DNA Adduct Formation from Acrylamide via ... - ACS Publications

Acrylamide (AA) is a high production volume chemical with many industrial ... with GA produced 5-7-fold higher whole body DNA adduct levels than with ...
0 downloads 0 Views 158KB Size
Download PDF
1328

Chem. Res. Toxicol. 2003, 16, 1328-1337

DNA Adduct Formation from Acrylamide via Conversion To Glycidamide in Adult and Neonatal Mice Gonc¸ alo Gamboa da Costa,†,‡ Mona I. Churchwell,† L. Patrice Hamilton,† Linda S. Von Tungeln,† Frederick A. Beland,† M. Matilde Marques,‡ and Daniel R. Doerge*,† Division of Biochemical Toxicology, National Center for Toxicological Research, Jefferson, Arkansas 72079, and Centro de Quı´mica Estrutural, Complexo I, Instituto Superior Te´ cnico, Av. Rovisco Pais, 1049-001, Lisboa, Portugal Received May 28, 2003

Acrylamide (AA) is a high production volume chemical with many industrial uses; however, recent findings of ppm levels in starchy foods cooked at high temperature have refocused worldwide attention on the neurotoxicity, germ cell mutagenicity, and carcinogenicity of AA. Oxidative metabolism of AA to its epoxide metabolite, glycidamide (GA), has been observed in experimental animals and humans and may be associated with many of the toxic effects of AA exposure, including formation of N7-(2-carbamoyl-2-hydroxyethyl)guanine (N7-GA-Gua) in vivo. This paper describes the characterization of two new GA-derived DNA adducts formed in vitro, N3-(2-carbamoyl-2-hydroxyethyl)adenine (N3-GA-Ade) and N1-(2-carboxy-2-hydroxyethyl)-2′deoxyadenosine. A sensitive method for quantification of N7-GA-Gua and N3-GA-Ade, based on LC with tandem mass spectrometry and isotope dilution, was developed and validated for use in measuring DNA adduct formation in selected tissues of adult and whole body DNA of 3 day old neonatal mice treated with AA and GA. In adult mice, DNA adduct formation was observed in liver, lung, and kidney with levels of N7-GA-Gua around 2000 adducts/108 nucleotides and N3-GA-Ade around 20 adducts/108 nucleotides. Adduct levels were modestly higher in adult mice dosed with GA as opposed to AA; however, treatment of neonatal mice with GA produced 5-7-fold higher whole body DNA adduct levels than with AA, presumably reflective of lower oxidative enzyme activity in newborn mice. DNA adduct formation from AA treatment in adult mice showed a supralinear dose-response relationship, consistent with saturation of oxidative metabolism at higher doses. These results increase our understanding of the mutagenic potential of GA and provide further evidence for a genotoxic mechanism in AA carcinogenesis.

Introduction AA1 (Figure 1) is a high production volume chemical (>200 Gg/year worldwide), whose polymeric forms are widely used in water treatment, crude oil processing, pulp paper processing, concrete, and grouts (1, 2). Monomeric AA is used extensively in biochemical applications for PAGE analysis of proteins; it is also a component of cigarette smoke (1-2 µg/cigarette) (3). More recently, AA has been measured in baked and fried starchy foods, notably french fries (up to 0.7 ppm), potato chips (up to 3.9 ppm), and bread (up to 1.9 ppm) (4, 5). Further studies have shown that Maillard browning reaction * To whom correspondence should be addressed. Tel: (870)543-7943. Fax: (870)543-7720. E-mail: [email protected]. † National Center for Toxicological Research. ‡ Instituto Superior Te ´ cnico. 1 Abbreviations: AA, acrylamide; Bis-Tris, bis(2-hydroxyethyl)iminotris(hydroxymethyl)methane; CNL, constant neutral loss; COSY, correlation spectroscopy; dA, 2′-deoxyadenosine; dG, 2′-deoxyguanosine; dR, 2′-deoxyribosyl; DEPT, distortionless enhancement by polarization transfer; ES, electrospray ionization; GA, glycidamide; LOD, limit of detection (S/N ) 3); LOQ, limit of quantitation (S/N ) 10); MS/MS, tandem mass spectrometry; MRM, multiple reaction monitoring; N3GA-Ade, N3-(2-carbamoyl-2-hydroxyethyl)adenine; N7-GA-Gua, N7(2-carbamoyl-2-hydroxyethyl)guanine; N1-GA-dA, N1-(2-carboxy-2hydroxyethyl)-2′-deoxyadenosine; N6-GA-dA, N6-(2-carboxy-2-hydroxyethyl)-2′-deoxyadenosine; RSD, relative standard deviation (SD/mean × 100); S/N, signal-to-noise ratio.

