The human GBM cell lines U251 and A172 were purchased from the Chinese Academy of Sciences Cell Bank (Shanghai, China). Normal human astrocyte (NHA) cells were kindly provided by Prof. Rolf Bjerkvig at the Department of Biomedicine, University of Bergen, Norway. Complete medium was used to maintain the GBM cell lines and NHA cells: Dulbecco’s modified Eagle’s medium (DMEM; Thermo Fisher Scientific; Waltham, MA, United States) supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific), streptomycin (100 g/ml), and penicillin (100 U/mL). Prof. Rolf Bjerkvig of the Department of Biomedicine at the University of Bergen, Norway kindly provided the patient-derived GBM stem-like cell (GSC) line P3#GBM. P3#GBM was obtained from Haukeland University Hospital (Bergen, Norway) following approval of the local ethics committee (approval number 2009/117) (Golebiewska et al., 2020). In addition, P3#GBM primary cells have been cultured in Prof. Rolf Bjerkvig’s laboratory since 2009, and several associated articles have been published (Wang et al., 2009; Han et al., 2020). P3#GBM cells were cultured in serum-free Neurobasal™ medium (Gibco/Thermo Fisher Scientific) supplemented with 2% B-27 Neuro Mix (Thermo Fisher Scientific), 20 ng/ml epidermal growth factor (EGF; Thermo Fisher Scientific), 10 ng/ml basic fibroblast growth factor (bFGF; PeproTech; Rocky Hill, NJ, United States), and streptomycin (100 g/ml), and the cells were incubated at 37°C in a humidified atmosphere composed of 95% air and 5% CO₂.

RB (purity ≥99.69%, CAS# 465-39-4), purchased from MedChemExpress (MCE; Monmouth, NJ, United States), was dissolved in dimethyl sulfoxide (DMSO; Sigma–Aldrich; St. Louis, MO, United States) and stored at −20°C as a stock solution (10 and 50 mM). Ruthenium red (RR; purity ≥95.0%, CAS# 11,103-72-3) purchased from MCE was dissolved in DMSO and stored at −20°C as a 10 mM stock solution. U0126 (CAS# 109,511-58-2) purchased from Beyotime (Shanghai, China) was dissolved in DMSO as a 10 mM stock solution and stored at −20°C. The stock solutions were diluted directly in complete medium, and the final DMSO concentration was <0.1%.

The following antibodies were used: MMP-2 (#87809S, CST; Boston, MA, United States), Vimentin (#5741S, CST), N-cadherin (#13116S, CST), Snail (#3879S, CST), phospho-ERK (p-ERK,Thr202/Tyr204, #4370S, CST), ERK (#4695S, CST), phosphor-JNK (p-JNK,Thr183/Tyr185, #4668S, CST), JNK (#9552S, CST), phospho-p38 (p-p38, Thr180/Tyr182, #4511S, CST), p38 (#8690S, CST), GAPDH (#5174S, CST), p21 (#2947S, CST), CDC25C (#4688S, CST), phospho-CDK1 (p-CDK1, Tyr15, #9111S, CST), CDK1 (#28439S, CST), Cyclin B1 (#12231S, CST), Src (#2109S, CST), phosphor-Src (p-Src, Ser17, #12432SS, CST), ATP1A1 (#23565S, CST) Cyclin B2 (#ab185622, Abcam; Cambridge, MA, United States), FAK (#ab40794, Abcam), phospho-FAK (p-FAK, Y397, #ab81298, Abcam), Paxillin (#ab32084, Abcam), and phospho-Paxillin (p-Paxillin, Y118, #ab109547, Abcam). Secondary antibodies against goat anti-rabbit/mouse were purchased from Zhongshan Golden Bridge Biotechnology (Zhongshan Golden Bridge; Beijing, China). The luminance was determined using a chemiluminescence imager (Bio–Rad ChemiDoc XRS+; Hercules, CA, United States) in accordance with the manufacturer’s protocol.

