Synthesis, Characterization, and 32P-Postlabeling ... - ACS Publications

1056

Chem. Res. Toxicol. 2005, 18, 1056-1070

Synthesis, Characterization, and 32P-Postlabeling Analysis of DNA Adducts Derived from the Environmental Contaminant 3-Nitrobenzanthrone Martin R. Osborne,†,‡ Volker M. Arlt,*,†,‡ Christian Kliem,§ William E. Hull,| Amin Mirza,⊥ Christian A. Bieler,§ Heinz H. Schmeiser,§ and David H. Phillips† Section of Molecular Carcinogenesis and Cancer Research UK Centre for Cancer Therapeutics, Institute of Cancer Research, Brookes Lawley Building, Cotswold Road, Sutton, Surrey SM2 5NG, United Kingdom, and Division of Molecular Toxicology and Central Spectroscopy Department, German Cancer Research Center, Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany Received February 22, 2005

3-Nitrobenzanthrone (3-NBA) is a potent mutagen and potential human carcinogen identified in diesel exhaust and ambient air particulate matter. 3-NBA forms DNA adducts in rodent tissues that arise principally through reduction to N-hydroxy-3-aminobenzanthrone (N-OHABA), esterification to its acetate or sulfate ester, and reaction of this activated ester with DNA. We detected 3-NBA-derived DNA adducts in rodent tissues by 32P-postlabeling and generated them chemically by acid-catalyzed reaction of N-OH-ABA with DNA, but their structural identification has not yet been reported. We have now prepared 3-NBA-derived adducts by reaction of a possible reactive metabolite, N-acetoxy-N-acetyl-3-aminobenzanthrone (N-Aco-N-Ac-ABA), with purine nucleosides and nucleotides, characterized them, and have shown that they are present in DNA treated with this 3-NBA derivative. Three of these adducts have been characterized as the C-C adduct N-acetyl-3-amino-2-(2′-deoxyguanosin-8-yl)benzanthrone, the C-N adduct N-acetyl-N-(2′-deoxyguanosin-8-yl)-3-aminobenzanthrone, and an unusual 3-acetylaminobenzanthrone adduct of deoxyadenosine, which involves a double linkage between adenine and benzanthrone (N1 to C1, N6 to C11b), creating a five-membered imidazo type ring system. According to IUPAC fused ring conventions, we propose the following systematic name for this adduct: (9′-(2′′-deoxyribofuranosyl))purino[6′,1′:2,3]imidazo[5,4-p](1,11b-dihydro-(N-acetyl-3-amino))benzanthrone. The 3′-phosphates of these novel adducts could be 5′-postlabeled using [γ-32P]ATP, although the efficiency of labeling was found to be low (less than 20%). However, none of these adducts could be detected in DNA from 3-NBA-treated rats by 32P-postlabeling. Two of these synthetic adducts were treated with alkali to generate nonacetylated adducts, and these were also shown by HPLC to differ from those adducts found in rat DNA. Therefore, a different approach to the synthesis of authentic standards is needed for the structural characterization of 3-NBA-derived DNA adducts formed in vivo.

Introduction (nitro-PAHs)1

Nitropolycyclic aromatic hydrocarbons are widespread environmental contaminants found in extracts from diesel and gasoline engine emissions, on the surface of ambient air particulate matter, in river sediments, and in grilled food (1-4). The increased lung * To whom correspondence should be addressed. Tel: +44-2087224405. Fax: +44-2087224052. E-mail: [email protected]. † Section of Molecular Carcinogenesis, Institute of Cancer Research. ‡ These authors contributed equally to the work. § Division of Molecular Toxicology, German Cancer Research Center. | Central Spectroscopy Department, German Cancer Research Center. ⊥ Cancer Research UK Centre for Cancer Therapeutics, Institute of Cancer Research. 1 NBA, 3-nitrobenzanthrone; 3-ABA, 3-aminobenzanthrone; N-OHABA, N-hydroxy-3-aminobenzanthrone; BA, benzanthrone; 3-Ac-ABA, 3-acetylaminobenzanthrone; N-Ac-N-OH-ABA, N-acetyl-N-hydroxy-3aminobenzanthrone; N-Aco-N-Ac-ABA, N-acetoxy-N-acetyl-3-aminobenzanthrone; nitro-PAH, nitropolycyclic aromatic hydrocarbon; dG, 2′-deoxyguanosine; dA, 2′-deoxyadenosine; dGp, 2′-deoxyguanosine 3′monophosphate; dAp, 2′-deoxyadenosine 3′-monophosphate; XO, xanthine oxidase; NAT, N,O-acetyltransferase; SULT, sulfotransferase; NQO1, NAD(P)H:quinone oxidoreductase; TLC, thin-layer chromatography; RAL, relative adduct labeling; ROE, rotating-frame Overhauser effect.

cancer risk from exposure to air pollutants and the detection of nitro-PAHs in lungs of nonsmokers with lung cancer have led to considerable interest in assessing their potential risk to humans (5-7). 3-Nitrobenzanthrone (3NBA, 3-nitro-7H-benz[de]anthracen-7-one; Scheme 1) belongs to this group of compounds and is present in diesel exhaust and airborne particulate matter (8-11), in surface soil, and in rainwater (12-14). It is one of the most mutagenic compounds ever tested in the Ames Salmonella typhimurium assay (8). 3-NBA is also a potent mutagen in transgenic mice and in human cells (15, 16), genotoxic in several short-term bioassays (8, 1720), and probably carcinogenic in rats (21). Furthermore, human occupational exposure to 3-NBA from diesel emissions is detectable and significant; 3-aminobenzanthrone (3-ABA, Scheme 1), the major metabolite of 3-NBA, was recently detected in the urine of salt mining workers occupationally exposed to diesel emissions (10). The principal mode of action of 3-NBA in rodents appears to be reduction and esterification to give reactive acetate and sulfate esters of N-hydroxy-3-aminobenzanthrone (N-OH-ABA; Scheme 1), which react with DNA

10.1021/tx0500474 CCC: $30.25 © 2005 American Chemical Society Published on Web 05/25/2005

3-Nitrobenzanthrone-Derived Adduct Standards

Chem. Res. Toxicol., Vol. 18, No. 6, 2005 1057

Scheme 1. Potential Pathways of Metabolic Activation and DNA Adduct Formation of 3-NBA and Its Metabolites 3-ABA or N-Ac-N-OH-ABA (See Text for Details)a

a N,O-AT, N,O-acetyl transfer (also catalyzed by NATs); P450, cytochrome P450; DAase, deacetylase; R ) -C(O)CH or -SO H; Me ) 3 3 -CH3; ?? ) unknown in vivo pathways.

Scheme 2. Structures of the Authentic DNA Adduct Standards Prepared for This Studya

a The acetylated DNA adducts were formed by the reaction of the corresponding nucleosides or nucleotides with N-Aco-N-Ac-ABA. Standards 1, 2, 9, and 10 were deacetylated with lithium hydroxide to give 3, 4, 11, and 12. For the novel five ring adducts, 9-12, two diastereomers with opposite optical rotation were obtained and are assumed to correspond to the two possible configurations (1S,11bR), as shown, and (1R,11bS) (see Figure 3). The atom numbering scheme shown is used in the text and for NMR assignments with a prefix A or G to refer to atoms in the adenine or guanine rings.

to give covalent adducts (16, 18, 22-27). We have previously shown that all 3-NBA-derived DNA adducts detected by 32P-postlabeling in rodents treated with 3-NBA are derived from reductive metabolites bound to purine bases (16, 18, 23, 27). However, these DNA adducts have not yet been structurally characterized. One DNA adduct derived from 3-NBA, which has been characterized to date, was synthesized by reacting Nacetoxy-N-acetyl-3-aminobenzanthrone (N-Aco-N-Ac-ABA; Scheme 1), the activated ester of N-acetyl-N-hydroxy-3-

aminobenzanthrone (N-Ac-N-OH-ABA; Scheme 1), with deoxyguanosine (dG) (28). The resulting adduct, N-acetyl3-amino-2-(2′-deoxyguanosine-8-yl)benzanthrone (structure 1; Scheme 2), features a C-C linkage between C8 of the guanine ring and C2 of benzanthrone (BA) (28). This adduct was detected by 32P-postlabeling in DNA of human HepG2 cells treated with 3-NBA (29). Collectively, these studies suggest that 3-NBA can be activated by two major pathways (Scheme 1). After nitroreduction to N-OH-ABA, the first pathway involves

1058

Chem. Res. Toxicol., Vol. 18, No. 6, 2005

Osborne et al.