Figure 1. Structures of AA and GA.

products derived from glucose and asparagine, a major amino acid present in potatoes and cereals, are responsible for the formation of AA in these foods (6-8). The classification of AA as a probable human carcinogen (2), based on findings of rodent carcinogenicity in several organs (9, 10), underscores the significance of these exposures. Although some suggestions have been made for hormonally mediated mechanisms of AA carcinogenicity in the rat (11), evidence for a genotoxic mechanism involving GA is more consistent. While AA is clastogenic and mutagenic in vivo, evidence for direct genotoxicity mediated by AA is limited. AA is nonmutagenic in Salmonella assays either in the presence or in the absence of a microsomal activating system; however, GA (Figure 1), the reactive epoxide metabolite of AA, is mutagenic, either with or without metabolic activation (12). Although the rate of reaction of AA with DNA bases in vitro is quite slow (13), the corresponding reaction of GA with DNA is

10.1021/tx034108e CCC: $25.00 © 2003 American Chemical Society Published on Web 09/12/2003

Acrylamide-Derived DNA Adducts in the Mouse

Figure 2. Structures of the GA-derived DNA adducts identified in this study.

much faster (14). Segerba¨ck et al. (15) previously characterized a depurinating DNA adduct, N7-GA-Gua, from the in vitro reaction of DNA with GA and detected this adduct in rodent tissues following a single intraperitoneal injection of [14C]-labeled AA. Studies in both rodents and humans have demonstrated the metabolism of AA to GA (16-18). Furthermore, the comparable induction of micronuclei in rodents treated with either GA or AA provided evidence that GA is the predominant genotoxin derived from AA (19). In light of the widespread exposure to AA, we have initiated studies to assess mechanisms of AA carcinogenicity. In this paper, we describe the synthesis and spectroscopic characterization of DNA adducts from reaction of GA with individual deoxynucleosides. In addition to the previously described depurinating adduct, N7-GA-Gua (15), a new depurinating adduct, N3-GA-Ade, and a novel 2′-deoxynucleoside adduct, N1-GA-dA, were detected from reactions of GA with deoxynucleosides and DNA (Figure 2). The synthesis of analogous 15N-labeled and unlabeled adducts provided the basis for a quantitative analytical method, based on LC with MS/MS and isotope dilution, that was used to quantify the major adducts derived from reaction of GA with DNA in vitro and for the analysis of depurinating DNA adducts formed in adult and neonatal mice following administration of AA or GA.

Experimental Section Reagents. Sigma Chemical Co. (St. Louis, MO) supplied the buffer reagents, deoxynucleosides, and DNA used in this study. AA was supplied by ICN (Cleveland, OH). 15N5-Labeled dA (>98 atom % 15N) was obtained from Cambridge Isotope Laboratories (Andover, MA). 15N5-Labeled GMP (>98 atom % 15N) was purchased from Spectra Stable Isotopes (Columbia, MD). GA was synthesized by epoxidation of acrylonitrile under basic conditions (20) and characterized by 1H NMR spectroscopy; the purity was assessed at approximately 98% using LC-UV (205 nm) and LC/ES/MS/MS (full scans).

