Benzene Metabolite 1,2,4-Benzenetriol Induces Halogenated DNA and Tyrosines Representing Halogenative Stress in the HL-60 Human Myeloid Cell Line

Background: Although benzene is known to be myelotoxic and to cause myeloid leukemia in humans, the mechanism has not been elucidated. Objectives: We focused on 1,2,4-benzenetriol (BT), a benzene metabolite that generates reactive oxygen species (ROS) by autoxidation, to investigate the toxicity of benzene leading to leukemogenesis. Methods: After exposing HL-60 human myeloid cells to BT, we investigated the cellular effects, including apoptosis, ROS generation, DNA damage, and protein damage. We also investigated how the cellular effects of BT were modified by hydrogen peroxide (H2O2) scavenger catalase, hypochlorous acid (HOCl) scavenger methionine, and 4-aminobenzoic acid hydrazide (ABAH), a myeloperoxidase (MPO)-specific inhibitor. Results: BT increased the levels of apoptosis and ROS, including superoxide (O2•−), H2O2, HOCl, and the hydroxyl radical (•OH). Catalase, ABAH, and methionine each inhibited the increased apoptosis caused by BT, and catalase and ABAH inhibited increases in HOCl and •OH. Although BT exposure increased halogenated DNA, this increase was inhibited by catalase, methionine, and ABAH. BT exposure also increased the amount of halogenated tyrosines; however, it did not increase 8-oxo-deoxyguanosine. Conclusions: We suggest that BT increases H2O2 intracellularly; this H2O2 is metabolized to HOCl by MPO, and this HOCl results in possibly cytotoxic binding of chlorine to DNA. Because myeloid cells copiously express MPO and because halogenated DNA may induce both genetic and epigenetic changes that contribute to carcinogenesis, halogenative stress may account for benzene-induced bone marrow disorders and myeloid leukemia.

volume 120 | number 1 | January 2012 • Environmental Health Perspectives Research Benzene, widely used in the chemical industry, is a common environmental contaminant found in gasoline, cigarette smoke, and coal tar. In humans, chronic exposure to benzene results in progressive deterioration in hemato poiesis, possibly leading to myelo dysplastic syndrome and acute myeloid leukemia (Aksoy 1989;Huff 2007).
Although the mechanisms of benzene tox icity remain unclear, it is considered to occur only after metabolic activation (Snyder and Hedli 1996;Whysner et al. 2004). In the liver, benzene is primarily metabolized by cytochrome P450 2E1 (CYP2E1) to benzene oxide, which is then converted by epoxide hydrolase to dihydrodiol. Subsequent process ing by dihydrodiol dehydrogenase yields cat echol (CT). Alternatively, by non enzymatic rearrangement, benzene oxide is converted to phenol, which can be oxidized by CYP2E1 to 1,4hydroquinone (HQ) and 1,4benzo quinone (Snyder and Hedli 1996). The path way for formation of 1,2,4benzenetriol (BT) in humans is not yet clearly understood; it has been suggested that BT may be formed by the hydroxylation of either HQ or CT (Henderson et al. 1989;Inoue et al. 1989).
Various metabolites of benzene are con sidered to bring out toxicity through the generation of reactive oxygen species (ROS), inhibition of topo isomerase, and subsequent induction of DNA damage (Whysner et al. 2004). Among benzene metabolites, the tri phenolic metabo lite BT reacts most actively with molecu lar oxygen (Lewis et al. 1988;Zhang et al. 1996). We also know that BT induces oxidative DNA damage and breaks DNA strands (Kawanishi et al. 1989;Kolachana et al. 1993;Lewis et al. 1988). Moreover, BT damages DNA more severely than does HQ, and benzene and CT per se have no detect able effects on DNA (Kawanishi et al. 1989). Because epidemiological studies of HQ and 1,4benzoquinone have never demon strated carcino genicity in humans, the International Agency for Research on Cancer (IARC) assigned their carcinogenic risk to humans as group 3: not classifiable as to carcinogenicity to humans (IARC 2011). Meanwhile, IARC has not evaluated the carcinogenicity of BT.
The heme enzyme myeloperoxidase (MPO), which is synthesized and secreted by neutrophils, monocytes, and other myel oid cells, is an important source of oxidants.
