Simultaneous Quantification of Three Lipid Peroxidation-Derived

Apr 29, 2010 - The increase in sensitivity of the analysis reduced the amount of DNA ... and 1,N2-εdGuo in tissue DNA by isotope dilution nanoLC-NSI/...
1 downloads 0 Views 1MB Size
Download PDF
Anal. Chem. 2010, 82, 4486–4493

Simultaneous Quantification of Three Lipid Peroxidation-Derived Etheno Adducts in Human DNA by Stable Isotope Dilution Nanoflow Liquid Chromatography Nanospray Ionization Tandem Mass Spectrometry Hauh-Jyun Candy Chen,* Guan-Jih Lin, and Wen-Peng Lin Department of Chemistry and Biochemistry, National Chung Cheng University, 168 University Road, Ming-Hsiung, Chia-Yi 62142, Taiwan Etheno DNA adducts are promutagenic DNA lesions derived from exogenous industrial chemicals, as well as endogenous sources including lipid peroxidation. Furthermore, levels of etheno adducts in tissue DNA are elevated in cancer-prone tissues. In this study, we have developed a highly sensitive and specific stable isotope dilution nanoflow LC-nanospray ionization tandem mass spectrometry (nanoLC-NSI/MS/MS) assay for simultaneous detection and accurate quantification of 1,N6etheno-2′-deoxyadenosine (εdAdo), 3,N4-etheno-2′deoxycytidine (εdCyt), and 1,N2-etheno-2′2 deoxyguanosine (1,N -εdGuo) in tissue DNA. Typically, [13C1,15N2]εdAdo, [15N3]εdCyd, and [13C1,15N2]1,N2εdGuo were added to calf thymus, human placenta, or human leukocyte DNA as internal standards, and the mixture was subjected to enzyme hydrolysis to form the nucleosides. The etheno adducts in DNA hydrolysate were enriched by a reversed phase solid-phase extraction column before analysis by nanoLC-NSI/MS/ MS under the highly selective reaction monitoring (HSRM) mode. This nanoLC-NSI/MS/MS assay achieved attomole-level sensitivity with the detection limit of 0.73, 160, and 34 amol for the respective standard εdAdo, εdCyd, and 1,N2-εdGuo injected on-column, while the quantification limit for the entire assay was 0.18, 4.0, and 3.4 fmol, respectively. The levels of εdAdo, εdCyd, and 1,N2-εdGuo in human placental DNA were 28.2, 44.1, and 8.5 adducts in 108 normal nucleosides, respectively. The levels of εdAdo, εdCyd, and 1,N2-εdGuo in 11 human leukocyte DNA samples were 16.2 ( 5.2, 11.1 ( 5.8, and 8.6 ( 9.1 (mean ( S.D.) in 108 normal nucleotides, respectively, starting from 30 µg of DNA or 1-1.5 mL of blood, and all the relative standard deviations were within 10%. An aliquot equivalent to 6 µg of DNA hydrolysate was used for analysis by this nanoLC-NSI/MS/MS. Thus, this highly sensitive and specific nanoLC-NSI/MS/MS method is suitable for accurate quantification of the * To whom correspondence should be addressed. Phone: (886) 5-242-8176. Fax: (886) 5-272-1040. E-mail: [email protected].

4486

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

three lipid peroxidation-derived etheno DNA adducts as noninvasive biomarkers in clinical studies for cancer risk assessment and for evaluation of preventive agents. Etheno (ε) DNA adducts, including 1,N6-etheno-2′-deoxyadenosine (εdAdo), 3,N4-etheno-2′-deoxycytidine (εdCyd), 1,N2etheno-2′-deoxyguanosine (1,N2-εdGuo), and N2,3-etheno-2deoxyguanosine (N2,3-εdGuo) (Scheme 1), are promutagenic DNA lesions that can derive from exogenous industrial chemicals, such as vinyl chloride, as well as endogenous sources, including lipid peroxidation.1-4 1,N2-εdGuo may be formed to a larger extent from endogenous oxidative processes. Among etheno adducts, 1,N2-εdGuo is a major product formed in DNA exposed to fatty acids or lipid peroxidation-derived aldehydes under peroxidizing conditions in vitro, whereas its isomer N2,3-εdGuo is originated predominately from vinyl chloride.5,6 1,N2-εdGuo was not detected in DNA treated with vinyl chloride in vitro nor in vivo, but it was formed to a large extent when DNA was treated with fatty acid or the lipid peroxidation-generated R,βunsaturated aldehyde 4-hydroxynonenal under peroxidizing conditions.6 Thus, 1,N2-εdGuo is an appropriate biomarker of endogenous lipid peroxidation-induced DNA damage, although much attention has been paid to εdAdo and εdCyd.1–4 Inflammation is a risk factor in carcinogenesis, with the hypothesis that inflammation-generated free radicals stimulate lipid peroxidation.4 Peroxidized lipids decompose and form reactive aldehydes, including R,β-unsaturated aldehydes, which generate etheno adducts under oxidatively stressed conditions.7,8 These etheno adducts are highly mutagenic in mammalian cells.9–11 The formation of etheno DNA adducts may contribute to genetic instability and cancer progression. Levels of etheno adducts in (1) Chung, F.-L.; Chen, H.-J. C.; Nath, R. Carcionogenesis 1996, 17, 2105– 2111. (2) Bartsch, H.; Nair, J. Eur. J. Cancer 2000, 36, 1229–1234. (3) Chen, H.-J. C. Chin. Pharm. J. 2004, 56, 1–16. (4) Nair, U.; Bartsch, H.; Nair, J. Free Radical Biol. Med. 2007, 43, 1109– 1120. (5) Swenberg, J. A.; Fedtke, N.; Ciroussel, F.; Barbin, A.; Bartsch, H. Carcinogenesis 1992, 13, 727–729. (6) Morinello, E. J.; Ham, A.-J. L.; Ranasinghe, A.; Sangaiah, R.; Swenberg, J. A. Chem. Res. Toxicol. 2001, 14, 327–334. (7) Chen, H.-J. C.; Chung, F.-L. Chem. Res. Toxicol. 1994, 7, 857–860. (8) Chen, H.-J. C.; Chung, F.-L. Chem. Res. Toxicol. 1996, 9, 306–312. 10.1021/ac100391f  2010 American Chemical Society Published on Web 04/29/2010