Chem. Res. Toxicol., Vol. 16, No. 10, 2003 1329 General Instrumentation. HPLC-UV analyses and separations of synthetic products were conducted with a Phenomenex Luna-C18 (2) column (4.6 mm × 250 mm, 5 µm, Phenomenex Co., Torrance, CA), using a Waters system (Waters Associates, Milford, MA) consisting of two model 510 pumps and a model 660 automated gradient controller, equipped with a Rheodyne model 7125 injector (Rheodyne, Cotati, CA), and a HewlettPackard 1050 diode array spectrophotometric detector (HewlettPackard Co., Palo Alto, CA). The peaks were monitored at 205 nm (GA) or 254 and 280 nm (DNA adducts). UV spectra were recorded on a Beckman DU-40 UV-vis spectrophotometer (Beckman Coulter, Fullerton, CA). 1H NMR spectra were obtained on a Varian Gemini 300 spectrometer (Varian, Inc., Palo Alto, CA) or a Varian Unity 300 spectrometer (Varian Deutschland Gmbh, Darmstadt, Germany), both operating at 300 MHz. The samples were introduced in 5 mm Shigemi microsample tubes (Shigemi, Inc., Allison Park, PA), susceptibility-matched to either methyl sulfoxide-d6 or methanol-d4. Labile protons were assigned following chemical exchange with D2O. Whenever necessary, two-dimensional 1H-1H COSY sequences were run to establish proton coupling patterns. 13C NMR spectra were run on the Varian Unity 300 instrument, operating at 75.4 MHz. Assignment of the guanine resonances in N7-GA-Gua was based on comparison with published data for purines, methylated purines, and methylated purine nucleosides (21), combined with direct identification of C8 by running a DEPT pulse sequence. Chemical shifts are reported in ppm downfield from tetramethylsilane, and coupling constants are reported in Hz. LC/MS Analyses. 1. Liquid Chromatography. The liquid handling system consisted of an Alliance HT pump (Waters Associates), an automated switching valve (TPMV, Rheodyne), and a second quaternary gradient pump (GP40, Dionex Co., Sunnyvale, CA). The analytical column (Luna C18-2, 2 mm × 150 mm, 3 µm, Phenomenex) was eluted at 200 µL/min with 2% acetonitrile in water, and the entire effluent was directed into the mass spectrometer. The effluent from the initial 4.1 min of each elution was diverted to waste, while the flow of mobile phase into the mass spectrometer was maintained by the second LC pump. The mobile phase was switched back to waste at the end of the run (total run time of 8.1 min). 2. Mass Spectrometry. A Quattro Ultima triple stage quadrupole mass spectrometer (Micromass, Manchester, U.K.), equipped with an ES source, was used with an ion source temperature of 120 °C, a desolvation gas temperature of 400 °C, and a constant cone voltage of 35 V. For MS/MS measurements, a collision cell gas pressure (Ar) of (2-4) × 10-3 mbar was used to acquire positive ions in either the CNL, product ion, or MRM modes. The MRM transitions monitored and the respective collision energies were as follows: N3-GA-Ade, m/z 223f178 (25 eV); [15N5]N3-GA-Ade (internal standard), m/z 228f183 (25 eV); N7-GA-Gua, m/z 239f152 (19 eV); [15N5]N7GA-Gua (internal standard), m/z 244f157 (19 eV); N1-GA-dA, m/z 340f224 (20 eV); [15N5]N1-GA-dA (internal standard), m/z 345f229 (20 eV); and N6-GA-dA, m/z 340f224 (20 eV); [15N5]N6-GA-dA (internal standard), m/z 345f229 (20 eV). Preparation and Quantification of Adduct Standards. 1. Adducts from Reaction with dA. dA (10 mg, 39.8 µmol) was reacted with a 17-fold molar excess of GA (60 mg) in 5 mL of 5 mM Bis-Tris and 0.1 mM EDTA buffer (pH 7.1), at 37 °C for 2 days. The reaction products were purified by preparative HPLC using a 30 min linear gradient of 0-10% acetonitrile in 10 mM ammonium formate (pH 7.0). Two adducts were isolated and characterized. 1.1. N1-GA-dA. Retention time, 13.1 min; λ max (100 mM TrisHCl, pH 7.4), 258 nm. 1H NMR (methanol-d4): δ 8.23 (1H, s, H2/H8), 8.10 (1H, s, H8/H2), 6.12 (1H, t, J ) 6.0, H1′), 4.29 (1H, d, Jgem ) 14.3, N1CHcHd), 4.22 (1H, m, H3′), 4.06 (1H, d, Jvic ) 7.9, CHOH), 3.98 (1H, dd, Jgem ) 14.3, Jvic ) 7.9, N1CHcHd), 3.69 (1H, m, H4′), 3.46 (1H, dd, Jgem ) 11.8, Jvic ) 3.4, H5′/ H5′′), 3.38 (1H, dd, Jgem ) 11.8, Jvic ) 3.9, H5′′/H5′), 2.40 (1H, m, H2′), 2.17 (1H, m, H2′′). MS(ES): m/z 340 [(M + H)+]. ES/