MPO catalyzes the formation of hypo chlorous acid (HOCl), a powerful oxidant derived from chloride ions and hydrogen peroxide (H 2 O 2 ). HOCl is a potent cytotoxin that plays key roles in host defense by oxidizing the cellular constituents of invading pathogens (Hurst and Barrette 1989). At the same time, HOCl is also capable of damaging proteins, lipids, and nucleic acids in host tissue (Heller et al. 2000). By damaging the DNA of host cells, MPOinduced DNA halogenation might contribute to the association between chronic inflammation and cancer (Marnett 2000).
Although benzene is known to be specif ically toxic to bone marrow in humans, the mechanism for this is not understood (Whysner et al. 2004). MPO has a much higher endog enous presence in bone marrow than in any other internal organ (Heller et al. 2000), but no previous study has examined the role of MPO derived HOCl in benzene toxicity.
We investigated the effect of MPOderived HOCl on the toxicity of BT in the HL60 human myeloid cell line. To examine DNA damage induced by BT, we used an immuno cytometric method to evaluate halogenated DNA, and we determined 8oxodeoxy guanosine (8oxodG) levels using highper formance liquid chromatography (HPLC) coupled with electrochemical detection (ECD). We found that BT generates HOCl via the H 2 O 2 -MPO-halide system; rather than gener ating 8oxodG, this HOCl halogenates DNA.
Determination of apoptosis by flow cytometry. We used annexin V-fluorescein isothiocyanate (FITC) and propidium iodide (PI) doublelabeling kits (TACS Annexin VFITC Kit; Trevigen, Gaithersburg, MD, USA) to detect phosphatidyl serine as a marker of apoptosis. HL60 cells suspended in RPMI 1640/10% FCS at 4 × 10 5 /mL were exposed to BT (25-100 µM) at 37°C in 5% CO 2 for 8 hr. For experiments with catalase (H 2 O 2 scavenger), cells were exposed to BT plus 250 U/mL catalase. For experi ments with ABAH (MPO inhibitor) and methio nine (HOCl scavenger), HL60 cells were pre incubated with RPMI 1640/10% FCS containing 100 µM ABAH or 25 mM methio nine for 24 hr; media was then replaced with new media containing the reagent plus BT. Unexposed HL60 cells were used as controls. After incubation, cells were harvested and washed and then stained with annexin V-FITC and PI according to the manufacturer's instructions. We evaluated the cells using a FACScan flow cytometer (Becton Dickinson, Mountain View, CA, USA), and data were analyzed using WinMDI software (version 2.9; Biology Software Net, La Habra, CA, USA). Determination of intra cellular ROS genera tion by flow cytometry. To detect intra cellular superoxide (O 2 •− ) and H 2 O 2 , we followed the method of Takeuchi et al. (1996). Briefly, HL60 cells suspended in phenolred-free RPMI 1640 at 4 × 10 5 /mL were incubated with hydroethidine (HE; Molecular Probes, Carlsbad, CA, USA) or dichlorofluorescin diacetate (DCFHDA; Molecular Probes). The probeloaded cells were then exposed to BT with or without 250 U/mL catalase for 30 min at 37°C. For experiments with ABAH or methionine, the cells were pre treated as described above and then suspended in phenolred-free RPMI 1640 at 4 × 10 5 /mL. After addition of the same concentration of ABAH or methionine, loaded with probes, the cells were exposed to BT for 30 min at 37°C. Non fluorescent HE is oxidized to fluorescent 2hydroxyethidium by O 2 •− , whereas DCFH is oxidized to dichloro fluorescein (DCF) by H 2 O 2 and peroxidases (Rothe and Valet 1990). Presence of 2hydroxy ethidium or DCF was measured by FACScan.
For selective detection of HOCl and the hydroxyl radical ( • OH), HL60 cells suspended in phenolredfree-RPMI 1640 at 4 × 10 5 / mL were incubated with 10 µM amino phenyl fluo rescein (APF; Sekisui Medical, Tokyo, Japan) or 10 µM hydroxyphenyl fluorescein (HPF; Sekisui Medical) and then exposed to BT for 30 min at 37°C with or without 250 U/mL catalase. For experi ments with ABAH or methio nine, cells were pre treated and exposed as described above. APF and HPF themselves are not highly fluorescent, but when reacted with HOCl (APF) or • OH (HPF) they exhibit strong dosedependent fluorescence, which can be used to differenti ate HOCl and • OH from H 2 O 2 , nitric oxide, and O 2 •− (Setsukinai et al. 2003). The specific ity and usefulness of these probes have been described previously (Kohanski et al. 2007;Nakazato et al. 2007). We meas ured the fluo rescence intensity of cells by FACScan and analyzed data using WinMDI software.