Scheme 1. Structures and Proposed Fragmentation Pathways for εdAdo, εdCyd, and 1,N2-εdGuo in MS/MSa

a

Asterisks (*) indicate the positions of the isotopes.

tissue DNA are elevated in cancerous and cancer-prone inflammatory tissues, including liver, lung, colon, pancreas and stomach,12-16 suggesting that these promutagenic lesions could drive cells to malignancy. Thus, measuring etheno adducts plays an imperative role in cancer formation and in evaluating the efficacy of preventive agents.1–4 Evidence showed the existence of repair enzymes for these etheno adducts in humans. After being repaired in the body, the urinary excretion of etheno adducts is associated with cigarette smoking, an oxidatively stressed condition, or cancer-prone liver diseases.17–23 Accurate quantification of low adduct levels in clinical DNA samples, because of their limited availability, is a challenging task from the analytical chemistry point of view. It is now feasible with the improvement in the sensitivity of the instrumental technology (9) Moriya, M.; Zhang, W.; Johnson, F.; Grollman, A. P. Proc. Natl. Acad. Sci. U. S. A 1994, 91, 11899–11903. (10) Akasaka, S.; Guengerich, F. P. Chem. Res. Toxicol. 1999, 12, 501–507. (11) Levine, R. L.; Yang, I. Y.; Hossain, M.; Pandya, G. A.; Grollman, A. P.; Moriya, M. Cancer Res. 2000, 60, 4098–104. (12) Nair, J.; Vaca, C. E.; Velic, I.; Mutanen, M.; Valsta, L. M.; Bartsch, H. Cancer Epidemiol., Biomarkers Prev. 1997, 6, 597–601. (13) Schmid, K.; Nair, J.; Winde, G.; Velic, I.; Bartsch, H. Int. J. Cancer 2000, 87, 1–4. (14) Speina, E.; Zielinska, M.; Barbin, A.; Gackowski, D.; Kowalewski, J.; Graziewicz, M. A.; Siedlecki, J. A.; Olinski, R.; Tudek, B. Cancer Res. 2003, 63, 4351–4357. (15) Frank, A.; Seitz, H. K.; Bartsch, H.; Frank, N.; Nair, J. Carcinogenesis 2004, 25, 1027–1031. (16) Nair, J.; Gansauge, F.; Beger, H.; Dolara, P.; Winde, G.; Bartsch, H. Antioxid. Redox. Signal. 2006, 8, 1003–1010.

of mass spectrometry by increasing the efficiency of ionization, transmission, and detection of ions, coupled with the miniaturized LC equipment with increased analyte concentration eluting into the MS source. The increase in sensitivity of the analysis reduced the amount of DNA samples required for the assay and is the goal of biomonitoring.24 Previous measurement of εdAdo and εdCyd was performed by 32P-postlabeling.12,14,25 To our knowledge, this is the first report analyzing the three lipid peroixidation-generated etheno adducts in the same DNA sample by MS-based method. So far, only a few reports adopted nanoflow liquid chromatography with nanospray ionization tandem mass spectrometry (nanoLC-NSI/ MS/MS) for analysis of DNA adducts in human tissues.26-30 In this report, we have developed a highly sensitive and specific assay for simultaneous detection and accurate quantification of εdAdo, εdCyd, and 1,N2-εdGuo in tissue DNA by isotope dilution nanoLC-NSI/MS/MS. This assay should be valuable in evaluation of these three adduct levels as biomarkers of lipid peroxidation-induced DNA damage and their role in carcinogenesis with minimum amount of DNA sample. Particularly, analysis of leukocyte DNA offers a noninvasive assay with an amount of blood (1-1.5 mL) that is practically accessible. EXPERIMENTAL SECTION Caution. 2-Chloroacealdehyde is a highly toxic and mutagenic volatile chemical and should be handled with gloves in a wellventilated hood. Materials. Calf thymus (Lot No. 105K7025) and human placental DNA (Lot No. 059H37791), εdAdo and all the enzymes used for DNA hydrolysis were obtained from Sigma Chemical Co. (St. Louis, MO) except that alkaline phosphatase was from Calbiochem Chemical Co. (La Jolla, CA). The isotope standards [13C1,15N2]εdAdo, [15N3]εdCyd, and [13C1,15N2]1,N2-εdGuo were synthesized by reacting the respective 2′-deoxyribonucleosides [13C1,15N2]dAdo, [15N3]dCyd, and [13C1,15N2]dGuo (Cambridge Isotope Laboratories, Andover, MA) with 2-chloroacealdehyde, followed by purification with a C18-OH SPE column and quantification by HPLC-UV as reported previously.21,22,31 All reagents are of reagent grade or higher. Human Leukocyte DNA isolation. The subjects of this study were healthy volunteers. They received a written warranty stating (17) Chen, H.-J. C.; Lin, T.-C.; Hong, C.-L.; Chiang, L.-C. Chem. Res. Toxicol. 2001, 14, 1612–1619. (18) Hanaoka, T.; Nair, J.; Takahashi, Y.; Sasaki, S.; Bartsch, H.; Tsugane, S. Int. J. Cancer 2002, 100, 71–75. (19) Chen, H.-J. C.; Hong, C.-L.; Wu, C.-F.; Chiu, W.-L. Toxicol. Sci. 2003, 76, 321–327. (20) Chen, H.-J. C.; Chiu, W.-L. Chem. Res. Toxicol. 2003, 16, 1099–1106. (21) Chen, H.-J. C.; Wu, C.-F.; Hong, C.-L.; Chang, C.-M. Chem. Res. Toxicol. 2004, 17, 896–903. (22) Chen, H.-J. C.; Chiu, W.-L. Chem. Res. Toxicol. 2005, 18, 1593–1599. (23) Nair, J.; Srivatanakul, P.; Haas, C.; Jedpiyawongse, A.; Khuhaprema, T.; Seitz, H. K.; Bartsch, H. Mutat. Res. 2010, 683, 23–28. (24) Singh, R.; Farmer, P. B. Carcinogenesis 2006, 27, 178–196. (25) Hagenlocher, T.; Nair, J.; Becker, N.; Korfmann, A.; Bartsch, H. Cancer Epidemiol., Biomarkers Prev. 2001, 10, 1187–1191. (26) Embrechts, J.; Lemiere, F.; Van Dongen, W.; Esmans, E. L.; Buytaert, P.; Van Marck, E.; Kockx, M.; Makar, A. J. Am. Soc. Mass Spectrom. 2003, 14, 482–491. (27) Liu, X.; Lovell, M. A.; Lynn, B. C. Anal. Chem. 2005, 77, 5982–5989. (28) Liu, X.; Lovell, M. A.; Lynn, B. C. Chem. Res. Toxicol. 2006, 19, 710–718. (29) Chen, H.-J. C.; Chen, Y.-C. Chem. Res. Toxicol. 2009, 22, 1334–1341. (30) Chen, H.-J. C.; Lin, W.-P. Anal. Chem. 2009, 81, 9812–9818.

