2-methoxyaniline (o-anisidine) is a potent carcinogen, causing tumors of the urinary bladder in both genders of F344 rats and B6C3F1 mice.1, 2 The International Agency for Research on Cancer has classified o-anisidine as a group 2B carcinogen,2 which is possibly carcinogenic to humans. Besides its carcinogenicity, it exhibits other toxic effects, including hematologic changes, anemia and nephrotoxicity.1, 2o-anisidine is used as an intermediate in the manufacturing of a number of azo and naphthol pigments and dyes, which are used for printing (90%) and for paper (3%) and textile (7%) dyeing.1, 3 Such a wide use of this aromatic amine could result in occupational exposure. Furthermore, it may be released from textiles and leather goods colored with these azo dyes and a wider part of the population may be exposed. This carcinogen is also a constituent of cigarette smoke.2, 4 This strongly suggests that o-anisidine ranks not only among occupational pollutants produced in the manufacturing of chemicals, but also among environmental pollutants; it can be assumed that human exposure is widespread. Indeed, o-anisidine was found in human urine in the general population in concentrations of 0.22 μg/l (median).5 In addition, hemoglobin adducts of o-anisidine were detected in blood samples of persons living in urban or rural areas of Germany.6, 7, 8 The adducts as well as o-anisidine in urine might originate not only from the sources mentioned above, but also from a possible o-anisidine precursor, o-nitroanisol (2-methoxynitrobenzene). This chemical was released into the environment in the course of an accident in a German chemical plant, causing subsequently local and regional contamination.6, 9, 10
It has not been determined if o-anisidine is a genotoxic carcinogen. In spite of potent rodent carcinogenicity, it is weakly mutagenic.11o-anisidine is mutagenic in Salmonella thyphimurium,12 an effect that has been associated with both peroxidative activation and the involvement of N-acetyltransferases.13, 14, 15, 16 The chemical induces chromosomal aberrations in Chinese hamster ovary cells,17 gene mutations in mouse lymphoma cells18 and intrachromosomal recombinations in Saccharomyces cerevisiae.19 A statistically significant DNA damage in urinary bladder of CD-1 mice exposed to o-anisidine determined by the single-cell gel electrophoresis (Comet) assay was detected.20 Moreover, Ashby et al.14, 21 demonstrated that a weak but significant increase in the frequency of mutations was induced in urinary bladder in transgenic lacI (Big Blue) mice treated with this carcinogen. However, the chemical is negative in other in vivo genotoxicity assays, including the mouse micronucleus test and the DNA single-strand break assay in rat liver, spleen, kidney and bladder.13, 22, 23
Recently, we found that o-anisidine is oxidatively activated by peroxidases in vitro to species binding to DNA,24, 25 which suggests a genotoxic mechanism of o-anisidine carcinogenicity. Nevertheless, knowledge on the in vivo DNA adduct formation by o-anisidine is necessary to confirm a genotoxic mechanism. Ashby et al.14 found that a single oral administration of o-anisidine (750 mg/kg) to B6C3F1 mice yielded no DNA adducts detectable by the 32P-postlabeling assay of bladder and liver DNA. The same authors found no DNA binding of 14C-ring-labeled o-anisidine in mice.14 The authors suggested that o-anisidine was mutagenic via the formation of radical species.
Theoretically, o-anisidine might be oxidized like other aromatic amines26 to a reactive N-hydroxyarylamine intermediate. Cytochrome P450 (CYP) enzymes are the most efficient enzymes participating in the oxidation of aromatic amines.26N-hydroxyarylamine intermediates of carcinogenic aromatic amines can further be metabolized by phase 2 enzymes, such as N,O-acetyltrasferases (NATs) or sulfotransferases (SULTs), leading to the formation of reactive esters, e.g., N-acetoxy- or N-sulfooxyarylamines, which undergo heterolysis of the N-O or S-O bond to produce electrophilic nitrenium ions capable of reacting with DNA to form DNA adducts.27, 28 However, the metabolism of o-anisidine by CYP enzymes has not yet been studied in detail. Preliminary experiments showed that this carcinogen might undergo O-demethylation29 that is CYP-dependent and yields formaldehyde.30 Other products formed by microsomal oxidation have not yet been characterized.
The present study was undertaken to resolve 3 principal problems: whether CYP-mediated N-hydroxylation of o-anisidine is involved in a metabolic activation of this carcinogen; to determine the capability of humans to activate o-anisidine; and to evaluate the potential of this carcinogen to form DNA adducts in vivo. Therefore, we here investigated whether human CYP enzymes (and which of them) oxidize o-anisidine to species capable of binding to DNA. In addition, we examined the formation of DNA adducts by o-anisidine and their tissue distribution in rats treated with this carcinogen.
