Advertisement

Functional analysis of ATM variants in a high risk cohort provides insight into missing heritability

  • Scott L. Baughan
    Affiliations
    Center for Molecular Medicine and Genetics, Wayne State University School of Medicine, Detroit, MI, United States
    Search for articles by this author
  • Fatima Darwiche
    Affiliations
    Center for Molecular Medicine and Genetics, Wayne State University School of Medicine, Detroit, MI, United States
    Search for articles by this author
  • Michael A. Tainsky
    Correspondence
    Corresponding author.
    Affiliations
    Center for Molecular Medicine and Genetics, Wayne State University School of Medicine, Detroit, MI, United States

    Department of Oncology, Wayne State University School of Medicine, Detroit, MI, United States

    Molecular Therapeutics Program, Karmanos Cancer Institute at Wayne State University School of Medicine, Detroit, MI, United States
    Search for articles by this author
Open AccessPublished:March 19, 2022DOI:https://doi.org/10.1016/j.cancergen.2022.03.003

      Highlights

      • Extensive in vitro analysis of several variants of unknown significance in ATM reveals several likely pathogenic variants.
      • Results of in vitro assays differ from predictions made via in silico algorithms.
      • Provides a method for identification and assessment of rare SNPs in hereditary cancer syndromes.
      • Wet lab functional assays provide useful data in determining individual SNP pathogenicity.

      Abstract

      Variants of unknown significance (VUS) remain a constant challenge in the diagnosis of hereditary cancer and the counseling of patients with pedigrees suggestive of such a syndrome. In order to assess some of this limitation, several variants in the DNA repair gene ATM were selected from a cohort of high risk individuals with negative genetic diagnoses. ATM has proven a challenge in the counseling of patients due to its nature as a moderate penetrance gene. In this study, six ATM missense mutations with a high likelihood for pathogenicity were assessed through a battery of experiments to yield high fidelity information on their biochemical effect on ATM activity. We report that several of these variants show signs of reduced ATM function indicative of likely pathogenicity. With further study, this data may be used in clinic, improving diagnosis, surveillance, and outcome for patients carrying these mutations.

      Keywords

      Introduction

      In 2021, an estimated 21,410 individuals will be diagnosed with ovarian cancer in the USA. In addition, 13,700 patients with ovarian cancer will die from the disease [
      • Siegel R.L.
      • Miller K.D.
      • Fuchs H.E.
      • Jemal A.
      Cancer Statistics, 2021.
      ]. The five year survival rate for ovarian cancer is currently 47%, a number reflecting high survival rates for early stage disease (up to 92% for stage 1A and 1B). However, unlike breast cancer, few ovarian cancers are diagnosed at early stage. Only 15% of patients present early, and the survival rates decline rapidly as the disease becomes more advanced [
      • Nash Z.
      • Menon U.
      Ovarian cancer screening: current status and future directions.
      ,
      • Domchek S.M.
      • Robson M.E.
      Update on Genetic Testing in Gynecologic Cancer.
      ]. About 60% of patients present with advanced disease and metastatic cancer for which the five year survival rate is below 30% [
      • Domchek S.M.
      • Robson M.E.
      Update on Genetic Testing in Gynecologic Cancer.
      ].
      The pattern of disease presentation can be addressed by early surveillance and prophylactic surgery for at risk patients but is not recommended for the general population, or even many patients at moderate risk [
      • Nash Z.
      • Menon U.
      Ovarian cancer screening: current status and future directions.
      ,
      • Domchek S.M.
      • Robson M.E.
      Update on Genetic Testing in Gynecologic Cancer.
      ]. Cancer genetic testing can play a key role in identifying patients for whom early surveillance and enhanced testing can be beneficial. This, in turn can drastically increase the survival of the most vulnerable patients by enhancing cancer detection at early stages when it is most treatable [
      • Fostira F.
      • Papadimitriou M.
      • Papadimitriou C.
      Current practices on genetic testing in ovarian cancer.
      ,
      • Afghahi A.
      • Kurian A.W.
      The Changing Landscape of Genetic Testing for Inherited Breast Cancer Predisposition.
      ].
      The diagnostic yield of testing for pathogenic mutations using gene panels has vastly increased over single gene testing [
      • Minion L.E.
      • Dolinsky J.S.
      • Chase D.M.
      • Dunlop C.L.
      • Chao E.C.
      • Monk B.J.
      Hereditary predisposition to ovarian cancer, looking beyond BRCA1/BRCA2.
      ,
      • Hirotsu Y.
      • Nakagomi H.
      • Sakamoto I.
      • Amemiya K.
      • Oyama T.
      • Mochizuki H.
      • et al.
      Multigene panel analysis identified germline mutations of DNA repair genes in breast and ovarian cancer.
      ,
      • Kraus C.
      • Hoyer J.
      • Vasileiou G.
      • Wunderle M.
      • Lux M.P.
      • Fasching P.A.
      • et al.
      Gene panel sequencing in familial breast/ovarian cancer patients identifies multiple novel mutations also in genes others than BRCA1/2.
      ,
      • Kapoor N.S.
      • Curcio L.D.
      • Blakemore C.A.
      • Bremner A.K.
      • McFarland R.E.
      • West J.G.
      • et al.
      Multigene Panel Testing Detects Equal Rates of Pathogenic BRCA1/2 Mutations and has a Higher Diagnostic Yield Compared to Limited BRCA1/2 Analysis Alone in Patients at Risk for Hereditary Breast Cancer.
      ]. Panel testing has been able to identify pathogenic mutations in 17% of BRCA1/2 negative high risk families and 24% of unselected ovarian cancer patients [
      • Bunnell A.E.
      • Garby C.A.
      • Pearson E.J.
      • Walker S.A.
      • Panos L.E.
      • Blum J.L.
      The Clinical Utility of Next Generation Sequencing Results in a Community-Based Hereditary Cancer Risk Program.
      ], and it is successful and cost effective enough that such testing has been proposed to benefit the general population regardless of family history [
      • Manchanda R.
      • Patel S.
      • Gordeev V.S.
      • Antoniou A.C.
      • Smith S.
      • Lee A.
      • et al.
      Cost-effectiveness of Population-Based BRCA1, BRCA2, RAD51C, RAD51D, BRIP1, PALB2 Mutation Testing in Unselected General Population Women.
      ]. There is growing evidence that such panel testing should be more routinely used for hereditary breast and ovarian cancer (HBOC) [
      • Graffeo R.
      • Livraghi L.
      • Pagani O.
      • Goldhirsch A.
      • Partridge A.H.
      • Garber J.E.
      Time to incorporate germline multigene panel testing into breast and ovarian cancer patient care.
      ,
      • Desmond A.
      • Kurian A.W.
      • Gabree M.
      • Mills M.A.
      • Anderson M.J.
      • Kobayashi Y.
      • et al.
      Clinical Actionability of Multigene Panel Testing for Hereditary Breast and Ovarian Cancer Risk Assessment.
      ]. The results of gene panel testing, when unambiguous, identify at risk individuals and change their clinical management, enabling increased surveillance [
      • Frost A.S.
      • Toaff M.
      • Biagi T.
      • Stark E.
      • McHenry A.
      • Kaltman R.
      Effects of Cancer Genetic Panel Testing on at-Risk Individuals.
      ]. Regardless of the results, care of patients with the disease can be affected for the better [
      • Knabben L.
      • Imboden S.
      • Mueller M.D.
      Genetic testing in ovarian cancer - clinical impact and current practices.
      ]. Despite this success, at least 49% of the heritable component of hereditary breast and ovarian cancer remains unknown, with many seemingly familial cases testing negative with the current panels [

      Genetics, A., Hereditary Cancer panels: white Paper 2018. 2018.