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4487

that the information was for research purposes only and that their personal information would be kept confidential. Freshly drawn humanbloodwasstoredinthepresenceof10%(v/v)citrate-dextrose solution (ACD) as anticoagulant and stored at 4 °C. An aliquot of 1.1 mL of the blood/anticoagulant solution was subjected to Blood Genimic Midi kit (Viogen, Sunnyvale, CA) following the modified procedures reported previously.31 The amount of DNA was quantified by a NanoDrop 1000 photometer (J&H Technology Co., Ltd., Wilmington, DE). The purity of DNA was checked by the absorbance ratio A260/A280 being between 1.8 and 2.0. Enzyme Hydrolysis of DNA. To the solution containing DNA (30 µg), aliquots containing 100 pg each of [13C1,15N2]εdAdo, [15N3]εdCyd, and [13C1,15N2]1,N2-εdGuo was added, and the mixture was subjected to the following enzyme hydrolysis conditions with or without modification. Method A. Nuclease P1 (2.4 units) in ammonium acetate (50 mM, pH 5.3) was added to the solution containing DNA and the isotope standards and incubated at 45 °C for 2 h. Phosphdiesterase I (0.0024 unit) in Tris-HCl (100 mM, pH 7.4) was then added to the reaction mixture and incubated at 37 °C for 2 h. Finally, alkaline phosphatase (0.6 unit) and adenosine deaminase (0.9 unit) were added, and the mixture was incubated at 37 °C for 1 h.31 Method B. DNase I (77.3 units) was added to the solution containing DNA and the isotope standards and incubated at 37 °C for 3 h. Snake venom phosphodiesterase (type VII, 0.0024 units) and alkaline phosphatase (0.06 units) were then added and incubated at 37 °C overnight. Micrococcal nuclease (0.61 unit), spleen phosphodiesterase (0.061 unit) in 20 mM sodium succinate, 10 mM calcium chloride (pH 6) buffer were added to the mixture was then incubated with at 37 °C for 4 h and continued with nuclease P1 (1.5 units) for an additional 2 h.32 Method C. Micrococcal endonuclease (3 units) and spleen phosphodiesterase (0.3 unit) in Tris-HCl buffer (50 mM, pH 6.8) were added to the solution containing DNA and the isotope standards and incubated at 37 °C for 4 h.33 Alkaline phosphatase (0.6 unit) and adenosine deaminase (0.9 unit) were added, and the mixture was incubated at 37 °C for additional 1 h. Adduct Enrichment by SPE column. The enzyme digest was filtered through a 0.22 µm syringe filter, and the etheno adducts were enriched by the reversed phase C18 SPE column [Bond Elut C18, 100 mg, 1 mL, Varian (Harbor City, CA)] collecting 1 mL of the 25% aqueous methanol fraction after washing with 3 mL of water and 1 mL of 3% aqueous methanol. The collected fraction was evaporated under vacuum by a centrifuge concentrator, reconstituted in 10 µL of 0.1% acetic acid, and passed through a 0.22 µm nylon syringe filter. Two microliters of the processed sample was subjected to the nanoLC-NSI/MS/ MS for analysis. NanoLC-NSI/MS/MS Analysis. A 2-µL injection loop was connected to a six-port switching valve injector into a LC system consisting of an UltiMate 3000 Nano LC system (Dionex, Amsterdam, Netherlands) and a C18 tip column (120 mm × 75 µm, 5 µm, 100 Å) packed in-house (magic C18, Michrom BioResource, (31) Chen, H.-J. C.; Chiang, L.-C.; Tseng, M.-C.; Zhang, L. L.; Ni, J.; Chung, F.-L. Chem. Res. Toxicol. 1999, 12, 1119–1126. (32) Doerge, D.-R.; Churchwell, M.-I.; Fang, J.-L.; Beland, F.-A. Chem. Res. Toxicol. 2000, 13, 1259–1264. (33) Nair, J.; Barbin, A.; Guichard, Y.; Bartsch, H. Carcinogenesis 1995, 16, 613–617.

