Exposure to the Fungicide Captan Induces DNA Base Alterations and Replicative Stress in Mammalian Cells

The classification of the fungicide captan (CAS Number: 133–06‐2) as a carcinogen agent is presently under discussion. Despite the mutagenic effect detected by the Ames test and carcinogenic properties observed in mice, the genotoxicity of this pesticide in humans is still unclear. New information is needed about its mechanism of action in mammalian cells. Here, we show that Chinese Hamster Ovary (CHO) cells exposed to captan accumulate Fpg‐sensitive DNA base alterations. In CHO and HeLa cells, such DNA lesions require the XRCC1‐dependent pathway to be repaired. Captan also induces a replicative stress that activated the ATR signaling response and resulted in double‐strand breaks and micronuclei. The replicative stress is characterized by a dramatic decrease in DNA synthesis due to a reduced replication fork progression. However, impairment of the XRCC1‐related repair process did not amplify the replicative stress, suggesting that the fork progression defect is independent from the presence of base modifications. These results support the involvement of at least two independent pathways in the genotoxic effect of captan that might play a key role in carcinogenesis. Environ. Mol. Mutagen. 60:286–297, 2019. © 2018 Wiley Periodicals, Inc.


INTRODUCTION
Captan is an agricultural fungicide that belongs to the phthalimide family. It was first registered in 1951 for tree fruit use and has been used commercially for over 60 years. Despite extensive studies its toxicity is still not consensual and it has been classified in several ways. It was classified in 1982 by the International Agency for Research on Cancer (IARC) as group 3 "Not classifiable as to its carcinogenicity to human" (IARC (International Agency for Research on Cancer) 1982). In the United States, the Environmental Protection Agency (EPA) reclassified it in 2004 from "probable human carcinogen" to "not likely carcinogenic to human" (Gordon 2007). In contrast, it is classified in category 2 H351 ("suspected of causing cancer") according to the European regulations (European Commission, EC Reg N 1272. Captan is authorized in many countries. It is used mainly on fruits such as apples and is detected as pesticide residues in food in European countries (Crepet et al. 2013;Lozowicka 2015).
Concerning occupational exposure, a recent analysis from the North American Pooled Project showed that captan is associated with an increased risk of multiple myeloma in farmers (odds ratio 3.14 with a confidence interval 95% of 1.30-7.58) (Presutti et al. 2016). However, this risk is reduced for duration of exposure greater than 7 years (Presutti et al. 2016). Moreover, this non-monotonic exposure response pattern and the previous absence of association between the highest level of captan exposure and risk of cancer revealed by the Agricultural Heath Study (AHS) (Greenburg et al. 2008) contributed to the US EPA revision in 2004. A risk assessment study was conducted by the European Food Safety Authority (EFSA) in 2009 which concluded in its non-genotoxic potential (European Food Safety Authority 2009). However, the mutagenic potency of captan has long been reported in Ames strains TA1535 and TA104 and also in the Escherichia coli WP2uvrA strain (Bridges 1975; Barrueco and de la Pena 1988;Ohta et al. 2002), revealing base-change substitution on AT or GC base pairs, as well as in mammalian cellular mutation assays (O'Neill et al. 1981;Edgar et al. 1985). In addition to its mutagenic activity, the genotoxic potential of captan is supported by the induction of DNA breaks leading to chromosomal aberrations in mammalian cells (Swenberg et al. 1976;Tezuka et al. 1980;Snyder 1992). Finally, a direct interaction of captan with DNA and/or proteins linked to chromatin such as histones, polymerases or topoisomerase II, has been suggested (Anderson and Rosenkranz 1974;Selsky and Matheson 1981;Dillwith and Lewis 1982;Pritchard and Lappin 1991;Snyder 1992;Grilli et al. 1995;Provan et al. 1995;Provan and Eyton-Jones 1996;Rahden-Staron 2002).
In contrast to cellular analyses, carcinogenesis assays were negative in rats and slightly positive in mice, which developed duodenal glandular tumors and squamous cell lesions of the forestomach [see for review ]. The difference observed in genotoxicity and mutagenesis between in vitro and in vivo situations might depend on the metabolism of captan and sequestration of reactive derivatives [for review see ]. Under in vivo conditions, the presence of thiol-rich environment such as cysteine or glutathione found in blood or liver, could explain the absence of genotoxicity by the rapid elimination of the reactive and volatile breakdown product of captan, the thiophosgene (Swenberg et al. 1976;Moriya et al. 1978;Gordon 2010). Thus, the increased tumor risk observed in mice might be due to chronic mucosal wounding and regenerative hyperplasia Thompson et al. 2017), leading to non-genotoxic effects inducing duodenal carcinogenesis. Nevertheless, Mac Cauley and colleagues have described a significant positive association between DNA damage in oral human leukocytes using alkaline comet assay and urinary concentration of the captan metabolite tetrahydrophthalimide (McCauley et al. 2008).
In this context, it is of great interest to throw new light on captan mechanisms of toxicity. To this end, we investigated captan potential genotoxicity in mammalian cell lines. Here, we show that captan induces DNA damage that requires a XRCC1-dependent repair process and an independent replicative stress characterized by a slowing down of fork velocity, resulting in the formation of double-strand breaks (DSBs) and genomic instability.