Table 1. HPLC Analysis of dG Modified with N-Aco-N-Ac-ABAa acetylated dG adducts retention time (min) λmax (nm) yield (absorbance units)

dG-N-Ac-ABA-1

dG-N-Ac-ABA-2 (1)

dG-N-Ac-ABA-3

dG-N-Ac-ABA-4

dG-N-Ac-ABA-5 (5)

19.5 257 26

21.0 275, 388 57

25.0 243, 275, 427 8

25.0 287, 403 9

26.0 259, 402 77

a Nucleoside adducts were separated on an ODS column with 20% acetonitrile in 0.05 M ammonium formate, increasing to 50% over 30 min. dG-N-Ac-ABA-3 and dG-N-Ac-ABA-4 were separated on Sephadex LH-20 but had identical retention times on HPLC.

the formation of a nitrenium ion yielding nonacetylated DNA adducts. The second pathway proceeds via the formation of N-Ac-N-OH-ABA and an N-acetyl-nitrenium ion yielding acetylated DNA adducts. Although it has been established by 32P-postlabeling that the major 3-NBA-derived DNA adducts formed in vivo arise from reaction of deoxyadenosine (dA) (adduct spots 1 and 2) and dG (adduct spots 3, 4, and 5) with reductive metabolites of 3-NBA not carrying an N-acetyl group (16, 18, 23, 27), the structures of these adducts remain to be fully characterized. The strategy used in the present study for the synthesis of 3-NBA-derived DNA adducts involves the reaction of N-OH-ABA or N-Aco-N-Ac-ABA with purine nucleosides, nucleotides, or DNA. In view of the current relevance of the 32Ppostlabeling method for the detection and quantitation of 3-NBA-derived DNA adducts observed in various in vitro bioassays, in cells and in vivo in rodents treated with 3-NBA (16-18, 20, 22-31), we describe the synthesis and characterization of three acetylated and two nonacetylated 3-NBA-derived DNA adducts (Scheme 2) for use as standards in the 32P-postlabeling assay. We also report the recoveries of the 32P-labeled standards in the nuclease P1 digestion and butanol extraction enrichment procedures of the 32P-postlabeling assay.

Materials and Methods Cautions: 3-NBA and its derivatives are mutagens and potential carcinogens and should be handled with care. Exposure to 32P should be avoided, by working in a confined laboratory area, with protective clothing, Perspex or plexiglass shielding, Geiger counters, and body dosimeters. Wastes must be discarded according to appropriate safety procedures. Instrumentation. 1H and 13C NMR spectra were recorded on Bruker Avance DPX-250 and AM-500 spectrometers. Chemical shifts are referenced to tetramethylsilane as the internal standard (0.0 ppm) or to the solvent signals of DMSO-d6 (2.4985 ppm for 1H, 39.50 for 13C). Spectrum analysis was carried out using the Bruker software WIN NMR and WIN-DAISY (iterative simulation of 500 MHz 1H spectra). Liquid chromatography-mass spectrometry (LC-MS) measurements were made on a Waters LCT Mass Spectrometer (Lockspray), equipped with an electrospray ion source (ESI) and Waters 2795 HPLC, using the chromatography system (c) described below. UV-visible spectrophotometry was carried out with a Beckman DU7 spectophotometer. Optical rotation was measured at 589 nm with a Perkin-Elmer 141 Polarimeter. 32P-Postlabeling HPLC analysis was carried out with a Waters 2690 HPLC system using the chromatography system (b) described below. Molecular modeling was performed with ChemOffice 2004 (CambridgeSoft Corp., Cambridge, MA) using the MM2 force field and with HyperChem 5.1 (HyperCube, Inc.) using the MM+ force field and the semiempirical methods AM1 and PM3. Chromatography. LC was carried out on a Waters apparatus. The systems used were as follows: (a) a reversed phase column (Jupiter ODS 4.6 mm × 250 mm; Phenomenex, United Kingdom) eluted with 0.05 M ammonium formate containing acetonitrile at a flow rate of 0.8 mL/min. 3-NBA derivatives were detected in the eluate by their UV absorbance at 254 nm; (b) a

phenyl-modified reversed phase column (Luna 5 µm phenylhexyl, 4.6 mm × 150 mm, Phenomenex) with a linear gradient of methanol (from 30 to 55% in 45 min) in aqueous 0.5 M sodium phosphate (pH 3.5) at a flow rate of 1 mL/min. Radioactivity eluting from this column was measured by monitoring Cerenkov radiation with a Flow Scintillation Analyzer (Packard, Dowers Grove, IL); (c) a reversed phase column (Gemini 5 µm C18, 4.6 mm × 50 mm, Phenomenex) eluted with a linear gradient of methanol (10-90% in 7.5 min) in 0.1% formic acid at a flow rate of 1 mL/min was used. Chemicals. The nucleosides dA and dG, the nucleotides 2′deoxyadenosine 3′-monophosphate (dAp) and 2′-deoxyguanosine 3′-monophosphate (dGp), calf thymus DNA, and salmon testis DNA were from Sigma Chemical Co. (Poole, United Kingdom). Enzymes and chemicals for the 32P-postlabeling assay were obtained from commercial sources as before (22, 23). Synthesis of 3-NBA, N-OH-ABA, N-Ac-N-OH-ABA, and N-Aco-N-Ac-ABA. 3-NBA was synthesized as reported previously (22). N-OH-ABA was prepared as follows: 20 mg of 3-NBA in 10 mL of diglyme was reduced by stirring with 50 µL of hydrazine hydrate and 10 mg of 5% palladium on charcoal under nitrogen for 30 min (26). N-Ac-N-OH-ABA and N-Aco-N-Ac-ABA were synthesized essentially as described (28). The authenticity of 3-NBA was confirmed by UV, ESI-MS, and 250 MHz 1H NMR spectroscopy. The authenticity of N-Ac-N-OH-ABA, N-Aco-NAc-ABA, and N-OH-ABA was confirmed by UV and ESI-MS spectroscopy. Preparation of Adducts of N-Acetyl-3-aminobenzanthrone with dG: dG-N-Ac-ABA-2 (Structure 1, Scheme 2) and dG-N-Ac-ABA-5 (Structure 5, Scheme 2). The dG adducts were prepared by reaction of N-Aco-N-Ac-ABA (30 mg) with dG (40 mg) in acetonitrile-water (55 °C, 18 h) as described (28). The reaction mixture was separated by LC on a Sephadex LH20 column (1.4 cm × 80 cm), eluted with 0.05 M ammonium formate in methanol-water, 10-80% methanol in 4 h, and fractions were further analyzed on an ODS column (system a). Adducts thus obtained are listed in Table 1. dG-N-Ac-ABA-2 (1) was identified by the shape of its UV-visible spectrum and by its 1H NMR spectrum, which agreed well with that reported previously for the C-C adduct N-acetyl-3-amino-2-(2′-deoxyguanosin-8-yl)benzanthrone (28). Adduct dG-N-Ac-ABA-5 (5) was the most abundant product. The yellow substance was sparingly soluble in water. The negative ion ESI-MS spectrum gave peaks at m/z 509 (int. 23%) for [M - Ac]- and 551 (100%) for [M - H]-, as expected for a dG-N-Ac-ABA adduct with the molecular formula C29H24N6O6 (calcd M ) 552.176). The adduct rapidly hydrolyzed in 0.1 M HCl (half-life of 6 min at 37 °C; single product) or 0.1 M NaOH (half-life of 12 min at 37 °C; two products), suggesting that substitution had occurred on the amide nitrogen. 1H and 13C measurements of dG-N-Ac-ABA-5 (5) in DMSO-d6 at 11.7 T and 30 °C gave spectra with a wide range of site-dependent line widths and signal integrals consistent with a ca. 2:1 ratio of two conformations in the slow to intermediate exchange limit. The structure of dG-N-Ac-ABA-5 (5) was confirmed by acquiring spectra at temperatures up to 110 °C (see Results). Preparation of Adducts of N-Acetyl-3-aminobenzanthrone with Deoxyguanosine-3′-phosphate: dGp-N-AcABA-2 (Structure 2, Scheme 2) and dGp-N-Ac-ABA-5 (Structure 6, Scheme 2). These dGp adducts were prepared essentially as described previously (29). To dGp (4 mg) in 0.4 mL of 0.3 M sodium acetate (pH 6.0) was added 4 mg of N-Aco-



Table 2. HPLC Analysis of dGps Modified with N-Aco-N-Ac-ABAa acetylated dGp adducts retention time (min) λmax (nm) yield (absorbance units)

dGp-N-Ac-ABA-1

dGp-N-Ac-ABA-2 (2)

dGp-N-Ac-ABA-4

dGp-N-Ac-ABA-5 (6)

26.0 256 2

25.0 275, 392 11

32.0 287, 403 1

31.0 259, 402 12

a Nucleotide adducts were separated on an ODS column with 10% acetonitrile in 0.05 M ammonium formate, increasing to 50% in 40 min.

