Endogenous Lipid Hydroperoxide-mediated DNA-adduct Formation in Min Mice*

Despite intensive research over the last two decades, there are still no specific markers of endogenous lipid hydroperoxide-mediated DNA damage. We recently demonstrated that heptanone-etheno-2′-deoxyguanosine adducts are formed in the DNA of rat intestinal epithelial cells that stably express cyclooxygenase-2. Heptanone-etheno adducts can only arise from the reaction of lipid hydroperoxide-derived 4-oxo-2(E)-nonenal with DNA. This raised the possibility that similar adducts would be formed in vivo in settings where cyclooxygenase-2 expression is increased. Therefore, DNA-adduct formation was studied in C57BL/6JAPCmin mice, a colorectal cancer mouse model in which cyclooxygenase-2 is up-regulated. 15(S)-Hydroperoxy-5Z,8Z,11Z,13E-eicosatetraenoic acid is the major lipid hydroperoxide produced endogenously by cyclooxygenase-2. It undergoes homolytic decomposition to the DNA-reactive bifunctional electrophile 4-oxo-2(E)-nonenal, which forms heptanone-etheno adducts with DNA. A quantitative comparison was made of the heptanone-etheno-DNA adducts present in C57BL/6J and C57BL/6JAPCmin mice. Using highly specific and sensitive methodology based on stable isotope dilution liquid chromatography/tandem mass spectrometry, we have detected the endogenous formation of heptanone-etheno adducts in mammalian tissue DNA for the first time. In addition, we found that there were statistically significant increased levels of the heptanone-etheno-2′-deoxyguanosine and heptanone-etheno-2′-deoxycytidine adducts in the C57BL/6JAPCmin mice when compared with the control C57BL/6J mice.

Likewise, NSAIDs have demonstrated their efficacy to inhibit carcinoma formation in animal models (5)(6)(7)(8). The molecular mechanism for the ability of NSAIDs to inhibit tumor formation is not known. However, most NSAIDs inhibit both COX-1 and COX-2 (9). The COX-1 enzyme, which is constitutively expressed in nearly all tissues in the body, predominantly plays a housekeeping role by mediating normal physiologic processes such as protection of the stomach and platelet aggregation (10). The other isoform, COX-2, is inducible by various mitogens, growth factors, and mediators of inflammation in certain cells and tissues (11).
Although COX-1 is constitutively present in normal colon tissue, COX-2 levels are low to undetectable. In contrast, COX-2 mRNA is up-regulated in tumor tissue, and COX-2 protein is expressed in high amounts (12). In rats that have been subjected to carcinogens, COX-2 mRNA and protein levels were increased in the tumors (13). Furthermore, cells overexpressing COX-2 have an altered cellular adhesion pathway, fail to undergo apoptosis, have increased invasiveness, and cell cycle prolongation (14 -16). All of these changes were reversed by the addition of an NSAID, suggesting that COX-2-derived metabolites play an important role in tumorigenesis (14 -16). In animal models, the absence of COX-2 expression inhibited polyp formation to 84% of control (17), and overexpression of the COX-2 gene in transgenic mice was sufficient to induce polyp formation in Ͼ85% of the mice (18).
Arachidonic acid is the preferred substrate for COX-2. It catalyzes the bis-dioxygenation of arachidonic acid to form PGG 2 , a cyclic endoperoxide with a hydroperoxide at C-15. This is followed by reduction of the C 15 -hydroperoxide to give PGH 2 (the peroxidase reaction). PGH 2 is the precursor to the formation of thromboxane A 2 , prostacyclin, and other PGs (9,19). When mice were fed eicosapentaenoic acid, an -3 fatty acid, this resulted in a significant reduction in intestinal arachidonic acid content and PGE 2 levels as well as a 64% reduction in polyp load (20). In addition, when cytosolic phospholipase 2, a phospholipase that has been well characterized as a major arachidonic acid releasing enzyme, was deleted in Min mice, there was an 83% reduction in polyp number (21).
The ability to inhibit polyp formation by NSAIDs and the reduction of arachidonic acid availability suggested that eicosanoid production may be very important in promoting carcinogenesis. In addition, eicosanoids have the ability to modulate various steps involved in carcinogenesis such as apoptosis and angiogenesis. However, eicosanoids are also produced by COX-1, which is present in both affected and unaffected tissue. This suggests that there may be a link between polyp formation and COX-2 up-regulation through a pathway that does not involve the biosynthesis of eicosanoids.
