Exploiting metabolic vulnerability in glioblastoma using a brain-penetrant drug with a safe profile

Abstract

Glioblastoma is one of the most treatment-resistant and lethal cancers, with a subset of self-renewing brain tumour stem cells (BTSCs), driving therapy resistance and relapse. Here, we report that mubritinib effectively impairs BTSC stemness and growth. Mechanistically, bioenergetic assays and rescue experiments showed that mubritinib targets complex I of the electron transport chain, thereby impairing BTSC self-renewal and proliferation. Gene expression profiling and Western blot analysis revealed that mubritinib disrupts the AMPK/p27^Kip1 pathway, leading to cell-cycle impairment. By employing in vivo pharmacokinetic assays, we established that mubritinib crosses the blood-brain barrier. Using preclinical patient-derived and syngeneic models, we demonstrated that mubritinib delays glioblastoma progression and extends animal survival. Moreover, combining mubritinib with radiotherapy or chemotherapy offers survival advantage to animals. Notably, we showed that mubritinib alleviates hypoxia, thereby enhancing ROS generation, DNA damage, and apoptosis in tumours when combined with radiotherapy. Encouragingly, toxicological and behavioural studies revealed that mubritinib is well tolerated and spares normal cells. Our findings underscore the promising therapeutic potential of mubritinib, warranting its further exploration in clinic for glioblastoma therapy.

Keywords: Hypoxia; Metabolic Reliance; Oxidative Phosphorylation; Radiotherapy; Reactive Oxygen Species.

Figure 1. Mubritinib alters mitochondrial respiration through complex I inhibition and impairs BTSC growth.

(A–C) Patient-derived BTSCs, BTSC53 (A), BTSC73 (B) and BTSC147 (C) were subjected to real-time Resipher analysis to measure the basal oxygen consumption rate (OCR) following mubritinib treatment. Data are presented as the means ± SEM, n = 3 independent biological experiments. Two-way ANOVA followed by Dunnett’s test vs vehicle control. BTSC53 (A): ***p₂₀ = 1.9e−10, ***p₁₀₀ = 1.2e−14, ***p₅₀₀ = 1.2e−14. BTSC73 (B): ***p₂₀ = 1.2e−14, ***p₁₀₀ = 1.2e−14, ***p₅₀₀ = 1.2e−14. BTSC147 (C): ***p₂₀ = 2.2e−11, ***p₁₀₀ = 1.2e−14, ***p₅₀₀ = 1.2e−14. (D, E) Basal mitochondrial respiration, maximal mitochondrial respiration and spare respiratory capacity (SRC) were measured in BTSC53 (D) and BTSC73 (E) following mubritinib treatment using the Resipher system. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. BTSC53 (D): *p_{basal (20)} = 0.0105; ***p_{basal (100)} = 0.0005; ***p_{basal (500)} = 0.0004; ***p_{maximal (20)} = 0.0005, ***p_{maximal (100)} = 5.1e−5, ***p_{maximal (500)} = 5.9e−5; **p_{SRC (20)} = 0.0018, ***p_{SRC (100)} = 0.0004, ***p_{SRC (500)} = 0.0006. BTSC73 (E): **p_{basal (20)} = 0.0025; ***p_{basal (100)} = 0.0007; ***p_{basal (500)} = 0.0004; ***p_{maximal (20)} = 2.3e−6, ***p_{maximal (100)} = 3.2e−7, ***p_{maximal (500)} = 2.3e−7; *p_{SRC (20)} = 0.021, **p_{SRC (100)} = 0.0038, **p_{SRC (500)} = 0.0031. (F) Multiple patient-derived BTSCs, BTSC12, BTSC25, BTSC50, BTSC53, BTSC73, BTSC75, BTSC100, BTSC119, BTSC147, BTSC198 and P3, were exposed to increasing concentrations of mubritinib (0 to 500 nM) for 7 days, followed by live cell counting. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. ***p_{BTSC12 (20)} = 0.0003, ***p_{BTSC12 (100)} = 1.4e−5, ***p_{BTSC12 (500)} = 7e−6; *p_{BTSC25 (20)} = 0.0316, **p_{BTSC25 (100)} = 0.0052, ***p_{BTSC25 (500)} = 0.0007; ***p_{BTSC50 (20)} = 0.0003, ***p_{BTSC50 (100)} = 2.6e−6, ***p_{BTSC50 (500)} = 1e−6; ***p_{BTSC53 (20)} = 1.1e−5, ***p_{BTSC53 (100)} = 2e−6, ***p_{BTSC53 (500)} = 1e−6; ***p_{BTSC73 (20)} = 0.0007, ***p_{BTSC73 (100)} = 3.1e−5, ***p_{BTSC73 (500)} = 2.6e−5; ***p_{BTSC75 (20)} = 0.0007, ***p_{BTSC75 (100)} = 3.2e−5, ***p_{BTSC75 (500)} = 6e−6; *p_{BTSC100 (100)} = 0.0384, **p_{BTSC100 (500)} = 0.001; ***p_{BTSC119 (20)} = 0.0004, ***p_{BTSC119 (100)} = 6.9e−5, ***p_{BTSC119 (500)} = 1e−5; ***p_{BTSC147 (20)} = 9.4e−8, ***p_{BTSC147 (100)} = 1.2e−8, ***p_{BTSC147 (500)} = 8e−9; ***p_{BTSC198 (20)} = 0.0008, ***p_{BTSC198 (100)} = 0.0003, ***p_{BTSC198 (500)} = 0.0002; ***p_{P3 (100)} = 0.0003, ***p_{P3 (500)} = 0.0001. (G–J) EdU incorporation was analysed by immunofluorescence imaging of BTSCs (#53, #73 and #147) after 4 days of mubritinib treatment at concentrations of 20 nM, 100 nM and 500 nM. Representative images of EdU (red) staining are shown (G). Nuclei were stained with DAPI (blue). Scale bar = 50 µm. The number of EdU-positive cells was quantified with Fiji software in BTSC53 (H), BTSCS73 (I) and BTSC147 (J). Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. BTSC53 (H): **p₂₀ = 0.0045, ***p₁₀₀ = 0.001, ***p₅₀₀ = 0.0006. BTSC73 (I): **p₂₀ = 0.0012, ***p₁₀₀ = 2.3e−5, ***p₅₀₀ = 5e−6. BTSC147 (J): **p₂₀ = 0.0033, ***p₁₀₀ = 0.0006, ***p₅₀₀ = 0.0003. (K) Pearson correlation analysis between the basal oxygen consumption rate (OCR) and BTSC sensitivity score to mubritinib following 7 days of treatment was performed. (l) A schematic diagram of the mitochondrial electron transport chain with ectopic NDI1 expression is presented. (M–O) BTSC53 (M), BTSC73 (N) and BTSC147 (O) expressing the control (CTL) or NDI1 vector were treated with vehicle control or 500 nM mubritinib and subjected to Resipher analysis to measure the basal OCR, maximal respiration and SRC. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. BTSC53 (M): **p_{basal (CTL vs mubritinib)} = 0.0049, ***p_{basal (mubritinib vs NDI1 + mubritinib)} = 0.0004, p_{basal (NDI1 vs NDI1 + mubritinib)} = 0.7841; **p_{maximal (CTL vs mubritinib)} = 0.0011, ***p_{maximal (mubritinib vs NDI1 + mubritinib)} = 0.0003, p_{maximal (NDI1 vs NDI1 + mubritinib)} = 0.5223; p_{SRC (NDI1 vs NDI1 + mubritinib)} = 0.8773. BTSC73 (N): **p_{basal (CTL vs mubritinib)} = 0.0041, **p_{basal (mubritinib vs NDI1 + mubritinib)} = 0.0035, p_{basal (NDI1 vs NDI1 + mubritinib)} = 0.9933; ***p_{maximal (CTL vs mubritinib)} = 6.1e−5, ***p_{maximal (mubritinib vs NDI1 + mubritinib)} = 0.0001, p_{maximal (NDI1 vs NDI1 + mubritinib)} = 0.8470; *p_{SRC (CTL vs mubritinib)} = 0.016, *p_{SRC (mubritinib vs NDI1 + mubritinib)} = 0.0467, p_{SRC (NDI1 vs NDI1 + mubritinib)} = 0.9615. BTSC147 (O): *p_{basal (mubritinib vs NDI1 + mubritinib)} = 0.0192, p_{basal (NDI1 vs NDI1 + mubritinib)} = 0.9528; **p_{maximal (CTL vs mubritinib)} = 0.0018, ***p_{maximal (mubritinib vs NDI1 + mubritinib)} = 0.0009, p_{maximal (NDI1 vs NDI1 + mubritinib)} = 0.9472; *p_{SRC (CTL vs mubritinib)} = 0.0335, *p_{SRC (mubritinib vs NDI1 + mubritinib)} = 0.0396, p_{SRC (NDI1 vs NDI1 + mubritinib)} = 0.9993. (P–R) The number of live BTSC53 (P), BTSC73 (Q) and BTSC147 (R) expressing control vector or NDI1 after 7 days of 500 nM mubritinib treatment was measured. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. BTSC53 (P): ***p_{CTL vs mubritinib} = 0.0003, ***p_{mubritinib vs NDI1 + mubritinib} = 0.0005, p_{NDI1 vs NDI1 + mubritinib} = 0.4954. BTSC73 (Q): ***p_{CTL vs mubritinib} = 0.0004, **p_{mubritinib vs NDI1 + mubritinib} = 0.0017, p_{NDI1 vs NDI1 + mubritinib} = 0.1875. BTSC147 (R): ***p_{CTL vs mubritinib} = 0.0002, ***p_{mubritinib vs NDI1 + mubritinib} = 0.0002, p_{NDI1 vs NDI1 + mubritinib} = 0.0904. (S) Pearson correlation analysis was performed between the BTSC sensitivity score to mubritinib following 7 days of treatment and mitochondrial (MTC) transcriptional subtype signature score. Source data are available online for this figure.

