4-Nitroquinoline 1-Oxide Forms 8-Hydroxydeoxyguanosine in Human Fibroblasts through Reactive Oxygen Species

4-Nitroquinoline 1-oxide (4NQO) is thought to elicit its carcinogenicity by producing DNA adducts after being metabolized to 4-hydroxyaminoquinoline 1-oxide, which forms 8-hydroxydeoxyguanosine (8OHdG), oxidative damage. To determine whether reactive oxygen species (ROS) are involved in the generation of 8OHdG by 4NQO, we used high-performance liquid chromatography and immunohistochemistry to measure the levels of 8OHdG in normal human fibroblasts treated with 4NQO. The extent of ROS induced by 4NQO was determined by using fluorescent probes to detect ROS, electron paramagnetic resonance spectrometry using a cell-free system, and measurement of intracellular glutathione (GSH) levels. In fibroblasts, 4NQO dose dependently increased 8OHdG levels. Hydrogen peroxide (H2O2) and superoxide were detected in cells treated with 4NQO by using dichlorofluorescin diacetate and hydroethidine, respectively. The addition of catalase to culture medium reduced 8OHdG levels and the intensity of dichlorofluorescin fluorescence, while 4NQO generated hydroxyl radicals in the cell-free system. These findings suggest that 4NQO treatment leads to formation of superoxide, H2O2, and hydroxyl radicals, resulting in the production of a substantial amount of 8OHdG in DNA. Neither the level of 8OHdG nor that of GSH had returned to the basal level 24 h after removal of 4NQO even at a concentration as low as 1 microM. Our results suggest that generation of ROS and depletion of GSH in cells are also important factors for the generation of 8OHdG by 4NQO. This paper describes practical and sensitive ways to detect ROS and 8OHdG and discusses a new functional pathway to elicit genotoxicity.

4-Nitroquinoline 1-oxide (4NQO) is a powerful chemical carcinogen (Kitano, 2000;Nagao and Sugimura, 1976) whose carcinogenic action is thought to be initiated by the enzymatic reduction of its nitro group.The four-electron reduction product 4-hydroxyaminoquinoline 1-oxide (4HAQO) is believed to be the proximate carcinogenic metabolite of 4NQO.When metabolized to an electrophilic reactant, selyl-4HAQO (Tada, 1975), it reacts with DNA to form stable quinoline monoadducts such as 3-(deoxyadenosin-N 6 -yl)-4AQO and N 4 -(guanosin-7-yl-4AQO) (Kohda et al., 1991), which are considered responsible for its mutagenicity and genotoxicity.Kohda et al. (1986) found that a substantial amount of 8hydroxydeoxyguanosine (8OHdG) was formed in DNA from Ehrlich ascites cells exposed to 4NQO and in calf thymus DNA treated with 4HAQO in vitro in the presence of seryl adenosine monophosphate (seryl-AMP).The pairing of 8OHdG with adenine as well as cytosine during DNA replication results in G:C to T:A transversion mutations (Moriya, 1993).In fact, G:C to T:A mutations were predominant among those found in 4NQO-induced tumors (Ide et al., 2001).Since 8OHdG can be generated by oxygen radical-producing agents, Kohda et al. (1987) tested the effect of radical scavengers on the 4HAQOinduced 8OHdG formation, but they could not detect any.They therefore concluded that the 8OHdG produced by 4HAQO can be attributed to 4HAQO-DNA adducts but is not mediated by reactive oxygen species (ROS).
Recently, it has been reported that 4NQO is a powerful inducer of the soxS (superoxide response gene) of Escherichia coli and an E. coli mutant strain that is deficient in either superoxide dismutase (SOD) or oxidative repair enzymes (Dxth, nfo-1 : kan: deficient in exonuclease III and endonuclease IV).These mutant strains were found to be hypersensitive to killing by 4NQO compared to their repair proficient counterparts (Nunoshiba and Demple, 1993).It is thus possible that formation of 8OHdG by 4NQO is mediated by ROS as well as by DNA adducts.
These findings prompted us to ascertain whether 8OHdG production by 4NQO is in part elicited by ROS formation.We detected the production of H 2 O 2 and superoxide in human fibroblasts treated with 4NQO, and we were able to provide direct evidence of the involvement of ROS in the generation of 8OHdG by 4NQO in addition to by DNA adducts, which were previously believed to be solely responsible for its carcinogenicity.Many carcinogens share common mechanisms for inducing strong genotoxic activity, and identification of these mechanisms would make a significant contribution to the development of countermeasures against environmental carcinogens.We utilized a sensitive method to detect 8OHdG at a much lower concentration than the minimum concentration needed for detection by high-performance liquid chromatography (HPLC) using the immunohistochemical method, thus providing a useful and practical way to screen possible environmental carcinogens.Involvement of oxidative stress in target cells and its carcinogenic potential are also discussed.

