Replication Protein a Availability During DNA Replication Stress Is A

Published OnlineFirst August 2, 2018; DOI: 10.1158/0008-5472.CAN-18-0618

Cancer Molecular Cell Biology Research

Replication Protein A Availability during DNA Replication Stress Is a Major Determinant of Cisplatin Resistance in Ovarian Cancer Cells Francois¸ Belanger 1, Emile Fortier1,2, Maxime Dube1,2, Jean-Francois¸ Lemay1, Remi Buisson3, Jean-Yves Masson3, Abdelhamid Elsherbiny1,2, Santiago Costantino1,4, Euridice Carmona5, Anne-Marie Mes-Masson2,5,6, Hugo Wurtele1,2,6, and Elliot Drobetsky1,2,6

Abstract

Intrinsic and acquired resistance to cisplatin remains a pri- via NER, specifically during S phase. Such defects in NER were mary hurdle to treatment of high-grade serous ovarian cancer attributable to RPA exhaustion arising from aberrant activation (HGSOC). Cisplatin selectively kills tumor cells by inducing of DNA replication origins during replication stress. Reduced DNA crosslinks that block replicative DNA polymerases. Single- RPA availability promoted Mre11-dependent degradation of stranded DNA (ssDNA) generated at resulting stalled replica- nascent DNA at stalled RF in cell lines exhibiting elevated tion forks (RF) is bound and protected by heterotrimeric sensitivity to cisplatin. Strikingly, defective S-phase NER, RF replication protein A (RPA), which then serves as a platform instability, and cisplatin sensitivity could all be rescued by for recruitment and activation of replication stress response ectopic overexpression of RPA. Taken together, our findings factors. Cells deficient in this response are characterized by indicate that RPA exhaustion represents a major determinant of extensive ssDNA formation and excessive RPA recruitment that cisplatin sensitivity in HGSOC cell lines. exhausts the available pool of RPA, which (i) inhibits RPA- dependent processes such as nucleotide excision repair (NER) Significance: The influence of replication protein A exhaus- and (ii) causes catastrophic failure of blocked RF. Here, we tion on cisplatin sensitivity harbors important implications investigated the influence of RPA availability on chemosensi- toward improving therapy of various cancers that initially tivity using a panel of human HGSOC cell lines. Our data respond to platinum-based agents but later relapse due to revealed a striking correlation among these cell lines between intrinsic or acquired drug resistance. Cancer Res; 78(19); 5561–73. cisplatin sensitivity and the inability to efficiently repair DNA 2018 AACR.

Introduction followed by combination chemotherapy with a platinum-based drug such as cisplatin (CDDP). This regimen yields a high initial High-grade serous ovarian cancer (HGSOC) is the most lethal response rate, often resulting in clinical remission. However, most gynecologic malignancy, with a 5-year survival rate of approxi- patients eventually relapse, at which point their cancer manifests mately 50% (1). Frontline treatment consists of debulking surgery strong resistance to CDDP (2). The therapeutic efficacy of CDDP is attributable to the capacity of its platinum atom to form covalent 1Centre de Recherche de l'Hopital^ Maisonneuve-Rosemont, Montreal, Canada. bonds involving the N7 position of purines in DNA generating 2Programme de Biologie Moleculaire, Universite de Montreal, Montreal, Canada. 1,2- and 1,3-intrastrand crosslinks (98%) in addition to a much 3CHU de Quebec Research Center, Oncology Axis, and Laval University Cancer lower yield (2%) of interstrand crosslinks (3). These adducts Research Center, Quebec City, Canada. 4Departement d'Ophtalmologie, Uni- strongly block the progression of DNA polymerases, inducing a 5 versite de Montreal, Montreal, Canada. Centre de Recherche du Centre Hospi- state of "replication stress" that selectively promotes the elimi- talier de l'Universite de Montreal, and Institut du Cancer de Montreal, Montreal, nation of rapidly proliferating cancer cells (4). Canada. 6Departement de Medecine, Universite de Montreal, Montreal, Canada. The mechanism of clinical resistance to CDDP is complex and Note: Supplementary data for this article are available at Cancer Research remains incompletely understood. The cytotoxicity of CDDP can Online (http://cancerres.aacrjournals.org/). be reduced by processes that counter its ability to react with DNA, F. Belanger and E. Fortier contributed equally to this article. including increased efflux/reduced uptake and inactivation by Current address for R. Buisson: Massachusetts General Hospital Cancer Center, sulfur-containing molecules (5). In addition, CDDP-induced Harvard Medical School, Boston, MA. intrastrand crosslinks are removed by nucleotide excision repair fi Corresponding Authors: Elliot Drobetsky, Centre de Recherche de l'Hopital^ (NER; ref. 6), and cells de cient in this pathway are sensitive to the Maisonneuve-Rosemont, 5415 boul de l'Assomption, Montreal H1T 2M4, drug (7, 8). Indeed, based on various lines of evidence, increased Canada. Phone: 514-252-3400; Fax: 514-252-3430; E-mail: NER capacity has been proposed as an important mechanism of [email protected]; and Hugo Wurtele, Centre de Recherche de CDDP resistance in ovarian cancer (5). ^ l'Hopital Maisonneuve-Rosemont, 5415 Boulevard de l'Assomption, Montreal In addition to mechanisms outlined above that remove DNA H1T 2M4, Canada. Fax: 514-252-3430; E-mail: [email protected] lesions or inhibit their formation, modulation of the cellular doi: 10.1158/0008-5472.CAN-18-0618 response to replication stress, i.e., pathways that contribute to 2018 American Association for Cancer Research. the stabilization/resolution of stalled replication forks (RF), is

www.aacrjournals.org 5561

Belanger et al.