1330 Chem. Res. Toxicol., Vol. 16, No. 10, 2003 MS/MS (m/z 340): 224 [(MH2 - dR)+], 178 [(M - dR - CO2)+], 136 [(Ade + H)+]. 1.2. N3-GA-Ade. Retention time, 15.0 min; λ max (10 mM ammonium formate, pH 7.0), 274 nm. 1H NMR (DMSO-d6): δ 8.18 (1H, s, H2), 7.87 (2H, bs, N6H2), 7.76 (1H, s, H8), 7.46 (1H, s, CONHaHb), 7.36 (1H, s, CONHaHb), 6.29 (1H, bs, OH), 4.72 (1H, dd, Jgem ) 13.4, Jvic ) 3.1, N3CHcHd), 4.41 (1H, dd, Jvic ) 9.3, J′vic ) 3.1, CHOH), 4.19 (1H, m, CHcHd). MS (ES): m/z 223 [(M + H)+]. ES/MS/MS (m/z 223): 206 [(MH - NH3)+], 178 [(MH - HCONH2)+], 136 [(Ade + H)+], 88 [(MH-Ade)+]. 2. Adduct from Reaction with dG. dG (100 mg, 374 µmol) was dissolved in 20 mL of water; an approximately 10-fold molar excess of GA (300 mg) was added, and the mixture was incubated for 4 days at 37 °C. The crystals that separated from the solution were briefly washed with boiling water and dried to yield a single adduct. 2.1. N7-GA-Gua. η, 29%; retention time (2.5% acetonitrile in 10 mM ammonium formate, pH 7.0), 7.7 min; λmax (10 mM ammonium formate, pH 7.0), 285 nm ( 7.1 mM-1 cm-1). 1H NMR (DMSO-d6): δ 10.84 (1H, s, N1H), 7.80 (1H, s, H8), 7.35 (1H, s, CONHaHb), 7.33 (1H, s, CONHaHb), 6.13 (2H, s, N2H2), 5.94 (1H, d, J ) 6.0, OH), 4.57 (1H, dd, Jgem ) 13.2, Jvic ) 3.0, N7CHcHd), 4.25 (1H, m, CHOH), 4.15 (1H, dd, Jgem ) 13.2, Jvic ) 9.2, CHcHd). 13C NMR (DMSO-d6): 173.50 (CONH2), 159.95 (guaC6), 154.72 (guaC2), 152.55 (guaC4), 143.98 (guaC8), 108.18 (guaC5), 70.28 (CHOH), 49.58 (N7CH2). MS (ES): m/z 239 [(M + H)+]. ES/MS/MS (m/z 239): 194 [(MH - HCONH2)+], 152 [(Gua + H)+]. 3. [15N]-Labeled Adduct Standards. 15N-Labeled adduct standards were prepared essentially as described for the nonlabeled adducts. Briefly, [15N5]dA (2.2 mg) was reacted with GA (14.9 mg) in 1 mL of water for 2 days at 37 °C. The [15N5]N1GA-dA and [15N5]N3-GA-Ade adducts were then isolated by preparative HPLC. Similarly, [15N5]N7-GA-Gua was synthesized from reaction of 15N5-labeled GMP (10 mg) with GA (40 mg) in 1 mL of water for 4 days at 37 °C and the adduct crystals were collected by filtration from the reaction mixture. 4. Assessment of Purity and Adduct Quantification. The chemical purity of each adduct was confirmed by HPLC-UV, with monitoring at both 254 and 280 nm, and by full scan positive ion ES/MS (m/z 100-600) detection. The isotopic distributions of [15N]-labeled adducts were determined using full scan LC-ES/MS (m/z 100-600). The content of unlabeled adducts present in each of the labeled analogues was 5 different amounts of unlabeled (1-5000 fmol) and a constant amount of labeled adduct (50-100 fmol). The plots were linear in all cases, with an r2 of at least 0.998. The MS response factor for each 15N /15N isotopomeric adduct pair was determined from the 5 0 respective slope of the calibration curve. The adduct standards were quantified spectrophotometrically on the basis of molar extinction coefficients of either the isolated unlabeled adduct (N7-GA-Gua) or the closely related structures. N3-GA-Ade and its labeled analogue were assumed to have molar extinction coefficients identical to the one determined experimentally for 3-methyladenine in 10 mM ammonium formate, pH 7.0 (273 nm ) 12.1 mM-1 cm-1), and the N1-GA-dA standards were quantified in 100 mM HCl using the molar extinction coefficient (257 nm ) 13.7 mM-1 cm-1 at pH 1) reported for 1-methyladenine (22). The molar extinction coefficient for N7-GA-Gua was measured experimentally in 10 mM ammonium formate, pH 7.0 (285 nm ) 7.1 mM-1 cm-1) and found to be comparable to the value (7.4 mM-1 cm-1) reported for 7-methylguanine (22). Modification of DNA with GA in Vitro. Salmon testis DNA was dissolved in 5 mM Bis-Tris and 0.1 mM EDTA (pH 7.1), and serial dilutions of a GA solution (100 mg/mL) in the same buffer were added to different aliquots; the resulting

