Chem. Res. Toxicol. 2002, 15, 1295-1301
1295
Quantification of Multiple DNA Adducts Formed through Oxidative Stress Using Liquid Chromatography and Electrospray Tandem Mass Spectrometry Mona I. Churchwell, Frederick A. Beland, and Daniel R. Doerge* National Center for Toxicological Research, Jefferson, Arkansas 72079 Received October 1, 2001
Damage to DNA can arise through covalent modification of bases by reaction with oxidants and products of lipid peroxidation derived through normal aerobic metabolism. Such premutagenic lesions, including 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxo-dG), pyrimido[1,2R]purine10(3H)one-2′-deoxyribose (M1-dG), etheno-2′-deoxyadenosine (-dA), and etheno-2′-deoxycytidine (-dC), are believed to be important in the development of human cancers related to diet, disease states, and lifestyle. We report the development of a method for concurrent quantification of these four adducts in DNA hydrolysates of 100 µg or less using on-line sample preparation coupled with liquid chromatography and tandem mass spectrometry. The sensitive detection permitted adduct quantification at levels below one adduct in 108 normal nucleotides and measurement of these adducts in DNA from untreated rodent liver and normal human liver samples. This methodology should prove useful in hypothesis-driven studies of cancer etiology in laboratory animals and humans.
Introduction A price is paid for the higher metabolic efficiency gained from the use of molecular oxygen as the terminal electron acceptor in oxidative phosphorylation - the production of reactive species derived from incomplete reduction (e.g., O2-‚, H2O2). Transition metal-catalyzed reactions can convert these molecules into the highly reactive hydroxyl radical. All classes of biological macromolecules (lipids, proteins, and nucleic acids) are susceptible to chemical modification by reactive oxygen species, but DNA is a particularly important cellular target because of the potential for cumulative heritable changes. The free radical theory of aging posits that accumulation of oxidative damage to cells is causative in many degenerative diseases associated with aging, including cancer (1). Chemical modification of DNA bases can occur through direct oxidation by hydroxyl radical or indirectly through reaction with products of membrane lipid peroxidation. Figure 1 shows the structures for several well-studied examples: 8-oxo-dG, M1-dG, -dA, and -dC.1 Studies have linked exposure to ionizing radiation and the concomitant formation of hydroxyl radical with increased levels of a number of oxidized bases in DNA, including 8-oxo-dG (2). It has also been shown that hydroxyl radical-mediated oxidation at the 4′-position of deoxyribose moieties in DNA forms base propenals that are * To whom correspondences should be addressed. Telephone: (870) 543-7943. Fax: (870) 543-7720. E-mail:
[email protected]. 1 Abbreviations: dA, 2′-deoxyadenosine; dC, 2′-deoxycytidine; dG, 2′-deoxyguanosine; ECD, electrochemical detection; -dA, 1,N6--2′deoxyadenosine; -dC, 3,N4--2′-deoxycytidine; IS, internal standard; LOD, limit of detection (signal/noise ratio ) 3); LOQ, limit of quantification (signal/noise ratio ) 10); M1-dG, pyrimido[1,2R]purine10(3H)one-2′-deoxyribose; MN, micrococcal nuclease; MRM, multiple reaction monitoring; 8-oxo-dG, 8-oxo-7,8-dihydro-2′-deoxyguanosine; RSD, relative standard deviation (SD/mean × 100); SPD, spleen phosphodiesterase.
10.1021/tx0101595
Figure 1. Structures of target oxidative DNA adducts and sites of labeling in internal standards.