Cell Lines and Treatments
HeLa human cells were maintained in Dulbecco's modified Eagle's medium (DMEM, Gibco, Life technologies). AA8 (XRCC1 +/+ ) and EM9 (XRCC1 −/− ) rodent cell lines were maintained in a mixture of 50% DMEM media and 50% F-12 media. All media were supplemented with 10% heat-inactivated calf serum and 0.5 mg/mL penicillin/streptomycin (P/S). Cells lines were grown in a humidified incubator at 37 C with 5% CO 2 . All cell lines were checked and were mycoplasma-free. Peripheral Blood Mononuclear cells (PBMC) collected from one volunteer [Établissement Français du Sang, (EFS), written consent for the use of blood samples for the research protocol obtained according to the regulation for blood transfusion of the French blood organization EFS, Toulouse (France)] were Ficoll-separated from whole blood and aliquoted (1*10 6 cells/aliquot) prior to cryopreservation in liquid nitrogen as described previously (Lebailly et al. 2015). For each experiment, aliquots of PBMC were defrosted, washed with phosphate-buffered saline (PBS) with 10% serum, centrifuged (200 g, 10 min) and resuspended in PBS. When needed, cells were treated with captan (Dr. Ehrenstorfer GmbH, Germany/ CIL-Cluzeau; purity of 98.5%, with impurities as inorganic salts, besides 0.02% ethanol and 0.01% dichloromethane). Captan was added to culture medium, supplemented with serum for all experiments, except for comet assays experiments, which were carried out without serum.
To induce DNA damage as products of purine oxidation detectable by formamidopyrimidine-DNA glycosylase (Fpg), cells were placed on ice and treated with the compound Ro 19-8022 (gift from Hoffman Laroche) (at 1 μM in PBS) for 2 min 30 s under visible light (1000 W-halogen). Ro 19-8022 plus light is an appropriate positive control for the Fpg-modified comet assay (Azqueta et al. 2013).

RNA Interference
Gene silencing was performed by transfection of siRNA (Sigma-Aldrich) in HeLa cells with INTERFERin ® according to the manufacturer's instructions (Polyplus). Briefly, siRNA were mixed with INTERFERin ® in Opti-MEM and incubated for 10 min at room temperature. Then, HeLa cells were transfected with the negative control siRNA (CAUGUCAUGU GUCACAUCU-dTdT), or with siRNA directed against XRCC1 (GGAAG AUAUAGACAUUGAG-dTdT), or ATR (CCUCCGUGAUGUUGCUU GA-dTdT). After 24 h of incubation, cells were counted and plated for further analysis.

