Enhancing the sensitivity of the thymidine kinase assay by using DNA repair‐deficient human TK6 cells

Abstract The OECD guidelines define the bioassays of identifying mutagenic chemicals, including the thymidine kinase (TK) assay, which specifically detects the mutations that inactivate the TK gene in the human TK6 lymphoid line. However, the sensitivity of this assay is limited because it detects mutations occurring only in the TK gene but not any other genes. Moreover, the limited sensitivity of the conventional TK assay is caused by the usage of DNA repair‐proficient wild‐type cells, which are capable of accurately repairing DNA damage induced by chemicals. Mutagenic chemicals produce a variety of DNA lesions, including base lesions, sugar damage, crosslinks, and strand breaks. Base damage causes point mutations and is repaired by the base excision repair (BER) and nucleotide excision repair (NER) pathways. To increase the sensitivity of TK assay, we simultaneously disrupted two genes encoding XRCC1, an important BER factor, and XPA, which is essential for NER, generating XRCC1 −/− /XPA −/− cells from TK6 cells. We measured the mutation frequency induced by four typical mutagenic agents, methyl methane sulfonate (MMS), cis‐diamminedichloro‐platinum(II) (cisplatin, CDDP), mitomycin‐C (MMC), and cyclophosphamide (CP) by the conventional TK assay using wild‐type TK6 cells and also by the TK assay using XRCC1 −/− /XPA −/− cells. The usage of XRCC1 −/− /XPA −/− cells increased the sensitivity of detecting the mutagenicity by 8.6 times for MMC, 8.5 times for CDDP, and 2.6 times for MMS in comparison with the conventional TK assay. In conclusion, the usage of XRCC1 −/− /XPA −/− cells will significantly improve TK assay.


| INTRODUCTION
Genotoxicity assessment is essential for developing medicines and ensuring the safety of industrial chemicals. in vitro assessment of genotoxicity precedes its in vivo evaluation in the development of drugs (Corvi and Madia, 2017). The Organization for Economic Co-operation and Development (OECD) has provided guidelines for the testing of chemicals using various in vitro genotoxicity tests including Ames test, the micronucleus test, the mouse lymphoma assay (MLA) and the thymidine kinase (TK) assay . The TK assay uses human lymphoblastoid TK6 cells harboring heterozygous for mutation at the TK gene (TK +/− ) and detects various mutations that inactivate the intact TK allelic gene including point mutations, long deletion, DNA recombination, and chromosome loss (Liber and Thilly, 1982;Koyama et al., 2006). The specificity of the TK assay is likely to be very high for the following reasons. The TFT selection works with extremely high specificity to kill the cells that lost the TK activity (Moore-Brown et al., 1981). The TK gene is a stably expressed house-keeping gene, and mutations of the TK gene, but not other mechanisms, cause the complete irreversible inactivation of the TK activity by the four-hour exposure to chemicals in the TK assay (Clements, 1995). However, the sensitivity of this assay is very low especially in the detection of the chemicals that generate point mutations due to the following reasons. The size of the exons in the TK gene is 0.7 kb, which accounts for only 1.1 × 10 −7 in the whole genome (6 × 10 6 kb). BER normally removes over 50% of the chemical-induced base damage within 0.5 hr (Hoch et al., 2017), replication forks in an asynchronous population of cells encounter only 4% (0.5/13) of the chemical-induced base damage considering 13 hr cell cycle time of TK6, and less than 0.04% (4 × 10 −4 ) of the chemicalinduced base damage causes mutations considering that the error-rate of translesion DNA synthesis (TLS) polymerases is approximately 10 −2 per base (McCulloch and Kunkel, 2008). Thus, chemical-induced base damage causes mutations in the coding sequences of the TK gene with less than 4.4 × 10 −11 probability (1.1 × 10 −7 × 4 × 10 −4 ).
Approximately 20% of the missense and nonsense mutations in cancer-related genes affect oncogenesis (Scott and Meldrum, 2005). If these mutations inactivate the TK gene also at~20% frequency, chemical-induced base damage inactivates the TK gene with less than 10 −11 probability. We therefore need to damage more than 10 5 nucleotides per cell to inactivate the TK gene at least in a single cell of 10 6 cells used for a TK assay. Collectively, the sensitivity of the TK assay is very low as we need to damage such a large number of nucleotides to detect the mutagenic potential of chemicals. We therefore hypothesized that the usage of DNA repair-deficient cells might increase the sensitivity of the currently used genotoxic assays, including TK assay. Chemicals induce point mutations by damaging nucleotides, inaccurate replication of such damaged nucleotides often happened during error-prone TLS (Sale et al., 2012). The vast majority of damaged nucleotides are repaired by the two major pathways, base excision repair (BER) and nucleotide excision repair (NER). BER removes lesions caused by the alkylation, hydrolysis, and oxidation of nucleotides. Typical BER is initiated by an incision of the DNA strand 5 0 to the damaged bases, generating single-strand breaks (SSBs) (Krokan and Bjoras, 2013). Their repair is facilitated by the x-ray repair crosscomplementing group 1 (XRCC1) protein, which provides docking sites for various BER effector enzymes (Thompson et al., 1990;Caldecott, 2003). Hypersensitivity of XRCC1 deficient cells to alkylating agents and H2O2, indicates the vital role of XRCC1 in BER (El-Khamisy et al., 2003). A typical alkylating agent is MMS, which modifies both guanine (to 7-methylguanine) and adenine (to 3-methlyladenine) generating base mispairing and replication blocks, respectively (Beranek, 1990). NER removes helix-destabilizing bulky adducts generated by crosslinking agents and UV (Aboussekhra et al., 1995). XPA recognizes such bulky adducts and is essential for initiating NER (De Vries et al., 1995). The capability of NER is evaluated by measuring cellular sensitivity to UV and crosslinking agents such as MMC and CDDP, which induce protein-DNA crosslinks, intrastrand, and interstrand crosslinks (Weng et al., 2010;Dasari and Bernard, 2014). Both XRCC1 and XPA prevent the induction of point mutations through absolutely accurate and error-free repair mechanisms (Lindahl, 1999). We therefore hypothesized that XRCC1/XPA double knockout cells may accumulate a higher number of mutations following a wide variety of base-damaging substances in comparison with wild-type cells.
In this study, we disrupted both XRCC1 and XPA genes in the TK6 cell line and used the resulting XRCC1 −/− /XPA −/− cells for the TK assay. We measured the mutagenicity of four mutagenic alkylating agents, MMS, CDDP, MMC, and CP following the OECD guideline for TK assay (TG-490). This TK assay using XRCC1 −/− /XPA −/− cells detected 2-8 times higher numbers of mutations when compared with the conventional TK assay using wild-type TK6 cells.