emerging as a key determinant of CDDP resistance in cancer (9). FBS, L-glutamine, and antibiotics. All cell lines were used within For example, germline mutations in BRCA1 or BRCA2 greatly 20 passages after thawing and routinely tested for mycoplasma increase susceptibility to ovarian and breast tumors (10). contamination by DAPI staining/fluorescence microscopy. Although BRCA1/2 were initially characterized for their key roles in DNA double-strand break (DSB) repair by homologous recom- Ectopic RPA expression and siRNA treatment bination (HR), recent data reveal that these proteins also stabilize The RPA expression plasmid (pAC-GFP-RPA; ref. 16), a generous stalled RF by inhibiting Mre11-dependent nucleolytic degrada- gift of Dr. J. Lukas (University of Copenhagen, Copenhagen, tion of nascent DNA (11, 12). Moreover, sensitivity to replication- Denmark), was transfected using Lipofectamine 2000 (Thermo- blocking drugs such as CDDP was recently shown to correlate Fisher). Stable clones were sorted by FACS in 96-well plates and strongly with the inability of BRCA1/2-defective cells to protect selected with 500 mg/mL Geneticin (ThermoFisher). siRNA smart- nascent DNA from nuclease activity (12). pools for knockdown of RPA1 and BRCA1 were purchased from Following exposure to replication stress–inducing drugs, Dharmacon and transfected using RNAiMax (ThermoFisher). uncoupling of DNA polymerases from the MCM helicase complex Pools of nontargeting (NT) duplexes were used as controls. at blocked RF generates regions of single-stranded DNA (ssDNA), which are rapidly coated by heterotrimeric replication protein A Cell irradiation and drug treatment (RPA; ref. 13). RPA-ssDNA acts as a recruitment/activation plat- Cell monolayers were irradiated with 254-nm UV (hereafter form for DNA damage response (DDR) factors that mitigate UV) at a fluence rate of 0.4 J/m2/s, or with ionizing radiation (dose replication stress, notably the apical kinase ataxia telangiectasia rate of 5 10 2 Gy/s) as previously described (24). The ATR and Rad3-related (ATR; ref. 14). Activated ATR phosphorylates inhibitor VE-821 (Cedarlane), hydroxyurea (HU; Bioshop multiple substrates, which cooperate to (i) prevent further origin Canada), cyclin-dependent kinase inhibitor roscovitine (Abcam), activation, (ii) stabilize stalled RF, and (iii) block cell-cycle or MRE11 inhibitor mirin (Sigma-Aldrich) were added to cell progression. Abrogation of ATR activity during replication stress monolayers in fresh media. leads to unrestricted firing of DNA replication origins, excessive accumulation of ssDNA, and greatly increased levels of chroma- NER assay tin-bound RPA (15). This eventually results in global exhaustion Removal of 6-4 pyrimidine-pyrimidone photoproducts of cellular RPA and extensive DSB formation at persistently stalled (6-4PP) as a function of cell cycle was quantified as described RF, a process recently termed "replication catastrophe" (16). (19). For repair determinations in the presence of roscovitine, However, the precise mechanisms leading to RF instability in cells were treated with 20 mmol/L of the drug (or mock treated) for response to RPA exhaustion, and their potential influence on 2 or 72 hours before the assay and during post-UV incubations. cancer chemoresistance, remain incompletely characterized. Under conditions of limited RPA availability, important RPA- dependent processes aside from DNA replication can be compro- Clonogenic survival mised, which might contribute to increased RF stalling/failure. Cells were seeded on 100 mm dishes and treated with CDDP Indeed, we and others have shown, in various mutant yeast strains (Cedarlane) or mitomycin C (MMC; Abcam) for 2 hours in and mammalian cell lines exhibiting replication stress response serum-free medium. Camptothecin (CPT; Bioshop Canada) treat- defects, that sequestration of RPA at stalled RF exhausts the ment was carried out for 2 hours in complete medium. Hydroxy- cellular pools of this complex, thus interfering with its ability to urea was added for the duration of the experiment. Clonogenic perform essential functions in NER during S phase (17, 18). survival was evaluated as previously described (19). Moreover, we previously demonstrated that a majority of human melanoma cell lines exhibit striking defects in NER exclusively Immunoblotting during S, although the molecular basis was not thoroughly Western blotting was performed with whole-cell extracts using investigated (19). Here, we probed the impact of RPA exhaustion standard protocols and the following antibodies: mouse anti- on DNA repair and RF stability, thereby elucidating the basis of RPA2 (Calbiochem; NA18), rat anti-tubulin (Abcam; ab6161), chemoresistance, in ovarian cancer cells. rabbit anti-RPA1 (Abcam; ab79398), mouse anti-RPA3 (Abcam; ab6432), rabbit anti-RPA2 phospho S33 (Abcam; ab211877), rabbit anti-Chk1 phospho S345 (Cell Signaling Technology; Materials and Methods #2348), mouse anti-Chk1 (Cell Signaling Technology; #2360), Cell culture mouse anti-BRCA1 (Santa Cruz Biotechnology; sc-6954), mouse Ovarian cancer cell lines were established from patient tumors anti-GAPDH (Santa Cruz Biotechnology; sc-365062), and rabbit (TOV) or ascites (OV) as described (20–23), and cultured in OSE anti-YY1 (Santa Cruz Biotechnology; sc-1703). medium (Wisent) supplemented with 10% FBS, L-glutamine, and antibiotics (ThermoFisher). U2OS osteosarcoma cells were pur- Quantification of CDDP-induced DNA adducts chased from the ATCC and cultured in DMEM (Corning) supple- Immediately after treatment with 20 mmol/L CDDP (2 hours mented with 10% FBS, L-glutamine, and antibiotics. U2OS and all in serum-free medium), genomic DNA was extracted in SNET ovarian cancer cell lines were authenticated in 2018 using short buffer (20 mmol/L Tris-HCl, pH 8, 400 mmol/L NaCl, 1% SDS, tandem repeat (STR) profiling by the McGill University Genome 5 mmol/L EDTA, and 0.4 mg/mL proteinase K), followed by Center (Montreal, Canada). The model ovarian cancer cell line phenol/chloroform extraction and RNase A (Qiagen) treatment. A2780 and its isogenic CDDP-resistant counterpart A2780Cis, both DNA was quantified with PicoGreen (ThermoFisher) using a of which were authenticated by the distributor (Sigma-Aldrich) by TBS-380 fluorimeter (Turner Biosystems). Note that 100 ng of STR profiling and used within 1 month following purchase, were DNA was analyzed by slot blotting with rat anti–cisplatin- grown in RPMI-1640 (Sigma-Aldrich) supplemented with 10% modified DNA antibody (Abcam, ab103261) as described (18).

5562 Cancer Res; 78(19) October 1, 2018 Cancer Research

RPA Availability and Chemoresistance

Immunofluorescence microscopy 3 hours. The CldU/IdU ratio of each fiber was normalized by Cells were seeded on glass slides in 6-well dishes and irradiat- dividing by the mean of CldU/IdU ratios observed immediately ed with UV or mock-treated. Posttreatment incubations were after CldU labeling (time 0). Imaging was performed using a carried out in the presence of 30 mmol/L BrdUrd (Sigma-Aldrich). DeltaVision Elite system (GE Healthcare) in conjunction with FIJI RPA and BrdUrd immunofluorescence were evaluated as de- software (NIH). Experiments were performed at least twice inde- scribed (18) using mouse anti-RPA2 (Calbiochem; NA18) and pendently, and a minimum of 125 fibers were counted for each rat anti-BrdUrd (Abcam; ab6326). Secondary antibodies were experiment (250 fibers total per condition). goat anti–mouse-Alexa488 (ThermoFisher, A10667) and chicken anti–rat-Alexa594 (ThermoFisher; A21471). Images were acquir- Pulsed-field gel electrophoresis ed using a DeltaVision Elite system (GE Healthcare). Intensity of Cells were treated with 10 mmol/L CDDP in complete med- nuclear RPA2 signals was evaluated using custom software, as ium for 24 hours, trypsinized, washed, and resuspended in PBS. described (18). At least 200 cells combined from 2 independent Agarose plugs composed of low melting point agarose (1% w/v) experiments were quantified for each condition. in PBS and containing 500,000 cells were prepared using a CHEF disposable plug mold (Bio-Rad). The plugs were digested with a Quantification of EdU foci by fluorescence microscopy solution of 1 mg/mL proteinase K, 100 mmol/L EDTA, 0.2% (w/v) Cells were seeded on glass slides in 6-well dishes and irradiated sodium deoxycholate, and 1% (w/v) N-lauroyl sarcosinate at with 20 J/m2 of UV or mock-irradiated, followed by 1-hour 37C with low agitation for 2 days. Plugs were then washed 3 incubation in complete media and then 30-minute incubation times for 1 hour with 20 mmol/L Tris HCl (pH 8.0), 50 mmol/L in the presence of 50 mmol/L EdU. Cells were then washed with EDTA, and once with 0.5X TBE for 20 minutes. Pulsed-field PBS and fixed with 3% formaldehyde in PBS on ice for 20 minutes. gel electrophoresis was performed in a 0.9% (w/v) agarose gel Samples were washed with cold PBS, permeabilized on ice for 10 in 0.5X TBE at 14C under the following conditions. Block I: minutes with 0.5% triton X-100 in PBS, and washed again with 12 hours at 5.5 V/cm, 0.1 to 30 seconds at 120C. Block II: PBS. Labeling was performed using click chemistry in PBS þ 2 12 hours at 3.6 V/cm, 0.1 to 5 seconds at 120C. The gel was mmol/L CuSO4,1mmol/L Alexa Fluor 647 Azide, and 10 mmol/L partially dried at room temperature for 30 minutes and then at L-ascorbate for 30 minutes at room temperature. Samples were 55 C for 30 minutes before staining overnight with SYBR Green washed with PBS and then labeled with 10 mg/mL DAPI in PBS. (ThermoFisher) diluted in 0.5X TBE. Image acquisition was per- Images were acquired using a DeltaVision Elite system (GE formed using a ChemiDoc imaging system (Bio-Rad). Healthcare) with a 100x objective and deconvoluted using soft- WoRx 7.0.0. The intensity of nuclear EdU signals and number of foci were quantified using custom software as described previ- Results ously (25). At least 200 cells combined from 2 independent Cisplatin resistance correlates with NER efficiency during S experiments were quantified for each condition. Statistical anal- phase in ovarian cancer cell lines ysis (Student t test) was used to compare the fold change in Our overarching goal was to investigate the potential influence þ – number of foci after UV relative to mock between SPR and SPR of replication stress–induced RPA exhaustion on CDDP resistance cell lines. Because different cell lines vary in the average size of in HGSOC cells. We employed a panel of cell lines including eight their nucleus, the number of foci per cell was normalized to an HGSOC (OV866(2), TOV2223G, OV1369(R2), OV90, OV3331, equal arbitrary surface for easier visualization on a box plot. (This OV1946, TOV2835EP, and TOV3041G) and two "non HGSOC" does not affect the statistical analysis above.) cell lines, i.e., one clear cell adenocarcinoma (TOV21G), and one endometrioid adenocarcinoma (TOV112D); see Supplementary Dual detection of gH2AX and RPA following CDDP treatment Table S1 for cell line characterization). Clonogenic survival assays by flow cytometry revealed a broad spectrum of CDDP sensitivities among cell lines Exponentially growing cultures were treated or mock-treated (Fig. 1A), with LD50 differing by as much as 180-fold, ranging with 5 mmol/L CDDP for 2 hours in serum-free medium, washed from 0.049 mmol/L for TOV21G to 9 mmol/L for TOV112D (Fig. with PBS, and incubated for 24 hours. Following immunolabel- 1A; Supplementary Table S2). As noted earlier, well-studied ing using anti-RPA1 (Abcam; ab79398) and anti–gH2AX Ser139 mechanisms contributing to CDDP resistance act by preventing (Millipore), flow cytometry was performed as described (16). the generation of DNA damage. We therefore quantified the initial induction of CDDP-induced intrastrand crosslinks in our HGSOC DNA fiber analysis cell lines by slot blot (Fig. 1B, left) and found no correlation with DNA fiber assays were performed as described (12). Briefly, for drug sensitivity (Fig. 1B, right). This suggests that the ability to DNA RF progression, cells were labeled for 15 minutes with respond to DNA damage, as opposed to mechanisms that prevent 5 mmol/L IdU, washed, UV-irradiated, and then labeled for 90 its induction, might play a primary role in chemoresistance minutes with 25 mmol/L CldU. For RF progression in the presence among cell lines of our panel. of HU, following labeling for 15 minutes with 5 mmol/L IdU, cells As mentioned earlier, defective responses to DNA replication were washed whereupon 0.2 mmol/L HU was added together stress cause excessive RPA sequestration at stalled RF, which with 25 mmol/L CldU for a further 60-minute incubation. For RF reduces the availability of this complex for NER specifically during progression experiments, the CldU/IdU ratio of each fiber in S (hereafter, S phase-specific nucleotide excision repair is referred treated samples was normalized by dividing by the mean of to as SPR, and SPR-deficient and -proficient cell lines are denoted – þ CldU/IdU ratios observed after mock treatment. For RF protection SPR and SPR , respectively). Because NER is an important deter- assays, cells were labeled with 5 mmol/L IdU for 20 minutes minant of CDDP sensitivity, we employed a flow cytometry– followed by 25 mmol/L CldU for 20 minutes. After washing, cells based immunoassay (26) to evaluate the efficiency of this path- were incubated with 30 mmol/L HU 100 mmol/L mirin for way as a function of cell cycle in our cell lines. This assay has