      ]. Failure to offer panel testing has led to the underdiagnoses of HBOC in Medicare and Medicaid patients, for whom insurance does not cover most testing panels and methods [
      • Yang S.
      • Axilbund J.E.
      • O'Leary E.
      • Michalski S.T.
      • Evans R.
      • Lincoln S.E.
      • et al.
      Underdiagnosis of Hereditary Breast and Ovarian Cancer in Medicare Patients: genetic Testing Criteria Miss the Mark.
      ,
      • Copur M.S.
      Universal Genetic Testing for All Breast Cancer Patients.
      ]. In addition, current guidelines may miss at least half of patients with actionable mutations [
      • Beitsch P.D.
      • Whitworth P.W.
      • Hughes K.
      • Patel R.
      • Rosen B.
      • Compagnoni G.
      • et al.
      Underdiagnosis of Hereditary Breast Cancer: are Genetic Testing Guidelines a Tool or an Obstacle?.
      ]. Overall, there is an increasing trend toward the discovery of new cancer predisposing mutations, and the inclusion of these on testing panels would further increase the panel's utility in the clinic [
      • Kraus C.
      • Hoyer J.
      • Vasileiou G.
      • Wunderle M.
      • Lux M.P.
      • Fasching P.A.
      • et al.
      Gene panel sequencing in familial breast/ovarian cancer patients identifies multiple novel mutations also in genes others than BRCA1/2.
      ,
      • Stafford J.L.
      • Dyson G.
      • Levin N.K.
      • Chaudhry S.
      • Rosati R.
      • Kalpage H.
      • et al.
      Reanalysis of BRCA1/2 negative high risk ovarian cancer patients reveals novel germline risk loci and insights into missing heritability.
      ].
      The protein kinase ATM plays an essential role in the repair of double strand breaks, a mechanism often associated with HBOC. Along with the related protein, ATR, ATM is a key activator of the DNA damage response, and the cellular response to other types of stress, including oxidative stress and reactive oxygen species, replication fork stress, telomere dysfunction, hypoxia, and hyperthermia, phosphorylating over 150 target proteins leading to widespread pathway activation, chromatin relaxation, and nucleosome remodeling [
      • Awasthi P.
      • Foiani M.
      • Kumar A.
      ATM and ATR signaling at a glance.
      ,
      • Blackford A.N.
      • Jackson S.P.
      ATM, ATR, and DNA-PK: the Trinity at the Heart of the DNA Damage Response.
      ,
      • Marechal A.
      • Zou L.
      DNA damage sensing by the ATM and ATR kinases.
      ]. Its numerous targets include Chk2, MRE11, Rad9, Rad50, p53, NBS1, DNA-Pks, and CtIP [
      • Awasthi P.
      • Foiani M.
      • Kumar A.
      ATM and ATR signaling at a glance.
      ,
      • Blackford A.N.
      • Jackson S.P.
      ATM, ATR, and DNA-PK: the Trinity at the Heart of the DNA Damage Response.
      ,
      • Choi M.
      • Kipps T.
      • Kurzrock R.
      ATM Mutations in Cancer: therapeutic Implications.
      ]. Upon DNA damage, auto-phosphorylation of ATM on S1981 in the FAT domain causes the inactive multimer form to disassociate, and ATM is stabilized in its active monomeric form [
      • Awasthi P.
      • Foiani M.
      • Kumar A.
      ATM and ATR signaling at a glance.
      ,
      • Blackford A.N.
      • Jackson S.P.
      ATM, ATR, and DNA-PK: the Trinity at the Heart of the DNA Damage Response.
      ,
      • Paull T.T.
      Mechanisms of ATM Activation.
      ]. Additional phosphorylation occurs on S367 and S1893 along with acetylation of K3016 to further stabilize active ATM, which then interacts with its numerous targets [
      • Awasthi P.
      • Foiani M.
      • Kumar A.
      ATM and ATR signaling at a glance.
      ,
      • Paull T.T.
      Mechanisms of ATM Activation.
      ].
      ATM mutations have long been implicated in the development of breast cancer, conferring a 60% risk by age 80 to individuals carrying germline mutations [
      • Graffeo R.
      • Livraghi L.
      • Pagani O.
      • Goldhirsch A.
      • Partridge A.H.
      • Garber J.E.
      Time to incorporate germline multigene panel testing into breast and ovarian cancer patient care.
      ]. Damaging ATM mutations are present in the heterozygous state in up to 2% of the general population, conferring a 2–4 fold risk increase for general cancers, and a 5–9 fold risk increase for cancers in women [
      • Choi M.
      • Kipps T.
      • Kurzrock R.
      ATM Mutations in Cancer: therapeutic Implications.
      ,
      • Jerzak K.J.
      • Mancuso T.
      • Eisen A.
      Ataxia-telangiectasia gene (ATM) mutation heterozygosity in breast cancer: a narrative review.
      ]. ATM is hypothesized to be a significant component of genetic risk specifically for hereditary breast and ovarian cancer syndrome (HBOC), and mutations have been found in many families with HBOC even as it presents a challenge to genetic counselors as a moderate penetrance gene [
      • Domchek S.M.
      • Robson M.E.
      Update on Genetic Testing in Gynecologic Cancer.
      ,
      • Stafford J.L.
      • Dyson G.
      • Levin N.K.
      • Chaudhry S.
      • Rosati R.
      • Kalpage H.
      • et al.
      Reanalysis of BRCA1/2 negative high risk ovarian cancer patients reveals novel germline risk loci and insights into missing heritability.
      ,
      • Choi M.
      • Kipps T.
      • Kurzrock R.
      ATM Mutations in Cancer: therapeutic Implications.
      ,
      • Jerzak K.J.
      • Mancuso T.
      • Eisen A.
      Ataxia-telangiectasia gene (ATM) mutation heterozygosity in breast cancer: a narrative review.
      ,
      • Southey M.C.
      • Goldgar D.E.
      • Winqvist R.
      • Pylkas K.
      • Couch F.
      • Tischkowitz M.
      • et al.
      PALB2, CHEK2 and ATM rare variants and cancer risk: data from COGS.
      ,
      • Tecza K.
      • Pamula-Pilat J.
      • Kolosza Z.
      • Radlak N.
      • Grzybowska E.
      Genetic polymorphisms and gene-dosage effect in ovarian cancer risk and response to paclitaxel/cisplatin chemotherapy.
      ,
      • Thorstenson Y.R.
      • Roxas A.
      • Kroiss R.
      • Jenkins M.A.
      • Yu K.M.
      • Bachrich T.
      • et al.
      Contributions of ATM mutations to familial breast and ovarian cancer.
      ,
      • DeLeonardis K.
      • Sedgwick K.
      • Voznesensky O.
      • Matloff E.
      • Hofstatter E.
      • Balk S.
      • et al.
      Challenges in Interpreting Germline Mutations in BARD1 and ATM in Breast and Ovarian Cancer Patients.
      ].  The increasing trend of ATM mutation discovery in HBOC is strongly suggestive of a role in the development of cancer, and individual ATM mutations may confer moderate risk for these cancers. These women, therefore, undergo enhanced screening, and they may be more responsive to PARP inhibitors for cancer treatment [
      • Fostira F.
      • Papadimitriou M.
      • Papadimitriou C.
      Current practices on genetic testing in ovarian cancer.
      ,
      • Jerzak K.J.
      • Mancuso T.
      • Eisen A.
      Ataxia-telangiectasia gene (ATM) mutation heterozygosity in breast cancer: a narrative review.
      ].
      In a previous study [
      • Stafford J.L.
      • Dyson G.
      • Levin N.K.
      • Chaudhry S.
      • Rosati R.
      • Kalpage H.
      • et al.
      Reanalysis of BRCA1/2 negative high risk ovarian cancer patients reveals novel germline risk loci and insights into missing heritability.
      ], we examined, using whole exome sequencing, a cohort of 48 women with a personal history of ovarian cancer and a significant family history of cancer but who were negative for pathogenic mutations in BRCA1 and BRCA2. Hypothesizing that variants of unknown significance in genes related to DNA repair and cell cycle control may account for their missing heritability, we found several single nucleotide polymorphisms (SNPs) classified as variants of unknown significance with a moderate to high predicted impact on function in our cohort. In addition, five families were re-contacted for clinical genetic testing and counseling based on new relevant mutations detected by our study. The majority of the SNPs in our cohort were found to be low frequency mutations of uncertain significance and poorly annotated in both literature and clinical databases [
      • Stafford J.L.
      • Dyson G.
      • Levin N.K.
      • Chaudhry S.
      • Rosati R.
      • Kalpage H.
      • et al.
      Reanalysis of BRCA1/2 negative high risk ovarian cancer patients reveals novel germline risk loci and insights into missing heritability.
      ]. In accordance with our hypothesis and data, others have found that retesting patients who were previously negative when assessed for BRCA1/2 specific mutations yields important information about risk, pointing to the further relevance of panel testing in the clinic [
      • Crawford B.
      • Adams S.B.
      • Sittler T.
      • van den Akker J.
      • Chan S.
      • Leitner O.
      • et al.
      Multi-gene panel testing for hereditary cancer predisposition in unsolved high-risk breast and ovarian cancer patients.
      ].
      Because of the likelihood of hereditary cancer in our patient cohort, we hypothesized that these SNPs could represent pathogenic variants. In our cohort, several missense mutations in ATM were found to be significantly overrepresented (Table 1) in addition to two rare truncations in ATM [
      • Stafford J.L.
      • Dyson G.
      • Levin N.K.
      • Chaudhry S.
      • Rosati R.
      • Kalpage H.
      • et al.
      Reanalysis of BRCA1/2 negative high risk ovarian cancer patients reveals novel germline risk loci and insights into missing heritability.
      ]. All mutations identified by NGS whole exome sequencing were validated by Sanger sequencing. Here, we report the in-depth functional analysis of these variants. The analysis of these variants demonstrates that by combining functional studies with small curated cohorts, our process can facilitate the discovery of variants likely responsible for missing heritability.
      Table 1Variants of unknown significance in ATM from a cohort of high risk ovarian cancer. Variants are listed in amino acid order from N to C-Terminus with rsID. All variants have a low minor allele frequency (MAF) and are listed in ClinVar as conflicting interpretations of pathogenicity. REVEL scores are shown, which combine several variant classification algorithms into one higher fidelity score [
      • Ioannidis N.M.
      • Rothstein J.H.
      • Pejaver V.
      • Middha S.
      • McDonnell S.K.
      • Baheti S.
      • et al.
      REVEL: an Ensemble Method for Predicting the Pathogenicity of Rare Missense Variants.
      ,
      • Tian Y.
      • Pesaran T.
      • Chamberlin A.
      • Fenwick R.B.
      • Li S.
      • Gau C.-.L.
      • et al.
      REVEL and BayesDel outperform other in silico meta-predictors for clinical variant classification.
      ]. REVEL scores are closer to 0 than 1, indicating that most algorithms predict that these variants are unlikely to be pathogenic. Scores from the EVE algorithm
      • Frazer J.
      • Notin P.
      • Dias M.
      • Gomez A.
      • Min J.K.
      • Brock K.
      • et al.
      Disease variant prediction with deep generative models of evolutionary data.
      are mostly intermediate, with lower scores predicting a less deleterious effect. .
      AAVariantProtein DomaindbSNP IDMAF (Exac)MAF(gnomAD)ClinVarREVELEVE
      S49Cc.146C>GSubstrate Bindingrs18000540.0110.0071Conflicting0.060.384
      S333Fc.998C>TNuclear Localizationrs289049190.0040.0015Conflicting0.1020.437
      F1463Cc.4388T>GNBN Bindingrs1383274060.0020.0014Conflicting0.2510.809
      D1853Vc.5558A>TNBN Bindingrs18016730.0060.0048Conflicting0.3440.475
      L2307Fc.6919C>TFAT Domainrs560098890.00190.0012Conflicting0.1890.628
      V2540Ic.7618G>API3K Domainrs352032000.000050.00004Conflicting0.3150.102

