New juglone derivatives (D1-D4) were prepared according to literature procedures (Clive et al., 2004; Mitchell et al., 2013; Li and Shen, 2020). Ag₂O (117 mg, 0.5 mmol) and alkyl halide (1.5 mmol) was added to the solution of juglone (174 mg, 1 mmol) in CH₂Cl₂ (5 mL). The reaction mixture was stirred at room temperature for 24 h. After filtration through celite and removal of the solvent in vacuo, the residue was subjected to flash column chromatography on silica gel (230–400 mesh) using n-hexane/ethyl acetate as eluent to give the product D1-D4.

All reactions were carried out in oven-dried glassware under an atmosphere of dry N₂ with the rigid exclusion of air and moisture using standard Schlenk techniques. Dichloromethane was freshly distilled from CaH₂ immediately before use. All other chemicals were purchased from either J&K Chemical Co. or used as received unless otherwise specified. ¹H and ¹³C{¹H} NMR (Nuclear magnetic resonance) spectra were recorded on a Varian Inova 400 spectrometer at 400 and 100 MHz, respectively. All signals were reported in ppm unit with references to the residual solvent resonances of the deuterated solvents for proton and carbon chemical shifts. Mass spectra were obtained on a Thermo Finnigan MAT 95 XL spectrometer, Shanghai Institute of Organic Chemistry, CAS.

GBM primary cells were isolated by specimens derived from patients in Huashan Hospital with full consent after approval from the local ethic committee. The molecular pathology of the specimen for in vivo experiment is IDH-wildtype, MGMT unmethylation, and TP53 mutation. U251 and U87 were purchased from China Academia Sinica Cell Repository (Shanghai, China). Primary GBM cells and glioma cell lines were cultured in Dulbecco’s modified Eagle’s Medium (DMEM; HyClone, Logan, UT, United States) supplemented with 10% fetal bovine serum (FBS; Gibco BRL, Gaithersburg, MD, United States). Cell cultures were maintained in a 5% CO₂ humidified incubator at 37°C.

GBM specimen was placed in ice and transferred to the lab within 1 h after surgical resection. The specimen was washed with PBS to remove the blood and necrotic tissue. Then the minced GBM tissue by a surgical knife blade was digested with 0.25% trypsin in the falcon tube at 37°C for 15 min and shaken every 5 min. The tissues were triturated into single cells by a 5 ml pipette and a 40 μm filter was used to remove tissue debris. Single cells were centrifuged and resuspended with 1 ml RBC lysis buffer at room temperature for 5 min. At last, GBM cells were resuspended and cultured in DMEM with 10% FBS. For subculture, cells were passaged once they reached 80–90% confluence. An intracranial implantation experiment was performed on cells within passages two and five to minimize genetic mutation.

Juglone (Sigma, America) and derivatives were dissolved in dimethyl sulfoxide (DMSO) and diluted in DMEM. Cell viability was assessed by the Cell Counting Kit-8 assay (CCK-8, Dojindo, Japan). Briefly, tumor cells that were cultured in DMEM with 10% FBS were seeded in 96-well plates at a density of 1×10⁴ cells/100ul/well and incubated overnight. Tumor cells were pretreated with and without NAC (2 mM, Beyotime, China) for 1 h. After treatment of juglone or its derivatives in different concentrations for 48 h, cells were incubated for 1 h with 10 μl CCK-8 per well. The optical density (OD) was measured at 450 nm with a microplate spectrophotometer (Bio-Rad, United States). Proliferation was examined using the EdU incorporation (Ribobio, China) assay, which was performed according to the manufacturer’s protocol, and the cells were examined under a fluorescence microscope. The experiment was triplicated, and each contained six replicates.

For apoptosis assay, glioma cells were treated with two kinds of juglone derivatives for 48 h, 1×10⁵ cells were harvested, resuspended in 100 µl binding buffer, and incubated with 5 µl Annexin V-FITC, 10 µl of PI (BD, San Joe CA) in darkness at room temperature for 15 min. After that, 400 µl of binding buffer was added before being tested on a FACS Calibur cytometer (BD, San Joe CA). FlowJo software (Tree Star, Ashland OR) was used to analyze the data. The experiments were triplicated.