www.aacrjournals.org Cancer Res; 78(19) October 1, 2018 5563

Belanger et al.

100 A B P > 0.3 10 TOV2223G OV1369(R2) CDDP (mmol/L) 020 TOV112D OV866(2) 8 OV1369(R2) 10 OV866(2) OV866(2) OV1369(R2) m mol/L) ( TOV2223G TOV2223G 6 OV90 OV1946 OV90 OV1946 OV90 OV3331 4 1 TOV2835EP TOV2835EP OV3331 TOV3041G Relative survival (%) 2 TOV3041G OV3331 Cisplatin LD50 TOV2835EP TOV21G OV1946 TOV3041G 0.1 0 0 5101520 0 246 Relative induction of CDDP-DNA Cisplatin concentration (μmol/L)

C D 100 * G1 * * No UV UV 0 h UV 6 h S 2 80 10 G S G G * 1 2 2 *

101 60 * 6-4PP OV866(2) 0 * 10 40 G1 G2 G1 G2 G1 G2 102 20 101 remaining at 6 h (%) 6-4PP 6-4PP

TOV3041G 0 100 G1 G2 G1 G2 G1 G2

OV90

DNA content A2780

OV1946 OV3331

TOV21G

OV866(2) A2780Cis

TOV112D

TOV2223G TOV3041G

OV1369(R2) TOV2835EP HGSOC Non Isogenic HGSOC pair

9 10 R2 = 0.92 R2 = 0.65 E TOV112D F OV866(2) P < 0.0001 P < 0.006 8 OV1369(R2) OV1369(R2) ) 2 m mol/L) OV866(2) 6 ( 6 TOV2223G TOV2223G TOV3041G OV90 TOV112D OV90 4 3 OV1946

OV3331 UV LD50 (J/m OV3331 Cisplatin LD50 2 TOV2835EP SPR+ TOV21G SPR+ SPR− TOV2835EP SPR− TOV21G OV1946 0 TOV3041G 0 0 20 40 60 80 100 0 20 40 60 80 100 6-4PP remaining at 6 h (%) 6-4PP remaining at 6 h (%)

Figure 1. Cisplatin resistance correlates with SPR efficiency in ovarian cancer lines. A, Sensitivity to CDDP was measured by clonogenic survival. Values represent the mean SEM of three independent experiments. B, Cisplatin LD50s do not correlate with initial DNA damage induction as analyzed by slot blot (see text for details). Band intensities from three independent experiments were quantified by densitometry. C, Flow cytometry–based assay for measurement of 6-4PP excision as a function of cell cycle. Representative bivariate dot plot of 6-4PP versus DNA content (propidium iodide) for SPR-proficient OV866(2) and SPR-deficient TOV3041G. D, 6-4PP repair as a function of cell cycle at 6 hours post-UV in 12 ovarian cancer cell lines. Cells are considered SPR-deficient when SPR is significantly slower relative to other phases (, P < 0.01; two-tailed paired t test). E, Cisplatin LD50s plotted against the percentage of unrepaired 6-4PP at 6 hours post-UV. F, UV LD50s were plotted against percentage of unrepaired 6-4PP. P value, F test of the linear regression.

5564 Cancer Res; 78(19) October 1, 2018 Cancer Research

RPA Availability and Chemoresistance

been fully optimized to quantify removal of UV-induced 6-4PPs; tions for UV and CDDP, NER plays only a limited role in removing however, we emphasize that results on 6-4PPs are extrapolatable the primary genotoxic adduct. The bifunctional alkylating agent to CDDP-induced intrastrand crosslinks because removal of MMC induces mainly DNA interstrand crosslinks resolved by the either adduct is dependent on NER. We found that five HGSOC Fanconi anemia pathway in collaboration with HR (28), and the – cell lines were SPR , i.e., exhibiting profound defects in NER topoisomerase I inhibitor CPT generates replication-associated during S versus G0–G1 or G2–M (Fig. 1C and D), whereas three DSB repaired by HR and nonhomologous end-joining (29). Sen- repaired efficiently in all phases. We also characterized the two sitivity to either MMC or CPT correlated significantly with SPR þ "non-HGSOC" cell lines, one of which was SPR and the other deficiency (Fig. 2A and B; Supplementary Fig. S1B–S1C; Supple- – – SPR (Fig. 1D). Remarkably, all SPR cell lines were significantly mentary Table S2); however, no correlation was noted for ionizing þ more sensitive to CDDP versus SPR counterparts (Fig. 1E; radiation, which induces DSB in all cell-cycle phases and does not Supplementary Table S2). We also measured NER efficiency in kill cells primarily by causing replication stress (Fig. 2C; Supple- A2780 and A2780Cis, an isogenic pair of ovarian cancer lines mentary Fig. S1D; Supplementary Table S2). Moreover, SPR defi- differing in CDDP sensitivity (27), and found that CDDP-resis- ciency correlated with sensitivity to HU, which causes replication tant A2780Cis exhibited significantly more efficient SPR versus stress without directly inducing DNA damage (Fig. 2D; Supple- the parental CDDP-sensitive counterpart A2780 (Fig. 1D). mentary Fig. S1E; Supplementary Table S2). The above results Because we evaluated NER capacity by quantifying repair of suggest that defective SPR, while presumably contributing to UV-induced DNA adducts, we also determined clonogenic sur- enhanced CDDP/UV sensitivity in ovarian cancer cells, might vival in response to this model mutagen and found a significant represent only one consequence of abnormal DNA replication correlation between UV LD50 and SPR capacity (Fig. 1F; Supple- stress responses among others that confer such sensitivity. mentary Fig. S1A and Supplementary Table S2). The above data indicate that SPR defects are frequent in HGSOC cell lines and SPR-deficient CDDP-sensitive HGSOC cell lines exhibit correlate strongly with CDDP and UV sensitivity. abnormal DNA replication dynamics We also evaluated sensitivity of our HGSOC cell lines to repli- Defective responses to replication stress often lead to excessive cation stress–inducing drugs for which, compared with the situa- recruitment of RPA at stalled RF (17, 18). We measured

A R2 = 0.68 B R2 = 0.75 P < 0.006 P < 0.01 1,000 1,000 OV1369(R2) OV1369(R2) OV866(2) OV90 TOV2223G OV90 TOV112D OV3331 OV866(2) TOV2223G TOV2835EP OV3331 100 100 TOV2835EP OV1946

TOV3041G OV1946 MMC LD50 (nmol/L) SPR+ SPR+ - - TOV3041G SPR TOV21G SPR

10 Camptothecin LD50 (nmol/L) 10 Figure 2. 0 20 40 60 80 100 0 20 40 60 80 100 ﬁ SPR ef ciency correlates with 6-4PP remaining at 6 h (%) 6-4PP remaining at 6 h (%) sensitivity to MMC, CPT, and HU, but not ionizing radiation, in ovarian cancer cell lines. The percentage of 6-4PP remaining at 6 hours post-UV was plotted against the LD50 of MMC R2 R2 (A), CPT (B), ionizing radiation (C), C = 0.017 D = 0.62 and HU (D). P value, F test of the linear P > 0.7 P < 0.02 regression. 2.0 180 OV1369(R2) OV1369(R2) TOV2835EP 150 OV866(2) 1.5 OV90 OV3331 OV90 120 TOV112D TOV2223G TOV3041G OV1946 μ mol/L) 1.0 OV866(2) 90 TOV2223G OV1946 TOV3041G 60 IR LD50 (Gy) 0.5 TOV21G

+ HU LD50 ( + SPR 30 SPR TOV2835EP SPR- SPR- 0 0 0 20 40 60 80 100 0 20 40 60 80 100

6-4PP remaining at 6 h (%) 6-4PP remaining at 6 h (%)

www.aacrjournals.org Cancer Res; 78(19) October 1, 2018 5565

Belanger et al.

