Juglone induces ferroptosis in glioblastoma cells by inhibiting the Nrf2-GPX4 axis through the phosphorylation of p38MAPK

LN229 and T98G cell lines were obtained from American Type Culture Collection (ATCC) and maintained in DMEM culture medium (BasalMedia, China) containing 5% and 10% fetal bovine serum and 1% penicillin/streptomycin, respectively. The culture temperature was 37 °C, and the CO2 concentration was 5%. The cells came from our laboratory stocks and, after resuscitation, these cells were used for no more than 10 generations.

Juglone, Chloroquine (CQ-1), Ferrostatain-1 (Fer-1), Liproxstatin-1 (Lip-1), Mdivi-1 (Mdivi), Necrostatin-1 (Nec-1) and Z-VAD-FMK (Z-VAD) were purchased from Selleck. 2ʹ,7ʹ-Dichlorodihydrofluorescein diacetate (DCFDA) were purchased from MCE. Antibodies against ACSL4, PTGS2, GPX4, Ki67, 4HNE were purchased from Abcam. Antibodies against XCT, FTH1, p38, pp38, JNK, p-JNK, GSK3α/β, p-GSK3α/β, MMp2 and MMp9 were purchased from Cell Signal Technology. Antibodies against TFRC were purchased from Invitrogen. Antibodies against α-Tubulin and GAPDH, as well as secondary antibodies, were purchased from Hua-an Biotechnology. Details and information of antibodies are provided in Additional file 1: Table S1.

Approximately 3000–5000 cells were inoculated into each well of a 96-well plate. After 24 h of treatment with different concentrations of juglone with or without inhibitor, the supernatant was discarded, 100 μL of DMEM medium containing 10% CCK-8 (APE × BIO, K1018) was added to each well, incubated for 1 h at 37 °C, then the absorbance was measured at a wavelength of 450 nm.

Approximately 1500–2000 cells per well were inoculated in a 6-well plate. Until the cells grew into visible colonies. Then LN229 or T98G cell lines were maintained in juglone medium for 24 h (cells pre-treated with inhibitor for 1 h), then the medium was replaced with normal growth medium. The same operation needed to be repeated approximately 2 times, with the cells maintained for about 15 days. Fixed with 4% polyformaldehyde for 30 min, stained with crystal violet solution (1% crystal violet in 95% ethanol) for 2 h.

RIPA buffer supplemented with protease inhibitors was used to extract total protein; protein quantitation was conducted using the BCA protein assay kit (Biosharp, China). Samples were electrophoresed on a 10% SDS polyacrylamide gel (SDS-PAGE), then were transferred onto an NC membrane. The membrane was then blocked with 5% non-fat milk for 1 h, followed by washing, incubation with primary and secondary antibodies, and finally detected with enhanced chemiluminescence. The Nuclear and Cytoplasmic Protein Extraction Kit (78833, Thermo Scientific) was used for nuclear Nrf2 and cytosol Nrf2 extraction according to the manufacturer’s instructions, GAPDH and histone H3 (1:3000; Abcam) were served as internal reference proteins.

LN229 and T98G cells treated with different concentrations of Juglone for 24 h were fixed with 4% polyformaldehyde for 15 min. After fixation, the cells were washed twice with PBS and treated with PBS solution of 0.2% Triton-X 100 for 15 min. Then, the cells were incubated with 5% skim milk powder at room temperature for 1 h. Primary antibody was incubated at 4 °C for 12 h, and the secondary antibody was incubated at room temperature for 1 h, DAPI staining was done for 10 min, followed by the addition of anti-fluorescence quenching agents and observed with a fluorescence detection microscope. Images were captured and analyzed subsequently with the MetaView software.

After the cells were treated with different concentrations of juglone for 24 h, the cells were collected, lysed with a lysis buffer to obtain the supernatant, and centrifuged to obtain the supernatant, followed by overnight incubation at 4 °C with 2 μL of the antibody and 20 μL of the Protein A/G beads (Thermo Scientific). The sample was then centrifuged, washed three times with IP buffer, then the resulting protein sample was resuspended in 1 × loading buffer and subjected to a metal bath at 100 °C for 10 min, and the immunoprecipitated protein was analyzed by immunoblotting with anti-ubiquitin antibody.