COX-2 can convert arachidonic acid into 15-hydroxy-5,8,11,13-(Z,Z,Z,E)-eicosatetraenoic acid (15-HETE) (22), which is derived from 15-HPETE, a prototypic -6 polyunsaturated fatty acid-derived lipid hydroperoxide. In settings of oxidative stress, where reducing pathways were compromised, 15 (24,25). HPNE was shown to be the immediate precursor of ONE and HNE (22, 23, 26 -28). It was recently demonstrated that HPNE also formed unsubstituted etheno adducts with double-stranded DNA (29). EDE also formed unsubstituted etheno adducts (25), and ONE formed heptanone-etheno adducts (30 -32) (Scheme 1). However, HPNE was 10 times more efficient in the formation of unsubstituted etheno adducts than EDE, suggesting that it is the primary source for the formation of these adducts (29). Unsubstituted etheno adducts are highly mutagenic in mammalian cells and have been detected in human DNA (33,34). Recently, we showed that the ONE-derived heptanone-etheno adduct, H⑀dCyd, was highly mutagenic in both bacteria and human cells. 3 Interestingly, there were substantial differences in mutational frequency and base substitution between bacteria and human cells. 3 In earlier studies we showed that vitamin C induces the decomposition of lipid hydroperoxides to DNA-reactive bifunctional electrophiles. This made it possible to develop an in vitro system to show that DNA damage occurred in rat intestinal epithelial cells stably expressing COX-2. There was a dose-dependent increase in H⑀dGuo adduct formation in the presence of vitamin C in these cells (35). H⑀dGuo adduct formation together with 15-(S)-HPETE biosynthesis was inhibited by a specific COX-2 inhibitor (35). Therefore, there is now substantial evidence that COX-2-mediated DNA damage may play a role in tumorigenesis.
Min mice are one of the most commonly used mouse models for colorectal cancer. These mice spontaneously develop large numbers of polyps. In addition, up-regulation of COX-2 in the small intestine in these mice has been well characterized (36). Therefore, this mouse model and its wild type, non-COX-2 expressing counterpart were used to examine whether heptanone-etheno adducts were present in the DNA. A quantitative stable isotope dilution LC/ESI/MS/MS assay was used to quantify heptanone-etheno adducts in the small intestinal tissue.

EXPERIMENTAL PROCEDURES
Materials and Reagents-Ammonium acetate, activated calf thymus, vitamin C, DNase I, and zinc chloride were purchased from Sigma-Aldrich. 15(S)-HPETE was obtained from Cayman Chemical (Ann Arbor, MI). High performance liquid chromatography grade water and acetonitrile were obtained from Fisher. Nuclease P1 and shrimp alkaline phosphatase (SAP) were obtained from Roche Diagnostics. 13 C 15 N-Labeled DNA with Ͼ99% isotopic purity, prepared by growing algae in 13 C 15 N-enriched media, was supplied by Spectra Stable Isotopes (Columbia, MD). Gases were supplied by BOC Gases (Lebanon, NJ).
Animals-C57BL/6J and C57BL/6JAPC min mice were obtained as gifts from Dr. Colin Funk of Queen's University, Kingston, ON, Canada. Mice were fed ad libitum and sacrificed by CO 2 asphyxiation. The entire small intestine was removed and immediately frozen at Ϫ80°C until required for DNA extraction.
Mass Spectrometry-Mass spectrometry for the quantitative analysis of five DNA adducts was conducted with a Finnigan TSQ Quantum Ultra AM spectrometer (Thermo Electron Corp., San Jose, CA) equipped with an ESI source in the positive ion mode. Operating conditions were: spray voltage, 4.5 kV; heated capillary temperature, 350°C. Nitrogen was used for the sheath gas and auxiliary gas set at 60 and 10 (in arbitrary units), respectively. Collision-induced dissociation was performed using argon as the collision gas at 1.5 milliTorr in the second (radio frequency-only) quadrupole. An additional direct current offset voltage was applied to the region of the second multipole ion guide (Q0) at 5 V to impart enough translational kinetic energy to the ions so that solvent adduct ions dissociate to form sample ions. MRM analysis at unit resolution was maintained for both precursor and product ions. The following MRM transitions were monitored: C⑀dGuo (m/z 476 3 360; 15   Liquid Chromatography-LC/UV chromatography for quantitative DNA base analysis was conducted using gradient system 1 on a Hitachi L-2130 Intelligent Pump equipped with a Hitachi L2450 diode array detector (Hitachi, San Jose, CA). The separation employed a Phenomenex Jupiter 5-m C18 column (250 ϫ 4.6 mm inner diameter, 5 m; Phenomenex, Torrance, CA). Solvent A was water, and solvent B was acetonitrile. The gradient was as follows; 0% B at 0 min, 0% B at 5 min, 8% B at 25 min, 100% B at 26 min, 100% B at 36 min, 0% B at 37 min, and 0% B at 52 min. The flow rate was 1 ml/min. The separation was performed at ambient temperature.