DCFDA fluorescent assay (Sigma Aldrich, United States) was used to label intracellular reactive oxygen species (ROS) and then detected with flow cytometry. Briefly, 5×10⁵ cells were pretreated with different concentrations of juglone derivatives for 6 h and then loaded with DCFDA (10 μM) probe. After incubation at 37°C for 30 min, cells were harvested, washed, resuspended with PBS, and fluorescence intensity was measured by flow cytometry. FlowJo software was used to analyze the mean fluorescence intensity (MFI). The experiments were repeated three times.

The potential targets of juglone (SMILES: C1 = CC2 = C(C(=O)C=CC2 = O)C (=C1)O) were obtained from SwissTargetPrediction (http://swisstargetprediction.ch/). The top 100 genes were selected for the construction of a protein-protein interaction (PPI) network (https://string-db.org/). The TSV format file was downloaded from string and imported into Cytoscape software (version 3.8.0) for visualization. Molecular docking was performed using UCSF Chimera (Pettersen et al., 2004). The structure data of CDC25B and DUSP1 for docking were obtained from alpha fold (https://alphafold.com/) (Jumper et al., 2021; Varadi et al., 2022).

After treating with different concentrations of juglone derivatives for 48 h, total protein of glioma cell lines of U87 and U251 were obtained from RIPA lysis buffer with 1% PMSF (Beyotime, China). The protein concentration was determined by BCA assay (Beyotime, China), and samples were separated on 10% SDS-PAGE, and then transferred onto NC membranes (0.45 μm, Millipore, United States). The membranes were incubated with primary Abs against cleaved-PARP (1:1000, Cell Signaling Technology, China) and β-actin (1:10000, Cell Signaling Technology, China) overnight at 4°C, followed by HRP conjugated secondary Ab (1:3000, Cell Signaling Technology, China). The protein bands were visualized using enhanced chemiluminescence (ECL, Millipore, United States) and a detection system (ChemiDoc Touch, Bio-Rad)

All animal procedures were conducted according to protocols approved by the Institutional Animal Care and Use Committee at Fudan University.

Female BALB/c-nu mice (3–4 weeks old) provided by SLAC Laboratory Animal Company (Shanghai, China) were used as orthotopic xenograft recipients. Mice were housed in an environment with a 12-h light/dark cycle. At least 1 week time was provided for mice to acclimatize new environment before experimentation. Human primary GBM cells were labeled with lentivirus expressing luciferase. Before intracranial transplantation, cells were digested into single cells and suspended with PBS at a density of 1×10⁵ cells/μL. Mice were anesthetized intraperitoneally with 10% chloral hydrate and secured into a stereotaxic apparatus. GBM cells in 10 uL PBS via a Hamilton syringe were injected into the right forebrain (2.5 mm lateral and 1 mm anterior to bregma, at a 2.5 mm depth from the skull surface). The mice were randomly divided into three groups (control group, D2, and D3 treatment group). The number in each group was five. D2 and D3 were dissolved in DMSO and diluted in PBS; the final concentration of DMSO was 20 mg/ml. PBS containing the same concentration of DMSO was used as vehicle control. Juglone derivatives treatment group was injected intraperitoneally with D2 and D3 (1 mg/kg) every 2 days, which was the same as the previous dosage. Bioluminescent imaging was performed on the twenty-eighth day after transplantation with IVIS-200 (Xenogen, United States) to test the tumor volume. Before anesthesia, D-luciferin (Yeason, China) was injected intraperitoneally at 150 mg/kg body weight. Images of different groups were captured with the same parameters. Bioluminescence values of intracranial tumors were quantitated using the Living Image software. Mice were euthanized when neurological symptoms appeared and perfused with 4% paraformaldehyde in PBS.