Clonogenic Assay
Cells were plated in triplicate at a density of 200 to 500 cells per well in 6-well plates. One day after seeding, cells were treated with captan at the indicated concentrations in complete medium and grown for 6-12 days. AA8 and EM9 cells were exposed to captan until fixed. For HeLa cells treated with siRNA, cells were exposed to captan 66 h to 90 h post-transfection during the most efficient time of siRNA-mediated protein depletion. Formed colonies were fixed and stained with a 0.25% methylene blue (Sigma) and 50% methanol solution. Colonies containing more than 50 cells were counted and the survival rate calculated. Survival rates are normalized to 100% for undamaged cells for each cell population.

Proliferation Assay
Cells were plated in duplicate in 24-well plates at a density of 40,000 and 50,000 cells per well for AA8 and EM9 cells, respectively. One day after seeding, cells were treated in duplicate with captan at the indicated concentrations in complete medium and grown for 48 h. Final cell number in each well was estimated by Malassez cell counting.

Micronucleus Assay
HeLa cells were grown in 12-well plates (2 × 10 4 cells/well) for 24 h before to be treated or not with captan for 24 h. AA8 and EM9 cells were grown in 24-well plates (2 × 10 4 cells/well) for 24 h before to be treated with captan for 24 h. Cells were fixed with 4% paraformaldehyde for 20 min after a PBS wash. They were then permeabilized with 0.5% Triton X-100 in PBS for 15 min, and nuclei were stained with 4.6-diamino-2-phenyl indole (DAPI) (100 nM).
Environmental and Molecular Mutagenesis. DOI 10.1002/em

Comet Assay
The alkaline comet assay was used to detect strand breaks and alkalilabile sites as previously described with minor modifications (Lebailly et al. 2015). Cells were embedded in 0.7% low melting point agarose and a parallel digestion with Fpg enzyme (gift from Serge Boiteux, CNRS, France) allowing the detection of Fpg-sensitive sites was performed, as already described (Azqueta et al. 2013). Fifty cells per slide and two slides per sample were analyzed. The extent of DNA damage was evaluated for each cell by measuring the intensity of all tail pixels divided by the total intensity of all pixels in the head and tail of the comet. The median of these 100 values was calculated and named % tail DNA. For comet cellular repair assay, the Net level of Fpg-sensitive sites (% Tail DNA) was obtained for each condition by subtracting the damage (% Tail DNA) obtained in the absence of Fpg from Fpg-exposed comets ( " + Fpg " − " − Fpg " = Net Fpg). To specifically calculate the effect of captan (% Tail DNA), Net Fpg-sensitive sites of non-treated (control) cells was subtracted from Net Fpg-sensitive sites of captan treated cells, corresponding to Net Fpg-sensitive sites above the control cells (Net Fpg of captan − Net Fpg of control cells = Net Fpg of captan above control).

Flow Cytometry Analyses γH2AX Staining
Two hours after treatment with 10 μM captan, AA8 and EM9 cell lines were fixed in PBS containing 4% paraformaldehyde for 15 min at room temperature, washed, then permeabilized with PBS containing 0.2% triton X-100 and 1% BSA for 20 min at room temperature. Antibody against γH2AX (1:300, clone JBW301, Millipore) was incubated for 3 h in PBS containing 1% BSA at room temperature. After three washes, cells were incubated with the secondary antibody anti-mouse Alexa Fluor 488 (1:1,000, A11017, Thermofisher) for 30 min at room temperature. Then, cells were washed three times and incubated in PBS containing DAPI (1 μg/mL) for 15 min before samples were processed using flow cytometry (MACSQuant, Miltenyi Biotec). At least 10,000 events were analyzed per sample using FlowJo software.

