EGFR promotes ALKBH5 nuclear retention to attenuate N6-methyladenosine and protect against ferroptosis in glioblastoma

Abstract

Growth factor receptors rank among the most important oncogenic pathways, but pharmacologic inhibitors often demonstrate limited benefit as monotherapy. Here, we show that epidermal growth factor receptor (EGFR) signaling repressed N⁶-methyladenosine (m⁶A) levels in glioblastoma stem cells (GSCs), whereas genetic or pharmacologic EGFR targeting elevated m⁶A levels. Activated EGFR induced non-receptor tyrosine kinase SRC to phosphorylate the m⁶A demethylase, AlkB homolog 5 (ALKBH5), thereby inhibiting chromosomal maintenance 1 (CRM1)-mediated nuclear export of ALKBH5 to permit sustained mRNA m⁶A demethylation in the nucleus. ALKBH5 critically regulated ferroptosis through m⁶A modulation and YTH N6-methyladenosine RNA binding protein (YTHDF2)-mediated decay of the glutamate-cysteine ligase modifier subunit (GCLM). Pharmacologic targeting of ALKBH5 augmented the anti-tumor efficacy of EGFR and GCLM inhibitors, supporting an EGFR-ALKBH5-GCLM oncogenic axis. Collectively, EGFR reprograms the epitranscriptomic landscape through nuclear retention of the ALKBH5 demethylase to protect against ferroptosis, offering therapeutic paradigms for the treatment of lethal cancers.

Keywords: ALKBH5; EGFR; GCLM; SRC; YTHDF2; cancer stem cell; ferroptosis; glioblastoma; glioblastoma stem cell; m(6)A.

Figure 1.. EGFR signaling regulates RNA m⁶A levels in glioblastoma

(A) Spatial glioblastoma patient sample (UKF#275) with respect to histology (left) and cell type distribution using the latest reference dataset (~10⁶ cells from 11 datasets) (right). Reference mapping was performed using SpaceXR and SPATA2 software. (B) Examples of spatial gene expression patterns of different GFRs. (C) Example of spatially resolved niches. (D) Spatially weighted correlation of niches and GFRs from 16 de novo glioblastomas. (E) EGF treatment decreases global m⁶A levels in GSCs. Verification of the m⁶A abundance in GSC RNA by dot blot (upper) and RNA level by methylene blue (below). (F) Colorimetric assay for global m⁶A levels treated with or without EGF (two-tailed t-test, ***p < 0.0001, n = 3). (G) EGF treatment decreases global m⁶A levels. Verification of the m⁶A abundance by immunofluorescent staining. (H) Representative LC-MS chromatogram and molecular structure of m⁶A and A. (I) EGF treatment decreases global m⁶A levels in GSCs RNA. Verification of m⁶A abundance by LC-MS (two-tailed t-test, ***p < 0.001, n = 4). See also Figure S1.

Figure 2.. EGF-EGFR signaling regulates ALKBH5 nuclear localization

(A) Immunoblot of m⁶A regulator protein levels in 1919 and MES20 cells after EGF treatment. (B) Immunoblot of m⁶A regulators in GSCs with or without EGFR inhibitor treatment. (C) Immunoblot of m⁶A regulator levels in subcellular fractions of GSCs with or without EGF treatment. (D, E) Subcellular localization of ALKBH5 in GSCs treatment with or without EGF (D) or upon EGFR knockdown (E) by immunofluorescence staining. (F) Immunoblot of ALKBH5 protein localization in GSCs with EGFR modulation. (G) Immunoblot of ALKBH5 protein localization in GSCs treated with or without erlotinib. (H) Cellular localization of ALKBH5 in GSCs with or without EGFRvIII overexpression by immunofluorescence staining. See also Figure S2.