4488

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

Auburn, CA). The pump output (30 µL/min) was split before the injection port to a flow rate of 0.3 µL/min. Mobile phase A was 0.1% acetic acid (pH 3.3), and mobile phase B was 0.1% acetic acid in acetonitrile. The elution started with isocratic 5% mobile phase B from 0 to 7 min and a linear gradient of 5% mobile phase B to 10% mobile phase B from 7 to 12 min, followed by a linear gradient of 10% mobile phase B to 70% mobile phase B from 12 to 22 min and maintained at 70% B in the next 15 min. The effluent was subjected to analysis by a triple quadrupole mass spectrometer, TSQ Quantum Ultra EMR mass spectrometer (Thermo Electron Corp., San Jose, CA), equipped with a nanospray ionization (NSI) interface. The column effluent enters the spray chamber through a tapered emitter constructed in-house from a 15 cm long, 75-µm i.d. fused-silica capillary (Polymicro Technologies, Phoenix, AZ). The capillary was holding vertically with a stone weighing 50 g. When a flame gun approached the capillary at the point 1.5 cm from the end, the capillary was pulled by gravity and broke to form the tip. The tip of the capillary was immersed in 50% hydrofluoric acid for 3 min and then rinsed with water. The tapered emitter was directly electrosprayed into the mass spectrometer under the positive ion mode for the NSI-MS/MS. The spray was monitored by a built-in CCD camera. The spray voltage was 1.5 kV, and the source temperature was at 220 °C. Argon was used as the collision gas in MS/MS experiments. The pressure of the collision cell was 1.5 mTorr, and the collision energy was 15 V. The adduct enriched sample was analyzed by nanoLC-NSI/MS/MS with the transition from the parent ion [M + H]+ focused in quadrupole 1 (Q1) and dissociated in a collision cell (quadrupole 2, Q2) yielding the product ion [M + H - dR]+ for εdAdo and εdGuo or [M + H - dR - CO HCN]+ for εdCyd, which was analyzed in quadrupole 3 (Q3) under the highly selective reaction monitoring (H-SRM) mode with the mass width of Q1 ) 0.2 m/z and Q3 ) 0.7 m/z, and a dwell time of 0.1 s. For εdAdo and [13C1,15N2]εdAdo, the H-SRM methods monitoring Q1 and Q3 were m/z 276.1 f m/z 160.0 and m/z 279.1 f m/z 163.0, respectively, with a collision energy of 20 V. For εdCyd and [15N3]εdCyd, the H-SRM methods monitoring Q1 and Q3 were m/z 252.0 f m/z 136.0 and m/z 255.0 f m/z 139.0, respectively, with a collision energy of 25 V; whereas a collision energy of 45 V was used for εdCyd and [15N3]εdCyd at m/z 252.0 f m/z 81.0 and m/z 255.0 f m/z 83.0, respectively. For 1,N2-εdGuo and [13C1,15N2]1,N2εdGuo, the H-SRM methods monitoring Q1 and Q3 were m/z 292.1 f m/z 176.1 and m/z 295.1 f m/z 179.1, respectively, with a collision energy of 25 V. Calibration Curves. The stock solutions of εdAdo, εdCyd, 1,N2-εdGuo, and their isotopomers (1.0 mg/mL) in water were stored at -80 °C. The sample solutions for calibration were freshly prepared by diluting the stock solutions in water for each analysis. Samples with various amounts of εdAdo, εdCyd, and 1,N2-εdGuo ranging from 0, 0.05, 0.1, 1.0, 5.0, 10, 100, and 200 pg each were added to the isotopes [13C1,15N2]εdAdo, [15N3]εdCyd, and [13C1,15N2]1,N2-εdGuo (100 pg each) as internal standards. Each sample was enriched by the C18 SPE column, evaporated, reconstituted in 10 µL of 0.1% acetic acid (pH 3.3), and 2 µL of the aliquot was analyzed the above-