Gamboa da Costa et al. solutions (1 ng to 1 mg GA per mg DNA in 1 mL) were incubated at 37 °C overnight. The GA-modified DNA was precipitated with 0.1 vol of 5 M NaCl and 3 vol of ice-cold ethanol, centrifuged, washed with 3 vol of ice-cold 70% ethanol, and redissolved in 5 mM Bis-Tris and 0.1 mM EDTA (pH 7.1) at a concentration of approximately 1 mg/mL, as determined by UV spectroscopy. The depurinating adducts were released from DNA by neutral thermal hydrolysis. Following optimization of the procedure to maximize adduct yields, aliquots of the GA-modified DNA solutions (100 µL) were routinely heated at 100 °C for 10 min in a DNA thermal cycler (model 480, Perkin-Elmer Co., Branchburg, NJ). The samples were then cooled to room temperature, and an aliquot of the [15N]-labeled adduct standards was added to each sample. The resulting sample was eluted through an Amicon Microcon 3 kDa molecular mass cutoff centrifugal size exclusion column (Millipore Co., Bedford, MA), which had been prewashed with water, by using centrifugation at 12 000g for 45 min at 21 °C. The adducts were subsequently quantified by LC/MS/MS. The deoxynucleoside adducts were also quantified by LC/MS/ MS following standard enzymatic hydrolysis of the DNA to deoxynucleosides (23). The hydrolysis mixtures from 100 µg aliquots of the GA-modified DNA were analyzed directly, without any adduct enrichment procedure. Thermal Lability Kinetics of the Depurinating Adducts. The thermal lability of the depurinating adducts was determined using a salmon testis DNA sample that had been reacted with GA at a level of 100 µg GA/mg DNA. The freshly precipitated GA-modified DNA was redissolved in 5 mM BisTris and 0.1 mM EDTA (pH 7.1; 1 mg/mL), and the solution was incubated at either 21 or 37 °C for up to 48 h. At each time point, 100 µL aliquots were removed, the DNA was precipitated, washed with ice-cold 70% ethanol, and subjected to thermal neutral hydrolysis as described above, and the released adducts were quantified by LC/MS/MS. Animal Handling Procedures and Isolation of in Vivo Modified DNA. Procedures involving care and handling of mice were reviewed and approved by the NCTR Laboratory Animal Care and Use Committee. All mice were obtained from the NCTR colony, and the basal diet was autoclaved NIH-31 (AA content estimated at approximately 0.1 µg/g by LC/MS/MS, D.R. Doerge, unpublished). Adult male C3H/HeNMTV mice (28.1 ( 1.1 g), adult female C57B1/CN mice (25.5 ( 3.3 g), and 3 day old B6C3F1 mice (approximately 2 g) were treated with a single intraperitoneal dose of an aqueous solution containing either AA or GA (50 mg/kg; 13 mg/mL); the injection volumes were 100 µL for the adult animals (median weight assumed to be 26 g for males and females) and 5 µL for the neonatal mice. Additional adult male mice were treated with 0, 1, or 10 mg/kg of AA, using the same volume of appropriately diluted solutions of AA. Six hours after dosing, the mice were sacrificed by exposure to gaseous carbon dioxide; the liver, lungs, and kidneys were removed from the adults, frozen immediately on dry ice, and stored at -80 °C until DNA extraction was performed. DNA was prepared from liver nuclei and from whole lung and kidney tissue homogenates. For neonatal mice, the whole body was cooled in liquid nitrogen and powdered for preparation of total DNA. The DNA isolation was performed, with comparable results, by a chloroform/phenol extraction procedure (24) or use of a Blood & Cell Culture Maxi kit (Qiagen Co., Valencia, CA), with minor modifications from the standard procedure. The DNA hydrolysis and adduct analysis procedures were conducted as described above for in vitro-modified DNA. Statistical Analyses. Statistical analyses of N7-GA-Gua and N3-GA-Ade adduct levels were conducted by ANOVA. When necessary, the data were ln transformed before the analysis to maintain homogeneous variances, a normal data distribution, or both. In instances where the transformation failed to give an equal variance or a normal distribution, analyses were performed by Kruskal-Wallis ANOVA on ranks. Pairwise comparisons were conducted by Student-Newman-Keuls procedure or Dunn’s method. P values less than 0.05 were consid-