MATERIALS AND METHODS
Cell culture.Human primary fibroblasts, designated N-1, N-2, and N-3, were grown from skin biopsies of three volunteers, a 50-year-old female, a 26year-old female, and a 29-year-old male.Informed consents for use of the fibroblasts were obtained from all participants.Human neonatal dermal fibroblasts were purchased from Cambrex (Walkersville, MD).All cell strains were maintained in Dulbecco's modified minimum essential medium (DMEM; Nissui, Tokyo, Japan) supplemented with 13% fetal bovine serum (FBS; HyClone, Irvine, CA) in humidified air with 5% carbon dioxide at 37°C.For this study, all fibroblasts were used within eight passages.
4NQO treatment.4NQO (Wako Pure Chemical, Osaka, Japan) was freshly prepared for each experiment from a stock solution (5mM in ethanol) and diluted with DMEM to various concentrations.The final concentration of EtOH in the medium did not exceed 1%.A preliminary experiment showed that the addition of 1% EtOH to the medium did not have any effect on the assay systems or cellular viability.Cells were rinsed with DMEM without FBS prior to treatment.Fibroblasts were treated with 1nM-50lM of 4NQO for 15, 30, 45, or 60 min at 37°C depending on the treatment conditions.Immediately after each treatment, cells were rinsed twice with PBS.
8OHdG measurement by means of HPLC using an electrochemical detector.For 8OHdG measurements, 3 3 10 6 cells were plated onto 100-mm tissue culture dishes in 10 ml of 13% FBS/DMEM.Eighteen hours later, cells were treated with 0.1-50lM of 4NQO for 1 h at 37°C and immediately afterward rinsed and detached by trypsin (0.1%) and EDTA (0.02%).The trypsinization was stopped by adding 1% FBS/DMEM.Cells were harvested, rinsed twice with ice-cold PBS, and stored as pellets containing 3 3 10 6 cells at ÿ80°C until use.All procedures of DNA extraction from cells were performed inside an EAN-140 anaerobic incubator (TABAI ESPEC Corp., Osaka, Japan) in order to prevent formation of oxidative DNA damage during the experimental procedure, as described elsewhere (Nakajima et al., 1996).Briefly, cells (3 3 10 6 ) were treated with RNase (100 lg/ml) (Roche Diagnostics, Mannheim, Germany) followed by the direct addition to the cell pellet of proteinase K (50 units/ml) and lysis buffer (Applied Biosystems, Foster, CA), which had been preset in the anaerobic incubator and left overnight.The mixture was then incubated at 60°C for 30 min in a heat block.Next, DNA was precipitated with 99.5% EtOH and rinsed twice with 80% EtOH.The extracted DNA was heat denatured and sequentially digested to nucleosides with nuclease P 1 (Seikagaku Corp., Tokyo, Japan) and alkaline phosphatase (Sigma, St. Louis, MO).The mixture was then subjected to ultrafiltration in order to remove proteins or polysaccharides by using a Microcon (YM-3, cut-off molecular weight: 3000; Millipore Corporation, Bedford, MA).8OHdG was detected using a HPLC system (Tosoh Corp., Tokyo, Japan) with an electrochemical detector (ECD) according to a previously described method (Takeuchi and Morimoto, 1993).The 8OHdG level was expressed as the number of 8OHdG molecules per 10 5 dG, determined simultaneously with an UV monitor coupled to the HPLC system.
Immunostaining of 8OHdG.For immunostaining of 8OHdG, 3 3 10 5 cells were plated on a coverslip in 6-well plates (Iwaki, Tokyo, Japan) in 13% FBS/DMEM.After 18 h, the cells were rinsed and then treated with 4NQO for 15 min or 1 h with or without 1000 units/ml SOD (Wako Pure Chemical) or 1000 units/ml catalase (Roche Diagnostics).Immunostaining of 8OHdG was carried out as previously described (Yarborough et al., 1996).After 4NQO treatment, the cells were washed twice and fixed with 75% EtOH at ÿ20°C for 10 min followed by treatment with RNase (100 lg/ml) at 37°C for 1 h and with proteinase K (10 lg/ml) at room temperature for 10 min.DNA was denatured with 4 N HCl for 7 min at room temperature, after which pH was adjusted with a 50mM Trizma base.After the cells had been washed, normal horse serum was used for blocking, followed by incubation with primary antibody against 8-oxo-dG, IF7 (Trevigen Inc., Gaithersburg, MD) diluted to 1:30 at 4°C overnight, and then incubated with biotin-conjugated anti-mouse goat IgG.Endogenous peroxidase was blocked by treating the cells with 3% H 2 O 2 (Wako Pure Chemical) in methanol for 30 min at room temperature.After washing with PBS, ABC reagent (Vector Laboratories, Burlingame, CA), avidin-conjugated horseradish peroxidase was treated for 30 min at 37°C followed by color development with diaminobenzidine (Vector Laboratories).The cells were washed with H 2 O and mounted with 50% glycerol in PBS.Digitized color images of each specimen were obtained with a microscope (E6TUW-21-1; Nikon, Tokyo, Japan) equipped with a digital camera (DXM1200; Nikon) and a digital imaging system (ATC-1, version 2.12; Nikon).The images were analyzed in an eight-bit gray-scale mode using ImageJ freeware (version 1.32).The formula used for the quantification of signal intensity of the 8OHdG immunostaining was as follows: 8OHdG index ¼ P ([X ÿ thresholds] 3 area [lm 2 ])/total cell number, where X is the staining density on the gray scale.
Trypan blue dye exclusion assay.Cell viability was determined with a trypan blue dye exclusion assay.When cells were treated with 4NQO for 8OHdG measurement by means of HPLC-ECD, part of each sample was removed and suspended in a dilution medium with trypan blue solution (GIBCO-BRL, Grand Island, NY), and the number of living and dead cells were counted.The viability was determined as the number of viable (unstained) cells/total number of cells and quantified as the percentage of viable cells at each concentration in comparison with that obtained without 4NQO treatment.
Detection of intracellular ROS production by 4NQO.For ROS detection using confocal microscopy, 3 3 10 5 cells were plated on a glass-bottom 35-mm dish (Matsunami Glass, Osaka, Japan) in 2 ml of culture medium at 37°C.After 18 h, the cells were rinsed with DMEM and then treated with 1.0lM 4NQO for 15, 30, 45, or 60 min or with 10lM for 60 min.Sixty-minute treatments were carried out with SOD or catalase as well.For the chase study, cells were treated with 1.0lM 4NQO for 1 h, rinsed twice, and chased with 13% FBS/DMEM for 1, 3, or 6 h.
ROS generation in cells was assessed by using the fluorescent probe 2#,7#dichlorodihydrofluorescin diacetate (DCFH-DA; Molecular Probes Europe BV, Leiden, The Netherlands) or hydroethidine (HEt) (Polysciences, Warrington, PA).These probes produce 2#,7#-dichlorofluorescein (DCF) or 2-hydroxyethidium ROLES OF REACTIVE OXYGEN SPECIES IN 4NQO CARCINOGENESIS (2OHEt) when oxidized by H 2 O 2 in the presence of peroxidase or superoxide, respectively, and these products can be detected by confocal microscopy (LeBel et al., 1992;Zhao et al., 2005).Immediately after 4NQO treatment for various periods, the cells were rinsed twice then labeled for 10 min with 3lM DCFH-DA or 3lM HEt in DMEM in a dark CO 2 incubator at 37°C, washed twice with PBS, and placed in PBS while confocal images were being acquired.
The confocal images were acquired immediately by means of laser excitation using a Zeiss LSM-510 laser scanning confocal microscope (Carl Zeiss, Eching, Germany) with a 320 objective lens (Plane-Neofluar 20 3/0.5 Ph2), equipped with an argon laser (458, 488, 514 nm), which provided excitation at 488 nm and emission at 530-600 nm for DCF and at > 585 nm for 2OHEt.Eight images were saved and digitized as 512 3 512 eight-bit pixels and averaged.
Electron paramagnetic resonance spin-trapping method.A cell-free model of an iron-GSH-4NQO system and a Fenton reaction system (Fe 2þ þ H 2 O 2 ) were used to determine whether 4NQO can induce oxidative stress by using electron paramagnetic resonance (EPR) spectrometry and a spin-trapping agent, 5,5-dimethyl-1-pyroline-N-oxide (DMPO) (Labotec, Tokyo, Japan).Spectra were recorded on a JES-TE100 ESR spectrometer (JEOL, Tokyo, Japan) with a 100-kHz field modulation.The magnetic field strength was calibrated with the hyperfine coupling constant (8.69 mT) of Mn 2þ ions doped in MgO powder.The g values of the observed EPR spectra were estimated with the third signal of Mn 2þ as the standard (g ¼ 2.034).All experiments were performed at room temperature (298 K) using a 9-mm quartz flat cell.In the experiments, samples were aspirated into the quartz flat cell, and ESR data acquisition started approximately 15 s after sample preparation.Data analyses were performed with a WIN-RAD EPR data analyzer (Radical Research, Inc., Tokyo, Japan).The X-band spectrometer settings were 5 mW of microwave power, a time constant of 0.1 s, and modulation amplitude of 0.079 mT.
Intracellular glutathione quantification.For glutathione (GSH) measurements, cells were plated and treated with 1nM-1lM of 4NQO for 1 h at 37°C and then stored as cell pellets (3 3 10 5 ) at ÿ80°C until use.
Intracellular GSH levels were measured with a Total GSH Quantification Kit (Dojindo, Kumamoto, Japan) based on the enzymatic recycling procedure, according to the manufacturer's recommended procedure.Briefly, cells were washed and lysed with 1% sulfosalicylic acid, the lysates were incubated on ice for 10 min, and the supernatants collected after centrifugation.GSH levels were measured by determining the rate of colorimetric change of 5,5#-dithiobis(2nitrobenzoic acid) in the presence of GSH reductase and NADPH.Color development was monitored at 405 nm with a kinetic method using a BIO-RAD Model 3550 microplate reader (Bio-Rad Laboratories Inc., Hercules, CA).GSH levels were calculated by using a standard curve.
Statistical analysis.Results are expressed as means ± SDs.A linear regression analysis was used to evaluate the association between 4NQO concentrations and cellular 8OHdG levels.For statistical comparisons among three cell strains, ANOVA was used.Student's t-test (for equality of means) was used for comparisons among 8OHdG levels, 8OHdG index, and GSH levels.Two-tailed tests were performed, and a difference of p < 0.05 was considered significant.StatView 5.0 software (SAS Institute, Cary, NC) was used for analysis of all the data.