Figure 3.. ALKBH5 Y71 phosphorylation is essential for its nuclear localization and functions in vitro and in vivo

(A-D) Immunoprecipitation of phosphorylated tyrosine in ALKBH5 from GSCs after EGF treatment (A), EGFR overexpression (B), EGFR knockout (C) and treatment with erlotinib (D). (E) EGFR-induced phosphorylated tyrosine residues in ALKBH5 protein. (F) Tyrosine phosphorylation in ALKBH5-depleted GSCs upon exogenous overexpression of ALKBH5 WT or Y71A or Y306A mutation, as determined by immunoprecipitation. (G) Cellular localization of ALKBH5 upon ALKBH5 knockdown in GSCs rescued with WT or mutated ALKBH5 by immunofluorescence staining. (H) ALKBH5 WT, but not the Y71F ALKBH5 mutant, decreases global m⁶A levels in GSCs. Verification of the m⁶A abundance by dot blot (upper) and RNA level by methylene blue (below). (I) Colorimetric assay for global m⁶A level in ALKBH5 knockdown GSCs upon ALKBH5 WT or mutation rescue (one-way ANOVA, **p < 0.01, n = 3). (J) Cell viability of GSCs upon ALKBH5 knockdown, either alone or in combination with ALKBH5 WT or Y71F overexpression. (one-way ANOVA, **p < 0.01, n = 3). (K) Survival curves of immunocompromised mice bearing intracranial xenografts driven from 1919 GSCs transduced with shALKBH5.1323 and ALKBH5 WT or Y71F overexpression. (Log-rank analysis, **p < 0.01). (L) Representative histology images of tumor-bearing brains. Brains were harvested after the presentation of the first neurological sign in any cohort. See also Figure S3.

Figure 4.. GCLM is a specific target of ALKBH5 in GSCs

(A) Overlap of ALKBH5 regulated genes that are highly expressed in GBM in general and GSCs specifically compared with normal brain or NSCs. (B-D) Immunoblotting detects GCLM expression in neural stem cells (B), astrocytes (C), and DGCs (D), and in GSCs. (E) Cell viability assay of GSCs transduced with shRNAs targeting GCLM (one-way ANOVA, **, P < 0.01, n = 3). (F) Survival curves show the time until the onset of neurological signs in intracranial xenografts derived from 1919 and MES20 transduced with two independent non-overlapping shRNAs (shGCLM.856 or shGCLM.937) targeting GCLM or a non-targeting shRNA (shCONT). (Log-rank analysis, *** p < 0.001,). (G) m⁶A abundance on GCLM mRNA in 1919 and MES20 GSCs as quantified by MeRIP-qPCR (one-way ANOVA **P < 0.01, n = 3). (H) Immunoblot of GCLM expression in ALKBH5 knockdown cell lines with ALKBH5 WT or Y71F mutation rescue. (I) GCLM expression in ALKBH5 knockdown GSCs with ALKBH5 WT or mutant Y71F expression, detected by immunofluorescence assay. (J) qPCR analysis of GCLM mRNA levels following ALKBH5. Student’s t-test with Holm-Sidak multiple test correction. ***, p < 0.001. (K) ALKBH5 knockdown promotes GCLM mRNA decay in GSCs. qPCR analysis of mRNA level in GSCs treated with actinomycin D. (two-way ANOVA with Sidak multiple test correction, ***, p < 0.001). (L) The binding of YTHDF2 protein with GCLM RNA was predicted with RBPsuite. (M) RIP-qPCR assay for the enrichment of YTHDF2 in GCLM transcript in GSCs. (One-way ANOVA, ****, p < 0.0001). (N) qPCR analysis of GCLM mRNA level following knockdown with shALKBH5.1323 and treatment with or without YTHDF2 inhibitor (DC-Y13-27). Two-way ANOVA with Sidak multiple test correction. ****, p < 0.0001. (O) Reduced and total GSH concentration in ALKBH5 knockdown GSCs with ALKBH5 WT or mutant Y71F expression (one-way ANOVA, ***P < 0.001, n = 3). (P) Survival curves of immunocompromised mice bearing intracranial xenografts driven from 1919, transduced with shALKBH5.1323 with or without GCLM overexpression. (Log-rank analysis, **, p < 0.01) (Q) Representative histology images of sections of tumor-bearing brains. Tumors were derived from 1919 cells transduced with or without GCLM knockdown. Brains were harvested after the presentation of first neurological sign in any cohort. See also Figure S4.