– þ chromatin-bound RPA at 3 hours post-UV in 3 SPR and 3 SPR flow cytometry assay (16), we found that 24 hours after CDDP – HGSOC lines using immunofluorescence microscopy. We treatment, the proportion of SPR cell lines displaying elevated observed significantly increased chromatin-bound RPA during signals for both chromatin-bound RPA and gH2AX (a marker of þ – þ þ S (BrdUrd cells) in SPR cell lines compared with SPR counter- DSB) was significantly higher in all cases compared with SPR parts (Fig. 3A); moreover, this could not be attributed to varia- counterparts (Fig. 4A). We also used pulsed-field gel electropho- tions in RPA expression (Fig. 3B). We next evaluated RF progres- resis to directly evaluate generation of DSB 24 hours after CDDP sion post-UV using DNA fiber analysis and observed significantly treatment (Fig. 4B). In agreement with the aforementioned result, – reduced incorporation of labeled nucleoside analogs (CldU) SPR cell lines manifested significantly higher DSB induction – þ þ post-UV in SPR versus SPR lines (Fig. 3C). These results suggest compared with SPR counterparts. Taken together, our results that elevated recruitment of RPA to chromatin, reflecting abnor- indicate that CDDP sensitivity in HGSOC cell lines strongly mal DNA replication dynamics during genotoxic stress, might correlates with abnormal activation of replication origins, SPR underlie defective SPR in CDDP-sensitive HGSOC cell lines. defects, and RF instability upon replication stress. Increased frequency of stalled RF due to aberrant activation of – þ replication origins, e.g., upon abrogation of ATR activity, can Nascent DNA at stalled RF is unstable in SPR versus SPR cause RPA exhaustion (16). Moreover, other carcinogenic HGSOC cell lines mechanisms including oncogene activation that increase the We used DNA fiber analysis to further evaluate RF dynamics in number of active RF (30, 31) might also diminish RPA availabil- our panel of cell lines upon exposure to low concentrations of HU, ity. No correlation was observed between SPR status and rate of a ribonucleotide reductase inhibitor that induces RF stalling proliferation, percentage of cells in S, or S-phase duration (Table without directly generating DNA lesions (34, 35). Consistent – – 1; Supplementary Fig. S2A–S2C), indicating that SPR cells do not with our observation that SPR HGSOC cell lines are sensitive þ – exhibit significantly higher rates of replication versus SPR coun- to HU (Fig. 2D), we observed that all three SPR HGSOC lines terparts in the absence of replication stress. We next evaluated the exhibit reduced replication progression rate in response to this þ þ number of active origins during replication stress in SPR versus drug compared with SPR counterparts (Fig. 5A). This further – SPR HGSOC lines by quantifying the number of sites of DNA supports the notion that the observed impairment in RF progres- – synthesis per nucleus, i.e., EdU foci representing replication sion upon genotoxin exposure in SPR cell lines is unlikely to origins, using fluorescence microscopy. This analysis revealed result solely from defective SPR. þ that in SPR cell lines, the number of EdU foci at 1 hour post- As mentioned previously, the HR factors BRCA1/2 and Rad51 UV is comparable with that in the absence of UV treatment (Fig. ensure protection of nascent DNA at stalled RF from Mre11- – 3D). However, SPR cell lines exhibit a significant increase in EdU dependent degradation (hereafter "RF protection"; refs. 12, 36), focus formation post-UV, suggesting dysregulated origin firing which in turn promotes overall RF progression upon exposure to during replication stress. To further validate these results, we replication-blocking genotoxins (37). Moreover, RF protection adapted a method reported elsewhere (32) to estimate the num- significantly mitigates the cytotoxicity of chemotherapeutics, ber of active RF in our cell lines upon replication stress. From the including CDDP (12). Given the strong association between SPR experiment immediately above, we evaluated total EdU signal efficiency and chemosensitivity, we explored possible correlations intensity after UV relative to mock treatment and found that it was between SPR status and RF protection in HGSOC cell lines. þ – similar in SPR versus SPR cells (Fig. 3E, left). Because the Quantification of RF protection was performed by successive – þ progression of individual RF is reduced in SPR versus SPR cell incubations with IdU and CldU, followed by prolonged incuba- – lines post-UV (Fig. 3C), the above data suggest that SPR cells tion with a concentration of HU that completely inhibits RF present elevated sites of EdU incorporation. To quantify this, progression. Under these conditions, nascent DNA degradation – nuclear EdU signal intensity was divided by the CldU/IdU ratios upon RF stalling leads to CldU/IdU ratios <1. Interestingly, SPR þ presented in Fig. 3C. For SPR cell lines, this value was below one, cell lines all manifest strong RF protection defects in response to consistent with robust inhibition of origin firing after UV. In HU, which were rescued by concomitant treatment with the – contrast, SPR cell lines displayed significantly higher ratios, Mre11 inhibitor mirin (Fig. 5B). Importantly, only one of the þ indicating increased origin activation relative to SPR counter- interrogated cell lines was found to have undetectable BRCA1 parts (Fig. 3E, right plot). expression, which probably explains its RF protection defect, We next evaluated the impact of aberrant origin activity on the whereas all six lines had normal BRCA2 and Rad51 levels – phenotypes of CDDP-sensitive (SPR ) HGSOC cell lines. CDK (Supplementary Fig. S3). Moreover, DNA sequencing did not activity permits activation of origins during S (33), and inhibition reveal mutations in BRCA1/2 in these cell lines (Supplementary of these kinases by roscovitine diminishes origin firing and RPA Table S1); therefore, the molecular basis of RF protection – recruitment to chromatin upon replication stress (16). Strikingly, defects in SPR cell lines OV1946 and TOV2835EP currently we found that roscovitine treatment significantly rescued SPR remains unknown. Nevertheless, our data indicate that RF defects (Fig. 3F), consistent with a key role of aberrant origin protection and SPR defects may often coexist in chemosensitive activity in abrogating SPR in ovarian cancer cell lines. We note that ovarian cancer cell lines, suggesting possible mechanistic over- phosphorylation of Chk1 (S345), RPA2 (S33), and gH2AX are all lap between these phenomena. þ – comparable in SPR and SPR cell lines following CDDP treatment (Fig. 3G). Thus, although the precise mechanism underlying RPA availability modulates chemosensitivity and nascent DNA – abnormal activation of replication origins in SPR cell lines stability in ovarian cancer cell lines remains unknown, it is unlikely to be attributable to defective Because nascent DNA degradation is expected to generate ATR/Chk1 activation upon replication stress. ssDNA and consequent RPA recruitment to stalled RF, we rea- Exhaustion of RPA due to elevated origin activity can eventually soned that RF protection defects might contribute to SPR inhibi- engender widespread induction of DSB at stalled RF (16). Using a tion by diminishing RPA availability. Contrary to this notion,

5566 Cancer Res; 78(19) October 1, 2018 Cancer Research

RPA Availability and Chemoresistance

P < 0.02 UV ACB IdU CldU SPR+ SPR- + -

No UV UV + SPR SPR 15 15 min 90 min

P < 0.001 DAPI 10 SPR- SPR+ 4

OV866(2) OV1369(R2) TOV2223G TOV112D OV1946 TOV21G TOV2835EP TOV3041G RPA1 2 BrdUrd RPA2 1

(arbitrary unit) 5 RPA3 0.5

RPA2 GAPDH

Nuclear RPA2 in BrdUrd Nuclear RPA2 0.25 0 0.125 UV: -+-+-+-+-+-+ 0.0625 Normalized CldU/IdU ratios

OV1946 OV866(2) OV1946 TOV2223G TOV3041G OV866(2) OV1369(R2) TOV2835EP TOV3041G TOV2223G TOV2835EPOV1369(R2) D No UV UV E n.s. P < 0.002 P < 0.0025 SPR+ SPR- SPR+ SPR- 600 n.s. + - No UV UV 100 1.50 SPR SPR SPR+ SPR- 75 1.25

400 DAPI

50 1.00

EdU 200 25 0.75 Number of EdU foci (% relative to mock) 0 0.50 (normalized for nuclear size) EdU incorporated after UV

0 EdU ratio/fork progression

OV1946 OV1946

OV866(2) OV866(2)