Detection of 8OHdG by HPLC-ECD and Cell Viability after 4NQO Treatment
An experiment was carried out to determine whether 4NQO treatment resulted in 8OHdG formation in fibroblasts obtained from normal individuals.Three human fibroblast strains were treated with different concentrations of 4NQO for 1 h, and 8OHdG levels were measured by HPLC-ECD.A large amount of 8OHdG was detected in cells exposed to 4NQO in a dosedependent manner from 1.0 to 50lM 4NQO (Fig. 1A).The slopes of the lines showed no significant differences among the three cell strains.Under our experimental conditions, 0.95lM 4NQO was required to produce one 8OHdG molecule per 10 5 dG in normal human fibroblasts (simple regression line for overall data of three strains: y ¼ ÿ0.60 þ 0.95 x, r 2 ¼ 0.993, and p < 0.001).The background levels in this experiment were from approximately 2-4 8OHdG molecules per 10 6 dG.However, production of 8OHdG at less than 1.0lM of 4NQO was hardly detectable by HPLC-ECD, probably because 8OHdG/10 5 dG levels for treatment with 1.0lM of 4NQO were too low to be detected with the HPLC-ECD method.A simultaneous cell viability assay using the dye exclusion method at 8OHdG quantification by HPLC-ECD showed about 90% viability at 15.0lM of 4NQO (Fig. 1B).