      Materials and methods

      Cohort and IRB: The sample consists of DNA from 48 BRCA1/2 mutation negative Caucasian OVCA patients from 47 families (one mother-daughter pair). Informed consent was signed and permission was obtained for the collection of blood samples and for access to medical records for all subjects. The protocol (HIC#024199MP2F(5R)) was approved following Full Board Review by the Human Investigation Committee at Wayne State University, Detroit, Michigan.
      ATM variants: pcDNA3.1(+)Flag-His-ATM WT was a gift from Michael Kastan (Addgene plasmid # 31,985; http://n2t.net/addgene:31985 ; RRID:Addgene_31,985) [
      • Canman C.E.
      • Lim D.S.
      • Cimprich K.A.
      • Taya Y.
      • Tamai K.
      • Sakaguchi K.
      • et al.
      Activation of the ATM kinase by ionizing radiation and phosphorylation of p53.
      ]. ATM variants were constructed using the NEB Q5 site directed mutagenesis kit. Following completion of the PCR, ATM variants were transformed into STBL2 Max Efficiency competent cells (Invitrogen), and grown for 2 days on LB agar + 100 µg/mL Carbenicillin at 30°C. Small colonies were selected and transferred to liquid culture in LB broth + 100 µg/mL Carbenicillin, and grown for 2 days at 30°C. Presence of the desired variants were confirmed by Sanger sequencing (Genewiz), and clones carrying the correct variant sequence were then assessed by full plasmid Sanger sequencing via several primers resulting in overlapping 500 bp high quality regions. Clones with no other alterations except the VUS were used for subsequent experiments. ATM-S1981A was expressed from hATMS1981A, which was a gift from Michael Kastan (Addgene plasmid # 32,300 ; http://n2t.net/addgene:32300 ; RRID:Addgene_32,300) [
      • Bakkenist C.J.
      • Kastan M.B.
      DNA damage activates ATM through intermolecular autophosphorylation and dimer dissociation.
      ]. These plasmids are available upon request from Addgene.
      Cell Culture: Hela-DRGFP cells were used as described previously [
      • Lopes J.L.
      • Chaudhry S.
      • Lopes G.S.
      • Levin N.K.
      • Tainsky M.A.
      FANCM, RAD1, CHEK1 and TP53I3 act as BRCA-like tumor suppressors and are mutated in hereditary ovarian cancer.
      ,
      • Chaudhry S.R.
      • Lopes J.
      • Levin N.K.
      • Kalpage H.
      • Tainsky M.A.
      Germline mutations in apoptosis pathway genes in ovarian cancer; the functional role of a TP53I3 (PIG3) variant in ROS production and DNA repair.
      ]. The same cell line was used for all experiments. Quality control for mycoplasma was completed once every 6 months throughout the experimental workflow. Cell line authentication was performed by the Biobanking and Correlative Sciences Core Lab of the Karmanos Cancer Institute.
      Lentiviral shRNA knockdown: ATM knockdown was accomplished using shRNA expressed from Lentiviral vectors targeting the 3′UTR (Sigma-Aldrich, TRCN0000010299, ready to use lentiviral particles). Hela-DRGFP cells were transduced with lentivirus at a MOI of 2 on the day of plating for experiments. Freshly thawed lentiviral supernatants were combined with counted cells in suspension and then plated as one unit. Knockdown efficacy of ∼.6 was confirmed by Western Blot. Verification of ATM knockdown and ATM VUS models is shown in Supplemental Figure S-1 and Fig. 1d-E.
      Fig. 1:
      Fig. 1ATM variants in HDR analysis. A: Diagram of HDR analysis in Hela-DRGFP. Treatment with iSce1 creates a double strand break, which, if repaired by HDR, results in the expression of GFP from a now functional allele. B: Comparison of GFP expression after pCMV3 (empty vector) and pCBASce1 transfection in Hela DRGFP. C: Six ATM variants were assessed for fluorescence after treatment with iSce1. S1891A is a known pathogenic variant lacking kinase activity. pATM-WT expressed wild-type ATM and was used as a positive control. Sce1 only shows baseline Hela-DRGFP fluorescence and is used for calibration of WT-ATM rescue. The group pCMV3 shows baseline Hela fluorescence without iSce1 treatment. All conditions receiving ATM shRNA (shATM) also receive iSce1. Assessment is completed using FACS (BD-LSR II). Statistics use alpha = 0.05, 1 tailed t-test, assuming unequal variances, comparisons made to pATM WT. D: Relative expression of ATM as a ratio to vinculin assessed via western blot and normalized to mock transfection with pCMV3 (loading control).  E. Western Blots showing ATM expression and Vinculin.
      Transfections: Plasmid transfections were carried out using JetPrime (Polyplus transfection) according to manufacturer's specification. For a 12-well plate, 70,000 Hela-DRGFP cells were seeded 1 day prior to transfection and checked for appropriate patterning before transfection. Transfections were carried out using 1 µg of plasmid (pCMV3 for control, or pATM-Flag-His (variant or WT)), 100 µL JetPrime buffer, and 4 µL of JetPrime per well. For HDR assays, an additional 500 ng or iSce1 or pCMV3 was included. This formula was scaled for kinase assay, colony survival assay, and protein harvesting for western blotting.
      Western Blotting: For ATM, protein lysates were collected on ice using RIPA buffer with 1.5% PITC and 1% SDS. Samples were incubated in RIPA 15 min on ice before shearing with a 21 ga. needle. Following shearing, samples were incubated on ice for 1 hour, then centrifuged at 15,000 rpm at 4°C for 15 min. Protein concentration was assessed via Bradford assay. Two hundred µg of each sample were loaded for gel electrophoresis, which was carried out using 6% polyacrylamide gels at 120 V x 20 min + 220 V x 45 min. Transfer was accomplished in Bjerrum Schafer-Neilsen transfer buffer to nitrocellulose at 180 mA for 18 hrs in cold room. Membranes were air dried for 15 min at room temperature to fix protein, then blocked in 5% milk-TBST overnight at 4°C. After 3 × 10 min wash in TBST, ATM primary antibody (1:500 ATM-AM9, Millipore-Sigma, 05–513) in 5% BSA-TBST was used, and incubated overnight at 4°C. Following another 3 × 10 min TBST wash, anti-mouse secondary (AF 680, Abcam, ab175774) was incubated for 1 hour at room temperature in 5% BSA-TBST. After a final 3 × 10 min TBST wash, imaging was carried out using a Li-cor Odyssey imager using Odyssey software (Li-Cor) for visualization and quantification. Vinculin was used as a loading control and was probed after ATM imaging to avoid oversaturation. The membrane was incubated with vinculin primary antibody (Invitrogen, 700,062) for 2 h at room temperature in 5% BSA-TBST, washed as before, then incubated for 1 hour at room temperature with anti-rabbit IgG secondary antibody (AF-790 DxR, Invitrogen, A11374) in BSA-TBST, washed again, and imaged on the Licor. p-CHK2-T68, p-p53-S15, and p-H2AX-S139 were probed concurrently. For these proteins, 100 ug of lysate was loaded onto a 12% polyacrylamide gel, then transferred to nitrocellulose membranes. After 15 min air drying, the membranes were blocked in 5% BSA-TBST for 1 hour, then probed for 2 h in 5%BSA-TBST with three antibodies (P-CHK2 Thr68 [Cell Signaling, 2661S] at 1:1000, P-P53 Ser15 [Cell Signaling, 9284S] at 1:1000 and P-H2AX Ser139 [Cell Signaling, 2577S] at 1:800) for 2 h at room temperature. After wash, anti-rabbit IgG secondary was used to probe for 1 hour at 1:10,000 in 5%BSA-TBST. The membranes were then washed and visualized on a Li-Cor Odyssey imager. The membranes were then stripped and re-blocked overnight at 4°C in 5% BSA-TBST and probed for loading control (beta actin, (anti-ACTB, Sigma, HPA041264) as before. pBRCA1 was blotted from a 6% gel run and transferred as ATM above. The membrane was blocked overnight at 4°C in 5% BSA-TBST, then probed in 5% BSA-TBST using pBRCA1-S1387 (Bethyl, A300–007A) at 1:1000 and vinculin at 1:10,000 at 4°C overnight. Following wash, the membrane was incubated with the anti-rabbit IgG secondary antibody for 1 hour at room temperature, washed again, and imaged on the Li-Cor Odyssey as above. p-p53-S20 and p21 were probed using a 12% gel as above for p-CHK2, p-p53-S15, and p-H2AX. The membrane was incubated with anti-p-p53-S20 (Cell Signaling, 9287S) in 5% BSA-TBST overnight at 4°C after blocking, then washed, probed with anti-rabbit secondary antibody and imaged. The membrane was then stripped, re-blocked for 1 hour using BSA-TBST, and probed for actin and p21 (Cell signaling, 2947S) for 2 h at room temperature, after which it was washed, probed with anti-rabbit IgG secondary antibody, washed again, and imaged as before.
      CDC25A-S178 and p-CDC25C-S216 were probed on separate 12% gels as above. Each gel was incubated separately with either CDC25A-S178 antibody (Bioss, bs-3095R) or CDC25C-S216 antibody (Cell Signaling, 2577S) for 2 h at room temperature after blocking in BSA-TBST. Anti-rabbit IgG secondary antibody was used for both, and actin was used as a loading control.
      Kinase assay: Hela cells were transfected with all ATM variants and controls as described above. Three days post transfection, 1 plate of cells for each group was treated with 2.5 µM Bleomycin (Selleck Chemical), with a duplicate kept as a control. After 4 h incubation, protein was purified from both plates and used for western blotting as described above.
      HDR assay: To obtain relative GFP expression in experimental groups, Hela-DRGFP cells were plated on day 1 at 70k/well in a 12 well plate, and transduced with lentiviral shRNA as described above. On day 2, transfection was performed using JetPrime as described above to introduce pCMV3, iSce1, and pATM variants. On days 3 and 4, the medium in the wells was changed to remove dead cells. Finally, on day 5 the cells were harvested in PBS for use in Flow Cytometry. Flow cytometry was performed using the Wayne State University Flow Cytometry Core as described before [
      • Lopes J.L.
      • Chaudhry S.
      • Lopes G.S.
      • Levin N.K.
      • Tainsky M.A.
      FANCM, RAD1, CHEK1 and TP53I3 act as BRCA-like tumor suppressors and are mutated in hereditary ovarian cancer.
      ], except using a BD-LSR instrument. Gates were used to select for singlets, GFP+ (FITC), and DAPI+ (live) cells.
      Colony Survival Assay: Hela Cells were transfected as described above. Two days post transfection, cells were re-suspended and counted, and then 300 cells were seeded onto each well of 6 well plates. A total of 5 concentrations with three replicates each were plated for each condition. One day post plating, cells were treated with bleomycin (0–10 µM) for 4 h. Media was then replaced with standard culture media and the colonies were allowed to grow for 7 days. After 7 days, colonies were fixed and stained using crystal violet. Counting was performed with a dissection miscrscope. Colonies of at least 20 cells were scored, each as 1 colony. IC50s and 95% confidence intervals were calculated using GraphPad Prism. Survival was calculated after controlling for plating efficiency.

      Results

      Assessment of ATM variants in homology directed repair

      ATM variants were initially assessed via homology directed repair (HDR) assay to provide a global view of decreased function in HDR. To model the ATM VUS, Hela DRGFP cells were transduced with lentiviral shRNA targeting the 3′UTR of ATM to perform knockdown. ATM cDNA from plasmids expressing either wild-type ATM or one of the six VUS was then used to restore ATM expression to previous levels (Fig. 1). An empty vector control was included to visualize HDR activity in unaltered Hela cells and to verify restoration of ATM signaling to wild-type levels. Transfection with iSce1 was performed to cut the DRGFP signaling plasmid was done, and the relative HDR activity of the models was then assessed after three days by flow cytometry using GFP expression as a marker for successful HDR. A negative control with no iSce1 was included and used to correct for any background HDR activity at the DRGFP allele. Restoration of complete HDR function was observed by the transfected pATM-WT plasmid in comparison to the unaltered control (pCMV3). The ATM knockdown control (shATM+ pCMV3), modeling ATM dysfunction, showed marked loss of HDR activity (p = 0.00067 vs. pATM-WT). Several of the variants we tested exhibited defects in HDR function compared to the wild type plasmid, with two variants exhibiting HDR function equivalent to the known pathogenic control allele, S1981A (p = 0.00078 vs. pATM-WT). S49C showed a significant decrease in HDR function (p = 0.03), but this variant likely retains some HDR function (see discussion below). S333F and D1853V were similar in exhibiting a significant decrease in HDR activity (p = 0.007 and p=  0.0009, respectively), but retain some HDR function. Some mechanism conferred by these mutations likely reduces the efficiency of ATM's kinase activity. F1463C was interesting in its appearance in our previous study in three patients from unrelated families. Thus, it was a high priority for analysis. This variant possessed lower HDR activity than the known pathogenic control, suggesting that it causes a near complete loss of function to the affected allele (p = 0.00003 vs. pATM-WT). Similarly, V2540I exhibited a near complete loss of HDR activity in this assay (p = 0.000014 vs. pATM-WT)). Because they showed such a large decrease, these two VUS are of prime interest as potential pathogenic variants Finally, L2307F showed complete retention of function, with levels of HDR activity similar to the wild-type control allele (p = 0.18). These data strongly suggest that there is severe loss of ATM function for several of the variants present in our high risk patient cohort.