N-Ac-ABA in 0.4 mL of acetonitrile, and the mixture was incubated for 18 h at 55 °C. The reaction mixture was separated by LC on ODS (system a) with 10-50% acetonitrile in 40 min, yielding the adducts listed in Table 2. Adduct dGp-N-Ac-ABA-2 (2) exhibited a UV spectrum very similar to that of dG-N-AcABA-2 (1) and was converted to it by dephosphorylation with alkaline phosphatase. The identity of dGp-N-Ac-ABA-2 (2) as the 3′-phosphate derivative of N-acetyl-3-amino-2-(2′-deoxyguanosin-8-yl)benzanthrone with the molecular formula C29H25N6O9P (calcd M ) 632.142) was confirmed by MS and 1H NMR spectroscopy. The ESI-MS spectrum in the positive ion mode gave peaks at m/z 286 (int. 64%) [M + H - dGp]+, 437 (20%) [M + H - dRp*]+, 633 (100%) [M + H]+, and 655 (51%) [M + Na]+, where dRp* represents neutral loss of the deoxyribosyl 3′-phosphate fragment C5H9O6P (mass 196) through cleavage of the ribose-guanine bond with H atom transfer to the guanine residue. The 1H NMR spectrum of dGp-N-Ac-ABA-2 (2) (250 MHz, DMSO-d6, 25 °C) was analyzed on the basis of the 500 MHz spectrum obtained for dG-N-Ac-ABA-2 (1). The assignable signals were resolved as follows: δ 11.34 (br s, 1H, N1-H), 8.736 (s, 1H, H1), 8.720 (dd, 1H, H6), 8.604 (dd, 1H, H4), 8.562 (d, 1H, H11), 8.370 (dd, 1H, H8), 8.002 (t, 1H, H5), 7.865 (td, 1H, H10), 7.669 (t, 1H, H9), 6.44 (br s, 2H, G2-NH2), 5.832 (dd, 1H, H1′), 4.718 (m, 1H, H3′), 3.729 (m, 1H, H4′), 2.68 (m, 1H, H2′a), 2.063 (br s, 3H, NAc CH3). Because of the modest quality of the spectrum, only crude estimates for the following JHH coupling constants (in Hz) could be obtained as follows: J4,5 ) 7.7, J4,6 ) 1.0, J5,6 ) 7.7, J8,9 ) 7.8, J8,10 ) 1.2, J9,10 ) 7.7, J10,11 ) 7.7, and J1′,2′ ) 8.1, 6.5. Signals for 3-NH, 3′-OH, and 5′-OH were too broad for detection; signals for H5′ and H2′b in the range of 3.0-3.6 ppm were obscured by large solvent peaks. Similarly, dGp-N-Ac-ABA-5 (6) was shown to be the 3′phosphate derivative of N-acetyl-N-(2′-deoxyguanosin-8-yl)-3aminobenzanthrone (5) on the basis of the UV spectrum, dephosphorylation with alkaline phosphatase, and MS and NMR data. The positive ion ESI-MS spectrum gave peaks at m/z 214 (int. 88%), 245 (28%), 437 (100%) [M + H - dRp*]+, 475 (25%), 553 (10%) [M - HPO3 + H]+, and 575 (11%) [M HPO3 + Na]+. The 1H NMR spectrum (250 MHz, DMSO-d6, 25 °C) was similar to that obtained at 500 MHz for dG-N-Ac-ABA-5 (5) with substantial line broadening due to the conformational dynamics at the 3-amino position. The signals could be assigned as follows: δ 8.67 (d, 1H, H6), 8.33 (d, 1H, H8), 8.02 (t, 1H, H5), 6.64 and 6.51 (s, 2H, G2-NH2, two conformers), 6.37 and 6.28 (br, 1H, H1′, two conformers), 5.00 (m, 1H, H3′), 2.19 and 1.92 (s, 3H, NAc CH3, two conformers). On the basis of their UV spectra and dephosphorylation with alkaline phosphatase, the adducts dGp-N-Ac-ABA-1 and dGpN-Ac-ABA-4 were also shown to be the 3′-phosphates of dG-NAc-ABA-1 and dG-N-Ac-ABA-4, respectively. Preparation of 3-Amino-2-(2′-deoxyguanosin-8-yl)benzanthrone (dG-ABA-2; Structure 3, Scheme 2) and 3-Amino2-(2′-deoxyguanosin-3’-phosphate-8-yl)benzanthrone (dGpABA-2; Structure 4, Scheme 2). dG-ABA-2 (3) and dGp-ABA-2 (4) were made by deacetylation of dG-N-Ac-ABA-2 (1) and dGpN-Ac-ABA-2 (2), respectively, as described (28). Briefly, 100 µg of dG-N-Ac-ABA-2 (1) was treated with 0.1 M LiOH in ethylene glycol-water (1 mL, 4:1) at 50 °C for 2 days. The reaction mixture was separated on an ODS column (system a); elution with 26% acetonitrile gave 90 µg of dG-ABA-2 (3) with a retention time of 20 min. Similarly, treatment of 70 µg of dGpN-Ac-ABA-2 (2) gave 50 µg of dGp-ABA-2 (4), eluted after 20

min with 18% acetonitrile. The UV-visible spectra of dG-ABA-2 (3) and dGp-ABA-2 (4) (λmax at 284 and 503 nm) were similar to that reported for 3-amino-2-(2′-deoxyguanosin-8-yl)benzanthrone (28). For dG-ABA-2 (3), the ESI-MS spectrum gave positive ions at m/z 395 (int. 100%) [M + H - dR*]+, 511 (12%) [M + H]+, and 533 (8%) [M + Na]+, where dR* represents neutral loss of a deoxyribosyl fragment C5H8O3 (mass 116); calcd M ) 510.165 for C27H22N6O5. For dGp-ABA-2 (4), positive ions were detected at m/z 395 (int. 100%) [M + H - dRp*]+ and 591 (8%) [M + H]+, where calcd M ) 590.132 for C27H23N6PO8. Preparation of dA-N-Ac-ABA-1 (Structure 9, Scheme 2). Forty milligrams of N-Aco-N-Ac-ABA in 4.5 mL of acetonitrile was added to 60 mg of dA in 6 mL of 0.01 M sodium phosphate, pH 7.0, and kept at 55 °C for 16 h. Then, 20 mL of 0.05 M ammonium formate was added, and the reaction mixture was run on a Sephadex LH20 column (1.4 cm × 80 cm), eluted at 1.5 mL/min with 0.05 M ammonium formate in methanol, 1040% in 120 min, to 70% methanol in 30 min. The adduct was eluted at 180-200 min. These fractions were evaporated to 10 mL and run in 0.15 mL lots on an ODS column (system a), eluted with 0.05 M ammonium formate in 19% MeCN, 0.8 mL/ min. The product was separated into two isomers, dA-N-AcABA-1a (25 min) and dA-N-Ac-ABA-1b (26 min), with a yield of about 3 mg of each. The isomers were identical in quantity, in UV spectrum (λmax 271.5 nm, with a tail around 320-380 nm), spectroscopic pKa (4.6), and in their positive ion ESI-MS spectra: m/z ) 421 (int. 21%, 1a; 10%, 1b) [M + H - dR*]+ and 537 (100%, 1a; 100%, 1b) [M + H]+, where calcd M ) 536.181 for C29H24N6O5. The two isomers gave essentially identical 1H NMR spectra at 250 MHz. Complete assignments for the 1H and 13C NMR data obtained at 11.7 T for isomer 1a were accomplished via 1H-1H COSY and ROESY 2D and 13C-1H correlations via one-bond and three-bond couplings (COLOC) (see Results). The optical rotations (RD, 22 °C, 1 g/L ethanolwater) of dA-N-Ac-ABA-1a and dA-N-Ac-ABA-1b were about +55 and -37°, respectively. Treatment with 0.1 M HCl at 37 °C removed deoxyribose (half-life of ca. 30 min) to give the true enantiomers Ade-N-Ac-ABA-1a and Ade-N-Ac-ABA-1b with RD ) +94 and -98°, respectively (22 °C, 1 g/L ethanol-water). Preparation of dAp-N-Ac-ABA-1 (Structure 10, Scheme 2). This adduct was prepared essentially as described previously (29). To dAp (4 mg) in 0.4 mL of 0.3 M sodium acetate (pH 6.0) was added 4 mg of N-Aco-N-Ac-ABA in 0.4 mL of acetonitrile, and the mixture was incubated for 18 h at 55 °C. Chromatography on an ODS column (system a, 10-50% acetonitrile gradient over 40 min) yielded one major adduct peak at a retention time of 24 min. UV spectrum: λmax 271 nm, with a tail around 320-380 nm. Positive ion ESI-MS: m/z 421 (int. 8%) [M + H - dRp*]+, 617 (100%) [M + H]+, and 639 (11%) [M + Na]+, in agreement with M ) 616.147 for C29H25N6O8P. The 1H NMR spectrum (250 MHz, DMSO-d , 25 °C) gave peaks at δ 6 9.41 (s, 1H, 3-NH), 8.612 (s, 1H, A2), 8.12 (s, 1H, A8), 8.078 (dd, 1H, H8), 8.066 (dd, 1H, H6), 7.695 (dd, 1H, H4), 7.684 (dd, 1H, H11), 7.642 (ddd, 1H, H10), 7.584 (t, 1H, H5), 7.513 (ddd, 1H, H9), 6.490 (d, 1H, H1), 6.262 (d, 1H, H2), 6.203 (dd, 1H, H1′), 4.771 (m, 1H, H3′), 4.043 (ddd, 1H, H4′), 3.553 (dd, 1H, H5′b), 3.535 (dd, 1H, H5′a), and 2.05 (br s, 3H, NAc CH3). The H2′a,b protons were obscured by solvent peaks. First-order analysis gave the following 1H coupling constants: J1,2 ) 3.2, J4,5 ) 7.8, J4,6 ) 1.2, J5,6 ) 7.4, J8,9 ) 7.2, J8,10 ) 1.2, J9,10 ) 7, J9,11 ) 1.3, J10,11 ) 7.6, J1′,2′a ) 7, J1′,2′b ) 7, and J3′,4′ ) 3.3.