The Cell Counting Kit-8 assay (CCK-8; Dojindo, Kumamoto, Japan) was performed to determine cell viability. To obtain an accurate IC₅₀ of each cell, we experimented with different concentration gradients. For P3#GBM cells, we used a concentration gradient of RB at 0, 0.5, 1, 1.5, 2, 2.5, 3 and 3.5 μM for 48 h. For the GBM cell lines U251 and A172, we used RB concentrations of 0, 1.5, 3, 4.5, 6, 7.5 and 9 μM for 48 h. For NHA cells, we used a concentration gradient of RB at 0, 5, 15, 20, 25, 30, 35, 40 and 45 μM for 48 h. All final DMSO concentrations were <0.1%. P3#GBM, U251 and A172 cells (4 × 10³ cells/well) were inoculated in 96-well plates and incubated at 37°C. After 24 h, the original culture medium was replaced with 100 μL of culture medium containing various doses of RB or vehicle control (diluted DMSO). After 24, 48 and 72 h of treatment, P3#GBM, U251 and A172 cells were incubated in 10 μL of CCK-8 reagent with 90 μL of serum-free DMEM for 1 h. An EnSight multimode plate scanner was used to detect the absorbance at 450 nm (PerkinElmer, Hopkinton, MA, United States).

U251 and A172 cells (800 cells/well) were inoculated in 6-well plates. After cell attachment, U251 and A172 cells were treated with DMSO or RB (2 or 4 μM), and then fresh medium was replaced once every 3 days for the duration of the expression. After incubation for 14 days, 4% paraformaldehyde was used to fix U251 and A172 colonies for 5 min, and then 0.5% crystal violet was used to stain U251 and A172 colonies for 15 min and washed three times with PBS. A bright-field microscope was used to count colonies with more than 50 cells (Leica; Solms, Germany).

5-ethynyl-2′-deoxyuridine (EdU) binding assays were used to detect the binding of EdU, a thymidine analog, to proliferating cells via a catalytic reaction between EdU and Apollo fluorescent dye (RiboBio; Guangzhou, China). The nuclei of the cells were counterstained with 4′,6-diamidino-2-phenylindole (DAPI). The number of EdU-positive cells in each well was determined using a fluorescence microscope (Leica, Solms, Germany).

Western blotting was performed as previously described (Yao et al., 2020). P3#GBM, U251 and A172 cells were treated with RB for 48 h and then lysed in RIPA lysis buffer (Beyotime) with protease and phosphatase inhibitors (Beyotime) for 30 min and sonicated to enhance lysis. The luminescence intensity was detected with a chemiluminescence imager (Bio–Rad ChemiDoc XRS+; Hercules, CA, United States).

P3#GBM, U251 and A172 (4 × 10⁵) were inoculated in 6-well plates. After treating cells with (diluted DMSO) or RB (2 and 4 μM) for 48 h, the cells were digested, centrifuged, and incubated overnight at 4°C in cold 70% ethanol. Samples were centrifuged, stained with propidium iodide (BD Biosciences; San Jose, CA, United States) for 10-15 min protected from light and subjected to cell cycle analysis by flow cytometry (Accuri C6, BD Biosciences). ModFit software (Becton Dickinson; San Diego, CA, United States) was used to identify P3#GBM, U251 and A172 cell cycle distributions.

To determine how RB affected GBM cell invasion, we carried out three invasion assays: transwell invasion assays, 3D spheroid invasion assays and GBM-brain organoid coculture invasion assays.

The transwell invasion assay was performed using 24-well Matrigel invasion chambers (Costar, Cambridge, MA, United States). Briefly, Matrigel was mixed with serum-free cell culture medium at 4°C in a 1:8 dilution ratio. One hundred microliters were applied uniformly to the surface of the polycarbonate membrane in the upper chamber and left at 37°C for 1 h to polymerize into a gel. The cells were suspended in inserts (2 × 10⁴ cells/200 μL). DMEM containing 30% fetal bovine serum was added to the lower chamber (600 μL). U251 and A172 cells were treated with vehicle control (diluted DMSO) or RB (2 or 4 μM) for 48 h. Fixation, staining, digital imaging and counting of cells under invading membranes.