Chemicals were obtained from the following sources: α-naphthoflavone (α-NF), NADP+, NADPH, ketoconazole, diethyldithiocarbamate (DDTC), sulfaphenazole, 3-[(3-cholamidopropyl)dimethylammonio]-1-propane sulfonate (CHAPS), dilauroyl phosphatidylcholine, dioleyl phosphatidylcholine, phosphatidylserine, glucose 6-phosphate, nuclease P1, deoxyguanosine 3′-monophosphate (dGp), acetyl-coenzyme A (acetylCoA), cytochrome c, phosphoadenosylphosphosulfate (PAPS) and chlorzoxazone from Sigma Chemical (St. Louis, MO); 7-pentoxy-,7-ethoxyresorufin, o-nitroanisole, o-anisidine, o-aminophenol (> 99% based on HPLC) from Fluka Chemie (Buchs, Switzerland); testosterone, 6β-hydroxytestosterone from Merck (Darmstadt, Germany); glucose 6-phosphate dehydrogenase from Serva (Heidelberg, Germany); bufuralol and its 1′-hydroxyderivative from Gentest (Woburn, MA); calf thymus DNA (CT-DNA) and glutathione from Roche Diagnostics (Mannheim, Germany); and bicinchoninic acid from Pierce (Rockford, IL). All chemicals were of analytical purity or better. 3-isopropenyl-3-methyldiamantane (3-IPMDIA) was synthesized according to Olah et al.31N-(2-methoxyphenyl)hydroxylamine was synthesized by the procedure similar to that described earlier.32 Briefly, to a solution of 2 g ammonium chloride and 90 mmol o-nitroanisole in 60% ethanol/water, 180 mmol zinc powder was added in small portions. After addition of the first portion at room temperature, the reaction starts; this can be monitored by the rising temperature in the flask. The reaction mixture was kept at 10–15°C using a cooling bath (ice/sodium chloride mixture) and slowly adding additional doses of zinc powder. After 1 hr, excess zinc was removed by filtration and ethanol was removed under reduced pressure. The product was extracted into 100 ml ethyl acetate and crystallized by adding hexane. The yield was 60%. N-(2-methoxyphenyl)hydroxylamine authenticity was confirmed by electrospray mass and CID spectra and high field proton NMR spectroscopy. The positive-ion electrospray mass spectrum exhibited the protonated molecule at m/z 140.1, while the CID of its ion fragments at m/z 125.2, 108.1 and 109.1. The 1H-NMR spectra were recorded at 400 MHz in dimethyl sulfoxide-d6. The central line of dimethyl sulfoxide at 2,500 ppm was used as reference line. The spectra showed the presence of the following protons: 8.28 (1H, d, J = 2.3 Hz, exchanged with CD3OD), 7.64 (1H, d, J = 1.5 Hz, exchanged with CD3OD), 7.01 (1H, m, Σ J = 9.6 Hz), 6.84 (2H, m, Σ J = 15.0 Hz), 6.75 (1H, m, Σ J = 16.9 Hz) and 3.75 (3H, s).
Ortho-[ring-U-14C]anisidine hydrochloride, with a purity of 99% as estimated by HPLC (which was used for repurification of the compound), and a specific activity of 134 mCi/mmol (4.96 GBq/mmol), was kindly provided by Dr. John Ashby and Dr. Paul A. Lefevre (Syngenta Central Toxicology Laboratory, Alderley Park, Cheshire, U.K.). Enzymes and chemicals for the 32P-postlabeling assay were obtained commercially from sources described previously.33, 34
Six male Wistar rats (125–150 g) were treated once a day for 5 consecutive days with o-anisidine dissolved in sunflower oil (0.15 mg/kg body weight i.p. per day). Two control animals received an equal volume of solvent only. Rats were placed in cages in temperature- and humidity-controlled rooms. Standardized diet and water were provided ad libitum. Animals were killed 24 hr after the last treatment by cervical dislocation.35 Seven organs (liver, kidney, lung, heart, spleen, brain and urinary bladder) were removed immediately after death, quickly frozen in liquid nitrogen and stored at −80°C until DNA isolation.
Human hepatic microsomal and cytosolic samples, Supersomes and assays
Nine human hepatic microsomal and 2 cytosolic samples were from Gentest (numbers of samples are shown in Tables I and III) and stored at −80°C. The donors ranged in age from 2 to 71 years and included 2 men and 7 women. A drug and/or alcohol abuse history of the donors is described in Gentest protocols. Each microsomal preparation was reanalyzed for specific CYP and NADPH:CYP reductase activities and 2 cytosolic preparations for SULT and NAT activities by assays described in the protocols of Gentest. Our data were similar to those reported by Gentest in their specification sheets.
Table I. o-Anisidine Metabolism in Human Hepatic Microsomes
Table III. Binding OF [14C]-Labeled o-Anisidine to DNA
DNA binding of activated [14C]o-anisidine (pmol/mg DNA)
Triplicate incubations were carried out in total volumes of 0.75 ml 50 mM sodium phosphate (pH 7.4) containing 1.0 mM [14C]o-anisidine dissolved in methanol (20 μl/0.75 ml incubation), 1 mM NADPH, 1 mg human hepatic microsomal protein, 1 mg human hepatic cytosolic protein, 1 mg CT-DNA at 37°C for 120 min.
Incubations contained additionally 2 mM acetylCoA or 0.2 mM PAPS.
Incubations were the same as in (1), but without NADPH.
Incubations were the same as in (1), but without the human hepatic micrisomal and cytosolic samples. Mixures were extracted with ethyl acetate, DNA isolated and [14C] radioactivity bound to DNA measured as described in Material and Methods. The numbers are averages ± SEM of 3 parallel experiments.
Supersomes, microsomes isolated from insect cells transfected with Baculovirus constructs containing cDNA of either one of the following human CYPs: CYP1A1, 1A2, 1B1, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1, 3A4 and 2E1 and 3A4 with cytochrome b5 and expressing NADPH:CYP reductase were from Gentest. Protein concentrations in the microsomal fractions were assessed using the bicinchoninic acid protein assay with bovine serum albumin as a standard.36 The concentration of CYP was estimated according to Omura and Sato37 based on the absorption of the complex of reduced CYP with carbon monoxide.