Figure 2. Mubritinib impairs the AMPK/p27^Kip1 and cell-cycle in BTSCs.

(A–C) Gene set enrichment analysis of deregulated genes in BTSC73 treated with 500 nM mubritinib for 24 h demonstrating enrichment of gene sets corresponding to cyclin D1-associated events in G1 (A), SCF SKP2-mediated degradation of p27/p21 (B) and E2F-mediated regulation of DNA replication (C) are shown. (D, E) Cell cycle distribution was assessed by flow cytometry after PI staining of BTSC53 (D) and BTSC73 (E) following 24 h of treatment with 500 nM mubritinib. Data are presented as the means ± SEM, n = 3 independent biological experiments. Unpaired two-tailed t test. BTSC53 (D): **p_G1 = 0.0046, **p_S = 0.0023. BTSC73 (E): **p_G1 = 0.0014, *p_S = 0.0131, *p_G2 = 0.0471. (F) BTSC53 and BTSC73 were treated for 24 h with 500 nM of mubritinib or vehicle control and subjected to immunoblotting using antibodies against p-AMPK and total AMPK. Tubulin was used as loading control. Densitometric quantifications of p-AMPK protein level normalized to the corresponding loading control are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. Unpaired two-tailed t test. **p_BTSC53 = 0.0022, **p_BTSC73 = 0.0025. (G) BTSC53 and BTSC73 were treated for 24 h, 48 h or 72 h with 500 nM of mubritinib or vehicle control and subjected to immunoblotting using an antibody against cyclin D1. Tubulin was used as loading control. Densitometric quantifications of cyclin D1 protein level normalized to the corresponding loading control are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. *p_{BTSC53 24h} = 0.0123, ***p_{BTSC53 48h} = 6.6e−5, ***p_{BTSC53 72h} = 4e−6, *p_{BTSC73 24h} = 0.0422, **p_{BTSC73 48h} = 0.0020, ***p_{BTSC73 72h} = 0.001. (H) Nuclear fractions from BTSC53 and BTSC73 treated for 72 h with 500 nM of mubritinib or vehicle control and subjected to immunoblotting using an antibody against p27^Kip1. H2AX is used as nuclear loading control. Densitometric quantifications of nuclear p27^Kip1 protein level normalized to the corresponding loading control are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. Unpaired two-tailed t test. **p_BTSC53 = 0.0049, **p_BTSC73 = 0.0023. (I) BTSC53 and BTSC73 were treated for 24 h, 48 h or 72 h with 500 nM of mubritinib or vehicle control and subjected to immunoblotting using an antibody against phosphorylated retinoblastoma protein (p-Rb). Tubulin was used as loading control. Densitometric quantifications of p-Rb protein level normalized to the corresponding loading control are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. ***p_{BTSC53 24h} = 0.0005, ***p_{BTSC53 48h} = 3.1e−6, ***p_{BTSC53 72h} = 7e−7, ***p_{BTSC73 24h} = 2e−6, ***p_{BTSC73 48h} = 7e−8, ***p_{BTSC73 72h} = 2e−8. (J) Proposed model for the mechanism by which mubritinib impairs the AMPK/p27^Kip1 axis and cell cycle in BTSCs. Source data are available online for this figure.

Figure 3. Mubritinib impairs the stemness of BTSCs without affecting the non-oncogenic normal hNPCs.

