ATM orchestrates the DNA-damage response to counter toxic non-homologous end-joining at broken replication forks

Mutations in the ATM tumor suppressor gene confer hypersensitivity to DNA-damaging chemotherapeutic agents. To explore genetic resistance mechanisms, we performed genome-wide CRISPR-Cas9 screens in cells treated with the DNA topoisomerase I inhibitor topotecan. Thus, we here establish that inactivating terminal components of the non-homologous end-joining (NHEJ) machinery or of the BRCA1-A complex specifically confer topotecan resistance to ATM-deficient cells. We show that hypersensitivity of ATM-mutant cells to topotecan or the poly-(ADP-ribose) polymerase (PARP) inhibitor olaparib reflects delayed engagement of homologous recombination at DNA-replication-fork associated single-ended double-strand breaks (DSBs), allowing some to be subject to toxic NHEJ. Preventing DSB ligation by NHEJ, or enhancing homologous recombination by BRCA1-A complex disruption, suppresses this toxicity, highlighting a crucial role for ATM in preventing toxic LIG4-mediated chromosome fusions. Notably, suppressor mutations in ATM-mutant backgrounds are different to those in BRCA1-mutant scenarios, suggesting new opportunities for patient stratification and additional therapeutic vulnerabilities for clinical exploitation.


Cell lines, culture conditions and treatments
Atm +/+ (wild type; WT) and Atm -/mouse embryonic stem cells (mESCs) were obtained from oocytes of Atm +/mice 4 , and NSG mESCs (Prkdc mutant) were obtained from NSG mice 5 . Unfertilised oocytes generated by superovulation were isolated at E0.5, with cumulus masses digested using hyaluronidase on a stereomicroscope with heat stage set at 37°C. Embryos were washed through 3 drops of M2 medium. For the oocyte activation, 150µl 100mM SrCl2 and 12µl 0.5M EGTA (pH=8) were added to 3ml of KSOM (GSM-5140, AMS Biotechnology). The activation medium was sterile filtered and pre-equilibrated in a 60 x 15mm centre well IVF dish without oil overlay in a humidified tissue culture incubator set at 37°C and 5% CO2. Oocytes were moved via mouth pipetting to the activation medium and incubated for 90 minutes. Oocytes were washed through 3 drops of M2 with any lysed or fragmented embryos removed before returning to pre-equilibrated KSOM to culture. Embryos were checked at day 3, with any 8 cell embryos transferred to KSOM supplemented with CHIR99021 (3µM; Abcam ab120890) and PD0325901 (1µM; Sigma PZ0162) for 24 hours before being returned to KSOM for further culture to blastocyst. Only well expanded blastocysts were selected for denuding. Blastocysts with blastocoels expanding to less than half of the embryo were allowed to culture on until well expanded. Blastocysts were frequently denuded on day 5 or 6 post activation. Zonas were removed using pre-warmed (37°C) acid Tyrode's solution (T1788, Sigma). An expanded blastocyst with large blastocoel would often take under a minute to lose the zona in warm acid Tyrode's. Embryos were moved to M2 and observed after 1 minute. On occasion the zona appears to go in but can be seen again after rehydrating in M2 and so were put back to acid as many times as necessary to fully denude the zona. Embryos were washed through 2 more dishes of M2 medium. Subsequently the parthenogenetic embryos were grown in 10378-016), sodium pyruvate (Gibco;11360-070), beta-mercaptoethanol (Sigma, N3148), non-essential amino acids (Gibco; 11140-035) and LIF 6,7 . Upon the first rounds of passage CHIR and PD were sequentially removed at a 3-passage interval to allow for adaptation. All plates and flasks were gelatinized before cell seeding.
Human immortalized RPE-1 hTERT PuroKO cells (see below) were grown in DMEM F-12 Ham (Sigma) supplemented with glutamine, fetal bovine serum, antibiotics and sodium pyruvate. U2OS cells were grown in DMEM (Sigma) supplemented with fetal bovine serum and antibiotics. HT29 cells were obtained from ECACC (Cat# 91072201). All cells were originally obtained from the ATCC cell repository, and we have authenticated cell lines used in our study by STR profiling, if not otherwise stated.
All cells are routinely tested to be mycoplasma free.
Samples treated for western blotting were irradiated with 10 Gy ionizing radiation (IR) or with addition of 1 µM camptothecin or 1 µM topotecan respectively (CPT; Sigma, TPT; Tocris Bioscience) to the medium. IR treatments were performed using a calibrated RX-650 fitted with a 0.5-mm aluminum filter for soft X rays. ATM inhibitor, ATR inhibitor and DNA PK inhibitor (ATMi; KU55933, Tocris Biosciences, ATRi; AZD6738, AstraZeneca, DNA PKi; NU7441, Tocris Biosciences) was added 1 h before genotoxic treatment, and samples were collected 1 h after application of DNA damaging conditions. siRNA transfection siRNA transfections were performed using Lipofectamine RNAiMAX (Life Technologies). Cells were reverse transfected at a final siRNA concentration of ∼60 nM, transfection was repeated 24 h after the first transfection and cells were assayed 48-72 h after transfection. As a negative control, we used siRNA oligonucleotides targeting Luciferase (siLuc).

Crystal violet sensitivity assays
Cells were seeded at 125 000 cells/well into 24-well plates, and 24 h after plating were treated with the appropriate drug concentration for 5 days, with daily medium and drug replacement. Topotecan and olaparib were from Tocris Bioscience. When IR treatments were performed, cells were seeded at 500 000 cells/well into 6-well plates, irradiated 24 h after plating, and kept growing until cells turned the culture medium yellow for 2 consecutive days. Surviving cells were fixed and stained with crystal violet.