Acrylamide-Derived DNA Adducts in the Mouse

Chem. Res. Toxicol., Vol. 16, No. 10, 2003 1331

Figure 3. Electrospray positive product ion mass spectrum of N3-GA-Ade. The major fragments derived from the protonated molecule are outlined in the structure. ered significant. Unless specifically noted, the data are reported as the mean ( SE.

Results Characterization of the Adducts from Reaction of GA with Deoxynucleosides. N7-GA-Gua (Figure 2), the major product from reaction between GA and dG, has been characterized previously by Segerba¨ck et al. (15) on the basis of UV, 13C NMR, and ES/MS spectral analyses. In that study, substitution at N7 of guanine was deduced from the pH dependence of the UV absorbance profile of the adduct, and nucleophilic attack at the methylene carbon of GA was inferred from the upfield location of the methylene carbon of the GA-derived substituent, as compared to the methine carbon. Our UV and ES/MS/MS analyses for N7-GA-Gua were generally consistent with the data of Segerba¨ck et al. (15); however, although the 13C NMR resonances of the GA moiety at δ 49.58 (CH2), 70.28 (CHOH), and 173.50 (CONH2) were similar to those reported previously, there were substantial differences in the resonances of the Gua carbons. This was presumably due to the fact that our data were obtained in DMSO-d6, whereas Segerba¨ck et al. (15) used deuterated trifluoracetic acid, in which the solvent acidity may have induced multiple deuteration equilibria through the nitrogen atoms of the Gua ring system, therefore affecting the overall π-electron density and consequently the chemical shift values. Thus, whereas four of the five Gua carbons were found by Segerba¨ck et al. (15) between 109 and 122 ppm, which are substantially upfield from the typical resonances of purine carbons (with the exception of C5), the spectrum recorded in DMSO-d6 (cf. Experimental Section) had Gua carbon resonances within the range expected for a substituted purine (21). As compared with guanosine and dG, the C5 resonance (δ 108.18 ppm) of N7-GA-Gua was shielded by 9-10 ppm and the C8 resonance (δ 143.98 ppm) was