8OHdG Detection by Immunohistochemistry
Although production of 8OHdG at less than 1.0lM of 4NQO was hardly detectable by HPLC-ECD, previous findings of a colony formation assay, where sensitivity to reproductive inactivation can be evaluated, showed that 0.4lM of 4NQO resulted in only a 10% survival rate for normal fibroblasts (Moriwaki et al., 1993).We therefore employed immunohistochemical staining as a more sensitive method for the detection of 8OHdG production at lower concentrations.
Cultured normal human neonatal fibroblasts were incubated with 0.01, 0.1, and 1.0lM of 4NQO for 1 h followed by immunostaining with an antibody against 8OHdG.The staining intensity of 8OHdG in cells treated with 1.0lM of 4NQO was stronger than that following treatment with 0.1lM (Fig. 2A).The 8OHdG-staining intensities were semiquantified and expressed as 8OHdG indices by calculating the signals with ImageJ 1.32 (Fig. 2B).8OHdG indices increased in a dosedependent manner at lower concentrations of less than 1.0lM and reached a plateau at concentrations above 1.0lM.

Inhibition of 8OHdG Formation by Antioxidants
It is reasonable to speculate that superoxide and H 2 O 2 are involved in the formation of 8OHdG by 4NQO since (1) 8OHdG is induced by agents generating ROS (Kasai and Nishimura, 1991), (2) 4NQO is a powerful inducer of the sox in E. coli (Nunoshiba and Demple, 1993), and (3) superoxide can produce H 2 O 2 in the presence of SOD.We therefore examined the effect of antioxidant on the formation of 8OHdG by 4NQO.
A low concentration of 0.4lM 4NQO yielded almost 100% cell viability (Fig. 1B) as determined with the dye exclusion assay which evaluates membrane integrity.However, the same concentration of 4NQO killed off more than 90% of cells according to the colony formation assay (Moriwaki et al., 1993), which evaluates replication ability.This biologically relevant and still viable dose of less than 1.0lM of 4NQO was used in the following assays to identify the mechanisms of 4NQO genotoxicity.
8OHdG was formed by treating cells with H 2 O 2 in a timeand a dose-dependent manner (Fig. 3A).The cells were then treated with 1.0lM of 4NQO for 15 min in the presence or absence of SOD or catalase in the culture medium, and 8OHdG was detected immunohistochemically.Antioxidants dramatically reduced 8OHdG-staining intensities (Fig. 3B).8OHdG indices were calculated in cells treated with 500lM H 2 O 2 , 0.1lM 4NQO, or 1.0lM 4NQO for 15 min with or without antioxidants.The addition of SOD or catalase significantly reduced the 8OHdG index (Fig. 3C) because the addition of catalase into the culture medium containing 0.1lM or 1.0lM of 4NQO completely inhibited 8OHdG production.It is further speculated that H 2 O 2 must be involved in the formation of 8OHdG by 4NQO treatment.

4NQO Induces Oxidative Stress in Human Fibroblasts
Our finding that the formation of 8OHdG by 1.0lM 4NQO treatment was completely inhibited in the presence of 1000 units/ml of catalase (Fig. 3C) strongly suggests that 4NQO produces 8OHdG at least partly via ROS at this low concentration.We therefore examined whether 4NQO induces ROS in cultured human skin fibroblasts by using DCFH-DA and HEt as fluorescent probes for H 2 O 2 and superoxide, respectively.Fluorescence of DCF was detected as short as a 15-min treatment with 1.0lM of 4NQO, and the signal increased in intensity until 30-45 min treatment (Fig. 4A, upper panel).Since fluorescence of DCF can be detected not only by hydrogen peroxide but also by NO-related radicals (Setsukinai et al., 2003), we tested whether the fluorescence of DCF diminishes as a result of catalase treatment.Addition of 1000 U/ml of catalase to the culture medium markedly reduced the fluorescence of DCF after a 60-min treatment with 4NQO (Fig. 4A), indicating that 4NQO treatment induced H 2 O 2 .Although the amount of ROS was greatly diminished by the 6-h chase, it remained the same during both the 1-and 3-h chase periods (Fig. 4A, lower panel).These observations suggest that ROS

