Mass Spectrometric Methodology for the Determination of

Matthias Baum , Richard N. Loeppky , Silke Thielen and Gerhard Eisenbrand ... Michael C. Byrns, Choua C. Vu, Jonathan W. Neidigh, José-Luis Abad, Rog...
0 downloads 0 Views 341KB Size
Download PDF
556

Chem. Res. Toxicol. 2005, 18, 556-565

Mass Spectrometric Methodology for the Determination of Glyoxaldeoxyguanosine and 6 O -Hydroxyethyldeoxyguanosine DNA Adducts Produced by Nitrosamine Bident Carcinogens Michelle K. Dennehy and Richard N. Loeppky* Department of Chemistry, University of Missouri, Columbia, Missouri 65211 Received July 22, 2004

N-Nitrosodiethanolamine (NDELA) is a bident carcinogen that undergoes both P-450 mediated R-hydroxylation and β-oxidation, leading ultimately to the formation of two prominent DNA adducts, glyoxaldeoxyguanosine (gdG) and O6-2-hydroxyethyldeoxyguanosine (OHEdG), in rat liver. HPLC coupled with electrospray ionization (ESI) and tandem mass spectrometry was used for both detection and quantification of gdG and OHEdG. The method, which is fast, sensitive, and unambiguous, is a significant improvement over the previous 32P-postlabeling methodology. A rapid procedure for the enzymatic hydrolysis of the DNA under acidic conditions preserved the integrity of the pH sensitive gdG adducts. Glyoxal and 3-nitroso-2-oxazolidinone generated gdG and OHEdG adducts, respectively, in calf thymus DNA (ct-DNA) in a concentration (range of 104) dependent manner permitting optimization. Isotopomeric internal standards were prepared from the modified guanine derivatives by enzymatic trans-glycosylation. Quantitative HPLC-ESI-MS/MS analysis employing selective reaction monitoring (SRM) for the loss of the deoxyribose fragment was utilized. Both adducts could be detected in the liver DNA of rats that were administered NDELA in a dose range of 0.4-0.8 mmol/kg. At the highest dose, gdG adducts (4.4-11 adducts/106 nuc.) were more abundant than OHEdG adducts (0.35-0.87 adducts/106 nuc.). Conversely, OHEdG adducts were produced in higher yields in ct-DNA than were gdG adducts at the same reagent concentrations.

Introduction

Scheme 1

We have previously shown that N-Nitrosodiethanolamine (NDELA1) 1 is a bident (two-toothed) carcinogen (1, 2). It and several other nitrosamines undergo metabolic activation by means of microsomal oxidation ([O]) to produce DNA binding fragments from both arms of the nitrosamine as shown in Scheme 1 (1-3). In vitro experiments have shown that microsomal oxidation occurs at both R- and β-positions (1). The former process generates the unstable R-hydroxynitrosamine 2 that decomposes to the highly reactive 2-hydroxyethyldiazonium ion 3, which is an efficient and indiscriminate alkylating agent. Glycolaldehyde 4 is also a product of this process. β-Oxidation leads to NHMOR 6 through its unstable aldehyde isomer 5. The further microsomal oxidation of either glycoladehyde or NHMOR results in the formation of glyoxal 7. DNA adducts, such as 6O-2* Corresponding author. Phone: (573) 882-4885. Fax: (573) 8822754. E-mail: [email protected]. 1 Abbreviations: AcP, acid phosphatase; Ct-DNA, calf thymus DNA; DNAse, deoxyribonuclease; gdG, glyoxaldeoxyguanosine or 3-(2′-deoxyβ-D-erythro-pentofuranosyl)-5,6,7,9-, tetrahydro-6,7-dihydroimidazo[1,2-a]purin-9-one; dG*, 8-13C-7,9-15N2-deoxyguanosine; gdG*, glyoxal8-13C-7, 9-15N2-deoxyguanosine; gG*, glyoxal-8-13C-7,9-15N-guanine; HPLC, high-performance liquid chromatography; HPLC-ESI-MS/MS, high performance liquid chromatography electrospray ionization tandem mass spectrometry; NDELA, N-nitrosodiethanolamine; NHMOR, N-nitroso-2-hydroxymorpholine; OHEdG, O6-hydroxyehtyldeoxyguanosine; OHEdG*, O6-1′,1′,2′,2′-D4-2-hydroxy-ethyldeoxyguanosine; OHEG, O6-hydroxyethylguanine; OHEG*, O6-1′,1′,2′,2′-D42-hydroxy-ethylguanine; PDEII, phosphodiesterase II; PNPase, purine nucleoside phosphorylase; SRM, single/selected reaction monitoring; TPase, thymidine phosphorylase.