(A–F) BTSC53 (A), BTSC73 (B), BTSC147 (C), BTSC12 (D), BTSC119 (E) and P3 (F) were subjected to extreme limiting dilution analysis (ELDA) to estimate the stem cell frequencies (SCF), 21 days following treatment with 500 nM mubritinib or vehicle control. Data are presented as the means ± SEM, n = 3 independent biological experiments. Chi-square test for ELDA plots and unpaired two-tailed t test for SCF were used. **p_BTSC53 = 0.0052, *p_BTSC73 = 0.0112, **p_BTSC147 = 0.0078, **p_BTSC12 = 0.0062, *p_BTSC119 = 0.0287, *p_P3 = 0.0454. (G–L) BTSC53 (G), BTSC73 (H), BTSC147 (I), BTSC12 (J), BTSC119 (K) and P3 (L) were treated for 4 days with 500 nM mubritinib or vehicle control and subjected to immunoblotting using the antibodies indicated on the blots. Vinculin and tubulin were used as loading controls. Densitometric quantifications of cleaved Notch1, Olig2 and SOX2 protein levels normalized to their corresponding loading controls are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. BTSC53 (G): ***p_{Cleaved Notch1} = 0.0008, ***p_Olig2 = 0.0007, **p_SOX2 = 0.0016. BTSC73 (H): **p_{Cleaved Notch1} = 0.0093, **p_Olig2 = 0.0036, **p_SOX2 = 0.0071. BTSC147 (I): **p_{Cleaved Notch1} = 0.0054, *p_Olig2 = 0.0188, *p_SOX2 = 0.0478. BTSC12 (J): **p_{Cleaved Notch1} = 0.0097, ***p_Olig2 = 0.0006, **p_SOX2 = 0.0011. BTSC119 (K): ***p_{Cleaved Notch1} = 7.4e−5, **p_Olig2 = 0.0033, **p_SOX2 = 0.0030. P3 (L): **p_{Cleaved Notch1} = 0.0017, ***p_Olig2 = 0.0005, ***p_SOX2 = 0.0005. (M) BTSC73 transduced with the control (CTL) or NDI1 vector were treated with vehicle control or 500 nM mubritinib and subjected to ELDA to estimate the SCF, 21 days following treatment. Data are presented as the means ± SEM, n = 3 independent biological experiments. Chi-square test for ELDA plots and one-way ANOVA followed by Tukey’s test for SCF were used. **p_{CTL vs mubritinib} = 0.0082, *p_{mubritinib vs NDI1 + mubritinib} = 0.0294, p_{NDI1 vs NDI1 + mubritinib} = 0.9946. (N) BTSC73 transduced with the CTL or NDI1 vector were treated with vehicle control or 500 nM mubritinib for 4 days and subjected to immunoblotting using the antibodies indicated on the blots. Vinculin was used as loading control. Densitometric quantifications of cleaved Notch1, Olig2 and SOX2 protein levels normalized to their corresponding loading controls are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. **p_{Cleaved Notch1 (CTL vs mubritinib)} = 0.00105, ***p_{Cleaved Notch1 (mubritinib vs NDI1 + mubritinib)} = 0.001, p_{Cleaved Notch1 (NDI1 vs NDI1 + mubritinib)} = 0.9999, ***p_{Olig2 (CTL vs mubritinib)} = 2.8e−5, ***p_{Olig2 (Mubritinib vs NDI1 + mubritinib)} = 1.4e−5, p_{Olig2 (NDI1 vs NDI1 + mubritinib)} = 0.7603, **p_{SOX2 (CTL vs mubritinib)} = 0.006, **p_{SOX2 (mubritinib vs NDI1 + mubritinib)} = 0.0022, p_{SOX2 (NDI1 vs NDI1 + mubritinib)} = 0.8297. (O) Human neural progenitor cells (hNPCs) were subjected to ELDA to estimate the SCF, 21 days following treatment with 500 nM mubritinib or vehicle control. Data are presented as the means ± SEM, n = 3 independent biological experiments. Chi-square test for ELDA plots and unpaired two-tailed t test for SCF were used. p = 0.7555. (P) hNPCs were treated with vehicle control or 500 nM mubritinib for 4 days and subjected to immunoblotting using the antibodies indicated on the blots. Vinculin and tubulin were used as loading controls. Densitometric quantifications of cleaved Notch1, Olig2 and SOX2 protein levels normalized to their corresponding loading controls are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. p_{Cleaved Notch1} = 0.4252, p_Olig2 = 0.6578, p_SOX2 = 0.9996. Source data are available online for this figure.

Figure 4. Mubritinib, a brain penetrant drug, delays BTSC tumourigenesis.

(A, B) Schematic diagram of mubritinib pharmacokinetic analysis is shown (A). C57BL/6N mice were treated by I.P with a single dose of mubritinib (6 mg/kg). The brain and plasma were collected at different time points (0.5 h to 48 h) after injection. Mubritinib was quantified by LC‒MS/MS (B). (C) Schematic diagram of the experimental procedure in which luciferase-expressing mGB2 cells were intracranially injected into C57BL/6N mice. One week after implantation, the mice were randomized into 2 groups: vehicle control or mubritinib (6 mg/kg). Mice were treated 3 times per week (Monday, Wednesday and Friday). Luciferase imaging was used to assess tumour progression. (D, E) Bioluminescence images (D) and quantification of luciferase activity (E) 6 weeks after implantation are presented. Data are presented as box plots showing 25th and 75th percentiles (box), median (centre line), minima and maxima (whiskers), n = 8 mice. Unpaired two-tailed t test. ***p = 0.001. (F) Kaplan–Meier (KM) survival plots were generated to evaluate the lifespan of the mice in each group, and the mice were collected at the end stage (log-rank test, n = 7 mice). (G–I) Luciferase-expressing BTSC147 were intracranially injected into RAGγ2C^−/− mice. Bioluminescence images (G) and quantification of luciferase activity (H) 9 weeks after implantation are presented. Data are presented as box plots showing 25th and 75th percentiles (box), median (centre line), minima and maxima (whiskers), n = 9 mice. Unpaired two-tailed t test. **p = 0.0037. KM survival plot was generated to assess animal lifespan (I). Log-rank test, n = 9 mice, ***p = 4.9e−5. (J) BTSC73 were intracranially injected into RAGγ2C^−/− mice, and a KM survival plot was generated to compare the lifespans of vehicle- and mubritinib-treated mice (log-rank test, n = 5 mice). (K–M) Representative immunofluorescence images and quantification of Olig2 (green) (K), SOX2 (red) (L) and Ki67 (red) (M) positive cells in BTSC73 intracranial xenografts in mice treated with mubritinib or vehicle are shown. Nuclei were stained with DAPI (blue). The number of Olig2-, SOX2- and Ki67-positive cells was quantified with Fiji software. Scale bar = 100 μm. Data are presented as the means ± SEM, n = 3 mice. Unpaired two-tailed t test. *p_olig2 = 0.0386, **p_SOX2 = 0.0075, *p_Ki67 = 0.0122. (N) Control or NDI1 vector expressing BTSC73 were intracranially injected into RAGγ2C^−/− mice and the mice were administered either vehicle control or mubritinib (6 mg/kg). KM survival plot was generated to assess animal lifespan (log-rank test, n = 5 mice). Source data are available online for this figure.

Figure 5. Mubritinib sensitizes BTSCs and GB tumours to IR.

