In Vitro PIG-A Gene Mutation Assay in Human B-Lymphoblastoid TK6 Cells

Abstract The X-linked PIG-A gene is involved in the biosynthesis of glycosylphosphatidylinositol (GPI) anchors. PIG-A mutant cells fail to synthesize GPI and to express GPI-anchored protein markers (e.g., CD59 and CD55). In recent years, in vitro PIG-A assay has been established based on the high conservation of PIG-A/Pig-a loci among different species and the large data from the in vivo system. The purpose of this study was to extend the approach for PIG-A mutation assessment to in vitro human B-lymphoblastoid TK6 cells by detecting the loss of GPI-linked CD55 and CD59 proteins. TK6 cells were treated with three mutagens 7,12-dimethylbenz[a]anthracene (DMBA), N-ethyl-N-nitrosourea (ENU), etoposide (ETO), and two nonmutagens: cadmium chloride (CdCl2) and sodium chloride (NaCl). The mutation rate of PIG-A gene within TK6 cells was determined on the 11th day with flow cytometry analysis for the negative frequencies of CD55 and CD59. The antibodies used in this production were APC mouse-anti-human CD19 antibody, PE mouse anti-human CD55 antibody, PE mouse anti-human CD59 antibody, and nucleic acid dye 7-AAD. An immunolabeling method was used to reduce the high spontaneous level of preexisting PIG-A mutant cells. Our data suggested that DMBA-, ENU-, and ETO-induced mutation frequency of PIG-A gene was increased by twofold compared with the negative control, and the effects were dose-dependent. However, CdCl2 and NaCl did not significantly increase the mutation frequency of PIG-A gene, with a high cytotoxicity at a dose of 10 mmol/L. Our study suggested that the novel in vitro PIG-A gene mutation assay within TK6 cells may represent a complement of the present in vivo Pig-a assay, and may provide guidance for their potential use in genotoxicity even in cells with a significant deficiency of GPI anchor.


Introduction
Genotoxicity tests are utilized to identify compounds with a potential risk for carcinogenicity and heritable mutations, assay (such as Muta Mouse, BigBlue mice, etc.), and in vitro gene mutation assay of mammalian cell lines (ICH 2012). The overall aim of these test systems is to generate a solid database for hazard identification with respect to mutagenicity and to assess the mechanisms of action of chemical carcinogens.
Currently, mutation assay for a rodent-based endogenous phosphatidylinositol class A gene (PIG-A in humans and Pig-a in rodents) has become a recognized method for detecting potential mutagenicity of exogenous compounds. 2 The PIG-A/Pig-a gene is involved in early synthesis of cell-surface glycosylphosphatidylinositol (GPI) anchors, which was usually employed by GPI anchor proteins (e.g., CD59, CD55, or CD90) to bind to the cell membrane. Of all the genes involved in the GPI anchor synthesis, only the PIG-A/Pig-a gene is located on the X chromosome. Therefore, in each cell, there is only one copy of the functional PIG-A/Pig-a gene, and a single inactivating mutation in this gene will result in the inactivation of the enzymatic function of the PIG-A protein and the subsequent deficiency of all GPI-linked surface proteins. 3,4 Flow cytometry is the most common method for diagnosis of GPI(À) frequency (PIG-A gene mutation rate) in TK6 human lymphoblastoid cells. [5][6][7][8][9] From a throughput standpoint perspective, flow cytometry analysis of Pig-a mutations may be useful for in vitro analysis of gene mutation and development of high-density dose-response data. 10 Evidence suggests that an in vitro trial may also be conceived of as the value of testing hypotheses about the results of an in vivo assay (e.g., investigating negative in vivo responses) and of prescreening compounds for in vivo testing. 11 Compared with the existing in vitro gene mutation assays (such as mouse lymphoma assay [MLA] test and hypoxanthine-guanine phosphoribosyl transferase [HPRT] gene mutation test), the in vitro PIG-A gene mutation test has the following three advantages: (1) being able to select human cells with complete function of P53 gene; (2) reducing the false-positive rate of in vitro genotoxicity test; and (3) shorter detection period (only 11 days) in comparison to MLA or HPRT gene mutation tests. Therefore, in vitro PIG-A gene mutation test using flow cytometry may be more objective, convenient, and high-throughput. 9,12 TK6 cells, known as GPI-anchor-negative cells, grow in suspension, and have good practical usage in genotoxicity testing, and this may be attributed to its well-known characteristics for flow cytometric analysis, as well as the available access of the standardized cultures for genetic toxicological assessment from cell repositories. [13][14][15] In this study, a simple and efficient immunomagnetic separation was explored in TK6 cells to obtain a relative low and stable GPI(À) background frequency. A PIG-A gene mutation assay was then conducted through flow cytometry assay for detecting the loss of GPI-linked CD55 and CD59 using mutagen and nonmutagenic compounds with different mechanisms of action. Our data suggested that the development and testing of methods for in vitro PIG-A gene mutation assay may represent a promising strategy to investigate the potential mutagenicity of chemical or physical mutagens.