Isolation of individual CYPs
CYP1A2, 2B4, 2C3 and 2E1 enzymes were isolated from liver microsomes of rabbits induced with β-naphthoflavone (CYP1A2), phenobarbital (CYP2B4), or ethanol (CYP2E1, 2C3) by procedures described by Haugen and Coon38 and Yang et al.39 Rabbit CYP3A6 was isolated from hepatic microsomes of rabbits induced with rifampicin.40 The procedure was analogous to that used for isolation of CYP2B4. Rat CYP2B2 was isolated from liver microsomes of rats pretreated with phenobarbital (PB) by the procedure as described.41 Recombinant rat CYP1A1 protein was purified to homogeneity by the procedure described previously42 from membranes of Escherichia coli transfected with a modified CYP1A1 cDNA in the laboratory of H.W. Strobel (University of Texas, Medical School of Houston, Houston, TX) by P. Hodek (Charles University, Prague, Czech Republic). Rabbit liver NADPH:CYP reductase was purified as described.43 Rabbit liver cytochrome b5 was prepared as described elsewhere.44
Unless stated otherwise, incubation mixtures used for study of the o-anisidine metabolism contained the following in a final volume of 100 μl: 100 mM sodium phosphate buffer (pH 7.4), 1 mM NADP+, 10 mM D-glucose 6-phosphate, 1 U/ml D-glucose 6-phosphate dehydrogenase, 10 mM MgCl2 and a human hepatic microsomal fraction containing 0.04–1.0 nmol CYP, as well as 0.1–0.5 mM o-anisidine dissolved in 1.0 μl methanol. The reaction was initiated by adding the substrate. Incubation mixtures, in which the efficiencies of Supersomes expressing human CYPs were tested, were the same except that only 10–50 pmol of CYP was used. Incubations containing purified CYP reconstituted with NADPH:CYP reductase and cytochrome b5 contained 50–200 pmol of each CYP. Briefly, CYP was reconstituted as follows: 0.5 μM CYP, 0.5 μM NADPH:CYP reductase, 0.5 μM cytochrome b5, 0.5 μg/μl CHAPS, 2.0 μg/μl liposomes [dilauroyl phosphatidylcholine, dioleyl phosphatidylcholine, phosphatidylserine (1:1:1)], 3 mM reduced glutathione and 50 mM HEPES/KOH, pH 7.4.45, 46 An aliquot of CYP of the reconstitution mixture was then added to incubation mixtures. After incubation in open glass tubes (37°C, 60 min), the incubation mixtures were terminated by addition of 100 μl of methanol and centrifuged at 5,000g for 5 min. The supernatants were collected and 20 μl aliquots applied onto high-performance liquid chromatography (HPLC) column, where metabolites of o-anisidine were separated. Another aliquot of the reconstitution mixture was used to estimate CYP activities with typical substrates: 7-ethoxyresorufin O-deethylation (CYP1A1/2), coumarin 7-hydroxylation (CYP2A6), 7-pentoxyresorufin O-depentylation (CYP2B6),47 tolbutamide methyl hydroxylation (CYP2C), bufuralol 1′-hydroxylation (CYP2D6), chlorzoxazone 6-hydroxylation (CYP2E1) and testosterone 6β-hydroxylation (CYP3A).48 In the control incubation, the CYP was omitted from the reconstitution mixture.
The HPLC was performed with a Bishoff HPLC pump with a LDC/Milton spectrophotometric detector set at 254 nm, and peaks were integrated with a Waters QA 1 integrator. The column used was a Nucleosil 100-5 C18 (Macherey-Nagel, Duren, Germany; 25 cm × 4.6 mm, 5 μm) proceeded by a C-18 guard column. Chromatography was under isocratic conditions of 20% methanol in 50 mM sodium phosphate, pH 8.0, with a flow rate of 0.7 ml/min. Two o-anisidine metabolites with retention times (r.t.) of 6.4 and 10.0 min (peaks 1 and 2 in Fig. 1) were separated by HPLC. To characterize the o-anisidine metabolite eluting at 10.0 min (peak 2), fractions containing this metabolite were collected from multiple HPLC runs, concentrated on a speed-vac evaporator, and analyzed by mass spectrometry. The metabolite was also identified by comparison of its chromatographic properties on HPLC with those of an authentic synthetic standard.
Positive-ion ESI mass spectra were recorded on a Finnigan LCQ-DECA quadrupole ion trap mass spectrometer (FinniganMAT, San Jose, CA). Metabolites (final concentration 1 pmol/μl) dissolved in methanol/water (1:1, v/v) were continuously infused through a capillary held at 1.8 kV into the dynamic Finnigan nanoelectrospray ion source via a linear syringe pump (Harvard Apparatus Model 22) at a rate of 1 μl/min. The ionizer and ion transfer optics parameters of the ion trap were as follows: spray voltage 1800 V, capillary temperature 150°C, capillary voltage 14 V, tube lens offset −22 V, octapole 1 offset −7.4 V, lens voltage −16 V, octapole 2 offset −11.3 V, octapole r.f. amplitude 450 V peak-to-peak (pp) and entrance lens voltage −66.9 V. Helium was introduced at a pressure of 0.1 Pa to improve the trapping efficiency of the sample ions. The spectra were scanned in the range m/z 50–800 and the gating time was set to accumulate and trap 1 × 107 ions. The mass isolation window for precursor ion selection was set to 2 amu and centered on the 12C isotope of the pertinent ion. The background helium gas served as the collision gas for the collision-induced dissociation (CID) experiment. The relative activation amplitude was 35% and the activation time was 30 msec. No broadband excitations were applied.
The following inhibitors were used with o-anisidine in human hepatic microsomes: α-NF, which inhibits CYP1A1 and 1A2; 3-IPMDIA, which inhibits CYP2B641, 49; DDTC, which inhibits CYP2A6 and 2E1; sulfaphenazole, which inhibits CYP2C; ketoconazole, which inhibits CYP3A4. Inhibitors were dissolved in 1.0 μl of methanol to yield final concentrations of 1–1,000 μM in the incubation mixture. The complete mixtures without o-anisidine were then incubated at 37°C for 10 min; o-anisidine was added and incubation continued for a further 60 min at 37°C. An equal volume of methanol alone was added to the control incubations.