ROLES OF REACTIVE OXYGEN SPECIES IN 4NQO CARCINOGENESIS
production continued for at least 3 h after the removal of 4NQO from the culture medium.
As shown in Figure 4B, the levels of 2OHEt-dependent fluorescence just after treatment with 1lM 4NQO increased depending on the length of incubation time (15, 30, 45, and 60 min).The fluorescence intensity after 60 min of treatment with 10lM 4NQO was stronger than that after the same period of treatment with 1lM 4NQO, indicating that the production by 4NQO of superoxide in cells was dose dependent.4NQOinduced 2OH-Et fluorescence diminished with time, and the fluorescence during the 3-h chase weakened to the same level as the basal level.

A Cell-Free System Detection of Oxidative Stress by 4NQO
4NQO is capable of accepting a single electron to form a radical anion of the general form R-NO Á ÿ 2 ; and the transfer of a single electron to oxygen can produce superoxides (Biaglow et al., 1977).Previous in vitro studies have shown that 4HAQO may generate ROS such as H 2 O 2 (Hozumi, 1969).We therefore carried out an experiment using EPR spectroscopy to determine whether 4NQO itself or only its metabolite from 4HAQO can generate ROS in a cell-free system.
EPR spectroscopy using a radical trap agent, DMPO, showed that induction of oxidative stress by the iron-GSH-4NQO system occurred even in a cell-free system.There were few differences between the EPR spectra obtained from the reaction mixture containing only ferrous ion, GSH, and DMPO under atmospheric conditions (Figs. 5B-5D) and those from control (Fig. 5A).
However, when 4NQO was added to the reaction mixture, prominent radical signals derived from DMPO were observed in a time-dependent manner (Figs.5E-5G).These signals consisted of a doublet of a triplet (six lines indicated with white arrows) with hyperfine splitting constants (hfsc) of a(N) ¼ 1.623 mT and a(H) ¼ 2.251 mT, which is assumed to be derived from DMPO-a-hydroxyethyl adducts.No EPR signals were detected for 4NQO alone (data not shown).Figure 5H shows DMPO-OH signals (four black arrows) derived from hydroxyl radicals under the Fenton reaction system.When EtOH was added to the DMPO-OH system at the same concentrations as for 4NQO, the DMPO-a-hydroxyethyl signal appeared in addition to the DMPO-OH signal (Fig. 5I).This finding is consistent with that of a previous study that hydroxyl radicals react with EtOH to form a-hydroxyethyl radicals (Rosen and Rauckman, 1981).The split pattern of the EPR spectrum with six signals shown in Figure 5I indicates that the signals were DMPO-a-hydroxyethyl adducts.Weak DMPO-OH signals were also detected in the spectrum shown in Figures 5G and 5I (indicated by black arrows).These results strongly suggest that hydroxyl radicals were generated by 4NQO in the cell-free system.

Different Effects of H 2 O 2 and 4NQO on Total GSH Level and 8OHdG Level
ROS can be induced by a very low concentration of 4NQO, such as 0.1lM, which is the equivalent of 500lM of H 2 O 2 in terms of its 8OHdG formation efficacy (Fig. 3C), indicating that 4NQO was much stronger than H 2 O 2 in this respect.It is thus likely that the 8OHdG-producing mechanisms of these two compounds are different.The fibroblasts were incubated with 1.0lM of 4NQO for 0-60 min (a-e) or with 1.0lM of 4NQO for 60 min and chased for 1, 3, and 6 h with fresh medium without 4NQO (g-i).Catalase (1000 U/ml) was added to the culture medium with 1.0lM of 4NQO during the 60-min treatment (f).(B) 4NQO-induced superoxide in fibroblasts.Superoxide-dependent cellular fluorescence images just after treatment with 1lM 4NQO are visualized in an incubation time-dependent manner (15, 30, 45, and 60 min) (k-n), whereas fibroblasts treated without 4NQO but with HEt did not show any fluorescence (j).Fluorescence images after treatment with 10lM 4NQO for 60 min was much higher than those treated with 1lM (o).Chase for 1, 3, and 6 h after treatment with 1lM of 4NQO for 60 min demonstrated that superoxide-dependent fluorescence subsided in 6 h (p-r).
Since GSH is known to remove H 2 O 2 (Biaglow et al., 1977) as well as to combine with 4NQO (Ishikawa et al., 1997), we examined the GSH levels (Fig. 6).Although GSH levels decreased in a dose-dependent manner to a similar extent both in H 2 O 2 -treated cells and 4NQO-treated cells (Figs. 6 A and  6B), the GSH level in 4NQO-treated cells did not return to the basal level during the 24-h chase (Fig. 6B), while that of H 2 O 2treated cells showed a complete recovery to the basal level (Fig. 6A).These results support the notion that 4NQO produces ROS and that the sustained low level of GSH after 4NQO treatment may represent impaired GSH production or enhanced GSH consumption.To assess the differences in reduction of 8OHdG, a chase study was performed using H 2 O 2 and 4NQO (Fig. 6C).Although the cells treated with 500lM H 2 O 2 and 0.1lM 4NQO showed equivalent removal of 8OHdG after the 24-h chase, none of the 8OHdG produced by 1.0lM 4NQO, which falls below the minimum level detectable by HPLC, was reduced to the basal level.