hydroxyethyl-2′-deoxyguanosine (OHEdG) 8 and 3-(2′deoxy-β-D-erythro-pentofuranosyl)-5,6,7,9-tetrahydro-6,7dihydroimidazo[1, 2-a]purin-9-one (gdG) 9 are formed in DNA by these processes. We previously reported the development and utilization of 32P-postlabeling procedures for the detection of both the gdG adduct 9 and the OHEdG adduct 8 and demonstrated their presence in the DNA of rats given NDELA and related nitrosamines (2). These 32P-postlabeling methods were extremely useful in the initial elaboration of the mechanism of NDELA carcinogenic action but possessed certain features that

10.1021/tx049802o CCC: $30.25 © 2005 American Chemical Society Published on Web 02/11/2005

HPLC-MS/MS Analysis of gdG and OHEdG DNA Adducts

rendered their use in more extensive mechanistic or epidemiological studies problematic. The gdG adduct is very pH sensitive and decomposes readily as the pH is increased above pH 6.5 to dG and glyoxal or products derived from it (2, 4). At pH 0.99). For gdG, pmol gdG ) 1.268 (pmol gdG measured) -0.7, and the error in the slope was (0.0.006. For OHEdG, the linear equation was pmol OHEdG ) 0.978 (pmol OHEdG) - 1.4, and the error in the slope was (0.003. The calibration curves are given in the Supporting Information. In the DNA samples, while we encountered some instrument variability, the practical detection limit for gdG was 1.5-2 pmol and for OHEdG it was 100-150 fmol. Under ideal instrument conditions 200 fmol of the gdG adduct could be detected at good S/N. The number of adducts per 106 nucleotides was determined as follows: widely available tabulated data give the % GC and % AT base pairs in the DNA of a given species. This permits the determination of an approximate molar mass for a nucleotide. The number of nucleotides in the DNA sample can then be determined from the DNA concentration. Areas for the adduct peak and its standard isotopomer were measured by two different methods, which did not yield statistically significantly different data. In the first case, the area was determined using standard software. This method, however, does not allow for the setting of a precise time-based integration window, which is the same for all runs. In the second method, the data were transferred to a spreadsheet, which permitted precise setting of the window and indexing of chromatograms. The areas were then determined by summing the detector response over the window width. The concentration of the adduct of interest was calculated by dividing the area of the analyte by the area of the standard and by multiplication of the resulting quotient by the concentration of the standard. These data were used with the regression equations obtained form the calibration curves to determine the final concentrations from which the units of adduct were calculated first as pmol determined and then adducts per 106 nucleotides.

Chem. Res. Toxicol., Vol. 18, No. 3, 2005 559

Results

Figure 1. Method for determination of the gdG an OHEG adducts by acidic DNA hydrolysis and HPLC-ESI-MS/MS using SRM is summarized.

Preparation of Synthetic Nucleoside Adduct Standards. To develop a method for the trace detection of the gdG and OHEdG DNA adducts, standards of these compounds were prepared synthetically. The gdG adduct was prepared directly from the reaction of dG with glyoxal (13). The OHEdG standard was made from O6hydroxyethylguanine (OHEG) by enzymatic glycosylation in which the enzymes, purine nucleoside phosphorylase, and thymidine phosphorylase transferred a ribosyl moiety from thymidine to OHEG, generating OHEdG (15). The target compound was obtained after the reaction mixture was concentrated, and fractions were collected using HPLC. The optimal reaction time (5 h) for the glycosylation of O6-hydroxyethylguanine was determined by the formation of O6-(2-hydroxyethyl)deoxyguanosine being monitored via HPLC. Preliminary HPLC-ESIMS/MS experiments showed that the two adducts of interest, gdG and OHEdG, both undergo major fragmentation with the loss of the ribosyl moiety as has been observed for similar compounds and is as shown in the inset of Figure 1. Synthesis of Isotopic Nucleoside Adduct Standards. To achieve superior accuracy in our HPLC-ESI-

MS/MS analytical scheme, we synthesized isotopomers of each adduct for use as internal standards that differed by three (gdG) or four (OHEdG) atomic mass units from those of the adducts being analyzed. Because of the availability of isotopic starting materials, available synthetic routes, and the minimization of contamination by unlabeled material, we chose 8-13C-7,9-15N2-glyoxaldeoxyguanosine (gdG*) and O6-1′,1′,2′,2′-D4-2-hydroxyethyldeoxyguanosine (OHEdG*) as standards for the detection and quantification of gdG and OHEG, respectively. The syntheses are outlined in Scheme 2. Purchased 8-13C-7,9-15N2-guanine, 98% 15N, and 98% 13C isotopic purity was converted to the labeled guanosine by a slight adaptation of the enzymatic transglycosylation procedures of Chapeau and Marnett (15) and Holmberg et al. (16). Reaction with glyoxal then gave the isotopic standard gdG*, which was purified by repetitive HPLC and collection. 2-Amino-6-chloropurine was reacted with the sodium salt of 1,1,2,2-D4-ethanediol (98% isotopic purity) to give the corresponding tetradeuterated isotopomer of O6-hydroxyethylguanine, which was converted into the corresponding deoxyriboside using the same

