12234–12251 Nucleic Acids Research, 2020, Vol. 48, No. 21 Published online 19 November 2020 doi: 10.1093/nar/gkaa1048 Targeting OGG1 arrests cancer cell proliferation by inducing replication stress Torkild Visnes 1,2,*,† , Carlos Benı́tez-Buelga 1,† , Armando Cázares-Körner1 , Kumar Sanjiv1 , Bishoy M. F. Hanna1 , Oliver Mortusewicz 1 , Varshni Rajagopal1 , Julian J. Albers 1 , Daniel W. Hagey3 , Tove Bekkhus1 , Saeed Eshtad 1 , Juan Miguel Baquero4 , Geoffrey Masuyer 5,6 , Olov Wallner1 , Sarah Müller1 , Therese Pham1 , Camilla Göktürk1 , Azita Rasti1 , Sharda Suman1 , Raúl Torres-Ruiz7,8 , Antonio Sarno9,10,11 , Elisée Wiita1 , Evert Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 J. Homan1 , Stella Karsten 1 , Karthick Marimuthu1 , Maurice Michel1 , Tobias Koolmeister1 , Martin Scobie1 , Olga Loseva1 , Ingrid Almlöf1 , Judith Edda Unterlass1 , Aleksandra Pettke1 , Johan Boström1,12 , Monica Pandey13 , Helge Gad 13 , Patrick Herr13 , Ann-Sofie Jemth 1 , Samir El Andaloussi3 , Christina Kalderén1 , Sandra Rodriguez-Perales7 , Javier Benı́tez4,14 , Hans E. Krokan9,10 , Mikael Altun1,12 , Pål Stenmark5,15 , Ulrika Warpman Berglund1 and Thomas Helleday1,13,* 1 Science for Life Laboratory, Department of Oncology and Pathology, Karolinska Institutet, S-171 76 Stockholm, Sweden, 2 Department of Biotechnology and Nanomedicine, SINTEF Industry, N-7465 Trondheim,Norway, 3 Department of Laboratory Medicine, Karolinska Institutet, Stockholm, Sweden, 4 Human Genetics Group, Spanish National Cancer Research Centre (CNIO), Madrid, Spain, 5 Department of Biochemistry and Biophysics, Stockholm University, SE-106 91 Stockholm, Sweden, 6 Department of Pharmacy and Pharmacology, Centre for Therapeutic Innovation. University of Bath, Bath BA2 7AY, UK, 7 Molecular Cytogenetics Group, Human Cancer Genetics Program, Spanish National Cancer Research Centre (CNIO), Madrid, 28029, Spain, 8 Josep Carreras Leukemia Research Institute and Department of Biomedicine, School of Medicine, University of Barcelona, Barcelona 08036, Spain, 9 Department of Clinical and Molecular Medicine, Norwegian University of Science and Technology, Trondheim, Norway, 10 The Liaison Committee for Education, Research and Innovation in Central Norway, Trondheim, Norway, 11 Department of Environment and New Resources, SINTEF Ocean, N-7010 Trondheim, Norway, 12 Science for Life Laboratory, Division of Clinical Physiology, Department of Laboratory Medicine, Karolinska Institutet, Stockholm, Sweden, 13 Weston Park Cancer Centre, Department of Oncology and Metabolism, University of Sheffield, Sheffield S10 2RX, UK, 14 Spanish Network on Rare Diseases (CIBERER), Madrid, Spain and 15 Department of Experimental Medical Science, Lund University, SE-221 00 Lund, Sweden Received February 28, 2020; Revised October 15, 2020; Editorial Decision October 16, 2020; Accepted October 28, 2020 ABSTRACT is well tolerated in non-transformed cells, we report here that OGG1 depletion obstructs A3 T-cell lym- Altered oncogene expression in cancer cells causes phoblastic acute leukemia growth in vitro and in vivo, loss of redox homeostasis resulting in oxidative validating OGG1 as a potential anti-cancer target. In DNA damage, e.g. 8-oxoguanine (8-oxoG), repaired line with this hypothesis, we show that OGG1 in- by base excision repair (BER). PARP1 coordinates hibitors (OGG1i) target a wide range of cancer cells, BER and relies on the upstream 8-oxoguanine-DNA with a favourable therapeutic index compared to glycosylase (OGG1) to recognise and excise 8-oxoG. non-transformed cells. Mechanistically, OGG1i and Here we hypothesize that OGG1 may represent an shRNA depletion cause S-phase DNA damage, repli- attractive target to exploit reactive oxygen species cation stress and proliferation arrest or cell death, (ROS) elevation in cancer. Although OGG1 depletion * To whom correspondence should be addressed. Tel: +44 1142159677. Email: t.helleday@sheffield.ac.uk Correspondence may also be addressed to Torkild Visnes. Tel: +47 92217744. Email: torkild.visnes@sintef.no. † The authors wish it to be known that, in their opinion, the first two authors should be regarded as Joint First Authors.  C The Author(s) 2020. Published by Oxford University Press on behalf of Nucleic Acids Research. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Nucleic Acids Research, 2020, Vol. 48, No. 21 12235 representing a novel mechanistic approach to tar- cancer target, which confirms targeting oxidative DNA re- get cancer. This study adds OGG1 to the list of BER pair as a concept for treatment of cancer. Furthermore, this factors, e.g. PARP1, as potential targets for cancer validates that BER inhibitors, other than PARP inhibitors, treatment. are effective as anti-cancer treatments. INTRODUCTION MATERIALS AND METHODS Transformed cancer cells are characterized by increased lev- els of DNA damage (1,2) as a result of lost redox home- Cell culture ostasis, oncogene-induced replication stress and impaired Adherent and suspension cell lines were cultured in RPMI DNA repair pathways (3,4). As a consequence, cancers may (61870-010 Thermo Fisher Scientific), McCoy’s (36600-021 become addicted to efficient repair and hence, targeting Thermo Fisher Scientific) or DMEM (10566-016 Thermo the DNA damage response and repair (DDR) pathways Fisher Scientific) media depending on the cell line. The Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 is an established anti-cancer strategy (5). The base exci- media was supplemented with 10% fetal bovine serum sion repair (BER) pathway, involving PARP1 and glycosy- (10500064, Thermo Fisher Scientific) and 100 U/ml Peni- lases such as 8-oxoguanine-DNA glycosylase 1 (OGG1), is cillin Streptomycin (15140122, Thermo Fisher Scientific) repairing the vast majority of DNA lesions. We and oth- and the cells were cultured at 37◦ C and 5% carbon dioxide. ers previously demonstrated that PARP inhibitors are ef- The BJ-Tert and BJ-Ras cell lines were provided by W. Hahn fective in treatment of homologous recombination defec- (Dana-Farber Cancer Institute), MEF Ogg1−/- cells from tive (HRD) cancers (6,7) and currently four different PARP M. Bignami (Istituto Superiore di Sanità, Rome, Italy), inhibitors are approved for treating several HRD cancers. HCT116 and HCT116+Chr3 human colon carcinoma cells By trapping PARP on DNA (8), clinically used PARP in- were obtained from Dr. Bert Vogelstein (2001, Johns Hop- hibitors interfere with replication forks (9,10), alternative kins, Baltimore, MD, USA), Hec59 and Hec59+Chr2 from end-joining (11–13) and other pathways (14,15). Whereas (37), LCL#1 and LCL#2 from J. Benitez (Spanish National PARP1-inhibitors also blocks BER (16), targeting the BER Cancer Research Centre, Madrid, Spain), and the rest of the pathway as such has not been validated as an anti-cancer cell lines were sourced from commercial suppliers American target. Most of the focus in the DDR inhibitor (DDRi) Type Culture Collection (ATCC) or the German Collec- field has been targeting proteins involved in signalling DNA tion of Microorganisms and Cell Cultures GmbH (DMSZ). double-strand breaks (DSBs) or replication stress (RS), All cultures were passaged a maximum of 25 times after such as ATR, ATM, CHK1 and WEE1 kinases currently thawing from stock vials and checked for mycoplasma con- evaluated in clinical trials (17). tamination using MycoAlert™ Mycoplasma Detection Kit It is well established that expression of oncogenes such (Lonza) every other month. as MYC and RAS lead to the generation of reactive oxygen species (ROS) and oxidative DNA damage (18–21). Thus, high levels of oxidized bases have been found in the genome RNA interference of cancer cells (22,23), which excrete oxidized bases and nu- cleotides into serum and urine serving as robust biomarkers 200 000 immortalized BJ-Tert and 100 000 transformed for cancer (22,23), also reviewed in (24,25). The most com- BJ-Ras-SV40 cells were seeded in six-well plates, incu- mon result of ROS DNA damage is the oxidation of gua- bated overnight and transfected with siRNA duplexes us- nine to 8-dihydro-7,8-oxoguanosine (8-oxodG) in DNA, ing Interferin (# 409–50, Polyplus Transfection) accord- repaired by OGG1. Whereas the presence of 8-oxodG in ing to instructions. After 48 h, 3000 BJ-Tert cells and DNA is miscoding, the signature C→A transversion mu- 1000 BJ-Ras cells per well were seeded in 96-well plates tation is surprisingly rare in human malignancies (26). This and reverse transfected with the same siRNA sequence indicates that high-ROS cancers may rely on efficient path- according to instructions. The final concentrations of ways to repair ROS-induced DNA damage. siRNA duplexes were 10 and 2.7 nM for the forward Surprisingly, Ogg1−/− mice are alive and grow old, albeit and reverse transfections, respectively, and the siRNA se- having increased incidence of lung cancer at the age of 18 quences were siOGG1#1: 5 -GGAUCAAGTAUGGACA months (27). Interestingly, OGG1 overexpression protects CUGAC-3 , siOGG1#2: 5 -GGACAAUCUUUCCGGU cells against Ras-induced senescence (28) and high OGG1 GGA-3 . AllStars negative control siRNA (SI03650318, Qi- expression is correlated with lower genomic instability in a agen) was used as transfection control. panel of adenocarcinoma cell lines (29). Moreover, the tran- Stable transfections with doxycycline-inducible small scriptional activity of genes (PCNA, KRAS, MYC, VEGF) hairpin RNA was performed with the plasmid pRSITEP- and transcription factors (NF-␬B) involved in cell prolifer- U6Tet-(sh)-EF1-TetRep-P2A-Puro-P2A-RFP670 (38) and ation, and initiation or progression of cancer can be modu- transduced into A3 and H460 cell lines as described (38). lated upon introduction and processing of 8-oxodG within The non-targeting shRNA sequence was identical to the promoters and 5 untranslated regions via BER (30–34), al- one described in (38), and the hairpins recognizing OGG1 together suggesting a role for OGG1 in cancer development. hybridized to the following sequences in the OGG1 open While targeting the BER pathway, with for instance reading frame shOGG1#1: 5 -GGAGTGGTGTACTAGC PARP inhibitors, is a validated strategy to treat cancer, GGATC-3 , shOGG1#2: 5 -GTGTGCGACTGCTGCG other BER targeting strategies have received surprisingly lit- ACAAG-3 and shOGG1#3: 5 -TGTGCCCGTGGATG tle attention (35,36). Here, we validate OGG1 as an anti- TCCATAT-3 . 