ATM regulates 3-methylpurine-DNA glycosylase and promotes therapeutic resistance to alkylating agents

Abstract

Alkylating agents are a first-line therapy for the treatment of several aggressive cancers, including pediatric glioblastoma, a lethal tumor in children. Unfortunately, many tumors are resistant to this therapy. We sought to identify ways of sensitizing tumor cells to alkylating agents while leaving normal cells unharmed, increasing therapeutic response while minimizing toxicity. Using an siRNA screen targeting over 240 DNA damage response genes, we identified novel sensitizers to alkylating agents. In particular, the base excision repair (BER) pathway, including 3-methylpurine-DNA glycosylase (MPG), as well as ataxia telangiectasia mutated (ATM), were identified in our screen. Interestingly, we identified MPG as a direct novel substrate of ATM. ATM-mediated phosphorylation of MPG was required for enhanced MPG function. Importantly, combined inhibition or loss of MPG and ATM resulted in increased alkylating agent-induced cytotoxicity in vitro and prolonged survival in vivo. The discovery of the ATM-MPG axis will lead to improved treatment of alkylating agent-resistant tumors.

Significance: Inhibition of ATM and MPG-mediated BER cooperate to sensitize tumor cells to alkylating agents, impairing tumor growth in vitro and in vivo with no toxicity to normal cells, providing an ideal therapeutic window.

Figure 1. An siRNA screen identifies base excision repair pathway members as sensitizers to TMZ in pediatric GBM

a-b. SJG2 and KNS42 cells were transfected with a 240 DNA siRNA pool library and 24 h post transfections, cells were cultured with or without TMZ (100uM). Cell viability was assessed by the almarBlue cell viability assay 72 h post siRNA transfection. Data were normalized using the standard z-score method by correcting the raw data for plate to plate variation. Significance of potential TMZ sensitizers were determined using z-score cut off values of less than −1.65 (dotted line), which corresponded to a p value of 0.05 in all three biological replicates (rep1-3). c. Venn diagram of genes common and unique to both cell lines from the siRNA screen. d. Heat map of cell viability to validate target genes in TMZ and non-TMZ conditions using SJG2, KNS42, NHAs and normal neural stem cells (NSC). TMZ dose used was 100 uM and viability was assessed using almarBlue assay on day 3. *p<0.05 using ANOVA followed by a post-hoc Dunnett’s test. Each heatmap box represents the average of three independent experiments. e. Immunoblotting to evaluate the protein expression levels of base excision repair pathway members in pediatric glioblastoma cell lines. f. Immunofluorescence of MPG in SJG2 cells showing nuclear expression of the protein. Scale bar = 16um. g. Pearson correlation plots of MPG (black) and MGMT (red) protein expression quantified by chemi luminescence densitometry versus IC50 values following 7 days of TMZ in pediatric GBM cell lines from supplemental figure 1a. Strong correlation is seen with MPG protein levels (r= 0.78 at 7 days) but not with MGMT levels (r= 0.03). All experiments were performed in triplicate, error bars represent standard error of the mean.

Figure 2. Base excision repair genes are up-regulated in pGBM and adult gliomas

a. Immunohistochemistry analysis of 72 pediatric GBM samples demonstrating strong nuclear, cytoplasmic or negative staining. Scale bar =20um. b. Kaplan Meir survival curve analysis of patients staining positive for MPG or negative. Log rank p=0.016. c–h. RNA gene-expression analysis of base excision repair proteins downstream of MPG in pediatric GBM (pGBM, n=53) compared to normal human brain (NB, n=172). ***P<0.001. i. Copy number alterations identify 20/40 BER DNA repair genes are significantly altered in 47 pediatric high grade gliomas. Gains were established as 2.5 more copies of tumor DNA per gene compared to matched normal controls and losses were established as 1.5 copies or less of tumor DNA compared to normal control tissue both corrected at a false discovery rate (q-value) of <5%.

Figure 3. MPG loss sensitizes pGBM cells to alkylating agents

a. Immunoblotting confirming MPG protein knockdown in pediatric GBM cell lines SJG2 and KNS42 following transfection with pooled stable clones generated from two unique MPG shRNA constructs. b. Cell count of pediatric GBM cell lines SJG2 and KNS42 expressing MPG shRNA or shRNA controls (con) exposed to increasing doses of TMZ (0-500 uM). Cell counts were performed 72h post TMZ treatment. c. Immunofluorescence of gamma H2AX in SJG2 cells expressing MPG shRNA or control (con) shRNA. Scale bar =16um. d. Activated Cleaved caspase 3/7 assay of SJG2 and KNS42 cells treated with TMZ post 24h treatment. **p<0.01,***p<0.001 e. Colony forming unit (CFU) assay in SJG2 and KNS42 cells cultured with or without 100 uM TMZ. CFUs were counted after 14 days. **p<0.01 knockdown versus control cells. f. Immunoblotting of MPG and LIG4 following siRNA treatment showing effective protein knockdown. g-h. Plot of cell viability of SJG2 (g) and KNS42 (h) cells transfected with MPG, LIG4 or dual siRNA following exposure to varying concentrations of TMZ. Viability was measured using the almarBlue viability assay and quantified after 7 days. Dual siRNA knockdowns were also compared to single knockdowns to evaluate additive effects of double knockdown, *p<0.05,**p<0.01,***p<0.001.