OV1946 OV1946

TOV2223G TOV3041G TOV2223G TOV3041G

OV866(2) OV866(2)

TOV2835EP TOV2835EP

OV1369(R2) OV1369(R2)

TOV2223G TOV3041G TOV2223G TOV3041G

TOV2835EP TOV2835EP

OV1369(R2) OV1369(R2)

P < 0.01 F 100 P < 0.025 G G1 Mock CDDP P < 0.01 S SPR- SPR+ SPR- SPR+ 80 G2

OV1946 TOV2835EP TOV3041G OV866(2) OV1369(R2) TOV2223G OV1946 TOV2835EP TOV3041G OV866(2) OV1369(R2) TOV2223G 40 p-Chk1(S345)

Chk1 20 p-RPA2(S33) 6-4PP remaining at 6 h (%) 6-4PP -n.s. γH2AX 0 Rosco: -2 h72 h - 2 h 72 h -2 h72 h H3 OV1946 TOV2835EP TOV3041G

Figure 3. SPR-defective HGSOC cell lines exhibit abnormal DNA replication dynamics. A, RPA2 recruitment to chromatin post-UV was measured by immunofluorescence microscopy. BrdUrd labeling distinguishes S-phase cells. Left, representative images for TOV2835EP. Right, fluorescence intensity of the nuclear RPA signal in HGSOC lines. Values for the mock treatment (black) are compared with values post-UV (blue). Red lines, median values. B, Immunoblot analysis of RPA subunit levels in SPR– and SPRþ ovarian cancer lines. C, Effect of UV (20 J/m2) on RF progression was measured using DNA fiber analysis. Red lines, median values. D, Effect of UV on the number of active RF following UV irradiation (see text for details). Left, fluorescence microscopy of EdU foci in TOV2835EP. Right, the number of EdU foci per nucleus in each line was quantified using custom software. E, Left, mean intensity of nuclear EdU signal post-UV, relative to mock, from D. Right, ratio of EdU incorporation post-UV (relative to mock-treated cells) divided by the normalized ratio of RF progression as measured in C to estimate the number of active origins post-UV. F, Effect of roscovitine on SPR following 2- or 72-hour treatment. P values, Student t test comparing mean between SPRþ and SPR– lines. G, Induction of phospho-Chk1 (S345), phospho-RPA2 (S33), and gH2AX 4 hours after CDDP treatment.

www.aacrjournals.org Cancer Res; 78(19) October 1, 2018 5567

Belanger et al.

Table 1. Rates of proliferation and S-phase duration of cell lines used in this control GFP-expressing clone. In agreement with our previous study finding using transient overexpression (18), stable RPA over- S-phase expression rescued SPR defects in the U2OS clone (Fig. 6A), Doubling duration confirming that limited availability of RPA is a major cause of SPR time (days) % cells in S (hours) fi OV866(2) þ 1.9 0.1 33 49.8 0.8 SPR de ciency in this system. Moreover, consistent with the fl OV1369(R2) þ 2.2 0.1 33 211 1 notion that RPA sequestration in uences nascent DNA stabil- TOV2223G þ 2.1 0.1 28 3 7.8 0.7 ity, we found (i) that SPR-defective U2OS cells present Mre11- TOV2835EP – 2.8 0.4 38 210.2 0.6 dependent RF protection defects upon HU treatment and (ii) OV1946 – 1.8 0.1 43 28.4 0.5 that both RF protection and progression defects in this cell line – TOV3041G 2.8 0.1 37 310.60.9 are rescued upon RPA overexpression (Fig. 6B). We emphasize U2OS RPA-GFP þ 1.2 0.1 50.0 0.7 9.3 0.2 U2OS GFP – 0.95 0.05 49.7 0.9 9.0 0.3 that the effects of RPA overexpression on NER and replication stress described immediately above are not likely caused by NOTE: Values are the mean SEM of three independent experiments. Quan- tification of S-phase duration is described in Supplementary Fig. S2A–S2C. differences in the rate of DNA synthesis, as no differences were noted in the rate of S-phase progression/duration, or the fraction of cells in S phase, between the RPA-overexpressing þ although siRNA-mediated knockdown of BRCA1 in 3 SPR cell and control U2OS clones (Table 1). lines caused strong Mre11-dependent RF protection defects (Fig. To further characterize the link between RPA availability and RF 5C), it did not compromise SPR (Fig. 5D). Importantly, exposure protection, we used siRNA to knockdown BRCA1 and RPA1, to VE-821, a potent ATR inhibitor, generated marked SPR defects either alone or in combination, in U2OS (Fig. 6C, left). BRCA1 þ in the 3 SPR cell lines as expected, ruling out the possibility that depletion, while not causing a significant increase in RPA recruit- these might somehow be immune to the impact of replication ment to chromatin following CDDP treatment as assessed by flow stress–induced RPA sequestration. Conversely, we found that cytometry (Fig. 6C, right), strongly exacerbated the existing RF mirin treatment, which resolves Mre11-dependent RF protection protection defect in U2OS (Fig. 6D). We note that the lack of effect defects (Fig. 5B), did not restore SPR in deficient cell lines (Fig. of BRCA1 depletion on chromatin-bound RPA is consistent with 5E). Overall, the above data indicate that Mre11-dependent results indicating that RF protection defects associated with degradation of nascent DNA at stalled RF does not engender SPR BRCA1 depletion do not compromise SPR in HGSOC cells (Fig. defects in HGSOC cell lines. 5C and D). Strikingly, siRNA-mediated knockdown of RPA1 We considered the alternative possibility that RPA exhaustion impaired RF protection to a similar extent as for BRCA1-depleted leading to compromised SPR might also contribute to RF pro- cells, which could be reversed by Mre11 inhibition (Fig. 6D). tection defects in HGSOC lines. For technical reasons, i.e., high Moreover, codepletion of RPA1 and BRCA1 caused an even cloning efficiency, we used U2OS osteosarcoma cells as model for stronger Mre11-dependent RF protection defect (Fig. 6D). These our initial experiments. We previously showed that U2OS exhibits data imply that reduced RPA availability (i) promotes Mre11- profound SPR defects that can be rescued by ectopic overexpres- dependent degradation of nascent DNA at stalled RF and (ii) can sion of RPA (18). We generated a U2OS clone stably expressing synergize with other genetic anomalies, e.g., BRCA1 deficiency, to the three RPA subunits in a stoichiometric manner, as well as a cause RF protection defects.

A 2 P < 0.03 B 10 40 P < 0.02 SPR- SPR+ 4 - + 1 SPR SPR 10 Mock CDDP H2AX

Mock - + - + g SPR SPR SPR SPR 0 30 10 3 2 10

1 20

10 H2AX/RPA1

g 2

OV1946 TOV3041G TOV2835EP OV1369(R2) TOV2223G OV866(2) OV1946 TOV3041G TOV2835EP OV1369(R2) TOV2223G OV866(2) H2AX CDDP g 0 10

2 10 10 DSB- % High Induction of DSB 1

1 10 (fold change mock/CDDP) 0 H2AX

g 0 0 10

CDDP + ATRi + CDDP 0 1 2 10 10 10 OV1946 RPA1 OV866(2) TOV3041G TOV2223G OV1946 TOV2835EPOV1369(R2) OV866(2) TOV3041G TOV2223G TOV2835EPOV1369(R2)

Figure 4. – SPR cell lines exhibit increased gH2AX and DNA DSB after CDDP treatment. A, RPA recruitment to chromatin and DSB formation after CDDP treatment. Left, representative bivariate plots of gH2AX and RPA for TOV3041G. ATR inhibition (VE-821) was used as positive control. Right, percentage of cells with high gH2AX and RPA chromatin levels for HGSOC lines. B, PFGE analysis of CDDP-induced DSB. Left, representative gel. The bands corresponding to DSB were quantiﬁed by densitometry. Right, fold induction of DSB signal after CDDP, relative to mock. Values are mean SEM of three independent experiments. P values, Student t test comparing mean between SPRþ and SPR– lines.

