive rates of Ki-67 were 49.6, 22.3, and 43.2% in the vehicle control group, 25 and 250 mg/kg RUEA groups, respectively. Moreover, the proliferation rate was higher in the 250 mg/kg group than in the 25 mg/kg group. No differences were found between the vehicle control group and the blank control group.

Next, we performed TUNEL assay to observe the apoptotic cells in the tumors. As shown in Figures 6A,C, RUEA extract treatment induced cell apoptosis. The average cell apoptosis rate in the vehicle control group was 27.4% which was much higher than in the blank control group. Moreover, apoptosis rates reached as high as 36.9 and 40.2% in the 25 and 250 mg/kg groups, respectively. Thus, our results suggested that RUEA reduced cell proliferation and induced apoptosis in OSCC.

Western blot results showed that compared with the vehicle control group, RUEA extract treatment obviously decreased Prx1 expression in the transplanted tumors. As shown in Figure 7, compared with the vehicle control group, RUEA extracts increased E-cadherin expression and decreased vimentin and Snail expression in the transplanted tumors. No significant differences were observed between the two control groups. Our results suggested RUEA extracts inhibited Prx1 expression and the EMT process in vivo.