ATM inhibition exploits checkpoint defects and ATM-dependent double strand break repair in TP53-mutant glioblastoma

Abstract

Determining the balance between DNA double strand break repair (DSBR) pathways is essential for understanding treatment response in cancer. We report a method for simultaneously measuring non-homologous end joining (NHEJ), homologous recombination (HR), and microhomology-mediated end joining (MMEJ). Using this method, we show that patient-derived glioblastoma (GBM) samples with acquired temozolomide (TMZ) resistance display elevated HR and MMEJ activity, suggesting that these pathways contribute to treatment resistance. We screen clinically relevant small molecules for DSBR inhibition with the aim of identifying improved GBM combination therapy regimens. We identify the ATM kinase inhibitor, AZD1390, as a potent dual HR/MMEJ inhibitor that suppresses radiation-induced phosphorylation of DSBR proteins, blocks DSB end resection, and enhances the cytotoxic effects of TMZ in treatment-naïve and treatment-resistant GBMs with TP53 mutation. We further show that a combination of G2/M checkpoint deficiency and reliance upon ATM-dependent DSBR renders TP53 mutant GBMs hypersensitive to TMZ/AZD1390 and radiation/AZD1390 combinations. This report identifies ATM-dependent HR and MMEJ as targetable resistance mechanisms in TP53-mutant GBM and establishes an approach for simultaneously measuring multiple DSBR pathways in treatment selection and oncology research.

Fig. 1. MMEJ activity is increased in GBM models with acquired TMZ-resistance.

A Illustration of BFP_MMEJ8 reporter and (B) validation in U251 cells treated with ART558 or POLQ-targeting sRNA (siPOLQ). Reporter expression was calculated as described in methods and normalized to DMSO control for ART558 or non-targeting siRNA (siNT) control for siPOLQ) (C) BFP_MMEJ8 activity in U251 treated with TMZ (48 h), washed, and transfected with BFP_MMEJ8 24 h later. D Clonogenic survival of U251 transfected with siNT or siPOLQ). E Western blot from (D). F–H Measurement of MMEJ, MMR, and MGMT activity in PDX lines. “TMZ” denotes acquired TMZ resistant subline (striped bars) derived from the parental line with the same number. I Relative viability (CellTiter Glo 2.0) of G22 and G22-TMZ treated with TMZ for 120 h with or without ART558 (1 µM). J MMEJ and HR assays in G22 and G22-TMZ treated as indicated. K Clonogenic survival of U251 transfected with indicated siRNA and treated with TMZ (96 h) followed by growth for 7–10 days. Data are presented as the mean of three independent experiments (except for (F) where n = 4), error bars show the standard deviation (SD), and p-values are from statistical comparison by unpaired two-tailed t-test. In (B–D) and (K) comparison is to control (DMSO or siNT) and in (F, G), comparison is to parental PDX line. Source data are provided as a source data file.

Fig. 2. Fluorescence-based multiplexed double strand break repair assays (FM-DSBR) simultaneously measure the activity of three DSBR pathways.

Illustrations of fluorescent reporter plasmids for NHEJ (A), HR (B), and MMEJ (C) and validation in TK6 wild type (WT) and knockout cell lines. D FM-DSBR analysis in U251 treated with ART558 (1 µM), NVB (100 µM), or AZD7648 (1 µM) for 4 h prior to transfection. E Representative flow cytometry plots from (D) indicating reporter expression for the three reporters plus an AmCyan-expressing transfection control plasmid. F FM-DSBR in U251 72 h after siRNA knockdowns (siNT: non-targeting siRNA). G Western blot of cells in (F). The samples derive from the same experiment but different gels for Mre11 and Lig4. These were processed in parallel. H Validation of POLQ knockdown by qRT-PCR using actin as a control. Data are presented as the mean of n = 3 independent experiments except for (B) where n = 4, error bars show SD, and p-values are from comparison to WT by unpaired two-tailed t test. Source data are provided as a source data file.

Fig. 3. TMZ resistance is associated with altered DSBR, especially enhanced HR and MMEJ.