In the 3D spheroid invasion assay, P3#GBM, U251 or A172 cells (3,000 cells/well) were inoculated in a U-shaped 96-well plate (Trevigen; Gaithersburg, MD, United States) to develop tumor spheroids for 3 days, and then tumor spheroids were embedded in invasive matrix (Trevigen; Gaithersburg, MD, United States). Medium containing vector control (diluted DMSO) or RB (2 or 4 μM) was replaced with fresh complete medium, and 0-h spheroids were pictured as a reference point to measure the invasive area. The relative invasion area of tumor spheroids was measured on 6^th day.

Our previous work described the protocol for culturing 18-days rat fetal brain organoids for the GBM brain organoid coculture invasion ex vivo system (Bjerkvig, Laerum and Mella 1986). After 3 weeks in culture, the differentiation of cells in normal brain organoids was complete, and brain organoids could be exposed to tumor organoids. U251-GFP and A172-GFP cells (n = 3,000), established using GFP lentiviruses, were plated in 96-well plates for 3 days to generate tumor spheroids and then cocultured with mature brain organoids for 24 h. GBM organoids were treated with diluted DMSO or 4 μM RB for 0, 48 and 96 h. Confocal microscopy captured coculture images of tumor cell invasion (Leica DMi8; Solms, Germany). ImageJ (National Institutes of Health; Bethesda, United States) software was used to analyze the GBM cell-related invasive area.

Na⁺-K⁺-ATPase activities were determined by a Na⁺-K⁺-ATPase kit (A070-2-2; Nanjing Jiancheng Bioengineering Institute; Nanjing, China). The sample and mixed reagents were added according to the manufacturer’s instructions and reacted precisely at 37°C for 10 min. Then, R4 was added and centrifuged for 10 min at 3,500 rpm, and the supernatant was collected. Then, the color developer was added to the supernatant at room temperature for 2 min, and R6 was added for 5 min at room temperature. An EnSight multimode plate scanner was used to detect the absorbance at 636 nm and 1 cm optical diameter (PerkinElmer, Hopkinton, MA, United States).

A total of 2–4 ×10⁵ GBM cells were inoculated in each 6-well plate, and the cells were treated with diluted DMSO or 4 μM RB for 48 h. After 48 h of treatment, the cells were washed three times with PBS and incubated with 10 μM 2′,7′-dichlorofluorescein diacetate (DCFH-DA, Sigma–Aldrich) for 30 min. After incubation, DCFH-DA was replaced with PBS. The fluorescence intensity of DCFH-DA was detected by flow cytometry (BD C6, San Jose, CA, United States). FlowJo software was used to analyze flow cytometry data (Treestar; Ashland, OR, United States), and at least 1 × 10⁵ cells were analyzed.

RNA sequencing (LC-Bio; Shanghai, China) was performed on RB-treated P3#GBM cells (4 μM RB) and U251 cells (4 μM RB) and their parent cell lines. Full data accompanying this experiment are available in the Sequence Read Archive (SRA). The accession number is PRJNA796785.

Fluo-4/AM (MCE) was added to the cells after they had been treated with diluted DMSO or RB (2 or 4 μM) for 48 h. The solution was kept in the dark at room temperature for 30 min. Heated Hank’s balanced salt solution was used to wash U251 and A172 cells. Intracellular Ca²⁺ levels in GBM cells were detected by measuring the fluorescence intensity of the calcium indicator Fluo-4/AM with fluorescence microscopy and flow cytometry. FlowJo software was used to analyze flow cytometry data.

Short interfering RNA sequences (siRNAs) silencing ATP1A1 (GenePharma Gene; Shanghai, China) were transfected into cells with Lipofectamine 3,000 reagent (Thermo Fisher Scientific). The silencing efficiency in U251 and A172 cells was assessed by immunoblotting 48 h after transfection. The siRNA sequence was as follows: siATP1A1#3 5′-GGA​GGC​UUC​UUU​ACU​UAC​UTT-3′.