560

Chem. Res. Toxicol., Vol. 18, No. 3, 2005

Dennehy and Loeppky

Scheme 2. Synthesis of Isotopic Adduct Standards

transglycosylation methodology. A priori, we did not know whether the enzymatic trans-glycosylation procedure (15) would work to produce the OHEdG adduct. The transglycosylation procedures proceeded well giving 60 and 68% yields of dG* and OHEdG*, respectively. We were impressed with the versatility of this methodology. Each standard was assayed for isotopic purity by both high-resolution MS and HPLC-ESI-MS/MS. In each case, we could not detect any contamination from the unlabeled substance. HPLC-ESI-MS/MS Method Development. A number of preliminary HPLC experiments was performed to obtain the best separation of the adducts of interest. A mobile phase containing 0.1% formic acid to facilitate ionization and acetonitrile and water with a C-18 150 × 2.1 mm 5 µm column at a flow rate of 0.3 mL/min produced the best results. Methanol could not be used as a solvent or a mobile phase because of the partial formation of gdG-methanol acetal adducts under the chromatographic conditions (4). Repetitive injection of gdG solutions at a given concentration as well as those containing the isotopic standards gave comparable areas and no indication that the glyoxal fragment was being lost through hydrolysis or other transformations in the mildly acidic solution for the duration of the analysis time, which did not exceed 24 h. Specific retention times (volumes) for the nucleosides of interest are reported in the Experimental Procedures and are shown graphically in the chromatographs. The chromatographic conditions were manipulated so that the unadducted nucleosides would not coelute with adducted nucleosides of interest, even though the latter were not completely separated in all cases. For example, although dA and dG were not entirely resolved, gdG elutes without interference of other nucleosides. In prior work, we synthesized the 2-hydroxyethyl adducts of the mono- and diphosphonucleotides of guanine at the 2-NH2, 7, and O6 positions and found them to be chromatographically wellseparated under chromatographic conditions similar to those here (21), but we did not check these samples or their corresponding nucleosides in this work. On the basis of our prior experience, however, we believe that the coelution of OHEdG with these related adducts is unlikely. The technique of selective reaction monitoring (SRM) in a triple quadrapole mass spectrometer was applied in this study. The two adducts, gdG and OHEdG, and their respective isotopomeric internal standards, which cochromatographed with them, were separated by several

minutes in the HPLC. This allowed two separate segments to be programmed into the method for a single run. The first SRM segment was set to scan for the loss of the deoxyribose fragment from gdG and gdG*, that of 326 f 210 m/z and 329 f 213 m/z, respectively. The second segment scanned for the same loss in the two OHEdG isotopomers, 312 f 196 m/z and 316 f 200 m/z. We evaluated the reliability and reproducibility of this methodology by making standard mixtures of both adducts and their isotopomers. The analyte concentration was varied, while the isotopic standard concentration was held constant. Calibration curves were established by linear regression of the data and were highly linear (r2 > 0.99). DNA Hydrolysis Method Development. The enzymatic digestion procedure used to hydrolyze DNA was adapted from the method of Marsch et al. (19). The process is short and proceeds well in mildly acidic solution where the gdG adduct is most stable. Ammonium acetate buffers were used in place of the corresponding sodium salts to avoid MS complications arising from Na+ adducts in the ESI source. Briefly, the DNA was digested for 20 min at pH 5 with DNAse, giving 3′oligodeoxyphosphonucleotides. Further treatment with PDE gave the 3′-monodeoxynucleotides. Incubation of the resulting mixture with acid phosphatase for 4 h generated the deoxynucleosides. HPLC-UV monitoring experiments with ct-DNA showed the procedure to be very effective. AcP from white potato was used preferentially over AcP from wheat germ as it cleaved the phosphate groups more efficiently in our system. The final method is summarized as depicted in Figure 1. Modification of Ct-DNA and HPLC-ESI-MS/MS Determination of Adducts. Ct-DNA was modified with glyoxal, a process that has been shown to occur in a concentration dependent manner (4). While there are various possible methods for producing hydroxyethyl adducts in DNA, the formation of OHEdG adducts in significant yield requires the in situ generation of the highly unstable 2-hydroxyethyldiazonium ion 3, which alkylates the base oxygen atoms with greater efficiency than more stable alkylating agents (e.g., ethylene oxide). The decomposition of 3-nitroso-2-oxazolidinone 10 (Scheme 3) is known to generate the 2-hydroxyethyldiazonium ion 3, which reacts rather indiscriminately with nucleophilic sites in DNA to form the OHEdG adduct (22), among other products. While 3-nitroso-2-oxazolidinone 10 has not been used for this purpose, the published half-life of 3-nitroso-2-oxazolidinone in phosphate buffer at pH 6.4

