Suppression of mitochondrial ROS by prohibitin drives glioblastoma progression and therapeutic resistance

Abstract

Low levels of reactive oxygen species (ROS) are crucial for maintaining cancer stem cells (CSCs) and their ability to resist therapy, but the ROS regulatory mechanisms in CSCs remains to be explored. Here, we discover that prohibitin (PHB) specifically regulates mitochondrial ROS production in glioma stem-like cells (GSCs) and facilitates GSC radiotherapeutic resistance. We find that PHB is upregulated in GSCs and is associated with malignant gliomas progression and poor prognosis. PHB binds to peroxiredoxin3 (PRDX3), a mitochondrion-specific peroxidase, and stabilizes PRDX3 protein through the ubiquitin-proteasome pathway. Knockout of PHB dramatically elevates ROS levels, thereby inhibiting GSC self-renewal. Importantly, deletion or pharmacological inhibition of PHB potently slows tumor growth and sensitizes tumors to radiotherapy, thus providing significant survival benefits in GSC-derived orthotopic tumors and glioblastoma patient-derived xenografts. These results reveal a selective role of PHB in mitochondrial ROS regulation in GSCs and suggest that targeting PHB improves radiotherapeutic efficacy in glioblastoma.

Fig. 1. PHB is highly expressed in GSCs.

a–c Immunoblot (IB) showing the expression of indicated proteins in glioma stem-like cells (GSCs) and matched non-stem tumor cells (NSTCs) (a), in GSCs, normal human astrocyte (NHA), and human neural progenitor cells (hNP1, 17231, and 15167) (b), or in GSCs, primary GBM cells and established glioma cell lines (c). d and e Representative immunofluorescent (IF) images of human primary glioblastoma (GBM) specimens stained with anti-PHB (green) and anti-SOX2 (red). Nuclei were counterstained with Hoechst (blue) (d, left). Scale bars, 40 μm. Quantifications of PHB staining intensity in SOX2+ (n = 110) and SOX2− (n = 100) cells (five random microscope fields from three tumors) are shown (d, right). Pearson’s correlation coefficient between PHB and SOX2 staining intensity in GBM cells is shown (e). f mRNA expression of PHB in GSCs (n = 19) relative to bulk tumor cells (BTCs, n = 7) from GEO profile (GSE86237) are shown. g IB showing the levels of PHB and PHB2 in human primary GBM tissues and adjacent normal brain tissues. h Immunohistochemical (IHC) staining of PHB in primary GBM and matched adjacent brain tissue. Scale bars, 100 μm. i–k IHC analysis of PHB in a glioma tissue microarray. Representative images and boxplots of histoscore of PHB in low grade and high grade gliomas are shown (i). Scale bars, 50 μm. (low grade, n = 94; high grade, n = 66). The percentages of recurrence of gliomas in tumors with low (n = 79) or high (n = 81) expression of PHB are shown (j). Kaplan–Meier survival analysis of patients with PHB low (n = 35) and PHB high (n = 31) expression in high-grade gliomas are shown (k). (Log-rank Mantel–Cox test). See Supplementary Table 1. Boxplots represent the median, 25th, and 75th percentiles. The maximum and minimum are connected to the center box through the vertical lines (whiskers). d, f, i Unpaired two-sided Student’s t-test (d, f), Welch’s two-sided t-test (i). Source data are provided as Source Data file.

Fig. 2. PHB promotes GSC self-renewal and tumor progression.