Figure 4. MPG is a substrate of ATM and phosphorylation of MPG is essential for function

a. Primary amino acid sequence of MPG across species demonstrates that the pSQ ATM substrate residue is evolutionarily conserved. b. Immunoblotting of ATM, phospho-ATM (pATM), ATR and phospho-ATR (pATR) in adult GBM (T98G) and pediatric GBM (SJG2, KNS42 and RES259) cell lines in the presence (+) and absence (-) of TMZ. Cells were treated with TMZ for 48h. c. Immunoblots demonstrating co-immunoprecipitation of ATM and MPG in adult (T98G) and pediatric (SJG2) GBM cell lines. Both proteins are detected following immunoprecipitation (IP) of either ATM or MPG. d. Immunoblots demonstrating detection of MPG following immunoprecipitation with a phospho-(serine/ threonine) (pSQ) ATM/ATR substrate antibody in both adult (T98G) and pediatric (SJG2) GBM cell lines in the absence (-) but not in the presence (+) of an ATM inhibitor (ATM inb, ku55933 used at 5uM). e. Denaturing immunoprecipitation of MPG and immunoblotting of pSQ substrate specific antibody in presence and absence of ATM inhibitor to demonstrate MPG is phosphorylated at the SQ residue. W.C.L=whole cell lysate. ATM inhibitor ku55933 was used at 5uM f. In vitro kinase ATM substrate assay. Immunoblots for thiophosphate ester to detect phosphate incorporation into MPG from a thiol labelled ATP analog in presence of ATM kinase. In vitro kinase assay was performed with wild-type MPG (MPG WT), mutant MPG (MPG S172G) in the presence (+) or absence (-) of cold ATP, ATP_γS, PNBM or with an ATM inhibitor (ku55933, 5uM) demonstrating MPG is directly phosphorylated by ATM kinase at serine 172. g. Fresh lysed whole cell lysates from e, were used for the MPG molecular beacon assay to confirm that inhibition of ATM resulted in reduced MPG glycosylase activity.

Figure 5. Phosphorylation of MPG is required for optimal its function in DNA repair

a. Immunoblot using anti-Flag antibody demonstrating robust expression of MPG flag tagged constructs in pediatric GBM cell line RES259. EV: empty vector, WT: wild type, S172G: MPG mutant in which the serine at residue 172 is replaced by glycine, S172D: MPG mutant in which the serine at residue 172 is replaced by aspartic acid, R182A: MPG mutant in which the arginine at residue 182 is replaced by alanine, B-actin: beta-actin. b. Quantification of molecular beacon MPG activity assay after transfection of empty vector (EV), wild-type (WT) or mutant MPG constructs (R182A, S172G or S172D) in presence of TMZ (100uM). c. Quantification (b) of comet tail assay (a) after transfection of empty vector (EV), wild-type (WT) or mutant MPG constructs (R182A or S172G) in presence (+) or absence (-) of TMZ (100uM). d. Cell viability assay after transfection of wildtype or mutant MPG constructs in the presence or absence of TMZ at 48h. e. Activated cleaved caspase 3/7 assay after transfection of wildtype or mutant MPG constructs in presence of TMZ (100uM). f. Immunoprecipitation of ATM/ATR pSQ substrates in 5 frozen pGBM operative samples. Immuno precipitates were additionally probed for MPG and detected in 3/5 samples. IgG was used as a negative control. g. Immunoprecipitation of MPG is 1% SDS followed by western blotting of pSQ from the 5 samples used in f to demonstrate that MPG is phosphorylated at the pSQ residue in clinical samples. *P<0.05,**P<0.01***P<0.001. All experiments were performed in triplicate with mean and SEM reported where appropriate.

Figure 6. Combined loss of MPG and ATM sensitizes pGBM cells to TMZ in vivo

a. Immunoblotting of SJG2 cells demonstrating effective MPG, ATM or dual knockdown. b. Kaplan Meir survival curve analysis of intracranial injected SJG2 cells expressing control shRNA, MPG, ATM or dual knockdown into NOD-SCID mice. c. Kaplan Meir survival curve analysis of intracranial injected SJG2 cells expressing control shRNA, MPG, ATM or dual knockdown into NOD-SCID mice treated with TMZ (65 mg/kg/5 days). d. Quantification of ki67 staining in mice (n=3 mice per group) from c. e. Kaplan Meir survival curve analysis of intracranial injected SJG2 into NOD-SCID mice treated with vehicle, TMZ (65mg/kg/5days), Methoxyamine (MA 100mg/kg/5days) and both TMZ+MA. Mice were treated upon confirmation of tumour by T2-MRI. f. Quantification of ki67 staining in mice (n=3 mice per group) from e. g. Kaplan Meir survival curve analysis of an orthotopic patient derived xenograft (PDX) mouse model from pGBM 462 cells. 20 mice were injected with 5 per each treatment arm: Vehicle, TMZ (65mg/kg/2 weeks), Methoxyamine (MA 100mg/kg/2 weeks) and both TMZ+MA (65mg/kg TMZ and MA 100mg/kg for 2 weeks). Mice were treated upon confirmation of tumour by T2-MRI. h. Hematoxylin and eosin stain (H&E stain) of represented tumors mice from each treatment arm confirming high grade glioma/GBM morphology. Scale bar, Large insets at 500um and small magnified insets at 25um. N denotes normal mouse brain and T denotes xenograft tumor growth.

Figure 7. Model of ATM-MPG mediated therapeutic resistance in pGBM

a. Western blot analysis demonstrating MPG knockdown by retroviral shRNA transduction and ATM knockdown by siRNA at 72h. Control cells were transduced with shRNA control retrovirus and treated with siRNA scramble controls. b. Control or double MPG and ATM knockdown astrocytes were treated with 100uM TMZ and assessed for viability at 72h. ns = non significance. c. Control or double MPG and ATM knockdown astrocytes were targeted by siRNA for DNA damage response genes redundant to either ATM or MPG and treated with 100uM TMZ and assessed for viability at 72h. All experiments were performed in triplicate with the standard error of the mean reported. d. Summary model of targeting the base excision and ATM pathways for sensitization to alkylating agents.