The incubation mixtures contained in a final volume of 0.75 ml: 50 mM sodium phosphate (pH 7.4), 0.5 mM [14C]o-anisidine dissolved in methanol (7.5 μl/0.75 ml incubation), 1 mM NADPH, 1 mg human hepatic microsomal protein, 1 mg human hepatic cytosolic protein and 1 mg CT-DNA (4 mM dNp). The reaction was initiated by adding [14C]o-anisidine. Control incubations were carried out either without activating system, without DNA, without NADPH, or without o-anisidine. After incubation (37°C, 60–120 min), all reaction mixtures were extracted twice with ethyl acetate (2 × 2 ml). DNA was isolated from the residual water phase by the phenol/chloroform extraction method as described earlier.24, 34, 50, 51 The 14C radioactivity of modified DNA was determined by liquid scintillation counting (Packard Tri-Carb 200 CA). The content of DNA was determined spectrophotometrically.52
The standard procedure,53 the nuclease P1 enrichment version54 and the 1-butanol extraction-mediated enrichment procedure55 of the 32P-postlabeling assay were performed as described earlier.25, 53, 54, 55, 56 Labeled DNA digests were separated by 2 chromatographic methods on polyethylenimine (PEI)-cellulose plates. In the first, essentially as described previously,54 except that D3 solvent was 3.5 M lithium formate, 8.5 M urea (pH 3.5); D4 solvent was 0.8 M lithium chloride, 0.5 M Tris, 8.5 M urea (pH 8.0), followed by a final wash with 1.7 M sodium phosphate (pH 6.0). D2 was omitted (method A). In the second, 32P-labeled adducts were also resolved by the modification described by Reddy et al.51 This procedure has been shown to be suitable for resolution of DNA adducts formed by N-(2-methoxyphenyl)hydroxylamine or by o-nitroanisole, reductively activated in vitro and in vivo.56 The solvents used in this case were D1, 2.3 M sodium phosphate (pH 5.77); D2 was omitted; D3, 2.7 M lithium formate, 5.1 M urea (pH 3.5); D4, 0.36 M sodium phosphate, 0.23 M Tris-HCl, 3.8 M urea (pH 8.0). After D4 development and brief water wash, the sheets were developed (along D4) in 1.7. M sodium phosphate (pH 6.0; D5), to the top of the plate, followed by an additional 30- to 40-min development with the TLC tank partially opened to allow the radioactive impurities to concentrate in a band close to the top edge (method B).56 Adduct levels were calculated in units of relative adduct labeling (RAL), which is the ratio of c.p.m. of adducted nucleotides to c.p.m. of total nucleotides in the assay.
Preparation of reference compounds and 32P-postlabeling analysis of adducts
An aliquot of 0.5 μmol dGp was incubated in 50 mM Tris-HCl buffer, pH 5.0, 7.0 and 7.4, with 20 μmol N-(2-methoxyphenyl)hydroxylamine without further activation at 37°C overnight in a total volume of 0.5 ml.56 After incubation and extraction with ethyl acetate, 20 μl aliquots were removed from the aqueous phase and directly used for 32P-postlabeling analysis; the standard procedure53 and the nuclease P1 version of the assay was utilized.54 Control incubations were carried out either without N-(2-methoxyphenyl)hydroxylamine or without dGp. Resolution of the adducts on a PEI-cellulose TLC plate was carried out by method B.
32P-postlabeling of adducts in DNA of organs of rats treated with o-anisidine
Whole rat organs were minced and DNA was isolated by the phenol/chloroform extraction method. 32P-postlabeling analysis was performed as described above. Resolution of DNA adducts on PEI-cellulose TLC plate was carried out by method B.
Cochromatography on PEI-cellulose
Adduct spots detected by the 32P-postlabeling assay in incubations with DNA and dGp in vitro and in DNA from treated rats showing similar properties on TLC were excised from the thin-layer plates and extracted as described.33, 34, 56 Cutouts were extracted with 2 800 μl portions of 6 M ammonium hydroxide/isopropanol (1:1) for a total of 40 min. The eluent was evaporated in a Speed-Vac centrifuge. For cochromatographic analyses, the extracts were dissolved in water so that equal amounts of radioactivity could be applied for each sample. Development of these adducts was carried out in D3 and D4 directions using 2 different solvents systems: (a) D3, 2.7 M lithium formate, 5.1 M urea (pH 3.5), and D4, 0.36 M sodium phosphate, 0.23 M Tris-HCl, 3.8 M urea (pH 8.0); (b) D3, 2.7 M lithium formate, 5.1 M urea (pH 3.5), and D4, 4 M ammonium hydroxide/isopropanol (1:1).
HPLC analysis of 32P-labeled 3′,5′-deoxyguanosine bisphosphate adducts
HPLC analysis was performed essentially as described previously.56, 57 Individual spots detected by the 32P-postlabeling assay were excised from thin-layer plates and extracted.58 The dried extracts were dissolved in 100 μl of methanol/phosphate buffer (pH 3.5) 1:1 (v/v). Aliquots (50 μl) were analyzed on a phenyl-modified reversed-phase column (250 × 4.6 mm, 5 μm Zorbax Phenyl; Säulentechnik Dr. Knauer, Berlin, Germany) with a linear gradient of methanol (from 40% to 80% in 45 min) in aqueous 0.5 M sodium phosphate and 0.5 M phosphoric acid (pH 3.5) at a flow rate of 0.9 ml/min. Radioactivity eluting from the column was measured by monitoring Cerenkov radiation with a Berthold LB 506 C-1 flow-through radioactivity monitor (500 μl cell; dwell time, 6 sec).
Statistical association between CYP-linked catalytic activities in human hepatic microsomal samples and levels of individual o-anisidine metabolites formed by the same microsomes were determined by the linear regression correlation coefficients using version 6.12 Statistical Analysis System software. Correlation coefficients were based on a sample size of 9. All p-values are 2-tailed and considered significant at the 0.05 level.