We constructed P3#GBM cells infected with lentivirus expressing fluorescein-GFP for animal studies. A total of 4 × 10⁵ fluorescein-GFP P3#GBM cells in a 10 μl cell suspension were stereotaxically inoculated into 4-week-old BALB/c nude male mouse brains (1 mm posterior to bregma, 2 mm right of midline suture, depth 1.5 mm) (Han et al., 2019; Johannessen et al., 2019). Male mice were chosen because of the uneven tumor formation in the animal model due to individual differences in sex hormones in female mice. On the 7^th day, after determining tumor size, the mice were randomly divided into two groups: the NC (n = 10) and RB groups (n = 10). Mice were administered daily intraperitoneal injections of NC (DMSO) or RB (10 mg/kg/day). Bioluminescence imaging was used to evaluate the size of the tumor every week (IVIS Spectrum, Hopkinton, MA, United States). During imaging, mice were intraperitoneally administered D-fluorescein and potassium salt D (150 mg/kg; Yeasen Biotech; Ltd., Shanghai, China) under isoflurane gas anesthesia. The survival of each mouse was recorded after death. Kaplan–Meier analysis was performed on the median survival of the tumor-bearing mice, and log-rank analysis was used to measure the statistical significance of the differences. Finally, the tumors were cut up and frozen in liquid nitrogen or placed in formalin for immunohistochemistry.

Tumors were separated from the sacrificed mice, fixed in 4% paraformaldehyde and embedded in paraffin. The paraffin-embedded samples were sectioned (4 μm) and secured on glass slides. The sections were heated in a microwave oven at pH 7.2 in 10 mM citric acid buffer for epitope retrieval. The slides were then incubated with primary antibodies (rabbit anti-Ki-67 1:200 dilution; rabbit anti-ATP1A1 1:100 dilution) overnight at 4°C, followed by incubation with HRP-conjugated secondary antibodies for 1 h at room temperature. Antibodies were detected using the substrate diaminobenzidine (Beyotime), and slides were counterstained with hematoxylin (Beyotime).

Two pathologists independently evaluated each immunostained section under light microscopy. Immunoreactive scores were calculated by multiplying the intensity of staining (from 0 to 4; 0 = negative, 1 = weakly positive; light yellow, 2 = moderately positive; yellowish-brown, 3 = strongly positive; brown) by the percentages of stained-positive cells (from 0 to 100).

At least three biological replicates of each experiment were performed. Data are provided as the mean ± SEM. Statistical analysis was performed using GraphPad Prism 8. The following p values were considered significant: ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

The chemical structure of RB is shown in Figure 1A. The half-maximal inhibitory concentration (IC₅₀) values were 2.29, 3.05 and 6.21 µM in the primary GBM cell line P3#GBM and in two GBM cell lines, U251 and A172. However, the IC₅₀ value was 32.66 µM in NHA cells. NHA cells showed proliferation inhibition only at higher concentrations compared to GBM cells, suggesting that RB may be more sensitive to inhibiting GBM cell proliferation at certain concentrations. To determine the cytotoxic effects and inhibitory activity of RB in GBM cells, we treated P3#GBM, U251 and A172 cells with RB at 2 or 4 µM concentrations for 24, 48 and 72 h in vitro. CCK-8 assays showed that RB significantly inhibited the proliferation of GBM cells in a concentration-dependent manner (Figure 1B and Supplementary Figure S1A).

To determine whether RB inhibited the proliferation of GBM cells, we performed a colony formation assay. The results showed that RB at 2 μM significantly reduced the number of U251 and A172 cell colonies, and no colonies appeared under 4 μM RB treatment (Figures 1C,D). These results were also confirmed in EdU assays, where the percentages of EdU-positive cells were significantly reduced in the U251 and A172 cells treated with 2 and 4 μM RB (Figures 1E,F), indicating that RB significantly inhibited the proliferation of human GBM cells.

To further determine whether proliferation inhibition induced by RB was associated with cell cycle arrest, we examined the effect of RB on the cell cycle of P3#GBM, U251 and A172 cells. Flow cytometry data indicated that RB significantly inhibited cell cycle progression in P3#GBM, U251 and A172 cells. A significant increase in the G2/M phase fraction was coupled with a decrease in the G0/G1 phase fraction (Figure 1G and Supplementary Figure S1B). Taken together, these results revealed that RB inhibited cell proliferation and induced G2/M-phase cell cycle arrest in GBM cells.