Cell Cycle Analysis
Two hours after captan treatment (10 μM), CHO cell lines were incubated with EdU (5 μM) for 15 min. Cells were collected by trypsinization and incorporated EdU was detected using the baseclick EdU flow cytometry kit (Sigma, BCK-FC488) according to the manufacturer's instructions. Cells were incubated in PBS containing DAPI (1 μg/mL) for 15 min before samples were processed using flow cytometry (MACSQuant, Miltenyi Biotec). At least 10,000 events were analyzed per sample using FlowJo software.

DNA Fiber Assay
CHO cell lines were plated for 24 h before sequential pulse labelling with 50 μM CldU (5-chlorodeoxyuridine, Sigma-Aldrich), then 100 μM IdU (5iododeoxyuridine, Sigma-Aldrich) for 20 min each, followed by a chase with 200 μM thymidine (Sigma-Aldrich) for 1 h. Then, cells were collected for DNA fiber assays and re-suspended in cold PBS. Two μl of cells resuspended in PBS (5 × 10 5 cells/ml) were spotted onto clean glass slides, mixed with 7.5 μL of lysis buffer (200 mM Tris-HCl pH 7.4, 50 mM EDTA, 0.5% SDS) and incubated horizontally for 5 min at room temperature. DNA was spread by tilting the slide manually at a 30-45 angle. The air-dried DNA was fixed in cold methanol/acetic acid (3:1) for 20 min. Slides were rehydrated in PBS twice and the DNA was denatured in 2.5 M HCl for 45 min at room temperature. The slides were washed twice in PBS, followed by incubation in blocking solution (1% BSA, 0.1% Tween 20, PBS) for 30 min at room temperature followed by primary antibodies (rat anti-CldU (1:75, OBT0030G, Biorad) and mouse anti-IdU (1:50, BD347583, Becton Dickinson)) for 1.5 h at room temperature. After three washes in PBS, the slides were fixed in 4% paraformaldehyde for 10 min, washed and then incubated in blocking solution for 10 min. Next, the slides were incubated with the secondary antibodies (anti-rat Alexa Fluor 546 (A11081) and anti-mouse Alexa Fluor 488 (A11017), Thermofisher, 1:200) for 1 h at room temperature. Subsequently, they were washed as before, incubated in blocking solution for 10 min, then with the single-strand DNA antibody overnight at 4 C (mouseanti ssDNA, 1:50, MAB3034, Millipore). After three washes in PBS, the slides were incubated in blocking solution for 10 min and finally with the secondary antibody (anti-mouse Alexa Fluor 350, 1:200, A21130, Thermofisher) for 2 h. Images were analyzed using NIS Elements-AR Nikon software. The specific DNA staining allowed the exclusion of any signal due to broken or overlapping DNA fibers. IdU track length was determined if flanked by a CldU track. At least 400 fibers per condition were measured.

Statistical Analysis
The results are expressed as the mean AE SEM. Statistical analysis was assessed using Prism 6 software (GraphPad). Student's t-test, Mann-Whitney or two-ways ANOVA tests, followed by post hoc tests were used when appropriate. A P value <0.05 was considered significant. For clonogenic survival assay, micronucleus assay, γH2AX and 53BP1 immunofluorescence analyses on siRNA-transfected HeLa and CHO (AA8 and EM9) cells, two-way ANOVA followed by Sidak's multiple comparison test were used in order to study treatment effects and cell line effect. This statistical approach was also used for γH2AX staining analyzed by flow cytometry and cell cycle analysis. For the analysis of "Fpg-sensitive sites" by Environmental and Molecular Mutagenesis. DOI 10.1002/em comet assay, two-tailed Student's t-test was used. For the other comet assays, two-way ANOVA followed by Sidak's multiple comparison tests were performed. For DNA fiber assay, statistical analysis was performed using two-tailed Mann-Whitney test.