Metabolism of o-anisidine by human hepatic microsomal samples
When o-anisidine was incubated with human hepatic microsomes in the presence of NADPH, 2 product peaks with r.t. of 6.4 and 10.0 min were observed by HPLC analysis (peaks 1 and 2 in Fig. 1a). On the basis of mass spectroscopy and cochromatography with the synthetic standard, the structure of one of the o-anisidine metabolites (peak 2 in Fig. 1a) was identified. In the positive-ion electrospray mass spectrum, the metabolite showed the protonated molecule at m/z 140.1 (Fig. 2), indicating the molecular mass of a hydroxylated derivative of methoxyaniline. The CID of this ion afforded a fragment at m/z 125.2 showing the mass difference equaled to 15, representing a methyl group. Other fragments at m/z 108.1 and 109.1 show the molecular masses of protonated methoxybenzene and N-phenylhydroxylamine, respectively. Collectively, these results indicate that the analyzed compound is a N-(2-methoxyphenyl)hydroxylamine metabolite. Indeed, the analyzed metabolite is identical with authentic N-(2-methoxyphenyl)hydroxylamine (by chromatography). Characterization of the metabolite eluting at 6.4 min (peak 1 in Fig. 1a) remains to be performed. However, the cochromatography of this metabolite with an o-aminophenol standard revealed that it is not O-demethylated o-anisidine, the o-aminophenol standard eluted at 7.4 min, nor was it one of the N-oxidation products 2-nitrosoanisole (2-methoxynitrosobenzene) or 2-nitroanisole (2-methoxynitrobenzene). Oxidation of o-anisidine with human microsomal systems to both metabolites was time-dependent and linear until 90 min (data not shown).
Involvement of CYP enzymes in o-anisidine oxidation in human hepatic microsomes
To identify the human CYPs capable of oxidizing o-anisidine, microsomal samples from livers of 9 human donors were used. Catalytic activities known to be associated with specific CYP enzymes (CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4) are described in Gentest protocols and show large individual variations. All human microsomal preparations oxidized o-anisidine (Table I). Nine human hepatic microsomal samples used in this study yielded qualitatively similar metabolic profiles; however, interindividual variations in levels of each metabolite were evident as well as variations in the ratios of both metabolites (Table I). Correlations between the individual CYP catalytic activities and the rates of formation of each of the o-anisidine metabolites in the same set of human hepatic microsomes were used to examine the role of specific human CYP enzymes in the metabolism of o-anisidine. Highly significant correlations were found between rates of chlorzoxazone 6-hydroxylase, a marker for CYP2E1, and the levels of N-(2-methoxyphenyl)hydroxylamine as well as a metabolite M1 (peak 1 in Fig. 1; Table II). No significant correlation was found between any other examined CYP activities (CYP1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4) and levels of o-anisidine metabolites.
Table II. Correlation Coefficients Between CYP-Linked Catalytic Activities and Levels of o-Anisidine Metabolites Formed by Human Hepatic Microsomes
Inhibition experiments supported the role of CYP2E1 in oxidizing o-anisidine in human hepatic microsomes. The human hepatic microsome sample with the highest efficiency to oxidize o-anisidine (sample HK34) was selected, and incubations were carried out in the absence and presence of specific inhibitors of CYP enzymes, α-NF for CYP1A1/2, 3-IPMDIA for CYP2B6, sulfaphenazole for CYP2C, DDTC for CYP2E1 and 2A6 and ketoconazole for CYP 3A4. Incubations with DDTC resulted in a 95% decrease in the levels of N-(2-methoxyphenyl)hydroxylamine, while inhibitors of other CYP enzymes were ineffective. Formation of metabolite M1 (peak 1 in Fig. 1) was inhibited by DDTC and ketoconazole, by 74% and 48%, respectively.
Oxidation of o-anisidine by purified CYP enzymes
To characterize further the role of individual CYPs in oxidation of o-anisidine, several CYP enzymes were purified, reconstituted with NADPH:CYP reductase and cytochrome b546 and used as the oxidation system. All reconstituted CYPs were active with their typical substrates (data not shown). Among the CYP enzymes tested, rabbit CYP2E1 was the most efficient enzyme oxidizing o-anisidine, while other CYP enzymes (rat CYP1A1, 2B2 and rabbit CYP1A2, 2B4 and 3A6) were less active (Fig. 3).
Oxidation of o-anisidine by recombinant human CYP enzymes
To investigate whether human recombinant CYPs oxidize o-anisidine, we utilized microsomes of baculovirus-transfected insect cells (Supersomes) containing recombinantly expressed human CYPs and NADPH:CYP reductase. All human CYPs used in the experiments efficiently oxidized their typical substrates (results not shown). Human recombinant CYP1A2, followed by CYP2B6 and 2E1, were the most efficient to metabolize o-anisidine. Among other CYPs tested in this study, CYP1A1, 2A6, 2D6 and 3A4 also oxidized o-anisidine, but to a lesser extent (Fig. 4).
o-anisidine is activated by human hepatic microsomes to form DNA adducts
Using [14C]-labeled o-anisidine and the 32P-postlabeling technique, we tested the efficiency of human hepatic microsomes to mediate the activation of o-anisidine to species binding to DNA. The microsomal sample HK34, which oxidized o-anisidine highly efficiently, was selected and incubated with DNA, [14C]o-anisidine and NADPH either without or in the presence of 2 human hepatic cytosolic samples with high activities of phase 2 enzymes, SULT and NATs, samples H856 and H806, respectively (for data see the Gentest protocol). The human hepatic subcellular fractions were effective in activating [14C]o-anisidine to metabolites binding to DNA (Table III). The binding was negligible when microsomes or NADPH were omitted from the incubation mixture. Addition of PAPS, the cofactor of SULTs, into the incubations caused a statistically significant increase of 50% in [14C]o-anisidine binding. Addition of acetylCoA, the cofactor of NATs, had no effect (Table III).
Since the DNA in these experiments was isolated from the reaction mixture by phenol/chloroform extraction, covalent binding of the active form(s) of o-anisidine to DNA should have occurred.34, 50, 52 In order to corroborate the covalent nature of [14C]-labeled o-anisidine-DNA binding, the 32P-postlabeling analysis was applied to the same DNA as another independent method in the present work.