1060


Preparation of dA-ABA-1 (Structure 11, Scheme 2) and dAp-ABA-1 (Structure 12, Scheme 2). These adducts were made by deacetylation of dA-N-Ac-ABA-1 (9) and dAp-N-AcABA-1 (10) with lithium hydroxide, respectively, as for the guanine derivatives reported above; however, the yields were lower. dA-N-Ac-ABA-1 (9) (0.4 mg; either isomer) gave a red substance, dA-ABA-1 (11) (0.07 mg), with λmax 285 (sh 320) and 504 nm; positive ion ESI-MS: m/z ) 379 (int. 100%) [M + H dR*]+, 495 (16%) [M + H]+, and 517 (10%) [M + Na]+, where calcd M ) 494.170 for C27H22N6O4. In an analogous manner, the deacetylation of 0.15 mg of dApN-Ac-ABA-1 (10) gave 0.04 mg of dAp-ABA-1 (12), with nearly the same UV-visible spectrum, λmax 283 (sh 320) and 505 nm. Its negative ion mass spectrum gave a single ion, m/z 493 [M H2PO3]-, where calcd M ) 574.137 for C27H23N6O7P. There was not enough of dA-ABA-1 (11) or dAp-ABA-1 (12) with which to obtain further evidence of their structure. Preparation of Other Nucleotide-ABA Adducts. dAp and dGp (4 µmol/ml) were incubated with 3-NBA (0.3 mM) enzymatically activated by xanthine oxidase (XO; 1 unit/ml) in 50 mM potassium phosphate buffer, pH 7.0, in the presence of 1 mM hypoxanthine as described (18). Aliquots of the incubation mixtures were used directly for the butanol extraction-mediated 32P-postlabeling procedure. Reaction of N-OH-ABA and N-Aco-N-Ac-ABA with DNA. N-OH-ABA was mixed with a solution of 50 mg of salmon testis DNA in 50 mL of 0.06 M sodium cacodylate, pH 6.0, and the mixture was incubated for 18 h at 60 °C under nitrogen (26). The mixture was extracted three times with 20 mL of ethyl acetate, and the DNA was precipitated from the aqueous phase with ethanol. Chemical modification of salmon testis DNA with N-Aco-N-Ac-ABA was performed as described previously (28). Digestion of N-OH-ABA- and N-Aco-N-Ac-ABA-Modified DNA to Nucleosides. Reacted DNA (5 mg) was dissolved in 1 mL of 0.01 M Tris (pH 7.0) and 0.01 M magnesium chloride and treated at 37 °C with 0.2 mg of DNase I (18 h; Sigma, United Kingdom), with 0.15 mL of 0.5 M Tris (pH 9.0), 0.05 units of Crotalus adamanteus venom phosphodiesterase (6 h; Sigma), and 6 units of alkaline phosphatase (18 h; Sigma). Treatment of Rats with 3-NBA. Female Wistar rats (n ) 3) were treated with a single dose of 3-NBA (2 mg/kg body weight, i.p.) in tricaprylin (0.5 g/L). The animals were killed after 24 h, and DNA was isolated from several tissues and analyzed as described recently (23). 32P-Postlabeling Thin-Layer Chromatography (TLC) Analysis. The approximate concentrations of DNA adduct standards were determined spectrophotometrically, assuming a maximum molar absorbance of 20000. Adduct standards dissolved in water were added to 4 µg of calf thymus DNA, digested, enriched, and analyzed as reported (22). For resolution of DNA adduct spots, the chromatographic conditions were used as follows: D1, 1.0 M sodium phosphate, pH 6.0; D3, 4 M lithium formate, 7 M urea, pH 3.5; D4, 0.8 M lithium chloride, 0.5 M Tris, 8.5 M urea, pH 8.0. Detection and quantitation of DNA adducts were performed using a Packard Instant Imager. DNA adduct spots were numbered as described previously (18, 23, 25, 32). DNA adduct levels (RAL, relative adduct labeling) were calculated from the adduct cpm, the specific activity of [γ-32P]ATP, and the amount of DNA (pmol of DNA-P) used. HPLC Analysis of 32P-Labeled Deoxyribonucleoside3′,5′-Bisphosphate Adducts. Individual adduct spots detected by the 32P-postlabeling TLC assay, or the origin after D1 only, were excised from the TLC plates, extracted with pyridinium formate buffer, and cochromatographed (system b) with standard bisphosphate adducts essentially as described (23).

Results Synthesis and Characterization of GuanineContaining Adducts. To characterize the 3-NBAderived DNA adducts formed in different in vitro bioassays, in cells and in vivo in rodents treated with 3-NBA, we needed to prepare modified nucleosides and nucle-

Osborne et al.