Determination of halogenated DNA by immuno cytometric analysis. To detect DNA damage by HOCl, we analyzed halo genated DNA using a novel monoclonal anti body (mAb2D3) that recog nizes the HOClmodified 2´deoxycytidine residue 5chloro2´deoxycytidine (5CldC; supplied by Y. Kawai, Nagoya University, Nagoya, Japan) (Kawai et al. 2004(Kawai et al. , 2008. HL60 cells were suspended in RPMI 1640/10% FCS at 1 × 10 6 /mL and then exposed to 50 µM BT with or without catalase. For experi ments with ABAH or methionine, cells were pre treated as described above and then exposed to BT for 1 hr or 4 hr at 37°C in 5% CO 2 . HL60 cells were exposed to 20 µM HQ or 1 mM NaOCl for 1 hr or 4 hr at 37°C in 5% CO 2 . After exposure, the cells were washed with phosphatebuffered saline (PBS) and then fixed in 4% paraformaldehyde (Wako Pure Chemical) at 4°C for 20 hr. We evalu ated halogenated DNA as described elsewhere (Kawai et al. 2004(Kawai et al. , 2008, with minor mod ifications. Briefly, the fixed cells were per meabilized by a 3min exposure, on ice, to PBS containing 0.3% Triton X100. The cells were then blocked with 2% bovine serum albumin (SigmaAldrich) in PBS contain ing 0.05% Tween 20 (TPBS). The cells were then incubated with mAb2D3 in TPBS for 1 hr at room temperature. After washing with TPBS, the cells were incubated in TPBS for 1 hr at room temperature with FITClabeled antimouse IgG (Dako, Kyoto, Japan). After incubation, cells were washed with TPBS and their fluo rescence intensity was meas ured by FACScan. The data were analyzed as described above.
Determination of 8oxodG by HPLC ECD. To detect oxidative DNA damage by • OH, we evaluated 8oxodG by HPLCECD. Cells were suspended in RPMI 1640/10% FCS at 1 × 10 6 /mL and exposed to 50 µM BT for 1, 2, or 4 hr; 20 µM HQ for 1 or 4 hr; or 20 µM CT for 2 hr, with all exposures at 37°C in 5% CO 2 . The cells were immediately chilled in an icewater bath, washed with ice cold PBS, and then stored for later analysis as cell pellets at -80°C. DNA was extracted from the cells with DNA Extractor WB Kit (Wako Pure Chemical) according to the manufac turer's instructions and enzymatically digested to nucleo sides, as described by Takeuchi et al. (1994). After HPLC separation, 8oxodG was detected by ECD, and deoxy guanosine (dG) was detected by ultraviolet absorption as described elsewhere (Takeuchi et al. 1994). 8oxodG level was expressed as the molar ratio of 8oxodG per 10 5 dG.
Immunocytochemical detection of halogenated tyrosines. To detect protein damage by HOCl, we analyzed halogenated tyrosines using rabbit antichlorotyrosine antibody (Hycult Biotech, Uden, the Netherlands) (Gujral et al. 2003) and mouse antidibromotyrosine monoclonal anti body (JaiCA, Shizuoka, Japan), which crossreacts with dichloro tyrosine (Kato et al. 2005). Cells were suspended in RPMI 1640/10% FCS at 4 × 10 5 /mL and then exposed to 50 µM BT at 37°C in 5% CO 2 for 4 hr. After exposure, the cells were washed with PBS and centrifuged with Shandon Cytospin 4 (Thermo Scientific, Kanagawa, Japan) at 1,000 rpm for 8 min. Centrifuged cells on slides were dried and fixed with cold acetone and then blocked with PBS containing 2% bovine serum albumin. To detect chloro tyrosine, cells were incubated with antichlorotyrosine antibody and then stained with Alexa Fluor 488-conjugated goat antirabbit antibody (Invitrogen, Tokyo, Japan) and 1 µg/mL PI. To detect dibromo/ dichlorotyrosine, cells were incubated with antidibromotyrosine antibody and then stained with Alexa Fluor 488-conjugated goat antimouse antibody and 1 µg/mL PI. The stained slides were examined by fluorescence microscopy.