HPLC-MS/MS Analysis of gdG and OHEdG DNA Adducts

Chem. Res. Toxicol., Vol. 18, No. 3, 2005 561

Scheme 3

and 25 °C is 50 min (22), which suggested that it would generate the adduct of interest in a short time. As a trial, we allowed an excess 3-nitroso-2-oxazolidinone to react with dG in buffer at pH 7 and 37 °C. The formation of OHEdG was monitored using HPLC-ESI-MS/MS, and the disappearance of 3-nitroso-2-oxazolidinone was followed in the UV chromatogram (on the MS system) of the same run. While the latter technique was not useful in obtaining highly accurate rate data for the decomposition of 3-nitroso-2-oxazolidinone, we observed qualitatively that the OHEdG adduct concentration increased as the nitroso compound concentration decreased. Moreover, we observed good pseudo-first-order kinetic data for the formation of OHEdG (k ) 7.9 ( 0.4 × 10-3 min-1). These data indicated that 3-nitroso-2-oxazolidinone would effectively introduce 2-hydroxyethyl groups into DNA.

Figure 3. HPLC-ESI-MS/MS of acid-hydrolyzed 3-nitroso2-oxazolidinone modified ct-DNA. Panel 1 is a UV chromatogram (λ ) 254 nm) that displays the HPLC separation of the digested DNA. The retention times of the nucleosides are shorter than those in Figure 2 as a shorter HPLC column was used. Panel 2 depicts MS/MS scan of OHEdG monitoring the 312 f 196 fragmentation. Panel 3 portrays the MS/MS scan of OHEdG* in which the 316 f 200 fragmentation is monitored.

Figure 2. HPLC-ESI-MS/MS of acid-hydrolyzed glyoxal modified ct-DNA. Panel 1 is a UV chromatogram (λ ) 254 nm) that displays the HPLC separation of digested DNA. Panel 2 depicts MS/MS scan of gdG monitoring the 326 f 210 fragmentation. Panel 3 portrays the MS/MS scan of gdG* in which the 329 f 213 fragmentation is followed.

Trial reactions and the application of the methodology depicted in Figure 1 showed that this was true. We allowed ct-DNA to react with either glyoxal or 3-nitroso-2-oxazolidinone over a concentration range of 104 in multiples of 10 to produce gdG or OHEdG adducts, respectively. The methodology described previously was used to liberate the nucleoside adducts and quantitate them. Known amounts of the internal standards were added at the end of the digestion and removal of the protein. Chromatograms obtained from typical runs of digested glyoxal or 3-nitroso-2-oxazolidinone modified DNA are shown in Figures 2 and 3, respectively. The adduct data are presented in Figure 4 and the tabulated data are given in the Supporting Information. The reaction of glyoxal (0.01-100 mM) with ct-DNA produced 0.15 gdG adducts per 106 nucleotides at the lowest concentration and 72 at the highest, while the same concentration range of 3-nitroso-2-oxazolidinone produced an adduct range of OHEdG of 6.4-126 per 106 nucleotides. Animal Experiments. Metabolic fragments from Nnitrosodiethanolamine (NDELA), among other nitrosamines, are known to alkylate DNA and form gdG and

562

Chem. Res. Toxicol., Vol. 18, No. 3, 2005

Dennehy and Loeppky

Figure 4. Number of adducts per 106 nucleotides detected by HPLC-ESI-MSMS in ct-DNA (1 mg/mL) after reaction with either glyoxal (9) or N-nitroso-2-oxazolidinone (b) over the concentration range given at 37 °C for 15 h are shown. The inset shows the entire concentration range, whereas the larger graph shows the data at lower adduct levels.

OHEdG adducts (2). To assess the ability of these adducts to serve as markers of exposure in humans and determine their utility in further carcinogenesis mechanism studies, a cursory animal study was performed. A primary goal of this pilot study was to determine whether our method was sensitive enough to detect the adducts in rats that were administered NDELA. Because a large proportion of the NDELA is known to be excreted unchanged (23), this is a relatively severe test. Eight Wistar rats, in pairs, were administered NDELA at doses of 0.0 (H2O control), 0.2, 0.4, and 0.8 mmol of NDELA/kg body wt. and euthanized at 4 h, their livers were removed, and DNA was isolated from them (2). The DNA was hydrolyzed to nucleosides using an acidic enzymatic digest described previously, and the resulting mixture was analyzed using our HPLC-ESI-MS/MS method. Not surprisingly, perhaps, this DNA was more difficult to workup and convert into nucleotide free nucleosides as can be seen in the UV monitored chromatographic trace shown in a representative sample (Figure 5). Nevertheless, we were able to detect and quantitate the levels of gdG and OHEdG DNA adducts in samples from the highest dose levels. No adducts were detected in the control rats or at a dose of 0.2 mmol/kg. Adducts could be detected, but not quantified due to low S/N, in one rat at 0.4 mmol/kg. At a dose of 0.8 mmol/kg, the adduct levels (adducts/106 nucleotides) were rat 1, gdG 4.4 ( 0.7 and OHEdG 0.35 ( 0.02 and rat 2, gdG 11 ( 1 and OHEdG 0.87 ( 0.02. Although the data were not as good, the adduct levels seen in the rat given 0.4 mmol/kg were approximated to be about half of those seen in rat 2, indicating in very limited sense that higher doses of NDELA increased adduct levels. Although we used only one set of animals, two animals at each dose level, the DNA isolation and hydrolysis were repeated several times. The data were determined from a single day’s MS run and are typical. Qualitatively, other day’s runs showed similar trends but were not judged suitable for comparison despite the use of both external, and of course, the internal standards because of variability in instrument response, from either day to day, or in some cases run to run. For each animal at a given NDELA dose, three different aliquots from a single DNA