Approximately 2 × 10⁵ cells were seeded onto 6-well plates. After 24 h of incubation, LN229 and T98 cells were treated with different methods for another 24 h, then replaced with HBSS containing 10 μmol/L DCFDA for incubation for 30 min. The cells were washed twice with PBS to remove DCFDA. Then the cells were resuspended in 500 μL HBSS and detected by flow cytometry, and the results are analyzed using FlowJo V10 software. Meanwhile, DCFDA was also used for immunofluorescence detection. LN229 and T98 cells were seeded in 6-well plate and subjected to different treatments for 24 h. Then they were stained with 10 μmol/L DCFDA for 1 h and then stained with Hoechst for 10 min. The fluorescence microscope was used to analyse the results.

The measurements of intracellular GSH and MDA levels were determined using GSH test kit (G263, Tong Ren, Japan) and MDA test kit (Sangon, China) respectively. All experimental operations were completed according to the kit operating instructions.

The ferroptosis-related genes were sourced from the FerrDb database (http://​www.​zhounan.​org/​ferrdb) [36]. The potential targets of juglone were predicted using Swiss Target Prediction (http://​www.​swisstargetpredi​ction.​ch/​) by applying the downloaded Juglone’s SDF and mol2 format [37]. The protein–protein interaction (PPI) network was obtained from the STRING database 11.0 (http://​string-db.​org/​) and was analyzed and visualized through Cytoscape 2.8.2 [38]. Further, the R language software was used to analyze GO and KEGG data and draw volcano plot and heatmaps.

The crystal structure corresponding to the Keap1 and p38 proteins was obtained from the RCSB PDB database (Structural Bioinformatics Protein Data Bank) [39]. The protein crystals obtained were processed separately using the Protein Preparation Wizard module in Schrödinger software. The 2D structure data file of the compound juglone was processed using the LigPrep module in Schrödinger to generate all its 3D chiral conformations. The prepared ligand compound Juglone was docked with the active sites of the p38 and Keap1 protein structures, respectively, and the scores were calculated.

Nrf2 overexpression vector was purchased from Genepharma (Shanghai, China). Overexpression plasmid (pcDNA3.1-Nrf2) was transfected into logarithmically growing LN229 and T98G cells using Lipofectamine 3000 (Thermo Fisher Scientific). After transfection, cells were collected and the increased expression level of Nrf2 was confirmed by western blot assays.

The si-RNA targeting Nrf2 and its negative control siRNA (si-control) were obtained from GenePharma (Shanghai, China). The sequences of the siRNAs are shown in Additional file 1: Table S2. Transfection was conducted according to Lipofectamine 3000 (Thermo Fisher Scientific) instructions.

LN229 cell suspension (4 × 10⁶ cells/100 μL) was subcutaneously injected into the axilla of 6-week-old nude mice (Sibeifu Biotechnology Co., Ltd, Beijing, China). When the tumor volume was visible (7 days after inoculation), all animals were randomly divided into four groups, each with 6 mice (3 females, 3 males), which were the control group (saline), juglone group (200 mg/kg/day), Fer-1 (0.8 mg/kg/day) group, and Fer-1 (0.8 mg/kg/day) + juglone (200 mg/kg/day) group, receiving abdominal injections every other day (3 times a week), with tumor size measured every 3 days, one mouse from each group died early during experimental procedure. 17 days after the treatment, the mice were euthanized and the tumors were removed. Then, subsequent Western blot analysis and immunohistochemical experiments were conducted with tumor tissues obtained from different pharmaceutical groups. This animal experiment was approved by the Ethical Committee of China Medical University (Approval No.: CMUXN2023016; Approval Date: June 28, 2023).

Statistical analyses were performed using GraphPad Prism9 (GraphPad software). All experiments were performed at least three independent times and results were expressed as the mean ± SD (standard deviation). p < 0.05 was considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.001).

To determine the anti-tumor effect of juglone on GBM cells, we treated two types of GBM cells, LN229 and T98G cells, with different concentrations of juglone. Its chemical structure is shown in Fig. 1A. We used the CCK-8 method to determine that juglone significantly inhibited GBM cells proliferation in a time‐dependent manner. The IC50 values of juglone are 11.62 μM (24 h), 9.417 μM (48 h), and 9.214 μM (72 h) in LN229 cells, and 28.80 μM (24 h), 23.52 μM (48 h) and 23.45 μM (72 h) in T98G cells, respectively. Therefore, we chose the IC50 value of juglone at 24 h for subsequent experiments. The drug concentration gradients were set at 5 μM, 7.5 μM, 10 μM (LN229) and 15 μM, 20 μM, 25 μM (T98G), respectively, for subsequent experiments (Fig. 1B). Compared with the control group, after the treatment with juglone, the cell colonies shrank, and the number of cells decreased in a dose-dependent manner (Fig. 1C, D), validating the inhibitory effect of juglone on the proliferation of GBM cells. Further Ki67 immunofluorescence detection results showed that in LN229 and T98G cells, the fluorescence intensity of Ki67 in the two types of glioma cells increased with the increase in drug concentration when induced by juglone (Fig. 1E, F). In summary, the above results indicated that juglone markedly inhibited cell proliferation and enhanced apoptosis in GBM.

