Quantitative Determination of N7-Methyldeoxyguanosine and O6

Anal. Chem. 2002, 74, 5376-5382

Quantitative Determination of N7-Methyldeoxyguanosine and O6-Methyldeoxyguanosine in DNA by LC-UV-MS-MS Yanan Yang, Dejan Nikolic, Steven M. Swanson, and Richard B. van Breemen*

Department of Medicinal Chemistry and Pharmacognosy, College of Pharmacy, University of Illinois at Chicago, 833 South Wood Street, Chicago, Illinois, 60612

The N-7 and O-6 positions of 2′-deoxyguanosine are the predominant sites of methylation by N-methyl-N-nitrosourea (MNU), which is used to produce a variety of experimental cancers in animal models. Here we report the development of a highly sensitive quantitative assay based on high-performance liquid chromatography-UVtandem mass spectrometry (LC-UV-MS-MS) to measure N7-methyl-2′-deoxyguanosine (N7-MedG) and O6methyl-2′-deoxyguanosine (O6-MedG) in DNA hydrolysates. Since this assay was selective for deoxyribonucleosides, potential interference from methylated RNA was eliminated. Isotopically labeled analogues, [2H3]N7-MedG and [2H3]O6-MedG, were synthesized and added to the DNA hydrolysates as internal standards. In-line UV absorbance detection was used for the quantitative analysis of the native deoxyribonucleoside dG, and MS-MS was used for the determination of N7-MedG and O6-MedG. The limits of detection for N7-MedG and O6-MedG were determined to be 64 and 43 fmol, respectively. The limits of quantification were 0.13 pmol for N7-MedG and 0.085 pmol for O6-MedG. The stabilities of N7-MedG and O6-MedG were also investigated. Although O6-MedG was stable at room temperature for at least 11 days, the half-life of N7MedG at room temperature was 2 days. Both adducts were stable at -20 °C. Calf thymus DNA and DNA from the livers of MNU-treated Sprague-Dawley rats were assayed using LC-UV-MS-MS, which was optimized for speed as well as for sensitivity. The levels of N7-MedG and O6-MedG in calf thymus DNA increased with MNU concentration and incubation time. The levels of N7-MedG and O6-MedG in the rat livers 2 h after treatment with a single dose of 50 mg/kg MNU were 95.2 N7-MedG/105 dG and 14.8 O6-MedG/105 dG. This LC-UV-MS-MS assay provides the sensitivity and speed required to evaluate the extent of methylated DNA lesions in animal models of cancer induced by the methylating agent MNU. Cancer is a disease believed to be caused by alterations in DNA sequence or aberrations in gene expression. DNA methylating * To whom correspondence should be addressed. Tel: (312) 996-9353. Fax: (312) 996-7107. E-mail: [email protected].

5376 Analytical Chemistry, Vol. 74, No. 20, October 15, 2002

agents are implicated in the etiology of human cancers1-3 and are often employed in animal models of the most common and deadly forms of cancer including breast,4 prostate,5 lung,6 and colon cancers.7 One of the most frequently used methylating agents in animal models is N-methyl-N-nitrosourea (MNU; Figure 1). MNU is a direct-acting, Sn-1-type methylating agent that induces a spectrum of DNA adducts including methylation at the N7- and O6-positions of guanine residues. Methylation at the N7position of dG is the most common adduct, accounting for over 70% of all methylation,8 and methylation at the O6-position is particularly significant since the O at position six is involved in hydrogen bonding with cytosine. Blocking the O with a methyl group allows guanine to improperly pair with thymine. After a round of DNA replication, the thymine correctly pairs with adenine, effecting a G to A transition mutation at the original site of methylation. This very mutation is known to activate certain proto-oncogenes such as members of the ras gene family.9 In addition, methylation at the O6-position of guanine can interfere with the activity of DNA-methyltransferases, which methylate cytosine residues at carbon five.10-12 The presence of 5-methylcytosine residues in CpG sites throughout the genome is an important regulator of gene expression, and the alteration of expression of oncogenes or tumor suppressor genes caused by aberrant cytosine methylation patterns is believed to contribute to carcinogenesis.13 A number of assays have been developed to measure alkylguanine or alkyldeoxyguanosine adducts. These include 13C (1) Kyrtopoulos, S. A.; Anderson, L. M.; Chhabra, S. K.; Souliotis, V. L.; Pletsa, V.; Valavanis, C.; Georgiadis, P. Cancer Detect. Prev. 1997, 21, 391-405. (2) Wachsman, J. T. Mutat. Res. 1997, 375, 1-8. (3) Kyrtopoulos, S. A. Mutat. Res. 1998, 405, 135-43. (4) Russo, J.; Russo, I. H. Breast Cancer Res. Treat. 1996, 39, 7-20. (5) Bosland, M. C. J. Cell. Biochem. Suppl. 1992, 16H, 89-98. (6) Moon, R. C.; Rao, K. V.; Detrisac, C. J.; Kelloff, G. J. Adv. Exp. Med. Biol. 1992, 320, 55-61. (7) Reddy, B. S. Hematol. Oncol. Clin. North Am. 1998, 12, 963-73. (8) Singer, B.; Grunberger, D. Molecular Biology of Mutagens and Carcinogens; Plenum: New York, 1983. (9) Barbacid, M. Annu. Rev. Biochem. 1987, 56, 779-827. (10) Hepburn, P. A.; Margison, G. P.; Tisdale, M. J. J. Biol. Chem. 1991, 266, 7985-7. (11) Tan, N. W.; Li, B. F. Biochemistry 1990, 29, 9234-40. (12) Bentivegna, S. S.; Bresnick, E. Cancer Res. 1994, 54, 327-9. (13) Wajed, S. A.; Laird, P. W.; DeMeester, T. R. Ann. Surg. 2001, 234, 10-20. 10.1021/ac020235o CCC: $22.00