Figure 5. HPLC-ESI-MS/MS of hydrolyzed rat liver DNA from a rat that was gavaged with 0.8 mmol of NDELA/kg body wt. Panel 1 depicts a UV chromatogram (λ ) 254 nm) in which the HPLC separation of the digested DNA is shown. Both nucleoside and nucleotide peaks are present. Panel 2 depicts MS/MS scan of gdG monitoring the 326 f 210 fragmentation. (The unknown was also seen in ct DNA.) Panel 3 portrays the MS/MS scan of gdG* in which the 329 f 213 fragmentation is monitored. In panel 4, the MS/MS scan of OHEdG is shown in which the 312 f 196 fragmentation is monitored. Panel 5 reveals the MS/MS scan of OHEdG* in which the 316 f 200 fragmentation is monitored.

HPLC-MS/MS Analysis of gdG and OHEdG DNA Adducts

digestion were analyzed, and the data were averaged to generate the values reported.

Discussion One of the most encouraging aspects of this research was our finding that the adapted DNA enzymatic hydrolysis methodology of Marsch et al. (19) quickly and reliably liberated the gdG and OHEdG from DNA. This is particularly important for the accurate assay of gdG, which as we have noted is particularly sensitive to hydrolytic removal of the glyoxal fragment at basic pH values, even those close to pH 7. Two attributes of our DNA hydrolysis procedure led to success: (1) the incubations can effectively be done at mildly acidic pH (6-6.5), where gdG is much more stable and (2) the procedure is much shorter than what we used in our 32P-postlabeling assay of the same adducts (2). The much shorter digestion times not only increase lab efficiency but lessen the probability that the glyoxal fragment will move from DNA and/or nucleotides to the enzymes or to the solution. Moreover, this DNA hydrolysis procedure proved to be much more reliable as we moved from one lot of biochemicals to another. This was a great problem in our prior 32P-postlabeling work with the gdG adduct (2). Because of its much more robust nature, the OHEdG adduct was less susceptible to these difficulties. The SRM HPLC-ESI-MS/MS methodology, while it does not have quite the sensitivity as the 32P-postlabeling procedure for some adducts, as has been discussed elsewhere, has other superb advantages. The most important among these is the degree of certainty that MS provides with respect to the identity of the adduct, which is enforced by, but independent of, chromatographic position. The 32P-postlabeling procedure achieves its greatest sensitivity and certainty with more hydrophobic (bulky) adducts, which chromatographically separate well from the labeled unmodified nucleotides. This is not true for gdG and OHEdG, which are quite hydrophilic. As we discuss next, in the animal experiments we were able to achieve equivalent sensitivity for the detection of the gdG adduct to what we observed in the 32Ppostlabeling experiments and perhaps at least a 100-fold greater detection limit for the OHEdG adduct. The stable isotopic internal standards necessary for this work were easily prepared through enzymatic transglycosylation methods (15). Because large quantities are not needed, HPLC purification of these substances was very effective and efficient. The HO group of the OHEdG O6-2-hydroxyethylguanine posed no problems in this enzymatic synthesis of the unlabeled and labeled OHEdG, which we had previously made by intricate a laborious organic synthetic procedures (21). Overall, the methodology summarized in Figure 1 was very efficient and reliable. It is much easier to perform in the laboratory than the 32P-postlabeling procedure (2). The use of HPLC-ESI-MS/MS with SRM for detection of the adducts of interest did involve some problems. These were principally related to enigmatic changes in instrument sensitivity from day to day and even during a run. As a result, in all of our work, we strove to perform all analyses in a contiguous block of time, to order the samples randomly in the autosampler, to analyze standard solutions frequently, to wash the injection system by solvent injection at frequent intervals, and to perform repetitive analyses of the same sample. Thus, from the