12236 Nucleic Acids Research, 2020, Vol. 48, No. 21 Viability Madrid. All samples were collected under written con- sent and institutional review board agreement. CD34+ Cells were seeded in 96- or 384-well plates and incu- cells was obtained from mononuclear cells were sepa- bated for 3 days for combination experiments or 5 days rated by fractionation in Ficoll-hypaque according to man- for single-drug exposure experiments. Resazurin (R7017, ufacturer’s recommendations (GE Healthcare). Purified SigmaAldrich) was added to a final concentration of CD34+ cells were obtained using a MACS CD34 Micro- 0.01 mg/ml resazurin and fluorescence was measured at Bead kit (Miltenyi Biotec) and were cultured in StemSpan ex530/em590 after incubation for 2, 4 or 6 h. Curves were SFM II (StemCell Technologies) containing 100 U/ml fitted using XLfit software (IDBS) or Prism 8.0 (Graphpad penicillin/streptomycin (Gibco) and a cytokine cocktail of Software), and EC50 values were determined. SCF (100 ng/ml), TPO (100 ng/ml), Flt3 ligand (100 ng/ml, Peprotech. Cells were cultured at 37◦ C, 5% CO2 and 5% O2 . Colony formation assays BJ-Tert and BJ-Ras cell lines were transfected with siRNA Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 Activation of T-cells using Phytohemagglutinin-L (PHA-L) for 48 h as described above, trypsinized, counted and seeded or dynabeads in 6-well plates at 300 cells per well and incubated until colony size surpassed 50 cells, followed by medium removal PMBCs and CD34- fraction (CB) were cultured in the and addition of 4% (w/v) methylene blue in methanol. presence of PHA-L (Sigma-Aldrich, ref: 11249738001) Following extensive washes in tap water and air drying, or Dynabeads® Human T-Activator CD3/CD28 (Ther- colonies with >50 cells were counted. mofisher scientific, ref: 11131D) for the activation and ex- H460 cells harboring knockdown constructs were treated pansion of human T cells according to the manufacturer’s with 500 ng/ml doxycycline for 48 h, trypsinized, counted instructions. and seeded in 6-cm diameter petri dishes for 9 days in the presence of an equivalent concentration doxycycline. Colonies exceeding 50 cells were fixed, stained and enumer- CD3 flow cytometry assay ated as above. The experiment was performed on blood cells from 4 differ- The indicated concentrations TH5487 were spotted in 6- ent healthy individuals, with three replicates each. Human well plates using a Tecan D300e digital dispenser. 200–500 peripheral blood mononuclear cells were isolated from fresh cells/well H460, ACHN, Ogg1−/− mouse embryonic fibrob- buffy coats obtained from healthy donors via the Karolin- lasts and MRC5 cells were then seeded at 200–500 cells/well ska Hospital, Stockholm, Sweden. For separation, Ficoll- and incubated for 7–11 days. Colonies exceeding 50 cells Paque PLUS density medium (17144003, GE Healthcare) were fixed, stained and enumerated as above. and SepMate separation tubes (85450, StemCell) were used, 250 HEK293T or HEK293T (KO) cells were counted and according to manufacturer’s instructions. Briefly, buffy coat seeded in 6-well plates for 8 days in the presence or absence diluted 1:1 with PBS and layered on 15 ml of Ficoll-Paque of TH5487 at indicated concentrations. Colonies were di- PLUS in the SepMate tubes was spun down for 10 min at rectly scanned after removing media and drying colonies. 12 000 × g. The upper layer of the tube content was then Area of HEK293T or HEK293T(KO) colonies was digitally poured into new falcon tubes and washed twice with PBS. colour contrasted and then analysed using ImageJ. Cells were seeded out in round bottom 96-well plates (83.3925.500, Sarstedt) in RPMI Medium 1640 containing PBMC isolation GlutaMAX™ (61870-010, ThermoFisher) supplemented with 10% human AB+ male heat inactivated clotted Peripheral blood mononuclear Cells (PBMC) were iso- whole blood serum (H5667, Sigma-Aldrich) and 100 U/ml lated from peripheral whole blood from controls using penicillin/streptomycin (15140122, Gibco). Non-activated Histopaque® -1077 (Sigma-Aldrich, San Luis, MO, USA) cells were seeded at a concentration of 1 × 106 cells/ml. For following the manufacturer’s instructions. Briefly, periph- CD3/CD28 activation, Dynabeads™ Human T-Activator eral blood was collected in heparin and diluted with CD3/CD28 (11131D, ThermoFisher) were mixed with 0.8 1 equal amount of PBS. Next, blood was centrifuged with × 106 cells/ml at a concentration of 0.75 beads per cell. Only Histopaque® -1077 at 400 × g for 30 min at room tempera- viable cells were counted, using a TC20™ automated cell ture and the PBMC layer was recovered. All steps were pro- counter (Bio-Rad) and Trypan blue (1450021, Bio-Rad). cessed within 4 h after blood extraction. The total volume of medium per well was 200 ␮l. DMSO The samples were obtained from healthy donors who and TH5487 were added directly into the wells, at a con- signed an appropriate informed consent and the proposal centration of 0.25% and 25 ␮M, respectively. After 3 days was approved by the ethics committee at the Fuenlabrada of activation and treatment, 50 ␮l fresh complete medium University Hospital, Madrid, Spain. The study was per- containing DMSO or TH5487 was added to the wells for an formed in accordance with the principles of the Declaration additional 3 days. of Helsinki. After 6 days of culture, 90 ␮l cells were moved to a Nunc™ 96-Well Polystyrene Conical Bottom MicroW- CD34+ isolation and culture ell™ Plate (249935, ThermoFisher), containing 10 ␮l of Isolation of total CD34+ cells was performed from um- cold CellWASH (349524, BD) per well, supplemented bilical cord blood samples (CB) from healthy donors dis- with 5 ␮l Precision Count Beads™ (424902, BioLegend) tributed from Centro de Transfusión de la Comunidad de and 0.5 ␮l of each of the following antibodies: PE-Cy™7 Nucleic Acids Research, 2020, Vol. 48, No. 21 12237 Mouse Anti-Human CD3 (563423, BD), PE Mouse Anti- 8-oxodG detection in genomic DNA Human CD8 (561950, BD), APC Mouse Anti-Human 8-oxodG detection in genomic DNA through LC–MS/MS CD71 (334108, Nordic Biosite) as an additional activation (39) and single-cell electrophoresis (40) were performed as marker and APC-H7 Mouse Anti-Human CD4 (560158, described. BD). The plate was incubated at 4◦ C for 30 min and then washed with cold CellWASH. The samples were then re- Immunofluorescence suspended in 100 ␮l of Annexin V Binding buffer (422201, Nordic Biosite), containing 0.5 ␮l FITC-conjugated An- A3 cells were seeded in a 96-well plate 24 h prior to start nexin V (556420, BD) and 1:500 SYTOX™ Blue Dead Cell of the experiment. Cells were then treated with 10 ␮M Stain (S34857, ThermoFisher) and incubated at RT for 15 TH5487, for 24, 48 and 72 h, respectively or with 2 mM hy- min, protected from light. The cells were then moved to droxyurea for 1 h. Final DMSO concentration was 0.2%. flow cytometry tubes containing 100 ␮l additional Cell- After the treatment, cells were transferred to an imaging WASH, incubated on ice for at least 10 min and then plate (black Falcon 96-well plate) pre-coated with 0.1% poly Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 run with a 3-laser Navios flow cytometer from Beckman L-lysine (Sigma). Cells were allowed to attach for 15 min- Coulter. utes and were then fixed with 4% paraformaldehyde. The 10 000 events from an FSC-SSC gate set on the typical liv- plate was washed with 0.05% Tween 20 in PBS, cells were ing lymphocyte population were collected and the data was permeabilized with 0.5% Triton-X100 in PBS and blocked compensated and analyzed using FlowJo 10.5.3. The CD3 with 2% bovine serum albumin in PBS. Incubation with positive cells were gated for and plotted in a Sytox Blue- primary antibody (mouse anti-␥ H2AX, Millipore, prod- Annexin V graph, where the double-negative cells were con- uct no. 05-636, dilution 1:1000) in 2% BSA/PBS was per- sidered viable. The number of CD3 positive, viable events formed at 4◦ C overnight. After washing with 0.05% Tween was quantified and then divided by the number of Precision 20 in PBS, cells were incubated with secondary antibody Count Beads™ collected per tube. This cell-per-bead num- (Alexa Fluor anti-mouse 647, Thermo Scientific, dilution ber was further divided by the mean of the DMSO viable 1:400) together with DAPI staining at room temperature CD3+ cell-per-bead population, in order to generate a rela- for 1 h. Plate was washed thoroughly with 0.05% Tween tive concentration of living CD3 positive cells in relation to 20 in PBS and imaged with ImageXpress (Molecular De- the DMSO samples. For gating, Fluorescence-Minus-One vices, CA, USA). Image analysis was completed with Cell (FMO) samples were used as technical controls. For com- profiler software. Mean intensities of ␥ H2AX signal in sin- pensation, single-stains and unstained samples were run gle nuclei were averaged and compared between the treat- fresh for every new experiment. Data was further analyzed ment and control. Statistical analysis was performed using and plotted using Microsoft Excel and GraphPad Prism GraphPad 7.0c software. Data from three independent ex- 8. periments were pooled and analyzed for statistical signifi- cance by one-way analysis of variance. Lentiviral generation of cell lines RNA-sequencing in A3 shNT cells The Photinus pyralis luciferase gene from the vector To extract bulk RNA from cells, the pellets were resus- pGL4.32 (Promega) was cloned into pENTR1A no CCDB pended in 500 ␮l of TRI Reagent. After 5 min, 100 ␮l of plasmid (w48-1, AddGene 17398), and shuttled into the chloroform was added and the tubes were shaken by hand pLenti-PGK-hygro DEST (w530-1, AddGene 19066) plas- for one minute. After 15 min incubation, the samples were mid, packaged into lentiviruses as described (38) and trans- centrifuged at 12 000 × g for 15 min at 4◦ C. 300 ␮l of the duced into A3 cells containing shRNA constructs as de- aqueous phase were then mixed thoroughly with 300 ␮l of scribed above and selected with 300 ␮g/ml hygromycin or Isopropanol, 30 ␮l of 3M Sodium Acetate and 1 ␮l of Pel- 20 ␮g/ml blastidicin for 10 days. let Paint (Merck 69049) and incubated over night at −20◦ C. The following day, the samples were centrifuged at 20 000 Immunoblotting × g g for 30 min and the pellets washed twice with 600 ␮l of 70% ethanol. After drying, the pellets were resuspended Cells were washed in cold PBS and lysed in RIPA in 15 ␮l of Elution Buffer and the concentration of RNA buffer (50 mM Tris–HCl pH 7.5, 150 mM NaCl, 1 mM was measured using the Qubit RNA High Sensitivity Assay EDTA, 1% NP-40, 0.5% sodium deoxycholate and 0.1% (Thermo Fischer Q32852) according to the manufacturer’s SDS supplemented with complete protease inhibitor cock- instructions. tail (04693116001, Roche), freeze-thawed once, incubated After diluting the samples, 2 ng of RNA were used as in- 20 min on ice followed by sonication at 80% amplitude, 0.7 put for the Smart-seq2 RNA-sequencing protocol (41) and cycle and 10 cycles in a UP100H (Hielscher Ultrasonics) 50 bp single ends were sequenced on an Illumina HiSeq and clarification by centrifugation. Proteins were separated 3000 sequencer. Reads were mapped to the ENSEMBL hu- and blotted with 4–12% polyacrylamide gels and the Trans- man transcriptome GRCh37 using Tophat 2.1.1 to generate Blot Turbo transfer system, respectively (BioRad). The fol- the read count matrix. lowing primary antibodies were used: rabbit anti-OGG1 (ab124741, Abcam) 1:2500, mouse anti-Actin (ab6276, Ab- Transcriptome data analysis in A3 cells cam) 1:10 000, goat anti-vinculin 1:1000, rabbit anti-histone 3 (ab1791, Abcam) 1:5000, mouse anti-␥ H2AX(p-Ser139) Differential expression analysis was performed with DE- (05-636, Millipore) 1:2000). Seq2 (v. 1.26.0) (42). DEGs were visualized using the En- 12238 Nucleic Acids Research, 2020, Vol. 48, No. 21 hancedVolcano package (v.1.4.0). For functional analysis deoxyuridine (EdU) (Thermo Fisher, A10044) for 20 min. Gene Sets were retrieved with the msigdbr package (v7.1.1). Next, cells were fixed in cold 70% ethanol overnight. The For Gene Set Enrichment Analysis only genes with a mean azide alkyne Huisgen cycloaddition (‘Click reaction’) was expression >5 normalized counts as determined by DESeq2 carried out according to the Click-iT EdU Imaging Man- were taken into account. GSEA was performed using the ual (Thermo Fisher) using various fluorescent dyes: Alexa gsea package (v.1.12.0) that implements the algorithms de- Fluor® 488 azide (Thermo Fisher, A10266), Alexa Fluor® scribed by Subramanian et al. (43). Heatmaps were gen- 647 azide (Thermo Fisher, A10277), or TAMRA azide erated using the ComplexHeatmap package (v.2.2.0) (44). (Thermo Fisher, T10182), depending on assay conditions. Mapping of log2 fold normalized counts onto the DNA Samples were incubated with Alexa Fluor® 488 Mouse Replication KEGG (Kyoto Encyclopedia of Genes and anti-H2AX (pS139) (BD Biosciences, 560445) overnight at Genomes) pathway was done using the pathview package 4◦ C and finally stained with DAPI solution [10 ␮g/ml DAPI (v.1.26.0) (45). All analyses were performed in R v.3.6.0. (Sigma, D9542), 0.1 mg/ml RNase A (Thermo Fisher, EN0531), 1% BSA in PBS] for 20 min at room temperature. Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 RNA extraction, quantification and real time PCR analysis Measurements were done using a Navios Flow Cytome- ter (Beckman Coulter). Data was analyzed using Kaluza® RNA was extracted from cultured cells using TRIzol® Flow Analysis Software (Beckman Coulter). Reagent (Thermo Fisher Scientific). RNA quantity and quality were assessed by NanoDrop® (ND-1000 V3.7.1; Thermo Fisher Scientific). The High-Capacity cDNA Re- Annexin V-staining verse Transcription Kit (Applied Biosystems) was utilized The annexin V staining experiment was performed using for cDNA synthesis following the manufacturer’s instruc- BD Pharmingen FITC Annexin V Apoptosis Detection Kit tions. Two microliters of cDNA at a final concentration I (cat.no 556547) according to the protocol given in the kit. of 10 ng/␮l were mixed with GoTaq® qPCR MasterMix Cells were washed with ice-cold PBS and then washed with 1× (Promega) and 1 ␮M cDNA primers of each pair of 1X Annexin buffer diluted in ice-cold PBS. To the washed primers (F/R) in a final volume reaction of 10 ␮l. Primers pellet of cells, PI and FITC Annexin V dye was added and used are listed in Supplementary Table S2. The amplifica- incubated at room temperature in dark for 15 mins. After tion conditions consisted of an initial step at 95◦ C for 10 the incubation, 400 ␮l of annexin buffer was added and min, followed by 40 cycles of 10 s at 95◦ C and 1 min at run in Navios flow cytometer (Beckman Coulter) to detect 60◦ C. Each qPCR was performed in duplicates or tripli- the apoptosis with blue laser(488 nm, filter 525/40 nm) and cates including no-template controls in an Abi QuantStu- dead cells with blue laser(488 nm, filter 620/30 nm). Kaluza dio 1 Flex Real-Time PCR System (Applied Biosystems). Flow Analysis (Beckman Coulter) was used for data analy- Relative transcript levels for each replication gene, was cal- sis. culated using the 2Ct method for qPCR analysis after normalization with the housekeeping gene beta-actin (B- actin-F 5 -CCTGGCACCCAGCACAAT- 3 ; B-actin-R 5 DNA fibre assay - GGGCCGGACTCGTCATACT-3 ). A3 Cells were exposed to either 0.1% DMSO or 10 ␮M TH5487 for the indicated times, pulse-labeled with 25 Oxidative DNA damage within MCM4 gene promoter ␮M 5-chloro-2 -deoxyuridine (CldU) for 30 min, washed We adapted the procedure described by O’Callaghan et al. with medium and pulse-labeled with 250 ␮M 5-iodo-2 - (46) to measure oxidative DNA damage at the MCM4 gene deoxyuridine (IdU) for 30 min. Alternatively, control or promoter region. This is a qPCR method based on differ- OGG1 shRNA expression was induced with 1 ␮g/ml doxy- ences in PCR kinetics between DNA template digested by cyline for 48 h or 96 h before pulse labelling. CldU was FPG and undigested DNA. We have modified the origi- detected by incubating acid-treated fiber spreads in Ibidi nal protocol to incorporate another purified human OGG1 ␮-Slide VI 0.4 (Ibidi, 80606) with rat anti-BrdU mono- to quantify the accumulation of base lesions specific for clonal antibody (AbD Serotec; cat# MCA2060), whereas OGG1, 8-oxoguanine. Conditions used for incubation were IdU was detected using mouse anti-BrdU monoclonal anti- 200 ng of gDNA together with 2.4 ␮M of hOGG1 dur- body (BD Biosciences; cat# 347580) for 1 h at 37◦ C. Slides ing 4 h. The quantitative real-time amplification of genomic were fixed with 4% paraformaldehyde and incubated with DNA was performed as described by O’Callaghan et al. goat anti-rat Alexa Fluor 555 or goat anti-mouse Alexa (46). Specific primers were used at a final concentration Fluor 488 for 1.5–2 hours. Fibers were examined using a of 100 nM (promoterMCM4-F: 5 -GCTGTGATTGGTGA Zeiss (Jena, Germany) LSM780 confocal laser scanning mi- GGCCC -3 ; promoterMCM4-R: 5 -CAAACCGCGAGA croscope with a 63× oil immersion objective. The lengths CCCAGAG-3 ) . Amplification cycling conditions were 10 of red (AF 555) or green (AF 488) labeled patches were min at 95◦ C, followed by 40 cycles of 95◦ C for 15 s and 60◦ C measured using the ImageJ software (National Institutes of for 1 min. CT method was run in an ABI quant studio 1 Health; http://rsbweb.nih.gov/ij/) and arbitrary length val- and all samples were loaded and analyzed in triplicate. ues were converted into micrometers. EdU flow cytometry analysis Mouse xenograft experiment At the indicated time points, compound/dox treated cells Animal experiments were conducted as per the European were incubated in medium containing 10 ␮M 5-ethynyl-2 - directive, ethical guideline and regulations of the regional Nucleic Acids Research, 2020, Vol. 48, No. 21 12239 animal ethical committee Stockholm 2010/63 (N8914). were used as mobile phases at a flow rate of 1 ml/min, with Upon arrival to animal facility, animals were acclimatized a gradient from 10–90% in 3 min; or Method B0597 × 3: in the animal house for a week with ad lib food and wa- Column Xterra MSC18 (50 × 3.0 mm); H2 O (containing ter, with a 12 h light cycle and the temperature and humid- 10 mM NH4 HCO3 ; pH 10) and MeCN were used as mobile ity set according to laboratory animal guidelines and reg- phases at a flow rate of 1 ml/min, with a gradient of 5–97% ulations. For the xenograft experiment involving inducible in 3 min. For LC–MS, detection was made by UV (254 or shRNA, 1 × 107 non-targeting luciferase transfected A3 214 nm) and MS (ESI+ ). Preparative LC was performed cells in 50% matrigel in PBS were injected into flank re- on a Gilson system using Waters C18 OBD 5 ␮m column gion of BALB/c nude mice. Later, mice from non-targeting (30 × 75 mm) with water buffer (50 mM NH4 HCO3 at pH and shRNA groups were randomly divided into two groups 10) and acetonitrile as mobile phases using a flow rate of containing eight mice per group. One group from both non- 45 ml/min. All final compounds were assessed to be >95% targeting and OGG1-targeting shRNA was given 1 mg/ml pure by LC–MS analysis. doxycycline in drinking water from day 15 till last sur- TH5487 was synthesized as described (39). Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 vival of mice (when tumor size was about to reached 1000 N-(3,4-Dichlorophenyl)-4-[4-[6-(hydroxymethyl)-3- mm3 ). For the experiment involving native A3 cells, 8 × pyridyl]-2-oxo-3H-benzimidazol-1-yl]piperidine-1-carbox 106 luciferase-transfected cells in 50% matrigel were injected amide (TH5796). subcutaneously into NOD-SCID mice. Treatment was initi- Step 1: A mixture of 7-bromo-3-(4-piperidyl)-1H- ated 14 days after inoculation when tumour volume reached benzimidazol-2-one;2,2,2-trifluoroacetic acid (1200 mg, approximately 100 mm3 . TH5487 was formulated in a vehi- 3.0 mmol), diisopropylethylamine (0.52 ml, 3.0 mmol), cle solution of 5% DMSO, 10% tween-80, 10% Cremophor and 3,4-dichlorophenylisocyanate (560 mg, 3.0 mmol) in and 75% water, and the indicated doses were administered DCM (20 ml) was stirred at 20◦ C for 16 h. The product by oral gavage twice daily on weekdays for 6 weeks or un- percipitated and was collected by filtration and washed til tumor size reached 1000 mm3 . Tumor volume was mea- with DCM, water, and then DCM again. The solid was sured using calipers (length × 0.52 × width × width) and dried at 40◦ C under vacuum. LCMS [M+H]+ 483. bioluminescence images were taken using an IVIS spectrum Step 2: A mixture of 4-(4-bromo-2-oxo-3H- in vivo imaging system (Perkin-Elmer) after intraperitoneal benzimidazol-1-yl)-N-(3,4-dichlorophenyl)piperidine- injection of D-luciferin (150 mg/kg in PBS). Following ter- 1-carboxamide (73 mg, 0.15 mmol), potassium carbonate mination, tumors were ground in liquid nitrogen and re- (2M, 0.38 ml), [6-(hydroxymethyl)-3-pyridyl]boronic acid suspended in tris-buffered saline with complete protease (69 mg, 0.45 mmol), and Pd(PPh3 )4 (8.7 mg, 0.0075 mmol) inhibitors (Roche), homogenized and subjected to the in- was stirred under nitrogen atmosphere in dioxane at 95◦ C dicated temperatures (47). Following lysis by three freeze- for 16 h. The mixture was thereafter concentrated and thaw cycles, samples were centrifuged at 20 000 × g for 30 purified by preparative LC, the product was isolated as the min, and OGG1 in the supernatant was detected by im- TFA salt. munoblotting. Survival plots were drawn using Graphpad LCMS [M+H]+ 512. 1 H NMR (400 MHz, DMSO-d6 ) ␦ Prism software. ppm 11.22 (s, 1 H), 8.90 (s, 1 H), 8.79 (d, J = 1.6 Hz, 1 H), 8.30 (dd, J = 8.2, 1.9 Hz, 1 H), 7.89 (t, J = 1.3 Hz, 1 H), 7.81 (d, J = 8.2 Hz, 1 H), 7.49 (d, J = 1.6 Hz, 2 H), 7.36 (d, Biochemical assay J = 7.9 Hz, 1 H), 7.16 (t, J = 7.6 Hz, 1 H), 7.11 (dd, J = 7.9, OGG1 activity was assessed in vitro as previously described 0.9 Hz, 1 H), 4.78 (s, 2 H), 4.47 (ddt, J = 16.3, 8.1, 3.9, 3.9 (39). Hz, 1 H), 4.31 (d, J = 13.6 Hz, 2 H), 2.99 (t, J = 12.3 Hz, 2 H), 2.26 - 2.39 (m, 2 H), 1.73 - 1.81 (m, 2 H). 4-(4-Bromo-2-oxo-3H-benzimidazol-1-yl)-N-(4-iodo-3- Synthetic chemistry experimental section methyl-phenyl)piperidine-1-carboxamide (TH6943). All reagents and solvents were purchased from Sigma- A mixture of 4-iodo-3-methyl-aniline (23 mg, 0.10 Aldrich, Combi-Blocks, Thermo Fischer Scientific, or mmol), diisopropylethylamine (0.035 ml, 0.20 mmol), VWR and were used without purification. The compound and trichloromethyl carbonochloridate (0.0060 ml, 0.050 7-bromo-3-(4-piperidyl)-1H-benzimidazol-2-one;2,2,2- mmol) was stirred in DCM (2.0 ml) at 20◦ C for 5 min. Then trifluoroacetic acid was synthesized as described (39). the mixture was added to a vial charged with 7-bromo-3- Unless otherwise stated, reactions were performed without (4-piperidyl)-1H-benzimidazol-2-one 2,2,2-trifluoroacetic care to exclude air or moisture. Analytical thin-layer acid (41 mg, 0.10 mmol), diisopropylethylamine (0.017 ml, chromatography was performed on silica gel 60 F-254 0.10 mmol) and DCM (2.0 ml). The resulting mixture was plates (E. Merck) and visualized under a UV lamp. Flash stirred at 20◦ C for 3 h. The solids were then filtered off and column chromatography was performed in a Biotage® washed with DCM, water, and then DCM again. The solid SP4 MPLC system using Merck silica gel 60 Å (40–63 was dried at 40◦ C under vacuum, no further purification ␮m mesh). 