LC/ESI/MS analysis for DNA adducts on gradient system 2 was performed using a Hitachi L-2130 Intelligent Pump. The separation employed a Phenomenex Luna 3-m C8 column (150 ϫ 4.6-mm inner diameter, 3 m). Solvent A was 5 mM ammonium acetate in water, and solvent B was 5 mM ammonium acetate in acetonitrile. The gradient was as follows: 6% B at 0 min, 6% B at 2 min, 9% B at 12 min, 55% B at 22 min, 80% B at 30 min, 80% B at 33 min, 6% B at 35 min, and 6% B at 45 min.
The flow rate was 0.3 ml/min. The separation was performed at ambient temperature.
Extraction of DNA from Mouse Tissue-DNA isolation from the small intestine of mice was performed using a Wako DNA extraction WB kit (Wako Chemicals, Richmond, VA). Briefly, tissue was homogenized in lysis     solution and then treated with an enzyme reaction solution, RNase, and protease to liberate the DNA from the nucleus. This was followed by NaI extraction and several wash steps to purify the DNA. Typically several mg of DNA were obtained. It was hydrolyzed as described below, and normal bases were quantified by LC/UV using system 1.
Hydrolysis of DNA and Isolation of DNA Adducts-The extracted DNA was combined and dissolved in Chelex-treated 10 mM Tris buffer (22) containing 100 mM MgCl 2 (pH 7.4, 1 ml). DNase I (556 units) dissolved in 10 mM MOPS containing 120 mM MgCl 2 (pH 7.0) was added and incubated at 37°C for 1.5 h. At the end of the incubation the pH was adjusted to 9.0 by adding 0.1 M glycine buffer (pH 10.0), nuclease P1 (15.5 units) was added, and incubation was continued for a further 2 h at 37°C. 50 mM Tris-HCl (pH 7.4) was then added followed by alkaline phosphatase (30 units) in SAP 10ϫ buffer, and incubation was conducted for 2 h at 37°C. The samples were filtered through a 0.2-m Costar cartridge. At this time an aliquot was removed for LC/UV analysis of the bases using gradient system 1. Quantitation of DNA bases was carried out by constructing standard curves of known amount of bases. The remaining sample was applied to a solid phase extraction cartridge (6 ml, Supelclean LC-18, Supelco, Bellefonte, PA) that had been prewashed with acetonitrile (18 ml) and water (18 ml). The cartridge was then washed with water (4 ml) and a methanol/water mixture (1 ml, 5:95 v/v). Adducts were eluted with an acetonitrile/water mixture (6 ml, 1:1 v/v). The eluates were evaporated to dryness under nitrogen.
Analysis of Adducts-The adducts were redissolved in water/acetonitrile (100 l 94:6 v/v). Aliquots (10 l) were analyzed by LC/ESI/ MRM/MS using gradient system 2. The 15 N 5 -or 15 N 3 -labeled internal standards were added to the samples before hydrolysis. Quantitation was performed from standard curves constructed by the ratio of known amounts of authentic standards and internal standards. For the MRM analysis, parent ions were monitored for the protonated molecular ion of the endogenous adduct and for its 15 N 5 or 15 N 3 internal standard. Product ions that were monitored corresponded to the loss of the 2Ј-deoxyribose moiety (BH 2 ϩ ) from the endogenous adduct and its 15 N 5 or 15 N 3 internal standard. Adduct levels were normalized to the amount of DNA as detected by base analysis mentioned above.
Artifact Formation during Isolation and Analysis-Duplicate small intestine samples from Min mice were spiked with 13 C 15 N-labeled DNA (520 g) before defrosting. The amount of unlabeled DNA in each sample was estimated as 650 g. Labeled and unlabeled DNA were extracted from the tissues, hydrolyzed in the presence of 15 N internal standards using DNase, nuclease P1, and alkaline phosphatase. DNA adducts were isolated by solid phase extraction as described above. The     (22) was reacted with vitamin C (2 mM) in the presence of calf thymus DNA (5 mM) for 24 h at 37°C. The reaction was placed on ice for 15 min. DNA was precipitated by cold ethanol and 2.9 M sodium acetate. It was hydrolyzed as described above, and normal bases were quantified by LC/UV using system 1.