S9 Metabolic Activation System and Test Article Dose Formulation Preparation
The mammalian liver post-mitochondrial fraction (S9) was purchased from Molecular Toxicology, Inc. (Moltox Inc., Boone, North Carolina, United States), which was prepared from male Sprague-Dawley rats that had been injected with Aroclor 1254 at 500 mg/kg. The liver microsome enzyme system (S9 mixture) was the mixture of S9 and S9 cofactors containing NADP (4 mmol/L), glucose-6-phosphate (5 mmol/L), KCl (30 mmol/L), MgCl 2 (10 mmol/L), PBS at pH 7.4, and deionized water. The final concentration of S9 in the culture medium was 1.0% (v/v). temperature for 30 minutes in the dark. After washing with staining buffer (PBS with 2% BSA [w/v], precooled) (1 mL Â 2), cell pellets were resuspended in 500 μL DNA staining solution (PBS with 2% BSA; 7-AAD for 2 μL/sample) for 10 minutes on ice to exclude dead cells, and then centrifuged. The supernatant was discarded and cells were fixed in 600 μL fixation buffer (PBS with 1% formaldehyde [w/v] and 2.5 μg/mL AD). Cells were placed on ice before flow cytometry analysis. All solutions or buffer was kept on ice after preparation.

Antibody Labeling
To ensure the accuracy, four samples of 2 Â 10 6 TK6 cells were separately prepared and labeled with individual antibody staining solution for further 30 minutes of incubation in the dark. The staining solutions included staining buffer with mouse anti-human CD19-APC (20 μL/sample), or staining buffer with mouse anti-human CD55-PE (12 μL/sample) and mouse anti-human CD59-PE (20 μL/sample), or DNA staining solution. After completion, all samples were rinsed with staining buffer (1 mL Â 2), and cells were fixed in 600 μL fixation buffer and placed on ice before analysis.

Flow Cytometry Assay for GPI(À) Frequency Analysis
All flow cytometric analyses were conducted with an Accuri C6 PLUS from Becton-Dickinson. Excitation and emission detection measurements of the applied fluorescence dyes were as follows: APC (640-670/14 nm) and PE (488-575/26 nm). Samples were analyzed with a middle rate or 65 μL/min. For each determination of the GPI(À) frequency (PIG-A gene mutation rate), at least 10 6 cells were collected.

Cleansing of Preexisting GPI(À) TK6 Cells by Immunomagnetic Separation
For cleansing of preexisting GPI(À) TK6 cells, $4 Â 10 6 TK6 cells were pelleted at 200 g for 5 minutes and resuspended in 2 mL staining solution. The cells were labeled at room temperature for 30 minutes in the dark. After labeling, the cells were washed with staining buffer twice, and then pelleted at 200 g for 5 minutes. The cell pellets were resuspended in 500 μL PBS (with 2% BSA and precooled) containing 40 μL anti-PE microbeads for 30 minutes on ice or at 2 to 8°C protected from light. After incubation, the cells were washed by 10 mL staining buffer (precooled), and resuspended in 2 mL staining buffer (precooled). LS columns were placed in the MACS Separator and prewetted with a 5 mL staining buffer (precooled) in a biosafety cabinet. Then 2 mL cell suspension was added into the LS columns and the cells were allowed to freely flow through the column by gravity. After all of the suspension was run out, 5 mL staining buffer (precooled) was loaded onto the LS column for three times. Then the LS column was taken out from the separator and 5 mL staining buffer (precooled) was added. The cells adsorbed in the LS column were extruded with a piston and collected. The cell suspension was centrifuged at 200 g for 5 minutes. Pellet was resuspended in 10 mL culture medium for expansion culture. After expansion, GPI(À) frequency was analyzed by flow cytometry. If GPI(À) frequency of TK6 cells was too high, the above steps were repeated and performed the cleaning steps again.