(A, B) BTSC53 (A) and BTSC73 (B) were treated with 500 nM mubritinib, irradiated with 2 Gy, and subjected to ELDA to estimate the SCF, 21 days following treatment. Data are presented as the means ± SEM, n = 3 independent biological experiments. Chi-square test for ELDA plots and one-way ANOVA followed by Tukey’s test for SCF were used. BTSC53 (A): ***p_{control vs mubritinib} = 0.0008, **p_{control vs IR} = 0.0098, ***p_{control vs IR + mubritinib} = 1.5e−5, ***p_{IR vs IR + mubritinib} = 0.0005, **p_{mubritinib vs IR + mubritinib} = 0.0058. BTSC73 (B): ***p_{control vs mubritinib} = 0.0002, **p_{control vs IR} = 0.0036, ***p_{control vs IR + mubritinib} = 1.6e−5, **p_{IR vs IR + mubritinib} = 0.0013, *p_{mubritinib vs IR + mubritinib} = 0.0318. (C, D) BTSC53 (C) and BTSC73 (D) were treated with 20 nM or 500 nM mubritinib, irradiated with 2 Gy or 4 Gy and subjected to live cell counting by PI staining followed by flow cytometry after 7 days of treatment. Data are presented as the means ± SEM, n = 3 independent biological experiments. Two-way ANOVA followed by Tukey’s test. BTSC53 (C): *p_{control vs mubritinib 20} = 0.0117, ***p_{control vs mubritinib 500} = 0.0001, *p_{2Gy vs 2Gy + mubritinib 20} = 0.0117, ***p_{2Gy vs 2Gy + mubritinib 500} = 0.0001, ***p_{4Gy vs 4Gy + mubritinib 500} = 0.0001, **p_{mubritinib 20 vs 2Gy + mubritinib 20} = 0.0029, **p_{mubritinib 500 vs 2Gy + mubritinib 500} = 0.0029, ***p_{mubritinib 20 vs 4Gy + mubritinib 20} = 3.2e−6, ***p_{mubritinib 500 vs 4Gy + mubritinib 500} = 3.2e−6. BTSC73 (D): **p_{control vs mubritinib 20} = 0.0019, ***p_{control vs mubritinib 500} = 4.1e−7, **p_{2Gy vs 2Gy + mubritinib 20} = 0.0019, ***p_{2Gy vs 2Gy + mubritinib 500} = 4.1e−7, **p_{4Gy vs 4Gy + mubritinib 20} = 0.0019, ***p_{4Gy vs 4Gy + mubritinib 500} = 4.1e−7, ***p_{mubritinib 20 vs 2Gy + mubritinib 20} = 6.8e−5, ***p_{mubritinib 500 vs 2Gy + mubritinib 500} = 6.8e−5, ***p_{mubritinib 20 vs 4Gy + mubritinib 20} = 4e−8, ***p_{mubritinib 500 vs 4Gy + mubritinib 500} = 4e−8. (E) Schematic diagram of the experimental procedure in which luciferase-expressing BTSC73 cells were intracranially injected into RAGγ2C^−/− mice. One week after implantation, the mice were randomized into 2 groups: vehicle control or mubritinib (6 mg/kg). Mice were treated 3 times per week (Monday, Wednesday and Friday). For IR groups, on day 12 and 16, mice received 2 Gy. (F, G) Bioluminescence images (F) and quantification of luciferase activity (G) 3 weeks after implantation are presented. Data are presented as box plots showing 25th and 75th percentiles (box), median (centre line), minima and maxima (whiskers), n = 5 mice. One-way ANOVA followed by Tukey’s test. ***p_{vehicle vs mubritinib} = 0.0002, ***p_{vehicle vs IR} = 0.0003, ***p_{vehicle vs mubritinib + IR} = 4.4e−5. (H, I) H&E stainings were performed on day 30 (H). Scale bar = 1 mm. Tumour volume was estimated based on H&E staining (I). Data are presented as the means ± SEM, n ≥ 3 mice. One-way ANOVA followed by Tukey’s test. **p_{vehicle vs mubritinib} = 0.0065, **p_{vehicle vs IR} = 0.0056, ***p_{vehicle vs mubritinib + IR} = 0.0003. (J) KM survival plot was graphed to evaluate mice lifespan in each group, mice were collected at end stage (log-rank test, n = 6 mice). (K) Survival extensions of mice bearing BTSC73-derived tumours treated with mubritinib, IR, or mubritinib + IR relative to those treated with the vehicle control were calculated. Data are presented as the means ± SEM, n = 6 mice. One-way ANOVA followed by Tukey’s test. **p_{mubritinib vs IR} = 0.0047, ***p_{mubritinib vs mubritinib + IR} = 2.1e−5, ***p_{IR vs mubritinib + IR} = 8e−8. (L, M) BTSC53 were intracranially injected into RAGγ2C^−/− mice. KM was plotted to evaluate mice lifespan in each group (log-rank test, n = 5 mice) (L), and survival extensions were calculated (M). Data are presented as the means ± SEM, n = 5 mice. One-way ANOVA followed by Tukey’s test. **p_{mubritinib vs mubritinib + IR} = 0.0032, ***p_{IR vs mubritinib + IR} = 0.0003. (N, O) Representative immunofluorescence images and quantification of phosphorylated histone H2AX (pH2AX) (green) (N) and cleaved caspase 3 (CC3) (green) (O) in BTSC73 intracranial xenografts in mice treated with vehicle control, mubritinib, IR, or mubritinib + IR are shown. Nuclei were stained with DAPI (blue). Quantification was performed with Fiji software. Scale bar = 100 μm. Data are presented as the mean ± SEM, n = 3 mice. One-way ANOVA followed by Tukey’s test. *p_{pH2AX (vehicle vs mubritinib + IR)} = 0.0118, **p_{pH2AX (mubritinib vs mubritinib + IR)} = 0.0084, *p_{pH2AX (IR vs mubritinib + IR)} = 0.0193, *p_{CC3 (vehicle vs mubritinib + IR)} = 0.0125, **p_{CC3 (mubritinib vs mubritinib + IR)} = 0.0062, *p_{CC3 (IR vs mubritinib + IR)} = 0.0229. Source data are available online for this figure.

Figure 6. Mubritinib alleviates tumour hypoxia and enhances oxidative stress to sensitize GB tumours to IR.