Chem. Res. Toxicol., Vol. 18, No. 3, 2005 563

MS perspective, our animal work should be viewed as a guide and a demonstration of principal rather than the definitive optimization of the procedure. Not surprisingly, we achieved our maximal detection limit when we worked with the pure nucleoside mixtures. We also obtained good results, as is demonstrated in Figure 4 when we analyzed adduct levels produced by modification of ct-DNA. With these samples, and even more with the samples from the animal experiments, there can be problems with ionization suppression that arises from the presence of other bases and nucleotides, even though we attempted to manage this as best we could by good chromatographic separation of the adducts of interest. The data of Figure 4 show that N-nitroso-2-oxazolidinone was more effective in producing adducts in ct-DNA than was glyoxal at the same concentration. In both cases, the number of adducts does not increase linearly with reagent concentration. We have observed this phenomenon in other cases, as have others (2, 24). Qualitatively, it can be explained by the fact that not all guanine residues are equally accessible due to the threedimensional structure of DNA. At the same reagent concentration, the yields of the OHEdG adduct from reaction with N-nitroso-2-oxazolidinone were greater than were the gdG adduct yields produced by glyoxal. This result was somewhat unexpected. Guanine bases in DNA have a high affinity for reaction with glyoxal (4), but we allowed ample time for glyoxal reaction, which is reversible. Although glyoxal is also known to generate cross-links in DNA (25, 26) (not detected here), it is quite selective in its reaction with guanine. On the other hand, the 2-hydroxyethyldiazonium ion 3 is vastly more reactive than glyoxal, but N-nitroso-2-oxazolidinone is not 100% efficient in producing it (22). Moreover, because of its great reactivity, the 2-hydroxyethyl diazonium ion 3 alkylates guanine at the nitrogen atoms, alkylates other bases and the phosphate oxygens, hydrolyzes to ethylene glycol, and undergoes rearrangement to acetaldehyde on decomposition (1). Ultimately, it must be the greater reactivity of this species that gives rise to more adduction of the guanines than glyoxal does. The most interesting finding to arise from this part of our study is the observation that about 10-fold higher levels of the gdG adduct were seen in comparison to OHEdG in both rats 5 and 6 in the animal studies. These data obviously contrast with what we observed in our in vitro experiments with ct-DNA. The most significant difference in these experiments is that the DNA binding agents are produced metabolically in the rat. Our in vitro microsomal metabolism experiments with NDELA showed that β-oxidation, which is required for glyoxal formation, is preferred over R-oxidation (1). Subsequent R-oxidation of NHMOR 6 at C-5 will give a different diazonium ion, but the other fragments from the R-hydroxylation of NHMOR 6 at either position should ultimately give either glyoxal or compounds that can metabolically give rise to it. In the in vitro ct-DNA experiments, the reactive diazonium ion was generated proximal to the target. We do not know how it, or its precursors, is transported to the DNA in vivo, but the probability of their destruction before arrival is relatively high because of the high intrinsic chemical instability of the 2-hydroxyethyldiazonium ion 3. Thus, the greater quantities of gdG found in the in vivo experiments could result from several factors but most probably the much shorter lifetime of

564

Chem. Res. Toxicol., Vol. 18, No. 3, 2005

the 2-hydroxyethyl diazonium ion. A comparison of the data obtained in this study with our previously published animal experiments shows that the levels of gdG adduct observed here for rats 1 and 2 (4-11 gdG/106 nuc.) are close to the same as what we observed using the 32P-postlabeling method (4-6 gdG/ 106 nuc.) (2). In each case, we were not able to observe gdG DNA adducts at NDELA doses below 0.8 mmol/kg body weight. In the prior study, because of the uncertainties posed by the two 32P-postlabeling chromatographic methods (TLC and HPLC), we employed standard additon (spiking) techniques, which were not used here. Even though we were only able to perform a cursory animal study with NDELA, the data from this study confirm those obtained from our prior work (2). As noted previously, the levels of OHEdG DNA adducts determined in our animals (0.4-0.9 OHEdG/106 nuc.) are lower than the gdG adducts and are lower that those measured in our prior work (2.3-24 OHEdG/ 106 nuc.) at comparable doses of NDELA. While we have no ready explanation for this difference, the important fact is that both methods show formation of DNA adducts from both arms of the NDELA molecule and confirm its bident carcinogenic nature. The HPLC-ESI-MS/MS method used here is, of course, more precise than 32P-postlabeling with respect to the identity of the adduct, yet data presented here are derived from a smaller number of animals. As noted previously, and as we have reviewed in our prior publications on the bioactivation of NDELA (1, 2, 4, 24, 27, 28), a large proportion of administered NDELA is excreted unchanged. From this perspective, the detection of any DNA adducts from it is significant. We believe that more recent instrument developments and perhaps further optimization of the animal derived DNA hydrolysis methodology could increase detection limits and permit more extensive dose response studies. Conclusions. Facile and sensitive methods for the detection of glyoxaldeoxyguanosine (gdG) and O6-(2hydroxyethyl)deoxyguanosine (OHEdG) adducts in DNA have been developed. Key components involve the use of an acidic enzymatic DNA hydrolysis, which minimizes the hydrolytic loss of the gdG adduct, and HPLC-MS methodologies for detection. More specifically, the ESIHPLC-MS/MS technique of selective reaction monitoring (SRM) increased the sensitivity of the overall method. Detection limits in the DNA derived samples were adequate but not ideal. Isotopically labeled internal standards, 8-13C-7,9-15N2-glyoxal-deoxyguanosine (gdG*) and 1,1,2,2-D-O6-(2-hydroxyethyl)deoxyguanosine (OHEdG*), were prepared very effectively by enzymatic trans-glycosylation procedures from the modified bases and were used in the quantification of their nonlabeled counterparts. The methodology is much faster, more reliable, and of equal or higher sensitivity than the 32Ppostlabeling labeling methods previously developed by our group for these same two adducts. The method developed proved sufficiently sensitive to detect and quantify the gdG and OHEdG adducts in the liver DNA of rats administered N-nitrosodiethanolamine (NDELA). gdG adducts were present at higher levels than OHEdG adducts. The levels of gdG DNA adducts were similar to those we reported previously where detection and quantitation were done by 32P-postlabeling, but lower levels of OHEdG were found.