DISCUSSION
Our study results demonstrated that a substantial amount of 8OHdG was formed in the cells treated with 4NQO.We could detect dose-dependent formation of 8OHdG by HPLC-ECD administered in doses between 1 and 50lM in three normal cell strains with similar slopes (Fig. 1A).Background levels in our study fell between approximately 2-4 8OHdG molecules per 10 6 dG, the equivalent of 0.5-1 per 10 6 nucleotides, which are levels similar to those obtained with the same method where DNA was isolated under anaerobic conditions (Takeuchi and FIG. 5. 4NQO induction of oxidative stress in a cell-free system.The effects of 4NQO on the induction of oxidative stress were determined with 90mM DMPO, 500lM FeSO 4 , and 500lM GSH in 10mM sodium phosphate buffer, pH 7.4.EPR spectra of a DMPO spin adduct obtained without 4NQO (A-D) were not different from those of control.However, when 500lM 4NQO was added to the system (E-G), prominent radical signals were observed in a time-dependent manner by 4NQO treatment.These signals consisted of a doublet of a triplet (white arrows) with hyperfine splitting constants (hfsc) of a(N) ¼ 1.623 mT and a(H) ¼ 2.251 mT, which is assumed to be the equivalent of a DMPO-a-hydroxyethyl signal.The effects of EtOH on the Fenton system Morimoto, 1993).On the other hand, the background level obtained from the DNA isolated with a conventional procedure was almost 10 times higher (Hattori et al., 1996).Although a 1h treatment with 1.0lM 4NQO had little effect on viability as observed in a dye exclusion assay (Fig. 1B), we and other investigators have found that treatment with 0.4lM 4NQO for 1 h resulted in a survival of only 10% in a colony formation assay (Mirzayans et al., 1989;Moriwaki et al., 1993).The dye exclusion assay is based on the principle that living cells maintain membrane integrity and therefore appear to be clear, whereas dead cells stain blue, so that the cell count should be performed before cells enter the replication cycle.On the other hand, the purpose of a colony formation assay is to evaluate reproductive capacity with the colony count performed about 10 days after the treatment.A much smaller dose can therefore impair the reproductive (or replicative) ability much more than the membrane integrity.Even if 4NQO causes DNA base damage, replication disability may not be detected within 24 h, while it also takes longer to develop biologically relevant causal effects such as mutation or cell death.Kakunaga (1974) reported that one cell generation is required for transformation to be established and more than four cell generations for expression of the transformed state in 4NQO-induced transformation to become detectable.These findings suggest that concentrations of less than 1.0lM of 4NQO cause considerable cell damage during the replication process and that this concentration should be sufficient and thus significant for genotoxicity where cellular replication is not totally impaired, yet highly mutable.
Since formation of 8OHdG generated by less than 1.0lM of 4NQO was hardly detectable by HPLC-ECD, we employed immunohistochemical staining, which is a more sensitive method, to detect 8OHdG formation at lower concentrations.Semiquantification of the 8OHdG-staining intensities showed that 8OHdG indices increased in a dose-dependent manner at lower concentrations as well (Fig. 2B).Immunostaining, especially with the ABC (avidin-biotin complex) method, has good detection sensitivity.However, if the antigen is in excess of the upper detection limit, all the antibodies may be consumed by the antigen.The signal intensity soon becomes saturated.On the other hand, the HPLC method is characterized by an efficient quantitative analysis at higher amounts of the substance, although the detection sensitivity at lower amounts is not so good since there is some background level of 8OHdG caused during DNA isolation manipulation.A small increase in 8OHdG can therefore not be detected because it is obscured by the background level.However, immunohistochemical staining made it possible to detect 8OHdG that cannot be detected because of artifacts during HPLC measurement.
8OHdG is well known as an oxidatively modified DNA molecule.It can pair with adenine as well as cytosine and causes G:C to T:A transversion during DNA replication (Moriya, 1993); these types of mutations are known to be generated by ROS (Cheng et al., 1992) and have also been observed in the Salmonella tryphimurium strain (Koch et al., 1994) treated with 4NQO or in experimental tumors induced by 4NQO (Ide et al., 2001).4NQO treatment thus induces 8OHdG, which causes mutations resulting in carcinogenesis.Kohda et al. clearly demonstrated that 4HAQO, the ultimate carcinogen derived from 4NQO, forms a quinone-DNA adduct.Ishikawa et al. (1984) found that the treatment of cells with 4NQO induces unscheduled DNA synthesis, indicating that 4NQO induces a DNA adduct, which is probably excised by nucleotide excision repair.The quinone-DNA adduct is believed to be a major source of 8OHdG formed by cells or DNA treated with 4NQO or 4HAQO.However, our finding that the antioxidant reduced the amount of 4NQO-induced 8OHdG implies that the 4NQO-induced 8OHdG formation may at least partly have been mediated by generation of ROS.Catalase completely suppressed 4NQO-induced 8OHdG formation, but SOD inhibited it only partially (Fig. 3).These results can be explained by the fact that, in the presence of SOD, superoxide produced by 4NQO generates H 2 O 2 , which continues generating hydroxyl radicals leading to 8OHdG production, even if it partially subsides (Fig. 7).The kinetics of intracellular ROS, where superoxide rapidly disappeared in 1-3 h, while H 2 O 2 remained at least until after the 6-h chase, underscore the validity of the proposed pathways (Fig. 7).Another explanation could be that hydrogen peroxide can more easily permeate the cell membrane than superoxide since the former permeates the cell membrane via the density gradient (Ohno and Gallin, 1985), while superoxide does so only via the ion channel (Lynch and Fridovich, 1978).Because hydrogen peroxide can freely diffuse through the cellular membrane, the intracellular hydrogen peroxide generated as a result is moved outside, where catalase can scavenge it.Since superoxide also diffuses via the anion channel, a similar situation may occur.