deshielded by 6-7 ppm, while the C2, C4, and C6 resonances were virtually unaffected. This observation is fully consistent with alkylation at N7, which should affect mostly the carbons (C5 and C8) within close proximity of the substitution site, further substantiating the conclusions drawn from the UV spectrum. The 1H NMR spectrum of N7-GA-Gua has not been reported previously. All of the expected Gua protons were observed as singlets at δ 10.84 (N1H), 7.80 (H8), and 6.13 (N2H2) ppm (cf. Experimental Section). In addition, all of the signals stemming from the GA substituent were present. The amide protons (δ 7.33 and 7.35 ppm) were clearly nonequivalent at room temperature. Typical diastereotopic behavior was also observed for the methylene protons, each detected as a doublet of doublets (δ 4.15 and 4.57 ppm) coupled to each other and to the methine proton (δ 4.25 ppm). At room temperature, the methine proton was also found to be coupled to the hydroxyl proton, which in turn was detected as a doublet (δ 5.94 ppm). Upon raising the temperature to 80-90 °C, the methine proton signal (a multiplet) collapsed reversibly to a broad singlet, whereas the geminal coupling of the methylene protons persisted. These observations are consistent with magnetic nonequivalence of the methylene protons, induced by the chirality of the methine carbon. Furthermore, the collapse of the methine coupling pattern at a relatively high temperature suggests a substantial barrier to rotation of the GA segment. Taken together, the observed coupling patterns confirm that attachment of the GA moiety to Gua occurred through the methylene carbon. The adenine-based adducts have not been reported previously (Figure 2). The ES positive product ion spectra, along with proposed fragmentation reactions for N3-GA-Ade and N1-GA-dA, are shown in Figures 3 and 4, respectively. It is noteworthy that while N3-GA-Ade retained the original carbamoyl substituent from GA,

1332 Chem. Res. Toxicol., Vol. 16, No. 10, 2003

Gamboa da Costa et al.

Figure 4. Electrospray positive product ion mass spectrum of N1-GA-dA. The major fragments derived from the protonated molecule are outlined in the structure.

hydrolysis to a carboxyl group occurred in N1-GA-dA. Although the isolated amounts of these adducts were insufficient for 13C NMR analyses, the 1H NMR spectra afforded sufficient structural information. Thus, in addition to the H2 and H8 protons, the spectrum of N1GA-dA in methanol-d4 indicated the unequivocal presence of all of the expected sugar resonances (cf. Experimental Section), which confirmed the structure as a dA derivative, instead of a depurinating adduct. The GA moiety had methylene and methine resonance patterns closely resembling those described above for N7-GA-Gua, again indicating reaction through the methylene carbon of GA. The 1H NMR spectrum of N1-GA-dA was also recorded in DMSO-d6, in an attempt to assign the labile protons; however, only a very broad signal was observed, at approximately 6.2 ppm, suggesting overlap of the N6H proton of dA with the hydroxyl proton of GA. The low concentration of the sample also precluded the detection of a downfield carboxyl proton, whose presence was inferred from the MS data. The 1H NMR pattern of the N1-GA-dA adduct, in particular the presence of the dR fragment, suggested attachment of the GA moiety to the N1 atom of dA. Further support to this structure was obtained from the UV spectrum, which was essentially identical to that of 1-methyladenosine (not shown), both at neutral (λmax ) 260 nm, 22) and either acidic (pH ) 1) or basic (pH ) 13) conditions. Extended treatment (>3 h) at pH 13 yielded a chromophore consistent with an N6-substituted adenosine derivative, which suggests the occurrence of a Dimroth rearrangement to N6-GA-dA (Figure 2), similarly to what has been reported for other N1-dA adducts from simple epoxides (14, 25). The 1H NMR spectrum of N3-GA-Ade (cf. Experimental Section), recorded in DMSO-d6, indicated the absence of a sugar residue and, therefore, a depurinating adduct structure. The GA segment had a pattern essentially identical to those of N7-GA-Gua and N1-GA-dA, and both the amide (δ 7.46 and 7.36 ppm) and the hydroxyl (δ 6.29)