otides as markers for these adducts, in sufficient quantity for structural characterization. We attempted to prepare dG-ABA adducts directly by reaction of dG with N-OHABA in acid solution but did not obtain satisfactory yields this way. We therefore decided to work with the more stable N-acetylated derivatives. The strategy was to prepare and characterize the acetylated adducts dG-NAc-ABA and dGp-N-Ac-ABA and then convert them to nonacetylated adducts by treatment with alkali. The reaction of N-Aco-N-Ac-ABA with dG gave at least five adducts (Table 1). Of the principal two adducts, one (dG-N-Ac-ABA-2; 1) has been identified as the C-C adduct N-acetyl-3-amino-2-(2′-deoxyguanosin-8-yl)benzanthrone (28), and the other (dG-N-Ac-ABA-5; 5) is the corresponding C-N adduct N-acetyl-N-(2′-deoxyguanosin8-yl)-3-aminobenzanthrone, which was characterized for the first time by variable temperature NMR as described below. The other three adducts (dG-N-Ac-ABA-1, dG-NAc-ABA-3, and dG-N-Ac-ABA-4) have not yet been identified. On positive ion MS, both dG-N-Ac-ABA-1 and dGN-Ac-ABA-4 gave a molecular ion at m/z 553 as expected for dG-N-Ac-ABA adducts; dG-N-Ac-ABA-3 gave a molecular ion at m/z 551 and may be an oxidized adduct. The third most abundant adduct, dG-N-Ac-ABA-1, was resolvable into two diastereoisomers, as was dA-N-AcABA-1 (9), and may therefore be a cyclic adduct analogous to it. The reaction of N-Aco-N-Ac-ABA with dGp gave the corresponding nucleotide adducts (Table 2). For adduct dG-N-Ac-ABA-5 (5), a series of 500 MHz 1 H spectra were acquired with the temperature increasing from 30 to 110 °C in 10 °C steps (see selected spectra in Figure 1). These spectra exhibited the broadening, coalescence, and line narrowing phenomena typical for the transition from the slow to the fast exchange limit for a conformational averaging process. At 30 °C, only the BA protons 6 and 8 showed resolved doublet structure while the NAc methyl group, the G2-NH2 protons and other BA protons (1, 2, 4, and 5) exhibited two signals in approximately a 2:1 ratio, indicating the presence of two major conformers in slow exchange. At 50 °C, the resonances for H9 and H10 of BA also became wellresolved (fast exchange averaging for sites with small chemical shift differences for the two conformers) while other BA resonances, the dRib signals, and especially the acetyl methyl resonance still exhibited considerable line broadening. At 60 °C, H11 of BA also became wellresolved while the NAc methyl group showed coalescence of the signals for the two conformers. Thus, H8-H11 of BA exhibited the smallest conformation-induced chemical shift effects, consistent with hindered rotation and conformational averaging occurring at the relatively distant trisubstituted 3-amino position. It was necessary to raise the temperature to 110 °C to obtain a high resolution spectrum for all BA and deoxyribose protons, whereby some residual broadening remained for the NAc methyl resonance and the BA H2. The 1H spectrum was analyzed completely by iterative simulation and line shape fitting (WIN-DAISY). The 1H NMR spectrum at 110 °C and the data of Tables 3 and 4 clearly demonstrated that the BA ring system was intact (nine protons at sp2 carbon sites), that N1-H and G2-NH2 of guanine and the acetyl methyl group were present, and that 3-amino NH protons and the H8 proton of the guanine ring were absent. The only structure consistent with these data is N-acetyl-N-(2′-deoxyguanosin-8-yl)-3-aminobenzanthrone. The 1H chemical shifts



Figure 1. 1H NMR spectra (500 MHz) of dG-N-Ac-ABA-5 (5) in DMSO-d6 at 30 (bottom), 60 (middle), and 110 (top) °C, showing the region 9.0-3.4 ppm. The assignment labels correspond to the numbering in Scheme 2 and Table 3. Table 3. 1H and

13C

Chemical Shift Assigments for dA-N-Ac-ABA-1 (Structure 9, Scheme 2) and dG-N-Ac-ABA-5 (Structure 5, Scheme 2)a dA-N-Ac-ABA-1 (9) (in DMSO-d6, 40 °C)

positionb 1 2 3 3a 4 5 6 6a 7 7a 8 9 10 11 11a 11b 11c 3-NH NAc (CdO) NAc (Me) A2/G2 A4/G4 A5/G5 A6/G6 A8/G8 G1-H G2-NH2 1′ 2′a (2′′) 2′b (2′′) 3′ 4′ 5′a 5′b 3′-OH 5′-OH

13C

shift

mult.c

60.45 116.52 131.35 128.45 127.26 127.96 127.09 132.40 184.11 130.62 126.74 128.07 134.27 121.94 145.41 67.11 136.93

D Dm S S Dd D Dd Sd St Sdd Dd Dd Dd Dd Sddd S S

169.30 23.28 143.97 144.02 119.27 147.31 138.20

Sq Q D Sddd Sd Sdd Dd

83.71 39.85

D T

70.57 87.91 61.54

D D T

1H

shift

dG-N-Ac-ABA-5 (5) (in DMSO-d6, 110 °C) mult.

6.486 6.291

d d

7.722 7.586 8.068

dd t dd

8.083 7.521 7.640 7.665

dd ddd ddd dd

9.343

s br

2.069 8.605

s br s

8.115

s

6.238 2.276 2.564 4.357 3.853 3.502 3.567 5.246 4.883

dd ddd ddd dddd ddd ddd ddd d t

13C

shift

mult.

124.36 126.26 140.24 127.79 130.77 126.94 129.01 129.21 182.00 129.88 126.73 128.32 133.32 123.56 134.68 127.78 126.36

D D S S D D D S S S D D D D S S S

171.02 21.50 152.91 149.89 115.03 155.42 138.33

S Q S S S S S

83.98 36.89

D T

70.82 87.86 61.76

D D T

1H

shift

mult.

8.751 7.885

dd d, br

8.752 8.018 8.708

dd dd dd

8.374 7.661 7.866 8.561

ddd ddd ddd dddd

2.119

s, br

10.65 6.10 6.357 2.152 3.170 4.500 3.937 3.610 3.715 4.85 4.55

s, brd s, br dd ddd ddd ddd ddd dd dd v br v br

a 1H shifts relative to TMS by iterative spectrum simulation; 13C shifts in italics denote unconfirmed assignments. b For atom numbering, see Scheme 2; for CH2 groups, protons a and b correspond to the lower and higher chemical shift, respectively; the conventional stereochemical notation 2′ and 2′′ for deoxyribose protons denotes the syn and anti positions, respectively, relative to the attached purine. c Lower case symbols represent multiplicities (mult.) due to resolved long-range C-H couplings in the 1H-coupled spectrum. d Measured at 60 °C.

of dG-N-Ac-ABA-5 (5) at 110 °C agreed well with the data presented by Takamura-Enya and co-workers (33) for the

nonacetylated adduct N-(2′-deoxyguanosin-8-yl)-3-aminobenzanthrone (temperature not specified).

1062


Table 4. 1H Couplings for dA-N-Ac-ABA-1 (Structure 9, Scheme 2) and dG-N-Ac-ABA-5 (Structure 5, Scheme 2) JHH in Hza coupled spins

dA-N-Ac-ABA-1 (9) (in DMSO-d6, 40 °C)

dG-N-Ac-ABA-5 (5) (in DMSO-d6, 110 °C)

1,2 1,11 4,5 4,6 5,6 8,9 8,10 8,11 9,10 9,11 10,11 1′,2′a 1′,2′b 2′a,2′b 2′a,3′ 2′b,3′ 3′,4′ 3′,3′-OH 4′,5′a 4′,5′b 5′a,5′b 5′a,5′-OH 5′b,5′-OH

3.34 NDb 7.85 1.30 7.70 7.68 1.54 NDb 7.36 0.99 8.02 6.24 7.39 -13.27 3.30 5.81 2.91 3.94 4.39 4.67 -11.80 5.34 4.54

8.04 0.43 8.39 1.26 7.28 7.87 1.51 0.53 7.17 1.06 8.12 6.59 7.74 -13.09 2.95 6.48 3.20 NDc 5.07 4.68 -11.75 NDc NDc

a Determined from resolution-enhanced 500 MHz spectra by iterative spectrum simulation (SD < 0.01 Hz); ND, not determined. b Coupling less than the line width. c Coupling removed by rapid -OH exchange with residual H2O.

At 30 °C, the 13C NMR spectrum showed a doubling of nearly all signals, consistent with the presence of two conformers in slow exchange. At 60 °C, single resonances were observed for nearly all carbons. The sample was stable for several days so that 13C NMR measurements and 2D C-H correlation experiments could be performed, leading to assignment of the narrow 13C signals observed for BA carbons 1, 6, and 8-11. Correlations could not be detected for the broad 13C signals for carbons 2, 4, and 5, which were assigned according to their relative line widths by analogy with the line widths for the corresponding assigned 1H signals (Figure 1). At 110 °C, all carbon signals could be resolved (Table 3), but there was substantial decomposition of the adduct after 14 h, and only 1D 13C spectra were obtained (broadband 1Hdecoupled and DEPT-135). Protonated carbons were assigned using the results obtained at 60 °C. Quaternary carbons were assigned by analogy with reference data for dG, 3-ABA, and the adduct dG-N-Ac-ABA-2 (1) (28). Assignments still considered to be tentative are indicated by italics in Table 3. In Table 3, the chemical shifts for the deoxyribose moiety in the dG adduct dG-N-Ac-ABA-5 (5) are similar to those for dG itself and for the dA adduct dA-N-Ac-ABA-1 (9) (see below) with two exceptions: in dG-N-Ac-ABA-5 (5), C2′ is shielded by ca. 3 ppm and H2′b (syn to the purine) is deshielded by ca. 0.6 ppm relative to dA-N-Ac-ABA-1 (9). These relatively strong perturbations were also observed for the dG adduct dG-N-AcABA-2 (1) (28). The phosphorylated adduct dGp-N-Ac-ABA-2 (2) gave a well-resolved 250 MHz 1H spectrum at 25 °C, which was very similar to the spectrum obtained from the nonphosphorylated adduct dG-N-Ac-ABA-2 (1). The only significant difference was the ca. 0.5 ppm increase in the chemical shifts for the deoxyribose protons 2′a and 3′ due to the phosphate group at position 3′. The phosphorylated adduct dGp-N-Ac-ABA-5 (6) gave a broad 1H spectrum