(A–D) BTSC73 (A, B) and BTSC53 (C, D) tumourspheres were treated with 500 nM mubritinib under normoxic (21% O₂) or hypoxic (1% O₂) conditions and subjected to labelling with hypoxia probe followed by live confocal imaging (A, C) or flow cytometric analysis (B, D). Representative confocal images of hypoxia probe (green) are shown. Nuclei were stained by Hoechst. Scale bar = 100 μm (A, C). The percentages of high hypoxia probe-positive cells quantified by flow cytometry are shown (B, D). Data are presented as the mean ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. BTSC73 (B): ***$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{c}ontrol1\%{\mathrm{O}}_2}=2.5\mathrm{e}5$, **$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{mubritinib}21\%{\mathrm{O}}_2}=0.0058$, ***$p_{\mathrm{control}1\%{\mathrm{O}}_2\mathrm{v}s\mathrm{mubritinib}1\%{\mathrm{O}}_2}=1.2\mathrm{e}5$. BTSC53 (D): ***$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{c}ontrol1\%{\mathrm{O}}_2}=3.1\mathrm{e}6$, **$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{mubritinib}21\%{\mathrm{O}}_2}=0.0022$, ***$p_{\mathrm{control}1\%{\mathrm{O}}_2\mathrm{v}s\mathrm{mubritinib}1\%{\mathrm{O}}_2}=1.2\mathrm{e}5$. (E) A schematic diagram illustrating the GFP hypoxia reporter is shown. (F, G) GFP hypoxia reporter-expressing BTSC73 tumourspheres were treated with 500 nM mubritinib under normoxic or hypoxic conditions and subjected to live confocal imaging (F) or flow cytometric analysis (G). Representative confocal images of GFP hypoxia reporter (green) are shown. Nuclei were stained by Hoechst. Scale bar = 100 μm (F). The percentages of high GFP hypoxia reporter-positive cells quantified by flow cytometry are shown (G). Data are presented as the mean ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. *$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{c}ontrol1\%{\mathrm{O}}_2}=0.03$, *$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{m}ubritinib21\%{\mathrm{O}}_2}=0.0114$, **$p_{\mathrm{control}1\%{\mathrm{O}}_2\mathrm{v}s\mathrm{m}ubritinib1\%{\mathrm{O}}_2}=0.0043$. (H) A schematic diagram of the in vivo intracranial tumour assay involving the GFP hypoxia reporter-expressing BTSC73 is presented. GFP hypoxia reporter-expressing BTSC73 were intracranially implanted into RAGγ2C^−/−, followed by treatment of mice with vehicle control or mubritinib (6 mg/kg) 3 times per week (Monday, Wednesday and Friday). (I) Representative images and quantifications of GFP staining of sections from GFP hypoxia reporter-expressing BTSC73 xenografts are shown. Scale bar = 1 mm, scale bar inset = 50 µm. The GFP-positive cells were quantified with Fiji software. Data are presented as the mean ± SEM, n = 3 mice. Unpaired two-tailed t test. **p = 0.0088. (J–M) BTSC73 (J, K) or BTSC53 (l, M) tumourspheres were treated with 500 nM mubritinib, IR 2 Gy, or combination of both for 3 days and subjected to labelling with H₂DCFDA and MitoSOX followed by live confocal imaging (J and l) or flow cytometric analysis (K, M). Representative confocal images of the H₂DCFDA (green) and MitoSOX (red) probes are shown. Nuclei were stained by Hoechst. Scale bar = 100 μm (J, l). The percentages of H₂DCFDA or MitoSOX positive cells are shown (K, M). Data are presented as the mean ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. BTSC73 (K): **p_{mitoSOX (control vs mubritinib + IR)} = 0.0013, **p_{mitoSOX (mubritinib vs mubritinib + IR)} = 0.0048, **p_{mitoSOX (IR vs mubritinib + IR)} = 0.0019, ***$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{control}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=3.3\mathrm{e}5$, ***$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{mubritinib}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=0.0002$, ***$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{IR}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=6.8\mathrm{e}5$. BTSC53 (M): **p_{mitoSOX (control vs mubritinib)} = 0.0026, *p_{mitoSOX (control vs IR)} = 0.0221, **p_{mitoSOX (control vs mubritinib + IR)} = 5.6e−5, *p_{mitoSOX (mubritinib vs mubritinib + IR)} = 0.0158, **p_{mitoSOX (IR vs mubritinib + IR)} = 0.0019, ***$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{control}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=0.0003$, **$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{mubritinib}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=0.0059$, **$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{IR}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=0.0032$. (N) Representative images of GB tumour sections from BTSC73 xenografts from mice treated with mubritinib, IR, or combination of mubritinib and IR and subjected to staining using OxyIHC oxidative stress detection kit are shown. Scale bar = 50 µm. OxyIHC was quantified with Fiji software. Data are presented as the mean ± SEM, n = 3 mice. One-way ANOVA followed by Tukey’s test. ***p_{control vs mubritinib + IR} = 0.0001, **p_{mubritinib vs mubritinib + IR} = 0.0011, **p_{IR vs mubritinib + IR} = 0.0025. (O) Schematic illustration showing that mubritinib alleviates hypoxia and enhances ROS generation in response to IR. Source data are available online for this figure.

Figure 7. Mubritinib sensitizes BTSCs and GB tumours to TMZ.

(A, B) BTSC53 (A) and BTSC73 (B) were treated with 20 nM or 500 nM mubritinib, 5 µM TMZ or a combination of mubritinib and TMZ and subjected to live cell counting by PI staining followed by flow cytometry after 7 days of treatment. Data are presented as the mean ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. BTSC53 (A): ***p_{control vs mubritinib 20} = 2.5e−8, ***p_{control vs mubritinib 500} = 9.3e−10, ***p_{control vs TMZ} = 6.73e−9, ***p_{TMZ vs TMZ + mubritinib 20} = 7.6e−5, ***p_{TMZ vs TMZ + mubritinib 500} = 5.9e−6, ***p_{mubritinib 20 vs TMZ + mubritinib 20} = 6.3e−6, ***p_{mubritinib 500 vs TMZ + mubritinib 500} = 0.0004. BTSC73 (B): ***p_{control vs mubritinib 20} = 5.3e−6, ***p_{control vs mubritinib 500} = 1.2e−7, ***p_{control vs TMZ} = 8.6e−6, **p_{TMZ vs TMZ + mubritinib 20} = 0.0010, ***p_{TMZ vs TMZ + mubritinib 500} = 5.1e−5, **p_{mubritinib 20 vs TMZ + mubritinib 20} = 0.0020; *p_{mubritinib 500 vs TMZ + mubritinib 500} = 0.0417. (C) Representative immunofluorescence images and quantification of phosphorylated histone H2A (pH2AX) (green) in BTSC73 treated with vehicle control, mubritinib, TMZ, or mubritinib + TMZ are shown. Nuclei were stained with DAPI (blue). Quantification was performed with Fiji software. Scale bar = 10 μm. Data are presented as the mean ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. *p_{control vs TMZ} = 0.0358, ***p_{control vs mubritinib + TMZ} = 0.0002, ***p_{mubritinib vs mubritinib + TMZ} = 0.0003, **p_{TMZ vs mubritinib + TMZ} = 0.0073. (D–K) Luciferase-expressing BTSC53 (D–G) or BTSC73 (H–K) were intracranially injected into RAGγ2C^−/− mice. The mice were randomized into 4 groups: vehicle control, mubritinib (6 mg/kg 3 times per week), TMZ (1 cycle of 5 days of 1 mg/kg), or combination of mubritinib (6 mg/kg 3 times per week) and TMZ (1 cycle of 5 days of 1 mg/kg). The first mubritinib injection was administered on day 11 after surgery, 24 h prior the start of TMZ treatment. Bioluminescence images (D, H) and quantification of luciferase activity (E, I) 4 weeks after implantation are presented. Data are presented as box plots showing 25th and 75th percentiles (box), median (centre line), minima and maxima (whiskers), n = 5 mice. One-way ANOVA followed by Tukey’s test. BTSC53 (E): **p_{vehicle vs mubritinib} = 0.0011, **p_{vehicle vs TMZ} = 0.0029, ***p_{vehicle vs mubritinib + TMZ} = 0.0006. BTSC73 (I): ***p_{vehicle vs mubritinib} = 1.7e−5, ***p_{vehicle vs TMZ} = 0.0001, ***p_{vehicle vs mubritinib + TMZ} = 1e−5. KM was plotted to evaluate mice lifespan in each group (F and J) (log-rank test n ≥ 4 mice), and survival extensions were calculated (G, K). Data are presented as the mean ± SEM, n ≥ 4 mice. One-way ANOVA followed by Tukey’s test. BTSC53 (G): ***p_{mubritinib vs mubritinib + TMZ} = 2.3e−7, ***p_{TMZ vs mubritinib + TMZ} = 6e−8. BTSC73 (K): *p_{mubritinib vs TMZ} = 0.0461, ***p_{mubritinib vs mubritinib + TMZ} = 9.5e−5, ***p_{TMZ vs mubritinib + TMZ} = 4e−6. Source data are available online for this figure.