© 2002 American Chemical Society Published on Web 09/06/2002

Figure 1. Structures of MNU, N7-MedG, O6-MedG, [2H3]N7-MedG, and [2H3]O6-MedG

nuclear magnetic resonance spectroscopy,14 immunoassay,15-17 gas chromatography/electron capture mass spectrometry (GC/MS),18 high-performance liquid chromatography (HPLC) with fluorescence detection,19 HPLC with electrochemical detection,20 HPLC purification followed off-line by 32P-postlabeling,21-24 and HPLC followed off-line by either CI or DCI mass spectrometry.25-27 The HPLC-electrochemical detection, HPLC-fluorescence, and GC/MS assays require cleavage of the glycosidic bond, preventing the ability to distinguish guanosine residues derived from DNA from those derived from RNA, which contaminates many DNA isolates. Furthermore, most of these assays including the NMR method are insufficiently sensitive to measure the low levels of alkylated species typically formed in vivo. Some of these published methods did not address methylated DNA adducts. Therefore, it is essential to develop more sensitive and selective on-line HPLC methods for qualitative and quantitative analysis of major DNA lesions induced by alkylating agents such as MNU. To address this problem, an LC-UV-MS-MS method was developed for measuring methylated guanosine residues derived (14) Chang, C.; Lee, C. G. Biochemistry 1981, 20, 2657-61. (15) Bianchini, F.; Montesano, R.; Shuker, D. E.; Cuzick, J.; Wild, C. P. Carcinogenesis 1993, 14, 1677-82. (16) Degan, P.; Montesano, R.; Wild, C. P. Cancer Res. 1988, 48, 5065-70. (17) Wild, C. P. Mutat. Res. 1990, 233, 219-33. (18) Saha, M.; Abushamaa, A.; Giese, R. W. J. Chromatogr., A 1995, 712, 34554. (19) Herron, D. C.; Shank, R. C. Cancer Res. 1980, 40, 3116-7. (20) Park, J. W.; Ames, B. N. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 7467-70. (21) Shields, P. G.; Povey, A. C.; Wilson, V. L.; Weston, A.; Harris, C. C. Cancer Res. 1990, 50, 6580-4. (22) Reddy, M. V.; Gupta, R. C.; Randerath, E.; Randerath, K. Carcinogenesis 1984, 5, 231-43. (23) Mustonen, R.; Forsti, A.; Hietanen, P.; Hemminki, K. Carcinogenesis 1991, 12, 1423-31. (24) Wilson, V. L.; Basu, A. K.; Essigmann, J. M.; Smith, R. A.; Harris, C. C. Cancer Res. 1988, 48, 2156-61. (25) Chang, C. J.; Ashworth, D. J.; Isern-Flecha, I.; Jiang, X. Y.; Cooks, R. G. Chem. Biol. Interact. 1986, 57, 295-300. (26) Ashworth, D. J.; Baird, W. M.; Chang, C. J.; Ciupek, J. D.; Busch, K. L.; Cooks, R. G. Biomed. Mass Spectrom. 1985, 12, 309-18. (27) Chae, W. G.; Chang, C. J.; Wood, J. M.; Cooks, R. G. Biol. Mass Spectrom. 1991, 20, 503-4.