1 H and 13 C NMR spectra were recorded on a was done. Bruker DRX-400 MHz spectrometer. Chemical shifts are LCMS [M+H]+ 555. 1 H NMR (400 MHz, DMSO-d6 ) ␦ expressed in parts per million (ppm) and referenced to the ppm 11.31 (br. s., 1 H), 8.64 (s, 1 H), 7.65 (d, J = 8.8 Hz, 1 residual solvent peak. Analytical LC–MS were performed H), 7.51 (d, J = 2.5 Hz, 1 H), 7.26 (br. d, J = 8.1 Hz, 1 H), on an Agilent MSD mass spectrometer connected to an 7.13–7.19 (m, 3 H), 6.96 (t, J = 8.1 Hz, 1 H), 4.35 - 4.45 (m, Agilent 1100 system with: Method ST1090A3: Column 1 H), 4.24 - 4.33 (m, 2 H), 2.89–2.99 (m, 2 H), 2.32 (s, 3 H), ACE 3 C8 (50 × 3.0 mm); H2 O (+ 0.1% TFA) and MeCN 2.20–2.31 (m, 2 H), 1.70 - 1.79 (m, 2 H). 12240 Nucleic Acids Research, 2020, Vol. 48, No. 21 X-ray crystallography frequently mutated RAS oncogenes leads to the generation of ROS and oxidative DNA damage (19,20), we wanted to A synthetic gene encoding hOGG1 [11–327] was expressed determine if OGG1, which initiates the BER process by rec- as a cleavable N-terminally His-tagged protein in E. coli ognizing and excising 8-oxodG, is important for the survival BL21(DE3) T1R pRARE2 at 18◦ C and purified using His- of transformed cells. To test this, we used siRNA to down- Trap HP (GE Healthcare) followed by gel filtration us- regulate OGG1 in the well characterized isogenic hTERT- ing HiLoad 16/60 Superdex 200 (GE Healthcare) after tag immortalized BJ fibroblasts transformed or not with SV40 cleavage (Thrombin, GE Healthcare). Protein was stored in largeT antigen and HRAS G12V (57). Five days after trans- 20 mM MES at pH 6.0, 200 mM NaCl, 10% glycerol, 0.5 fection, we could observe a marked decrease in viability of mM TCEP, at 20 mg/ml and stored at −80◦ C. transformed cells, but little effect in non-transformed cells Samples for co-crystallization were prepared by pre- (Figure 1A). Moreover, OGG1 knockdown decreased the incubation of hOGG1 (18 mg/ml) with 2–4 mM of the lig- colony-forming ability of BJ-Ras, but not BJ-Tert cells, indi- ands (from a stock solution of 100 mM in DMSO). Crys- cating that OGG1 protected cells against oncogene-induced Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 tals of the hOGG1:TH5487 complex were obtained from stress (Figure 1B and C). a hanging-drop vapor diffusion setup against 0.12 M Alco- To further investigate if OGG1 depletion inhibited hols, 0.1 M Buffer System 2 pH 7.5 and 48% v/v Precipitant growth in human transformed and cancer cell lines we Mix 4 (Morpheus screen (48), Molecular Dimensions, UK). tested the colony forming ability in previously validated A drop of 2 ␮l of sample was mixed with equal amount of OGG1 CRISPR-Cas9 depleted epithelial kidney embry- reservoir and incubated at 16◦ C. Crystals grew within 24 onic cells (HEK293T(KO)) (39). Whereas the number of hours and were frozen in liquid nitrogen for data collec- colonies was similar in HEK293T (KO) cells compared to tion. Diffraction data (Supplementary Table S3) were col- the parental HEK293T OGG1 proficient cells, the colony lected at station I04 of the Diamond Light Source (Didcot, area generated by the HEK293T (KO) cells was significantly UK). A complete dataset was collected from a single crys- smaller (Supplementary Figure S1A, S1B and S1C). In hu- tal at 100 K. Raw data images were processed and scaled man cancer cell lines, depletion of OGG1 in H460 lung can- with xia2 (49), DIALS (50), and Aimless (51) using the cer cells stably transfected with doxycycline-inducible small CCP4 suite 7.0 (52). Molecular replacement was performed hairpin RNA (shRNA) constructs targeting OGG1 reduced with the coordinates of human OGG1 (PDB code 1EBM), clonogenic ability (Figure 1D and F). A3 T-cell acute lym- to determine initial phases for structure solution in Phaser phoblastic leukemia cells transduced with the same con- (53). The working models were refined using Refmac5 (54) structs divided normally for 48 h after addition of doxycy- and manually adjusted with Coot (55) Validation was per- cline followed by slower proliferation and a loss of viability formed with Molprobity (56). Figures were drawn with Py- thereafter (Figure 1E and F). MOL (Schrödinger, LLC, New York). We then stably expressed luciferase in the same A3 in- The atomic coordinates and structure factors (codes ducible cell line, which was injected subcutaneously into 6RLW) have been deposited in the Protein Data Bank (http: Balb/C nude mice. Doxycycline was added to the drink- //wwpdb.org). ing water when tumor volume had reached 200 mm3 . This caused a regression in tumor size in xenografts harboring Statistical analysis OGG1-targeting shRNA, whilst tumors with non-targeting shRNA were unaffected (Figure 2A and B, Supplemen- Statistical analyses were performed with Prism 8.0 (Graph- tary Figure S2). After seven weeks, only mice injected with Pad Software). Statistical significance Figures 1A, B, D, E, OGG1-knockdown cells remained alive (Figure 2C). These 3D, E, F, 4D, E, G, H, J, K, 5A, C, D, E, F, I, K, 6D, E, F data show that OGG1 protects cancer cells from oncogenic and supplementary Figures S5A, B, S6A, S7A, B was deter- stress in vivo and validate OGG1 as anti-cancer target. mined using unpaired, two-sided t-tests (*P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001) Statistical signif- icance for Figure 3C and Supplementary Figure S4H was OGG1 active site inhibitors suppress cancer cell growth determined using Paired t-test (****P < 0.0001). Statisti- We previously described TH5487 as a selective small cal significance for Supplementary Figures S1B, S1C, S4E molecule inhibitor of OGG1 (39). To further understand and S4F was determined using Mann–Whitney unpaired the precise function of this inhibitor, we here determined test for nonparametric distribution (*P < 0.05, **P < 0.01, the X-ray crystal structure of human OGG1 in complex ***P < 0.001 and ****P < 0.0001). Normality of the dis- with TH5487 (Figure 3A), showing that it binds in the ac- tribution for each data set was tested using Kolmogorov– tive site, in an opposite orientation from the natural sub- Smirnov test. strate (39), with the iodophenyl tail of TH5487 occupying the place of the 8-oxoguanine base. The OGG1 protein un- RESULTS dergoes a conformational change when binding DNA, into a closed structural form. Here, we observe the benzimida- Validation of OGG1 as an anti-cancer target zolone core interacts with a lipophilic exo-site outside the While PARP inhibitors effectively block BER (16), target- active site, stabilized by Ile152 and Leu323 in addition to a ing the BER pathway as such has not been validated as ␲-stacking interaction with His270. Also, the carbonyl side an anti-cancer target since PARP inhibition also has impli- group of TH5487 makes a hydrogen bond with the Ile152 cations on replication forks (9,10), alternative end-joining ␣-amino backbone. Finally, the bromine atom of TH5487 (11–13) and other pathways (14). Since expression of the makes a water-mediated interaction with the side chain of Nucleic Acids Research, 2020, Vol. 48, No. 21 12241 Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 Figure 1. OGG1 knockdown is selectively toxic in oncogene-expressing cells and cancer cell lines. (A) Viability of BJ fibroblast cells immortalized with telomerase (BJ-Tert) or telomerase, SV40 large T protein and HRAS G12V oncogene (BJ-Ras) after OGG1 depletion for 5 days. Data are average ±SD of 10 technical replicates representative of two independent experiments. (B) Clonogenic survival of BJ-Tert and BJ-Ras cells following OGG1 depletion. Data are average ±SD of 8–10 replicates from three independent experiments. (C) Representative immunoblot analysis of OGG1 expression in BJ-Tert and BJ-Ras cells 48 h after transfection with siRNA. (D) Clonogenic survival of H460 cells stably transfected with doxycycline inducible shRNA constructs targeting OGG1 (sh1–sh3) or a non-specific sequence (shNT). Cells were seeded in the presence or absence of 500 ng/ml doxycycline and colonies were enumerated after 10–12 days. Data are average ±SD of six technical replicates from two independent experiments. All values are normalized to the number of colonies in medium free of doxycycline. (E) Proliferation of A3 cells stably transfected with doxycycline-inducible shRNA constructs targeting OGG1 (sh2–3) or a non-specific sequence (NT). Cells were seeded in medium with or without 250 ng/ml doxycycline and counted daily. Cultures reaching a density of more than one million cells per ml were added fresh medium to maintain cell growth and normalized to the starting density. (F) Representative immunoblot analysis of OGG1 expression in H460 and A3 cells stably transfected with shRNA constructs targeting OGG1 and treated with doxycycline as in D and E. Data are average ±SD of four technical replicates from two independent experiments. Statistical significance was determined using unpaired, two-sided t-tests (**P < 0.01, ***P < 0.001 and ****P < 0.0001), in all cases comparing against the distribution of the corresponding non-specific RNAi- sequence. A B C 1250 Doxycycline shNT 100 Tumor volume (mm3) shNT+dox shNT shNT 1000 sh2 shNT+dox Survival (%) sh2+dox 75 sh2 750 sh2+dox 50 shOGG#2 500 250 25 0 0 0 6 12 18 24 30 36 0 25 50 75 100 125 Days Days Figure 2. OGG1 knockdown reduces tumor growth in vivo. (A) A3 cells harbouring luciferase and a doxycycline-inducible shRNA construct targeting OGG1 or a non-specific sequence were injected subcutaneously in mice. Doxycycline was added to the drinking water at day 7, and tumor growth were monitored twice a week thereafter. Data are average ± SD, n = 8 per group. (B) Bioluminescence of luciferase expressing A3 cells in five representative mice imaged 28 days after grafting. (C) Survival of animals grafted with A3 cells. Mice were euthanized when tumor size reached 1000 mm3 . 