XRCC1Deficiency Induces Hypersensitivity to Captan
Since captan was reported to induce base substitution reversion in Ames test strains (TA102 and TA104) and WP2uvrA6 Escherichia coli strains used to detect base substitution mutagenesis, we first tested the effect of XRCC1 defect, a scaffold protein involved in base excision repair (BER) (Kubota et al. 1996). HeLa cells were depleted for XRCC1 protein by transfection with siRNA (Fig. 1A, left panel). Compared to control cells transfected with a scramble siRNA, XRCC1-deficient cells showed a significant higher sensitivity to 1 μM of captan (Fig. 1A,  right panel). Similarly, EM9 CHO cells mutated in XRCC1 were more sensitive than the wild type AA8 cells to captan treatment (Fig. 1B). These results suggest that captan sensitivity is associated with the induction of DNA damage that is repaired by XRCC1. HeLa cells exposed to 10 μM of captan for 24 h exhibited a higher percentage of micronucleus-containing cells compared to untreated cells (Supporting Information Fig. S1), suggesting that captan induces DNA damage that can generate genetic instability. Moreover, AA8 and EM9 CHO cells exposed to captan for 24 h exhibit a dose-dependent increase of micronucleus formation, with a significant greater impact on EM9 cells (Fig. 1C). This data demonstrates that XRCC1 is involved in the prevention of genetic instability induced by captan.
In order to confirm that captan induces DNA damage in mammalian cells, an alkaline comet assay was performed on human peripheral lymphocytes. When exposed to 1 μM of captan for 2 h, these lymphocytes present more DNA damage than the non-treated cells (Supporting Information Fig. S2), confirming the genotoxic activity of the fungicide. Then, the involvement of XRCC1 in the repair of DNA damage induced by captan was analyzed through alkaline and Fpgmodified comet assays. In CHO cells exposed to 1 μM of captan for 2 h, no damage was observed through alkaline comet assay (Fig. 1D), probably due to the high basal DNA damage already detected in this cellular background. In the same way, no DNA damage through alkaline comet assay can be observed in these cells treated with Ro 19-8022, a chemical agent known to generate 8-oxoguanine (Supporting Information Fig. S3A). Next, the Fpg-modified comet assay was performed to reveal Fpg-sensitive sites which mainly represent oxidative base modifications [Reviewed in (Collins 2014)], usually repaired by the BER pathway. Similar to the control treatment with Ro 19-8022 (Supporting Information Fig. S3A), captan exposure for 2 h induced Fpg-sensitive DNA lesions, as shown in AA8 and EM9 CHO cells (Fig. 1D). To determine whether XRCC1 is involved in the repair of such lesions, Fpg-modified comet assay was also performed on AA8 and EM9 CHO cells released for 1 h in fresh media after captan or Ro 19-8022 exposure. In control cells, repair of DNA lesions induced by captan or Ro 19-8022 was almost complete within 60 min (Fig. 1E for captan and Supporting Information Fig. S3B for Ro 19-8022). In contrast, repair of Fpg-sensitive lesions was severely impaired in XRCC1-deficient cells after captan or Ro 19-8022 exposure (Fig. 1E for captan and Supporting Information Fig. S3B for Ro 19-8022). These results indicate the involvement of XRCC1 in the repair of DNA damage induced by captan.