from MNU-treated DNA either in vitro or in vivo. We measured N7-methyldeoxyguanosine (N7-MedG) since this is by far the most abundant adduct produced in vivo.8 We also examined O6-methyl2′-deoxyguanosine (O6-MedG) due to its putative role in carcinogenesis. In the present report, a sensitive, quantitative, and rapid assay is described for the analysis of N7-MedG and O6-MedG derived exclusively from DNA and its application to DNA samples isolated from MNU-exposed rats. This novel method should facilitate studies on cancer initiation induced by methylating agents. EXPERIMENTAL SECTION Reagents. Ammonium acetate, methanol, chloroform, and ammonia (aqueous) were purchased from Fisher Scientific (Fair Lawn, NJ). O6-Methyl-2′-deoxyguanosine, dG, dA, dC, T, iodomethane, iodomethane-d3 (99.5+ atom % deuterium), methyld3 alcohol-d (99.8 atom % deuterium), trifluoroacetatic anhydride, and the enzymes RNase A, RNase T1, protease K, nuclease P1, and alkaline phosphatase were purchased from Sigma Chemical (St. Louis, MO). Deionized water was obtained on-site using a Milli-Q water purification system (Millipore Corp., Bedford, MA). Synthesis of N7-MedG and [2H3]-N7-MedG. N7-MedG was synthesized based on the method of Ashworth et al.25 dG (200 mg, 0.75 mmol) was dissolved in 2 mL of DMSO at 4 °C, and 0.2 mL (32 mmol) of iodomethane or iodomethane-d3 was added dropwise. The mixture was stirred for 6 h. The product solution was applied to a silica gel chromatography column and eluted with isocratic MeOH/CHCl3/NH3(aq) (50:50:1) as the mobile phase. Fractions were collected and monitored by using TLC [Rf ) 0.12 (N7-MedG/[2H3]-N7-MedG)]. The fractions containing N7MedG or [2H3]N7-MedG were combined, evaporated to dryness under reduced pressure using a rotary evaporator, and then recrystallized from anhydrous methanol. The yields were 60% (120 mg, 0.43 mmol) N7-MedG and 30% (60 mg, 0.21 mmol) [2H3]-N7MedG (99.5+ atom % deuterium). Analytical Chemistry, Vol. 74, No. 20, October 15, 2002