MS Analysis of C57BL/6J Mice-DNA was isolated from the entire small intestine of C57BL/6J mice. The DNA was subjected to enzyme hydrolysis in the presence of 15 N 5 -or 15 N 3 -labeled internal standard. A portion of hydrolysate was removed for the quantitation of normal bases, which was used to determine the amount of DNA in the sample. The endogenous DNA adducts were isolated using solid phase extraction from the remaining hydrolysate. LC/ESI/MS/MS in the MRM mode was conducted. A typical chromatogram for the C57BL/6J mouse DNA adducts and their internal standards is shown in Figs. 3 and 4. The DNA adducts were quantified using a calibration curve and normalized by the amount of DNA that was extracted from the small intestine. The number of adducts detected for the C57BL/6J mice (n ϭ 7) is C⑀dGuo (below detection limit), C⑀dCyd (below detection limit), H⑀dAdo (below detection limit), H⑀dCyd (0.1 adducts/10 7 normal bases), and H⑀dGuo (0.6 adducts/10 7 normal bases) ( Table 1 and Fig. 5).
Artifact Formation during Isolation and Analysis of DNA Adducts-The potential artifactual formation of heptanone-etheno-DNA adducts was examined by adding 13 C 15 N-labeled DNA to the tissue sample before DNA extraction was conducted. DNA was then isolated from the tissue and hydrolyzed using the normal work up procedure. 15 N 5 -Labeled H⑀dGuo, 15 N 5 -labeled H⑀dAdo, and 15 N 3 -labeled H⑀dCyd internal standards were added to the DNA hydrolysates. Amounts of the three 13 C 15 N-labeled heptanone-etheno adducts were all Ͻ0.05% of the relevant 15 N internal standard. This showed that no artifactual adduct formation had occurred (Fig. 8,  A-C). In contrast, the expected amounts of endogenous H⑀dGuo (Fig. 8A) and H⑀dCyd (Fig. 8C) were observed. As expected, the amount of endogenous H⑀dAdo was below the limit of detection of the assay (Fig. 8B).
DNA Adducts Formed in Calf Thymus DNA-15(S)-HPETE was decomposed in the presence of vitamin C and calf thymus DNA. LC/ESI/MRM/MS was conducted for the quantification of the adducts formed. The most predominant adducts formed were the heptanoneetheno adducts, specifically H⑀dCyd (650 adducts/10 6 normal bases). H⑀dAdo (300 adducts/10 6 normal bases) was the next most abundant, and H⑀dGuo (15 adducts/10 6 normal bases) was the least abundant. No carboxynonanone-etheno adducts were detected (Fig. 9).

DISCUSSION
Over the last two decades there has been a substantial effort to identify lipid hydroperoxide-derived endogenous DNA adducts as potential biomarkers for cancer. Using highly specific and sensitive stable isotope dilution methodology in combination with analysis by gas chromatography/MS or LC/MS, it has been possible to quantify ⑀Gua, ⑀dGuo,

Endogenous DNA Adducts in Min Mice
⑀dAdo, ⑀Ade, ⑀dCyd, ⑀Cyt, and the propano adduct M 1 G in mammalian tissue DNA or urine ( Table 2). Recently we showed that the unsubstituted etheno adducts most likely arise from HPNE, the initial product of homolytic lipid hydroperoxide decomposition (25). They can also arise from EDE (29) or 2,3-epoxy-4-hydroxynonanal, the oxidation product of HNE (37), and oxidized trans,trans-2,4-decadienal (38,39). However, EDE is much less reactive toward DNA than HPNE. Furthermore, the formation of 2,3-epoxy-4-hydroxynonanal and oxidized trans,trans-2,4-decadienal requires in vivo oxidation to occur. The possibility that in vivo oxidation of HNE and trans,trans-2,4-decadienal occurs more rapidly than detoxification by glutathione S-transferases and aldo-keto reductases seems highly unlikely (40). Unfortunately, unsubstituted etheno adducts are also formed during metabolism of environmental chemical such as vinyl chloride (41), vinyl fluoride (42), chloroethylene oxide, and chloroacetaldehyde (43). This makes it difficult to distinguish endogenous production from environmental exposure. Malondialdehyde can arise during thromboxane A 2 biosynthesis as well as during lipid peroxidation so it may contribute to the formation of M 1 G (44). In addition, M 1 G can also arise from base propenals formed during free radical-mediated damage to the sugar backbone of DNA (45). Therefore, M 1 G is not a specific marker of lipid hydroperoxide-mediated DNA damage. These additional pathways for formation may have contributed to the lack of correlation of M 1 G formation with oxidative stress in the CCl 4 rat model (46).