The Replicative Stress Response Is Crucial for Survival from Captan-Related Genotoxicity
To further characterize the genotoxic stress induced by captan, we performed immunofluorescence analyses in HeLa cells to investigate histone H2AX phosphorylation (γH2AX), which is a general DNA damage marker mainly associated with DSBs and replication stress (RS), and 53BP1 recruitment, which is more specific to DSBs (de Feraudy et al. 2010; Vignard et al. 2013) (Fig. 2A). After 2 h of captan exposure, a dose-dependent increase in both γH2AXand 53BP1-positive cells was observed (Fig. 2B). Except for the non-treated condition where the γH2AX signaling may be too low in some cells to be properly detected, all the 53BP1 positive cells were also γH2AX positive after captan treatment. However, a subclass of captan-treated cells accumulating a γH2AX signal was devoid of 53BP1 foci ( Fig. 2A,B). This suggests that captan-related DNA damage generates a RS that subsequently results in DSB formation. To test this hypothesis, HeLa cells exposed to captan for 2 h were colabeled with γH2AX and FANCD2, a protein that spontaneously form foci in S-phase cells (Liu et al. 2000;Taniguchi et al. 2002). As in non-treated HeLa cells where a weak γH2AX staining related to a low RS signaling is only visible in FANCD2 positive cells, the increased subclass of cells accumulating γH2AX after captan exposure is limited to the FANCD2 positive population (Supporting Information Fig. S4). Taken together, these data suggest that captanmediated DNA damage induces a RS and subsequently DSB signaling specifically during S-phase.
To investigate whether captan exposure induces RS, we assessed the induction of CHK1 phosphorylation at S345 (pCHK1) and of FANCD2 monoubiquitylation, two wellestablished markers of RS (Liu et al. 2000;Taniguchi et al. 2002). HeLa cells exposed to 20 μM of captan for 2 h exhibited higher pCHK1 and FANCD2 monoubiquitylation compared to non-treated cells, demonstrating that captan-exposed cells undergo RS (Fig. 2C). We next confirmed that the RS response is crucial after captan treatment by addressing the sensitivity of cells deficient in ATR, an essential kinase for surviving RS (Saldivar et al. 2017). As expected, siRNA-mediated ATR downregulation resulted in loss of viability in HeLa cells exposed to captan (Fig. 2D). Moreover, ATR-deficient cells treated with 10 μM of captan for 24 h presented a higher percentage of micronuclei compared to control cells (Fig. 2E), indicating that the RS response is important to prevent genetic instability after captan. In addition, the proportion of 53BP1-positive cells after captan exposure was higher after ATR depletion (Fig. 2F), providing further evidence that these DSBs derive from captan-induced RS. Taken together, these data demonstrate that captan exposure induces an RS that requires ATR to prevent DSB formation and genetic instability.

Captan Impairs Replication Fork Progression in an XRCC1-Independent Manner
The early RS induced by captan should rapidly mitigate the proliferative rate of exposed cells. Remarkably, AA8 and EM9 CHO cells treated with captan for 48 h show a similar dose-dependent proliferation defect (Supporting Information Fig. S5), suggesting that the captan induced RS does not merely relies on the presence of the DNA lesions that require XRCC1 to be repaired. To obtain better insights into the RS induced by captan and investigate Environmental and Molecular Mutagenesis. DOI 10.1002/em the impact of XRCC1 deficiency in this process, we analyzed γH2AX distribution through the cell cycle by flow cytometry in AA8 and EM9 CHO cells. We observed a significant increase in γH2AX staining in the S phase of both cell lines in the same range, mostly at the onset of the S phase (Fig. 3A). These data confirm the RS induction by captan highlighted in Figure 2 but suggest that it could be independent of the repair defects caused by XRCC1 Environmental and Molecular Mutagenesis. DOI 10.1002/em deficiency. Then, we assessed the impact of captan treatment on the cell cycle, especially in the S phase, by performing EdU incorporation experiments followed by flow cytometry analysis. In the absence of treatment, almost all AA8 and EM9 CHO cells in the S phase incorporated EdU (Fig. 3B). However, captan induced a huge loss of global EdU intensity only 2 h after treatment ( Fig. 3B and Supporting Information Fig. S6). Consequently, the proportion of EdU-negative S phase cells increased significantly in the same range in both cell lines (Fig. 3B). Impairment of EdU incorporation during replication suggests a slowing down of replication fork velocity. To directly address this question, we performed DNA fiber assays by successive double-pulse labelling with two nucleotide analogues, CldU followed by IdU incorporation, to measure replication fork speed in the DNA fiber. AA8 and EM9 CHO cell lines were treated with captan only during the second pulse labelling (IdU incorporation) and the track lengths were measured (Fig. 3C). We observed a significant decrease in IdU track length, reflecting a slowing down of replication fork velocity in the presence of captan. Moreover, the decrease in replication fork speed was similar between AA8 and EM9 cells, suggesting that the captan-related replication fork slowdown is independent of the repair defects caused by XRCC1 deficiency. To test whether XRCC1 could influence the fork restart, we performed a DNA fiber assay on AA8 and EM9 CHO cell lines exposed to captan only during the first pulse labelling (CldU), then quantified the fork restart by measuring the IdU track length (Fig. 3D). As shown, the replication fork restart was similar in both cell lines, suggesting that XRCC1 depletion does not affect the replication fork restart. Altogether, these results demonstrate that during the S phase, captan exposure inhibits replication fork progression, leading to H2AX phosphorylation, independently of BER pathway.