5568 Cancer Res; 78(19) October 1, 2018 Cancer Research

RPA Availability and Chemoresistance

Wash Wash Wash ABHU HU IdU CldU IdU CldU +/- Mirin

15 min 60 min 20 min 20 min 3 h

P < 0.001 P < 0.002 P < 0.001 SPR- SPR+ SPR- 4 4 SPR+ SPR- Mirin-treated 2 2 1 1 0.5 0.25 0.5

0.125 0.25 0.0625

Normalized CldU/IdU ratios 0.03125 Normalized CldU/IdU ratios 0.125

OV1946 OV1946 OV1946 OV866(2) OV866(2) TOV3041G TOV2223G TOV2223G TOV3041G TOV3041G TOV2835EPOV1369(R2) OV1369(R2) TOV2835EP TOV2835EP

C D 120 G1 S OV1369(R2)TOV2223G OV866(2) 100 siBRCA1: -+-+-+ G2 80 BRCA1

Tubulin 60 40 Wash Wash HU 20

IdU CldU +/- Mirin remaining at 6h (%) 6-4PP

20 min 20 min 3 h 0

siNT ATRi siNT ATRi siNT ATRi 4 P < 0.0001 P < 0.0001 P < 0.0001 siBRCA1 siBRCA1 siBRCA1 OV1369(R2) TOV2223G OV866(2) 2

1 120 E G1 100 S 0.5 G2 80 0.25 60 Normalized CldU/IdU ratios 0.125 Mirin: -+ - +- + 40

OV1369(R2) TOV2223G OV866(2) 20 6-4PP remaining at 6 h (%) 6-4PP 0 DMSO Mirin DMSO Mirin DMSO Mirin

OV1946 TOV2835EP TOV3041G

Figure 5. Nascent DNA at stalled RF is unstable in SPR-deficient HGSOC cell lines. A, Effect of HU on RF progression by DNA fiber analysis. B, RF protection upon HU treatment ( mirin) by DNA fiber analysis. C, Top plot, depletion of BRCA1 in SPRþ lines using siRNA versus NT controls. Bottom, RF protection mirin in BRCA1 knockdown SPRþ lines. D, 6-4PP removal after BRCA1 knockdown. Cells treated with ATR inhibitor were used as positive control for defective SPR. E, 6-4PP removal mirin. For bar graphs, values represent the mean SEM of three independent experiments. P values, Student t test (A, B, D,andE)or Mann–Whitney U test (C). For DNA fiber experiments, red lines are median values.

www.aacrjournals.org Cancer Res; 78(19) October 1, 2018 5569

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2018 American Association for Cancer Research. 5570 Downloaded from acrRs 81)Otbr1 2018 1, October 78(19) Res; Cancer DNA by measured was BRCA1 and/or RPA1 against siRNA Mann GPol scnrl.Lf,imnbo fRAsbnt.RcmiatRA n P2rti 2 laaesqec n aeahge W P3is RPA3 MW. higher a have and sequence cleavage P2A a RPA1. retain and/or RPA2 and RPA1 Recombinant 6-4PP. subunits. of removal RPA Right, of GFP. immunoblot to fused Left, control). as only (GFP P viaiiymdltsceoestvt n acn N stability. DNA nascent and chemosensitivity modulates availability RPA 6. Figure B lne tal. et elanger ﬁ – Whitney e assay. ber C EF RPA3-GFP A Tubulin RPA3 RPA2 RPA1 E, Tubulin siBRCA1: Clone: U RPA3 RPA1 RPA2 lnso SPR-de of Clones siRPA1: BRCA1 H, et( test RPA1 Fpoeto uigH treatment HU during protection RF YY1 H cancerres.aacrjournals.org GFP B GFP 34 and Published OnlineFirstAugust2,2018;DOI:10.1158/0008-5472.CAN-18-0618 -+-+ -+ RPA-GFP -RPA3 -RPA3-GFP D RPA2 RPA2-P2A RPA1 RPA1-P2A

Normalized CldU/IdU ratios -+ .FrDNA For ). RPA-GFP 61112 0.125 0.25 ﬁ 0.5 in V96wr eetdta vrxrs RPA-GFP. overexpress that selected were OV1946 cient 1 2 4 B, 0mn2 i 3h 20min 20 min Fpoeto n rgeso sasa nFg 5. Fig. in as assays progression and protection RF l Cl4 Cl3 IdU ﬁ GFP Wash siBRCA1: e xeiet,rdlnsaemda values. median are lines red experiments, ber siRPA1: P <0.0001

CldU Fold increase in chromatin RPA 6-4PP remaining at 6 h (%) 10 100 2 0 4 6 8 20 40 60 80 l l1Cl12 Cl11 Cl6 Wash 0 ﬂ RPA-GFP wcytometry. ow +/- Mirin 6-4PP remaining at 6 h (%) -+ -+-+ GFP P <0.0001 20 40 60 80 mirin. on September 23, 2021. © 2018American Association for Cancer Research. HU 0 -+ l Cl4 Cl3 I, GFP RPA GFP lngncsria olwn ipai treatment. cisplatin following survival Clonogenic P <0.0001 D, G G A, S Fpoeto nUO uigH ramn olwn iN gis BRCA1 against siRNA following treatment HU during U2OS in protection RF 2 1 l l1Cl12 Cl11 Cl6 ln fUO a eetdfrsal tihoercoeepeso fRPA-GFP of overexpression stoichiometric stable for selected was U2OS of clone A RPA-GFP I D C, B F, P idn ocrmtnatrcsltnteteti 2Sfollowing U2OS in treatment cisplatin after chromatin to binding RPA G G S

Relative survival (%) 2 1 -P eoa post-UV. removal 6-4PP 100

10 Normalized

1 Normalized CldU/IdU ratios 0.125 iN:N RA RPA1 BRCA1 NT siRNA: Cisplatin concentration( CldU/IdU ratios 0.25 0 0.125 Mirin: 0.5 0.25 5mn60min 15 min 0.5 G 1 2 IdU RPA Cl12 GFP Cl3 GFP Cl4 RPA Cl11 RPA Cl6 1 2 GFP Normalized P <0.0001 123 Wash

CldU/IdU ratios -+-+-+ 0.0625 * 0.125 GFP RPA 20 min 20 min 3 h 3 min 20 20 min 0.25 CldU IdU HU 0.5 Wash 1 2 4 P * aus Student values, G, l Cl4 Cl3 CldU GFP okporsinps-Vby post-UV progression Fork 15 min 90 min 90 15 min Wash IdU 20 min 20 min 3 h 3 min 20 20 min P <0.0001 IdU mmol/L) Normalized CldU/IdU ratios * 0.125 +/- Mirin Wash UV 0.25 l l1Cl12 Cl11 Cl6 4 0.5 HU RPA-GFP BRCA1 1 2 CldU RPA1 CldU -+ P <0.0001 * P <0.0001 t GFP Wash et( test P <0.001 acrResearch Cancer +/- Mirin GFP RPA A HU and F – I )or Published OnlineFirst August 2, 2018; DOI: 10.1158/0008-5472.CAN-18-0618