12242 Nucleic Acids Research, 2020, Vol. 48, No. 21 Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 Figure 3. The OGG1 inhibitor TH5487 is selectively toxic to oncogene-expressing cells and cancer cell lines. (A) Close-up view of ligand TH5487 (green) binding to human OGG1. Important residues in the binding site are marked, hydrogen bond interactions are shown in black dashed line, water (red sphere)- mediated interactions in grey dashed line. (B) Comparison between the binding of ligand TH5487 (green) to human OGG1 (gray) with the structure of TH5675 (pink) bound to mouse OGG1 (yellow, PDB 6G3Y). (C) Viability of BJ-Tert and BJ-Ras cells exposed to the indicated concentrations TH5487 for five days. Data are average ±SD of four technical replicates from two independent experiments. (D) Clonogenic survival of cell lines exposed to TH5487. The cancer cell lines ACHN and H460, and the non-transformed cell lines MRC5 and Ogg1−/− mouse embryonic fibroblasts (MEF) were incubated for 6–11 days in the presence of the indicated concentrations TH5487, followed by colony enumeration. Data are average ±SD values of four technical replicates, representative of three independent experiments. (E) TH5487 selectively decreases viability of cancer cell lines. EC50-values of 34 cancer (red) and 7 non-transformed cell lines (blue). Cells were exposed to a dilution series of TH5487 for five days followed by a viability assessment using resazurin. All cell lines were tested in two to ten independent experiments. Each point represents the EC50-value from one experiment (average of two or three technical replicates). (F) Comparative analysis of EC50 values for TH5487 in cancer- and non-transformed cell lines, and CD34+ hematopoietic stem cells, CD34- fraction from cord blood or PBMCs from healthy donor blood. All primary blood cells were tested upon activation with Dynabeads/Phytohemagglutinin or not and exposed to a dilution series of TH5487 to calculate EC50 values by resazurin assay. Data for blood cells are average ±SD from two to four independent donors, and are significantly different from the hematological cancer cell lines tested, using a two-sided unpaired t-test (*P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001, ns, non-significant). Nucleic Acids Research, 2020, Vol. 48, No. 21 12243 Ser326. This confirms the overall molecular mechanism of and 10 ␮M TH5487 treatment, which was more substan- this class of inhibitors (Figure 3B and Supplementary Fig- tial at 20 ␮M (Figure 4D and E). To further characterize ure S3), where OGG1 adopts a closed conformation upon the cell proliferation defect, we released the cells into fresh binding TH5487, thereby blocking OGG1 from accessing medium after treatment. Similar to A3 OGG1 knockdown substrate DNA lesions in DNA and chromatin (39). cells, A3 cells exposed for 6 days to 10 ␮M TH5487 were To study if TH5487 selectively inhibited growth of able to divide normally after replacing with fresh cell cul- transformed cells, we incubated the pair of isogenic ture media (Figure 4F to H) showing proliferation defects BJ-fibroblasts with the inhibitor. TH5487 causes a due to OGG1 depletion or targeting with TH5487 to be re- concentration-dependent decrease in viability in BJ-Ras versible. This suggests that TH5487 might primarily cause fibroblasts (Figure 3C), in line with reduced survival cytostatic rather than cytotoxic effects. after RNAi-mediated OGG1 downregulation (Figure 1A). To test whether the antiproliferative effect of TH5487 was Furthermore, the isogenic hTERT-immortalized cells are related to OGG1 inhibition, OGG1 knockdown cells treated less sensitive to TH5487 than BJ-Ras (EC50 = 29.5 ± 4.7 with doxycycline for 5 days were challenged for an addi- Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 ␮M and EC50 = 11.4 ± 5.7 ␮M, respectively, P < 0.05) tional 48 h with 10 ␮M TH5487. However, no significant (Figure 3C), in line with high tolerance to OGG1 siRNA additional toxic effect was detected (Figure 4I–K). Alter- (Figure 1A). We then characterized a subset of adherent cell natively, A3 OGG1 shRNA cells were exposed to doxycy- lines in colony formation assays and find TH5487 causes cline alone or in combination to TH5487 for 6 days and we a concentration-dependent loss of clonogenic potential found that TH5487 had additional toxic effect for one of in the cancer cell lines ACHN and H460, whereas the the constructs (A3 sh2) but not for the other (A3 sh3) (Sup- non-transformed cell lines MRC5 and Ogg1−/− mouse plementary Figure S5A and S5B). Interestingly, the combi- embryonic fibroblasts were unaffected by treatment up to nation of TH5487 and doxycycline in A3 sh2 knockdown 20 ␮M (Figure 3D). These results suggest that TH5487 cells led to a partial recovery of OGG1 protein levels, which may selectively kill cancer cells. To test this more broadly, might explain why TH5487 has an additional effect for this we determined the sensitivity of a large panel of cancer specific construct (Supplementary Figure S5C and S5D). and non-transformed cell lines and found overall that Finally, to further investigate potential off-target effects of TH5487 caused loss of viability in cancer cells while being the compound in an OGG1 knockout model, we evaluated better tolerated in non-transformed cell lines (Figure 3E the clonogenic potential of HEK293T or HEK293T(KO) and Supplementary Table S1). Moreover, this selectivity cells exposed to TH5487 for a period of 8 days. Here, we was not unique to TH5487, since the TH5487 inhibitor found that the colony area was significantly reduced in analogues TH5796 and TH6943 displayed a similar growth the parental HEK293T OGG1 proficient cells compared inhibition specific to cancer cells (Supplementary Figure to HEK293T (KO) cells (Supplementary Figure S5E and S4A to S4F and Supplementary Table S1). S5F). These results suggest that the antiproliferative proper- Finally, TH5487 was well tolerated by a set of hematopoi- ties of TH5487 in cancer cells are primarily related to OGG1 etic primary cells derived from healthy donors (CD34+ inhibition although residual off target effects at higher con- cells, resting and activated CD34- cells from umbilical cord centrations (20 ␮M or more) cannot be excluded, since it is blood (Figure 3F). Taken together, TH5487 caused loss of a common mechanism of action of cancer drugs (58). proliferation to a higher degree in the tested cancer cell lines (EC50 17.9 ±6.9 ␮M) compared to non-transformed cell Targeting OGG1 results in replication stress lines (EC50 36.9 ± 7.5 ␮M) (Figure 3F). Hematological cancer cell lines were particularly sensitive compared to pri- Given the general high levels of oncogene-induced ROS and mary blood cells (EC50 14.7 ± 4.7 ␮M and 35.1 ± 20 ␮M, associated oxidative DNA damage in cancer (18–23), we P = 0.00029). were curious if the loss of viability was preceded by ge- nomic accumulation of oxidized DNA. To test this, we ex- posed A3 cells to 10 ␮M TH5487 for up to five days, puri- Anticancer properties overlap between OGG1 loss and fied genomic DNA, and measured the genomic content of OGG1 inhibitors 8-oxodG. Unexpectedly, 8-oxodG levels did not rise above We selected A3 cells as a model cell line to characterize assay background levels during the experiment (Figure 5A). anti-proliferative properties mediated by either OGG1 in- Nor did 8-oxodG rise above background in OGG1-depleted hibition by TH5487 or by shRNA knockdown of OGG1. A3 cells (Supplementary Figure S6A). Using a more sensi- We found that similar to OGG1 knockdown (Figure 1E), tive method, we could detect genomic OGG1 substrates in TH5487 inhibited proliferation of A3 and other T-cell acute A3 cells using the modified Comet assay. Treatment with leukoblastic leukemia cell lines with an EC50 of ∼10 ␮M 10 ␮M TH5487 caused a ∼50% increase in OGG1-induced (Figure 4A and Supplementary Table S1). A3 cells exposed tail length, indicating that TH5487 induced a small increase to 10 ␮M TH5487 showed slower proliferation for up to 72 in OGG1 substrate levels in these cells (Figure 5B and C). h (Figure 4B), followed by a mild decrease in cell viability The overall increase in nuclear 8-oxodG in DNA is minor (Figure 4C). The same dose of TH5487 was well tolerated in comparison to KBrO3 control and unlikely to explain by resting and activated peripheral blood mononuclear cells growth inhibition induced by OGG1 inhibitors or protein (PBMCs) and primary T-cells from healthy donors (Supple- loss. mentary Figure S4G to S4K). Using Annexin V-propidium In spite of few nuclear oxidative DNA lesions, we iodide staining, we could demonstrate a slight but signifi- found unexpectedly a significant increase in phosphorylated cant increase in apoptosis after OGG1 knockdown and 5 ␥ H2AX, detected after 24 h treatment with 10 ␮M TH5487 12244 Nucleic Acids Research, 2020, Vol. 48, No. 21 Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 Figure 4. Anticancer properties overlap between OGG1 loss and OGG1 inhibitors. (A) Viability of T-cell acute leukoblastic leukemia cell lines treated for 5 days with the indicated doses TH5487. Data are average ±SD of independent experiments (A3, n = 5; Jurkat, n = 3; MOLT-4, n = 4; CCRF-CEM, n = 3). (B) Relative cell numbers and (C) viability (%) for A3 cells treated for 5 days with 10 ␮M TH5487. Data are average ±SD of six replicates from three independent experiments. (D) Induction of apoptosis in A3 cells treated for 72 h TH5487 and stained for Annexin V. Data are average ±SD of three independent experiments. (E) A3 cells transfected with shRNA targeting OGG1 were treated with doxycycline for 6 days and stained for Annexin V. Data are average ±SD of two independent experiments. (F) Scheme for recovery experiment in which A3 OGG1-shRNA cells or A3 cells were silenced or inhibited for 6 days with doxycycline (200 nM) or TH5487 (10 ␮M). Then, 250 000 cells/ml were seeded in fresh media and recovered for 24 h and 144 h by measuring cell number (G) and viability (H). Data are average ±SD of 2–6 technical replicates representative of two independent experiments (I). Scheme for off-target effect evaluation in which A3 OGG1-shRNA cells were exposed to doxycycline for six days and after that, cells were exposed to TH5487 (10 ␮M) for 48 h followed by measurements of cell numbers (J) and viability by trypan blue exclusion (K). Data are average ±SD of two technical replicates from four independent experiments. Statistical significance was determined using unpaired, two-sided t-tests (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns, non-significant). Nucleic Acids Research, 2020, Vol. 48, No. 21 12245 Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 Figure 5. Effect of TH5487 on DNA lesions, DNA damage markers and DNA replication. (A) 8-oxodG accumulation in A3 cells. A3 cells were treated with 20 mM KBrO3 for 1 h or with 10 ␮M TH5487 for the indicated times, and the amount of genomic 8-oxodG was quantified with LC-MS/MS. Data are average ±SD of five replicates from two independent experiments. (B) Comet assay. A3 cells were treated with 10 ␮M TH5487 for 72 h and strand-breaks and OGG1 substrate lesions were analyzed with the OGG1-modified Comet assay. Representative images of cells are shown. (C) Violin blot of Comet tail moment. Cells were treated as in B, and the tail moment of the cells were analyzed using blinded automatic analysis (n = 200 per condition from two independent experiments). The full line indicates median, and the dotted lines quartiles. (D) Violin blot of phosphorylated ␥ H2AX intensity. A3 cells were treated with 0.1% DMSO or 10 ␮M TH5487 for the indicated times and stained for phosphorylated ␥ H2AX. Treatment with 2 mM Hydroxyurea for 1 h was used as positive control. At least 33 000 nuclei per group were quantified from three independent experiments. The full line indicates median, and the dotted lines quartiles. All values are normalized to that of the mean value of the 1 h non-treated sample. (E) Relative induction of positive ␥ H2AX cells in OGG1 shRNA depleted cells for 72 h or OGG1 inhibited with 10 ␮M of TH5487 for 24, 48 and 72 h. Data are average ±SD of two technical replicates representative of 1–2 independent experiments (F) Relative cell cycle distribution of positive ␥ H2AX gated cells along G1, S or G2/M cell cycle phases after 72 h treatment with 10 ␮M TH5487. Data are average ±SD of two technical replicates from two independent experiments. (G) Experimental setup of DNA fiber assay. A3 cells were treated with 0.1% DMSO or 10 ␮M TH5487 for 48 h or 72 h, alternatively A3 shOGG#2 were treated with doxycycline for 48 or 96 h followed by addition of 5-chloro-2 -deoxyuridine (CldU) or 5-iodo-2 -deoxyuridine (IdU) to the medium. Representative images of DNA replication fibers are shown. (H) Distribution of fork speed in CldU-labelled A3 cells treated with DMSO or TH5487 for 48 or 72 h. (I) Total fork speed in DMSO and TH5487-treated cells. (J) Distribution of fork speed in CldU-labelled A3 shOGG1#2 cells treated or not with doxycycline for 48 or 96 h. (K) Total fork speed in A3 shOGG1#2 cells treated or not with doxycycline. Data shown as average ± SD from three independent experiments. At least 300 forks were scored per condition. Statistical significance was determined using unpaired, two-sided t-tests (**P < 0.01, ***P < 0.001 and ****P < 0.0001; ns, non-significant). 