Figure 8. Mubritinib does not impact normal non-oncogenic cells and has a safe profile in vivo.

(A) Schematic diagram of the in vitro analysis of the effects of mubritinib on normal cells. (B) Representative phase-contrast images of human neural progenitor cells (hNPCs) treated with 20 nM mubritinib, 500 nM mubritinib, or 10 µM TMZ for 7 days. Scale bars = 100 μm. (C, D) hNPCs were treated with increasing concentration of mubritinib (0–500 nM) or 10 µM TMZ for 7 days followed by live cell counting (C) and Annexin V/PI double staining (D). Representative flow cytometry plots and the percentages of dead cells (PI positive) are shown (D). 1 µM staurosporine (STS) is used as positive control. Data are presented as the mean ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. Live cell number (C): p_{control vs mubritinib 20} = 0.6277, p_{control vs mubritinib 100} = 0.7447, p_{control vs mubritinib 500} = 0.1854, ***p_{control vs TMZ} = 5e−10, ***p_{control vs STS} = 1e−9. Dead cells (D): p_{control vs mubritinib 20} = 0.5555, p_{control vs mubritinib 100} = 0.6174, p_{control vs mubritinib 500} = 0.9989, ***p_{control vs TMZ} = 4.5e−6, ***p_{control vs STS} = 1e−10. (E, F) HepaSH human hepatocytes were treated with mubritinib or the hepatotoxic drug bosentan, as a positive control. Representative phase-contrast images are shown (E). Scale bars = 100 μm. Cytotoxicity was measured by MTT (F). Data are presented as the mean ± SEM, n ≥ 2 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. p_{control vs mubritinib 20} = 0.8229, p_{control vs mubritinib 100} = 0.9219, p_{control vs mubritinib 500} = 0.9933. (G) Cell viability of mouse cortical neurons was measured by MTT after mubritinib treatment. Data are presented as the mean ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. p_{control vs mubritinib 20} = 0.9997, p_{control vs mubritinib 100} = 0.9660, p_{control vs mubritinib 500} = 0.9925. (H, I) Representative immunofluorescence images of SOX2 in the SVZ of mouse brain treated with vehicle control or mubritinib (H). Scale bars = 100 μm. The number of SOX2-positive cells was quantified with Fiji software (I). Data are presented as the mean ± SEM, n = 3 mice. Unpaired two-tailed t test. p = 0.2026. (J) A schematic diagram of the in vivo toxicological studies is presented. Biochemical analysis and treadmill fatigue tests were performed on RAGγ2C^−/− mice treated for 3 months with vehicle control or mubritinib (6 mg/kg) 3 times per week (Monday, Wednesday and Friday). (K) The body weights of the mice are presented. Data are presented as the mean ± SEM, n = 5 mice. Two-way ANOVA followed by Sidak’s test. p = 0.4354. (L–P) Biochemical analyses to evaluate liver (L), kidney (M), pancreatic function (N), tissue damage (O) and lactic acidosis (P) were performed. Data are presented as the mean ± SEM, n = 3 mice. Unpaired two-tailed t test. p_AST = 0.5737, p_ALT = 0.7168, p_ALP = 0.4601, p_{total proteins} = 0.7949, p_albumin = 0.0531, p_glucose = 0.8712, p_creatin = 0.5971, p_{uric acid} = 0.1361, p_urea = 0.9663, p_amylase = 0.7867, p_LDH = 0.6259, p_{lactic acid} = 0.5728. (Q) A treadmill fatigue test was performed. The speed and distances ran by the mubritinib-treated and vehicle control groups were measured. Data are presented as box plots showing 25th and 75th percentiles (box), median (centre line), minima and maxima (whiskers), n = 5 mice. Unpaired two-tailed t test. p_distance = 0.5490, p_speed = 0.4946. Source data are available online for this figure.

Figure EV1. Mubritinib decreases the proliferation of patient- and murine-derived BTSCs without inducing cell death.

(A) Basal mitochondrial respiration, maximal mitochondrial respiration and spare respiratory capacity (SRC) were measured in BTSC73 following mubritinib treatment using the high-resolution respirometer Oroboros. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs control. **p_{basal (20)} = 0.0012; ***p_{basal (100)} = 2.9e−5; ***p_{basal (500)} = 1.1e−5, *p_{maximal (20)} = 0.0228, ***p_{maximal (100)} = 0.0009, ***p_{maximal (500)} = 0.0006, *p_{SRC (100)} = 0.0488, *p_{SRC (500)} = 0.0377. (B) The murine glioblastoma cells, mGB2, were exposed to increasing concentrations of mubritinib (0 to 500 nM), followed by live cell counting. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs control. ***p_{mubritinib 100} = 0.0009, ***p_{mubritinib 500} = 4.4e−5. (C) The percentages of dead cells (PI positive) were measured by PI staining followed by flow cytometry in BTSCs treated with increasing concentrations of mubritinib for 7 days. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs control. p-values are higher than 0.05 for all conditions. (D) The percentages of dead cells (PI positive) and early apoptotic cells (Annexin V positive and PI negative) were measured by Annexin V/PI double staining followed by flow cytometry in BTSCs treated with increasing concentrations of mubritinib for 7 days. (E–G) Pearson correlation analysis was performed between basal OCR and the percentage of live cells in BTSC following 7 days treatment with mubritinib at 20 nM (E) 100 nM (F) and 500 nM (G). (H) Pearson correlation analysis was performed between maximal OCR and sensitivity score to mubritinib following 7 days of treatment. (I–K) mRNA levels of NDI1 gene were assessed by RT-qPCR in BTCS53 (I), BTSC73 (J) and BTSC147 (K) transduced with control or NDI1 vector. Data are presented as the means ± SEM, n = 3 independent biological experiments. (L–N) Basal OCR was measured by Resipher system in BTSC53 (L), BTSC73 (M) and BTSC147 (N) expressing the control (CTL) vector or NDI1 following treatment with vehicle control or 500 nM mubritinib. Data are presented as the means ± SEM, n = 3 independent biological experiments. Two-way ANOVA followed by Tukey’s test. BTSC53 (L): ***p_{CTL vs mubritinib} = 1.8e−12, p_{NDI1 vs NDI1 + mubritinib} = 0.1702. BTSC73 (M): ***p_{CTL vs mubritinib} = 4.9e−13, p_{NDI1 vs NDI1 + mubritinib} = 0.6586. BTSC147 (N): ***p_{CTL vs mubritinib} = 1e−10, p_{NDI1 vs NDI1 + mubritinib} = 0.0828. (O) Proliferation curve was generated using the CFSE assay in BTSC73 expressing the CTL vector or NDI1 following treatment with vehicle control or 500 nM mubritinib. Data are presented as the means ± SEM, n = 3 independent biological experiments. Two-way ANOVA followed by Tukey’s test. *p_{3d (CTL vs mubritinib)} = 0.0267, *p_{4d (CTL vs mubritinib)} = 0.0265, *p_{5d (CTL vs mubritinib)} = 0.0312, **p_{6d (CTL vs mubritinib)} = 0.0011, p_{NDI1 vs NDI1 + mubritinib} > 0.05 at all the time points. (P, Q) mRNA levels of NDUFS7 gene were assessed by RT-qPCR in BTCS73 (P), BTSC53 (Q) electroporated with either siCTL or siNDUFS7. Data are presented as the means ± SEM, n = 3 independent biological experiments. Unpaired two-tailed t test. ***p_BTSC73 = 5e−8, ***p_BTSC53 = 4.8e−6. (R, S) BTSC73 (R) and BTSC53 (S) were electroporated with either siCTL or siNDUFS7 and treated with 500 nM of mubritinib for 7 days, followed by live cell counting. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. BTSC73 (R): ***p_{siCTL vs siCTL + mubritinib} = 1.6e−6, ***p_{siCTL vs siNDUFS7} = 5.3e−6, p_{siNDUFS7 vs siNDUFS7 + mubritinib} = 0.0690. BTSC53 (S): ***p_{siCTL vs siCTL + mubritinib} = 3e−5, ***p_{siCTL vs siNDUFS7} = 6.8e−6, p_{siNDUFS7 vs siNDUFS7 + mubritinib} = 0.5667. (T) BTSCs and differentiated progeny counterparts (Diff) were subjected to immunoblotting using the antibodies indicated on the blots. Vinculin is used as loading control. (U–X) The number of live stem and differentiated cells from BTSC12 (U), BTSC53 (V), BTSC73 (W) and BTSC147 (X) was measured following 500 nM mubritinib treatment. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. BTSC12 (U): ***p_{stem control vs mubritinib} = 0.0002, *p_{diff control vs mubritinib} = 0.0426, **p_{stem mubritinib vs diff mubritinib} = 0.0078. BTSC53 (V): ***p_{stem control vs mubritinib} = 9.4e−5, **p_{stem mubritinib vs diff mubritinib} = 0.0010. BTSC73 (W): ***p_{stem control vs mubritinib} = 0.0001, *p_{diff control vs mubritinib} = 0.0322, **p_{stem mubritinib vs diff mubritinib} = 0.0049. BTSC147 (X): ***p_{stem control vs mubritinib} = 5.1e−5, ***p_{stem mubritinib vs diff mubritinib} = 0.0007.