      Effect of atm variants on colony formation

      Successful repair of DNA damage increases the cell's ability to survive mutagenic stimuli, including DNA damaging drugs commonly employed in chemotherapy. We used the radiomimetic bleomycin [
      • Povirk L.F.
      DNA damage and mutagenesis by radiomimetic DNA-cleaving agents: bleomycin, neocarzinostatin and other enediynes.
      ,
      • Bolzán A.D.
      • Bianchi M.S.
      DNA and chromosome damage induced by bleomycin in mammalian cells: an update.
      ] to assess cell viability in our ATM variants, as ATM deficiency sensitizes cells to the effects of ionizing radiation [
      • Bakkenist C.J.
      • Kastan M.B.
      DNA damage activates ATM through intermolecular autophosphorylation and dimer dissociation.
      ,
      • Hammond E.M.
      • Muschel R.J.
      Radiation and ATM inhibition: the heart of the matter.
      ,
      • Momcilović O.
      • Choi S.
      • Varum S.
      • Bakkenist C.
      • Schatten G.
      • Navara C.
      Ionizing radiation induces ataxia telangiectasia mutated-dependent checkpoint signaling and G(2) but not G(1) cell cycle arrest in pluripotent human embryonic stem cells.
      ,
      • Pandita T.K.
      • Lieberman H.B.
      • Lim D.S.
      • Dhar S.
      • Zheng W.
      • Taya Y.
      • et al.
      Ionizing radiation activates the ATM kinase throughout the cell cycle.
      ] (Fig. 2). To ensure comparable data, the same cell model system was used as in the HDR assay, but without transfection of the iSce1 plasmid to ensure that double strand breaks were strictly proportional to the amount of bleomycin used. As before [
      • Lopes J.L.
      • Chaudhry S.
      • Lopes G.S.
      • Levin N.K.
      • Tainsky M.A.
      FANCM, RAD1, CHEK1 and TP53I3 act as BRCA-like tumor suppressors and are mutated in hereditary ovarian cancer.
      ], the resulting variant models were first assessed for plating efficiency, a baseline measure of their ability to form a colony from a single cell. Plating efficiency can be reduced in cells with dysfunction of critical genes, indicating relative fragility. The assessment of plating efficiency is also used to control for the effect of the drug treatments on each variant model. Here, the ATM knockdown cells and the cells expressing the kinase deficient control variant S1981A showed significant reduction in their viability after plating but without any treatment with bleomycin (p = 0.0003 and p = 0.0004, respectively). It is noteworthy that several of the variants showed significant reductions in viability after plating when compared to the ATM wild type model, although all of these were similar or above the plating efficiency of the empty vector control transfected cells (F1463C p = 0.01, D1853V  p = 0.025, L2307F p = 0.003, V2540I p = 0.004). Bleomycin IC50 was assessed from the colony survival assay data, and none of the VUS were found to differ significantly from the wild type model in this metric, although the knockdown model did exhibit significantly reduced IC50 (p= <0.05, IC50 = 1.11 [95% CI 0.82 – 1.45] vs. IC50 3.51 [95% Ci 2.34–5.5] for ATM-WT). Because there was widespread cell death at higher concentrations, we considered whether the surviving percentage at the mid-point concentrations of bleomycin (1.5 and 2.5 µM, before high lethality is uniformly present) might be more illuminating for specific defects, even if the overall IC50 did not vary significantly between variants. In this analysis, it was found that at both concentrations of bleomycin, the ATM knockdown model showed significantly reduced survival compared to the wild type model, which in turn did not significantly vary from the empty vector control, indicating that the overall experiment was sensitive enough to detect a defect in ATM signaling, and that the wild type plasmid did fully restore ATM function, rendering analysis of these concentrations useful. At 1.5 µM bleomycin, the variants V2540I and F1463C showed significantly reduced survival (p = 0.046, and p = 0.033, 47% and 46% survival relative to plating, respectively). Similar early reduced viability was present in the ATM knockdown cells and those expressing S1981A (p = 0.02 and p = 0.008, 37% and 32% survival relative to plating, respectively). The reduced viability of the F1463C was even more apparent at the higher concentration of 2.5 µM bleomycin (p = 0.002 vs. ATM-WT, 13% survival relative to plating) again comparable to the ATM knockdown cells (p = 0.003 vs. ATM-WT, 11% survival relative to plating). These results are notable in light of the severe reduction in HDR activity seen for these two variants. The fact that they are also the least likely to recover from more mild doses of bleomycin compared to the ATM WT control indicates that these ATM variants may be unable to repair minimal DNA damage as efficiently, thus cells carrying them may be more susceptible to accumulating unrepaired mutations as a result of fewer double strand breaks. Such inability to repair and recover from smaller amounts of DNA damage could cause the elevated genetic risk seen in HBOC families.
      Fig. 2:
      Fig. 2Colony Survival Analysis of ATM Variants. A: Plating efficiency as a percent of the control group, pCMV3, for each variant model. B. IC50 for each variant model calculated using GraphPad Prism Software using the [inhibitor] vs. normalized response (variable slope) model. Significance calculated by GraphPad for 95% confidence. C-D: Percentage survival for each model group at the specified concentration of Bleomycin, as expressed as a percentage of plating efficiency. For all panels, N = 3, statistics use alpha = 0.05, 1 tailed t-test assuming unequal variance, comparisons made to pATM-WT. .

      Assessment of kinase targets for ATM variants

      The major biochemical function of ATM in the DNA damage response is as a protein kinase. Upon activation, ATM phosphorylates and modulates the activity of over 150 unique targets, many with redundant activation from different pathways [
      • Awasthi P.
      • Foiani M.
      • Kumar A.
      ATM and ATR signaling at a glance.
      ,
      • Blackford A.N.
      • Jackson S.P.
      ATM, ATR, and DNA-PK: the Trinity at the Heart of the DNA Damage Response.
      ,
      • Paull T.T.
      Mechanisms of ATM Activation.
      ,
      • Smith J.
      • Tho L.M.
      • Xu N.
      • Gillespie D.A.
      The ATM-Chk2 and ATR-Chk1 pathways in DNA damage signaling and cancer.
      , ]. In order to investigate a potential mechanism by which our variants of interest may disrupt the DNA damage signaling cascade, we analyzed eight such targets. These targets were BRCA1 at S1387 [
      • Ouchi T.
      BRCA1 phosphorylation: biological consequences.
      ], H2AX at S139 [
      • Burma S.
      • Chen B.P.
      • Murphy M.
      • Kurimasa A.
      • Chen D.J.
      ATM phosphorylates histone H2AX in response to DNA double-strand breaks.
      ],CHK2 at T68 [
      • Xu X.
      • Tsvetkov L.M.
      • Stern D.F.
      Chk2 activation and phosphorylation-dependent oligomerization.
      ,
      • Matsuoka S.
      • Rotman G.
      • Ogawa A.
      • Shiloh Y.
      • Tamai K.
      • Elledge S.J.
      Ataxia telangiectasia-mutated phosphorylates Chk2 in vivo and in vitro.
      ], p53 at S15 [
      • Zhao H.
      • Traganos F.
      • Darzynkiewicz Z.
      Phosphorylation of p53 on Ser15 during cell cycle caused by Topo I and Topo II inhibitors in relation to ATM and Chk2 activation.
      ,
      • Banin S.
      • Moyal L.
      • Shieh S.
      • Taya Y.
      • Anderson C.W.
      • Chessa L.
      • et al.
      Enhanced phosphorylation of p53 by ATM in response to DNA damage.
      ] and S20 [
      • Shieh S.Y.
      • Ahn J.
      • Tamai K.
      • Taya Y.
      • Prives C.
      The human homologs of checkpoint kinases Chk1 and Cds1 (Chk2) phosphorylate p53 at multiple DNA damage-inducible sites.
      ], CDC25A at S178, and CDC25C at S216 [
      • Matsuoka S.
      • Huang M.
      • Elledge S.J.
      Linkage of ATM to cell cycle regulation by the Chk2 protein kinase.
      ], and the transcriptional activation of CDKN1a, which encodes the p53 induced protein cip1/waf1/p21 (p21) [
      • Georgakilas A.G.
      • Martin O.A.
      • Bonner W.M.
      p21: a Two-Faced Genome Guardian.
      ] (Fig. 3).
      Fig. 3:
      Fig. 3Analysis of the Kinase Activity of ATM Variants. A: Representative western blots of all phosphorylation targets analyzed and loading controls. Blots are grouped by membrane, with loading controls present for each separate membrane-target.  B: Graph of relative induction for each target. Significance uses alpha =0.05, 1 tailed t-test assuming unequal variants. .
      Fig. 3:
      Fig. 3Analysis of the Kinase Activity of ATM Variants. A: Representative western blots of all phosphorylation targets analyzed and loading controls. Blots are grouped by membrane, with loading controls present for each separate membrane-target.  B: Graph of relative induction for each target. Significance uses alpha =0.05, 1 tailed t-test assuming unequal variants. .
      Anova analysis of our measurements of phosphorylation of BRCA1-S1387 showed significant differences between variants (Single Factor Anova, p = 0.01, F = 3.22, Fcrit=2.39). Compared to the ATM-WT control, both ATM knockdown and ATM-S1981A exhibited markedly reduced phosphorylation in response to DNA damage (p = 0.046 and p = 0.015, respectively). Replacement with the VUS V2540I showed a reduction in induced phosphorylation similar to the negative controls (p = 0.037). Surprisingly, the L2307F variant showed a notable and significant increase in phosphorylation of BRCA1-S1387 (p = 0.021). Visible but non-significant increases were also observed for cells expressing the variants S49C and F1463C (p = 0.26 and p = 0.15, respectively). This target on the BRCA1 protein is shared with two other kinases, and the resultant increase may reflect compensatory activation of those kinases (ATR and DNA-PKcs) in response to changes in signaling by this variant. Unfortunately, we were unable to acquire antibodies for the more ATM specific residues in BRCA1, S1423 and S1524 [
      • Ouchi T.
      BRCA1 phosphorylation: biological consequences.
      ].
      Induction of p-H2AX-s139 was significantly reduced in the ATM knockdown cells, and those expressing the S1981A and S333F variants (p = 0.04, p = 0.035, and p = 0.027, respectively) when compared directly to ATM-WT. However, these comparisons risk error in that the overall dataset didn't register as significant by Single Factor Anova (p = 0.17).
      Analysis of p-p53-S15, likewise, did not register as significant overall by Anova (p = 0.056). However, with the much narrower gap to significance, we think that the significant pairwise comparisons hold more weight. For this target, ATM knockdown, S1981A, and V2540I were significant (p = 0.038, p = 0.032, and p = 0.032, respectively). Additionally, phosphorylation at this site was notably reduced for S49C (p = 0.051), and D1853V (p = 0.07).
      A major phosphorylation target for ATM is CHK2-T68. We tested p-CHK2-T68 for our variants, which showed significant differences among variants in activation (Single Factor ANOVA, p = 0.000799, F = 543, Fcrit =2.39). The ATM knockdown cells, and those expressing S1981A and S333F variants showed significant decreases in induction (p = 0.022, 0.026, and 0.041 respectively). Additionally, compared to ATM-WT, phosphorylation by the variant ATM S49C was decreased but not significantly (p = 0.057) as was phosphorylation by the variant V2540I (p = 0.056).
      No differences were detected among the experimental conditions in p-p53-S20 induction (Single Factor ANOVA, p = 0.72, F = 0.67, Fcrit=2.39), including the variants affected for CHK2 phosphorylation above, indicating that the reduction in p-CHK2 protein kinase activity for S333F was not sufficient to register any change in p-p53-S20 induction.
      Given its central role in the DNA damage response of the activation of TP53, we investigated whether any of the ATM variants affected the expression of p21. (Despite expressing the HPV E6 protein, HeLa cells do express detectable, functional p53 [
      • Hoppe-Seyler F.
      • Butz K.
      Repression of endogenous p53 transactivation function in HeLa cervical carcinoma cells by human papillomavirus type 16 E6, human mdm-2, and mutant p53.
      ,
      • Raycroft L.
      • Wu H.Y.
      • Lozano G.
      Transcriptional activation by wild-type but not transforming mutants of the p53 anti-oncogene.
      ]) The induction of p21 was assessed and found not to differ significantly among any of the variants (Single factor Anova, p = 0.22, F = 1.48, Fcrit=2.39). Visually, more p21 accumulation was present, though not by a significant factor in the ATM knockdown cells, and those expressing the S1981A, S49C, and V2540I variants (p = 0.17, 0.13, 0.14, and 0.23, respectively). It is possible that this is the result of more unrepaired double strand breaks in these models, causing more cell cycle arrest as reflected in the accumulation of p21. However, accumulation of p21 is not currently known to be evidence of pathogenicity for ATM variants.
      Considering that some of the variants may affect CHK2 signaling, we also investigated CDC25A and CDC25C phosphorylation by the variants S49C, V2540I, and F1463C (Supplemental figure S-2, S-3). While there was no difference in CDC25C phosphorylation, CDC25A phosphorylation was significantly reduced for cells expressing both variants S49C and V2540I, but not for the ATM knockdown cells with the empty vector, a surprising result.