Statistical analysis. Data are presented as mean + SE. Statistical analyses were performed using PASW Statistics software (version 18.0; SPSS, Inc., Tokyo, Japan). Treatment effects were established by non parametric Wilcoxon tests. Data for DNA damage were analyzed using analysis of variance, followed by Fisher's protected least significant difference test for post hoc comparisons of individual treat ments. pValues < 0.05 (two tailed) were considered significant.

Levels of apoptosis and intra cellular ROS after BT exposure.
We found more annexin V-positive and PInegative cells, considered to be apoptotic, in HL60 cells that had been exposed to 50 µM BT for 8 hr volume 120 | number 1 | January 2012 • Environmental Health Perspectives ( Figure 1B) than in controls ( Figure 1A). The percentage of apoptotic cells in HL60 cells exposed to 50 µM BT was significantly greater than in unexposed cells ( Figure 1C, inset). Apoptosis increased depending on the concentration of BT ( Figure 1C).
We determined intracellular ROS flow cytometrically using ROSsensitive fluores cent probes: HE for O 2 •− , DCFHDA for H 2 O 2 , APF for HOCl, and HPF for • OH. BT increased the intra cellular levels of each of these ROS ( Figure 1D).

Effect of ROS scavengers and MPO inhibitor on apoptosis and intra cellular ROS.
Apoptosis was inhibited by catalase, ABAH, and methio nine (Figure 2A). Catalase also inhibited the generation of O 2 •− , H 2 O 2 , HOCl, and • OH induced by BT expo sure ( Figure 2B). Although ABAH inhib ited the BTinduced increase of HOCl and • OH, it further increased the generation of O 2 •− ( Figure 2C). We could not determine the effect of ABAH on H 2 O 2 because peroxi dases, which are required for the conversion of DCFH to DCF, were inhibited by ABAH (Matsugo et al. 2006).
Levels of halogenated DNA. Using flow cytometry after immunostaining, we meas ured the level of halogenated DNA in HL60 cells exposed to 50 µM BT. HL60 cells exposed to BT for 1 hr showed about the same levels of halogenated DNA as control (unexposed) cells. However, after 4 hr expo sure to BT, increased levels of halogenated DNA were apparent ( Figure 3A,B). Although catalase, methionine, and ABAH inhibited these increases (Figure 3C), the levels of halo genated DNA were still higher than those in control cells. HL60 cells exposed to 1 mM NaOCl for 1 hr and for 4 hr had significantly more halogenated DNA ( Figure 3B). In con trast, exposure to HQ did not increase the level of halogenated DNA ( Figure 3B).
Levels of 8oxodG. Using HPLCECD, we measured 8oxodG levels in HL60 cells exposed to 50 µM BT, 20 µM HQ, or 20 µM CT. HL60 cells exposed to 20 µM CT for 2 hr had significantly more 8oxodG. However, BT and HQ had about the same 8oxodG levels as the control cells ( Figure 4).
Detection of halogenated tyrosines. To confirm the induction of halogenative stress in HL60 cells by BT, we detected HOCl induced protein damage in the form of halogenated tyrosines. After 4 hr exposure to 50 µM BT, levels of both chloro tyrosine ( Figure 5A,B) and dibromo/dichloro tyrosine ( Figure 5C,D) were elevated.

Discussion
Because the findings of in vivo and in vitro research so strongly implicate the involvement of ROS in benzeneinduced toxicity (Snyder and Hedli 1996), we designed a study to inves tigate the carcinogenic mechanism of benzene, focusing on BT, a benzene metabo lite that generates ROS by autoxidation (Kawanishi et al. 1989;Zhang et al. 1996). We specif ically examined the cyto toxic effects of BT on a human myeloid cell line, a class of cells from the organ mainly affected by benzene. Representative dot graphs of unexposed HL-60 cells (control; A) and cells exposed to 50 μM BT for 8 hr (B). Cells in the lower right quadrant, which were stained with annexin V but not PI, were considered to be apoptotic; percentages of these cells in are shown in the figure. (C) Percentages of apoptotic HL-60 cells after exposure to BT for 8 hr; data presented are mean + SE from two independent experiments conducted in duplicate. Inset, percentages of apoptotic cells in controls or HL-60 cells exposed to 50 μM BT for 8 hr (mean + SE of 15 independent experi ments conducted in duplicate). (D) Fluorescence intensities, corresponding to levels of various ROS, in controls or cells exposed to 50 μM BT for 30 min (mean + SE from 5-7 independent experiments conducted in duplicate). Fluorescence intensity is shown in arbitrary units. **p < 0.01, and # p < 0.001, compared with control.