Figure 5.. ALKBH5 inhibits ferroptosis via GCLM

(A) Morphological assay of ferroptosis in GSCs upon GCLM knockdown. (B) SYTOX staining to detect cell death in GSCs. (C) Quantification of SYTOX positive cells from panel B (one-way ANOVA, ****p < 0.0001, n = 4). (D) Electron microscopy detects mitochondrial morphology in GSC 1919 targeting GCLM. (E) Electron microscopy detects mitochondrial morphology in GSC 1919 targeting ALKBH5. (F) Cell death assay of GSCs transduced with shRNAs targeting ALKBH5 or a shCONT. (G) SYTOX staining to detect cell death in ALKBH5 knockdown GSCs treated with or without reduced GSH (one-way ANOVA, ****p < 0.0001, n = 5). (H) Cell viability assay of ALKBH5 depleted GSCs, treated with or without reduced GSH (one-way ANOVA, ***p < 0.001, n = 3). (I, J) ROS assay of ALKBH5 knockdown GSCs, treated with or without reduced GSH by fluorescence microplate assay (I) and flow cytometry measurement (J) (one-way ANOVA, ***p < 0.001, n = 3). (K) Lipid ROS detection using C11 BODIPY 581/591 (C11) in GSCs. (L) Quantification of the levels of lipid ROS in panel K (one-way ANOVA, ***p < 0.001, n = 6). See also Figure S5.

Figure 6.. Pharmacologic targeting of ALKBH5 augments anti-tumor efficacy of EGFR

(A) Structures of ALKBH5i1, ALKBH5i2, and erlotinib. (B) Role of ALKBH5i on global m⁶A levels in GSCs. Verification of the m⁶A abundance by Colorimetric assay (two-tailed t-test, ***p < 0.001, **p < 0.01, n = 3). (C) Calculation and visualization of synergy scores for drug combinations of ALKBH5i and erlotinib. (D) In vivo bioluminescent imaging of tumors from respective experimental groups. (E) Survival curves of tumor-bearing mice from orthotopic intracranial xenograft implantation of 1919, treated with PBS, ALKBH5i1 (50 mg/kg), Erlotinib (20 mg/kg), or their combined treatment. (Log-rank test, **p < 0.01) (F) Survival curves of tumor-bearing mice from orthotopic intracranial xenograft derived from 1919 GSCs, treated with PBS, ALKBH5i2 (25 mg/kg), and erlotnib (20 mg/kg) or their combined treatment. (Log-rank test, **p < 0.01). (G) Survival curves of tumor-bearing mice from orthotopic intracranial xenograft implantation of MES20, treated with PBS, ALKBH5i1 (50 mg/kg), EGFR inhibitor (50 mg/kg), or their combined treatment. (Log-rank test, **p < 0.01) (H-I) AST activity (H) and ALT (I) in mice with erlotinib (50 mg/ kg) or ALKBH5i1 treatment (50 mg/kg). (J) Histological analysis of liver, lung, and kidney of mice with erlotinib (50 mg/kg) or ALKBH5i1 (50 mg/kg) treatment. (K, L) Survival analysis of patient cohorts stratified into high vs. low expression levels of EGFR, ALKBH5 transcriptional score with the median as the cutoff in TCGA (K) and CGGA (L) GBM-LGG RNA-seqV2 dataset. P-values were calculated with log-rank test. See also Figure S6.

Figure 7.. Targeting GCLM induces ferroptosis and generates an anti-tumor efficacy in GSCs

(A) Molecular structure of BSO. (B) Viability assay of paired GSCs and DGCs treated with the GCLM inhibitor. (C, D) Synergy score of GSC 1919 treated with ALKBH5i1 (C) or ALKBH5i2 (D) and BSO. (E) Luciferase image of tumors from respective experimental groups. (F) Survival curves of tumor-bearing mice from orthotopic intracranial xenograft implantation of 1919, treated with ALKBH5i1 (25 mg/kg), BSO (50 mg/kg), or their combined treatment. (Log-rank test, **p < 0.01) (G) Survival curves of tumor-bearing mice from orthotopic intracranial xenograft implantation of 1919, treated with ALKBH5i2 (25 mg/kg), BSO (50 mg/kg), or their combined treatment. (Log-rank test, **, p < 0.01). (H) Survival curves of tumor-bearing mice from orthotopic intracranial xenograft implantation of MES20, treated ALKBH5i1 (50 mg/kg), BSO (50 mg/kg), or their combined treatment. (Log-rank test, **p < 0.01) (I) Expression of EGFR and GCLM at single cell level. (J, K) Spatially weighted correlation analysis of EGFR and GSH metabolism in glioblastoma specimens (one-way ANOVA, **p < 0.01, ***p < 0.001, n = 16). See also Figure S7.