      Discussion

      In this study, we assessed the functional deficiencies conferred by the ATM variants S49C, S333F, F1483C, D1853V, L2307F, and V2540I (Table 2), all of which currently have conflicting interpretations of pathogenicity. Compared to our negative controls (ATM knockdown + empty vector and the kinase deficient variant ATM-S1981A), which showed broad defects across assays, we found that the variants S333F, F1463C, D1853V, and V2540I exhibited markedly reduced activity. Cells expressing S333F intriguingly were defective in HDR and also had reduced phosphorylation of two targets in response to DNA damage, suggesting a potential mechanism for this variant. By contrast, the variant L2307F showed almost entirely wild type activity, with reduction only in plating efficiency during colony survival assay, but a surprising increase in BRCA1 phosphorylation. S49C showed an intermediate phenotype, with S49C expressing cells exhibiting a significant reduction in HDR activity only, while many phosphorylation targets had defects approaching significance. Interestingly, despite low to intermediate REVEL scores for all variants (Table 1), the in silico assessments both alone [
      • Stafford J.L.
      • Dyson G.
      • Levin N.K.
      • Chaudhry S.
      • Rosati R.
      • Kalpage H.
      • et al.
      Reanalysis of BRCA1/2 negative high risk ovarian cancer patients reveals novel germline risk loci and insights into missing heritability.
      ] and combined failed to identify likely damaging variants. Rather than assessing the effects of many variants against a panel of control variants in a high throughput study, we studied the individual effects of several variants in detail using specific in vitro biochemical assessments of the variant activity. Multiple replications of each assay combined with the ability to then compare the variant effects across assays and enable the assessment of variant phenotype with high confidence.
      Table 2Comparison of ATM Variant activity in functional assays showing defects by significance. Values approaching significance are listed as numerical p values. S1981A is a known pathogenic variant, and shATM+pCMV3 represents a knockdown of ATM only. Both negative controls show defects across assays. Comparisons are made to ATM-WT, thus ATM-WT control is not shown. Magnitude effect is shown for measurements at or near significance and represents a fraction of the measurement for ATM-WT. ns = not significant. .
      ATM VUS Assay ComparisonGlobalColony SurvivalKinase Assay
      VariantHDRPlatingIC501.5 µM2.5 µMp-Chk2 T68p-P53 S15p-p53 S20p-H2AX S139p-BRCA1 S1387p21
      S49C0.42, p = 0.03nsnsnsnsns, 0.607, p = 0.057ns, 0.514, p = 0.051nsnsnsns
      S333F0.53, p = 0.007nsnsnsns0.505, p = 0.041nsns0.448, p = 0.027nsns
      F1463C0.37, p = 0.000030.62, p = 0.01ns0.72, p = 0.0330.39, p = 0.002nsnsnsnsnsns
      D1853V0.78, p = 0.00090.58, p = 0.025nsnsnsnsns, 0.716, p = 0.07nsnsnsns
      L2307Fns0.74, p = 0.003nsnsnsnsnsnsns3.19, p = 0.021ns
      V2540I0.37, p = 0.0000140.74, p = 0.004ns0.71, p = 0.046nsns, 0.597, p = 0.0560.420, p = 0.032nsns0.34, p = 0.037ns
      S1981A (Negative Control)0.46, p = 0.000780.40, p = 0.0003ns0.48,  p = 0.008ns0.422, p = 0.0260.412, p = 0.032ns0.385, p = 0.040.34, p = 0.015ns
      shATM+pCMV3 (knockdown)0.51, p = 0.000670.29, p = 0.00040.31, p<0.050.56, p = 0.020.30, p = 0.0030.398, p = 0.0220.517, p = 0.038ns0.541, p = 0.0350.18, p = 0.046ns
      Our investigation of S49C is particularly interesting in light of this variant's previous association with breast cancer [

      Buchholz, T.A., M.M. Weil, C.L. Ashorn, E.A. Strom, A. Sigurdson, M. Bondy, et al., A Ser49Cys variant in the ataxia telangiectasia, mutated, gene that is more common in patients with breast carcinoma compared with population controls. Cancer, 2004. 100(7):1345–51 DOI: https://doi.org/10.1002/cncr.20133.

      ,
      • Stredrick D.L.
      • Garcia-Closas M.
      • Pineda M.A.
      • Bhatti P.
      • Alexander B.H.
      • Doody M.M.
      • et al.
      The ATM missense mutation p.Ser49Cys (c.146C>G) and the risk of breast cancer.
      ], and lung cancer [
      • Schneider J.
      • Illig T.
      • Rosenberger A.
      • Bickeböller H.
      • Wichmann H.E.
      Detection of ATM gene mutations in young lung cancer patients: a population-based control study.
      ]. The intermediate nature of this variant is further evident in the difference in OR for breast cancer between heterozygotic carriers and homozygotic carriers [
      • Fletcher O.
      • Johnson N.
      • dos Santos Silva I.
      • Orr N.
      • Ashworth A.
      • Nevanlinna H.
      • et al.
      Missense variants in ATM in 26,101 breast cancer cases and 29,842 controls.
      ], and was also indeterminate for lung and gastric cancers in uranium miners [
      • Schneider J.
      • Philipp M.
      • Yamini P.
      • Dörk T.
      • Woitowitz H.J.
      ATM gene mutations in former uranium miners of SDAG Wismut: a pilot study.
      ]. Also of note is our assessment of L2307F, which had previously been associated with breast cancer [

      Mangone, F.R., E.C. Miracca, H.E. Feilotter, L.M. Mulligan, and M.A. Nagai, ATM gene mutations in sporadic breast cancer patients from Brazil. Springerplus, 2015. 4:23 DOI: 10.1186/s40064-015-0787-z.