Osborne et al.

at 250 MHz and 25 °C that was consistent with the exchange broadened spectrum of the nonphosphorylated adduct dG-N-Ac-ABA-5 (5) (Figure 1, 30 °C). The presence of the phosphate group at deoxyribose position 3′ in dGp-N-Ac-ABA-5 (6) was confirmed by the 0.5 ppm increase in chemical shift for H3′ relative to its value for the adduct dG-N-Ac-ABA-5 (5). Finally, samples of dG-N-Ac-ABA-2 (1) and dGp-N-AcABA-2 (2) were treated with alkali to give the deacetylated adducts dG-ABA-2 (3) and dGp-ABA-2 (4), which were characterized by their UV-visible and mass spectra. The 1H NMR spectrum of dG-ABA-2 (3) has been published previously (28). Unfortunately, dG-ABA-5 (7) and dGp-ABA-5 (8) could not be made in the same way, because of the lability of these adducts in alkali. Synthesis and Characterization of AdenineContaining Adducts. Reaction of N-Aco-N-Ac-ABA with dA gave one major adduct, dA-N-Ac-ABA-1 (9), in the form of two diastereoisomers with opposite but unequal optical rotations (see Materials and Methods). Thus, we reasoned that the dA adduct could not be the C-C-linked N-acetyl-3-amino-2-(2′-deoxyadenosin-8-yl)benzanthrone or the C-N-linked N-acetyl-3-(2′-deoxyadenosin8-yl)aminobenzanthrone, i.e., the analogues of the dG adducts discussed above. Furthermore, the possible adenine-N6 to BA-C2 linkage was also inconsistent with the isolation of two stereoisomers, and we concluded that the BA moiety was linked to the adenine group by two bonds to give a rigid, asymmetric core structure with inherent optical activity. The 1H NMR spectrum shown in Figure 2 and the 1H and 13C NMR data summarized in Tables 3 and 4 exhibited the following key features: (i) the amino NH and the acetyl group of BA were present, but no NH or NH2 protons from adenine were detected; (ii) both H2 and H8 of adenine were present with approximately their normal chemical shifts; (iii) the ribosyl moiety was intact; and (iv) the BA moiety was significantly altered in the region around C1 and C2 while sp2 carbon sites C3-C11 and their attached protons were largely unaffected. Specifically, the signals for the aromatic sp2 sites C1-H and C2-H and one quaternary sp2 carbon in BA had been lost. Instead, an unusual set of signals appeared as follows: a quasi-olefinic sp2 C-H (C/H shifts of 116.52/ 6.291 ppm), which was assigned to the BA C2 site, an allylic sp3 C-H (60.45/6.486 ppm) assigned as the BA C1 site, and a quaternary sp3 carbon at 67.11 ppm assigned as C11b. The two C-H protons were directly coupled with a small vicinal 3J1,2 coupling of 3.34 Hz and exhibited relatively strong mutual dipolar interaction (Overhauser effects via ROESY 2D), consistent with the location of these protons on adjacent carbons with an HCCH dihedral angle in the range of 60-120°. The proton assigned as H1 showed rotating-frame Overhauser effect (ROE) interactions in the order H11 > A2 ∼ H2 while H2 showed ROEs with H1 ∼ A2 > amide NH. In particular, the adenine A2 proton showed large and equal ROE to the BA protons H1 and H2. Taken together, these results clearly indicated that the aromatic character of the BA ring system had been lost in the vicinity of C1 and C2, specifically that C1 and C11b represented attachment points for the linkage with the adenine moiety. One possibility would be the formation of a three-membered aziridine ring system involving C1, C11b, and the N6 of adenine (34). This structure would have the correct carbon and proton types, consistent with



Figure 2. Expanded regions of the 500 MHz 1H NMR spectrum of dA-N-Ac-ABA-1 (9) in DMSO-d6 at 40 °C. The assignment labels correspond to the numbering in Scheme 2 and Table 3.

Figure 3. Modeled 3D structure (Chem3D, MM2 force field) of the (1S,11bR) diasteromer of dA-N-Ac-ABA-1 (9) is shown in a crossed-eyes stereoscopic view. In the vicinity of the adenine-BA linkage, the BA protons 1, 2, and 11 and the adenine proton A2 have positions that result in a set of short interproton distances (1-2, 1-11, 1-A2, and 2-A2) that are consistent with the observed 2D ROESY cross-peak intensities.

the NMR data, including the absence of a proton at N6. Furthermore, the aziridine ring would be roughly perpendicular to the BA ring system and could be formed by attachment from above or below the BA ring, leading to two diastereomers. However, modeling studies showed that such a structure was not consistent with the ROE and J coupling data mentioned above. For example, the distances between the adenine proton A2 and the BA protons 1 and 2 would be unequal and in the range of 4-6 Å, inconsistent with the observed ROEs, while the H1-H2 dihedral angle would be 23°, requiring a coupling of ca. 8 Hz. The alternative aziridine with linkage at C1 and C2 could be ruled out immediately on the basis of the observed carbon hybridization scheme. A further problem with aziridines would be their conformationally labile structure with nitrogen inversion and rotation of the adenine moiety, which would be expected to lead to

broadening of the NMR spectrum at high field, and this has not been observed. Another possible structure would involve single-bond linkages between adenine N1 and BA-C1 and between N6 and C11b, resulting in the formation of a fivemembered imidazole-like ring system. Modeling with the semiempirical methods AM1 and PM3 indicated that the minimum energy for the five ring structure is on the order of 6-14 kcal/mol lower than for the aziridine. Two well-defined diastereomers can be formed with configuration (1S,11bR), as shown in Figure 3, or (1R,11bS). The interproton distances in the model are A2-H1 ) 2.78 Å, A2-H2 ) 2.65 Å, H1-H2 ) 2.57 Å, H1-H11 ) 2.3 Å, and A2-H11 ) 3.47 Å, which are in excellent agreement with all ROE data. The H1-H2 dihedral angle is 68°, consistent with the small J coupling observed. A structure with the reversed linkage, i.e., N1-C11b and N6-

1064


Figure 4. Autoradiographic profiles of DNA adducts obtained by using the nuclease P1 (upper panels) or butanol (lower panels) enrichment version of the 32P-postlabeling assay. Adduct profiles obtained from salmon testis DNA treated with (A) N-OH-ABA and (B) N-Aco-N-Ac-ABA. (C) Adduct profiles obtained from lung DNA of rats treated with 2 mg of 3-NBA per kg body weight [these profiles are representative of adduct profiles obtained with DNA from other rat tissues including liver, kidney, spleen, heart, and colon (23)].

C1 can be definitely ruled out on the basis of the ROEs for proton A2. A detailed analysis of J couplings, ROEs, and modeling results for the deoxyribose moiety indicates that the deoxyribose ring is preferentially in the 2E or 2 T3 conformation with H1′ and H2′b pseudoaxial and H3′ pseudoequatorial and that the syn conformer for the nucleoside predominates (ROE for A8 to H1′ and H2′b; no ROE from H1′ to A2). Finally, the 3-amino group exhibits the preferred orientation shown in Figure 3 since the NH proton shows a much stronger ROE to H4 than to H2. According to IUPAC fused ring nomenclature (rule FR-4.4), we propose the following name for dA-N-AcABA-1 (9): (9′-(2′′-deoxyribofuranosyl))purino[6′,1′:2,3]imidazo[5,4-p](1,11b-dihydro-(N-acetyl-3-amino))benzanthrone. The phosphorylated adduct dAp-N-Ac-ABA-1 (10) was characterized by 250 MHz 1H NMR at 25 °C. The spectrum could be readily assigned based on the analysis of the spectrum obtained at 500 MHz for the nonphosphorylated adduct dA-N-Ac-ABA-1 (9). The key difference was the 0.4 ppm increase in chemical shift for H3′ in the phosphorylated derivative. Reaction of N-OH-ABA with DNA and Comparison of Products with Nonacetylated Nucleotide Adduct Standards. Salmon testis DNA was reacted in vitro with N-OH-ABA at 60 °C for 18 h under nitrogen and digested to 3′-nucleotides for the 32P-postlabeling assay as described (26). The patterns of adducts obtained by either of the two enrichment versions of the 32Ppostlabeling method (Figure 4A) were the same as those obtained in DNA isolated from 3-NBA-treated rats (Figure 4C) (18, 23). In each case, 3-NBA induced essentially the same DNA adducts consisting of a cluster of four adduct spots on TLC (spots 1-3 and 6) detected after enrichment using nuclease P1 digestion and a cluster of up to five adduct spots (spots 1-5) detected after enrichment using butanol extraction. The same adduct spots were obtained from dAp (spots 1 and 2) and dGp (spots 3-5) after in vitro incubations of the nucleotides with 3-NBA in the presence of XO, a mammalian nitroreductase (Figure 5F,G).