Dennehy and Loeppky

Acknowledgment. We gratefully acknowledge the National Institutes of Environmental Health Sciences (NIEHS) for the support of this research by means of Grant RO1 ES05953. Supporting Information Available: Calibration curve for the determination of gdG and OHEdG DNA adducts and table of the number of gdG and OHEdG adducts in glyoxal and 3-nitroso-2-oxazolidinone modified ct-DNA. This material is available free of charge via the Internet at http://pubs.acs.org.

References (1) Loeppky, R. N., and Goelzer, P. (2002) Microsome-Mediated Oxidation of N-Nitrosodiethanolamine (NDELA), a Bident Carcinogen. Chem. Res. Toxicol. 15, 457-469. (2) Loeppky, R. N., Ye, Q., Goelzer, P., and Chen, Y. (2002) DNA Adducts from N-Nitrosodiethanolamine and Related β-Oxidized Nitrosamines in Vivo: 32P-Postlabeling Methods for Glyoxal- and O6-Hydroxytheyldeoxyguanosine Adducts. Chem. Res. Toxicol. 15, 470-82. (3) Okazaki, O., Persmark, M., and Guengerich, F. P. (1993) NNitroso-N-methylvinylamine: Reaction of the epoxide with guanyl and adenyl moieties to yield adducts derived from both parts of the molecule. Chem. Res. Toxicol. 6, 168-173. (4) Loeppky, R. N., Cui, W., Goelzer, P., Park, M., and Ye, Q. (1999) Glyoxal-guanine DNA adducts: Detection, stability, and formation from nitrosamines in Cyclic DNA Adducts II (Singer, B., and Bartsch, H., Eds.) pp 155-168, IARC, Lyon. (5) Huber, C. G., and Oberacher, H. (2002) Analysis of Nucleic Acids by On-Line Liquid Chromatography-Mass Spectrometry. Mass Spectrom. Rev. 20, 310-343. (6) Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F., and Whitehouse, C. M. (1990) Electrospray inoizationsprinciples and practice. Mass Spectrom. Rev. 9, 37-70. (7) Chaudhary, A. K., Nokubo, M., Ogelsby, T. D., Marnett, L. J., and Blair, I. A. (1995) Characterization of Endogenous DNA Adducts by Liquid Chromatography/Electrospray Ionization Tandem Mass Spectrometry. J. Mass Spectrom. 30, 1157-66. (8) Andrews, C. L., Voutros, P., and Harsch, A. (1999) Analysis of DNA adducts using high-performance separation techniques coupled to electrospray ionization mass spectrometry. J. Chromatogr. A 856, 515-26. (9) Paehler, A., Richoz, J., Soglia, J., Vouros, P., and Turesky, R. J. (2002) Analysis and Quantification of DNA Adducts of 2-Amino3,8-dimethylimidazo[4,5-f]quinozaline in Liver of Rats by Liquid Chromatography/Electrospray Tandem Mass Spectrometry. Chem. Res. Toxicol. 15, 551-561. (10) Tretyakova, N., Chiang, S., Walker, V., and Swenberg, J. A. (1998) Quantitative analysis of 1,3-butadiene-induced DNA adducts in vivo and in vitro using liquid chromatography electrospray ionization tandem mass spectrometry. J. Mass Spectrom. 33, 363376. (11) Gangl, E., Turesky, R., and Vouros, P. (1999) Determination of in Vitro- and in Vivo-Formed DNA Adducts of 2-Amino-3methylimidazo[4,5-f]quinoline by Capillary Liquid Chromatography/Microelectrospray Mass Spectrometry. Chem. Res. Toxicol. 12, 1019-1027. (12) Loeppky, R. N., and Xiong, H. (1994) The synthesis of deuteriumlabeled N-nitrosodiethanolamine and N-nitroso-2-hydroxymorpholine. J. Labelled Compd. Radiopharm. 24, 1099-1110. (13) Chung, F. L., and Hecht, S. S. (1985) Formation of the cyclic 1,N2glyoxal-deoxyguanosine adduct upon reaction of N-nitroso-2hydroxymorpholine with deoxyguanosine. Carcinogenesis (London) 6, 1671-1673. (14) Ashby, J., Paton, D., Styles, J. A., Greatbanks, D., and Wright, B. (1982) Synthesis of N7-hydroxyethylguanine and O6-hydroxyethylguanine Markers for the reaction of ethylene oxide (EO) with DNA. Mutat. Res. 103, 257-261. (15) Chapeau, M.-C., and Marnett, L. J. (1991) Enzymatic Synthesis of Purine Deoxynucleoside Adducts. Chem. Res. Toxicol. 4, 636638. (16) Holmberg, I., Stal, P., and Hamberg. (1999) Quantitative Determination of 8-Hydroxy-2′-deoxyguanosine in Human Urine by Isotope Dilution Mass Spectrometry: Normal Levels in Hemochromatosis. Free Radical Biol. Med. 26, 129-35. (17) Swenson, D. H., Frei, J. V., and Lawley, P. D. (1979) Synthesis of 1-(2-Hydroxyehtyl)-1-nitrosourea and Comparison of Its Carcinogenicity with that of 1-Ethyl-1-nitrosourea. J. Natl. Cancer Inst. 63, 1469-73.