5377

Synthesis of [2H3]O6-MedG. Labeled O6-MedG was synthesized based on the method of Faithi.28 dG (100 mg, 0.37 mmol) was dissolved in 2 mL of dry pyridine with 0.5 mL of trifluoroacetic anhydride (TFAA) and kept in an ice bath. To this solution, 60 mL of methyl-d3 alcohol-d (CD3OD) containing 0.94 g (16 mmol) of sodium methoxide-d3 (NaOCD3) was added dropwise, and the solution was stirred for 24 h. The mixture was treated with pyridine hydrochloride (3 mL of pyridine/1 mL of concentrated HCl), and then the excess acid was destroyed by addition of 1 g of sodium bicarbonate. After filtration, the solution was concentrated and applied to a 30-g silica chromatography column. The column was washed with 200 mL of chloroform to remove byproducts, and then [2H3]O6-MedG was eluted with methanol/ chloroform (1:9). Fractions were collected and checked by using TLC [Rf ) 0.33 ([2H3]O6-MedG), methanol/chloroform (1:9)]. The fractions containing [2H3]O6-MedG were evaporated to dryness using a rotary evaporator and then recrystallized from ethyl acetate. The yield of [2H3]O6-MedG was 10% (10 mg, 0.04 mmol, 99.8 atom % deuterium). Methylation of Calf Thymus DNA by MNU. Calf thymus DNA (100 µg) was added to 50 µL of 100 mM Tris buffer at pH 7.4 containing either 0.1 or 1 mM MNU. Each solution was incubated at 37 °C for 10, 20, 30, or 40 min. After incubation, 25 µL of 5 M NaCl and 288 µL of ethanol (-20 °C) were added to precipitate DNA. The solid DNA was washed with 70% ethanol three times and kept frozen until hydrolysis. DNA Extraction and Purification from Rat Liver. All work conducted with animals was performed in compliance with the policies set forth in the Guide for the Care and Use of Laboratory Animals. Seven-week-old female Sprague-Dawley rats were purchased from Harlan Sprague-Dawley (Indianapolis, IN), housed under controlled temperature, humidity, and lighting conditions. The animals were fed Teklad 8640 diet and water ad libitum. Rats received a single intraperitoneal injection of MNU (50 mg/kg) or saline vehicle. Two hours after injection, the rats were sacrificed by CO2 asphyxiation. The median lobe of the liver was excised and homogenized for 1 min using a Brinkmann (Westbury, NY) Polytron in 3 mL of cell lysing buffer (320 mM sucrose, 10 mM Tris, 5 mM MgCl2, 50 mM D-mannitol, 10 mM Triton X-100) at pH 7.4 and 4 °C. Another 27 mL of cell lysing buffer was added, and the solution was vortex mixed. The sample solution was centrifuged for 15 min (2500 rpm, 664g) at 4 °C, and the supernatant was decanted carefully. The nuclei pellet was transferred to a 1.5-mL Eppendorf microcentrifuge tube, and 900 µL of 1% SDS, 1 mM EDTA, 10 µL of 1 M Tris (pH 7.4), and 90 µL of 5 M NaCl were added followed by vortex mixing. Then, 16 µL of RNase T1 (7500 units/mL in water) and 200 µL of RNase A (4 mg/mL in 10 mM Tris pH 7.4) were added, and the solution was incubated for 30 min at 37 °C. A 40-µL aliquot of freshly prepared protease K (40 mg/mL in water) was added to the solution, which was incubated at 37 °C for 30 min. The solution was transferred to a 5-mL tube, and 180 µL of 5 M NaCl and 15 µL of 1 M Tris pH 8.0 were added. The solution was extracted with equal volumes of 1-butanol. The aqueous layer was transferred to a 5-mL tube, and ice-cold 100% ethanol (3.2 mL) was added to precipitate DNA (70% ethanol in the final solution). The DNA was washed twice with 70% ethanol and stored at -20 °C (28) Fathi, R. Tetrahedron Lett. 1990, 31, 319-322.

5378

Analytical Chemistry, Vol. 74, No. 20, October 15, 2002

until hydrolysis. DNA Hydrolysis. DNA was dissolved in water at ∼2 mg/ mL, and a 50-µL aliquot (containing ∼100 µg of DNA) was mixed with 50 µL of buffer (50 mM ammonium acetate, 0.2 mM ZnCl2, pH 5.3), 10 µL of nuclease P1 (0.4 unit/µL), and 8 µL of alkaline phosphatase (1 unit/µL in 10 mM Tris pH 7.4) and incubated at 37 °C for 30 min. A 60-µL aliquot of the DNA digest and 10 µL of internal standard mixture (containing 27 ppb [2H3]N7-MedG and 15 ppb [2H3]O6-MedG) were combined in a 30 000 molecular weight cutoff ultrafiltration centrifuge tube (Amicon; Bedford, MA). The solution was centrifuged for 15 min at 12 000 rpm and 4 °C just before LC-UV-MS-MS analysis. LC-UV-MS-MS Quantification of Methylated Nucleosides. The HPLC system consisted of a Waters (Milford, MA) model 2690 pump, autoinjector, model 2487 dual-absorbance UV detector, and a YMC (Wilmington, NC) ODS-AQ C18 column (2.0 × 250 mm, 5 µm), and guard column (4.0 × 20 mm). Absorbance detection at 260 nm was used for on-line HPLC quantitation of dG in the DNA hydrolysate, and a standard curve was prepared using standard solutions of dG in the initial mobile phase. The flow rate was 200 µL/min, and the injection volume was 25 µL. The solvent system consisted of isocratic 25 mM ammonium acetate, pH 7.0/methanol, 95:5 (v/v), for 12 min followed by a linear gradient from 5 to 30% methanol over 12-23 min. The mobile phase was held at 30% methanol for 2 min before returning to the conditions over 3 min. Multiple reaction monitoring (MRM) was carried out using a Micromass (Manchester, U.K.) Quattro II triple quadrupole mass spectrometer equipped with positive ion electrospray ionization. Nitrogen was used as the drying and nebulizing gas at 8 and 0.8 L/min, respectively. The cone voltage was 40 eV. Collision-induced dissociation (CID) was carried out using argon at 1.5 µTorr with a collision energy of 25 eV. The dwell time per ion channel was 25 ms during MRM. Stability Studies. To investigate the stability of the methylated deoxyguanosines, a solution containing 0.554 ppm N7-MedG and 0.491 ppm O6-MedG in 1.0 mL of 100 mM Tris buffer, pH 7.4, was stored in the dark at room temperature. Identical solutions were stored in the dark at -20 °C or at 37 °C. Aliquots (60 µL) were removed daily for 11 days from the room temperature and -20 °C solutions and at 0.5, 1, 2, 4, 6, 17, 24, and 48 h from the 37 °C incubation and then analyzed for stability by using LCUV-MS-MS. Recovery Studies. Aliquots (180 µL each) of an enzymatic hydrolysate of 1 mg/mL calf thymus DNA were spiked with 20 µL of one of four standard solutions, so that the concentration of N7-MedG was 0.11, 0.55, 1.22, or 2.80 ppm and the concentration of O6-MedG was 0.10, 0.49, 0.98, or 2.48 ppm. After ultrafiltration through a 30 000 molecular weight cutoff membrane, a 60-µL aliquot of each solution was mixed with 10 µL of internal standard, and the concentration of the methylated guanosine was determined using LC-UV-MS-MS. Water (180 µL) spiked with 20 µL of the corresponding standard solution was analyzed as a control for the determination of recovery. Four dG standard solutions consisting of 91.1, 121, 152, and 202 ppm were filtered through a 30 000 molecular weight cutoff ultrafiltration membrane, and the concentration of dG in each ultrafiltrate was determined using HPLC with UV absorbance detection at 260 nm as described