Osborne et al.

As a second independent chromatographic procedure to confirm identities of adduct spots, we employed reversed phase HPLC analysis. When the whole DNA digest was labeled and analyzed by HPLC, the pattern shown in Figure 6A was obtained. The same pattern of adducts was found in DNA of 3-NBA-treated rats (data not shown). By excising areas containing major adduct spots from postlabeled samples after TLC and analyzing the radioactive material on HPLC, singly or in combination, we established that spots 1 and 2 derived from N-OH-ABA were eluted with retention times of 41.0 and 33.0 min, respectively, identical to the retention time of the 3-NBA-derived dAp reference compounds prepared with XO. Similarly, spots 3 (retention time 24.0 min), 4 (retention time 27.5 min), and 5 (retention time 39.0 min), coeluted with the 3-NBA-derived dGp reference compounds prepared with XO (data not shown). We then compared these adducts with standards prepared by deacetylation of the corresponding acetylated adducts, whose structures were known. Aliquots of dGpABA-2 (4) and dAp-ABA-1 (12) were labeled directly and analyzed by 32P-postlabeling. As shown in Figure 5A,D, dGp-ABA-2 (4) and dAp-ABA-1 (12) each produced a single major spot, assigned spots SG1 and SA1, respectively. To determine whether dGp-ABA-2 (4) and dApABA-1 (12) are formed in vivo in rats treated with 3-NBA or in vitro in DNA modified with N-OH-ABA, cochromatographic analyses on TLC plates were performed. Spot SG1 was indistinguishable from spot 3, and spot SA1 was indistinguishable from spot 1 on TLC plates (data not shown). However, when equal amounts of radioactivity of adduct spot SA1 and spot 1 were mixed and analyzed by HPLC, the adducts were well-separated (Figure 7A). Similarly, adduct spot SG1 was partially separated from spot 3 (Figure 7B). These results indicate that neither dGp-ABA-2 (4) nor dAp-ABA-1 (12) is formed in DNA from rats treated with 3-NBA or in DNA modified with N-OH-ABA. It is also clear that two independent chromatographic systems should be used with the 32Ppostlabeling method to compare authentic adduct standards with adducts formed under biologically relevant conditions. Reaction of N-OH-ABA with DNA and Comparison of Products with Nonacetylated Nucleoside Adduct Standards. N-OH-ABA-modified DNA was enzymatically hydrolyzed to the nucleosides, and the hydrolysate was analyzed by HPLC (Figure 8A). Because all four normal nucleosides were eluted within 10 min, peaks appearing after that time probably represent DNA adducts derived from N-OH-ABA. The authentic nucleoside standards dG-ABA-2 (3) and dA-ABA-1 (11), each of which had a retention time 32.0 min in this system, were not found in the DNA hydrolysate, confirming the negative results described above using the 32P-postlabeling assay. As shown in Figure 8A, the most abundant adducts from N-OH-ABA-modified DNA eluted with retention times of 30.0 (assigned peak p1) and 33.5 (assigned peak p2) min, respectively. The positive ion ESI-MS spectra corresponding to these two peaks were similar and consistent with the formula C27H22N6O5 (M ) 510.165) for dG-ABA adducts. Peak p1 gave m/z 395 (int. 100%) [M + H - dR*]+, 475 (15%), 511 (4%) [M + H]+, and 533 (7%) [M + Na]+. Peak p2 gave m/z 395 (100%) [M + H - dR*]+, 475 (14%), 511 (34%) [M + H]+, and 533 (7%) [M + Na]+. The structural identification of



Figure 5. Autoradiographic profiles obtained from adduct standards (A) dGp-ABA-2 (4), (B) dGp-N-Ac-ABA-2 (2), (C) dGp-N-AcABA-5 (6), (D) dAp-ABA-1 (12), and (E) dAp-N-Ac-ABA-1 (10). Autoradiographic profiles of DNA adducts obtained after incubation of (F) dGp and (G) dAp with 3-NBA after activation with XO. The butanol enrichment procedure of the 32P-postlabeling method was used.

Figure 6. Separation of 32P-labeled nucleoside 3′,5′-bisphosphate adducts on a phenyl-modified reversed phase column (system b; 23) by using the nuclease P1 (upper panels) or butanol (lower panels) enrichment of the 32P-postlabeling assay. The origin of TLC after D1 only was excised and extracted from the plates, dissolved, and injected on HPLC. Origins were from DNA digests of salmon testis DNA treated with (A) N-OH-ABA and (B) N-Aco-N-Ac-ABA.

the adducts corresponding to peak p1 and p2 awaits further investigation. Reaction of N-Aco-N-Ac-ABA with DNA and Comparison of the Products with Acetylated Nucleotide Adduct Standards. Salmon testis DNA was reacted in vitro with N-Aco-N-Ac-ABA at 60 °C for 18 h under nitrogen, digested with enzymes to 3′-nucleotides, and analyzed by 32P-postlabeling. The DNA adduct pattern on TLC consisted of a cluster of up to four adducts spots, assigned spots Ac1-Ac4 (Figure 4B). Adduct spot Ac1 was the predominant adduct formed, and adduct spot Ac4 was only detectable after enrichment by butanol extraction (Figure 4Bb). Areas containing major adduct spots from postlabeled samples after TLC separation, or from the origin after D1 only, were excised and extracted from the TLC plate, dissolved, and analyzed by HPLC. Several peaks were detected and identified by cochromatographic analysis with individual adduct spots (Figure 6B). Eluates of adduct spots Ac1, Ac2, and Ac3 eluted with retention times of 23.5, 38.5, and 23.5 min, respectively; adduct spot Ac4 could not be extracted from the TLC plate in sufficient amount for cochromatographic analysis by HPLC. Samples of adduct standards dGp-N-Ac-ABA-2 (2), dGp-N-Ac-ABA-5 (6), and dAp-N-Ac-ABA-1 (10) were 32Plabeled directly and analyzed by TLC. As shown in Figure 5B,C,E, each of these standards produced a single spot, assigned spots SG2, SG3, and SA2, respectively. Spot

SA2 came at the same position as spot SA1 for the nonacetylated adduct dAp-ABA-1 (12). However, cochromatographic analysis on HPLC proved, as expected, that spot SA1 (retention time 32.0 min) and spot SA2 (retention time 38.5 min) were different compounds (data not shown). Spot SA2 migrated outside the region of interest where N-Aco-N-Ac-ABA-derived DNA adducts were located. Digests of DNA modified with N-Aco-N-Ac-ABA were spiked with aliquots of dGp-N-Ac-ABA-2 (2) and dGp-NAc-ABA-5 (6) and were analyzed using either the nuclease P1 enrichment or the butanol extraction procedure of the 32P-postlabeling assay. Spot Ac1 (Figure 4B) and spot SG2 were not distinguishable on TLC plates (data not shown), or by HPLC, where a single peak eluting with a retention time of 23.5 min was observed (data not shown). Thus, the results confirmed the findings obtained by TLC analysis demonstrating that spot Ac1 was derived from dGp-N-Ac-ABA-2 (2). Spot Ac4 (Figure 4Bb) and spot SG3 were indistinguishable by TLC analysis (data not shown), indicating that spot Ac4 was derived from dGp-N-Ac-ABA-5 (6), however, as explained above, cochromatographic analysis of this spot could not be performed on HPLC. Reaction of N-Aco-N-Ac-ABA with DNA and Comparison of the Products with Acetylated Nucleoside Adduct Standards. N-Aco-N-Ac-ABA-modified DNA was enzymatically hydrolyzed to the corresponding nu-

1066


Osborne et al.