Determination of Doubling Times and Spontaneous Mutation Rate of TK6 Cells
Preexisting GPI(À)-cleansed TK6 cells were named TK6 GPIþ cells. For doubling time determination, TK6 GPIþ cells were seeded in triplicate at a density of 10 5 and 2 Â 10 5 cells/mL in a total volume of 25 mL medium. Cell growth was followed for 3 days by determining cell numbers every day. The doubling time was calculated according to Eq. (1): where D.T. is the doubling time, t 2 the cell harvest time, t 1 the cell inoculation time, Ch the cell harvest density, and Ci the cell inoculation density.
TK6 GPIþ cells were cultured and passaged for 40 days and 10 6 cells were subcultured in 25 mL RPMI-1640 every 2 days. GPI(À) frequency was tested every 4 days, and every 8 days for the last two times. Based on the TK6 doubling time, the GPI(À) frequencies were plotted over the corresponding population doublings. After linear regression was applied (fit curve by GraphPad Prism 7), the resulting slope value was equivalent to the spontaneous mutation rate (μ) that was calculated according to Eq. (2): where α is the slope of the fit curve, D.T. the doubling time of TK6 GPIþ cell, and 24 the 24 hours.

Optimized Cytotoxicity Determination
The cytotoxicity of test chemicals was measured to design appropriate dose ranges for mutagenicity testing by relative increase in cell counts (RICC). Cell density was determined by cell counter methods and RICC was calculated at 24 and 48 hours in the group without S9. In the group with S9 (24 hour À S9 group), TK6 GPIþ cells were treated with articles for 3 hours (3 hour þ S9 group) and rinsed with 10 mL PBS twice followed by cell density determination at 24 and 48 hours. Treatments which showed RICC between 10 and 20% of control and lower at 24 and 48 hours were selected as the top concentrations. If not limited by cytotoxicity, 10 mmol/L or precipitate concentration was chosen as the top concentration. RICC was calculated according to suspension was mixed by gentle vortex. Cells were directly cultured for 24 hours at 37°C, 5% CO 2 in a humidified environment for the 24 hour À S9 group. Then, the cells were rinsed with PBS and resuspended with 10 mL medium after 3 hours of treatment and with a continuous culture for 24 hours for the 4 hour þ S9 group. After 24 hours, cells were washed with PBS. A total of 2 Â 10 6 cells were subcultured every 2 days in 25 mL culture medium to allow phenotype expression. On day 11, flow cytometry was performed to detect GPI(À) frequencies.

Test Result Evaluation Criteria
The biological relevance of the results was carefully examined. The response was considered positive when the following conditions were met: the test agents induced a twofold increase over negative control in the frequency of GPI(À) cells in a dose-dependent manner with Microsoft excel 2007.

Template of GPI(À) Frequency Analysis by Flow Cytometry
The occurrence of GPI mutations is extremely rare, which requires very sensitive methods to detect. To achieve this, two independent GPI-anchored proteins, CD55 and CD59, were labeled. The GPI(þ) cells were separated from GPI(À) cells by staining CD55 and CD59 with PE-conjugated antibodies, and an APC-conjugated CD19 antibody was used to identify most of the TK6 cells as B-lymphocytes.
In this experiment, the gating procedure for flow cytometric analyses was set up using parallel samples and the results are presented in ►Fig. 1. Cells were collected by FSC/SSC scatter plot (►Fig. 1A, gate cells) and agglomeration cells were excluded by gating in FSC-A versus FSC-H scatter plots (►Fig. 1B, gate single cells). Dead cells were excluded via the uptake of 7-AAD, and healthy cells were collected (►Fig. 1C). Finally, GPI(þ) and GPI(À) cells were separated. The GPI(þ) cells showed positive APC and PE-fluorescent signals and were presented in the upper right quadrant (►Fig. 1D). Cells with nonspecific events lacking the APC signal were presented in the upper left quadrant and lower left quadrant, and were excluded from analysis (►Fig. 1D). In this study, a total of 10 6 APC-positive cells were collected to analyze the GPI(À) frequency.

Cleansing of Preexisting GPI(À) TK6 Cells
Flow cytometric analysis showed that the spontaneous mutation rate of the GPI(À) frequency in TK6 cells was $23.4% (►Fig. 2A), which is a very high rate, and the basic mutation frequency of GPI(À) cells was reduced to 0.18% (►Fig. 2B) after being cleared by the first magnetic bead adsorption, and 13.5 Â 10 À6 (►Fig. 2C) after the second clearance, suggesting that immunomagnetic separation is a very useful method to separate and cleanse the GPI(À) TK6 cells. The cells after cleansing of mutant GPI(À) cells are called TK6 GPIþ cells.