HPLC-MS/MS Analysis of gdG and OHEdG DNA Adducts (18) Mirvish, S. S., Weisenburger, D. D., Salmasi, S., and Kaplan, P. A. (1987) Carcinogenicity of 1-(2-hydroxyethyl)-1-nitrosourea and 3-nitroso-2-oxazolidone administered in drinking water to male MRC-Wistar rats: induction of bone, hematopoietic, intestinal, and liver tumors. J. Natl. Cancer Inst. 78, 387-93. (19) Marsch, G. A., Mundkowski, R. G., Morris, B. J., Lisa, M. M., Hartman, M. K., and Guengerich, F. P. (2001) Characterization of Nucleoside and DNA Adducts Formed by S-(1-Acetoxymethyl)glutathione and Implications for Dihalomethane-Glutathione Conjugates. Chem. Res. Toxicol. 14, 600-608. (20) Anonymous. (2000) TSQ Basic Operations, ThermoQuest Institute, Section 7. (21) Loeppky, R. N., and Gu, F. The synthesis of O6-(2-hydroxyethyl)deoxyguanosine nucleotides. Manuscript in preparation. (22) Singer, S. S. (1985) Decomposition Reactions of (Hydroxyalkyl)nitrosoureas and Related Compounds: Possible Relationship to Carcinogenicity. J. Med. Chem. 28, 1088-1093. (23) Preussmann, R., Habs, M., Habs, H., and Schmaehl, D. (1982) Carcinogenicity of N-nitrosodiethanolamine in rats at five different dose levels. Cancer Res. 42, 5167-5171.

Chem. Res. Toxicol., Vol. 18, No. 3, 2005 565 (24) Park, M., and Loeppky, R. N. (2000) In Vitro DNA Deamination by R-Nitrosaminoaldehydes Determined by GC/MS-SIM Quantitation. Chem. Res. Toxicol. 13, 72-81. (25) Kasai, H., Iwamoto-Tanaka, N., and Fukada, S. (1998) DNA modifications by the mutagen glyoxal: adduction to G and C, deamination of C and GC and GA cross-linking. Carcinogenesis 19, 1459-1465. (26) Brock, A. K., Kozekov, I. D., Rizzo, C. J., and Harris, T. M. (2004) Coupling Products of Nucleosides with the Glyoxal Adduct of Deoxyguanosine. Chem. Res. Toxicol. 17, 1047-1056. (27) Loeppky, R. N., Fuchs, A., Janzowski, C., Humberd, C., Goelzer, P., Schneider, H., and Eisenbrand, G. (1998) Probing the mechanism of the carcinogenic activation of N-nitrosodiethanolamine with deuterium isotope effects: in vivo induction of DNA single strand breaks and related in vitro assays. Chem. Res. Toxicol. 11, 1556-1566. (28) Loeppky, R. N. (1999) The mechanism of bioactivation of Nnitrosodiethanolamine. Drug Metab. Rev. 31, 175-193.

TX049802O