Figure 7. Separation of 32P-labeled nucleoside 3′,5′-bisphosphate adducts on a phenyl-modified reversed phase column (23). Individual adducts were excised and extracted from the TLC plates, dissolved, and injected on HPLC. (Aa) Spot 1 (Figure 4A) from salmon testis DNA treated with N-OH-ABA. (Ab) Spot SA1 (Figure 5D). (Ac) Equal amounts of spot 1 (Figure 4A) from salmon testis DNA treated with N-OH-ABA and spot SA1 (Figure 5D). (Ad) Equal amounts of spot 1 (Figure 5G) obtained after incubation of dAp with 3-NBA after activation with XO and spot SA1 (Figure 5D). (Ae) Equal amounts of spot 1 (Figure 4C) obtained from lung DNA from 3-NBA-treated rats and spot SA1 (Figure 5D). (Ba) Spot 3 (Figure 4A) from salmon testis DNA treated with N-OH-ABA. (Bb) Spot SG1 (Figure 5A). (Bc) Equal amounts of spot 3 (Figure 4A) from salmon testis DNA treated with N-OH-ABA and spot SG1 (Figure 5A). (Bd) Equal amounts of spot 3 (Figure 5F) obtained after incubation of dGp with 3-NBA after activation with XO and spot SG1 (Figure 5A). (Be) Equal amounts of spot 3 (Figure 4C) obtained from lung DNA from 3-NBA-treated rats and spot SG1 (Figure 5A).

cleosides, and the DNA hydrolysate was analyzed by HPLC. Several peaks (peaks p3-p7) were observed in the HPLC chromatogram shown in Figure 8B. All four normal nucleosides were eluted within 10 min. Peak p6 (retention time, 27.0 min) was the major adduct found and had the same retention time and UV absorption spectrum as those of dG-N-Ac-ABA-2 (1). Peak p4 (reten-

tion time, 25.5 min) had the same retention time and UV spectrum as dA-N-Ac-ABA-1 (9). Using cochromatographic analysis and the UV spectrum in Table 1, the adduct corresponding to peak p5 (retention time, 26.0 min) was identified as dG-N-Ac-ABA-1, whose structure has not yet been determined. Similarly, the adduct corresponding to peak p7 (retention time, 31.0 min) was



dAp-N-Ac-ABA-1 (10) was not recovered under the conditions normally used in the enrichment procedures. It is also clear from Table 5 that both nonacetylated standards, dGp-ABA-2 (4) and dAp-ABA-1 (12), were recovered only with poor efficiency (average recovery 0.42.7%); the limit of detection was one adduct in 106 nucleotides.

Discussion

Figure 8. HPLC analysis of the enzymatic digests of salmon testis DNA modified with (A) N-OH-ABA and (B) N-Aco-N-AcABA. The charts show the elution of nucleosides from an ODS column, as monitored by UV absorption at 254 nm. The normal nucleosides were eluted at 5-10 min (not shown). The principal adduct peaks for N-Aco-N-Ac-ABA-modified DNA are as follows: p3, unidentified adduct, probably a ring-opened guanine derivative; p4, dA-N-Ac-ABA-1a; p5, dG-N-Ac-ABA-1; p6, dGN-Ac-ABA-2 (1); and p7, dG-N-Ac-ABA-5 (5).

identified as dG-N-Ac-ABA-5 (5). The adduct corresponding to peak p3 awaits further investigation. Recoveries of Acetylated and Nonacetylated Adduct Standards by 32P-Postlabeling. The acetylated nucleotide adduct standards dGp-N-Ac-ABA-2 (2), dGpN-Ac-ABA-5 (6), and dAp-N-Ac-ABA-1 (10) and the nonacetylated standards dGp-ABA-2 (4) and dAp-ABA-1 (12) were added as their tetrabutylammonium salts at different concentrations to calf thymus DNA and tested for their recoveries in the 32P-postlabeling assay. Adduct recoveries were determined with both enrichment procedures of the 32P-postlabeling assay. In a typical experiment, 4 µg of DNA (12.3 nmol nucleotides) was used. To create adduct levels commonly observed in in vivo samples, adduct standards were added to DNA so that ratios in a range between 1/105 (123 fmol standard) and 1/108 (0.123 fmol standard) adduct/normal nucleotides were achieved. Then, the mixture was digested, enriched, labeled, separated by TLC, and analyzed. Recoveries (in %) are shown in Table 5. Of the five standards investigated, dGp-N-Ac-ABA-5 (6) showed the best recovery of all five standards, with an average recovery of around 15-20% over a 100-fold range of adduct concentrations using enrichment with butanol extraction; the limit of detection was one adduct in 107 nucleotides. However, dGp-N-Ac-ABA-5 (6) adduct was only barely detectable after enrichment with nuclease P1 digestion (limit of detection was one adduct in 105 nucleotides), which is indicative of the C8-dG nature of this adduct (35). The standard dGp-N-Ac-ABA-2 (2) showed an average recovery of 1-10% over a 100-fold range of adduct concentrations using both enrichment procedures, nuclease P1 digestion, and butanol extraction; the limit of detection was one adduct in 107 nucleotides. In contrast, adduct

The 32P-postlabeling assay is an ultrasensitive method for detecting DNA adducts that is applicable to a very wide range of DNA lesions. It requires only microgram amounts of DNA and is capable of detecting some types of DNA adducts at levels as low as one adduct in 1010 normal nucleotides in this amount of material (36-38). There are several advantages of the 32P-postlabeling assay over other methods. It does not require the use of radiolabeled test compounds and is applicable to a wide range of chemicals and types of DNA lesions. Prior structural characterization of adducts is not required, although some assumptions about their likely chromatographic properties may be necessary. Thus, the 32Ppostlabeling assay is widely used in human biomonitoring (39). Several human biomonitoring studies using the detection of DNA adducts by the ultrasensitive 32Ppostlabeling method have reported higher levels of bulky DNA adducts among subjects heavily exposed to diesel exhaust (40, 41). This may be predictive of cancer risk. We found that 3-NBA forms multiple characteristic DNA adducts in vivo in 3-NBA-treated rodents (16, 18, 23, 27). However, the method is not without its limitations (36). DNA lesions that are not chemically stable as mononucleotides will not be detected reliably. The method does not provide structural information of adducts, and identification of adducts often relies on demonstrating their cochromatography with characterized synthetic standards. Such standards can provide the means for determining the efficiency of labeling and detection, whereas in the absence of standards adduct levels may be underestimated (42). Nitro-PAHs require metabolism to reactive electrophilic species to exert their genotoxic activity (3, 4). The covalent binding of these nitro-PAH metabolites to cellular DNA has been well-recognized to play an important role in the initiation of mutagenesis and carcinogenesis (4, 43). Like other nitro-PAHs, 3-NBA forms specific DNA adducts in various in vitro bioassays, in cells, and in vivo in rodents (16-18, 20, 22-31). The formation of reactive acetate or sulfate esters of N-OH-ABA catalyzed by N,Oacetyltransferases (NATs) or sulfotransferases (SULTs) after nitroreduction of 3-NBA seems to be the major pathway of bioactivation leading to the formation of covalent DNA adducts (Scheme 1). Because it was unclear whether some of these adducts were acetylated at the amino group, we prepared adducts derived from both 3-ABA and 3-acetylaminobenzanthrone (3-Ac-ABA). The most prominent adduct obtained by reacting N-Aco-N-Ac-ABA with dG, and a major adduct found in N-Aco-N-Ac-ABA-treated DNA, was dG-N-Ac-ABA-5 (5), in which the 8-position of guanine is linked to the N3 nitrogen atom of 3-Ac-ABA. This was to be expected, as treatment of DNA with other arylacetamides gives 8-substituted guanine adducts (43, 44). Moreover, dGpN-Ac-ABA-5 (5) is sensitive to digestion with nuclease P1 (Table 5), which is also indicative of C8-dG arylamine-

1068


Osborne et al.

Table 5. Recoveries (%) of the Synthesized Adduct Standards under Typical

32P-Postlabeling

Conditionsa

recovery (%) 32P-postlabeling

assay nuclease P1

butanol

adduct level (adduct/ normal nucleotides)

dGp-ABA-2 (4)

dGp-N-Ac-ABA-2 (2)

dGp-N-Ac-ABA-5 (6)

dAp-ABA-1 (12)

dAp-N-Ac-ABA-1 (10)

1/105 1/106 1/107 1/108 1/105 1/106 1/107 1/108

0.36 (0.17-0.63) 0.64 (0.57-0.70) ND ND 2.4 (1.7-2.9) 2.3 (1.8-3.2) ND ND

7.5 (4.6-11.1) 5.5 (1.7-8.6) 0.72 (0.70-0.74) ND 8.1 (5.8-9.9) 11.2 (4.3-20.5) 7.7 (6.9-8.5) ND

Synthesis, Characterization, and 32P-Postlabeling ... - ACS Publications

Recommend Documents