Figure EV2. Transcriptional classification scores of BTSCs and correlation analysis with sensitivity to mubritinib.

(A, B) Heatmaps illustrating the hierarchical clustering of genes related to Wang (A) and Garofano (B) glioblastoma transcriptional subtypes. Dominant transcriptional subtype score is underlined and highlighted in bold. (C–H) Pearson correlation analysis was performed between the BTSC sensitivity score to mubritinib following 7 days of treatment and mesenchymal (C), classical (D), proneural (E), glycolytic/plurimetabolic (GPM) (F), neuronal (NEU) (G) and proliferative/progenitor (PPR) (H) transcriptional subtype signature scores.

Figure EV3. Validation of the effect of mubritinib on the stemness of murine-derived and additional patient-derived BTSCs.

(A–D) BTSC25 (A), BTSC50 (B), BTSC75 (C) and BTSC100 (D) were subjected to ELDA to estimate the SCF, 21 days following treatment with 500 nM mubritinib or vehicle control. Data are presented as the means ± SEM, n = 3 independent biological experiments. Chi-square test for ELDA plots and unpaired two-tailed t test for SCF were used. *p_BTSC25 = 0.038, *p_BTSC50 = 0.0196, *p_BTSC75 = 0.016, **p_BTSC100 = 0.0022. (E) The murine-derived BTSCs (mGB2) were subjected to ELDA to estimate the SCF, 7 days following treatment with 500 nM mubritinib or vehicle control. Data are presented as the means ± SEM, n = 3 independent biological experiments. Chi-square test for ELDA plots and unpaired two-tailed t test for SCF were used. **p_mGB2 = 0.0018. (F) mGB2 cells were treated for 4 days with 500 nM mubritinib or vehicle control and subjected to immunoblotting using the antibodies indicated on the blots. Vinculin was used as loading control. Densitometric quantifications of Olig2 and SOX2 protein levels normalized to their corresponding loading controls are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Dunnett’s test vs vehicle control. ***p_Olig2 = 0.0003, ***p_SOX2 = 0.0007. (G) BTSC53 transduced with the control (CTL) or NDI1 vector were treated with vehicle control or 500 nM mubritinib and subjected to ELDA to estimate the SCF, 21 days following treatment. Data are presented as the means ± SEM, n = 3 independent biological experiments. Chi-square test for ELDA plots and one-way ANOVA followed by Tukey’s test for SCF were used. **p_{CTL vs mubritinib} = 0.0069, *p_{mubritinib vs NDI1 + mubritinib} = 0.0121, p_{NDI1 vs NDI1 + mubritinib} = 0.8008. (H) BTSC53 transduced with the CTL or NDI1 vector were treated with vehicle control or 500 nM mubritinib for 4 days and subjected to immunoblotting using the antibodies indicated on the blots. Vinculin was used as loading control. Densitometric quantifications of cleaved Notch1, Olig2 and SOX2 protein levels normalized to their corresponding loading controls are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. *p_{Olig2 (CTL vs mubritinib)} = 0.0174, *p_{Olig2 (Mubritinib vs NDI1 + mubritinib)} = 0.0142, p_{Olig2 (NDI1 vs NDI1 + mubritinib)} = 0.9986, ***p_{SOX2 (CTL vs mubritinib)} = 0.0002, ***p_{SOX2 (mubritinib vs NDI1 + mubritinib)} = 0.0001, p_{SOX2 (NDI1 vs NDI1 + mubritinib)} = 0.8669, ***p_{Cleaved Notch1 (CTL vs mubritinib)} = 0.0004, ***p_{Cleaved Notch1 (mubritinib vs NDI1 + mubritinib)} = 0.0004, p_{Cleaved Notch1 (NDI1 vs NDI1 + mubritinib)} > 0.9999.

Figure EV4. Mubritinib sensitizes GB tumours to ionizing radiation.