Clonogenic survival assays
The day before treatment, cells were seeded in 6-well plates at 500 or 1000 cells/well, dilutions per dose and three replicates per condition. For inhibition of ATR and DNA PK, ATR inhibitor (AZD6738, 75 nM, AstraZeneca) and DNA PK inhibitor (NU7441, 200 nM, Tocris Biosciences) was added 1 h before genotoxic treatment. Upon treatment with the appropriate drug concentration for 5 -7 days, cells were stained with crystal violet, and the number of colonies per well was counted and normalized to the initial number of cells. For all experiments, data were normalized to the untreated conditions to take into account variations in plating efficiency.

Generation of Cas9 expressing cells
Atm +/+ and -Atm -/-mESCs were transfected with pPB-LR5.1-EF1a-blast2ACas9 8 and the piggyBac transposase vector pCMV-HyPBase 9 using TransIT-LT1 transfection reagent (Mirus) and following manufacturer's instructions (all transfections described in this work were performed using the same reagent). 48 h after transfection, selection was applied with 10 µg/ml blasticidin (Thermo-Fisher; R21009) for 6 days, and resistant colonies were isolated. Cas9 expression was tested by western blotting (WB) using 4-12% Bis-Tris SDS polyacrylamide gels (used for all WB applications in this work). Clones expressing and not expressing Cas9 were tested by transient transfection of pU6-Msh6, a construct produced by cloning of sgRNA DNA sequence templates targeting the mouse Msh6 gene into the pU6-sgRNA plasmid (a gift from W. Skarnes, The Wellcome Trust Sanger Institute, Cambridge, UK). 48 h after transfection, cells were treated with 2 µM 6-thioguanine (6-TG; Sigma) for 5 days, with daily medium and drug replacement. Disruption of Msh6 causes resistance to 6-TG 10 , and was used as surrogate for Cas9 activity. Cells were then allowed to recover with no drug for 5 more days, and survivors were stained with crystal violet. Sequences of all sgRNA templates used in this work are in Supplementary Table 2. Lentivirus production and transduction. Lentiviral production and transduction was as previously described 8  Screening for resistance to topotecan 10 × 10 7 mESCs/genotype were independently infected with the genome-wide gRNA lentiviral library at an MOI of 0.1-0.2, at a library coverage 1000X. Three days after infection, puromycin (10 µM; Gibco A11138-02) was added to the media. Upon established puromycin resistance, cells were partitioned in 3 independent replicates and cultured for 10 additional days. Upon passage a minimum of 50 x 10 6 cells/technical replica was maintained at a library coverage of 500X. For each of the 3 technical replicates, 50 x 10 6 cells were pooled and the representation sample was saved; 11 × 10 7 library-infected mESCs/genotype (in 11 15-cm plates; 1x10 6 cells/plate) and 2 x 10 7 non-library infected mESCs/genotype were treated with topotecan (400 nM for Atm +/+ ; 50 nM for Atm -/-) for 6 days, and further cultured for 4 additional days. Surviving cells were pooled per technical replicate, and genomic DNA was extracted and used for PCR templates. 1 plate per condition was fixed and stained with crystal violet and is represented in Supplementary Figure 1c. The screen was independently repeated and similar results were obtained (data not shown).
HT-29 colorectal cancer cells were infected with lentiviral particles containing the whole-genome sgRNA library, subjected to puromycin selection, and passaged to ensure loss of affected protein products. Puromycin-resistant cells were exposed to 10nM ATMi (AZD0156) and 0.3nM Irinotecan (SN-38) for 21 days, and resistant pools were isolated. Genomic DNA was extracted from these and from parallel cell cultures treated in the absence of topotecan, and DNA libraries were prepared and sequenced.

Illumina sequencing of guide RNAs and statistical analysis
Genomic DNA was extracted and guide RNAs sequenced as described previously 8 .
Single-end Illumina sequencing reads of 19 nucleotides were counted for each gRNA using in-house written software. Depleted or enriched genes in the drug-treated samples were determined from a comparison of read counts with the respective representation sample using the software package MAGeCK 11 version 0.5.3. A gene set enrichment analysis using MAGeCK indicated overrepresented pathways as annotated in the Molecular Signatures Database (MSigDB) 12 Figure S1 and Supplementary Table 2). Transfected populations aimed to produce Atm -/-Lig4 -/and Atm -/-Xrcc4 -/-mESCs were cultured for 5 passages, and then plated into 6-well plates at 500 000 cells/well. 24 h after seeding, cells were treated with 100 nM topotecan for 5 days with daily medium and drug replacement. Cells were then allowed to recover with no drug and surviving colonies were picked into 96-well plates, and expanded for WB testing. One colony from each colonies were transferred to a 96-well plate, triplicated, and genomic DNA was extracted from one of the replicas as described previously 13 . Diagnostic PCR was performed using 1 µl of genomic DNA as template, and PCR products were digested using AfeI to identify edited products (Supplementary Figure S2F). Sequences of diagnostic PCR oligonucleotides used in this work are in Supplementary Table 3.
To produce Lig4, Prkdc, Xrcc4 and Xrcc5 (Ku80) mutants in the WT background,  Table 2) as described above, except that they were tested on WB for absence of protein product directly with no diagnostic PCR step. ATM -/-LIG4 -/cells were produced by transfecting Table 2), running diagnostic PCRs (Supplementary Table 3), and testing selected clones on WB for absence of protein product as described above.

Mass spectrometry analysis
Phosphoproteome analysis was performed as described previously 18  The ten most intense ions were sequentially isolated and fragmented by higher energy In Xlf the (*) symbol denotes possible targeted exon due to incorrect annotation of the Xlf gene in the GRCm38.p4 genome assembly.