12246 Nucleic Acids Research, 2020, Vol. 48, No. 21 as well as following shRNA depletion of OGG1 (Figure 5D after an increase in dose to 40 mg/kg after the first week and E). Interestingly, the distribution of ␥ H2AX positive (Figure 7A). Since we were unable to demonstrate OGG1 cells after treatment by OGG1 inhibitors during cell cycle target engagement in A3 cells in the derived xenograft tu- revealed that the DNA damage induced by TH5487 was mor (Figure 7B), we hypothesized that the lack of in vivo confined to S-phase cells (Figure 5F), indicating that OGG1 TH5487 efficacy was due to lack of OGG1 target engage- deficiency may affect DNA replication. However, the A3 ment. cell cycle distribution after OGG1 perturbation remained As TH5487 was shown to be stable in mouse serum (39), largely constant, although we observed an accumulation we investigated if TH5487 interacted with serum proteins. in the sub-G1 population (Supplementary Figure S6B and Thus, we incubated bovine serum albumin proteins together S6C). Yet, OGG1 inhibition induced a marked reduction in with TH5487 and observed that bovine serum albumin in- replication fork rate at both 48- and 72 h time points (Fig- duced a marked loss of inhibition of OGG1 enzyme ac- ure 5G to I), largely mimicked by OGG1 depletion by RNAi tivity (Figure 7C). This suggests that albumin competed at both 48- and 96-h time points (Figure 5G, J and K). with OGG1 enzyme for TH5487 and severely reduced its Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 Furthermore, experiments where cells were pulsed with 5- efficacy. Thus, we conclude that systemic administration of ethynyl-2 -deoxyuridine (EdU) followed by click-chemistry TH5487 is not feasible with the current molecule and new mediated detection of newly incorporated nucleotides, sup- compounds or formulation strategies have to be developed ported that DNA synthesis was reduced in OGG1 inhib- for effective treatment of cancer. ited and OGG1-depleted cells (Supplementary Figure S7). Together, these data demonstrate that the replicative poten- DISCUSSION tial of OGG1 perturbed cells may be caused by replication stress in the form of a lowering of replication fork rate and Despite the inherent high levels of ROS and DNA dam- DNA damage during S-phase. age in cancer, and the clinical success of BER pathway in- To explore how OGG1 inhibition causes replication hibitors, DNA repair pathways removing oxidative DNA stress we performed RNA sequencing with A3 cells treated damage have so far not been pharmacologically targeted with 10 ␮M TH5487 or DMSO for 24 h. Gene Set En- (35). We and others (39,60,61) recently reported the first richment Analysis (GSEA) of the differentially expressed examples of cell active inhibitors targeting OGG1, and genes revealed that ‘DNA replication’ was one of the most here we provide the first structural details of how TH5487 downregulated gene signatures (Figure 6A and Supplemen- binds to the human enzyme (Figure 3A and B). Impor- tary Figure S8). Genes encoding key proteins and enzymes tantly, we show that multiple cancer cell lines are sensi- necessary for DNA replication were downregulated upon tive to loss or inhibition of OGG1, which could relate to TH5487 treatment (Figure 6B and C). We validated a panel an underlying addiction to a functional OGG1 in can- of replication associated genes following OGG1 knock- cer. This may in part explain why Ogg1 knockout mice down, confirming a ∼50% downregulation after 96 h for are largely spared from cancer (62), and the relative rar- all eight investigated transcripts (Figure 6D). These results ity of the expected C→A mutation signature in human suggest that TH5487 and OGG1 knockdown induce early cancers (26) in spite of the well-demonstrated fundamen- alterations in the A3 transcriptional profile that include tal role of ROS in hyperplasias. Surprisingly, after treat- downregulation in DNA replication signature. These are ment with OGG1 inhibitor the absolute levels of genomic unlikely to be explained by a lack of replicating cells, since 8-oxodG remained low, at or above background levels cells in S-phase were not depleted from the cell population of one 8-oxodG per million guanines (corresponding to in any experimental condition (Supplementary Figure S6B about 3000 damaged bases per diploid genome), even af- and S6C). ter prolonged treatments with TH5487. This finding sug- Interestingly, the promoter sequence of several genes in- gests that a sizeable number of 8-oxodG in bulk chromoso- cluded in the ‘DNA replication’ gene set contains one or mal DNA of cancer cells must be very low, even in stressed several SP1 transcription factor binding motifs (https://epd. cells. This may have broad implications for 8-oxodG as a epfl.ch//index.php). Since OGG1 has been reported to be biomarker in cancer research since it questions the use- required for recruitment of transcription factors such as fulness and applicability of indirect detection of genomic SP1 to the promoter of genes containing motifs of recog- 8-oxodG in whole cells, using single-cell electrophoresis nition for this transcription factor (59), we exposed A3 cells and antibody- or avidin-staining. Instead, novel methods to or OGG1-shRNA A3 cells to TH5487 for 72 h or doxycy- map and study specific genomic locations enriched with 8- cline for 96 h, respectively, and assessed promoter oxida- oxodG lesions have been developed. For example, mapping tion at the SP1 binding motif contained within the MCM4 of 8-oxodG in mouse embryonic fibroblasts (MEF) and promoter. We observed a significant accumulation of ox- MCF10A breast epithelial cells, has revealed that 8-oxodG idative DNA damage at this position that correlated with is enriched at promoters and 5 and 3 untranslated regions a slight but significant downregulation of MCM4 mRNA rather than randomly distributed (63–65). In MEF cells, levels (Figure 6E and F). ∼10 000 regions of 8-oxodG were enriched in WT mouse Since our data indicated that OGG1 inhibitor treatment embryonic fibroblasts compared to ∼18 000 regions when slowed A3 cell proliferation, we wanted to test if treatment Ogg1 was knocked out (63). Interestingly, 21 out of the with TH5487 could suppress growth of A3 xenograft tu- 113 replication genes transcriptionally downregulated by mors in mice. We orally administered 20 mg/kg TH5487 in OGG1 inhibitors (ATG, CDK1, CDK2, CDK2AP1, DNA2, established subcutaneous tumors of A3 cells. However, we E2F7, FEN1, MCM3, MCM4, NAP1L1, NFIC, NUCKS1, were not able to detect xenograft growth inhibition, even PDS5A, POLD4, RAC1, RAD51, RAD9A, RFC4, SENP2, Nucleic Acids Research, 2020, Vol. 48, No. 21 12247 Figure 6. TH5487 treatment induces a downregulation of DNA replication genes (A) Gene Set Enrichment analysis plot of the DNA Replication Gene Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 Ontology Gene Set. A3 shNT cells were treated with 10 ␮M TH5487 for 24 h or DMSO as control, and RNA sequencing was performed. Genes with a mean normalized count > 5 were ranked by log2 fold change. Genes towards the top of the ranked gene list are upregulated after TH5487 treatment, genes with higher ranks are downregulated. NES: normalized enrichment score, P value has been adjusted for multiple testing using the Benjamini–Hochberg method. (B) Log2 fold changes in expression level of DNA Repair genes of TH5487 treated A3 control cells compared to DMSO treated cells. Only genes with a mean normalized count >5 were analyzed. Visualization of the KEGG DNA Replication pathway. (C) Heatmap of the leading edge genes from the DNA Replication Gene Ontology Gene Set. Genes and samples were unsupervisedly clustered using hierarchical clustering (euclidean distance, complete linkage). Rows indicate technical replicates (n = 3). A negative z-score indicates relative downregulation, a positive relative upregulation of the gene. (D) Relative mRNA expression levels of replication-associated genes in OGG1-depleted cells (normalized to A3 shNT and expressed in percentage). A3 shNT or sh2-cells were treated with doxycycline for 72 h. Data shown as average ± SEM from four technical replicates from two independent experiments. (E) Oxidation at MCM4 promoter. The indicated cells were treated with 10 ␮M TH5487 or 250 ng/ml doxycycline for 72 and 96 h, respectively, DNA was purified and treated or not with OGG1 followed by PCR amplification of the MCM4 promoter. Data shown are the difference in Ct values caused by OGG1 enzymatic treatment or buffer. Data represent average ± SEM of three technical replicates from two independent experiments. (F) Relative expression of MCM4 mRNA levels in OGG1 perturbed cells expressed in percentage. Cells were treated as in D. and data represent average ± SEM 3–4 technical replicates from two independent experiments. Statistical significance was determined using a two-sided unpaired t-test (ns, non-significant, *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001). 