protons of GA, as well as the exocyclic amine protons of Ade (δ 7.87 ppm), were clearly detected. In addition, the H2 (δ 8.18 ppm) and the H8 (δ 7.76 ppm) resonances were consistent with N3 alkylation of the purine (26-28). Additional confirmation of the assigned structure stemmed from analysis of the UV spectrum (not shown), which was virtually identical to that of 3-methyladenine at neutral pH (λmax ) 274 nm). Furthermore, the changes in the UV absorbance pattern induced by alterations in the pH to either acidic (pH 1) or basic (pH 13) conditions (not shown) corresponded closely to the changes observed with 3-methyl- and 3-ethyladenine (Weis, C. C., Mays, J. B., and Swenson, D. H. Personal communication). Small amounts of adducts isomeric with N1-GA-dA, but eluting with much longer retention times, were observed in reactions of GA with dA but were not formed to a significant degree with DNA. These adducts were presumably diastereomers arising from spontaneous Dimroth rearrangement to N6-GA-dA, because they had elution and MS characteristics identical to those of the product from incubation of N1-GA-dA at pH 13 (vide supra). Similar findings were reported by Plna et al. (25) for adenine adducts of propylene oxide. An additional adduct was formed during reactions of GA with dA. Using LC/ES/MS/MS in the CNL mode (-m/z 116) and product ion scans ([M + H]+ ) m/z 322), a prominent cyclic adduct, 1,N6-(2-hydroxypropanoyl)-2′deoxyadenosine (Figure 2), presumably formed from N1GA-dA by intramolecular condensation, was tentatively identified from analyses conducted immediately after reaction of GA with dA; however, this adduct rapidly hydrolyzed to form N1-GA-dA and was not observed in hydrolysates of DNA extensively modified with GA (data not shown). This cyclic adduct and the structurally related GA-deoxycytidine adduct were previously reported from the reaction of GA with dA and dC, respectively (14).

Acrylamide-Derived DNA Adducts in the Mouse

Chem. Res. Toxicol., Vol. 16, No. 10, 2003 1333

Table 1. DNA Adduct Levels (Expressed as Adducts/108 Nucleotides), as Determined by LC/MS/MS, in DNA Samples Modified with GA in Vitro reaction N1-GA-dA as ratioa N7-GA-Guab N3-GA-Adeb N1-GA-dAc N6-GA-dAd 100 10-1 10-2 10-3 10-4 10-5 10-6

>130 000 17 000 1400 140 17 2.0 NDe

2200 240 19 1.9 NDe

25 000 NDe

25 000 f

310

a Expressed as GA/DNA (mg/mg). Details of the modification procedure are outlined in the Experimental Section. b Determined following neutral thermal hydrolysis of the DNA (cf. the Experimental Section). c Determined following enzymatic hydrolysis of the DNA to deoxynucleosides (cf. the Experimental Section). d Determined following enzymatic hydrolysis of the DNA to deoxynucleosides and treatment at pH 13 (cf. the Experimental Section). e ND, not detected. f Not measured.

Modification of Salmon Testis DNA with GA in Vitro. Salmon testis DNA was reacted with GA at 37 °C for 24 h at ratios of 1 ng to 1 mg GA per mg DNA. The N7-GA-Gua and N3-GA-Ade adduct levels were determined using LC/MS/MS after release of the adducts from DNA by neutral thermal hydrolysis (Table 1). In addition, the N1-GA-dA adduct levels were measured using LC/ MS/MS after enzymatic hydrolysis of the DNA to nucleosides by using an on-line sample cleanup procedure similar to that previously described for polar oxidative DNA adducts (29). To assess the thermal lability of the depurinating adducts, incubations of salmon testis DNA that had been reacted with GA (100 µg GA/mg DNA) were conducted at either 21 or 37 °C for up to 48 h. The DNA was then subjected to neutral thermal hydrolysis. The half-lives for N7-GA-Gua at 21 and 37 °C were determined to be 750 and 42 h, respectively; for N3-GA-Ade, the half-lives were 69 and 14 h, respectively. The predominant adduct formed was N7-GA-Gua (Table 1), which was detected at levels 10-70-fold higher than either N3-GA-Ade or N1-GA-dA. The presence of N1-GA-dA was confirmed and quantified by conversion to N6-GA-dA at pH 13 (25) using a slight modification of the LC conditions used for N1-GA-dA. No significant amounts of N6 adduct were seen before pH 13 treatment. The limitations in method sensitivity precluded measurement of N7-GA-Gua in samples reacted with