a PHB knockout GSCs were generated by CRISPR-Cas9 system. IB showing levels of indicated proteins in Ctrl and PHB KO GSCs. b GSCs stably transduced with Tet-on-inducible-shPHB were treated with DOX (100 ng/ml) or vehicle control. Expression of indicated proteins were assessed by IB. c, d Quantifications of tumorsphere numbers (2000 cells/well) formed by Ctrl or PHB KO GSCs (c, right), or control or PHB inducible-KD GSCs (d) (mean ± SD, n = 4, biologically independent experiments). Representative images of tumorspheres are shown (c, left). Scale bar, 100 μm. e In vitro extreme limiting dilution assays (ELDAs) show that PHB KO decreased the frequency of tumorsphere formation in GSCs. f Inducible-KD of PHB had limit effects on cell growth of NHA (left) and hNP1 (right), as measured by cell viability assay. g` Ctrl or PHB KO 4121 GSCs (5 × 10⁴/mouse) were implanted into the brains of nude mice (nu/nu, n = 6). Kaplan–Meier survival curve of mice is shown (Log-rank Mantel–Cox test). IB showing the efficiency of PHB KO in xenografts (top). h, i GSCs transduced with Tet-on-inducible-shPHB and Luciferase reporter were implanted into the brains of nude mice (nu/nu). Mice were treated with vehicle control or DOX (2 mg/ml in drinking water) to induce expression of shPHB from day 0. GBM xenografts (4121 GSCs) were tracked by bioluminescence (h, left). Bioluminescent quantification of tumor growth is shown (h, right) (mean ± SEM, n = 6, biologically independent mice). Kaplan–Meier survival curves of mice are shown (i) (4121 GSCs, n = 8; 387 GSCs, n = 7; Log-rank Mantel-Cox test). IB showing the efficiency of PHB knockdown in the xenografts (top). j The in vivo serial transplantation assay shows that PHB KO inhibits GSC self-renewal in vivo. Kaplan–Meier survival curves of mice implanted with indicated GSCs (4121) are shown (top) (n = 5; Log-rank Mantel–Cox test). Summary of mice medium survival in the serial transplantation assay is shown (bottom). k Co-IF staining of PHB (green) and SOX2 (red) in GBM xenografts derived from Ctrl or PHB KO GSCs (4121). Quantifications of SOX2+ cells are shown (right) (mean ± SD, images n = 8, from 4 biologically independent samples). Nuclei were counterstained with Hoechst (blue). Scale bars, 40 μm. Unpaired two-sided Student’s t-test (c–e, h), Welch’s two-sided t-test (k). Source data are provided as Source Data file.

Fig. 3. PHB specific mediates low levels of mitochondrial peroxide in GSCs.

a–e The peroxide levels, as indicated by DCFDA fluorescence, were measured by flow cytometry in PHB KO GSCs (a), PHB inducible-KD GSCs (b), GSCs and matched NSTCs (c), GSCs, hNP1 and NHA (d), or PHB KD NHA and hNP1 (e) (mean ± SD, n = 3, biologically independent experiments). f Ctrl or PHB KO 4121 GSCs were treated with vehicle control or NAC (5 mM) for 36 h. Flow cytometry analysis of peroxide by DCFDA staining is shown (left). IB of PHB, SOX2, and Olig2 levels are shown (middle). Quantifications of tumorsphere numbers (2000 cells/well) formed by GSCs are shown (right) (mean ± SD, n = 3, biologically independent experiments). g Co-IF staining of PHB (green) and 8-OHdG (red) in primary GBM specimens are shown. Quantifications of 8-OHdG staining intensity in PHB− (n = 139) and PHB+ (n = 127) cells are shown (right). (Boxplots represent the median, 25th, and 75th percentiles). Nuclei were counterstained with Hoechst (blue). Scale bars, 40 μm (up), 20 μm (down). h and i RNA-seq analysis in control and PHB KO 4121 GSCs. In h, the heatmap shows relative expression levels of genes downregulated or upregulated in the indicated cells (p < 0.05, FC > 2). It includes, respectively, 185 and 740 genes downregulated and upregulated in PHB KO compared to control GSCs. Raw data were log₂ transformed. A relative color scheme used the minimum and maximum values in each row to convert values to colors. In i, overrepresented Gene Ontology (GO) terms from RNA-seq analysis of upregulated gene sets (top) and downregulated gene sets (bottom) in PHB KO compared to control GSCs. j Quantitative real-time PCR (Q-PCR) analysis of mRNA levels of indicated genes in Ctrl and PHB KO 4121 GSCs (mean ± SD, n = 3, biologically independent experiments). k Q-PCR analysis of mRNA levels of indicated genes in PHB KO (top) and PHB inducible-KD (bottom) 4121 GSCs treated with vehicle control or NAC (5 mM) (mean ± SD, n = 3, biologically independent experiments). Welch’s two-sided t-test (a–c, e–g, j, k), Unpaired two-sided Student’s t-test (f, right). Source data are provided as Source Data file.