described nanoLC-NSI/MS/MS conditions. The equations of the calibration curves were obtained by linear regression. Assay Accuracy. The accuracy of the quantification was validated by adding three known amounts of εdAdo (20, 40, and 70 pg), εdCyd (10, 20, and 50 pg), and 1,N2-εdGuo (2.0, 4.0, 10 pg) to human placental DNA (30 µg) and analyzed for their levels by the nanoLC-NSI/MS/MS assay described above. RESULTS AND DISCUSSION Synthesis and Characterization of Standards. Accurate quantification of trace analytes in a complex matrix relies on the use of isotopomers that are structurally identical to the analytes as internal standards. The stable isotope internal standards serve as carrier for the minute amounts of analytes going through the sample pretreatment procedures, and their recovery can be properly adjusted because their physical and chemical properties are identical to those of the analytes except for their mass. They can also identify analyte peaks in the complicated chromatograms for the analytes based on the nearly identical retention times of LC-MS if isotopes other than deuterium are used. Standard εdAdo was commercially available, while εdCyd and 1,N2-εdGuo were synthesized from the parent 2′-deoxyribonucleoside with 2-chloroacetaldehyde as reported21,22 and purified by collection from reversed phase HPLC with a photodiode array detector. The isotopically labeled internal standards [13C1,15N2]εdAdo, [15N3]εdCyd, and [13C1,15N2]1,N2-εdGuo were synthesized the same way as described,21,22,31 except that they were purified by a disposable solid-phase extraction column (SPE) to avoid cross-contamination from standard εdAdo, εdCyd, and 1,N2-εdGuo. The purity of these three internal standards was confirmed by injecting 10 times of the amount of isotopes used for DNA samples, and no adducts were detected by the nanLCNSI/MS/MS. These heavy isotopes are quantified by the molar responses of their light isotopomers on HPLC-UV. Assay Development. The isotope internal standards [13C1,15N2]εdAdo, [15N3]εdCyd, and [13C1,15N2]1,N2-εdGuo were added to the DNA solution at the very beginning of the assay so that the recovery of the entire procedures can be monitored. The DNA solution containing the isotope standards was subjected to enzyme hydrolysis, and the efficiency for releasing the adducts from DNA was optimized by examining three different hydrolysis methods. To increase the detection sensitivity, it is required to enrich the analyte prior to LC/MS analysis to reduce the degree of ionization suppression by the matrix. The three etheno adducts were enriched from the DNA enzyme hydrolysate by a reversed-phase solid-phase extraction (SPE) column under conditions to collect all three adducts in one fraction and yet to eliminate as much normal nucleosides and interferences as possible. The enriched adducts were then analyzed by the nanoLC-NSI/MS/MS system, which was optimized to eliminate the interferences. In analysis of deoxynucleoside adducts by tandem mass spectrometry under the positive-ion mode, the transition from the protonated parent ion [M + H]+ to the deglycosylated base ion [M + H - deoxyribose]+ (or [B + H]+) is frequently used because of its high sensitivity. With the advancement of the technology in triple quadrupole mass spectrometers, the mass width in our instrument can be narrowed to 0.2 m/z in quadrupole 1 (Q1), while that for quadrupole 3 (Q3) is 0.7 m/z,

Figure 1. NanoLC-NSI/MS/MS chromatograms of εdAdo, εdCyd, and 1,N2-εdGuo in human placental DNA hydrolyzed by method A and analyzed under the H-SRM mode.

to increase the specificity, and consequently decrease the sensitivity, of the so-called highly selective reaction monitoring (H-SRM). Figure 1 shows the chromatograms for analyzing εdAdo, εdCyd, and 1,N2-εdGuo in human placental DNA under the H-SRM transitions monitoring Q1 and Q3 were m/z 276.1 f m/z 160.0 and m/z 279.1 f m/z 163.0, respectively, for εdAdo and [13C1,15N2]εdAdo. For εdCyd and [15N3]εdCyd, the H-SRM methods monitoring Q1 and Q3 were m/z 252.0 f m/z 136.0 and m/z 255.0 f m/z 139.0, respectively. For 1,N2-εdGuo and [13C1,15N2]1,N2-εdGuo, the SRM methods monitoring Q1 and Q3 were m/z 292.1 f m/z 176.1 and m/z 295.1 f m/z 179.1, respectively. However, in the H-SRM transition for εdCyd, an interference peak at 11.79 min was observed at the retention time of εdCyd, 11.97 min (Figure 1, third panel). We originally suspected that the interference peak was dAdo, which has the same molecular weight with εdCyd, and that we did not add enough adenosine deaminase. However, the interference peak did not disappear or became smaller with 10 times more adenosine deaminase. Therefore, we tried to use different H-SRM transitions and increased the collision energy from 25 to 45 V to further fragment 3,N4-ethenocytosine base (εCyt). The daughter ion scan spectrum of εdCyd shows the loss of deoxyribose ion [M + H deoxyribose]+ at m/z 136.0 and the loss of CO and HCN groups at m/z 81.0 (Figure 2, upper panel), whereas the fragment ions at m/z 139.0 and 83.0 agree with the fragmentation pattern and isotope labeling on [15N3]εdCyd (Figure 2, lower panel). Using this new H-SRM transitions for εdCyd and [15N3]εdCyd at m/z 252.0 f m/z 81.0 and m/z 255.0 f m/z 83.0, respectively, the level of εdCyd can be quantified at 11.94 min (Figure 1, fifth panel), but it cannot be done with the original transitions described above (Figure 1, third panel). The sensitivity of this new H-SRM transitions is about 7 times lower than that using the transitions Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4489

Table 1. Levels of εdAdo, εdCyd, and 1,N2-εdGuo in Human Placental DNA Hydrolyzed by Methods A, B, and C adducts levels (adducts/108 nucleotides),a,b mean ± SD (RSD, %) method A method B method C

2

εdAdo

εdCyd

1,N -εdGuo

27.8 ± 0.2 (0.6%) 20.7 ± 1.0 (4.6%) 23.3 ± 0.9 (4.1%)

44.4 ± 0.7 (1.6%) 28.6 ± 0.8 (2.7%) 34.4 ± 3.9 (11%)

8.1 ± 0.2 (3.0%) 8.2 ± 0.4 (4.7%) 6.5 ± 0.7 (11%)

a Each experiment started with 30 µg of human placental DNA, and an equivalent of 6 µg of DNA hydrolysate was subjected to the nanoLCNSI/MS/MS analysis. b Adduct levels are presented as mean ± standard deviation (SD) from triplicate experiments. The percentage standard deviation (RSD) is expressed in parentheses.