To investigate whether RB affects cell invasion, we performed transwell assays and 3D spheroid-based invasion assays. The results showed that in both transwell (Figures 2A,B) and 3D spheroid-based invasion assays (Figures 2C,D), treatment with RB significantly reduced GBM cell invasion by ∼2–4-fold compared with that in the control cells. Next, to further demonstrate the inhibitory effect of RB on GBM cell invasion, we conducted coculture experiments with GBM spheroid and brain organoid coculture assays. GBM cells were cultured for 3 days to develop tumor spheroids, which were then cocultured with mature rat brain organoids and treated with diluted DMSO or RB (4 μM). As shown in Figures 2E,G, after 48 and 96 h of treatment, the invasive area by GBM spheroids was decreased significantly in the 4 μM RB-treated U251 spheroids compared to the control U251 spheroids. We found the same results in 4 μM RB-treated A172 spheroids, where RB also reduced the area invaded by tumor cells after 48 and 96 h (Figures 2F,G).

In addition, we examined the effect of RB on the protein expression levels of several mesenchymal markers in GBM cells. We found that RB inhibited GBM cell invasion by decreasing the expression of Vimentin, N-cadherin, Snail and MMP-2 (Figure 2H). These data suggested that RB suppressed the invasion of GBM cells.

As shown in Figure 1G and Supplementary Figure S1B, flow cytometry analysis indicated that G2/M phase arrest occurred in GBM cells following 2 or 4 μM RB treatment. To elucidate the mechanisms of RB-induced G2/M arrest in GBM cells, we performed whole transcriptome sequencing in vehicle control (diluted DMSO) and 4 μM RB-treated P3#GBM and U251 cells. The differentially expressed mRNAs were chosen using the cutoffs of p value <0.05 and log₂ fold change >1. Figure 3A shows that 913 and 812 mRNAs had upregulated expression in P3#GBM and U251 cells, respectively, and that the number of intersecting genes with upregulated expression between the two GBM cell lines was 509. To shed light on the biological functions of the intersecting genes, we performed KEGG analysis of the intersecting genes with upregulated expression and found that the MAPK pathway was enriched as the top pathway (Figure 3B). In light of the fact that MAPK families have been shown to play a vital role in cycle transition (Zhang and Liu 2002; Wang et al., 2007), we predicted that the activation of the MAPK pathway might be a key determinant of RB-induced G2/M phase arrest. Then, we examined which kinds of MAPKs mediate RB-induced G2/M phase arrest in GBM cells. As shown in Figure 3C, the total amount of ERK1/2, JNK and p38 MAP kinases did not change significantly. The significant activation of ERK1/2 was observed in P3#GBM, U251 and A172 cells, but not JNK and p38. Only activation of ERK was consistent with the sequencing result that MAPK signaling pathway was significantly upregulated.

Activation of the MAPK/ERK pathway has been reported to induce cancer cells to arrest in G2/M phase by regulating the protein expression levels of p21 and CDC25C (Zhang and Liu 2002; Wang et al., 2007). The levels of several G2/M-related proteins in GBM cells treated with RB were then determined using western blotting. Cyclin B1, Cyclin B2, and CDC25C levels were decreased, whereas p21 and Tyr15-phospho-CDK1 levels were increased and total CDK1 remained unchanged (Figure 3D), indicating that G2/M phase arrest induced by RB was associated with altered levels of key checkpoint-related proteins.

We hypothesized that the activation of the MAPK/ERK pathway contributed to RB-mediated G2/M phase arrest. To ascertain the function of ERK in RB-induced G2/M phase arrest, we pretreated P3#GBM and U251 cells for 6 h with the specific inhibitor of ERK, U0126 (10 μM). As shown by flow cytometry, the ratios of cells in the G2/M phase were markedly increased in cultures treated with 4 μM RB compared to the DMSO control, but RB-mediated G2/M phase arrest was significantly rescued by U0126 in P3#GBM and U251 cells (Figure 3E and Supplementary Figure S1C). Next, we examined whether an inhibitor of ERK modulates the expression of key checkpoint proteins in RB-induced G2/M phase arrest. As shown in Figure 3F, the protein expression level of p-ERK was significantly attenuated by treatment with the ERK inhibitor U0126, indicating that the MAPK/ERK pathway was engaged in RB-induced GBM cell G2/M phase arrest. U0126 restored RB-induced inhibition of Cyclin B1, Cyclin B2 and CDC25C expression and blocked Tyr15-phospho-CDK1 and p21 expression in P3#GBM and U251 cells. These results suggested that RB induced GBM cells to arrest in G2/M phase via the MAPK/ERK/CDC25C/p21/CDK1/Cyclin B pathway.