Figure 2. LC-UV-MS-MS analysis of a hydrolysate of calf thymus DNA that had been treated with 1 mM MNU. (A) Positive ion electrospray LC-MS-MS with CID and MRM of stable isotope-labeled internal standards [2H3]N7-MedG and [2H3]O6-MedG). (B) Positive ion electrospray LC-MS-MS with CID and MRM of the nucleosides N7-MedG and O6-MedG released from the calf thymus DNA. (C) LC-UV chromatogram showing the elution of the native nucleosides released during the hydrolysis of the calf thymus DNA.

above. The corresponding unfiltered solutions of dG were analyzed as controls for the determination of recovery. RESULTS AND DISCUSSION Since the methylated nucleosides N7-MedG and O6-MedG should be present at only the low-ppm or high-ppb level compared to native nucleosides in DNA hydrolysates, an analytical method for their quantification requires low limits of detection for the methylated nucleosides as well as lack of interference from the abundant native nucleosides. In this LC-UV-MS-MS assay, HPLC was used to separate the deoxynucleosides to prevent interference while electrospray tandem mass spectrometry provided high selectivity and low limits of detection for the measurement of N7-MedG and O6-MedG. In addition, in-line UV absorbance detection was used for the simultaneous quantitative analysis of the native nucleosides. As a result, the levels of the methylated deoxyguanosines could be determined relative to unchanged dG. Figure 2 shows the LC-UV-MS-MS chromatograms of a hydrolysate of calf thymus DNA that had been treated with 1 mM MNU for 60 min. The native deoxyribonucleosides, detected using UV absorbance detection at 260 nm, were completely separated from each other and from N7-MedG and O6-MedG, which were detected using MS-MS. The retention times were 9.8, 22.3, 23.4, and 26.8 min for dG, dA, dC and T, and 15.7 and 31.3 min for N7-MedG and O6-MedG, respectively. As a charged nucleoside, N7-MedG eluted earlier than dG in the reversed-phase chromatogram, while methylation at the O6-position resulted in a longer retention time for O6-MedG. The inclusion of ammonium acetate as a weak ion pair agent in the mobile phase was necessary to obtain suitable peak shape, resolution, and reproducible retention times. During positive ion electrospray mass spectrometry, N7-MedG and O6-MedG formed abundant protonated molecules of m/z 282 while [2H3]N7-MedG and [2H3]O6-MedG formed ions of m/z 285. MS-MS with CID of the each of these protonated molecules produced only one abundant fragment ion, [MH - 116]+, which was formed by the elimination of deoxyribose with a concomitant