The standard procedure and 2 different enhancing versions of the 32P-postlabeling assay (the nuclease P1 version and enrichment of DNA adducts by 1-butanol extraction) were initially used to analyze o-anisidine-DNA adducts. In addition, 2 variations of multidirectional chromatographic systems were utilized to separate 32P-labeled adducts by TLC: method A is essentially the chromatographic systems originally described by Reddy and Randerath54 and resolves bulky adducts, while method B is a modification developed by Reddy et al.51 for smaller, more polar, adducts.50, 51 This method was found to be suitable to detect adducts in DNA formed by N-(2-methoxyphenyl)hydroxylamine,56 the reductive metabolite of o-nitroanisole.59 We could show that o-anisidine-DNA adducts detectable by 32P-postlabeling are formed after activation of o-anisidine with human hepatic microsomes. The nuclease P1 version of the assay was suitable for detection of o-anisidine-derived DNA adducts (Fig. 5a), while no adducts were visible after 1-butanol extraction and by using the standard procedure of the 32P-postlabeling assay. One major and one minor adduct spots were detected when TLC plates were developed using method B (spots 1 and 2 in Fig. 5). Although the exact nature of these adducts has not been elucidated yet, the chromatographic properties of the adducts on TLC indicate that o-anisidine metabolites containing only one benzene ring derived from N-(2-methoxyphenyl)hydroxylamine are covalently linked to DNA,50, 51, 56 because no adducts were detectable using method A, which resolves more hydrophobic bulky adducts,33, 34, 54, 55 or by using the butanol extraction procedure, which also extracts predominantly hydrophobic bulky adducts.54, 55 Control incubations performed without NADPH or without microsomes were free of adduct spots (not shown). The quantitative analyses of adducts revealed that the amount of adducts detectable by 32P-postlabeling was 2 orders of magnitude lower than that detected by methods utilizing [14C]o-anisidine (Tables III and IV).
Table IV. Quantitative Analysis of DNA Adducts Formed by o-Anisidine Activated with Human Hepatic Microsomes In Vitro
DNA adduct content determined by 32P-postlabeling1
RAL (adducts/107 nucleotides)
The numbers represent means ± SEM (n = 3) of triplicate in vitro incubations (2 postlabeling analyses of each sample). RAL, relative adduct labeling (the number of adducts per normal nucleotides in modified DNA). The total adduct content is the sum of the RAL of individual adducts.
During the enzymatic oxidation of o-anisidine with human microsomes, N-(2-methoxyphenyl)hydroxylamine was formed. Such a reactive compound is assumed to undergo decomposition to produce the nitrenium ion and might therefore be the proximate o-anisidine species responsible for modification of DNA. To confirm this hypothesis, N-(2-methoxyphenyl)hydroxylamine has been synthesized and reacted with dGp, initially at physiologic pH 7.4.56 As shown in Figure 5(b), the nuclease P1 version of 32P-postlabeling analysis resulted in detection of 2 dGp adduct spots formed in this reaction, migrating similarly to those generated in DNA modified by o-anisidine activated with human hepatic microsomes (Fig. 5a). To determine whether the adducts are identical to the adducts formed between dGp and N-(2-methoxyphenyl)hydroxylamine (dGp standard), the spots were extracted and analyzed by cochromatography on PEI-cellulose plates in direction D3 and D4 using 2 different solvent systems. These experiments showed that the 32P-labeled o-anisidine adducts were stable under the alkaline extraction conditions used and that both major adducts in DNA formed by o-anisidine were chromatographically indistinguishable from those of the dGp standard (data not shown).
Reactions of N-hydroxylamines with DNA are acid-catalyzed.60 We therefore investigated whether under acidic conditions more N-(2-methoxyphenyl)hydroxylamine-derived dGp adducts are detectable by 32P-postlabeling and compared adduct levels of reactions at pH 5 and pH 7. The major adduct resulting from N-hydroxyarylamines, including those from N-hydroxyanilines,61 are the C8 adducts of dGp. These adducts are frequently dephosphorylated by nuclease P1 and therefore will not be detected by the nuclease P1 version of the 32P-postlabeling assay. Using the standard procedure of the 32P-postlabeling analysis with the TLC method B, an additional adduct (spot 3 in Fig. 5d), accounting for more than 90% of the total RAL level and exhibiting sensitivity to dephosphorylation by nuclease P1 (Table V), was detected. One order of magnitude higher yield of dGp adducts at pH 5 than at pH 7 was obtained with this procedure (Table V).
Table V. Quantitative Analysis of Adducts Formed by N-(2-Methoxyphenyl) Hydroxylamine with Deoxyguanosine 3′-Monophosphate at pH 5.0 AND 7.0 In Vitro
dGp adduct content determined by 32P-postlabeling1 RAL (adducts/105 nucleotides)
The numbers represent means ± SEM (n = 3) of triplicate in vitro incubations (2 postlabeling analyses of each sample). RAL, relative adduct labeling (the number of adducts per unmodified dGp). The total adduct content is the sum of the RAL of individual adducts.
DNA adduct formation by o-anisidine in rats detected by 32P-postlabeling
To see if o-anisidine forms DNA adducts in vivo, DNA was isolated from several organs of Wistar rats treated with a total dose of 0.75 mg o-anisidine/kg body weight and analyzed by the 32P-postlabeling assay. Using the nuclease P1 version of the 32P-postlabeling assay, o-anisidine-specific DNA adduct patterns, similar to those found in vitro, were detected in the target organ for the o-anisidine carcinogenic effect, the urinary bladder (Fig. 5c) and also in liver, kidney and spleen. No DNA adducts were detected in lung, heart and brain. Likewise, DNA samples of control (solvent-treated) rats were free of these adduct spots even after prolonged exposure times (not shown). Two minor adducts, which were located close to the bottom of the plate, were detected in DNA of most rat organs, also in rats that had not been treated with o-anisidine. No adducts were detectable in DNA of analyzed tissues by using the standard and butanol extraction procedures of the assay.