Figure 4A reveals that a total of 2018 and 1938 mRNAs had downregulated expression in P3#GBM and U251 cells, respectively, and that the number of intersecting genes with downregulated expression between the two GBM cell lines was 904. To probe the mechanism of RB-mediated inhibition of GBM cell invasion, we performed KEGG pathway enrichment of intersecting genes with downregulated expression and found that focal adhesion was enriched as the top pathway (Figure 4B).

Src, FAK, and Paxillin are the main elements of focal adhesion, a high-intensity attachment points of cells to the extracellular matrix formed by the leading edge of the invading or attached cells (Stallcup 2017; Schiffer et al., 2018). We hypothesized that RB inhibits GBM cell invasion partly by suppressing the Src/FAK/Paxillin focal adhesion pathway in GBM cells. Our results showed that RB treatment decreased the phospho-Src, FAK, and Paxillin levels, although it had no effect on focal adhesion-related total Src, FAK, and Paxillin levels (Figure 4C), suggesting that RB inhibited GBM cell invasion by suppressing the Src/FAK/Paxillin focal adhesion pathway.

A previous study reported that maintained Ca²⁺ influx can destroy the components of FAK’s focal connections in the presence of Src deficiency, markedly inhibiting cell invasion and motility (Jeong et al., 2020). We further investigated whether RB altered intracellular Ca²⁺ levels to suppress the Src/FAK/Paxillin focal adhesion pathway, thereby inhibiting the invasion of GBM cells.

To determine whether Ca²⁺ in cells was a key target for RB treatment, we first assessed intracellular Ca²⁺ levels using flow cytometry and fluorescence microscopy. As shown in Figures 4D,E and Supplementary Figures S2A,B, intracellular Ca²⁺ levels were significantly elevated in the 4 μM RB-treated cells compared with the control cells. To investigate whether Ca²⁺ elevation was a crucial factor in the RB-inhibited focal adhesion pathway, we first treated RB-treated U251 and A172 cells with RR (3 μM), an L-type calcium current blocker, and decreased Ca²⁺ levels were observed after treatment with the combination of 4 μM RB and 3 μM RR compared to RB treatment alone (Supplementary Figures S2C,D). These same results were also confirmed by flow cytometry, where 3 μM RR blocked the Ca²⁺ elevation induced by RB (Supplementary Figures S2E,F). Furthermore, transwell assays revealed that the combination of 4 μM RB and 3 μM RR significantly rescued invasion in U251 and A172 cells compared to RB treatment alone (Figure 4F and Supplementary Figures S2G), suggesting that elevated Ca²⁺ inhibited invasion in treated cells. Additionally, western blot revealed that the activity of focal adhesion-related proteins inhibited by RB could be rescued by reducing intracellular Ca²⁺ by RR (Figure 4G). Taken together, our results demonstrated that RB inhibited GBM cell invasion partly by elevating intracellular Ca²⁺ levels to suppress the Src/FAK/Paxillin focal adhesion pathway.

We found that RB-mediated G2/M phase arrest inhibited GBM cell proliferation by activating the MAPK/ERK pathway, which altered p21/CDC25C/CDK1/Cyclin B protein expression, and RB inhibited GBM invasion by increasing intercellular Ca²⁺ to suppress the Src/FAK/Paxillin focal adhesion pathway. Thus, exploring RB target proteins in GBM cells would provide insight into the mechanisms underlying the inhibition of GBM malignant progression. The traditional Chinese medicine Chansu is known to be a specific inhibitor of Na⁺-K⁺-ATPase. Moreover, previous research showed that RB could alter intracellular K⁺ and Na⁺ concentrations by affecting the function of Na⁺-K⁺-ATPase (Hao et al., 2011). To predict the most likely Na⁺-K⁺-ATPase-related target proteins of RB, we used BioSolveIT SeeSAR software, which indicated that RB may target ATP1A1 in GBM cells (Figure 5A).