      ], but was benign in our analyses. Previously, the variant D1853V was studied with other variants in an in vitro assessment, and the authors concluded that the variant was fully functional [
      • Navrkalova V.
      • Sebejova L.
      • Zemanova J.
      • Kminkova J.
      • Kubesova B.
      • Malcikova J.
      • et al.
      ATM mutations uniformly lead to ATM dysfunction in chronic lymphocytic leukemia: application of functional test using doxorubicin.
      ]. However, this study only assessed p21 induction as a functional endpoint, finding no difference in this metric. Likewise, we did not see any difference in p21 expression, but our study was much more extensive, and we showed defects across multiple assays for this variant. Furthermore, in contrast to our results [
      • Stafford J.L.
      • Dyson G.
      • Levin N.K.
      • Chaudhry S.
      • Rosati R.
      • Kalpage H.
      • et al.
      Reanalysis of BRCA1/2 negative high risk ovarian cancer patients reveals novel germline risk loci and insights into missing heritability.
      ], a previous study reported that D1853V had little predictive value for breast cancer [
      • Moslemi M.
      • Moradi Y.
      • Dehghanbanadaki H.
      • Afkhami H.
      • Khaledi M.
      • Sedighimehr N.
      • et al.
      The association between ATM variants and risk of breast cancer: a systematic review and meta-analysis.
      ], and a second study failed to find an association with prostate cancer [
      • Angèle S.
      • Falconer A.
      • Edwards S.M.
      • Dörk T.
      • Bremer M.
      • Moullan N.
      • et al.
      ATM polymorphisms as risk factors for prostate cancer development.
      ]. This variant was seen in association with cancer in 3/23 patients in a study of Hodgkin's lymphoma in children [
      • Liberzon E.
      • Avigad S.
      • Yaniv I.
      • Stark B.
      • Avrahami G.
      • Goshen Y.
      • et al.
      Molecular variants of the ATM gene in Hodgkin's disease in children.
      ]. The elevated frequency was surprising considering the rarity of this mutation. This same study found one patient carrying the F1463C mutation as well, but with the P604S variant on the same allele, hampering assessment of the individual variants. F1463C had also previously been noted in a study of patient derived acute lymphoblastic leukemia cell lines, which patients with ataxia telangiectasia are at elevated risk for [
      • Liberzon E.
      • Avigad S.
      • Stark B.
      • Zilberstein J.
      • Freedman L.
      • Gorfine M.
      • et al.
      Germ-line ATM gene alterations are associated with susceptibility to sporadic T-cell acute lymphoblastic leukemia in children.
      ]. However, these studies are statistical studies of small populations, and the predictive ability of any variant may differ widely from population to population and between studies. Additionally, rare variants are harder to study in populations at the statistical level, thus in vitro assessment of the variant is much more likely to yield accurate, biochemically relevant data about any variant. Assessment of variants in ATM in particular is hampered by ATM's nature as a moderate penetrance gene. Compared to genes like BRCA1 or BRCA2, however, ATM mutations are low penetrance in terms of cancer risk [
      • Byrnes G.B.
      • Southey M.C.
      • Hopper J.L.
      Are the so-called low penetrance breast cancer genes, ATM, BRIP1, PALB2 and CHEK2, high risk for women with strong family histories?.
      ]. The risk varies with the cancer type [
      • Hall M.J.
      • Bernhisel R.
      • Hughes E.
      • Larson K.
      • Rosenthal E.T.
      • Singh N.A.
      • et al.
      Germline Pathogenic Variants in the Ataxia Telangiectasia Mutated (<em>ATM</em>) Gene are Associated with High and Moderate Risks for Multiple Cancers.
      ].
      The frequencies of the alleles that showed functional deficits in our assays are also of note. In gnomAD, a database aggregating the allelic frequencies of healthy individuals, all five of the candidates for pathogenic variants are present at very low frequencies, slightly lower than other databases (Table 1). It is very intriguing that for several of the variants, homozygotes are present in the gnomAD database and none of these individuals are known to have ataxia telangiectasia (AT). This leads to the conclusion that none of the variants for which non-AT homozygotes are documented are complete loss of function alleles. Indeed in our data we observed each of these variants presenting with a significant, but not complete loss of function. The variant S49C, for which our experiments suggest at worst an intermediate phenotype, is present in 12 individuals in the homozygous state. S333F is present in one individual as a homozygote, F1463C in 7 homozygotic individuals, and D1853V is present in 6 homozygotic individuals. Interestingly, the variant for which our study suggested the worst phenotype, V2540I, is present in only 14 individuals in gnomAD, with no homozygotes present.
      There are several possible reasons for this discrepancy that are able to incorporate all of the data. The first is that ATM is a moderate penetrance gene, and deleterious mutations in ATM, especially missense mutations that hamper ATM's functions rather than a null allele, will likely retain enough activity to prevent AT. The disease AT itself has a highly variable presentation, with variant AT coined for patients with mild symptoms [
      • Gilad S.
      • Chessa L.
      • Khosravi R.
      • Russell P.
      • Galanty Y.
      • Piane M.
      • et al.
      Genotype-Phenotype Relationships in Ataxia-Telangiectasia and Variants.
      ]. Age of symptom onset and presentation are known to be influenced by the genotype of AT patients and whether activity of the ATM protein is retained [
      • Levy A.
      • Lang A.E.
      Ataxia-telangiectasia: a review of movement disorders, clinical features, and genotype correlations.
      ,
      • Taylor A.M.R.
      • Lam Z.
      • Last J.I.
      • Byrd P.J.
      Ataxia telangiectasia: more variation at clinical and cellular levels.
      ,
      • Micol R.
      • Ben Slama L.
      • Suarez F.
      • Le Mignot L.
      • Beauté J.
      • Mahlaoui N.
      • et al.
      Morbidity and mortality from ataxia-telangiectasia are associated with ATM genotype.
      ,
      • Jacquemin V.
      • Rieunier G.
      • Jacob S.
      • Bellanger D.
      • d'Enghien C.D.
      • Laugé A.
      • et al.
      Underexpression and abnormal localization of ATM products in ataxia telangiectasia patients bearing ATM missense mutations.
      ]. Numerous case reports exist for such patients who are either diagnosed late or otherwise do not follow the natural history of AT, including patients with mild or missing symptoms as a result of a mutation with partial function [
      • OriScott YaelDinur-Schejter
      • Upton Julia
      • Feanny Stephen
      An atypical presentation of ataxia telangiectasia in a school-aged boy secondary to an intronic mutation.
      ,
      • Worth P.F.
      • Srinivasan V.
      • Smith A.
      • Last J.I.
      • Wootton L.L.
      • Biggs P.M.
      • et al.
      Very mild presentation in adult with classical cellular phenotype of ataxia telangiectasia.
      ,
      • Stankovic T.
      • Kidd A.M.J.
      • Sutcliffe A.
      • McGuire G.M.
      • Robinson P.
      • Weber P.
      • et al.
      ATM Mutations and Phenotypes in Ataxia-Telangiectasia Families in the British Isles: expression of Mutant ATM and the Risk of Leukemia, Lymphoma, and Breast Cancer.
      ,
      • Alterman N.
      • Fattal-Valevski A.
      • Moyal L.
      • Crawford T.O.
      • Lederman H.M.
      • Ziv Y.
      • et al.
      Ataxia-telangiectasia: mild neurological presentation despite null ATM mutation and severe cellular phenotype.
      ,
      • Stewart G.S.
      • Last J.I.
      • Stankovic T.
      • Haites N.
      • Kidd A.M.
      • Byrd P.J.
      • et al.
      Residual ataxia telangiectasia mutated protein function in cells from ataxia telangiectasia patients, with 5762ins137 and 7271T–>G mutations, showing a less severe phenotype.
      ]. In some cases, these patients have a null allele and a missense mutation, including one patient not diagnosed with AT until the age of 60 [
      • Newrick L.
      • Sharrack N.
      • Hadjivassiliou M.
      Late-onset ataxia telangiectasia.
      ]. Frequently, cases of variant AT occur when a patient is the carrier of a hypomorphic ATM allele, which retains some function but otherwise is pathogenic [
      • Fiévet A.
      • Bellanger D.
      • Rieunier G.
      • Dubois d'Enghien C.
      • Sophie J.
      • Calvas P.
      • et al.
      Functional classification of ATM variants in ataxia-telangiectasia patients.
      ]. However, in comparison to AT patients, who display mostly null mutations, missense mutations in ATM are vastly overrepresented in The Cancer Genome Atlas [
      • Yamamoto K.
      • Wang J.
      • Sprinzen L.
      • Xu J.
      • Haddock C.J.
      • Li C.
      • et al.
      Kinase-dead ATM protein is highly oncogenic and can be preferentially targeted by Topo-isomerase I inhibitors.
      ]. This same observation has been made specifically for ATM mutations in sporadic leukemia compared to leukemia from AT patients [
      • Vorechovský I.
      • Luo L.
      • Dyer M.J.
      • Catovsky D.
      • Amlot P.L.
      • Yaxley J.C.
      • et al.
      Clustering of missense mutations in the ataxia-telangiectasia gene in a sporadic T-cell leukaemia.
      ].
      It stands to reason, then, that a missense ATM allele may have enough activity to prevent the phenotype of AT while at the same time conferring an elevated lifetime risk of cancer through some impairment of ATM activity. Such an intermediate phenotype is exactly what is suggested by our experiments for the variants S333F, F1463C, D1853V, and possibly S49C, though this latter variant is the most likely to have a benign clinical phenotype, based both on our data, allele frequency, and the data of others reviewed in this discussion. Of our variants, V2540I showed the broadest defects across assays, and possesses the lowest allele frequency with no known homozygotic carriers in gnomAD. Thus, this variant is a clear candidate for pathogenicity.
      That these variants have different levels of impairment as seen in the assays conducted likely indicates that the cancer predisposition for patients carrying these mutations will be different. However, how different is difficult to quantify at this stage, and will require further study to explore.
      Mechanistically, we explored the effect of each of the six ATM VUS on the phosphorylation of several targets. Our study of kinase targets revealed significant impairment of several phosphorylation events for both the ATM knockdown and kinase dead (S1981A) controls, and similar defects for one of our most likely pathogenic variants from other assays, V2540I. While we did not see any defects in the phosphorylation of downstream targets for F1463C, the specific event causing impairment for this variant may be present in an unstudied phosphorylation target, in the activation of the ATM protein itself, or some other mechanism. Defective kinase activity was also revealed in several assays for S333F, possibly explaining the mechanism for the reduction in HDR activity for this variant.
      Other variants showed non-significant changes in phosphorylation activity that was nonetheless visible, such as the reduction in CHK2-T68, p53-S15, and H2AX-139 phosphorylation seen for S49C.  It is entirely possible that western blotting is not sensitive enough to discriminate between the small differences in kinase activity seen across assays for S49C. Alternatively, these and possibly additional small but not statistically significant reductions in signaling may be responsible for the reduction in DNA repair ability observed for S49C in the HDR assay. The significant reduction was seen in the phosphorylation of CDC25A, an end stage event in the DNA damage response pathway, may be reflective of this occurrence.
      Of the kinase targets studied, only p-p53-S20 was completely without utility in visualizing differences among the VUS. This is likely due to its downstream nature, being phosphorylated by CHK2 after CHK2 is activated by ATM.
      Overall, these data lead us to hypothesize that the mechanism by which most, if not all, of the variants which showed reduced function in the HDR and colony survival assays is impaired phosphorylation of one or more downstream targets of ATM in the DNA damage response pathway. Importantly, the specific deficiency in kinase activity appears to vary among the germline ATM variants observed in our cohort.
      Exploring the exact mechanism of function loss for each variant would be an extensive process, as ATM has many targets for its kinase activity, and these in turn have even more downstream targets that may be the exact mechanism of the defect. Additionally, some of the loss of function could stem from a small, but target wide defect in ATM's kinase activity, affecting each target only slightly but with an additive affect over the many targets. Finally, the defect may not be related to the kinase function of ATM at all, but could affect folding, multimerization, disassociation from the inactive multimer, or lead to early degradation of the misfolded protein. Such extensive testing for defects is beyond the scope of this work to evaluate the clinical significance of germline missense mutations. Our data here provide strong evidence according to the ACMG guidelines, PS3 (“Well-established in vitro or in vivo functional studies supportive of a damaging effect on the gene or gene product”), and PM2 (“Absent from controls [or at extremely low frequency if recessive]…in Exome Sequencing Project, 1000 Genomes Project, or Exome Aggregation Consortium”) for pathogenicity for V2540I (Table 1) [
      • Richards S.
      • Aziz N.
      • Bale S.
      • Bick D.
      • Das S.
      • Gastier-Foster J.
      • et al.
      Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.
      ]. Replication of this data in clinical laboratories could provide additional justification for the inclusion of these variants as risk alleles.
      Even with more knowledge of damaging mutations, moderate penetrance genes like ATM will continue to pose a challenge to physicians and genetic counselors because of the variability in phenotypic presentation. Prophylactic surgery is currently not recommended for carriers of pathogenic mutations in ATM, further limiting prevention options [
      • Domchek S.M.
      • Robson M.E.
      Update on Genetic Testing in Gynecologic Cancer.
      ]. The best strategy to address this limitation will continue to be the use of the patient's pedigree, individual preference, and the ability of the counselor to interpret the variant data for the benefit of the patient such that the individual can make an informed decision on the direction of her treatment. This further underscores the importance of tests like those we have presented here. Functional data on the VUS enables clinicians to reach conclusions about their individual patient's risk.
      Most importantly, this work, along with our previous studies [
      • Stafford J.L.
      • Dyson G.
      • Levin N.K.
      • Chaudhry S.
      • Rosati R.
      • Kalpage H.
      • et al.
      Reanalysis of BRCA1/2 negative high risk ovarian cancer patients reveals novel germline risk loci and insights into missing heritability.
      ,
      • Lopes J.L.
      • Chaudhry S.
      • Lopes G.S.
      • Levin N.K.
      • Tainsky M.A.
      FANCM, RAD1, CHEK1 and TP53I3 act as BRCA-like tumor suppressors and are mutated in hereditary ovarian cancer.
      ,
      • Chaudhry S.R.
      • Lopes J.
      • Levin N.K.
      • Kalpage H.
      • Tainsky M.A.
      Germline mutations in apoptosis pathway genes in ovarian cancer; the functional role of a TP53I3 (PIG3) variant in ROS production and DNA repair.
      ], indicates that a cohort of patients with negative genetic diagnoses but pedigrees suggestive of hereditary cancer syndromes can be effectively utilized to curate high yield variant pools for analysis of potentially pathogenic mutations. Such analyses provide an approach that is capable of finding and interrogating the “missing” hereditary mutations that continue to elude clinicians and yield false negative results on cancer gene testing panels, thereby addressing the problem directly. By using arrays of relevant in vitro functional studies, these studies avoid the high false positive and false negative error rates associated with high throughput variant studies and reduce the risk of random error in the results through replication and exclusion of unnecessary assessments by curating the test pool to likely variants found in the patient population. Further research is needed to increase the scale and efficiency of such work. Based on minor allele frequencies and estimated incidence, conclusive data on the six variants investigated in this study could help inform the decisions of 534 ovarian cancer patients and their offspring and a further 8254 breast cancer patients and their offspring. Therefore, based on the cost-benefit of identifying a single pathogenic mutation, small scale but thorough in vitro analysis of variants of unknown significance is useful in promoting human health and increasing the accuracy and utility of cancer genetic testing.

      CRediT authorship contribution statement

      Scott L. Baughan: Methodology, Project administration, Resources, Investigation, Validation, Formal analysis, Writing – original draft, Writing – review & editing, Visualization. Fatima Darwiche: Investigation, Validation. Michael A. Tainsky: Conceptualization, Funding acquisition, Supervision, Data curation, Methodology, Project administration, Resources, Writing – review & editing, Visualization.

      Acknowledgements

      This project was supported by The Barbara and Fred Erb Endowed Chair in Cancer Genetics to MAT and from the Graduate School of Wayne State University, SLB.

      Appendix. Supplementary materials

      References

        • Siegel R.L.
        • Miller K.D.
        • Fuchs H.E.
        • Jemal A.
        Cancer Statistics, 2021.
        CA Cancer J Clin. 2021; 71: 7-33https://doi.org/10.3322/caac.21654
        • Nash Z.
        • Menon U.
        Ovarian cancer screening: current status and future directions.
        Best Pract Res Clin Obstet Gynaecol. 2020; 65: 32-45https://doi.org/10.1016/j.bpobgyn.2020.02.010
        • Domchek S.M.
        • Robson M.E.
        Update on Genetic Testing in Gynecologic Cancer.
        J Clin Oncol. 2019; 37: 2501-2509https://doi.org/10.1200/jco.19.00363
        • Fostira F.
        • Papadimitriou M.
        • Papadimitriou C.
        Current practices on genetic testing in ovarian cancer.
        Ann Transl Med. 2020; 8: 1703https://doi.org/10.21037/atm-20-1422
        • Afghahi A.
        • Kurian A.W.
        The Changing Landscape of Genetic Testing for Inherited Breast Cancer Predisposition.
        Curr Treat Options Oncol. 2017; 18: 27https://doi.org/10.1007/s11864-017-0468-y
        • Minion L.E.
        • Dolinsky J.S.
        • Chase D.M.
        • Dunlop C.L.
        • Chao E.C.
        • Monk B.J.
        Hereditary predisposition to ovarian cancer, looking beyond BRCA1/BRCA2.
        Gynecol Oncol. 2015; 137: 86-92https://doi.org/10.1016/j.ygyno.2015.01.537
        • Hirotsu Y.
        • Nakagomi H.
        • Sakamoto I.
        • Amemiya K.
        • Oyama T.
        • Mochizuki H.
        • et al.
        Multigene panel analysis identified germline mutations of DNA repair genes in breast and ovarian cancer.
        Mol Genet Genomic Med. 2015; 3: 459-466https://doi.org/10.1002/mgg3.157
        • Kraus C.
        • Hoyer J.
        • Vasileiou G.
        • Wunderle M.
        • Lux M.P.
        • Fasching P.A.
        • et al.
        Gene panel sequencing in familial breast/ovarian cancer patients identifies multiple novel mutations also in genes others than BRCA1/2.
        Int J Cancer. 2017; 140: 95-102https://doi.org/10.1002/ijc.30428
        • Kapoor N.S.
        • Curcio L.D.
        • Blakemore C.A.
        • Bremner A.K.
        • McFarland R.E.
        • West J.G.
        • et al.
        Multigene Panel Testing Detects Equal Rates of Pathogenic BRCA1/2 Mutations and has a Higher Diagnostic Yield Compared to Limited BRCA1/2 Analysis Alone in Patients at Risk for Hereditary Breast Cancer.
        Ann Surg Oncol. 2015; 22: 3282-3288https://doi.org/10.1245/s10434-015-4754-2
        • Bunnell A.E.
        • Garby C.A.
        • Pearson E.J.
        • Walker S.A.
        • Panos L.E.
        • Blum J.L.
        The Clinical Utility of Next Generation Sequencing Results in a Community-Based Hereditary Cancer Risk Program.
        J Genet Couns. 2017; 26: 105-112https://doi.org/10.1007/s10897-016-9985-2
        • Manchanda R.
        • Patel S.
        • Gordeev V.S.
        • Antoniou A.C.
        • Smith S.
        • Lee A.
        • et al.
        Cost-effectiveness of Population-Based BRCA1, BRCA2, RAD51C, RAD51D, BRIP1, PALB2 Mutation Testing in Unselected General Population Women.
        J Natl Cancer Inst. 2018; 110: 714-725https://doi.org/10.1093/jnci/djx265
        • Graffeo R.
        • Livraghi L.
        • Pagani O.
        • Goldhirsch A.
        • Partridge A.H.
        • Garber J.E.
        Time to incorporate germline multigene panel testing into breast and ovarian cancer patient care.
        Breast Cancer Res Treat. 2016; 160: 393-410https://doi.org/10.1007/s10549-016-4003-9
        • Desmond A.
        • Kurian A.W.
        • Gabree M.
        • Mills M.A.
        • Anderson M.J.
        • Kobayashi Y.
        • et al.
        Clinical Actionability of Multigene Panel Testing for Hereditary Breast and Ovarian Cancer Risk Assessment.
        JAMA Oncol. 2015; 1: 943-951https://doi.org/10.1001/jamaoncol.2015.2690
        • Frost A.S.
        • Toaff M.
        • Biagi T.
        • Stark E.
        • McHenry A.
        • Kaltman R.
        Effects of Cancer Genetic Panel Testing on at-Risk Individuals.
        Obstet Gynecol. 2018; 131: 1103-1110https://doi.org/10.1097/aog.0000000000002531
        • Knabben L.
        • Imboden S.
        • Mueller M.D.
        Genetic testing in ovarian cancer - clinical impact and current practices.
        Horm Mol Biol Clin Investig. 2019; 41https://doi.org/10.1515/hmbci-2019-0025
      1. Genetics, A., Hereditary Cancer panels: white Paper 2018. 2018.