12248 Nucleic Acids Research, 2020, Vol. 48, No. 21 Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 Figure 7. OGG1 inhibitors does not inhibit tumor growth in vivo. (A) Lack of TH5487 efficacy in vivo. A3 cells were injected subcutaneously in mice, allowed to graft for three days and then administered 20–40 mg/kg TH5487 orally, once a day on weekdays. Twelve days into the treatment, the dose was increased to 40 mg/kg. Data are average ± SD (n = 9 per treatment group). (B) TH5487 target engagement in A3 xenograft. Tumors were excised following termination, and TH5487 target engagement was analyzed by cellular thermal shift assay, using an antibody specific for the human OGG1 protein. Data are average ± SD (n = 4 different tumors per group). (C) Effect of bovine serum albumin on OGG1 inhibition by TH5487. A dilution series of TH5487 was incubated with OGG1 enzyme and a DNA reporter oligonucleotide containing an OGG1-substrate, in the presence or absence of 1% (w/v) bovine serum albumin. Data are average ± SD of four technical replicates from two independent experiments. SIN3A, SMC3), were included in the list of chromosomal of benefit. On the other hand, the ready reversibility of gene locations having 8-oxodG-enriched peaks in Ogg1−/− - TH5487 in cell proliferation supports the cytostatic nature MEF (63). This observation, together with our data show- of OGG1 inhibitors (Figure 4). Since cancer intervention ing that OGG1 inhibition/depletion induces accumulation using cytostatic drugs alone is often ineffective, combina- of oxidative DNA damage at MCM4 gene promoter, and tion strategies using OGG1 inhibitors together with other downregulation of MCM4 mRNA levels, supports that cancer-selective or cytotoxic drugs should be rationalized OGG1 might play a role in 8-oxodG epigenetic regulation and tested in preclinical cancer models to address the util- of gene expression. This adds to the growing list of obser- ity of OGG1 inhibitors for cancer treatment. For that, sig- vations where DNA oxidation in promoter sequences mod- nificant improvements to its formulation may be required ulates gene transcription. for the compound to show in vivo efficacy in longer term Indeed, our transcriptional study to evaluate the impact experiments such as cancer xenograft studies. In particu- of TH5487 in A3 cells revealed that differential gene ex- lar, the affinity to albumin reported here, which cause more pression are observed 24 h after OGG1 inhibition. OGG1 than 99% of the molecule to associate with plasma pro- inhibition induced transcriptional downregulation of genes teins (39) is a challenge that must be solved for TH5487 involved in different biological processes such as cell cycle to be useful for in vivo applications where OGG1 must be control and DNA replication (Figure 6 and Supplemen- targeted over a long period of time to induce cell death tary Figure S8). This occurred earlier compared to some and inhibit proliferation. Alternatively, differences between of the replication stress phenotypes associated with OGG1 mouse strains, formulation or mode of administration may inhibition/downregulation observed in other assays. For ex- account for the different outcomes of in vivo experiments ample, reduction in EdU incorporation was only detectable involving TH5487. Moreover, although TH5487 is well tol- after 96 h treatment with TH5487 or doxycycline in A3 cells erated by mice (39), the possibility for off-target effects or OGG1-shRNA A3 cells respectively (Supplementary Fig- at high doses (see e.g. Figure 4D) must be taken into ure S7). In addition, we have found reduced transcriptional account. levels for all genes encoding for MCM2–7 complex after In conclusion, we validate OGG1 as one more to the list TH5487 treatment (Figure 6) followed by a progressive ac- of existing BER factors, e.g., PARP1, as potential targets cumulation of ␥ H2AX positive A3 cells during S-phase at for treatment of cancer. 72 h (Figure 5). These results are not surprising considering that MCM proteins are required for processive DNA repli- DATA AVAILABILITY cation and that loss of MCM function causes DNA dam- age and genome instability (66,67). This suggests that tran- The atomic coordinates and structure factors (codes 6RLW) scriptional downregulation of DNA replication genes might have been deposited in the Protein Data Bank (http: be the earliest phenotype associated with TH5487 induced //wwpdb.org). Transcriptome data is deposited at the replication stress in A3 cells. Gene Expression Omnibus with the accession number Taking into consideration the selectivity of TH5487 for GSE155782. The data that support the findings of this study arresting cancer cell proliferation (Figure 3), and that am- are available from the corresponding author upon reason- plification and/or upregulation MCMs members have been able request. identified as biomarkers of progression and negative out- come in several cancer types (68), including OGG1 in- SUPPLEMENTARY DATA hibition in clinical prevention strategies for cancer treat- ment, based on the MCMs expression status, could be Supplementary Data are available at NAR Online. Nucleic Acids Research, 2020, Vol. 48, No. 21 12249 ACKNOWLEDGEMENTS TEF SEP project 102020885 (to T.V.); Vinnova (to A.C.- K and P.S.); the Torsten and Ragnar Söderberg Founda- We thank the scientists at station I04 of the Diamond Light tion (to T.H.) and the Helleday Foundation (to C.B.). Fund- Source, Didcot, Oxfordshire (UK) for their support during ing for open access charge and project support: European data collection (allocation MX15806). We are grateful to Union’s Horizon 2020 research and innovation program un- professors Bjørn Dalhus, Arne Klungland, Julian Walfrids- der the Marie Sklodowska-Curie grant agreement [722729 son and Magnar Bjørås for helpful discussions, Dr Jaromir to A.P., B.M.F.H., T.H.]; European Research Council [ERC Mikes at the Scilifelab FACS facility for sorting RFP670 TAROX-695376 to T.H.]; Swedish Research Council (to positive cells, and Professor Margherita Bignami for the T.H. and P.S.); Swedish Cancer Society (to T.H. and P.S.); kind gift of Ogg1-/- mouse embryonic fibroblasts. Swedish Children’s Cancer Foundation (to T.H.); Swedish Author contributions: T.V., A.C.K. and T.H. conceived and Pain Relief Foundation (to T.H.). initiated the study. T.V. and C.B.B. contributed equally Conflict of interest statement. T.V., A.C.-K., O.W., T.K. and to the study. T.V, C.B-B., A.C.K., M.A., P.S. and T.H. T.H. are listed as inventors on a provisional U.S. patent Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 designed the experiments. A.C.K., T.V. and C.K was re- application no. 62/636983, covering OGG1 inhibitors. The sponsible for project administration at different times. T.V., patent is fully owned by a non-profit public foundation, the A.C.K., C.B.-B. and T.H. wrote the manuscript with assis- Helleday Foundation, and T.H., H.G. and U.W.B. are mem- tance from the other authors. T.V. and C.B.-B. performed bers of the foundation board developing OGG1 inhibitors most of the experiments. K.S. performed Comet assay, and toward the clinic. An inventor reward scheme is under dis- animal experiments together with U.W.B., S.S., C.G., A.R., cussion. The remaining authors declare no competing fi- T.P., S.M. and B.M.F.H. performed Western Blotting to- nancial interests. gether with T.V., J.E.U., A.C.K., M.P. and others. T.B. deter- mined OGG1i EC50 in cells together with C.B.B, T.V., S.M., V.R., S.K., I.A. and E.W., T.V., E.W. and B.M.F.H. per- REFERENCES formed colony formation assays, Fibre assay was performed by O.M. J.M.B. performed isolation of PBMC together with 1. Bartkova,J., Hořejšı́,Z., Koed,K., Krämer,A., Tort,F., Zieger,K., Guldberg,P., Sehested,M., Nesland,J.M., Lukas,C. et al. (2005) DNA S.K. A.P. performed immunostaining of ␥ H2AX. Flow cy- damage response as a candidate anti-cancer barrier in early human tometry experiments were conducted by K.M., S.E, S.K., tumorigenesis. Nature, 434, 864–870. C.B.-B. and V.R., T.V., A.S. and H.E.K. designed and per- 2. Halazonetis,T.D., Gorgoulis,V.G. and Bartek,J. (2008) An formed mass spectrometry experiments, X-ray crystallogra- oncogene-induced DNA damage model for cancer development. Science, 319, 1352–1355. phy was designed, performed and analysed by G.M., E.J.H. 3. Hanahan,D. and Weinberg,R.A. (2011) Hallmarks of cancer: the next and P.S. TH5487 off-target determination was conducted by generation. Cell, 144, 646–674. S.M., C.B.-B. and V.R. Recovery experiments for TH5487 in 4. Negrini,S., Gorgoulis,V.G. and Halazonetis,T.D. (2010) Genomic A3 cells were conducted by C.B.B., V.R. and S.M. OGG1 instability –– an evolving hallmark of cancer. Nat. Rev. Mol. Cell and APE1 protein purification was conducted by O.L. and Biol., 11, 220–228. 5. Jackson,S.P. and Helleday,T. (2016) Drugging DNA repair. Science, A.-S.J. T.V. and E.W. performed in vitro OGG1 activity as- 352, 1178–1179. says. Design and synthesis of OGG1i was done by A.C.K., 6. Bryant,H.E., Schultz,N., Thomas,H.D., Parker,K.M., Flower,D., O.W., T.K., M.S. and M.M. Plasmid design, cloning, trans- Lopez,E., Kyle,S., Meuth,M., Curtin,N.J. and Helleday,T. (2005) fection was conducted by T.V, H.G, J.E.U, J.B. and P.H. Specific killing of BRCA2-deficient tumours with inhibitors of poly(ADP-ribose) polymerase. Nature, 434, 913–917. TH5487 target engagement experiments were performed 7. Farmer,H., McCabe,N., Lord,C.J., Tutt,A.N.J., Johnson,D.A., by T.P., C.K. and A.C.K., CD34+ stem cell isolation and Richardson,T.B., Santarosa,M., Dillon,K.J., Hickson,I., Knights,C. PBMC isolation was conducted by R.T.-R., S.R.-P., J.M.B., et al. (2005) Targeting the DNA repair defect in BRCA mutant cells S.K., J.B. and C.B.-B., C.B.-B., S.E.A. D.H. performed the as a therapeutic strategy. Nature, 434, 917. RNA sequencing. J.J.A. and C.B.B. analyzed the RNA se- 8. Murai,J., Huang,S.N., Das,B.B., Renaud,A., Zhang,Y., Doroshow,J.H., Ji,J., Takeda,S. and Pommier,Y. (2012) Trapping of quencing data. PARP1 and PARP2 by clinical PARP inhibitors. Cancer Res., 72, 5588–5599. 9. Maya-Mendoza,A., Moudry,P., Merchut-Maya,J.M., Lee,M., FUNDING Strauss,R. and Bartek,J. (2018) High speed of fork progression induces DNA replication stress and genomic instability. Nature, 559, S.R.P.’s laboratory is partially funded by funds from the 279–284. Spanish National Research and Development Plan, Insti- 10. Ronson,G.E., Piberger,A.L., Higgs,M.R., Olsen,A.L., Stewart,G.S., McHugh,P.J., Petermann,E. and Lakin,N.D. (2018) PARP1 and tuto de Salud Carlos III and FEDER [PI17/02303 to S.R.- PARP2 stabilise replication forks at base excision repair intermediates P]; R.T.R. is supported by a fellowship from the AECC through Fbh1-dependent Rad51 regulation. Nat. Commun., 9, 1–12. scientific foundation; J.M.B. is supported by Spanish Min- 11. Mansour,W.Y., Rhein,T. and Dahm-Daphi,J. (2010) The alternative istry of Education, Culture and Sport [FPU15/01978]; end-joining pathway for repair of DNA double-strand breaks J.B.’s laboratory is partially funded by FEDER funds, requires PARP1 but is not dependent upon microhomologies. Nucleic Acids Res., 38, 6065–6077. H2020 BRIDGES project and the Spanish Network on 12. Patel,A.G., Sarkaria,J.N. and Kaufmann,S.H. (2011) Rare Diseases (CIBERER) [FIS PI16/00440]; Faculty of Nonhomologous end joining drives poly(ADP-ribose) polymerase Medicine at the Norwegian University of Science and (PARP) inhibitor lethality in homologous recombination-deficient Technology and the Central Norway Regional Health Au- cells. Proc. Natl. Acad. Sci. U.S.A., 108, 3406–3411. 13. Couto,C.A.-M., Wang,H.-Y., Green,J.C.A., Kiely,R., Siddaway,R., thority supports [46056921 to A.S. and H.E.K.]; Svan- Borer,C., Pears,C.J. and Lakin,N.D. (2011) PARP regulates hild and Arne Must’s Fund for Medical Research (to A.S. nonhomologous end joining through retention of Ku at and H.E.K.); Norwegian Research Council (to T.V.); SIN- double-strand breaks. J. Cell Biol., 194, 367–375. 12250 Nucleic Acids Research, 2020, Vol. 48, No. 21 14. Schriewer Jacqueline,M., Bien,P.C., Joseph,B. and Schumacker,P.T. 33. Amente,S., Bertoni,A., Morano,A., Lania,L., Avvedimento,E.V. and (2013) ROS-Mediated PARP activity undermines mitochondrial Majello,B. (2010) LSD1-mediated demethylation of histone H3 lysine function after permeability transition pore opening during 4 triggers Myc-induced transcription. Oncogene, 29, 3691–3702. myocardial ischemia–reperfusion. J. Am. Heart Assoc., 2, e000159. 34. Redstone,S.C.J., Fleming,A.M. and Burrows,C.J. (2019) Oxidative 15. Helleday,T. (2011) The underlying mechanism for the PARP and modification of the potential G-Quadruplex sequence in the PCNA BRCA synthetic lethality: clearing up the misunderstandings. Mol. gene promoter can turn on transcription. Chem. Res. Toxicol., 32, Oncol., 5, 387–393. 