DNA adducts in individual organs were quantified by measurement of the adduct count rates and expressed as RAL (Table VI). The highest levels of o-anisidine-derived DNA adducts were in urinary bladder, followed by liver, kidney and spleen (Table VI). Total adduct levels were in a range from 0.08 to 4.1 adducts per 107 nucleotides.
Table VI. Quantitative Analysis of DNA Adducts in Various Organs of Rats Treated with A 0.75 mg/kg Total Dose of o-Anisidine
The 2 DNA adduct spots, generated from o-anisidine in vivo, migrated again similarly to those formed in DNA or dGp modified by N-(2-methoxyphenyl)hydroxylamine in vitro. Cochromatography of these adducts on PEI-cellulose plates confirmed their identities (data not shown).
As a second independent chromatographic procedure to characterize the adduct spots, we employed reversed-phase HPLC analyses. We compared the chromatographic properties of adduct spots 1 and 2 formed by o-anisidine in DNA of the urinary bladder in vivo with those of adducts formed by o-nitroanisole in DNA of the urinary bladder of rats treated with this carcinogen. These adducts had been identified in vitro as adducts of dGp with N-(2-methoxyphenyl)hydroxylamine.56 Individual spots were isolated from the PEI plates and subjected to HPLC on a phenyl-modified column eluted with a gradient of methanol in phosphate buffer in high molarity and low pH. The major adduct in DNA generated by o-anisidine in the urinary bladder (spot 1 in Fig. 5c) eluted with an r.t. of 4.4 min, corresponding to the r.t. of 4.3 min of adduct spot 1 generated by o-nitroanisole in the urinary bladder of rats treated with this carcinogen (Fig. 6a and c). When equal amounts of radioactivity of both adduct spots were mixed, a single peak was found (Fig. 6b). As shown in Figure 6(d) and (e), spot 2 in DNA formed by o-anisidine in the bladder (Fig. 5c) produced one peak of radioactivity (r.t. of 5.2 min), corresponding to adduct spot 2 formed in DNA of the urinary bladder of rats treated with o-nitroanisole (r.t. of 5.2 min).
Because o-anisidine was found to be oxidized also by peroxidases in vitro to a diimine metabolite capable of generating dGp adducts detectable by the conventional 32P-postlabeling chromatography suitable for hydrophobic bulky adducts (method A),24, 25 we analyzed DNA of the urinary bladder from rats exposed to o-anisidine also with this method. In order to digest DNA effectively, 3 times higher amounts of an MN/SPD digestion mixture were used in the experiments.24, 25 No DNA adducts were detectable by this version of the 32P-postlabeling assay (results not shown).
We show for the first time that human hepatic microsomes can metabolize carcinogenic o-anisidine. The results of this study clearly demonstrate that human hepatic microsomes catalyze N-hydroxylation of o-anisidine, thereby activating this carcinogen to species binding to DNA. All human hepatic microsomal samples used in the study produced 2 metabolites. While one of these metabolites has been identified to be N-(2-methoxyphenyl)hydroxylamine, the structure of another o-anisidine metabolite has not yet been characterized.
DNA binding of o-anisidine activated with human hepatic microsomes in vitro was unambiguously proven by using 2 direct independent methods, namely, 32P-postlabeling and [14C]-labeled o-anisidine. Total DNA binding determined by the nuclease P1 version of 32P-postlabeling was 2 orders of magnitude lower than that determined with [14C]-labeled o-anisidine. One of the reasons for these discrepancies could be a poor efficiency of the labeling reaction for o-anisidine-derived adducts during the 32P-postlabeling method, which is the conversion of the adducted nucleoside 3′-phosphates to their corresponding 3′,5′-bisphosphates by T4 polynucleotide kinase. Indeed, recent results of Mourato et al.62 and Gonçalves et al.63 indicated that deoxyguanosine 3′-phosphate adducts formed from carcinogenic aromatic amines containing only one benzene ring were labeled at substantially lower efficiency than dGp. The discrepancies might also be caused by incomplete digestion of modified DNA, loss of material during the experimental manipulations used in 32P-postlabeling, compounds retained at the origin of the PEI-cellulose TLC plates, or loss of adducts during the enzymatic enrichment. We found that nuclease P1 digestion led to pronounced dephosphorylation of one adduct (spot 3 in Fig. 5d), the major adduct formed from dGp and N-(2-methoxyphenyl)hydroxylamine detectable by the 32P-postlabeling assay without any enrichments. Therefore, the loss of this adduct during enrichment might be the predominant reason why there was a huge discrepancy between the adduct levels reported by measurement of the radioactivity associated with the incorporation of [14C]-labeled o-anisidine (Table III) and those determined by 32P-postlabeling (Table IV). Because of its sensitivity to nuclease P1, which is a general feature of N-(deoxyguanosin-8-yl)arylamine 3′-phosphates, this adduct might be N-(deoxyguanosin-8-yl)-2-methoxyaniline. The other 2 adducts detected by the 32P-postlabeling technique might be speculated to be N2- and O6-substituted deoxyguanosine adducts, with the N2 substituted being the most prominent. Nevertheless, detailed chemical characterization of these dGp adducts has to be performed.