Fig. 4. PHB associates with and stabilizes PRDX3 by inhibiting its ubiquitin–proteasome degradation.

a, b IB showing levels of indicated proteins in GSCs with PHB KO (a) or PHB inducible-KD (DOX, 100 ng/ml) (b). c Co-immunoprecipitation (Co-IP) with anti-Flag M2 beads in Flag-vector and Flag-PRDX3 expressing GSCs and IB for PHB, PHB2, and PRDX3 are shown. d Co-IP with anti-PHB (left) or anti-PRDX3 (right) antibody in GSCs and IB for PHB, PHB2, and PRDX3 are shown. Immunoglobulin G (IgG) was used as a control antibody for IPs. e IB showing the CHX (50 μg/ml) chase analysis of PRDX3 protein degradation at indicated time points in GSCs with or without PHB inducible-KD (DOX, 100 ng/ml). Quantifications of relative protein levels of PRDX3 are shown (right) (mean ± SEM, n = 5 (0, 4, 8 h) or n = 3 (12 h), biologically independent experiments, Two-way ANOVA). f IB showing the levels of PRDX3 and PHB in PHB KO (top) or PHB inducible-KD GSCs (bottom) treated with vehicle control or MG132 (10 μM) for 12 h. g Flag-PRDX3 expressing GSCs with or without PHB inducible-KD (DOX, 100 ng/ml) were treated with MG132 (10 μM) for 12 h. Cell lysates were immunoprecipitated with anti-Flag antibody and IB with anti-ubiquitin-Lys48 (Ub-k48) or anti-ubiquitin (Ub) antibody. h IB showing that overexpression of PHB inhibits ubiquitination of PRDX3 in GSCs (4121). i Ectopic expression of Flag-PRDX3 rescued the induction of peroxide levels and the inhibition of cell growth by PHB depletion in GSCs. IB showing the levels of PRDX3 and PHB in 4121 GSCs (left). The peroxide levels, as indicated by DCFDA fluorescence, were measured by flow cytometry (middle, mean ± SD, n = 3, biologically independent experiments, Welch’s two-sided t-test). Cell growth of GSCs was assessed by cell viability assay (right, n = 3, biologically independent experiments, two-way ANOVA). j Q-PCR analysis of mRNA levels of indicated genes in Ctrl or PHB inducible-KD GSCs (4121) with Flag-vector or Flag-PRDX3 overexpression (mean ± SD, n = 3, biologically independent experiments, unpaired two-sided Student’s t-test). k Proposed model for PHB-PRDX3-mediated regulation of mitochondrial ROS homeostasis. In GSCs, PHB is highly expressed, associated with and stabilizes PRDX3 to maintain mitochondrial ROS homeostasis. Loss of PHB increases PRDX3 ubiquitin–proteasome degradation, elevates ROS levels, and subsequently triggers gene expression to induce GSC differentiation and death. Source data are provided as Source Data file.

Fig. 5. PHB promotes GSC radio-resistance.

a–d Nude mice (nu/nu) were intracranially implanted with GSCs (a, b, 4121 GSCs; c, d 387 GSCs) transduced with Luciferase/Tet-on-inducible-shPHB. Mice were randomly grouped (n = 6 for each group) and treated with control, IR (3 Gy, once a week, 4 times), DOX (2 mg/ml in drinking water), or the combined treatment from day 19 (a, b) or day 14 (c, d) after implantation, as shown by schematic representation (a, c top). GBM xenografts were tracked by bioluminescence and the representative images are shown (a, c bottom). Bioluminescent quantifications of tumor growth are shown (a, c right, mean ± SEM, unpaired two-sided Student’s t-test). Kaplan–Meier survival plots of mice are shown (b, d; Log-rank Mantel–Cox test). e IF staining of Tunel (top) or cleaved-caspase3 (bottom) in GBM xenografts from (c) are shown (left). Quantifications of Tunel+ or cleaved-caspase3+ cells are shown (right) (mean ± SD, n = 5, biologically independent samples, Unpaired two-sided Student’s t-test). Scale bars, 40 μm. f, g The peroxide levels, as indicated by DCFDA fluorescence, were measured by flow cytometry in 4121 GSCs with indicated treatments. IR, 3 Gy for 48 h (f) or 72 h (g) (mean ± SD, n = 3, biologically independent experiments, Welch’s two-sided t-test). h Cell apoptosis was measured by flow cytometry in 4121 GSCs with indicated treatments. IR, 3 Gy for 48 h (left) or 72 h (right). i IB showing levels of cleaved-PARP, cleaved-caspaspe3, caspase3, and PHB in GSCs with indicated treatments. IR, 3 Gy for 48 h (left) or 72 h (right). Source data are provided as Source Data file.