| Test chemicals
We purchased MMS and CDDP from Nacalai Tesque Inc (Kyoto, Japan), MMC from Sigma-Aldrich Inc. (CA), and CP from FUJIFILM Wako Pure Chemical Co. (Tokyo, Japan). We purchased liver S9 prepared from SD rats treated with phenobarbital and 5,6-benzoflavone from BoZo Research Center Inc (Tokyo, Japan). All test chemicals were dissolved in phosphate-buffered saline PBS purchased from Takara Bio Inc. (Shiga, Japan). All test chemicals were prepared immediately before the TK test.

| TK gene mutation assay
We examined mutation frequencies as described (Koyama et al., 2011) and following the OECD guidelines (TG-490). In brief, we incubated cells for 4 hr either with CDDP, MMC, or MMS in the absence of S9 mix or with CP together with S9 mix, washed with PBS, resuspended in a fresh medium and cultured for 3 days to allow for the expression of the TK deficient phenotype. To determine the plating efficiency of cells treated with mutagens, we seeded cells at 1.6 cells/well in 96-microwell plates in the absence of TFT at day 0 (PE0) and day 3 (PE3). To count the number of cells carrying TK −/− allelic genes, we seeded the cells at day three into 96-microwell plates at 40,000 cells/well in a medium containing 3.0 μg/ml trifluorothymidine (TFT), which kills only TK + cells carrying an intact TK gene. All plates were incubated at 37 C in 5% CO 2 in a humidified incubator. To The DT-A-pA/loxP/PGK-hisD R -pA/loxP or DT-A-pA/loxP/PGK-Bsr R -pA/loxP vector was digested with both ApaI and AflII, which cut at the 5 0 and 3 0 of the selection marker genes, respectively. We combined the digested DNAs with the 5 0 -and 3 0 -arms using the Seamless Cloning and Assembly Kit (Thermo Fisher Scientific). We transfected 6 μg each of the TALEN-expression plasmids and 2 μg each of the two gene-targeting vectors carrying hisD and Bsr R into 4 × 10 6 TK6 cells using the Neon Transfection System (Life Technologies) with three times 1,350 V pulse with a 10 ms pulse width. at 1 μg/ml CP and 6.6 ± 0.8 × 10 −6 at 3 μg/ml CP (Figure 2b). Spontaneous MF of XRCC1 −/− /XPA −/− cells was 9.5 ± 1.3 × 10 −6 , and their MF was 12.7 ± 1.3 × 10 −6 at 1 μg/ml CP and 20.9 ± 6.1 × 10 −6 at 3 μg/ml CP (Figure 2b). Collectively, after comparing the slopes of the minimum dose-response, we found that the usage of XRCC1 −/− / XPA −/− cells for the TK assay increases the detection of the mutagenicity associated with 1 and 3 μg/ml CP by 4.5 folds in comparison with the conventional TK assay, p-value = .02 (Table S1). However, the difference between spontaneously arising MFs and induced ones at the minimum concentration of CP was not statistically significant in wildtype or XRCC1 −/− /XPA −/− cells (Figure 2b). while wild-type TK6 shows a significant induction of mutations at 0.05 μg/ml MMC (p-value = .030) but not at 0.025 μg/ml MMC (p-value = .063) (Figure 4b).
RS was more than 50% (Figure 5a). The dose-response slope for XRCC1 −/− /XPA −/− showed 8.5 fold increase than wild-type TK6 at F I G U R E 3 The mutation frequency (MF) of wild-type and XRCC1 −/− /XPA −/− TK6 cells induced by MMS. Data are shown as in Figure 2. Error bars represent SD from at least three independent experiments. The statistically significant difference between spontaneously arising MFs and induced ones was calculated by the Student's t test. We defined p-value < .05 as statistically significant and mark such difference with *. (ns) p-value was not significant 0.25, 0.5 μM CDDP, p-value <.0001 (Table S4) To check the ability of XRCC1 −/− /XPA −/− cells to detect weak mutagens with more sensitivity, we determine the minimum doses of the genotoxic agents whose doses increased the MF in a statistically significant manner. We found that the TK assay using XRCC1 −/− / XPA −/− TK6 enhanced a significant detection of mutations at the minimum concentrations of DNA cross-linking agents MMC (0.025 μg/ml) and CDDP (0.5 μM) while the wild-type TK6 could not detect similar significant increase (Figures 4b and 5b). Similarly, the TK assay using  Figure 2. Error bars represent SD from at least three independent experiments. The statistically significant difference between spontaneously arising MFs and induced ones was calculated by the Student's t test. We defined p-value < .05 as statistically significant and mark such difference with *. (ns) p-value was not significant type and XRCC1 −/− /XPA −/− cells (Figure 2b). CP generates DNA cleavage, crosslinks, and adducts (reviewed in Ozolins c, 2010), and MMS is a mono-alkylating agent (Sobol et al., 2002). MMC and CDDP, on the other hand, are DNA cross-linking agents generating a variety of lesions, including protein-DNA crosslinks, intrastrand crosslinks, and interstrand crosslinks (Tomasz, 1995;Jordan and Carmo-Fonseca,-2000;Lorenti Garcia et al., 2009). XPA and XRCC1 participate in the repair of crosslinks, and these pathways have an overlapping role in removing a fraction of the crosslink DNA lesions (Zheng et al., 2003;Mustra et al., 2007;Zhang and Walter, 2014;Semlow et al., 2016).
This overlapping role may explain why the usage of XRCC1 −/− /XPA −/− cells increased the sensitivity of the TK assay for detecting the mutagenicity of MMC and CDDP to greater extents when compared with MMS, which induces the DNA lesions that are repaired exclusively by XRCC1-dependent BER (Op Het Veld et al., 1998). In summary, the usage of XRCC1 −/− /XPA −/− cells is advantageous, particularly when the TK assay examines the mutagenicity of crosslinking agents.
Enhancing the performance of the in vitro genotoxicity testing would lead to more reliance on the in vitro tests and less of a need to use in vivo tests. The EURL EVCAM has requested the improvement of the individual in vitro genotoxicity detection assays to increase their overall performance and as a consequence, minimize or even prevent the use of animals for detecting mutagenic chemicals (Corvi and Madia, 2017;EURL ECVAM, 2013a). A significant concern of the mammalian cell-based gene mutation tests is their limited sensitivity, and these tests need very high concentrations of chemicals, whose genotoxicity might not be extrapolated to the genotoxicity of environmentally appropriate lower levels of the substances (Elespuru et al., 2009).
The usage of DNA repair-deficient cells may solve this problem by increasing the sensitivity of the metazoan cell-based gene mutation tests (Ji et al., 2009;Yamamoto et al., 2011;Nishihara et al., 2016). Moreover, this usage gives an insight into molecular mechanisms underlying the mutagenicity of chemicals. For example, higher induced MF in XRCC1 −/− /XPA −/− cells than wild-type cells indicate that relevant compounds cause the DNA lesions that are repaired by either XRCC1-dependent BER or XPA-dependent NER. We propose the following two-step examination of genotoxic chemicals; first, the identification of a wide variety of genotoxic chemicals using a few mutants such as XRCC1 −/− /XPA −/− and double-strand-break repair mutant cells (Hsieh et al., 2019), and subsequently, characterization of molecular mechanisms underlying the identified genotoxicity using cells deficient in individual repair pathways that repair specific lesions. Our proposal for the multistep process is not for routinely evaluating genotoxicity but for investigating mutagenic mechanisms for special purposes, for example, the elucidation of molecular mechanisms underlying the mutagenesis of different chemicals in academic research. In conclusion, we propose that the usage of XRCC1 −/− /XPA −/− cells will improve the sensitivity of the TK assay; however, further validation is still needed before requesting it for regulatory use.

ACKNOWLEDGMENTS
We are grateful to M. Kato and A. Kobayashi for technical assistance.