We also investigated the capacity of o-anisidine to form DNA adducts in vivo with the 32P-postlabeling assay. The 2 DNA adducts formed by o-anisidine in DNA of rats treated with o-anisidine, detected by the nuclease P1 version of the assay, exhibited the same chromatographic properties as the adducts generated by activated o-anisidine in vitro. Therefore, these o-anisidine-derived DNA adducts might be identical with the deoxyguanosine adducts derived from N-(2-methoxyphenyl)hydroxylamine determined in in vitro incubations. While structural characterization of these adducts remains to be resolved, these results demonstrate a genotoxic mechanism of o-anisidine carcinogenicity. The highest level of DNA adducts was found in the urinary bladder; however, DNA adduct formation was seen also in liver, kidney and spleen, but at levels more than one order of magnitude lower than in the urinary bladder. This is highly consistent with the carcinogenic activity of o-anisidine; it causes neoplastic transformation predominantly in the urinary bladder.1, 2 If the levels of DNA adducts expressed as RAL were normalized using the [14C]o-anisidine-DNA from the in vitro incubations (Tables III and IV), the binding was the order of 24 adducts per 106 nucleotides in DNA of the urinary bladder. This is similar to what has been observed for the levels of adducts in the urinary bladder of mice fed carcinogenic regimens of aromatic amines.64, 65
The same deoxyguanosine adducts as those found in DNA of tissues of rats treated with o-anisidine were detected in DNA of the urinary bladder of rats treated with o-nitroanisole, an oxidized counterpart of o-anisidine, indicating that the reactive N-(2-methoxyphenyl)hydroxylamine metabolite of both carcinogens is critical for the formation of DNA lesions in target organs. DNA adduct levels formed in bladders, kidneys and livers of rats treated by either carcinogens are comparable, but 2-fold higher amounts of adducts was found in spleen after exposure of rats to o-nitroanisole56 than those exposed to o-anisidine.
The present study documents the role of specific human microsomal enzymes in activation pathways of o-anisidine. CYP2E1 seems to be the principal enzyme responsible for activation of o-anisidine in human liver. The role of this CYP enzyme in o-anisidine activation was supported by strong correlation coefficients between the rates of chlorzoxazone 6-hydroxylation, a CYP2E1-dependent reaction, and the rate of o-anisidine oxidation. This was confirmed also by the inhibition of N-(2-methoxyphenyl)hydroxylamine formation by DDTC, an inhibitor of CYP2E1. It should be noted that the interpretation of the results of inhibitors is sometimes difficult, because one inhibitor may be more effective with one substrate than another. Nevertheless, utilization of microsomes containing human recombinant CYP2E1 as well as pure rabbit CYP2E1 fully corroborated the role of CYP2E1 in the metabolism of o-anisidine. Besides the CYP2E1, several other CYP enzymes might also participate in o-anisidine oxidation in humans, because the human recombinant CYP1A2 and, to a lesser extent, CYP2B6, 1A1, 2A6, 2D6 and 3A4 oxidized this carcinogen. The discrepancy between the negative correlations found in human hepatic microsomes and the data using human recombinant CYPs remains to be explained. The lack of correlation of o-anisidine metabolite levels and enzyme activities of CYP1A1, 2B6, 2A6 and 2D6 may be attributable to their low activity caused by their low expression in human livers.66 However, the negative results on participation of CYP1A2 enzyme that is expressed in human liver in large quantities (more 10% of the total hepatic CYP complement)66 are rather surprising. One of the reasons for the observed discrepancies might be the different substrate specificities of recombinant and authentic human CYP enzymes.
The finding that the human recombinant CYP1A1, 2B6 and 2D6 efficiently activate o-anisidine may be of great significance. Human exposure to o-anisidine is thought to occur primarily via the respiratory tract. Although the total CYP content of the lungs is low compared to the liver, the specific activity based on mg protein can be higher than in the liver.67 CYP enzymes present in lungs may hence play an important role in extrahepatic oxidation of o-anisidine. In humans, CYP1A1 is mainly an extrahepatic enzyme expressed in the gastrointestinal and urinary tract and in lung upon induction,66 while CYP2B6 is expressed in several extrahepatic tissues, including lungs.68 The reports on pulmonary expression of CYP2D6 are contradictory. Guidice et al.69 observed expression of this enzyme in lung tissue at mRNA and protein levels to varying extent, depending on the lung tissue examined. Other authors found no expression of CYP2D6 in the lung samples they analyzed by immonohistochemistry or RT-PCR.68, 70
The present results clearly indicate that one of the pathways of bioactivation of o-anisidine in vivo in rats is its N-hydroxylation catalyzed by CYP enzymes. The DNA adduct formation is due to the reactivity of N-(2-methoxyphenyl)hydroxylamine, which forms spontaneously DNA-binding arylnitrenium ions (Tables IV–VI). However, phase 2 enzymes, namely, SULTs, present in the human hepatic cytosol, are also involved through O-sulfonation, resulting in the formation of a highly reactive electrophilic nitrenium ion. In contrast to SULTs, NATs have a marginal impact on o-anisidine-DNA binding under the used conditions (Table III). These results are consistent with the finding that the highest level of o-anisidine-DNA adducts were found in the urinary bladder of rats treated with this carcinogen. It is well known that N-sulfonyloxyesters of many N-hydroxy-arylamines formed in the liver are unstable under acidic conditions in the urinary bladder and decompose to electrophilic nitrenium ions capable of reacting with DNA in this target organ. In addition, in human bronchial epithelial cells and alveolar macrophages, SULT1A1 and SULT1A2 are expressed.71, 72 Their expression in the human respiratory tract could contribute significantly to the metabolic activation of o-anisidine. However, the precise kinetics of sulfonation of o-anisidine metabolites in human tissue awaits further investigation.
One of the most important results found in our present study is the finding that activation of o-anisidine by the human enzymatic system is analogous to that observed in rats. Human hepatic microsomal samples formed DNA adducts chromatographically indistinguishable from the adducts formed in rats in vivo. Our results, showing for the first time an analogy in the o-anisidine activation to species forming DNA adducts catalyzed by human enzymes and in rats in vivo, strongly suggest a carcinogenic potential of this rodent carcinogen for humans.
Finally, our study can form the basis for chemical characterization of the o-anisidine adducts found in vivo and subsequently for development of methods to monitor human exposure. To better understand the potential role of o-anisidine-DNA adducts in induction of cancer, our results require confirmation by larger animal studies, monitoring the dose-dependent formation and persistence of o-anisidine-DNA adducts in target tissues after inhalative and oral exposure.
The authors thank Dr. J. Ashby and Dr. P.A. Lefevre for their gift of the [14C]o-anisidine. This article is dedicated to Professor Dr. Sylva Leblová.