DISCUSSION
Despite the mutagenic effect of captan reported on bacteria testing strains [e.g. see (Bridges 1975;Barrueco and de la Pena 1988;Ohta et al. 2002)], the genotoxic effect of captan on eukaryotic cells, and in particular on human cells, is still under debate.
Our data indicate that the susceptibility of mammalian cells to captan is linked to its genotoxic potential, as revealed by the sensitivity of XRCC1-deficient cells. XRCC1 polymorphism has been associated with a higher risk of cancers (Hu et al. 2005;Hung et al. 2005). We show here that in absence of XRCC1, captan exposure of mammalian cells induces more micronucleus formation, indicating that XRCC1 may prevent genetic instability after captan. Indeed, XRCC1 plays a major role in maintaining chromosomal stability, its deficiency leading to an elevated level of sister chromatid exchange and chromosomal aberrations (Thompson et al. 1982;Zdzienicka et al. 1992). XRCC1 is a scaffold protein that interacts with different components of the BER machinery to stabilize or stimulate them (Caldecott 2014). BER is a versatile system capable of dealing with a broad range of base modifications, including oxidation, alkylation or deamination as well as abasic sites and single-strand breaks (Krokan and Bjoras 2013). The modified alkaline comet assay allowed us to shed the light on the presence of Fpg-sensitive DNA sites after captan exposure. As Fpg is a bacterial glycosylase mainly active across oxidized purines (Boiteux et al. 2017), our results corroborate bacterial mutagenicity studies with strains that detect oxidative compounds. Moreover, the sensitivity and the repair defects of CHO cells defective in XRCC1 protein demonstrate the role of BER in overcoming the genotoxic effect of captan. The XRCC1 deficient EM9 cells show moderate hypersensitivity to captan, which might suggest that other DNA repair pathways could be required to survive captan exposure. However, EM9 cells have previously been reported to exhibit massive or slight hypersensitivity to base modifications depending on the ROS-generating or methylating agent, which may be correlated to differential XRCC1 requirement in many BER sub-pathways implying different DNA glycosylases (Barrows et al. 1991;Horton et al. 2008). While the highlighting of Fpg-sensitive sites and the moderated sensitivity of XRCC1 deficient cells after captan exposure argue for the formation of oxidized bases such as 8-oxoguanine, our data cannot exclude that other lesions such as alkylated bases are induced. Further studies will be necessary to better characterize the base lesions induced by captan and the cellular mechanisms involved for repair. In summary, these results indicate that captan induces Fpg-sensitive DNA lesions in mammalian cells, presumably altered bases, which require an XRCC1-dependent BER process to be repaired, as depicted in the model (Fig. 4).
The cellular consequences of captan genotoxicity were first analyzed by co-immunodetection of γH2AX and 53BP1. After captan exposure, HeLa cells accumulated more γH2AXthan 53BP1-positive cells. 53BP1 is specifically recruited to DSBs (Schultz et al. 2000), whereas γH2AX can form foci after RS independently of DSB formation (Ray Chaudhuri et al. 2012). This suggests that captan induces an RS that may subsequently result in S-phase-dependent DSBs. Indeed, captan-treated HeLa cells rapidly exhibit enhanced Chk1 phosphorylation and FANCD2 monoubiquitinylation, two hallmarks of RS signaling (Liu et al. 2000;Taniguchi et al. 2002). These data are strengthened by the role of ATR, the major kinase governing the cellular response to RS (Liu et al. 2000;Taniguchi et al. 2002) to survive captan exposure. Finally, the increase in micronuclei-and 53BP1-positive cells after captan exposure in ATR-deficient cells compared to control cells demonstrates that an effective RS response is necessary to prevent DSB formation and genetic instability in response to captan, as depicted in the model (Fig. 4). This indicates that the most dramatic cellular consequences of captan exposure arise during the S phase.