RPA Availability and Chemoresistance

– We sought to perform similar experiments in SPR HGSOC cell situation for RPA exhaustion in cells treated with ATR inhibitors lines, but for unknown reasons could only generate RPA-over- (16). Although RPA is required to promote recruitment/activity of expressing clones in OV1946. We selected three independent a multitude of DDR factors at stalled RF (14), we note that the OV1946 clones overexpressing recombinant RPA (Fig. 6E). precise mechanisms through which reduced RPA availability Although GFP-expressing controls exhibited profoundly reduced compromises RF stability remain unclear. Notably, recent prote- SPR, RPA overexpression completely restored NER proficiency omic surveys of RF-associated proteins in situations where RPA (Fig. 6F). In addition, RPA overexpression rescued RF protection pools are exhausted did not reveal striking changes in protein and progression defects in response to HU (Fig. 6G and H), as well recruitment to stalled RF (43). Alternatively, it is possible that as markedly increased CDDP resistance (Fig. 6I). These data unprotected ssDNA at stalled RF might be more susceptible to indicate that reduced RPA availability contributes to RF instability breakage arising from enzymatic activities present in HGSOC and chemosensitivity in HGSOC cells. cells. For example, recently published data demonstrate that ssDNA at stalled RF is frequently targeted by the APOBEC family of cytidine deaminases, eventually leading to DSB induction via Discussion base excision repair–mediated incisions (44). Elucidation of the Here, we sought to elucidate the role of DNA replication stress– complete spectrum of molecular mechanisms leading to DSB induced RPA exhaustion in promoting CDDP sensitivity using a induction at RF upon RPA exhaustion in HGSOC cells will require panel of patient-derived HGSOC cell lines. We initially turned our further investigation. þ – attention to NER, which, as discussed earlier, (i) is believed to We found that, relative to SPR counterparts, SPR HGSOC cell represent an important determinant of CDDP resistance in ovar- lines exhibit extensive Mre11-dependent degradation of nascent ian cancer and (ii) is sensitive to replication stress–induced DNA at HU-stalled RF, raising the possibility that regions of fluctuations in RPA availability during S phase. Specifically, we ssDNA thus generated might cause RPA sequestration and asso- exploited a cell-cycle–specific repair assay to demonstrate that a ciated phenotypes. However, we provide several lines of evidence majority of HGSOC cell lines exhibit strong defects in NER during arguing against this possibility: (i) Abrogation of BRCA1, which S phase relative to G0–G1 and G2–M. Moreover, remarkably, these causes strong RF protection defects, is not sufficient to trigger – SPR cell lines are all significantly (as much as 180-fold) more either RPA exhaustion or inhibition of SPR, and (ii) preventing þ sensitive to CDDP relative to SPR counterparts. Together with nascent DNA degradation at stalled RF via pharmacologic inhi- our previous observations in melanoma and model cancer cell bition of Mre11 did not rescue SPR. Although RF protection lines (19, 26), this suggests (i) that reduced SPR capacity repre- defects do not cause RPA exhaustion, the converse appears to be – sents a common feature of human cancers and (ii) that SPR true since we found that siRNA-mediated knockdown of RPA1 tumors in vivo, although challenging at present to identify as such, results in RF protection defects comparable with those observed in may respond much better to replication-blocking chemothera- BRCA1-depleted cells. These findings are generally in accord with peutic drugs including CDDP. recently published data indicating that disruption of RPA1 inter- Our data indicating that SPR defects in HGSOC lines can be action with ssDNA due to PTEN downregulation diminishes RF attributed to sequestration of RPA at aberrantly activated repli- progression rates upon HU exposure (45). The mechanisms re- cation origins are consistent with recently published results show- sponsible for the impact of RPA exhaustion on Mre11-dependent ing that defective ATR signaling causes RPA exhaustion at least in degradation of nascent DNA at RF are unknown. Phosphorylated part due to misregulated origin firing (16). Although our results RPA32 interacts with PALB2, which itself promotes BRCA2 do not support a causative role for ATR/Chk1 activation defects in recruitment (46, 47); however, as mentioned above, a recent compromising SPR in HGSOC cells, we cannot exclude that study revealed that RPA availability does not significantly alter downstream components of S-phase checkpoint-dependent inhi- protein recruitment at stalled RF (43). We note that histone bition of origin firing might be involved. In addition, activation of posttranslational modifications strongly influence nascent DNA various oncogenes, e.g., Ras, Myc, and cyclin E1, dysregulates stability (12). It is therefore possible that diminished RPA avail- replication origin initiation programs in human cells (31, 38), ability might influence chromatin structure at RF, in turn causing although their impact on RPA availability and SPR remains increased susceptibility to nuclease-mediated nascent DNA unknown. Collectively, our data suggest that yet-to-be-identified degradation. cancer-associated genetic alterations resulting in failure to prevent Importantly, we show that ectopic RPA overexpression protects – unscheduled origin activity might enhance chemosensitivity in the SPR HGSOC cell line OV1946 from CDDP-induced cell HGSOC cells by causing RPA exhaustion and SPR defects. We note death while rescuing SPR proficiency and RF stability. The notion that although RPA is well characterized for its essential roles in that RPA availability influences CDDP resistance is in agreement DNA replication and repair, this multifunctional complex has with published data showing that cell killing by this drug is also recently been implicated in transcription (39–41). Thus, it is potentiated by mutations that cripple RPA binding to DNA interesting to speculate that generation of ssDNA at sites of RNA (48). Moreover, our data indicate that RPA sequestration can polymerase stalling at CDDP-damaged sites in DNA might (i) cause RF protection defects, which are also expected to strongly sequester RPA and (ii) promote replication stress via collisions contribute to chemosensitivity (12). In this regard, it is interesting between the blocked transcriptional machineries and DNA poly- to note that high RPA expression, which based on our results merases (42). might potentially overcome RF instability upon replication stress, Defective SPR is not the sole consequence of RPA exhaustion, as has been reported to be predictive of adverse outcome in ovarian this complex is involved in a number of other important cellular cancer treatment (48, 49). This raises the interesting possibility processes. Indeed, we found that exposure to CDDP causes that elevated RPA availability might be selected for during the increased RPA recruitment to chromatin and DSB formation in course of chemotherapy treatment, leading to reversal of chemo- – þ SPR versus SPR HGSOC lines, which is reminiscent of the sensitivity. Overall, our results lead us to propose that modulation

www.aacrjournals.org Cancer Res; 78(19) October 1, 2018 5571

Belanger et al.

of RPA protein levels/dynamics during S is a critical determinant Administrative, technical, or material support (i.e., reporting or organizing of CDDP resistance in ovarian tumors via multiple mechanisms. data, constructing databases): M. Dube, E. Drobetsky Cellular pathways that influence RPA availability may therefore be Study supervision: H. Wurtele, E. Drobetsky expected to harbor useful biomarkers for predicting treatment Acknowledgments response in HGSOC. This study was funded by operating grant #364096 from the Canadian Institutes of Health Research (CIHR) awarded to E. Drobetsky, H. Wurtele, and Disclosure of Potential Conflicts of Interest A.-M. Mes-Masson. J.-Y. Masson is recipient of a CIHR Foundation Award. No potential conflicts of interest were disclosed. H. Wurtele, E. Fortier, and J.-F. Lemay are supported by salary awards from the Fonds de Recherche du Quebec-Sante (FRQS). The Centre de Recherche de Authors' Contributions l'Hopital^ Maisonneuve-Rosemont and Centre Hospitalier de l'Universitede Montreal (CRCHUM)/Institut du Cancer de Montreal receive institutional Conception and design:F.Belanger, E. Fortier, M. Dube, J.-Y. Masson, funding from the FRQS. For ovarian tumor banking, A.-M. Mes-Masson is A.-M. Mes-Masson, H. Wurtele, E. Drobetsky supported by the Banque de Tissus et de Donnees of the Reseau de Recherche sur Development of methodology: F. Belanger, E. Fortier, A. Elsherbiny, H. Wurtele, le Cancer and FRQS, affiliated with the Canadian Tumor Repository Network E. Drobetsky (CTRNet). Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.):F.Belanger, E. Fortier, M. Dube, J.-F. Lemay, R. Buisson, A. Elsherbiny, E. Drobetsky The costs of publication of this article were defrayed in part by the Analysis and interpretation of data (e.g., statistical analysis, biostatistics, payment of page charges. This article must therefore be hereby marked advertisement computational analysis):F.Belanger, E. Fortier, M. Dube, J.-F. Lemay, in accordance with 18 U.S.C. Section 1734 solely to indicate R. Buisson, J.-Y. Masson, A. Elsherbiny, S. Costantino, H. Wurtele, E. Drobetsky this fact. Writing, review, and/or revision of the manuscript:F.Belanger, E. Fortier, M. Dube, R. Buisson, S. Costantino, E. Carmona, A.-M. Mes-Masson, H. Wurtele, Received February 26, 2018; revised June 11, 2018; accepted July 27, 2018; E. Drobetsky published first August 2, 2018.