Several years ago we identified ONE as a major product of homolytic lipid hydroperoxide decomposition (28). It is extremely reactive toward DNA bases forming heptanone-etheno adducts (31)(32)(33). Such adducts cannot arise from the metabolism of environmental chemicals. Furthermore, we have recently demonstrated that H⑀dCyd is highly mutagenic in both mammalian and bacterial systems. 3 Therefore, heptanoneetheno adducts could potentially serve as specific biomarkers of endogenous lipid hydroperoxide-mediated DNA damage. Using the Min mouse model of colon cancer, we have now demonstrated that heptanone-etheno adducts are present in both affected and unaffected colon tissue. DNA was isolated from the entire small intestine of C57BL/6J and C57BL/6JAPC min mice and then hydrolyzed in the presence of 15 N 5 -or 15 N 3 -labeled internal standards. The entire small intestine was analyzed to overcome the difference in the distribution of polyps (17). LC/ESI/MRM/MS was conducted to provide maximal sensitivity and specificity. Quantitation was performed using standard stable isotope dilution methodology. In separate experiments it was shown that the adducts were not generated as an artifact during the isolation and hydrolysis of the DNA. A typical chromatogram for the DNA adducts and their stable isotope internal standards is shown in Figs. 1 and 2.
A quantitative comparison of adduct formation in C57BL/6J and C57BL/6JAPC min mice DNA was made. There were statistically significant increased levels of the adducts that we were able to quantitate in the C57BL/6JAPC min mice versus the C57BL/6J mice (Table 1 and Fig.  5). These data correlate with our cell culture experiments (35), showing that increased levels of COX-2, as is the case in the small intestine of the C57BL/6JAPC min mice (36), result in increased levels of the adducts. H⑀dGuo was increased from 0.6 adducts/10 7 normal bases to 1.8 adducts/107 bases (Table 1). H⑀dCyd was also increased from 0.1 adducts/10 7 normal bases to 1.1 adducts/10 7 normal bases ( Table 1). The peaks for C⑀dGuo and C⑀dCyd were in the range of noise and close to the zero point of the assay, making it impossible to accurately quantify any adducts that may have been present (Figs. 3 and 6). H⑀dAdo levels were close to the lower limit of detection, making the analysis difficult (Figs. 4 and 7). However, there appeared to be a trend toward an increase in the Min mice.
These data suggest that the H⑀dGuo and H⑀Cyd adducts may be good markers for the detection of colorectal cancer. The major adducts identified in calf thymus DNA after treatment with 15(S)-HPETE were the heptanone-etheno adducts. H⑀dCyd was found at levels of 650 adducts/ 10 6 normal bases. H⑀dAdo was approximately half as abundant, with 300 adducts/10 6 normal bases, and H⑀dGuo was by far the least abundant heptanone adduct, with only 15 adducts/10 6 normal bases. The unsubstituted etheno adducts were much less abundant then the heptanone-etheno adducts (data not shown). No carboxynonanone-etheno adducts were detected. This was expected because 15(S)-HPETE cannot rearrange to form the 9,12-dioxo-10(E)-dodecenoic acid precursor necessary to form the carboxynonanone-etheno adducts. When comparing DNA adducts in calf thymus DNA versus the mouse tissue, there was a difference in the most predominant adducts formed. This suggests that there are repair enzymes for these adducts and that the repair mechanisms favor certain adducts. H⑀dGuo was the most abundant adduct in tissue and, therefore, may be useful as a cell and tissue biomarker. H⑀dAdo and H⑀dCyd were the least abundant adducts, suggesting that they were repaired more readily. Therefore, they are potential candidates for urinary biomarkers.
In summary, we have detected the endogenous formation of heptanone-etheno DNA adducts for the first time in mammalian tissue. Using highly specific and sensitive LC/ESI/MRM/MS methodology, it was possible to quantify the amount of adducts present in the colon tissue of normal C57BL/6J mice and C57BL/6JAPC min mice, a colorectal cancer model. There were statistically significant increased levels of adducts in the Min mice when compared with the wild type mice. This suggests that heptanone-etheno adducts in colon tissue DNA may serve as a biomarker of increased risk for colorectal cancer. Furthermore, there was a difference in the predominant adducts formed between an in vitro system and the in vivo mouse model, which suggests that DNA repair enzymes may favor certain DNA adducts. The adducts that are efficiently repaired may ultimately be excreted in the urine. Analysis of these adducts may provide biomarkers of endogenous lipid hydroperoxide-mediated DNA damage in a similar manner to the well characterized urinary DNA adducts of aflatoxin that have served so well as biomarkers of aflatoxin-mediated DNA damage (52).