The whole brain, heart, liver, and kidney of each mouse were collected. Fixation with 4% paraformaldehyde in PBS, dehydration with gradient ethanol, embedded in paraffin, and section cut in 5 μm thickness were performed. The sections were stained with hematoxylin and eosin (H&E). Cell morphology of different tissue was observed under a light microscope.

All quantified data were presented as mean ± SEM. For comparison between the two groups, two-tailed student’s t-tests were used to calculate p values. A p value < 0.05 was considered statistically significant.

The phenomenon that juglone is easy to be oxidated has previously been observed. Juglone solution took on dark brown color changes in a time-dependent manner when preserved at 4°C in an EP tube (Figure 1A). The maximum absorbance wavelength of the samples changed, which indicated several compounds existed after oxidation (Figure 1B). The cytotoxicity of juglone oxidation products was assessed by using the CCK-8 assay. Interestingly, juglone oxidation products had pro-tumor effects in low concentrations and antitumor effects in high concentrations. The cytotoxic effects of juglone were decreased dramatically in the same concentration after oxidation (Figure 1C). To figure out what the oxidation product was, we tested the solutions of juglone in fresh, 1 day, and 1 week after dissolution by using mass spectrometry (Supplementary Material S1). The peak of juglone could be detected in fresh solution (Molecular Weight: 174kD), while many unknown compounds could be found, and juglone itself decreased obviously after oxidation (Figure 1D).

The chemical protocol of synthesizing the target compounds is shown in Figure 2A. Compounds D1-D4 showed the characteristic peaks in 1H NMR spectra at corresponsive ppm respectively. 5-Methoxy-1,4-naphthoquinone (D1): Yellow solid. Yield: 84%. ¹H NMR (400 MHz, CDCl₃): δ 7.70 (m, 2H), 7.31 (d, J = 8.4 Hz, 1H), 6.86 (m, 2H), 4.00 (s, 3H). These data are identical with those reported in the literature (Mitchell et al., 2013) (Figure 2B, Figure 2C). 5-Allyloxy-1,4-naphthoquinone (D2): Yellow solid. Yield: 82%. ¹H NMR (400 MHz, CDCl₃): δ 7.73 (d, J = 7.6 Hz, 1H), 7.65 (t, J = 8.4 Hz, 1H), 7.29 (d, J = 8.4 Hz, 1H), 6.87 (m, 2H), 6.10 (m, 1H), 5.66 (d, J = 17.6 Hz, 1H), 5.37 (d, J = 10.4 Hz, 1H), 4.73 (m, 2H). These data are identical with those reported in the literature (Clive et al., 2004) (Figure 2D, Figure 2E). 5-Buthoxy-1,4-naphthoquinone (D3): Brown oil. Yield: 81%. ¹H NMR (400 MHz, CDCl₃): δ 7.70 (d, J = 7.2 Hz, 1H), 7.65 (t, J = 8.0 Hz, 1H), 7.29 (d, J = 8.4 Hz, 1H), 6.85 (m, 2H), 4.14 (t, J = 6.4 Hz, 2H), 1.89 (m, 2H), 1.60 (m, 2H), 1.01 (t, J = 7.6 Hz, 3H). ¹³C{¹H} NMR (100 MHz, CDCl₃): δ 185.2, 184.0, 159.1, 140.7, 135.9, 134.7, 133.8, 119.6, 118.8, 118.7, 69.0, 31.0, 19.0, 13.7. HRMS (EI) Calcd for C₁₄H₁₅O₃ ⁺ (M + H⁺): 231.1016. Found: 231.1019 (Figure 2F, Figure 2G). 5-Benzyloxy-1,4-naphthoquinone (D4): Orange solid. Yield: 86%. ¹H NMR (400 MHz, CDCl₃): δ 7.74 (d, J = 7.6 Hz, 1H), 7.65 (t, J = 8.4 Hz, 1H), 7.58 (d, J = 7.6 Hz, 2H), 7.42 (t, J = 8.4 Hz, 2H), 7.34 (d, J = 8.4 Hz, 2H), 6.89 (s, 2H), 5.30 (s, 2H). These data are identical with those reported (Li and Shen, 2020) (Figure 2H, Figure 2I). The purity of compounds (>95%) was measured with GC-MS (Supplementary Material S2).