Our observation that antioxidants inhibit the formation of 4NQO-induced 8OHdG contradicts the finding by Kohda et al. (1987) that the formation of 8OHdG residue in DNA treated in vitro with 4HAQO in the presence of seryl-AMP was not affected by the addition of antioxidants except for 10% EtOH.However, the concentration of 4HAQO they used was 5 3 10 ÿ4 M, which is 500 times higher than that used in our study.The discrepancy is therefore likely to be due to the difference in the concentration of 4NQO or to the quantitative ratio of 4NQO to antioxidant.It is further possible that 8OHdG formation is mediated not only by the quinone-DNA adduct through 4HAQO but also directly through 4NQO and mediated by ROS.
Previous in vitro studies have demonstrated that 4HAQO is capable of generating ROS such as superoxide and H 2 O 2 (Hozumi, 1969).However, the concentration of 4HAQO required for a 10-fold induction of soxS is 100-200lM, while that of 4NQO is 5lM or less (Nunoshiba and Demple, 1993).Moreover, 4NQO in Ehrlich ascites cells incubated in the presence of glucose stimulated oxygen consumption, but 4HAQO had no effect on cellular respiration (Biaglow et al., 1977).Our study provided direct evidence of the generation of ROLES OF REACTIVE OXYGEN SPECIES IN 4NQO CARCINOGENESIS ROS in skin fibroblasts treated with 4NQO because this treatment enhanced DCF generation, while the DCF-dependent fluorescence was completely suppressed by catalase (Fig. 4A).This indicates that H 2 O 2 is produced by the 4NQO treatment, which also enhanced 2OHEt-dependent fluorescence, a specific indication of superoxide production (Zhao et al., 2005).As far as we know, ours is the first study to directly demonstrate that 4NQO induces ROS formation in the cells.
We also used EPR spectroscopy using a radical trap agent, DMPO, to determine whether 4NQO in a cell-free system undergoes redox cycling and generates ROS.We added GSH and iron to this assay system based on the metabolic reduction mechanism of 4NQO proposed by Biaglow et al. (1977).We used GSH as a representative electron reductant because it is universally present in cells and the similar condition in vivo can be obtained.Since DMPO is unstable, four lines derived from DMPO-OH, the autooxidation product of DMPO, appear while the reaction is in progress.Since ROS generated by 4NQO had to be detected before the appearance of the autooxidized DMPO-OH, we used a high concentration of 4NQO to accelerate the reaction and make the 4NQO-derived signal distinguishable from the autooxidation signal of DMPO.When 4NQO was added to an iron-GSH-4NQO system, prominent radical signals derived from DMPO-a-hydroxyethyl adducts, which are presumably developed by the reaction between DMPO-OH signals and EtOH, appeared in a time-dependent manner (Figs.5E-5G) in addition to DMPO-OH signals (Fig. 5G).These results suggest that Fenton-like reactions involving an interaction between free radicals, ROS and transition metal ions, were accelerated by 4NQO and that hydroxyl radicals were indeed generated in the cell-free system.
By treating cells with 1lM of 4NQO, H 2 O 2 and superoxide were produced (Figs.4A and 4B), with H 2 O 2 remaining within the cells at least after the 3-h chase in our study (Fig. 4A).However, a previous report indicated the half-life of hydrogen peroxide as approximately 10 min when Epstein Barr virus (EBV) transformed lymphocytes were treated with 20mM hydrogen peroxide (Takeuchi and Morimoto, 1993).Under those experimental conditions, catalase activity was reported to be 14.73 U/mg protein, where one enzyme unit is equivalent to 1 mmol of product formation or 1 mmol of substrate disappearance/min.On the other hand, catalase activity of human fibroblasts was reported to be 9.4 U/mg, lower than of EBV-transformed lymphocytes (Shindo and Hashimoto, 1998).Thus, the half-life of hydrogen peroxide under our experimental conditions should be less than 10 min.4NQO may reduce the activity of the ROS-removal system, including GSH, catalase, and SOD, or it may induce rapid repeats of the redox cycle (Fig. 7), which may contribute to an increase in the quantity and duration of H 2 O 2 production.The depletion of GSH by 1.0lM of 4NQO was similar to that obtained with 500lM of H 2 O 2 (Fig. 6).GSH is likely to be employed in the detoxification of electrophilic compounds by forming a conjugated form of GSH with the compound.GSH-4NQO conjugate in cells (Stanley and Benson, 1988) is then exported outside the cells (Morrow et al., 2000), probably through the plasma membrane by means of an multidrug resistance-associated protein (MRP)/glutathione S-conjugate (GS-X) pump (Gomi et al., 1997).It is also likely to be compartmentalized into intracellular vesicles, being released by exocytosis.In addition to the usual consumption of GSH via oxidation by ROS, 4NQO-GSH conjugation may be involved in the prolonged low level of GSH after 4NQO treatment.Furthermore, 4NQO-GSH conjugates may suppress the synthesis of intracellular GSH, since the formation of these conjugates results in the generation of nitrite (NO ÿ 2 ) (Stanley and Benson, 1988) which inactivates c-glutamylcysteine synthase (Han et al., 1995).All these events may contribute to the large amount of 8OHdG formed by 4NQO.However, the sustained decrease of GSH during the 24-h chase cannot be explained by these theories.Since ROS produced by 4NQO had already disappeared after the 6-h chase (Fig. 4B), the continued reduction of GSH could not have been due to a cyclical production of ROS.If 4NQO blocks GSH at the transcriptional level, however, it could cause the prolonged decrease of the intracellular GSH level even after the 24-h chase.
To conclude, we were able to identify and clarify the complex mechanisms of DNA damage induced by a powerful carcinogen, 4NQO.A combination of ROS formation and GSH depletion is robustly involved in 8OHdG formation by 4NQO.Many carcinogens including environmental carcinogens share common pathways to elicit its genotoxic activity.Further detailed studies of these mechanisms can thus be expected to contribute to the prevention of cancer development by environmental carcinogens and to provide clues for the development of countermeasures against environmental or iatrogenic carcinogenesis.