437–446. 16. Strom,C.E., Johansson,F., Uhlen,M., Szigyarto,C.A., Erixon,K. and 35. Visnes,T., Grube,M., Hanna,B.M.F., Benitez-Buelga,C., Helleday,T. (2011) Poly (ADP-ribose) polymerase (PARP) is not Cázares-Körner,A. and Helleday,T. (2018) Targeting BER enzymes involved in base excision repair but PARP inhibition traps a in cancer therapy. DNA Repair (Amst.), 71, 118–126. single-strand intermediate. Nucleic Acids Res., 39, 3166–3175. 36. Michel,M., Visnes,T., Homan,E.J., Seashore-Ludlow,B., 17. Brown,J.S., O’Carrigan,B., Jackson,S.P. and Yap,T.A. (2017) Hedenström,M., Wiita,E., Vallin,K., Paulin,C.B.J., Zhang,J., Targeting DNA repair in Cancer: Beyond PARP inhibitors. Cancer Wallner,O. et al. (2019) Computational and experimental Discov., 7, 20–37. druggability assessment of human DNA glycosylases. ACS Omega, 4, 18. Vafa,O., Wade,M., Kern,S., Beeche,M., Pandita,T.K., 11642–11656. Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 Hampton,G.M. and Wahl,G.M. (2002) c-Myc can induce DNA 37. Umar,A., Koi,M., Risinger,J.I., Glaab,W.E., Tindall,K.R., damage, increase reactive oxygen species, and mitigate p53 function: a Kolodner,R.D., Boland,C.R., Barrett,J.C. and Kunkel,T.A. (1997) mechanism for oncogene-induced genetic instability. Mol. Cell, 9, Correction of hypermutability, 1031–1044. N-Methyl-N -nitro-N-nitrosoguanidine resistance, and defective 19. Lee,A.C., Fenster,B.E., Ito,H., Takeda,K., Bae,N.S., Hirai,T., DNA mismatch repair by introducing chromosome 2 into human Yu,Z.-X., Ferrans,V.J., Howard,B.H. and Finkel,T. (1999) Ras tumor cells with mutations in MSH2 and MSH6. Cancer Res., 57, proteins induce senescence by altering the intracellular levels of 3949–3955. reactive oxygen species. J. Biol. Chem., 274, 7936–7940. 38. Eshtad,S., Mavajian,Z., Rudd,S.G., Visnes,T., Boström,J., Altun,M. 20. Maya-Mendoza,A., Ostrakova,J., Kosar,M., Hall,A., Duskova,P., and Helleday,T. (2016) hMYH and hMTH1 cooperate for survival in Mistrik,M., Merchut-Maya,J.M., Hodny,Z., Bartkova,J., mismatch repair defective T-cell acute lymphoblastic leukemia. Christensen,C. et al. (2015) Myc and Ras oncogenes engage different Oncogenesis, 5, e275. energy metabolism programs and evoke distinct patterns of oxidative 39. Visnes,T., Cázares-Körner,A., Hao,W., Wallner,O., Masuyer,G., and DNA replication stress. Mol. Oncol., 9, 601–616. Loseva,O., Mortusewicz,O., Wiita,E., Sarno,A., Manoilov,A. et al. 21. Kotsantis,P., Petermann,E. and Boulton,S.J. (2018) Mechanisms of (2018) Small-molecule inhibitor of OGG1 suppresses oncogene-induced replication stress: jigsaw falling into place. Cancer proinflammatory gene expression and inflammation. Science, 362, Discov., 8, 537–555. 834–839. 22. Matsui,A., Ikeda,T., Enomoto,K., Hosoda,K., Nakashima,H., 40. Berglund,U.W., Sanjiv,K., Gad,H., Kalderén,C., Koolmeister,T., Omae,K., Watanabe,M., Hibi,T. and Kitajima,M. (2000) Increased Pham,T., Gokturk,C., Jafari,R., Maddalo,G., Seashore-Ludlow,B. formation of oxidative DNA damage, 8-hydroxy-2 -deoxyguanosine, et al. (2016) Validation and development of MTH1 inhibitors for in human breast cancer tissue and its relationship to GSTP1 and treatment of cancer. Ann. Oncol., 27, 2275–2283. COMT genotypes. Cancer Lett., 151, 87–95. 41. Picelli,S., Björklund,Å.K., Faridani,O.R., Sagasser,S., Winberg,G. 23. Toyokuni,S., Okamoto,K., Yodoi,J. and Hiai,H. (1995) Persistent and Sandberg,R. (2013) Smart-seq2 for sensitive full-length oxidative stress in cancer. FEBS Lett., 358, 1–3. transcriptome profiling in single cells. Nat. Methods, 10, 1096–1098. 24. Dizdaroglu,M. (2015) Oxidatively induced DNA damage and its 42. Love,M.I., Huber,W. and Anders,S. (2014) Moderated estimation of repair in cancer. Mut. Res./Rev. Mut. Res., 763, 212–245. fold change and dispersion for RNA-seq data with DESeq2. Genome 25. Nakabeppu,Y. (2014) Cellular levels of 8-Oxoguanine in either DNA Biol., 15, 550. or the nucleotide pool play pivotal roles in carcinogenesis and 43. Subramanian,A., Tamayo,P., Mootha,V.K., Mukherjee,S., survival of cancer cells. Int. J. Mol. Sci., 15, 12543–12557. Ebert,B.L., Gillette,M.A., Paulovich,A., Pomeroy,S.L., Golub,T.R., 26. Alexandrov,L.B., Nik-Zainal,S., Wedge,D.C., Aparicio,S.A.J.R., Lander,E.S. et al. (2005) Gene set enrichment analysis: a Behjati,S., Biankin,A.V., Bignell,G.R., Bolli,N., Borg,A., knowledge-based approach for interpreting genome-wide expression Børresen-Dale,A.-L. et al. (2013) Signatures of mutational processes profiles. Proc. Natl. Acad. Sci. U.S.A., 102, 15545–15550. in human cancer. Nature, 500, 415–421. 44. Gu,Z., Eils,R. and Schlesner,M. (2016) Complex heatmaps reveal 27. Sakumi,K., Tominaga,Y., Furuichi,M., Xu,P., Tsuzuki,T., patterns and correlations in multidimensional genomic data. Sekiguchi,M. and Nakabeppu,Y. (2003) Ogg1 knockout-associated Bioinformatics, 32, 2847–2849. lung tumorigenesis and its suppression by Mth1 gene disruption. 45. Luo,W. and Brouwer,C. (2013) Pathview: an R/Bioconductor Cancer Res., 63, 902–905. package for pathway-based data integration and visualization. 28. Ramdzan,Z.M., Vadnais,C., Pal,R., Vandal,G., Cadieux,C., Bioinformatics, 29, 1830–1831. Leduy,L., Davoudi,S., Hulea,L., Yao,L., Karnezis,A.N. et al. (2014) 46. O’Callaghan,N., Baack,N., Sharif,R. and Fenech,M. (2011) A RAS Transformation Requires CUX1-Dependent Repair of qPCR-based assay to quantify oxidized guanine and other Oxidative DNA Damage. PLoS Biol., 12, e1001807. FPG-sensitive base lesions within telomeric DNA. BioTechniques, 51, 29. Romanowska,M., Maciag,A., Smith,A.L., Fields,J.R., 403–412. Fornwald,L.W., Kikawa,K.D., Kasprzak,K.S. and Anderson,L.M. 47. Molina,D.M., Jafari,R., Ignatushchenko,M., Seki,T., Larsson,E.A., (2007) DNA damage, superoxide, and mutant K-ras in human lung Dan,C., Sreekumar,L., Cao,Y. and Nordlund,P. (2013) Monitoring adenocarcinoma cells. Free Radic. Biol. Med., 43, 1145–1155. drug target engagement in cells and tissues using the cellular thermal 30. Pastukh,V., Roberts,J.T., Clark,D.W., Bardwell,G.C., Patel,M., shift assay. Science, 341, 84–87. Al-Mehdi,A.-B., Borchert,G.M. and Gillespie,M.N. (2015) An 48. Gorrec,F. (2009) The MORPHEUS protein crystallization screen. J. oxidative DNA “damage” and repair mechanism localized in the Appl. Crystallogr., 42, 1035–1042. VEGF promoter is important for hypoxia-induced VEGF mRNA 49. Winter,G. (2010) xia2: an expert system for macromolecular expression. Am. J. Physiol. Lung Cell Mol. Physiol., 309, crystallography data reduction. J. Appl. Cryst. J. Appl. Crystallogr., L1367–L1375. 43, 186–190. 31. Cogoi,S., Ferino,A., Miglietta,G., Pedersen,E.B. and Xodo,L.E. 50. Parkhurst,J.M., Winter,G., Waterman,D.G., Fuentes-Montero,L., (2017) The regulatory G4 motif of the Kirsten ras (KRAS)gene is Gildea,R.J., Murshudov,G.N. and Evans,G. (2016) Robust sensitive to guanine oxidation: implicationson transcription. Nucleic background modelling in DIALS. J. Appl. Crystallogr., 49, Acids Res., 46, 661–676. 1912–1921. 32. Pan,L., Hao,W., Zheng,X., Zeng,X., Abbasi,A.A., Boldogh,I. and 51. Evans,P. (2006) Scaling and assessment of data quality. Acta Ba,X. (2017) OGG1-DNA interactions facilitate NF-␬B binding to Crystallogr. D Biol. Crystallogr., 62, 72–82. DNA targets. Sci. Rep., 7, 43297. 52. Collaborative Computational Project, Number 4 (1994) The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D Biol. Crystallogr., 50, 760–763. Nucleic Acids Research, 2020, Vol. 48, No. 21 12251 53. McCoy,A.J., Grosse-Kunstleve,R.W., Adams,P.D., Winn,M.D., during pseudomonas aeruginosa infection. J. Immunol., 205, Storoni,L.C. and Read,R.J. (2007) Phaser crystallographic software. 2231–2242. J. Appl. Crystallogr., 40, 658–674. 62. Sampath,H. (2014) Oxidative DNA damage in disease––insights 54. Murshudov,G.N., Skubák,P., Lebedev,A.A., Pannu,N.S., gained from base excision repair glycosylase-deficient mouse models. Steiner,R.A., Nicholls,R.A., Winn,M.D., Long,F. and Vagin,A.A. Environ. Mol. Mutagen., 55, 689–703. (2011) REFMAC5 for the refinement of macromolecular crystal 63. Ding,Y., Fleming,A.M. and Burrows,C.J. (2017) Sequencing the structures. Acta Crystallogr. D. Biol. Crystallogr., 67, 355–367. mouse genome for the oxidatively modified base 55. Emsley,P., Lohkamp,B., Scott,W.G. and Cowtan,K. (2010) Features 8-oxo-7,8-dihydroguanine by OG-Seq. J. Am. Chem. Soc., 139, and development of Coot. Acta Crystallogr. D. Biol. Crystallogr., 66, 2569–2572. 486–501. 64. Gorini,F., Scala,G., Di Palo,G., Dellino,G.I., Cocozza,S., Pelicci,P.G., 56. Chen,V.B., Arendall,W.B., Headd,J.J., Keedy,D.A., Lania,L., Majello,B. and Amente,S. (2020) The genomic landscape of Immormino,R.M., Kapral,G.J., Murray,L.W., Richardson,J.S. and 8-oxodG reveals enrichment at specific inherently fragile promoters. Richardson,D.C. (2010) MolProbity: all-atom structure validation for Nucleic Acids Res., 48, 4309–4324. macromolecular crystallography. Acta Crystallogr. D. Biol. 65. Amente,S., Di Palo,G., Scala,G., Castrignanò,T., Gorini,F., Crystallogr., 66, 12–21. Cocozza,S., Moresano,A., Pucci,P., Ma,B., Stepanov,I. et al. (2018) Genome-wide mapping of 8-oxo-7,8-dihydro-2 -deoxyguanosine Downloaded from https://academic.oup.com/nar/article/48/21/12234/5992293 by guest on 12 February 2021 57. Hahn,W.C., Counter,C.M., Lundberg,A.S., Beijersbergen,R.L., Brooks,M.W. and Weinberg,R.A. (1999) Creation of human tumour reveals accumulation of oxidatively-generated damage at DNA cells with defined genetic elements. Nature, 400, 464–468. replication origins within transcribed long genes of mammalian cells. 58. Lin,A., Giuliano,C.J., Palladino,A., John,K.M., Abramowicz,C., Nucleic Acids Res., 47, 221–236. Yuan,M.L., Sausville,E.L., Lukow,D.A., Liu,L., Chait,A.R. et al. 66. Bai,G., Smolka,M.B. and Schimenti,J.C. (2016) Chronic DNA (2019) Off-target toxicity is a common mechanism of action of cancer replication stress reduces replicative lifespan of cells by drugs undergoing clinical trials. Sci. Transl. Med., 11, eaaw8412. TRP53-Dependent, microRNA-assisted MCM2-7 downregulation. 59. Ba,X., Bacsi,A., Luo,J., Aguilera-Aguirre,L., Zeng,X., Radak,Z., PLoS Genet., 12, e1005787. Brasier,A.R. and Boldogh,I. (2014) 8-Oxoguanine DNA 67. Alvarez,S., Dı́az,M., Flach,J., Rodriguez-Acebes,S., Glycosylase-1 augments proinflammatory gene expression by López-Contreras,A.J., Martı́nez,D., Cañamero,M., facilitating the recruitment of site-specific transcription factors. J. Fernández-Capetillo,O., Isern,J., Passegué,E. et al. (2015) Replication Immunol., 192, 2384–2394. stress caused by low MCM expression limits fetal erythropoiesis and 60. Tahara,Y., Auld,D., Ji,D., Beharry,A.A., Kietrys,A.M., Wilson,D.L., hematopoietic stem cell functionality. Nat. Commun., 6, 8548. Jimenez,M., King,D., Nguyen,Z. and Kool,E.T. (2018) Potent and 68. Wang,Y., Chen,H., Zhang,J., Cheng,A.S.L., Yu,J., To,K.F. and selective inhibitors of 8-oxoguanine DNA glycosylase. J. Am. Chem. Kang,W. (2020) MCM family in gastrointestinal cancer and other Soc., 140, 2105–2114. malignancies: from functional characterization to clinical 61. Qin,S., Lin,P., Wu,Q., Pu,Q., Zhou,C., Wang,B., Gao,P., Wang,Z., implication. Biochim. Biophys. Acta Rev. Cancer, 1874, 188415. Gao,A., Overby,M. et al. (2020) Small-molecule inhibitor of 8-oxoguanine DNA glycosylase 1 regulates inflammatory responses