Cell viability of U87 and U251 were evaluated by CCK-8 assay after treatment with four juglone derivatives for 48 h. As shown in Figure 3A and Figure 3B, D1, D2, and D3 could dramatically decrease the viability of glioma cells. IC50 of four kinds of derivatives for U87 were 3.99, 3.28, 7.60, and 11.84 μM, respectively. In the U251 cell line, IC50 were 7.00, 5.43, 8.64, and 18.05 μM, respectively (Table 1). D2 and D3, which had better cytotoxicity and lipid-solubility, were chosen for further experiments. EdU assay was used to evaluate the D2 and D3 effects on glioma cell proliferation. As shown in Figures 3C,D, both D2 and D3 could attenuate cell proliferation in a dose-dependent manner.

To further reveal the underlying mechanism of anti-glioma effects, a pharmaceutical network was established among the predicted targets of juglone. SwissTargetPrediction showed juglone had a broad spectrum of potential targets, most of which are enzymes (25.0%), protease (15.0%), and kinase (13.0%, Figure 4A). KEGG analysis showed apoptosis pathway can be most significantly enriched (Figure 4B). Apart from apoptosis, the protein-protein interaction (PPI) network indicated that juglone could affect biological processes such as cell cycle, metabolism, immune reaction, and epigenomic status to a large extent (Figure 4C). CDC25B and DUSP1 were the two most likely candidates of juglone targets (Table 2), which were reported to be associated with apoptosis (Miyata et al., 2001; Robitaille et al., 2017). Molecular docking further provided interaction details between juglone and these two targets. Figure 4D showed juglone could insert into the pocket of CDC25B and interact with L477 and R479. In the meantime, juglone could protrude into a hydrophobic pocket of DUSP1, and the aromatic ring of juglone could interact with the hydrophobic residues of DUSP1 (A33 and F287) via hydrophobicity (Figure 4E).

To further validate the effect of juglone derivatives on apoptosis, U87 and U251 cells were stained with Annexin V/PI after treatment of D2 (2 μM, 6 μM) and D3 (5 μM, 10 μM). Flow cytometry analysis showed that D2 and D3 could induce apoptosis and increase the percentage of Annexin V+/PI + cells both in U87 and U251 groups (Figure 5A). To substantiate these phenomena, the expression of cleaved-PARP was evaluated by western blot. As shown, cleaved-PARP was up-regulated after treatment of D2 and D3 in a dose-dependent manner, which indicated these derivatives could induce apoptosis in gliomas (Figure 5B).

As reported in our previous work, juglone could induce ROS generation via p38-MAPK pathway activation. In this study, ROS production was also measured with a ROS assay kit by flow cytometry. As demonstrated, D2 and D3 could significantly induce ROS generation in U87 and U251 cells (Figure 6A). In addition, NAC, a ROS scavenger, reversed the cytotoxic effect, indicating the involvement of ROS generation in the anti-glioma effect of D2 and D3 (Figure 6B).

To investigate whether D2 and D3 could effectively inhibit glioma in vivo, the orthotopic glioblastoma model was first established (n = 15) and then assigned to the following groups randomly: control (n = 5), D2 (n = 5), D3 (n = 5). Three days after brain implantation of human primary glioma cells infected with lentivirus expressing luciferase into nude mice, vehicle, D2 (1 mg/kg) or D3 (1 mg/kg) were administrated intraperitoneally every other day. Both D2 and D3 had an inhibitory effect on glioma growth confirmed with in vivo imaging systems 28 days later after tumor transplantation (Figures 7A,B). HE staining showed that both D2 and D3 could inhibit tumor growth (Figures 7B,D), and no obvious histological harm to the heart and kidney could be observed in the group D2 and D3 compared to the vehicle group. However, there was partial necrosis of liver cells in the D2 and D3 groups (Supplementary Material S5). Kaplan–Meier analysis of survival data demonstrated a statistical difference between the control and the D2 or D3 group (Figure 7E).