We further explore the role of ferroptosis induced by juglone in GBM cells. To examine the effects of different cell death inhibitors, LN229 and T98G cells were cotreated with juglone and various inhibitors of cell death for 24 h.CCK-8 experiments show that Fer-1 and Lip-1, inhibitors of ferroptosis, can significantly reverse the inhibitory effect of juglone on LN229 and T98G cell activity (Fig. 2A), implying that ferroptosis is the major mechanism through which juglone induces death in GBM cells. Characteristics of ferroptosis are accumulation of ROS and lipid peroxides. By using flow cytometry technique to measure ROS in LN229 and T98G cells, we found that Juglone can dose-dependently enhance the accumulation of ROS in GBM cells (Fig. 2B, C). At the same time, immunofluorescence experiments also yielded the same result, namely, juglone significantly enhanced the green fluorescence intensity of GBM cells (Fig. 3A, B). As glutathione (GSH) consumption is a necessary process for ferroptosis, through GSH detection, we verified the predicted result that juglone can reduce the content of GSH (Fig. 2D). MDA and 4HNE are by-products of lipid peroxidation, and our results show that an increase in juglone concentration can trigger an increase in these two peroxides, with a significant statistical difference compared with the control group (Fig. 2E–G). We measured the expression of several key proteins relating to ferroptosis. WB experiments showed that in LN229 and T98 cells treated with juglone, the expression level of PTGS2 increased and the expression level of GPX4 decreased (Fig. 3C, D). Similarly, plate cloning experiments show that Fer-1 can reverse the reduction in the number of T98 and LN229 cell colonies induced by juglone (Fig. 3E, Additional file 1: Fig. S1). To further confirm whether juglone induces ferroptosis in GBM cells, we used transmission electron microscopy to detect morphological changes in mitochondria in T98G cells after 24 h of juglone treatment. The results show that when the juglone concentration is 20 μM, some mitochondrial membranes are significantly thickened, the mitochondria shrunk, there is a decrease or disappearance of mitochondrial cristae, and some mitochondria are swollen. As the drug concentration increased to 25 μM, more mitochondria shrank and the mitochondrial membrane ruptured, which is consistent with the characteristics of mitochondrial damage in ferroptosis (Fig. 3F). Simultaneously, after 24 h of juglone treatment, RNA was extracted from T98G cells and transcriptome sequencing was carried out. The results of the differential gene analysis obtained by comparing with the blank control group and the ferroptosis-related genes dataset from FerrDb were incorporated into the R package “clusterProfiler” for GSEA enrichment analysis. The results show a statistically significant difference (P < 0.05), suggesting a potential association between gene changes under the effect of juglone and the ferroptosis pathway. The overall results show that ferroptosis is the main form of cell death induced by juglone in GBM cells.

Using the Swiss Target Prediction database, 102 genes for potential drug target of juglone were selected. The intersection of the predicted drug targets and the ferroptosis-related genes from FerrDb database identified 27 key targets (Fig. 4A), which were analyzed using the String database and visualized through the Cytoscape software to identify the entire protein–protein interaction network (Fig. 4B). Through GO annotation, proteins were sorted into two main categories (biological processes and molecular functions). GO and KEGG enrichment analysis suggested that juglone induced ferroptosis is related to the MAPK family activity (Fig. 4D). Meanwhile, we selected the targets with the highest correlation and combined with existing literature, the expression of these key targets was verified through WB experiments. The results revealed that the expression levels of p-p38 and mmp2 significantly increased with the increase of drug concentration (Fig. 4C, Additional file 1: Fig. S3). We introduced GPX4 into the STRING network, as shown in Fig. 4G, and it indicated that p38 might interact with the Nrf2 -GPX4 axis, suggesting that this pathway may play a major role in juglone-induced ferroptosis. Transcriptome sequencing showed that, compared to the blank control group, juglone treatment resulted in significant upregulation of 2519 genes and significant downregulation of 190 genes (Fig. 4E). As the heatmap displayed within the intersection results of the FerrDb database showed, The gene expression levels Keap1 was significantly higher in the juglone treatment group (Fig. 4F). It is known that Keap1 is a negative regulator of Nrf2. WB analysis confirmed that with the increase in drug concentration of juglone, the expression levels of Keap1 protein increased and those of Nrf2 protein decreased significantly (Fig. 4H, I), consistent with this result, Nrf2 in both nucleus and cytoplasm decreased (Additional file 1: Fig. S4). Considering that Nrf2 is mainly degraded in the cell through ubiquitination, the immunoprecipitation experiment was conducted to detect the ubiquitination modification of Nrf2 protein in different juglone concentrations. The results showed that the level of Nrf2 ubiquitination increased with the concentration of juglone (Fig. 4J). We then conducted docking studies to verify the binding ability of juglone with p38 and Keap1, respectively. Juglone penetrates deep into the active pocket of p38 protein, forming a hydrophobic interaction with the residues ALA51, VAL38, PHE169, LEU167, ILE84, etc., of the p38 protein, and forming a hydrogen bond with the residue ALA51. Binding energy calculation showed that the binding energy of juglone-P38 was − 19.77 kcal/mol, indicating that juglone stably binds to p38. Similarly, juglone binds to the surface of the active pocket of Keap1 protein, forming a hydrophobic interaction with the residues LEU139, ALA138, LEU134, LEU110, LEU137, etc. of Keap1 protein, forming a hydrogen bond with the residues ASN414, SER602, and forming a π–π bond with the residue TYR334. The binding energy of Juglone to Keap1 was − 26.49 kcal/mol, indicating that juglone stably binds to p38 (Fig. 4K).