PI (no. of cells) PI (no. of cells)
4.44% , APF (for HOCl), and HPF (for • OH) in cells exposed to BT or BT plus catalase for 30 min (mean + SE from three to five independent experiments conducted in duplicate). (C) Effects of ABAH on ROS generated by BT, as shown by fluorescence intensity of HE, APF, and HPF in cells exposed for 30 min to BT or BT plus ABAH (mean + SE from three to five independent experiments conducted in duplicate). *p < 0.05, and **p < 0.01 compared with the corresponding cells exposed to BT alone.   (Tomono et al. 2009), suggests that BT generates HOCl. Moreover, BT exposure increased the amount of halo genated DNA and halogenated tyrosines detected by immuno logical examinations, which confirms generation of HOCl by BT. To the best of our knowledge, no previous studies have evaluated the generation of HOCl by benzene and its metabolites.
To investigate the mechanism of MPO mediated apoptosis in HL60 cells, we coin cubated HL60 cells with BT and catalase or pre treated cells with ABAH or methionine, and then exposed them to BT. These reagents drastically suppressed the level of BTinduced apoptosis. This strongly implicates the H 2 O 2 -MPO-HOCl system in the induction of apop tosis by BT. Catalase inhibited BTinduced generation of ROS, and ABAH specifically inhibited BTinduced generation of HOCl and • OH. This inhibition indicates that HOCl was generated via the H 2 O 2 -MPO-HOCl system. The inhibition of HOCl genera tion by both catalase and ABAH demon strates that, in HL60 cells exposed to BT, H 2 O 2 was cer tainly metabolized to HOCl by MPO. It also suggests that HOCl might trigger BTinduced apoptosis of HL60 cells. In contrast, ABAH further increased the BTinduced generation of O 2 •− , which indicates an accumulation of O 2 •− caused by the inhibition of MPO and also indicates that O 2 •− and probably H 2 O 2 do not directly trigger apoptosis.
We then investigated whether this cyto toxicity of BT was related to the induction of DNA damage. We evaluated the halogenation of DNA by HOCl, and 8oxodG induction by • OH. Although CT exposure increased 8oxodG as previously reported (Oikawa et al. 2001), BT exposure did not. In HL60 cells exposed to BT, however, we did detect more halogenated DNA. Furthermore, cata lase, ABAH, and methionine clearly inhibited DNA halogenation. These findings indicate that HOCl was generated by MPO after BT exposure, and that this HOCl was the likely culprit in DNA halogenation. We also tested whether HQ induces DNA damage. However, 20 µM HQ did not increase either Figure 3. Induction of halogenated DNA in HL-60 cells by BT. (A) Histograms showing results for control HL-60 cells (gray shaded area) and cells exposed for 4 hr to 50 μM BT (black line) or 1 mM NaOCl (gray dotted line). FL1-H, height of green fluorescence. (B) Fluorescence intensities (arbitrary units) of controls or cells exposed to 50 μM BT, 20 μM HQ, or 1 mM NaOCl for 1 or 4 hr; data presented are mean + SE from 4 independent experiments conducted in duplicate. (C) Fluorescence intensities of controls or cells exposed for 4 hr to 50 μM BT, 50 μM BT with catalase (BT+Cat), 50 μM BT with ABAH (BT+ABAH), or 50 μM BT with methionine (BT+Met) (mean + SE of 3 independent experiments conducted in duplicate). *p < 0.05 compared with the corresponding control. **p < 0.01, and # p < 0.001 compared with control. ## p < 0.05, † p < 0.01 compared with the corresponding HL-60 cells that were exposed to BT alone. volume 120 | number 1 | January 2012 • Environmental Health Perspectives the halogenated DNA or 8oxodG. These results may help explain why HQ does not induce leukemia in humans (Levitt 2007). HQ is known to autoxidize more slowly and simply than does BT (Kawanishi et al. 1989), possibly accounting for the difference between BT and HQ; further study is required to confirm this. Among halogenated nucleosides resulting from reaction with HOCl, 5CldC, 5chloro uracil, 8chlorodeoxy adenine, and 8chloro deoxy guanine have been identified (Whiteman et al. 1997), with 5CldC being the predomi nant carbonchlorinated nucleo side product (Henderson et al. 1999). The mAb2D3 mono clonal antibody that we used in this experi ment recognizes mainly 5CldC (Kawai et al. 2004(Kawai et al. , 2008. Exposing DNA to HOCl causes large increases in pyrimidine oxidation, with no evidence of purine oxidation (i.e., 8oxodG) (Whiteman et al. 1997). This effect is con sistent with our finding of halogen damaged DNA, such as 5CldC, and no evidence of increased 8oxodG. In contrast, exposing DNA to HOCl has been reported to increase the 8oxodG level (Ohnishi et al. 2002); however, that finding may have been due to the addition of diethylene triamine penta acetic acid to the reaction mixture. In another study, Kolachana et al. (1993) reported the induc tion of 8oxodG in HL60 cells by BT. The discrepancy between our results and those of that study may be explained by the difference in the precision of HPLCECD methods. In unexposed cells, we detected levels of 8oxodG about 0.2 per 10 5 dG. By contrast, assuming that the average molecular weight of nucleo tides is 300, the 8oxodG level in the other report was about 9.6 per 10 5 dG.
Oxidative DNA damage has been impli cated in carcino genesis (Weitzman and Gordon 1990). Although normal cells are able to efficiently repair the products of most pro mutagenic HOClmediated damage, no repair activity has been identified for 5chloro cytosine, probably because 5chlorocytosine mimics 5methyl cytosine (Lao et al. 2009). The presence of 5chloro cytosine, which can be mis recognized by cellular machinery as 5methyl cytosine, would alter methylation patterns. In addition, 5chloro cytosine is eas ily transformed to 5chloro uracil (Theruvathu et al. 2009). Because 5chloro uracil residues can pair with adenine as well as guanine (Kim et al. 2010), 5chloro uracil-adenine base pairing might induce genetic mutation. Consequently, halogenated DNA is potentially able to induce both epigenetic and genetic changes that con tribute to carcinogenesis. Inoue et al. (1989) detected BT in human urine after benzene exposure; the urinary concentration of BT linearly correlated with the degree of benzene exposure, reaching > 50 mg/L (396 µM) in a worker exposed to 210 ppm benzene. In addition, Aksoy (1989) reported on Turkish workers who were chronically exposed to up to 650 ppm ben zene, some of whom developed leukemia. We believe that it is plausible that 50 µM BT, the concentration used in the present study, could have been present in the workers.
Individuals with MPO polymorphism -463G→A in the promoter region, which reduces MPO expression, have decreased risk for various cancers (Cascorbi et al. 2000). Lan et al. (2004) reported that benzeneexposed workers with the -463G genotype showed greater hemato toxicity than did workers with the -463A genotype. These findings suggest important roles in myelo toxicity or carcino genesis for MPOcatalyzed reactions toward HOCl.
In the present study we have constructed a novel hypothesis ( Figure 6) that exposure to BT increases O 2 •− generation, possibly by autoxidation. The O 2 •− is chemically or enzy matically converted to H 2 O 2 , which is then metabolized to HOCl by MPO; this HOCl halogenates DNA and proteins, thus inducing myelo toxicity or leukemo genesis. The high expression of MPO from myeloid cells, along with the fact that halogenated DNA can cause gene mutation and epigenetic changes, may explain how benzene is involved in bone mar row dis orders or myeloid leukemia. A previ ous study of benzene toxicity reported that MPO plays a role in the bio activa tion of ben zene's phenolic metabolites (Eastmond et al. 2005). Here, we show for the first time that a benzene metabolite, BT, is capable of generat ing HOCl and consequent halogenative dam age via the H 2 O 2 -MPO-HOCl system. Our findings lend strong support to the hypothesis that BTinduced DNA halogenation is a pri mary reaction in leukemo genesis associated with benzene.