The latter finding prompted us to characterize the RS induced by captan in more detail. Furthermore, we wondered whether the induction of the Fpg-sensitive sites by captan sustains the replication defects. The slowing down of the Fig. 4. The fungicide captan induces two independent genotoxic effects. Left part: captan induces base modifications that require XRCC1 to be repaired. Right part: captan is also responsible for a replicative stress, impacting replicative-fork speed and leading to double-strand breaks. Both effects contribute to the establishment of genomic instability.
replication fork observed from only 20 min after captan treatment was not exacerbated by XRCC1 deficiency. This suggests that in parallel to induction of the Fpg-sensitive DNA lesions detected by the Fpg-modified comet assay, captan also inhibits DNA replication, which is in good agreement with previous studies. Indeed, biochemical analyses have shown that captan could be a general inhibitor of both eukaryotic and prokaryotic DNA polymerases (Dillwith and Lewis 1982;Freeman-Wittig et al. 1989;Snyder 1992). These reports showed that the fungicide impairs initiation of polymerization, but not the elongation step of DNA synthesis, in an irreversible manner by competing with DNA regarding polymerase loading. Consequently, dissociation of the polymerase from the DNA template would promote its inhibition by captan. This observation could explain the difficulties encountered by replicative forks to restart after captan treatment. It also raises the question of potential differences in captan sensitivity between replicative polymerases and specialized DNA polymerases, which are involved in DNA repair, translesion synthesis and replication. On the other hand, we cannot exclude that captan may induce different type of DNA lesions which repair is independent of XRCC1, and that may impede replication fork progression. Indeed, the repair of Fpg-sensitive sites induced by captan is less efficient than those induced by Ro 19-8022, suggesting a more complex situation. In the same way, the previously reported interaction of captan with DNA or proteins associated to chromatin , such as the topoisomerase II inhibition by captan (Rahden-Staron 2002), may be responsible for the RS we report in this study. Further analyses on the possible role of other DNA repair pathways will be necessary to address these issues.
DNA strand breaks and micronuclei are induced as a consequence of RS, which in turn leads to genomic instability. Our study demonstrates that captan exposure in mammalian cells induces various genotoxic effects, corroborating previous in vivo reports (Legator 1969;Feng and Lin 1987;Snyder 1992) and suggesting that these mechanisms could play a role in carcinogenesis. Moreover, our in vitro results on human peripheral lymphocytes, showing DNA damage with alkaline comet assay after captan treatment (Supporting Information Fig. S2), corroborate human in vivo study which has reported a significant positive association between DNA damage using alkaline comet assay and urinary concentration of captan (McCauley et al. 2008). In view of recent epidemiological data pointing to a carcinogenic effect in farmers exposed to captan (Presutti et al. 2016), more data are needed to evaluate the human health risk that this fungicide potentially represents.

ACKNOWLEDGMENT
Dr. Sandrine Ellero-Simatos is greatly acknowledged for her advice on statistical analysis. The authors were supported by grants from the Cancer Plan for PestiBG (n ENV201401) (EBR).

AUTHOR CONTRIBUTIONS
AFV, JV, and EBR designed the study. LCA, MM, MC, AVF, JV, and EBR performed the experiments, analyzed the data, prepared draft figures and tables. AFV, JV, and EBR prepared the manuscript draft with important intellectual input from GM and BS. All authors approved the final manuscript.

CONFLICTS OF INTEREST
The authors declare that there are no conflicts of interest.