A FM-DSBR analysis of G22 (parental) and G22-TMZ (acquired TMZ resistance) PDX lines. B, C Survival analysis of mice (n = 8 for G22 and n = 10 for G22-TMZ) implanted intracranially with the indicated PDX line and treated with placebo, TMZ, or radiotherapy. D FM-DSBR analysis of G59 and G59-TMZ acquired resistant line. E, F Survival analysis of mice implanted with G59 (n = 9) or G59-TMZ (n = 10) and treated as described for (B, C). G Survival of GBM patients treated with temozolomide and stratified by median expression of the indicated gene using cBioportal (n = 46 per group)^–. In (A, D), data are presented as the mean of three independent experiments, error bars show SD, and p-values are from multiple unpaired two-tailed t-tests with Holm-Šidák correction for multiple testing. For (B, C, E, F) p-values are from Mantel Cox test. In (G) p-values are from Log rank test in cBioPortal and q-values (also from cBioPortal) employ Benjamini Hochberg procedure to correct for false discovery. Source data are provided as a source data file.

Fig. 4. Fluorescence-based multiplexed double strand break repair (FM-DSBR) analysis identifies dual HR/MMEJ inhibitors in GBM.

A–C FM-DSBR in U251 cells treated with indicated inhibitor for 2 h and transfected with BFP_NHEJ, Cherry_HR, GFP_MMEJ6, and pMax Am Cyan plasmid as an internal control for transfection efficiency (n = 3 independent experiments except Buparlisib where n = 2). Cells were analyzed 24 h after transfection by flow cytometry and reporter expression was calculated as previously described and normalized to DMSO control. The target of each inhibitor is displayed above and dashed lines separate columns representing data for different inhibitors. See Fig. S8 for representative flow cytometry data and gating scheme. D FM-DSBR in SF295 under the same conditions as U251. E, F Analysis of HR by DR-GFP reporter and MMEJ by EJ2-GFP reporter in U251 reporter cell lines. Cells were treated with drug, immediately transfected with pCBASceI plasmid and pMax BFP transfection control and analyzed after 72 h. Data are presented as the mean of 3 independent experiments, except for (F) where n = 4 for DMSO, AZD1390, and Panobinostat). Error bars show SD. In (A–C), statistical comparison was to DMSO by one-way ANOVA with Dunnett’s multiple comparisons test. In (E, F) statistical comparison was to DMSO using unpaired two-tailed t-test. *: p < 0.05, **: p < 0.01, ***: p < 0.001. Source data and exact p-values are available in the source data file.

Fig. 5. AZD1390 inhibits HR and MMEJ and potentiates cell killing by DNA-damaging therapy in treatment-naïve and treatment resistant GBM cells.

A, B Clonogenic survival of U251 treated with TMZ or radiation in combination with: DMSO control, AZD1390 (10 nM), Birabresib (50 nM), Panobinostat, abbreviated “Pan” (2 nM), novobiocin (NVB, 25 µM), or BO2 (5 µM). C Analysis of MMR and MGMT activity in U251 treated with 100 nM AZD1390, Birabresib, or Panobinostat for 2 h and transfected with mOrange_G:G (MMR) and mPlum_O⁶-MeG (MGMT) with pMax_BFP transfection control. D Clonogenic survival of U251 transfected with indicated siRNA, treated with DMSO, TMZ, or TMZ/AZD1390 for 96 h. E FM-DSBR of U251 transfected with indicated siRNAs. F Clonogenic survival of T98G treated with AZD1390 (10 nM) in combination with radiation or TMZ (n = 2). G Relative viability of G22 and G22-TMZ treated (120 h) with vehicle (DMSO), AZD1390 (50 nM), TMZ, or combination. H FM-DSBR in G22 or G22-TMZ following 1 h DMSO or AZD1390 (50 nM) treatment. I Relative viability of G43, same as (F). Data are represented as the mean of three independent experiments and error bars show SD. Statistical comparisons are by unpaired two-tailed t-test. In (A, B, G, I), p-values are from comparison to vehicle control. For (A, B), the p-value is displayed for only the high dose. In (G), p-values are below each point and in (I), p-values are above the point. Holm-Šidák correction for multiple testing was applied in (D, E, G–I). Source data are provided as a source data file.

Fig. 6. AZD1390 suppresses DSB end resection and radiation-induced phosphorylation of DSB end protection factors.