capable of producing M1-dG (3). Alternatively, M1-dG can be formed by reaction of dG moieties in DNA with malondialdehyde, a product of lipid peroxidation and thromboxane biosynthesis (4). In addition, mechanisms have been proposed for the formation of etheno-DNA adducts through reaction of DNA bases with epoxy derivatives of R,β-unsaturated aldehydes derived from decomposition of polyunsaturated fatty acid hydroperoxides (5, 6). Oxidative DNA adducts represent premutagenic lesions in bacterial and eukaryotic cells because of their miscoding capabilities (-dC, -dA, ref 7; M1-dG, ref 8; 8-oxo-dG, ref 9). A number of enzyme systems that repair
This article not subject to U.S. Copyright. Published 2002 by the American Chemical Society Published on Web 09/28/2002
1296
Chem. Res. Toxicol., Vol. 15, No. 10, 2002
oxidative DNA adducts have been conserved across evolution consistent with an important role for efficient removal of such damage (10). Evidence from studies of humans and experimental animals suggests that steadystate levels of some oxidative DNA adducts (M1-dG, ref 11; -adducts, ref 12), but not 8-oxo-dG (13) can be affected by the composition of dietary fat. Similarly, agedependent accumulation of 8-oxo-dG in rodents has been reported (14). Chronic inflammation is another significant source of reactive oxidants, and studies have linked the inflammation associated with viral infections with increased levels of 8-oxo-dG and increased incidences of hepatocellular carcinoma in rodents (15) and humans (16). These findings and others form the basis for the hypothesis that oxidative DNA adducts are important causative factors in, and biomarkers for, a number of chronic degenerative disease states in humans and animals, including cancers that develop in the absence of known carcinogen exposure and may be related to dietary and lifestyle factors (14, 17). Progress on quantitative analysis of oxidative DNA adducts has been uneven. The relatively high steadystate levels of 8-oxo-dG (ca. one in 105 normal nucleotides) have facilitated measurement in many laboratories using LC-ECD (18) and more recently, LC-ES/MS/MS (19). However, etheno-DNA adducts and M1-dG are present at much lower levels (ca. one in 107-9 normal nucleotides), requiring considerably more analytical sensitivity. Most measurements of etheno-DNA adducts have been conducted using 32P-postlabeling in conjunction with immunoaffinity purification (20). Significant progress has been made in the analysis of M1-dG using GC-NICI/MS (21) and more recently using a slot-blot immunoassay (22). While immunochemical and 32P-postlabeling methods can be highly sensitive, they are often hampered by limitations in specificity and interassay precision. Alternatively, MS methods can be highly specific and precise, particularly when performed with stable-labeled internal standards, but typically have lower sensitivity. We have recently used LC-ES/MS/MS for the quantification of -dA and -dC in rodent and human DNA at levels approaching one adduct in 108 normal nucleotides (23). Another significant aspect of this method was the ability to analyze directly crude enzymatic hydrolysates from 100 µg of DNA or less using on-line solid-phase extraction for automated analysis. We now report extension of these procedures to permit concurrent analysis of 8-oxo-dG, M1-dG, -dA, and -dC from a single aliquot of hydrolyzed DNA.
Experimental Section Reagents. Sigma Chemical Co. (St. Louis, MO) supplied the 8-oxo-dG, thymidine kinase, and purine nucleoside phosphorylase, DNA, and DNA hydrolytic enzymes used in this study. Preparation of MDA-Modified DNA. Salmon testes DNA containing undetectable endogenous levels of M1-dG was reacted with different concentrations of MDA (0.2, 2, 20, 200, and 2000 µM) using the conditions previously described (22). The M1-dG adduct levels in normal 108 nucleotides were determined to be ND, 5, 29, 230, 2800, and 82 000, respectively. DNA Isolation and Hydrolysis Procedures. DNA was isolated from human liver (ca. 0.5 g) obtained from the Cooperative Human Tissue Network that is funded by the National Cancer Institute. Alternatively, human liver samples were obtained from the Liver Tissue Procurement and Distribution System (University of Minnesota, Minneapolis, MN). Liver
Churchwell et al. tissue, obtained following surgery, was quick frozen in liquid nitrogen, shipped on dry ice, and stored at -70 °C until analyzed. Assurances were given that samples were isolated and quick-frozen at -70 °C within 2 h postsurgery. A determination of normal, as opposed to diseased, status of the liver was made by pathological examination and determination of viruses (HBV, HCV) after surgery. Male and female Sprague-Dawley rats were obtained from the NCTR colony. The same DNA isolation, quantification, and hydrolysis procedures (MN/SPD) were used as those previously described for rodent liver DNA (23). Hydrolysis of DNA was shown by using LC-UV analysis to be quantitative for calf thymus and mouse liver DNA (23). Particular care was taken to ensure that the temperature of tissue homogenates and DNA solutions remained at or below ca. 4 °C whenever possible to minimize artifactual formation of 8-oxodG (24). For example, livers were homogenized in ice-cold sucrose buffer, centrifuges were refrigerated, and mitochondrial pellets were suspended in ice-cold sucrose buffer; however, as in previous studies, DNA hydrolysis reactions were conducted at 37 °C and extractions were done at room temperature (note: extractions conducted on ice had no observed effects on adduct levels). Adenosine deaminase incubations, conducted at 4 °C, were used to reduce the concentration of dA in DNA hydrolysates by conversion to inosine (23). Preparation of Labeled Internal Standards. Labeled 8-oxo-dG and M1-dG were synthesized enzymatically using thymidine kinase/nucleoside phosphorylase (25). The 13C5,15N2thymidine (Cambridge Isotope Laboratories, Andover, MA) was uniformly labeled in the deoxyribose moiety and the 13C5-labeled adducts were purified by using preparative HPLC. The M1-dG was composed of both N-7 and N-9 regioisomers. Unlabeled M1dG was similarly prepared from unlabeled thymidine. Uniformly labeled 15N5--dA and 15N3--dC were prepared as previously described (23). The chemical concentrations of labeled adducts were determined by using LC with UV detection (260 nm) in comparison with authentic unlabeled analogues. The isotopic distributions were determined using full scan LC-ES/MS (m/z 100-600). No unlabeled adducts were present in the labeled analogues (