References 1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2017. CA Cancer J Clin 17. Tsaalbi-Shtylik A, Moser J, Mullenders LHF, Jansen JG, de Wind N. 2017;67:7–30. Persistently stalled replication forks inhibit nucleotide excision repair in 2. Davis A, Tinker AV, Friedlander M. "Platinum resistant" ovarian cancer: trans by sequestering Replication protein A. Nucleic Acids Res 2014;42: what is it, who to treat and how to measure benefit? Gynecol Oncol 4406–13. 2014;133:624–31. 18. Belanger F, Angers J-P, Fortier E, Hammond-Martel I, Costantino S, 3. Jung Y, Lippard SJ. Direct cellular responses to platinum-induced DNA Drobetsky E, et al. Mutations in replicative stress response pathways are damage. Chem Rev 2007;107:1387–407. associated with S phase-specific defects in nucleotide excision repair. J Biol 4. Murai J. Targeting DNA repair and replication stress in the treatment of Chem 2016;291:522–37. ovarian cancer. Int J Clin Oncol 2017;22:619–28. 19. Belanger F, Rajotte V, Drobetsky EA. A majority of human melanoma cell 5. Galluzzi L, Senovilla L, Vitale I, Michels J, Martins I, Kepp O, et al. lines exhibits an S phase-specific defect in excision of UV-induced DNA Molecular mechanisms of cisplatin resistance. Oncogene 2012;31: photoproducts. PloS One 2014;9:e85294. 1869–83. 20. Provencher DM, Lounis H, Champoux L, Tetrault M, Manderson EN, Wang 6. Wood RD, Araujo SJ, Ariza RR, Batty DP, Biggerstaff M, Evans E, et al. DNA JC, et al. Characterization of four novel epithelial ovarian cancer cell lines. damage recognition and nucleotide excision repair in mammalian cells. In Vitro Cell Dev Biol Anim 2000;36:357–61. Cold Spring Harb Symp Quant Biol 2000;65:173–82. 21. Ouellet V, Zietarska M, Portelance L, Lafontaine J, Madore J, Puiffe M-L, 7. Furuta T, Ueda T, Aune G, Sarasin A, Kraemer KH, Pommier Y. Transcrip- et al. Characterization of three new serous epithelial ovarian cancer cell tion-coupled nucleotide excision repair as a determinant of cisplatin lines. BMC Cancer 2008;8:152. sensitivity of human cells. Cancer Res 2002;62:4899–902. 22. Letourneau IJ, Quinn MCJ, Wang L-L, Portelance L, Caceres KY, Cyr L, et al. 8. Selvakumaran M, Pisarcik DA, Bao R, Yeung AT, Hamilton TC. Enhanced Derivation and characterization of matched cell lines from primary and cisplatin cytotoxicity by disturbing the nucleotide excision repair pathway recurrent serous ovarian cancer. BMC Cancer 2012;12:379. in ovarian cancer cell lines. Cancer Res 2003;63:1311–6. 23. Fleury H, Communal L, Carmona E, Portelance L, Arcand SL, Rahimi K, 9. Dobbelstein M, Sørensen CS. Exploiting replicative stress to treat cancer. et al. Novel high-grade serous epithelial ovarian cancer cell lines that reflect Nat Rev Drug Discov 2015;14:405–23. the molecular diversity of both the sporadic and hereditary disease. Genes 10. Venkitaraman AR. Cancer susceptibility and the functions of BRCA1 and Cancer 2015;6:378–98. BRCA2. Cell 2002;108:171–82. 24. Rouget R, Auclair Y, Loignon M, Affar EB, Drobetsky EA. A sensitive flow 11. Schlacher K, Christ N, Siaud N, Egashira A, Wu H, Jasin M. Double-strand cytometry-based nucleotide excision repair assay unexpectedly reveals that break repair-independent role for BRCA2 in blocking stalled replication mitogen-activated protein kinase signaling does not regulate the removal fork degradation by MRE11. Cell 2011;145:529–42. of UV-induced DNA damage in human cells. J Biol Chem 2008;283: 12. Ray Chaudhuri A, Callen E, Ding X, Gogola E, Duarte AA, Lee J-E, et al. 5533–41. Replication fork stability confers chemoresistance in BRCA-deficient cells. 25. Auclair Y, Rouget R, Belisle JM, Costantino S, Drobetsky EA. Requirement Nature 2016;535:382–7. for functional DNA polymerase eta in genome-wide repair of UV-induced 13. Byun TS, Pacek M, Yee M, Walter JC, Cimprich KA. Functional uncoupling DNA damage during S phase. DNA Repair 2010;9:754–64. of MCM helicase and DNA polymerase activities activates the ATR-depen- 26. Auclair Y, Rouget R, Affar EB, Drobetsky EA. ATR kinase is required dent checkpoint. Genes Dev 2005;19:1040–52. for global genomic nucleotide excision repair exclusively during 14. Marechal A, Zou L. RPA-coated single-stranded DNA as a platform for post- S phase in human cells. Proc Natl Acad Sci U S A 2008;105: translational modifications in the DNA damage response. Cell Res 17896–901. 2015;25:9–23. 27. Behrens BC, Hamilton TC, Masuda H, Grotzinger KR, Whang-Peng J, Louie 15. Cimprich KA, Cortez D. ATR: an essential regulator of genome integrity. KG, et al. Characterization of a cis-diamminedichloroplatinum(II)-resis- Nat Rev Mol Cell Biol 2008;9:616–27. tant human ovarian cancer cell line and its use in evaluation of platinum 16. Toledo LI, Altmeyer M, Rask M-B, Lukas C, Larsen DH, Povlsen LK, et al. analogues. Cancer Res 1987;47:414–8. ATR prohibits replication catastrophe by preventing global exhaustion of 28. Verweij J, Pinedo HM. Mitomycin C: mechanism of action, usefulness and RPA. Cell 2013;155:1088–103. limitations. Anticancer Drugs 1990;1:5–13.

5572 Cancer Res; 78(19) October 1, 2018 Cancer Research

RPA Availability and Chemoresistance

29. Ryan AJ, Squires S, Strutt HL, Johnson RT. Camptothecin cytotoxicity 40. Fujimoto M, Takaki E, Takii R, Tan K, Prakasam R, Hayashida N, et al. RPA in mammalian cells is associated with the induction of persistent assists HSF1 access to nucleosomal DNA by recruiting histone chaperone double strand breaks in replicating DNA. Nucleic Acids Res 1991;19: FACT. Mol Cell 2012;48:182–94. 3295–300. 41. Zhang H, Gan H, Wang Z, Lee J-H, Zhou H, Ordog T, et al. RPA interacts 30. Hills SA, Diffley JFX. DNA replication and oncogene-induced replicative with HIRA and regulates H3.3 deposition at gene regulatory elements in stress. Curr Biol CB 2014;24:R435–444. mammalian cells. Mol Cell 2017;65:272–84. 31. Srinivasan SV, Dominguez-Sola D, Wang LC, Hyrien O, Gautier J. Cdc45 is 42. Brambati A, Colosio A, Zardoni L, Galanti L, Liberi G. Replication and a critical effector of myc-dependent DNA replication stress. Cell Rep transcription on a collision course: eukaryotic regulation mechanisms and 2013;3:1629–39. implications for DNA stability. Front Genet 2015;6:166. 32. Ge XQ, Blow JJ. Chk1 inhibits replication factory activation but allows 43. Dungrawala H, Rose KL, Bhat KP, Mohni KN, Glick GG, Couch FB, et al. The dormant origin firing in existing factories. J Cell Biol 2010;191:1285–97. replication checkpoint prevents two types of fork collapse without regu- 33. Bousset K, Diffley JF. The Cdc7 protein kinase is required for origin firing lating replisome stability. Mol Cell 2015;59:998–1010. during S phase. Genes Dev 1998;12:480–90. 44. Buisson R, Lawrence MS, Benes CH, Zou L. APOBEC3A and APOBEC3B 34. Desany BA, Alcasabas AA, Bachant JB, Elledge SJ. Recovery from DNA activities render cancer cells susceptible to ATR inhibition. Cancer Res replicational stress is the essential function of the S-phase checkpoint 2017;77:4567–78. pathway. Genes Dev 1998;12:2956–70. 45. Wang G, Li Y, Wang P, Liang H, Cui M, Zhu M, et al. PTEN regulates RPA1 35. Zellweger R, Dalcher D, Mutreja K, Berti M, Schmid JA, Herrador R, et al. and protects DNA replication forks. Cell Res 2015;25:1189–204. Rad51-mediated replication fork reversal is a global response to genotoxic 46. Murphy AK, Fitzgerald M, Ro T, Kim JH, Rabinowitsch AI, Chowdhury D, treatments in human cells. J Cell Biol 2015;208:563–79. et al. Phosphorylated RPA recruits PALB2 to stalled DNA replication forks 36. Mijic S, Zellweger R, Chappidi N, Berti M, Jacobs K, Mutreja K, et al. to facilitate fork recovery. J Cell Biol 2014;206:493–507. Replication fork reversal triggers fork degradation in BRCA2-defective cells. 47. Yazinski SA, Comaills V, Buisson R, Genois M-M, Nguyen HD, Ho CK, et al. Nat Commun 2017;8:859. ATR inhibition disrupts rewired homologous recombination and fork 37. Costanzo V.Brca2, Rad51 and Mre11: performing balancing acts on rep- protection pathways in PARP inhibitor-resistant BRCA-deficient cancer lication forks. DNA Repair 2011;10:1060–5. cells. Genes Dev 2017;31:318–32. 38. Maya-Mendoza A, Ostrakova J, Kosar M, Hall A, Duskova P, Mistrik M, et al. 48. Neher TM, Bodenmiller D, Fitch RW, Jalal SI, Turchi JJ. Novel irreversible Myc and Ras oncogenes engage different energy metabolism programs and small molecule inhibitors of replication protein A display single-agent evoke distinct patterns of oxidative and DNA replication stress. Mol Oncol activity and synergize with cisplatin. Mol Cancer Ther 2011;10:1796–806. 2015;9:601–16. 49. Levidou G, Ventouri K, Nonni A, Gakiopoulou H, Bamias A, Sotiropoulou 39. Sikorski TW, Ficarro SB, Holik J, Kim T, Rando OJ, Marto JA, et al. Sub1 and M, et al. Replication protein A in nonearly ovarian adenocarcinomas: RPA associate with RNA polymerase II at different stages of transcription. correlation with MCM-2, MCM-5, Ki-67 index and prognostic significance. Mol Cell 2011;44:397–409. Int J Gynecol Pathol 2012;31:319–27.

www.aacrjournals.org Cancer Res; 78(19) October 1, 2018 5573

Replication Protein A Availability during DNA Replication Stress Is a Major Determinant of Cisplatin Resistance in Ovarian Cancer Cells

François Bélanger, Emile Fortier, Maxime Dubé, et al.

Cancer Res 2018;78:5561-5573. Published OnlineFirst August 2, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-18-0618

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2018/08/02/0008-5472.CAN-18-0618.DC1

Cited articles This article cites 49 articles, 16 of which you can access for free at: http://cancerres.aacrjournals.org/content/78/19/5561.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/78/19/5561.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/78/19/5561. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.