hydrogen transfer (Figure 3). Therefore, the loss of the deoxyribose moiety from the protonated molecule, m/z 282 f 166, was selected for MRM of N7-MedG and O6-MedG. The analogous transition, m/z 285 f 169, was selected for MRM of the internal standards [2H3]N7-MedG and [2H3]O6-MedG, which coeluted with the corresponding unlabeled species during LC-MS-MS (Figure 2). The LC-MS-MS standard curves for N7-MedG and O6-MedG and the HPLC-UV standard curve for dG showed excellent linearity (see Figure 4). For example, over the concentration range 0.48-249 pmol injected on-column, the correlation coefficient R2 was 0.9995 for N7-MedG (N ) 3). For O6-MedG, the correlation coefficient was 0.9999 (N ) 3) over the concentration range 0.43221 pmol injected on-column. Over the concentration range 4.7518.9 nmol of dG injected on-column, the correlation coefficient R2 was 0.9975 (N ) 3). The concentrations of dG used for the HPLC-UV standard curve were much higher than those of N7MedG or O6-MedG measured using LC-MS-MS, because relatively few guanine residues on DNA were methylated by MNU. Following enzymatic hydrolysis of DNA, ultrafiltration was used to separate the low molecular weight nucleosides from the enzymes according to our method designed to minimize artifactual oxidation of nucleosides such as dG.29 Since the levels of N7-MedG and O6-MedG released from DNA were measured relative to the amount of unmethylated or native dG, it was necessary to evaluate the recovery of all three species. The recoveries for these deoxynucleosides at different concentrations are summarized in Table 1. Depending upon concentration, the recoveries were 92100, 88-100, and 98-100% for O6-MedG, N7-MedG, and dG, respectively. The limits of detection of N7-MedG and O6-MedG, defined as a signal-to-noise ratio of 3:1 (N ) 3), were determined to be 64 and 43 fmol, respectively. The limits of quantification for N7-MedG and O6-MedG, defined as a signal-to-noise ratio of 10:1, were (29) Hua, Y.; Wainhaus, S. B.; Yang, Y.; Shen, L.; Xiong, Y,.; Xu, X.; Zhang, F.; Bolton, J. L.; van Breemen, R. B. J. Am. Soc. Mass Spectrom. 2000, 12, 80-87.

Analytical Chemistry, Vol. 74, No. 20, October 15, 2002

5379

Figure 3. (a) Positive ion MS-MS with CID of (A) N7-MedG and (B) [2H3]N7-MedG. (b) Positive ion MS-MS with CID of (A) O6-MedG and (B) [2H3]-O6-MedG

determined to be 0.13 and 0.085 pmol, respectively. The coefficients of variance (CV) for the LC-MS-MS determinations of N7-MedG and O6-MedG were 6.7 and 7.9% (N ) 6), respectively. Intraassay reproducibility, determined by the standard deviation of five injections of the same standard solution, was 2.3% for N7MedG and 1.7% for O6-MedG. The interassay CV of N7-MedG and O6-MedG were determined to be 6.5 and 3.4% (N ) 5), respectively. Since decomposition of N7-MedG or O6-MedG might occur during sample preparation or storage prior to analysis, stability studies were carried out at 37 °C, room temperature, and freezer storage conditions. The results of the stability studies are summarized in Figure 5. Both N7-MedG and O6-MedG were stable in the dark at -20 °C for at least 11 days. Also, O6-MedG was stable during storage for at least 11 days in the dark at room temperature and at 37 °C for at least 48 h. However, N7-MedG decomposed to 18% of its original concentration after 11 days in the dark at room temperature. The half-life of N7-MedG was determined to be 2 days at room temperature (Figure 5A) and 24 h at 37 °C (Figure 5380 Analytical Chemistry, Vol. 74, No. 20, October 15, 2002

5C). However, no change in the concentration of N7-MedG was detected after 30 min at 37 °C. Based on these results, a fast and low-temperature sample preparation procedure is required. Although RNA contamination of DNA has the potential to interfere with many DNA analyses, the selectivity of this LCUV-MS-MS assay prevents this problem. First, enzymatic digestion is used to release deoxyribonucleosides from DNA, and these compounds remain chemically distinct from any ribonucleosides that might remain in the samples. Methods that rely on the analysis of free nucleic acid bases cannot distinguish between bases derived from DNA and RNA at this point. Second, N7-MedG, O6-MedG, and dG have HPLC retention times that are different from their corresponding ribonucleosides. Third, the molecular weights of N7-MedG and O6-MedG are different from their ribonucleoside analogues and may be distinguished by their protonated molecules during the first stage of mass spectrometry. Finally, the fragment ions of N7-MedG and O6-MedG recorded during the final stage of mass spectrometry in this MS-MS assay are formed by the elimination of deoxyribose. Interference by