Doubling Time and Spontaneous Mutation Rate of TK6 GPIþ Cells
Cell doubling time has a decisive influence on the efficiency of genotoxicity testing. A stable cell doubling time is also a sign of good cell status. Our result showed that TK6 GPIþ cells had a doubling time of 15.5 AE 1.2 hours.
The spontaneous GPI(À) rate was determined by cell division in the culture. TK6 GPIþ cells were tested for spontaneous GPI(À) rate up to 40 days, and the mean spontaneous GPI(À)

Optimized Cytotoxicity Determination
To detect the mutagenic effect, the dose of the treatment agents should be sufficiently high. However, the cytotoxici-ty should not be less than 10% for the highest concentration, otherwise the cells could not be recovered in 11 days' culture. In our study, the doses were selected according to the higher RICC in 24 or 48 hours after cell exposure to the test agents (►Fig. 4). If a reduction of RICC was not limited up to 10 mmol/L or precipitation concentration, as in the case of NaCl, 10 mmol/L was selected as the highest concentration. The cytotoxicity of ETO, DMBA, and CdCl 2 reached the maximum at 24 hours. Therefore, the concentrations of ETO and CdCl 2 were selected based on the RICC of 24 hours. However, the concentrations of ENU were selected based on the RICC of 48 hours (►Table 2). Furthermore, the kinetics of RICC tested in 11 days' culture of all test articles (including EMS and B[a]P as positive) were determined. Our data suggested that the trend of RICC initially rises, then falls, and finally returns gradually to normal level in 11 days (►Fig. 5).   In Vitro PIG-A Gene Mutation Assay in Human B-Lymphoblastoid TK6 Cells Zhou et al. e81

Negative and Positive Effects
Four experiments were performed in this study with 1% (v/v) dimethylsulfoxide (DMSO) or deionized water as negative control and 200 μmol/L EMS (24 hour À S9) or 16 μmol/L B[a] P (4 hour þ S9) as positive control. The PIG-A mutation rate of the negative control groups ranged from 67.6 to 75 Â 10 À6 with an average of 71.4 Â 10 À6 , while the PIG-A mutation rate of the positive control groups ranged from 142.3 to 282.6 Â 10 À6 with an average of 228.2 Â 10 À6 . The high variability of the positive control groups may be attributed to the test article or manipulation (►Fig. 6). Sodium chloride (NaCl): NaCl is a commonly used inorganic salt and a known nongenotoxic compound selected as a negative test article. NaCl was not cytotoxic to TK6 GPIþ cells   In Vitro PIG-A Gene Mutation Assay in Human B-Lymphoblastoid TK6 Cells Zhou et al. e82 and the solubility of NaCl in the medium was higher than 10 mmol/L, so the highest concentration of NaCl was selected as 10 mmol/L. In this experiment with the treatment without in vitro ÀS9 for 24 hours, NaCl did not cause a concentration correlation of the PIG-A gene mutation rate (►Fig. 7). The results of this NaCl were classified as negative in this experiment, consistent with the results of Ames and MLA tests for NaCl. 16 N-nitroso-N-ethylurea (ENU): ENU acts as a potent alkylating agent of the nitrosourea class and it is a direct mutagenic agent that does not require metabolic activation. 17 ENU is known to induce transitions and reversals after O 2À and O 4À alkylation of thymine residues and O 6À alkylation of guanine residues. 18 ENU was positive under mouse lymphoma TK gene mutation assay (MLA) test and HPRT gene mutation test in the absence of S9 conditions. In this experiment, 400 and 800 μmol/L of ENU induced a concentration-dependent increase of PIG-A gene mutation rate in TK6 GPIþ cells and more than twofold of the negative control (►Fig. 7), which was a positive result and was consistent with the published results of in vivo/in vitro PIG-A mutation assays. 7 Etoposide (ETO): ETO is a topoisomerase II inhibitor that, upon entry into the nucleus, forms a drug-enzyme-DNA complex with DNA that interferes with DNA topoisomerase II, resulting in irreparable DNA damage. 19,20 ETO was classified as a carcinogen (group 2A) that may be carcinogenic to humans by International Agency for Research on Cancer (IARC). In in vitro genotoxicity studies, ETO was weakly positive for strain TA98 in the Ames test, induced high frequency of small and large colony mutants in the MLA test, and induced chromosomal structural aberrations in the chromosomal aberration test. 21,22 In the present experiments, the PIG-A gene mutations in TK6 cells induced by 0.055 and 0.11 μmol/L of ETO were more than twofold higher than those in negative controls under conditions without metabolic activation system and with a dose-related manner in the absence of the S9-mix (►Fig. 7), confirming the roles of ETO as a mutagen.
Cadmium chloride (CdCl 2 ): CdCl 2 is a cadmium heavy metal salt, and IARC classifies cadmium and cadmium-like compounds as Class I carcinogens (possibly carcinogenic to humans). 22 However, studies on CdCl 2 have found that it is not a direct DNA-acting agent, and the possible carcinogenic mechanism of CdCl 2 is to stimulate and promote cell division by inhibiting DNA repair mechanisms and upstream signals of apoptosis through epigenetic pathways. 23 The Ames test of CdCl 2 had both negative and positive results (without S9 metabolism), while the results of the Bhas42 cell transformation assay of CdCl 2 suggested that CdCl 2 is a cancer promoter and noninitiator. 24 Our data showed that a high cytotoxicity with RICC of 7.2% was induced by 2.2 μmol/L of CdCl 2 (►Fig. 7). However, no significant and concentrationdependent increase was noted at all tested concentrations of CdCl 2 in the PIG-A gene mutation rate. Thus, CdCl 2 was tested as a nonmutagen in this experiment.
7,12-dimethyl-benz[a] anthracene (DMBA): DMBA is a polycyclic aromatic hydrocarbon that is oxidized by P450 enzymes (CYP1A1 and 1B1) to form DNA adducts covalently with DNA, which can lead to DNA damage and gene mutations. DMBA is commonly used in animal tests to induce skin and mammary tumors, as well as leukemia and other tumors. 25 A significant and concentration-dependent increase was noted in the PIG-A gene mutation rate in TK6 cells at concentrations of 7.5, 15, and 30 μmol/L DMBA in the presence of S9-mix (►Fig. 7), suggesting that DMBA is a mutagen, which is consistent with the published findings.