Figure 2. Daughter ion scan spectra of εdCyd and [15N3]εdCyd.

monitoring [M + H]+ to [M + H - deoxyribose]+ ions based on the peak areas of [15N3]εdCyd in these two H-SRM transitions (Figure 1, fourth and sixth panels). There are a number of enzyme hydrolysis conditions used in the literature for releasing DNA adducts from double-stranded DNA. We previously reported that different enzyme hydrolysis conditions could affect the efficiency of adduct release from DNA.29,30 It could be the result of the types and the amounts of enzymes used. In this study, we compared three enzyme hydrolysis conditions used for etheno adducts.31–33 Method A was used for εdAdo,31 and method B was for εdAdo and εdCyd analysis by LC-ESI/MS/MS.32 Method C was modified from that used for analyzing εdAdo and εdCyd by the 32P-postlabeling technique33 with the addition of alkaline phosphatase and adenosine deaminase with the amount used in method A. The former converts the adducted nucleotide 3′-monophosphates to the corresponding nucleosides and the latter transforms normal dAdo into deoxyinsosine, which mostly eluted earlier in the SPE column to minimize interfering with the nanoLC/NSI/ MS/MS analysis because dAdo has the same molecular weight as εdCyd. Because the enzyme hydrolysis conditions employed for 1,N2-εdGuo by Loureiro and co-workers34 used less amount of enzymes and shorter incubation time than those of methods A-C, it was not included for comparison. Table 1 summarizes the adduct levels of a human placental DNA sample measured using methods A, B, and C under the H-SRM conditions described above. It shows that method A releases 19%-55% more etheno adducts than methods B and C except that the levels of 1,N2-εdGuo measured by method A and (34) Loureiro, A. P.; Marques, S. A.; Garcia, C. C.; Di Mascio, P.; Medeiros, M. H. Chem. Res. Toxicol. 2002, 15, 1302–1308.

4490

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

B are similar. In addition, reproducibility of the analysis using method A is generally lower than using methods B and C. Thus, Method A was used throughout this study. Assay Sensitivity and Calibration. Monitored under the optimized H-SRM transitions, this nanoLC-NSI/MS/MS assay achieved attomole-level sensitivity with the on-column detection limit of 0.73, 160, and 3.4 amol (S/N ) 3) for the respective standard εdAdo, εdCyd, and 1,N2-εdGuo. Except εdCyd for which the sensitivity was sacrificed to overcome the interference problem, the ultrahigh sensitivity for εdAdo and 1,N2-εdGuo was mainly attributed to the nanoLC system using a 75 µm i.d. column, coupled with the nanospray (or nanoelectrospray) ionization source, by increasing the concentration of the analyte and efficient ionization in the MS. Lowering the column diameter further is not practical in use because it results in high column back pressure. An ultraperformance LC (UPLC) system with sub-2-µm particles in a column with an inner diameter of 100 µm or smaller coupled with NSI/MS/MS, providing high column resolution with high assay, should be a valuable tool in the field of DNA adduct analysis in which clinical tissues are limited. The calibration curves were obtained by mixing a fixed amount of εdAdo, εdCyd, and 1,N2-εdGuo (100 pg each) with various amounts of standard etheno adducts, going through the C18 SPE enrichment, and analyzed by the nanoLC-NSI/MS/MS. The calibration curves demonstrated good linearity with the correlation coefficient (R2) of 0.9983, 0.9948, and 0.9984 for εdAdo, εdCyd, and 1,N2-εdGuo, respectively (Figure 3). The calibration curves pass through the origin because no analytes were detectable in the sample containing only the isotopes. The quantification limit for the entire assay, defined as the lowest amount of analyte showing linearity, was 0.18, 4.0, and 3.4 fmol of εdAdo, εdCyd, and 1,N2-εdGuo, respectively, corresponding to 0.19, 4.2, and 3.6 adducts in 108 normal nucleotides, respectively, starting with 30 µg of DNA (Table 2). The fact that only 2 µL of the reconstituted eluant from the SPE column (in 10 µL) was injected to the LC/MS system contributes to the difference between detection and quantification limits. Assay Validation. A control sample containing the three isotopes only was passed through the SPE column and analyzed to make sure the LC system is clean and the ionization efficiency of the MS is optimal. A positive control sample was performed the same way as the real sample except that DNA was omitted. It contains the three isotopes, buffers, and hydrolytic enzymes. It

Figure 3. Calibration curves of (A) εdAdo, (B) εdCyd, and (C) 1,N2εdGuo ranging from 0.05 to 200 pg in the presence of [13C1,15N2]εdAdo, [15N3]εdCyd, and [13C1,15N2]1,N2-εdGuo (100 pg each). Table 2. Limits of Detection and Quantification of Etheno Adducts by nanoLC-NSI/MS/MS under the H-SRM Mode analyte (H-SRM transitions)

LOD

LOQ

εdAdo (m/z 276.1 f 160.0) εdCyd (m/z 252.0 f 81.0) 1,N2-εdGuo (m/z 292.1 f 176.1)

0.2 fg (0.73 amol) 40 fg (160 amol) 10 fg (34 amol)

50 fg (180 amol) 1.0 pg (4.0 fmol) 1.0 pg (3.4 fmol)

was reported that trace amounts of εdAdo was present in the hydrolytic enzymes (DNase and phosphodiesterase).32 Instead of dialyzing nearly all the enzymes used for digestion, this control experiment took into account the contribution of any artifact