Figure 4. Calibration curves for the positive ion electrospray LCMS-MS measurement of (A) N7- MedG and (B) O6-MedG and for the HPLC-UV analysis of (C) dG. Each data point represents the mean of three determinations ( standard deviation. Table 1. Recoveries of N7-MedG, O6-MedG, and dG Following Ultrafiltration Determined Using LC-UV-MS-MSa N7-MedG

O6-MedG

dG

concn (ppm)

recovery (%)

concn (ppm)

recovery (%)

concn (ppm)

recovery (%)

0.11 0.55 1.11 2.80

88 98 100 88

0.10 0.49 0.98 2.48

100 100 94 92

91.1 121 152 202

99 100 99 98

a Each recovery value represents the average of two determinations and is expressed as a percent of control. N7-MedG and O6-MedG concentrations were measured using LC-MS-MS, and dG concentrations were determined using LC-UV as described in the Experimental Section.

ribonucleosides will not occur since ribose weighs 16 u more than deoxyribose. Since RNA contamination does not interfere with this assay, the DNA isolation and purification procedures could be simplified

Figure 5. Stability of (A) N7-MedG and (B) O6-MedA in 100 mM Tris buffer at -20 °C and room temperature and (C) at 37 °C.

and shortened. Using our streamlined experimental protocol, DNA purification and hydrolysis could be completed in ∼2 h. Furthermore, all sample handling steps except for the hydrolysis incubations were carried out at ∼4 °C. Finally, LC-UV-MS-MS analyses were carried out on the same day as the DNA isolation and hydrolysis. For the in vitro application of this LC-UV-MS-MS assay, calf thymus DNA was treated with MNU, and untreated DNA was used as a control. No N7-MedG or O6-MedG was detected in the hydrolysates of the calf thymus DNA control. In the MNU-treated samples, methylation of guanine occurred more at N7 than at the O6 position since the N7 position of guanine is more electronegative.24 For example, exposure of calf thymus DNA to 1 mM MNU for 40 min produced 9.0 N7-MedG/105 dG, which is almost 10-fold higher than the corresponding level of 1.0 O6-MedG/105 dG. The time courses for methylation of guanine at the N7 and O6 positions by treatment of calf thymus DNA with 0.1 and 1 mM Analytical Chemistry, Vol. 74, No. 20, October 15, 2002

5381

time points and both concentrations of MNU, levels of N7-MedG exceeded those of O6-MedG. As an example of the in vivo application of this LC-UV-MSMS assay, Sprague-Dawley rats were treated with either a single intraperitoneal injection of MNU or saline vehicle. Two hours after injection, liver tissues were removed and assayed for O6-MedG and N7-MedG. In the control group, N7-MedG was below the limit of detection, and the background level of O6-MedG was 0.6/105 dG. In the MNU-treated group; there were 95.2 N7-MedG/105 dG and 14.8 O6-MedG/105 dG. CONCLUSION The levels of N7-MedG and O6-MedG in hydrolyzed DNA can be determined accurately using LC-UV-MS-MS with positive ion electrospray, CID, and MRM. Isotopically labeled internal standards that coelute with the analytes were used for peak confirmation and precise quantitation to correct for sample losses during ultrafiltration and sample preparation. To prevent decomposition of N7-MedG (the least stable methylated product), samples must be kept cold and processed quickly. This new LCMS-MS method is suitable for the analysis of methylated DNA from in vivo as well as in vitro systems and might be useful for cancer research.

Figure 6. Time course for the methylation of calf thymus DNA by MNU. LC-UV-MS-MS was used for the quantitative analysis of (A) N7-MedG and (B) O6-MedG relative to dG following enzymatic hydrolysis of the DNA.

MNU are shown in Figure 6. Methylation proceeded rapidly for the first 10 min and then slowed considerably by 20 min. At all

5382 Analytical Chemistry, Vol. 74, No. 20, October 15, 2002

ACKNOWLEDGMENT This research was supported by Grants RR10485 (R.R.v.B.), CA083124 (R.R.v.B.), AG020820 (S.M.S.), and CA071895 (S.M.S.) from the National Institutes of Health.

Received for review April 10, 2002. Accepted July 31, 2002. AC020235O