        • Yang S.
        • Axilbund J.E.
        • O'Leary E.
        • Michalski S.T.
        • Evans R.
        • Lincoln S.E.
        • et al.
        Underdiagnosis of Hereditary Breast and Ovarian Cancer in Medicare Patients: genetic Testing Criteria Miss the Mark.
        Ann Surg Oncol. 2018; 25: 2925-2931https://doi.org/10.1245/s10434-018-6621-4
        • Copur M.S.
        Universal Genetic Testing for All Breast Cancer Patients.
        Oncology (Williston Park). 2019; 33
        • Beitsch P.D.
        • Whitworth P.W.
        • Hughes K.
        • Patel R.
        • Rosen B.
        • Compagnoni G.
        • et al.
        Underdiagnosis of Hereditary Breast Cancer: are Genetic Testing Guidelines a Tool or an Obstacle?.
        J Clin Oncol. 2019; 37: 453-460https://doi.org/10.1200/jco.18.01631
        • Stafford J.L.
        • Dyson G.
        • Levin N.K.
        • Chaudhry S.
        • Rosati R.
        • Kalpage H.
        • et al.
        Reanalysis of BRCA1/2 negative high risk ovarian cancer patients reveals novel germline risk loci and insights into missing heritability.
        PLoS One. 2017; 12e0178450https://doi.org/10.1371/journal.pone.0178450
        • Awasthi P.
        • Foiani M.
        • Kumar A.
        ATM and ATR signaling at a glance.
        J Cell Sci. 2015; 128: 4255-4262https://doi.org/10.1242/jcs.169730
        • Blackford A.N.
        • Jackson S.P.
        ATM, ATR, and DNA-PK: the Trinity at the Heart of the DNA Damage Response.
        Mol Cell. 2017; 66: 801-817https://doi.org/10.1016/j.molcel.2017.05.015
        • Marechal A.
        • Zou L.
        DNA damage sensing by the ATM and ATR kinases.
        Cold Spring Harb Perspect Biol. 2013; 5https://doi.org/10.1101/cshperspect.a012716
        • Choi M.
        • Kipps T.
        • Kurzrock R.
        ATM Mutations in Cancer: therapeutic Implications.
        Mol Cancer Ther. 2016; 15: 1781-1791https://doi.org/10.1158/1535-7163.mct-15-0945
        • Paull T.T.
        Mechanisms of ATM Activation.
        Annu Rev Biochem. 2015; 84: 711-738https://doi.org/10.1146/annurev-biochem-060614-034335
        • Jerzak K.J.
        • Mancuso T.
        • Eisen A.
        Ataxia-telangiectasia gene (ATM) mutation heterozygosity in breast cancer: a narrative review.
        Curr Oncol. 2018; 25: e176-e180https://doi.org/10.3747/co.25.3707
        • Southey M.C.
        • Goldgar D.E.
        • Winqvist R.
        • Pylkas K.
        • Couch F.
        • Tischkowitz M.
        • et al.
        PALB2, CHEK2 and ATM rare variants and cancer risk: data from COGS.
        J Med Genet. 2016; 53: 800-811https://doi.org/10.1136/jmedgenet-2016-103839
        • Tecza K.
        • Pamula-Pilat J.
        • Kolosza Z.
        • Radlak N.
        • Grzybowska E.
        Genetic polymorphisms and gene-dosage effect in ovarian cancer risk and response to paclitaxel/cisplatin chemotherapy.
        J Exp Clin Cancer Res. 2015; 34: 2https://doi.org/10.1186/s13046-015-0124-y
        • Thorstenson Y.R.
        • Roxas A.
        • Kroiss R.
        • Jenkins M.A.
        • Yu K.M.
        • Bachrich T.
        • et al.
        Contributions of ATM mutations to familial breast and ovarian cancer.
        Cancer Res. 2003; 63: 3325-3333
        • DeLeonardis K.
        • Sedgwick K.
        • Voznesensky O.
        • Matloff E.
        • Hofstatter E.
        • Balk S.
        • et al.
        Challenges in Interpreting Germline Mutations in BARD1 and ATM in Breast and Ovarian Cancer Patients.
        Breast J. 2017; 23: 461-464https://doi.org/10.1111/tbj.12764
        • Crawford B.
        • Adams S.B.
        • Sittler T.
        • van den Akker J.
        • Chan S.
        • Leitner O.
        • et al.
        Multi-gene panel testing for hereditary cancer predisposition in unsolved high-risk breast and ovarian cancer patients.
        Breast Cancer Res Treat. 2017; 163: 383-390https://doi.org/10.1007/s10549-017-4181-0
        • Canman C.E.
        • Lim D.S.
        • Cimprich K.A.
        • Taya Y.
        • Tamai K.
        • Sakaguchi K.
        • et al.
        Activation of the ATM kinase by ionizing radiation and phosphorylation of p53.
        Science. 1998; 281: 1677-1679https://doi.org/10.1126/science.281.5383.1677
        • Bakkenist C.J.
        • Kastan M.B.
        DNA damage activates ATM through intermolecular autophosphorylation and dimer dissociation.
        Nature. 2003; 421: 499-506https://doi.org/10.1038/nature01368
        • Lopes J.L.
        • Chaudhry S.
        • Lopes G.S.
        • Levin N.K.
        • Tainsky M.A.
        FANCM, RAD1, CHEK1 and TP53I3 act as BRCA-like tumor suppressors and are mutated in hereditary ovarian cancer.
        Cancer Genet. 2019; (235-236:57-64)https://doi.org/10.1016/j.cancergen.2019.04.061
        • Chaudhry S.R.
        • Lopes J.
        • Levin N.K.
        • Kalpage H.
        • Tainsky M.A.
        Germline mutations in apoptosis pathway genes in ovarian cancer; the functional role of a TP53I3 (PIG3) variant in ROS production and DNA repair.
        Cell Death Discov. 2021; 7: 62https://doi.org/10.1038/s41420-021-00442-y
        • Povirk L.F.
        DNA damage and mutagenesis by radiomimetic DNA-cleaving agents: bleomycin, neocarzinostatin and other enediynes.
        Mutat Res. 1996; 355: 71-89https://doi.org/10.1016/0027-5107(96)00023-1
        • Bolzán A.D.
        • Bianchi M.S.
        DNA and chromosome damage induced by bleomycin in mammalian cells: an update.
        Mutat Res Rev Mutat Res. 2018; 775: 51-62https://doi.org/10.1016/j.mrrev.2018.02.003
        • Hammond E.M.
        • Muschel R.J.
        Radiation and ATM inhibition: the heart of the matter.
        J. Clin. Invest. 2014; 124: 3289-3291https://doi.org/10.1172/JCI77195
        • Momcilović O.
        • Choi S.
        • Varum S.
        • Bakkenist C.
        • Schatten G.
        • Navara C.
        Ionizing radiation induces ataxia telangiectasia mutated-dependent checkpoint signaling and G(2) but not G(1) cell cycle arrest in pluripotent human embryonic stem cells.
        Stem Cells. 2009; 27: 1822-1835https://doi.org/10.1002/stem.123
        • Pandita T.K.
        • Lieberman H.B.
        • Lim D.S.
        • Dhar S.
        • Zheng W.
        • Taya Y.
        • et al.
        Ionizing radiation activates the ATM kinase throughout the cell cycle.
        Oncogene. 2000; 19: 1386-1391https://doi.org/10.1038/sj.onc.1203444
        • Smith J.
        • Tho L.M.
        • Xu N.
        • Gillespie D.A.
        The ATM-Chk2 and ATR-Chk1 pathways in DNA damage signaling and cancer.
        Adv Cancer Res. 2010; 108: 73-112https://doi.org/10.1016/b978-0-12-380888-2.00003-0
        • Caspari T.
        How to activate p53.
        Curr Biol. 2000; 10: R315-R317https://doi.org/10.1016/s0960-9822(00)00439-5
        • Ouchi T.
        BRCA1 phosphorylation: biological consequences.
        Cancer Biol. Ther. 2006; 5: 470-475https://doi.org/10.4161/cbt.5.5.2845
        • Burma S.
        • Chen B.P.
        • Murphy M.
        • Kurimasa A.
        • Chen D.J.
        ATM phosphorylates histone H2AX in response to DNA double-strand breaks.
        J Biol Chem. 2001; 276: 42462-42467https://doi.org/10.1074/jbc.C100466200
        • Xu X.
        • Tsvetkov L.M.
        • Stern D.F.
        Chk2 activation and phosphorylation-dependent oligomerization.
        Mol Cell Biol. 2002; 22: 4419-4432https://doi.org/10.1128/mcb.22.12.4419-4432.2002
        • Matsuoka S.
        • Rotman G.
        • Ogawa A.
        • Shiloh Y.
        • Tamai K.
        • Elledge S.J.
        Ataxia telangiectasia-mutated phosphorylates Chk2 in vivo and in vitro.
        Proc Natl Acad Sci U S A,. 2000; 97: 10389-10394https://doi.org/10.1073/pnas.190030497
        • Zhao H.
        • Traganos F.
        • Darzynkiewicz Z.
        Phosphorylation of p53 on Ser15 during cell cycle caused by Topo I and Topo II inhibitors in relation to ATM and Chk2 activation.
        Cell Cycle. 2008; 7: 3048-3055https://doi.org/10.4161/cc.7.19.6750
        • Banin S.
        • Moyal L.
        • Shieh S.
        • Taya Y.
        • Anderson C.W.
        • Chessa L.
        • et al.
        Enhanced phosphorylation of p53 by ATM in response to DNA damage.
        Science. 1998; 281: 1674-1677https://doi.org/10.1126/science.281.5383.1674
        • Shieh S.Y.
        • Ahn J.
        • Tamai K.
        • Taya Y.
        • Prives C.
        The human homologs of checkpoint kinases Chk1 and Cds1 (Chk2) phosphorylate p53 at multiple DNA damage-inducible sites.
        Genes Dev. 2000; 14: 289-300
        • Matsuoka S.
        • Huang M.
        • Elledge S.J.
        Linkage of ATM to cell cycle regulation by the Chk2 protein kinase.
        Science. 1998; 282: 1893-1897https://doi.org/10.1126/science.282.5395.1893
        • Georgakilas A.G.
        • Martin O.A.
        • Bonner W.M.
        p21: a Two-Faced Genome Guardian.
        Trends Mol Med. 2017; 23: 310-319https://doi.org/10.1016/j.molmed.2017.02.001
        • Hoppe-Seyler F.
        • Butz K.
        Repression of endogenous p53 transactivation function in HeLa cervical carcinoma cells by human papillomavirus type 16 E6, human mdm-2, and mutant p53.
        J Virol. 1993; 67: 3111-3117https://doi.org/10.1128/jvi.67.6.3111-3117.1993
        • Raycroft L.
        • Wu H.Y.
        • Lozano G.
        Transcriptional activation by wild-type but not transforming mutants of the p53 anti-oncogene.
        Science. 1990; 249: 1049-1051https://doi.org/10.1126/science.2144364
      2. Buchholz, T.A., M.M. Weil, C.L. Ashorn, E.A. Strom, A. Sigurdson, M. Bondy, et al., A Ser49Cys variant in the ataxia telangiectasia, mutated, gene that is more common in patients with breast carcinoma compared with population controls. Cancer, 2004. 100(7):1345–51 DOI: https://doi.org/10.1002/cncr.20133.