First, we tested whether RB regulated Na⁺-K⁺-ATPase enzyme activity in GBM cells and found that treatment with 4 μM RB significantly enhanced Na⁺-K⁺-ATPase activity by ∼4-fold in U251 cells and by ∼6-fold in A172 cells (Figure 5B), indicating that RB significantly increased Na⁺-K⁺-ATPase activity in GBM cells. ROS and Na⁺-K⁺-ATPase develop a feed-forward cycle (Yan et al., 2013; Sodhi et al., 2015; Yan et al., 2017). Activation of Na⁺-K⁺-ATPase could block the Na⁺-K⁺-ATPase/ROS amplification loop as well as increase ATP1A1 expression (Yan et al., 2017). Whole transcriptome sequencing analysis suggested that peroxisomes were inactivated in GBM cells following treatment with RB (Figure 4B and Supplementary Figure S3A). Thus, we detected the levels of intracellular ROS in P3#GBM, U251 and A172 cells treated with RB. As expected, the levels of intracellular ROS were significantly decreased following treatment with RB (Supplementary Figures S3B,C). Furthermore, we tested ATP1A1 expression and found that RB significantly increased the expression of ATP1A1 (Figure 5C). These results showed that activation of Na⁺-K⁺-ATPase with RB upregulated ATP1A1 expression by blocking the Na⁺-K⁺-ATPase/ROS amplification loop. ATP1A1 may be the target by which RB inhibits the malignant progression of GBM.

To further clarify the target of RB in GBM cells, we knocked down ATP1A1 using siRNA in U251 and A172 cells. Our results showed that siATP1A1#3 efficiently knocked down ATP1A1 at the protein level in U251 and A172 cells (Figure 5D). In the cells transfected with ATP1A1 siRNAs, RB treatment markedly alleviated G2/M phase arrest induced by RB alone (Figure 5E and Supplementary Figure S3D). Next, after silencing ATP1A1, RB treatment markedly blocked the protein levels of p21, CDC25C, p-ERK and CDK1 compared to RB treatment alone in U251 and A172 cells (Figure 5F).

Furthermore, in cells transfected with ATP1A1 siRNAs, RB treatment substantially alleviated intracellular Ca²⁺ levels (Figure 5G and Supplementary Figure S3E) and restored the invasiveness of GBM cells (Figure 5H and Supplementary Figure S3F). Furthermore, in U251 and A172 cells transfected with ATP1A1 siRNAs, treatment with RB substantially restored the protein expression of MMP2 and the phosphorylation of FAK and Paxillin compared to RB treatment alone (Figure 5I). These results showed that RB inhibited the malignant progression of GBM by targeting the ATP1A1 signaling cascade and increasing Na⁺-K⁺-ATPase activity in GBM cells.

The efficacy of RB was evaluated using an orthotopic tumor modeling approach from P3#GBM-luciferase-expressing cells inoculated in BALB/c nude mice bregma. As shown in Figure 6A, after the 7^th day of implantation, mice were selected randomly into two groups (n = 10/group): the NC (DMSO) and RB groups (10 mg/kg/day), and tumor growth was evaluated using luciferase bioluminescence. The bioluminescence values of the two groups of mice showed a significant difference by 21 days (Figure 6B). On the 21st day after implantation, bioluminescence values were nearly 42% lower in the RB-treated group than in the DMSO-treated group (Figure 6C). Kaplan–Meier analysis was performed on the median survival of the tumor-bearing mice, and log-rank analysis was used to measure the statistical significance of the differences. As shown in Figure 6D, RB prolonged the median survival of the tumor-bearing mice from 26 to 29 days (p = 0.0082). We also investigated the expression of Ki-67 and ATP1A1 by immunohistochemical (IHC) staining. The expression of Ki-67 was markedly lower under RB treatment, suggesting that RB suppresses the proliferation of GBM in vivo (Figures 6E,F). The expression of ATP1A1 was increased after RB treatment, suggesting that RB inhibited the growth of GBM by targeting Na⁺-K⁺-ATPase.