(A, B) 2 Gy or 4 Gy irradiated BTSC73 (A) BTSC147 (B) were subjected to real-time Resipher analysis to measure the basal OCR 16 h following irradiation. Data are presented as the means ± SEM, n = 3 independent biological experiments. Two-way ANOVA followed by Dunnett’s test vs vehicle control. BTSC73 (A): *p_{control vs 2 Gy} = 0.0357, ***p_{control vs 4 Gy} = 6e−6. BTSC147 (B): ***p_{control vs 2 Gy} = 6.8e−6, ***p_{control vs 4 Gy} = 3.3e−6. (C) Schematic diagram of the experimental procedure in which luciferase-expressing BTSC73 cells were intracranially injected into RAGγ2C^−/− mice. 11 days after implantation, the mice were randomized into 2 groups: vehicle control or mubritinib (6 mg/kg). Mice were treated 3 times per week (Monday, Wednesday and Friday). For IR groups, 24 h after the first mubritinib injection, mice received IR (2 Gy). On day 16, mice were subjected to another cycle of IR at 2 Gy. (D, E) Bioluminescence images (D) and quantification of luciferase activity (E) are presented. Data are presented as box plots showing 25th and 75th percentiles (box), median (centre line), minima and maxima (whiskers), n = 5 mice. One-way ANOVA followed by Tukey’s test. *p_{vehicle vs mubritinib} = 0.0244, *p_{vehicle vs IR} = 0.0258, *p_{vehicle vs mubritinib + IR} = 0.0222. (F) KM survival plot was graphed to evaluate mice lifespan in each group, mice were collected at end stage (log-rank test, n = 5 mice). (G) Survival extensions of mice bearing BTSC73-derived tumours treated with mubritinib, IR, or mubritinib + IR relative to those treated with the vehicle control were calculated. Data are presented as the means ± SEM, n = 5 mice. One-way ANOVA followed by Tukey’s test. **p_{mubritinib vs IR + mubritinib} = 0.0020, ***p_{IR vs IR + mubritinib} = 0.0002. (H, I) Gene set enrichment analysis of deregulated genes in BTSC73 (H) and BTSC147 (I) treated with 500 nM of mubritinib for 24 h demonstrates enrichment of gene sets corresponding to DNA repair pathways. Permutation test was used to calculate p-values, which were then corrected using the Benjamini-Hochberg method to obtain adjusted p-values (p.adjust).

Figure EV5. Mubritinib alleviates tumour hypoxia and enhances oxidative stress to sensitize GB tumours to IR.

(A) BTSC147 tumourspheres were treated with 500 nM mubritinib and subjected to labelling with hypoxia probe (green) followed by live fluorescent imaging. Nuclei were stained by Hoechst. Scale bar = 100 μm. (B) BTSC147 tumourspheres were treated with 500 nM mubritinib under normoxic (21% O₂) or hypoxic (1% O₂) conditions and subjected to labelling with hypoxia probe followed by flow cytometric analysis. The percentages of high hypoxia probe-positive cells are shown. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. ***$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{m}ubritinib21\%{\mathrm{O}}_2}=0.0005$, ***$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{c}ontrol1\%{\mathrm{O}}_2}=7.4\mathrm{e}7$, ***$p_{\mathrm{control}1\%{\mathrm{O}}_2\mathrm{v}s\mathrm{m}ubritinib1\%{\mathrm{O}}_2}=0.0002$. (C–E) BTSC73 (C), BTSC53 (D) and BTSC147 (E) tumourspheres were treated with 500 nM mubritinib under normoxic or hypoxic conditions and subjected to labelling with hypoxia probe followed by flow cytometric analysis. The mean fluorescence of hypoxia probe is shown. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. BTSC73 (C): ***$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{c}ontrol1\%{\mathrm{O}}_2}=2.8\mathrm{e}5$, ***$p_{\mathrm{control}1\%{\mathrm{O}}_2\mathrm{v}s\mathrm{m}ubritinib1\%{\mathrm{O}}_2}=2.2\mathrm{e}5$. BTSC53 (D): ***$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{c}ontrol1\%{\mathrm{O}}_2}=0.0002$, ***$p_{\mathrm{control}1\%{\mathrm{O}}_2\mathrm{v}s\mathrm{m}ubritinib1\%{\mathrm{O}}_2}=0.0002$. BTSC147 (E): *$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{m}ubritinib21\%{\mathrm{O}}_2}=0.0148$, ***$p_{\mathrm{control}21\%{\mathrm{O}}_2\mathrm{v}s\mathrm{c}ontrol1\%{\mathrm{O}}_2}=0.0006$, ***$p_{\mathrm{control}1\%{\mathrm{O}}_2\mathrm{v}s\mathrm{m}ubritinib1\%{\mathrm{O}}_2}=0.0009$. (F, G) BTSC147 (F) and BTSC12 (G) tumourspheres were treated with 500 nM mubritinib, IR 2 Gy, or combination of both for 3 days and subjected to labelling with H₂DCFDA and MitoSOX followed by flow cytometric analysis. The percentages of H₂DCFDA or MitoSOX positive cells are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. BTSC147 (F): **p_{mitoSOX (control vs mubritinib)} = 0.0023, ***p_{mitoSOX (control vs mubritinib + IR)} = 3.1e−5, **p_{mitoSOX (mubritinib vs mubritinib + IR)} = 0.0065, ***p_{mitoSOX (IR vs mubritinib + IR)} = 0.0004. *$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{control}\mathrm{vs}\mathrm{mubritinib}\right)}$  = 0.0287, ***$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{control}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=0.0002$, *$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{mubritinib}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=0.0129$, **$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{IR}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=0.0020$. BTSC12 (G): **p_{mitoSOX (control vs mubritinib)} = 0.005, *p_{mitoSOX (control vs IR)} = 0.0237, ***p_{mitoSOX (control vs mubritinib + IR)} = 0.0002, *p_{mitoSOX (mubritinib vs mubritinib + IR)} = 0.0453, **p_{mitoSOX (IR vs mubritinib + IR)} = 0.0089. **$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{control}\mathrm{vs}\mathrm{mubritinib}\right)}=0.0040$, ***$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{control}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=0.0001$, *$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{mubritinib}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=0.0386$, **$p_{{\mathrm{H}}_2\mathrm{DCFDA}\left(\mathrm{IR}\mathrm{vs}\mathrm{mubritinib}+\mathrm{IR}\right)}=0.0013$. (H, I) BTSC73 (H) and BTSC53 (I) tumourspheres were treated with 500 nM mubritinib, IR 2 Gy, or combination of both for 3 days under hypoxic conditions (1% O₂) and subjected to labelling with MitoSOX followed by flow cytometric analysis. The percentages of MitoSOX positive cells are presented. Data are presented as the means ± SEM, n = 3 independent biological experiments. One-way ANOVA followed by Tukey’s test. BTSC73 (H): ***p_{control vs mubritinib} = 1.7e−5, *p_{control vs IR} = 0.0226, ***p_{control vs mubritinib + IR} = 1.3e−7, ***p_{mubritinib vs mubritinib + IR} = 6.4e−5, ***p_{IR vs mubritinib + IR} = 7.7e−7. BTSC53 (I): ***p_{control vs mubritinib} = 4.5e−6, ***p_{control vs IR} = 0.0006, ***p_{control vs mubritinib + IR} = 6.6e−7, *p_{mubritinib vs mubritinib + IR} = 0.0241, ***p_{IR vs mubritinib + IR} = 3.2e−5. (J) BTSC73 treated with 500 nM mubritinib, irradiated with 2 Gy, and subjected to ELDA, under hypoxic condition (1% O₂), to estimate the SCF. Data are presented as the means ± SEM, n = 3 independent biological experiments. Chi-square test for ELDA plots and one-way ANOVA followed by Tukey’s test for SCF were used. *p_{control vs mubritinib} = 0.0419, ***p_{control vs mubritinib + IR} = 0.0007, *p_{mubritinib vs mubritinib + IR} = 0.0435, *p_{IR vs mubritinib + IR} = 0.0206.