A RPA staining in U251 pre-treated with DMSO or AZD1390 (1 h) followed by DMSO, 2 µM camptothecin (CPT), or 10 Gy X-rays and collected 1 h later. B Representative flow cytometry plots from (A). C, D RPA staining data from (A) as a function of cell cycle phase, as determined by DNA content (PI staining). E NHEJ, HR, and MMEJ in asynchronous (Asynch) U251, or U251 arrested with palbociclib (G1), aphidicolin (S), or nocodazole (mitosis, M) for 18 h prior to AZD1390 treatment (100 nM, 1 h) and FM-DSBR. F, G HR and MMEJ activity in U251 transfected with indicated siRNA 72 h prior to treatment with DMSO or AZD1390 (100 nM, 1 h) and transfected with HR/MMEJ reporters. H Western blot for experiments in (F, G). The samples derive from the same experiment; however, the blots for Mre11 and CtIP are from the same gel, while BLM and GAPDH are from another gel. I Volcano plot of phosphorylated peptides detected by LC-MS/MS in SF295 cells treated with DMSO or AZD1390 (100 nM) for 1 h and then treated with 6 Gy (n = 3 biologically independent experiments). The x-axis represents the fold-difference in expression, with higher values representing greater abundance in the DMSO control.  J Selected phosphorylated peptides from (G). In (A, C–G, F), data are presented as the mean of three independent experiments, error bars show SD, and p-values are from unpaired two-tailed t-test. Holm-Šidák correction for multiple testing was applied in (C–G). Source data are provided as a source data file.

Fig. 7. AZD1390 suppresses elevated ATM-dependent HR and MMEJ in TP53-mutant GBM.

A Viability of GBM lines treated with TMZ (25 µM), AZD1390 (10 nM), or both for 6 days and normalized to vehicle (PBS). WT: wild type. B, C Cherry_HR and GFP_MMEJ6 expression (normalized to DMSO control) for cells treated with AZD1390 (1 h, 100 nM) before transfection. D Box and whisker plot showing NHEJ, HR, and MMEJ activity in TP53-WT or TP53-mutant GBMs (n = 6 GBM lines per group). The box displays the 25–75th percentile, the median is indicated by a solid line, and the whiskers (error bars) display the range (min. and max.). E Pathway analysis of genes upregulated in TP53-mutant GBMs. The top ten pathways are ranked by fold enrichment, false discovery rate (FDR) is displayed for each. F Heatmap showing NHEJ, HR, and MMEJ activity in the indicated cell line after pre-irradiation with 4 Gy X-rays. Data are reported as fold-change relative to 0 Gy control according to the color scale at right with the mean (n = 3) displayed in each cell. G mCherry HR and BFP_MMEJ8 activity in G14-GFP or G14-p53DD following pre-irradiation with 0 Gy or 4 Gy. H mCherry_HR and BFP_MMEJ8 activity in G14 cells stably expressing GFP or dominant negative p53DD (residues 300–393). Cells were treated with DMSO or AZD1390 (100 nM) and transfected 1 h later. Data are presented as the mean of three independent experiments, error bars show SD in all panels except for (D) where they show the range. In (A, D, H), p-values are from unpaired two-tailed t-test. In (B, C), p-values are from one-way ANOVA with Dunnett’s test for multiple comparisons. * indicates p < 0.05. Exact p-values for (B, C) and all source data are available in the source data file.

Fig. 8. A dysfunctional G2/M checkpoint renders TP53-mutant GBMs sensitive to ATM inhibition.

A Cell cycle profiles of U251 or A172 treated with DMSO (0.1%), AZD1390 (10 nM), TMZ (10 µM) or TMZ/AZD1390 for 72 h, fixed, and stained with propidium iodide. Experiment was repeated thrice with similar results. B Cell cycle distribution of U251 and A172 after the indicated treatment for 24 h (first four bars) or 48 h (last bar). C Immunostaining of U251 or A172 cells with anti-phospho S10 H3 antibody after indicated treatment in the presence or absence of AZD1390 (10 nM) and (D) Immunostaining with antibody against cleaved PARP (Asp214) 48 h after treatment. Immunostaining of G14-GFP and G14-p53DD cells 24 h after indicated treatment using anti-phospho S10 H3 (E) or anti-cleaved PARP Asp214 (F). G Survival analysis of TP53-WT (n = 99) or TP53-mutant (n = 51) GBM patients stratified by median expression of RHNO1 or HUS1 and compared using cBioportal, p-values from Log rank test with q-values employing Benjamini Hochberg procedure to correct for false discovery. In (B–F), data are presented as the mean of three independent experiments, error bars show SD, and p-values are from unpaired two-tailed t-test with Holm-Sidak correction for multiple comparisons. Source data are provided as a source data file.

Fig. 9. Model for effect of ATMi on GBM cells following formation of a DSB by radiation or TMZ.

Cell cycle progression is shown from left to right. Model created using Biorender.