FIG. 1 .
FIG. 1. Dose-dependent 8OHdG formation and cytotoxity of 4NQO.(A) Normal human skin fibroblast strains from three different individuals (N-1-N-3) were treated with various concentrations of 4NQO for 1 h.The amount of 8OHdG per 10 5 dG was determined with HPLC-ECD.(B) Cell viability in terms of dye exclusion assay showed inverse correlation with concentration of (4NQO lM).Each set of data represents the mean of three strains.The results are shown as means ± SDs.All values were obtained from three determinants in two or three independent experiments; *p < 0.05, ***p < 0.001, Student's t-test.

FIG. 2 .
FIG. 2. Immunohistochemical detection of 4NQO-induced 8OHdG.(A) Human fibroblasts were treated for 1 h with 0.01, 0.1, and 1lM or without 4NQO.Presence of 8OHdG within the nucleus was demonstrated by immunostaining with anti-8OHdG antibody.(B) Five randomly selected fields were photographed, and the staining intensities were analyzed by ImageJ 1.32 as described in ''Immunostaining of 8OHdG''.All values were derived from two or three independent experiments.The results are shown as means ± SDs.

FIG. 3 .
FIG. 3. Antioxidants inhibited 8OHdG formation.(A) The 8OHdG formation of fibroblasts treated with H 2 O 2 was detected by immunostaining and are shown as 8OHdG indices.(B) The presence of 8OHdG was immunohistochemically detected in cells incubated with 1.0lM 4NQO (a), 1.0lM 4NQO with 1000 units/ml of catalase (b), 1lM 4NQO with 1000 units/ ml of SOD (c), and without 4NQO (d) for 15 min.(C) The 8OHdG indices of the cells treated with H 2 O 2 or 4NQO for 15 min in the presence or absence of antioxidants, 1000 units/ml of SOD, or 1000 units/ml of catalase.All values were derived from three independent experiments.The results are shown as means ± SDs; **p < 0.01, Student's t-test.
FIG. 5. 4NQO induction of oxidative stress in a cell-free system.The effects of 4NQO on the induction of oxidative stress were determined with 90mM DMPO, 500lM FeSO 4 , and 500lM GSH in 10mM sodium phosphate buffer, pH 7.4.EPR spectra of a DMPO spin adduct obtained without 4NQO (A-D) were not different from those of control.However, when 500lM 4NQO was added to the system (E-G), prominent radical signals were observed in a time-dependent manner by 4NQO treatment.These signals consisted of a doublet of a triplet (white arrows) with hyperfine splitting constants (hfsc) of a(N) ¼ 1.623 mT and a(H) ¼ 2.251 mT, which is assumed to be the equivalent of a DMPO-a-hydroxyethyl signal.The effects of EtOH on the Fenton system (Fe 2þ þ H 2 O 2 ) were measured with 50mM DMPO, 10lM FeSO 4 , 100lM H 2 O 2 , and 10lM diethylenetriamine-N,N,N#,N$,N$-pentaacetic acid (DTPA) (DOJINDO, Kumamoto, Japan) with (I) or without (H) 1% (vol/vol) EtOH.DMPO-OH signals (black arrows) were observed in samples with EtOH.In the control experiment, PBS was used in place of EtOH.

FIG. 7 .
FIG. 7. Schematic model of 8OHdG production by 4NQO.4NQO induces 8OHdG through ROS as well as through quinone-DNA adduct.The prolonged reduction in GSH levels after 4NQO treatment (Fig.6) may be also involved in the formation of the markedly large quantity of 8OHdG.