derived from the system. We did not observe any detectable amount of etheno adducts in our control and positive control experiments. The overall recovery of the adducts after the entire assay procedures was approximately 59%, 63%, and 91% for εdAdo, εdCyd, and 1,N2-εdGuo, respectively, by comparing the peak areas of [13C1,15N2]εdAdo, [15N3]εdCyd, and [13C1,15N2]1,N2εdGuo in the DNA samples with those of the isotope standards (20 pg each). The recoveries in the control and positive control experiments are similar to those in DNA samples, suggesting that the loss was because of the SPE column, and no matrix effect was observed. If the matrix lowered the ionization efficiency, the peak areas of the isotopes in the DNA samples should be lower than those in the control sample. The precision in quantification of εdAdo, εdCyd, and 1,N2εdGuo was examined by analysis of human placental DNA (30 µg) in triplicates, and the experiments were performed in three different days following the optimized assay procedures using enzyme hydrolysis method A. The results showed that the relative standard deviations for both intraday and interday analyses were within 7% (Table 3), revealing the high reproducibility of the assay. The accuracy of the quantification was validated by adding three known amounts of εdAdo, εdCyd, and 1,N2-εdGuo to human placental DNA (30 µg) and analyzed for their levels. Linear regression gave a y-intercept of 6.77, 10.2, and 2.25 pg for εdAdo, εdCyd, 1,N2-εdGuo, respectively (not shown). The amount of εdAdo, εdCyd, and 1,N2-εdGuo in 30 µg of human placental DNA without addition of standards was 7.23, 10.5, and 2.22 pg, respectively, corresponding to 27.8, 44.4, and 8.1 adducts in 108 total normal nucleotides. Considering that the ratio of adenine/thymine base pairs versus cytosine/guanine pairs in human DNA is 1.5 and that 1 mg DNA contains 3.145 µmol of normal nucleotides in total, the adduct levels presented in total normal nucleotides can be converted into 0.93 εdAdo in 106 dAdo, 2.12 εdCyd in 106 dCyd, and 0.41 1,N2-εdGuo in 106 dGuo by dividing the numbers by a factor of 1.5/5 for dAdo and dThd nucleotides and the factor being 1/5 for dCyd and dGuo nucleotides. These results confirmed the excellent accuracy and quality control of this assay. Although the sources of the human placental DNA are different (with different lot numbers), the level of εdAdo is in the same order as that of our previous report.31 We also applied this nanoLC/NSI/MS/MS assay to untreated calf thymus DNA (30 µg) from a commercial source and level of εdAdo was determined as low as 2.6 in 108 normal nucleotides, while those of εdCyd and 1,N2-εdGuo were below the quantification limit (S/N ≈ 2). This result provides evidence that this assay was very sensitive and artifact was unlikely to contribute to the levels determined in human placental and leukocyte DNA (described below). Analysis of Leukocyte DNA Samples. Measurement of DNA adducts in leukocyte DNA isolated from human blood represents a noninvasive biomarker to reflect the extent of DNA damage in the whole body. To examine the possibility of using leukocyte as the surrogate tissue for biomonitoring, it requires a highly sensitive and specific assay because the DNA adduct levels in Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4491

Table 3. Precision in Quantification of Etheno Adducts by nanoLC-NSI/MS/MS Analysis adducts levels (adducts/108 nucleotides) mean ± SD (RSD, %)a,b εdAdo εdCyd 2 1,N -εdGuo

day 1

day 2

day 3

interday variation

27.8 ± 0.2 (0.6%) 44.4 ± 0.7 (1.6%) 8.1 ± 0.2 (3.0%)

29.5 ± 0.8 (2.7%) 44.4 ± 1.1 (2.4%) 8.8 ± 0.5 (6.0%)

27.2 ± 1.7 (6.4%) 43.5 ± 1.3 (3.0%) 8.7 ± 0.6 (7.1%)

28.2 ± 1.2 (4.2%) 44.1 ± 0.5 (1.2%) 8.5 ± 0.4 (4.5%)

a Each experiment started with 30 µg of human placental DNA, and an equivalent of 6 µg of DNA hydrolysate was subjected to the nanoLCNSI/MS/MS analysis. b Adduct levels are presented as mean ± standard deviation (SD) from triplicate experiments. The percentage standard deviation (RSD) is expressed in parentheses.

Table 4. Levels of εdAdo, εdCyd, and εdGuo in Human Leukocyte DNA Samples adducts levels (adducts/108 nucleotides)a,b mean ± SD (RSD, %) 2

sample

εdAdo

εdCyd

1,N -εdGuo

1 2 3 4 5 6c 7c 8 9 10 11 Mean ± SD

16.6 ± 0.03 (2.2%) 16.8 ± 0.09 (6.0%) 14.4 ± 0.12 (5.0%) 20.8 ± 0.06 (3.2%) 15.6 ± 0.09 (5.3%) 5.6 ± 0.03 (2.3%) 14.4 ± 0.06 (5.0%) 15.7 ± 0.4 (2.3%) 17.4 ± 0.4 (2.4%) 13.8 ± 0.9 (6.9%) 29.2 ± 2.1 (7.2%) 16.2 ± 5.2

15.1 ± 0.08 (5.7%) 10.1 ± 0.04 (3.1%) 14.9 ± 0.12 (8.4%) 6.1 ± 0.06 (9.8%) 5.9 ± 0.06 (8.4%) 21.8 ± 0.16 (6.8%) 14.9 ± 0.12 (8.4%) 5.6 ± 0.1 (1.1%) 16.3 ± 0.4 (2.2%) 5.6 ± 0.1 (1.4%) 5.7 ± 0.3 (4.9%) 11.1 ± 5.8

NDc 25.0 ± 0.08 (3.6%) 7.7 ± 0.06 (8.5%) 7.5 ± 0.08 (10%) NDc 3.4 ± 0.04 (8.7%) 7.7 ± 0.06 (8.4%) 6.3 ± 0.1 (1.6) 3.9 ± 0.04 (1.2%) 5.5 ± 0.04 (0.7%) 24.7 ± 2.3 (9.1%) 8.6 ± 9.1

a Each experiment started with 30 µg of human leukocyte DNA, and an equivalent of 6 µg of DNA hydrolysate was subjected to the nanoLC-NSI/MS/MS analysis. b Adduct levels are presented as mean ± standard deviation (SD) from triplicate experiments. The percentage standard deviation (RSD) is expressed in parentheses. c Not detectable.