Next, we investigated whether inhibition of p38 kinase phosphorylation would reverse the inhibitory effect on Nrf2 expression and its downstream GPX4 expression induced by juglone. Through WB analysis, we observed that the p38 phosphorylation inhibitor SB203580 inhibited the downregulation of Nrf2 and GPX4 in LN229 and T98 cells induced by juglone (Fig. 5A). Meanwhile, Plate clone formation experiment showed that SB203580 partially rescued the death of LN229 and T98G cells induced by juglone (Fig. 5B, C). In addition, the consumption of GSH and accumulation of MDA were also rescued by SB203580 (Fig. 5D, E), and SB203580 also rescued the ROS accumulation induced by juglone (Fig. 5F, G).

To further verify whether the Nrf2/GPX4 pathway is involved in the ferroptosis of GBM induced by juglone, we added the Nrf2 protein agonist t-BHQ or used the Nrf2 overexpression vector to transiently transfect LN229 and T98G cells. As shown in Fig. 5H and Additional file 1: Fig. S3, the above treatments confirmed that the increased expression level of Nrf2 protein rescued the expression of GPX4 from the juglone-induced reduction. In addition, treatment with t-BHQ activated Nrf2 expression, partially rescued the growth inhibition of LN229 and T98G cells induced by juglone in the plate clone experiment (Fig. 5I, J). Consistent with this, the accumulation of ROS was successfully eliminated by overexpression of Nrf2 and activation of t-BHQ (Fig. 5M, N). Similar results were also observed for the levels of GSH and MDA (Fig. 5K, L). In order to further explore the effect of juglone on Nrf2 in GBM cells, we designed two si-RNAs targeting Nrf2. The experimental results showed that knocking down Nrf2 enhanced the cytotoxic effect of juglone on T98G cells, and at the same time, knocking down Nrf2 further increased the accumulation of MDA and the consumption of GSH induced by juglone in T98G cells. These results indicate that the Nrf2/GX4 axis is involved in the ferroptosis process induced by juglone, and juglone can induce GBM cells ferroptosis by inhibiting the Nrf2/GPX4 pathway through activating p38 phosphorylation.

For an in-depth understanding of juglone’s anti-tumor in vivo, we chose to establish xenograft tumor model of nude mice and administer the drug every other day by intraperitoneal injection according to the experimental scheme. Although we observed that juglone treatment did not significantly affect the weight of the mice (Fig. 6B), we still observed a significant reduction in tumor volume and weight in the treatment group. Notably, the reduction in tumor volume and weight were reversed by the ferroptosis inhibitor Fer-1 (Fig. 6C–E). In our study, we also observed a similar reversal of the MDA and GSH levels in subcutaneous tumors in mice treated with juglone (Fig. 6G). To assess the effects of juglone more comprehensively, we performed further WB and immunohistochemical experiments. The results show that juglone significantly inhibits the expression of Ki67, GPX4, and Nrf2, and this inhibition is significantly reversed by the injection of Fer-1 (Fig. 7A–D). Thus, it can be seen that juglone inhibits tumor growth in vivo by inducing ferroptosis. Based on all these results, we have proposed a mechanistic model to explain the potential role of juglone in GBM cells (Fig. 8).