Fig. 6. Pharmacological targeting PHB inhibits GSC growth and tumorigenesis.

a The association of RocA and PHB was analyzed by RocA-conjugated sepharose pull-down in GSCs. b Co-IP of PHB in 4121 GSCs treated with RocA (10 nM) and MG132 (10 μM) for 12 h and IB for PHB and PRDX3 are shown. c IB showing the levels of indicated proteins in GSCs treated with increasing dose of RocA for 12 h. d Dose–response curves of RocA treatment in multiple GSC lines, hNP1 and NHA. Cells were treated with increasing dose of RocA for 48 h. IC₅₀ values of RocA were measured using nonlinear regression analysis of dose–response curves (mean ± SD, n = 3, biologically independent experiments). e, g The peroxide levels, as indicated by DCFDA fluorescence, were measured by flow cytometry in Ctrl or PHB inducible-KD 4121 GSCs (DOX, 100 ng/ml) (e), or in Flag-vector or Flag-PRDX3 expressed 4121 GSCs (g) treated with increasing dose of RocA for 18 h (mean ± SD, n = 3, biologically independent experiments, two-way ANOVA). f, h 4121 GSCs cultured with or without 5 mM NAC (f), or expressed with Flag-vector or Flag-PRDX3 (h), were treated with indicated doses of RocA for 3 days. Cell growth was assessed by cell viability assay. Data were normalized to the untreated cells of each group (mean ± SD, n = 4 (f) or n = 3 (h), biologically independent experiments, unpaired two-sided Student’s t-test, no adjustment). i–k Nude mice (nu/nu) intracranially implanted with 4121 GSCs (Luciferase) were randomly grouped (n = 6) at day 16 and treated with or without RocA (2.5 mg/kg, every 3 days, 8 times in total), as shown by schematic representation (i, top). GBM xenografts were tracked by bioluminescence and the representative images are shown (i, left). Bioluminescent quantification of tumor growth is shown (i, right; mean ± SEM, Welch’s two-sided t-test). Kaplan–Meier survival plot of mice is shown (j, Log-rank Mantel–Cox test). IF staining of SOX2 in GBM xenografts with indicated treatments and the representative images are shown (k, left). Quantifications of SOX2+ cells are shown (k, right) (mean ± SD, images n = 10, from five biologically independent samples, Unpaired two-sided Student’s t-test, no adjustment). Nuclei were counterstained with Hoechst (blue). Scale bars, 40 μm. Source data are provided as Source Data file.

Fig. 7. Pharmacological targeting PHB increases sensitiveness of GSCs to IR.

a The peroxide levels, as indicated by DCFDA fluorescence, were measured by flow cytometry in 4121 GSCs with indicated treatment for 48 h. IR, 3 Gy (mean ± SD, n = 3, biologically independent experiments, Welch’s two-sided t-test). b Cell apoptosis was measured by flow cytometry in 4121 GSCs with indicated treatments for 48 h. IR, 3 Gy. c IB showing levels of cleaved-PARP, cleaved-caspaspe3, and caspase3 in 4121 GSCs with indicated treatments for 48 h. IR, 3 Gy. d Cell growth of 4121 GSCs treated with indicated doses of IR for 48 h in the absence or presence of RocA (5 nM). Data were normalized to the untreated cells of each group (mean ± SD, n = 4, biologically independent experiments, unpaired two-sided Student’s t-test). e–g Nude mice (nu/nu) intracranially implanted with GSCs (Luciferase) were randomly grouped (4121 GSCs, n = 6; 387 GSCs, n = 8, for each group) and treated with control, IR (3 Gy, once a week, three times), RocA (2.5 mg/kg, every 3 days, six times), or the combined treatment from day 7 (4121 GSC xenografts) or day 9 (387 GSC xenografts). GBM xenografts were tracked by bioluminescence and the representative images are shown (e). Bioluminescent quantifications of tumor growth are shown (f) (mean ± SEM, Welch’s two-sided t-test). Kaplan–Meier survival plots of mice are shown (g, 4121 GSCs, left; 387 GSCs, right) (Log-rank Mantel–Cox test). h, i NOD/SCID mice were subcutaneously implanted with GBM PDX tumors. Mice were randomly grouped (n = 8 for each group) at day 11 and treated with control, IR (3 Gy, once a week, 4 times), RocA (2.5 mg/kg, every 3 days, eight times in total) or the combined treatment. Tumor volume was measured (h) (mean ± SEM, Two-way ANOVA). Kaplan–Meier survival plot of mice is shown (i). (Log-rank Mantel–Cox test). Source data are provided as Source Data file.