Figure 4. NanoLC-NSI/MS/MS analysis of εdAdo, εdCyd, and 1,N2εdGuo in human leukocyte DNA under the H-SRM mode. The level of εdAdo, εdCyd, and 1,N2-εdGuo in this sample (number 11) was 27.1, 5.4, and 27.0 in 108 normal nucleotides.

blood tend to be lower than in other tissues due to its relatively short lifetime. Analyzed by the nanoLC-NSI/MS/MS, all three etheno adducts were detected in all the samples analyzed starting from 30 µg of leukocyte DNA. Only an equivalent of 6 µg of DNA hydrolysate was loaded on the column and analyzed by nanoLC-NSI/MS/MS. Figure 4 shows representative nanoLC-NSI/MS/MS chromatograms of the H-SRM transitions for εdAdo, εdCyd, 1,N2-εdGuo, and their isotope-labeled internal standards. The level of εdAdo, εdCyd, and 1,N2-εdGuo in this sample was 27.1, 5.4, and 27.0 in 108 normal nucleotides. Similar to human placental DNA, εdCyd in leukocyte DNA can only be quantified using the transition of m/z 252.0 f 81.0. Thus, the interference peak is not tissue-dependent. Table 4 summarizes the levels of εdAdo, εdCyd, and 1,N2-εdGuo in 11 human leukocyte samples from healthy volunteers. The levels of εdAdo, εdCyd, and 1,N2-εdGuo were 16.2 ± 5.2, 11.1 ± 5.8, and 8.6 ± 9.1, respectively, and all the RSD were within 10%. The levels of εdAdo and εdCyd in human leukocyte DNA fall into the adduct level range reported previously,12 while those for 1,N2-εdGuo has never been reported in the literature. Thus, simultaneous quantification of three adducts offers a convenient 4492

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

comparison of levels between adducts in a single experiment, and it saves labor, consumables, and instrument time. Because the repair rates of these three etheno adducts vary in different organs, this assay is useful in evaluating which adduct is a valid biomarker in certain pathophysiological conditions. 32 P-Postlabeling is an ultrasensitive method widely used for detection of DNA adducts because the amount of DNA needed is typically in the range of 20-50 µg. It is often combined with immunoaffinity (IA) column to trap the etheno adducts before 32 P-postlabeling. Nair and co-workers detected εdAdo and εdCyd in levels ranging from 0-27 adducts per 108 parent nucleosides in liver from humans and untreated rodents and suggested that these adducts originated from endogenous lipid peroxidation.33 Levels of εdAdo and εdCyd in human leukocyte DNA, measured by 32P-postlabeling after IA enrichment, range from 0-902 εdAdo and 0.4-716 εdCyd per 108 parent nucleosides, respectively.12–14,25 Levels of εdAdo and εdCyd determined in this study fall in the range reported. However, IA enrichment could lead to loss of the trace amount of adducts present in the biological samples and the batch-to-batch variation and aging of the IA column could be problematic, especially in an assay without an appropriate internal standard. Our assay avoids the use of the IA purification step and employs the disposable SPE for adduct enrichment to ensure the system is free of cross-contamination. Furthermore, the trace amount of adduct is carried through the

entire assay procedures by the isotope standards. Mass spectrometry offers characterization of the analyte and provides sensitive and accurate quantification by incorporation of the stable isotope of the analyte as the internal standard (i.e., the isotope dilution method), which can monitor and correct the analyte loss through the assay procedures. Morinello and co-workers employed IA/gas chromatography/electron capture negative chemical ionization high-resolution mass spectrometry for simultaneous quantification of 1,N2-εGua and N2,3-εGua after acid hydrolysis of DNA and derivatization.6 The quantification limit for 1,N2εGua was 15 fmol in 250 µg DNA, but 1,N2-εGua was not detected in 300 µg of untreated calf thymus DNA. Loureiro and co-workers applied a narrow-bore LC column with ESI/ MS/MS to determine levels of 1,N2-εdGuo in DNA from commercial calf thymus, cultured mammalian cells, and untreated female rat liver.34 The 1,N2-εdGuo levels are in the range of 1.7-5.2 adducts in 107 dGuo from 130-350 µg DNA. The current trend to increase assay sensitivity and subsequently reduce the amount of DNA required is to use nanobore LC column with nanospray ionization MS/MS for quantification of trace amounts of adducts in DNA samples.24 With the high sensitivity of the present assay, only 6 µg of DNA hydrolysate was analyzed by the nanoLC-NSI/MS/MS. The requirement of 30 µg of leukocyte DNA, which is because only 2 µL of the reconstituted solution (in 10 µL) was injected to the LC/MS system, is equivalent to 1-1.5 mL of blood because a

range of 25 - 35 µg/mL DNA is isolated from freshly drawn blood. CONCLUSION In conclusion, this is the first study for simultaneous detection and quantification of the three lipid peroxidation-derived etheno DNA adducts in human DNA with the highly sensitive and specific stable isotope dilution nanoLC-NSI/MS/MS assay. The requirement for the small amount of sample (30 µg of DNA or 1-1.5 mL of blood) makes this assay clinically feasible in assessing the possibility of measuring etheno adduct levels in human leukocyte DNA as noninvasive biomarkers for DNA damage resulting from endogenous lipid peroxidation and for evaluating their roles in cancer formation and prevention. ACKNOWLEDGMENT This work was financially supported by National Science Council of Taiwan (grants NSC 96-2628-M-194-006 and NSC 972113-M-194-007-MY3) and National Chung Cheng University (to H.-J.C.C.). The authors thank Ms. Chia-Yen Wu for initiating this study and technical assistance.

Received for review February 11, 2010. Accepted April 15, 2010. AC100391F

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4493