Conclusions
The aim of this study was to develop an in vitro PIG-A gene mutation assay in TK6 cells via detection of the loss of GPIlinked CD55 and CD59 proteins by flow cytometry. Since mutations occur at a low frequency, the sensitivity of mutation detection methods is particularly important. To establish a highly sensitive PIG-A gene mutation assay, cleansing of preexisting GPI(À) TK6 cells was performed, and TK6 GPIþ cells were obtained by immunomagnetic sorting. Subsequently, the doubling time and spontaneous mutation rate of TK6 GPIþ cells were characterized. We reduced the effect of the spontaneous mutation rate on the sensitivity of the assay by controlling the subculture time for TK6 GPIþ cells that had a higher spontaneous mutations rate. We also found that RICC was important in the assay, with some compounds showing greater differences in RICC at 24 and 48 hours. If we selected the RICC at 24 hours as the concentration selection criterion, some chemicals may induce all cells to die on day 4 or 5 of the formal assay. ENU, ETO, DMBA, CdCl 2 , and NaCl were selected as subjects to validate the in vitro PIG-A assay, and they showed a positive result that each result of test articles was as expected.
TK6 cells have a normal P53 status which contributes to the maintenance of genomic integrity through recombinational repair. 26 The TK6 cell line is mainly used in in vitro mammalian cell gene mutation tests using the thymidine kinase gene. The mutagens (X-rays, EMS, methyl methanesulfonate, and mitomycin C) have been proved to dosedependently induce TK mutations in TK6 cells. 27 In contrast to the low spontaneous mutation frequency of TK mutation, the spontaneous mutation frequency of PIG-A mutation was high, and it still elevated during culture after cleansing of preexisting GPI(À) cells. Therefore, the passage time of TK6 cells should be controlled within 4 weeks. Besides the PIG-A mutations, PIG-L mutations also cause GPI-deficient isolates in the TK6 cell line, as PIG-L is heterozygous in these cells. 28 Thus, although the TK6 cell assay may not be totally analogous to the in vivo Pig-a assay, it has been shown that the sensitivity of the TK6 cell assay may benefit from having both an X-linked and an autosomal reporter of mutation in terms of increasing the types of mutations that the assay can detect.
In summary, an in vitro PIG-A gene mutation detection method based on flow cytometry and a TK6 GPIþ cell line was successfully established and preliminarily validated in this study. The method uses automated flow cytometry analysis to achieve high-throughput detection. The in vitro PIG-A gene mutation assay has a shorter experimental period and is relatively simple to perform in comparison to MLA assay and HPRT gene mutation assay. Therefore, the in vitro PIG-A mutation assay is expected to complement the in vivo Piga assay with some distinct advantages compared with other in vitro mammalian mutagenicity tests.