        • Stredrick D.L.
        • Garcia-Closas M.
        • Pineda M.A.
        • Bhatti P.
        • Alexander B.H.
        • Doody M.M.
        • et al.
        The ATM missense mutation p.Ser49Cys (c.146C>G) and the risk of breast cancer.
        Hum Mutat. 2006; 27: 538-544https://doi.org/10.1002/humu.20323
        • Schneider J.
        • Illig T.
        • Rosenberger A.
        • Bickeböller H.
        • Wichmann H.E.
        Detection of ATM gene mutations in young lung cancer patients: a population-based control study.
        Arch Med Res. 2008; 39: 226-231https://doi.org/10.1016/j.arcmed.2007.08.004
        • Fletcher O.
        • Johnson N.
        • dos Santos Silva I.
        • Orr N.
        • Ashworth A.
        • Nevanlinna H.
        • et al.
        Missense variants in ATM in 26,101 breast cancer cases and 29,842 controls.
        Cancer Epidemiol Biomarkers Prev. 2010; 19: 2143-2151https://doi.org/10.1158/1055-9965.epi-10-0374
        • Schneider J.
        • Philipp M.
        • Yamini P.
        • Dörk T.
        • Woitowitz H.J.
        ATM gene mutations in former uranium miners of SDAG Wismut: a pilot study.
        Oncol Rep. 2007; 17: 477-482
      3. Mangone, F.R., E.C. Miracca, H.E. Feilotter, L.M. Mulligan, and M.A. Nagai, ATM gene mutations in sporadic breast cancer patients from Brazil. Springerplus, 2015. 4:23 DOI: 10.1186/s40064-015-0787-z.

        • Navrkalova V.
        • Sebejova L.
        • Zemanova J.
        • Kminkova J.
        • Kubesova B.
        • Malcikova J.
        • et al.
        ATM mutations uniformly lead to ATM dysfunction in chronic lymphocytic leukemia: application of functional test using doxorubicin.
        Haematologica. 2013; 98: 1124-1131https://doi.org/10.3324/haematol.2012.081620
        • Moslemi M.
        • Moradi Y.
        • Dehghanbanadaki H.
        • Afkhami H.
        • Khaledi M.
        • Sedighimehr N.
        • et al.
        The association between ATM variants and risk of breast cancer: a systematic review and meta-analysis.
        BMC Cancer. 2021; 21: 27https://doi.org/10.1186/s12885-020-07749-6
        • Angèle S.
        • Falconer A.
        • Edwards S.M.
        • Dörk T.
        • Bremer M.
        • Moullan N.
        • et al.
        ATM polymorphisms as risk factors for prostate cancer development.
        Br J Cancer. 2004; 91: 783-787https://doi.org/10.1038/sj.bjc.6602007
        • Liberzon E.
        • Avigad S.
        • Yaniv I.
        • Stark B.
        • Avrahami G.
        • Goshen Y.
        • et al.
        Molecular variants of the ATM gene in Hodgkin's disease in children.
        Br J Cancer. 2004; 90: 522-525https://doi.org/10.1038/sj.bjc.6601522
        • Liberzon E.
        • Avigad S.
        • Stark B.
        • Zilberstein J.
        • Freedman L.
        • Gorfine M.
        • et al.
        Germ-line ATM gene alterations are associated with susceptibility to sporadic T-cell acute lymphoblastic leukemia in children.
        Genes Chromosomes Cancer. 2004; 39: 161-166https://doi.org/10.1002/gcc.10306
        • Byrnes G.B.
        • Southey M.C.
        • Hopper J.L.
        Are the so-called low penetrance breast cancer genes, ATM, BRIP1, PALB2 and CHEK2, high risk for women with strong family histories?.
        Breast Cancer Res. 2008; 10: 208https://doi.org/10.1186/bcr2099
        • Hall M.J.
        • Bernhisel R.
        • Hughes E.
        • Larson K.
        • Rosenthal E.T.
        • Singh N.A.
        • et al.
        Germline Pathogenic Variants in the Ataxia Telangiectasia Mutated (<em>ATM</em>) Gene are Associated with High and Moderate Risks for Multiple Cancers.
        Cancer Prev Res. 2021; 14: 433-440https://doi.org/10.1158/1940-6207.capr-20-0448
        • Gilad S.
        • Chessa L.
        • Khosravi R.
        • Russell P.
        • Galanty Y.
        • Piane M.
        • et al.
        Genotype-Phenotype Relationships in Ataxia-Telangiectasia and Variants.
        The American Journal of Human Genetics. 1998; 62: 551-561https://doi.org/10.1086/301755
        • Levy A.
        • Lang A.E.
        Ataxia-telangiectasia: a review of movement disorders, clinical features, and genotype correlations.
        Mov Disord. 2018; 33: 1238-1247https://doi.org/10.1002/mds.27319
        • Taylor A.M.R.
        • Lam Z.
        • Last J.I.
        • Byrd P.J.
        Ataxia telangiectasia: more variation at clinical and cellular levels.
        Clin. Genet. 2015; 87: 199-208https://doi.org/10.1111/cge.12453
        • Micol R.
        • Ben Slama L.
        • Suarez F.
        • Le Mignot L.
        • Beauté J.
        • Mahlaoui N.
        • et al.
        Morbidity and mortality from ataxia-telangiectasia are associated with ATM genotype.
        J Allergy Clin Immunol. 2011; 128 (382-9.e1)https://doi.org/10.1016/j.jaci.2011.03.052
        • Jacquemin V.
        • Rieunier G.
        • Jacob S.
        • Bellanger D.
        • d'Enghien C.D.
        • Laugé A.
        • et al.
        Underexpression and abnormal localization of ATM products in ataxia telangiectasia patients bearing ATM missense mutations.
        Eur J Hum Genet. 2012; 20: 305-312https://doi.org/10.1038/ejhg.2011.196
        • OriScott YaelDinur-Schejter
        • Upton Julia
        • Feanny Stephen
        An atypical presentation of ataxia telangiectasia in a school-aged boy secondary to an intronic mutation.
        LymphoSign Journal. 2020; 7: 57-60https://doi.org/10.14785/lymphosign-2020-0004
        • Worth P.F.
        • Srinivasan V.
        • Smith A.
        • Last J.I.
        • Wootton L.L.
        • Biggs P.M.
        • et al.
        Very mild presentation in adult with classical cellular phenotype of ataxia telangiectasia.
        Mov Disord. 2013; 28: 524-528https://doi.org/10.1002/mds.25236
        • Stankovic T.
        • Kidd A.M.J.
        • Sutcliffe A.
        • McGuire G.M.
        • Robinson P.
        • Weber P.
        • et al.
        ATM Mutations and Phenotypes in Ataxia-Telangiectasia Families in the British Isles: expression of Mutant ATM and the Risk of Leukemia, Lymphoma, and Breast Cancer.
        The American Journal of Human Genetics. 1998; 62: 334-345https://doi.org/10.1086/301706
        • Alterman N.
        • Fattal-Valevski A.
        • Moyal L.
        • Crawford T.O.
        • Lederman H.M.
        • Ziv Y.
        • et al.
        Ataxia-telangiectasia: mild neurological presentation despite null ATM mutation and severe cellular phenotype.
        Am J Med Genet A. 2007; 143a: 1827-1834https://doi.org/10.1002/ajmg.a.31853
        • Stewart G.S.
        • Last J.I.
        • Stankovic T.
        • Haites N.
        • Kidd A.M.
        • Byrd P.J.
        • et al.
        Residual ataxia telangiectasia mutated protein function in cells from ataxia telangiectasia patients, with 5762ins137 and 7271T–>G mutations, showing a less severe phenotype.
        J Biol Chem. 2001; 276: 30133-30141https://doi.org/10.1074/jbc.M103160200
        • Newrick L.
        • Sharrack N.
        • Hadjivassiliou M.
        Late-onset ataxia telangiectasia.
        Neurol Clin Pract. 2014; 4: 365-367https://doi.org/10.1212/cpj.0000000000000008
        • Fiévet A.
        • Bellanger D.
        • Rieunier G.
        • Dubois d'Enghien C.
        • Sophie J.
        • Calvas P.
        • et al.
        Functional classification of ATM variants in ataxia-telangiectasia patients.
        Hum Mutat. 2019; 40: 1713-1730https://doi.org/10.1002/humu.23778
        • Yamamoto K.
        • Wang J.
        • Sprinzen L.
        • Xu J.
        • Haddock C.J.
        • Li C.
        • et al.
        Kinase-dead ATM protein is highly oncogenic and can be preferentially targeted by Topo-isomerase I inhibitors.
        Elife. 2016; : 5https://doi.org/10.7554/eLife.14709
        • Vorechovský I.
        • Luo L.
        • Dyer M.J.
        • Catovsky D.
        • Amlot P.L.
        • Yaxley J.C.
        • et al.
        Clustering of missense mutations in the ataxia-telangiectasia gene in a sporadic T-cell leukaemia.
        Nat Genet. 1997; 17: 96-99https://doi.org/10.1038/ng0997-96
        • Richards S.
        • Aziz N.
        • Bale S.
        • Bick D.
        • Das S.
        • Gastier-Foster J.
        • et al.
        Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.
        Genet Med. 2015; 17: 405-424https://doi.org/10.1038/gim.2015.30
        • Ioannidis N.M.
        • Rothstein J.H.
        • Pejaver V.
        • Middha S.
        • McDonnell S.K.
        • Baheti S.
        • et al.
        REVEL: an Ensemble Method for Predicting the Pathogenicity of Rare Missense Variants.
        Am J Hum Genet. 2016; 99: 877-885https://doi.org/10.1016/j.ajhg.2016.08.016
        • Tian Y.
        • Pesaran T.
        • Chamberlin A.
        • Fenwick R.B.
        • Li S.
        • Gau C.-.L.
        • et al.
        REVEL and BayesDel outperform other in silico meta-predictors for clinical variant classification.
        Sci Rep. 2019; 9: 12752https://doi.org/10.1038/s41598-019-49224-8
        • Frazer J.
        • Notin P.
        • Dias M.
        • Gomez A.
        • Min J.K.
        • Brock K.
        • et al.
        Disease variant prediction with deep generative models of evolutionary data.
        Nature. 2021; 599: 91-95https://doi.org/10.1038/s41586-021-04043-8