Cyclooxygenase-2-mediated DNA Damage* □ S

Rat intestinal epithelial cells that express the cy-clooxygenase-2 (COX-2) gene permanently (RIES cells) were used as a model of in vivo oxidative stress. A targeted lipidomics approach showed that 15( S )-hydroxyei-cosatetraenoic acid (15( S )-HETE) was the major hydroxylated non-esterified lipid formed in unstimulated intact cells. The corresponding hydroperoxide, 15( S )-hydroperoxyeicosatetraenoic acid (15( S )-HPETE) undergoes homolytic decomposition to the DNA-reactive bifunctional electrophile 4-oxo-2( E )-nonenal, a precursor of heptanone-etheno-2 (cid:2) -deoxyguanosine. This etheno adduct was identified in the DNA of RIES cells. A dose-dependent increase in adduct levels was observed in the presence of vitamin C. This suggested that vitamin C increased lipid hydroperoxide-mediated 4-oxo-2( E )-non-enal formation in the cells. The selective COX-2 inhibitor NS-398 was protective against cellular DNA damage but was less effective if vitamin C was present. Pros-taglandin E 2 and 15( S )-HETE biosynthesis were com- pletely inhibited by 110 (cid:1) M NS-398 in the intact RIES cells. No inhibition of COX-1 was

Lipid hydroperoxide-derived bifunctional electrophiles are formed by three distinct pathways (Scheme I). The first pathway involves ␣-cleavage of an alkoxy radical and results in formation of t-EDE and c-EDE (16,17). The second pathway involves formation of HPNE, which undergoes reduction to HNE or a 2-electron oxidation to ONE (11-13, 18 -20). The third pathway appears to involve the intermediate formation of 9-hydroperoxy-12-oxo-10-decenoic acid, which undergoes a 2-electron oxidation to DODE (14), a reaction observed in the conversion of HPNE to ONE (11)(12)(13).
In a series of studies, ONE was shown to mediate the formation of heptanone-etheno-dGuo, heptanone-etheno-dAdo, and heptanone-etheno-dCyd adducts (21)(22)(23)(24), whereas EDE was found to mediate the formation of unsubstituted etheno-dGuo and etheno-dAdo adducts (17). However, HPNE is almost 10 times more efficient than EDE in forming unsubstituted etheno adducts (25). Therefore, HPNE appears to the major source of unsubstituted etheno adducts that arise from lipid peroxidation. Etheno-dGuo is mutagenic in mammalian cells (AA8 Chinese hamster ovary), inducing base pair mutations, with a preference for G to A transitions (26). Etheno-dAdo is more mutagenic in human cells (HeLa) than 8-oxo-dGuo, inducing A to T transversions in experiments using modified double-and single-stranded DNA substrates (27). In contrast to ONE, HNE is not very reactive toward DNA (28). However, both HNE and ONE can readily modify nucleophilic amino acid residues in proteins (29 -33). For example ONE forms a novel cyclic peptide in histone H4 by reaction with proximal histidine and lysine residues in the protein (31). DODE is the presumed intermediate in the formation of 9-carboxynonan-2-one-etheno DNA adducts with dGuo (14), 2Ј-deoxyadenosine, and 2Ј-deoxycytidine (15).
Numerous studies have implicated COXs as mediators of carcinogenesis. In addition, there is a correlation between the use of non-steroidal anti-inflammatory drugs and a reduction in risk for colon and breast cancer (34 -36). COX-2 is up-regulated in many tumors when it is absent from unaffected surrounding tissue (37). COX-mediated formation of PGs has long been assumed to play a role in tumorigenesis (38). However, little attention has been given to the potential for the formation of genotoxic bifunctional electrophiles that result from COXmediated lipid hydroperoxide formation. We reasoned that cells permanently expressing the COX-2 gene (RIES cells) would produce HPETEs or HPODEs by COX-2-mediated metabolism of AA or LA. This would then make it possible to test whether the intracellular formation of lipid hydroperoxides could induce DNA adduct formation. RIE cells were chosen because they have been used extensively in studies of the mechanism by which COX-2 is involved in adhesion, cell proliferation, and apoptosis (39,40). Using RIES cells, it was possible to determine whether the lipid hydroperoxides were derived from AA or from LA by analyzing the resulting HETEs and HODEs using a targeted lipidomics approach that we developed recently (41). This method is based on stable isotope dilution LC/electron capture APCI/MS/MS methodology, which makes it possible to employ normal phase chiral chromatography with extremely high sensitivity (42). Specificity of the approach comes from the use of MRM analysis of a specific parent ion and a specific product ion for each analyte coupled with enantio-selective separation of the resulting lipids. If the lipid hydroperoxides were derived from AA it would then be possible (using specific inhibitors) to determine whether they arose from constitutive COX-1 or from COX-2. We report the use of targeted chiral lipidomics methodology to examine lipid hydroperoxide formation in RIES cells together with stable isotope dilution LC/MS quantitation of DNA adducts that were formed. We also report the regulation of DNA adduct formation by vitamin C, which has been shown previously to activate lipid hydroperoxides to endogenous genotoxins in vitro (11).  MS-A Finnigan TSQ 7000 triple stage quadrupole mass spectrometer (Thermo Electron, San Jose, CA) equipped with an APCI source was used in the studies. For full-scan and MRM analyses, unit resolution was maintained for both parent and product ions. For the lipidomics profile, the instrument was operated in the negative ion mode. Operating conditions for the TSQ 7000 were vaporizer temperature at 500°C and heated capillary temperature at 230°C, with the corona discharge needle set at 16 A. The sheath gas (nitrogen) and auxiliary gas (nitrogen) pressures were 40 p.s.i. and 10 (arbitrary units), respectively. Collision-induced dissociation was performed using argon as the collision gas at 2.7 mtorr in the Rf-only quadrupole. Targeted chiral LC/electron capture APCI/MRM/MS analysis was conducted using PFB derivatives of 21 lipids and 7 heavy isotope analog internal standards. For DNA adduct analysis, the instrument was operated in the positive ion mode. The TSQ 7000 operating conditions were vaporizer temperature at 550°C and heated capillary temperature at 160°C, with the corona discharge needle set at 8 A. The sheath gas (nitrogen) and auxiliary gas (nitrogen) pressures were 80 p.s.i. and 3 (arbitrary units), respectively. Collision-induced dissociation was performed using argon as the collision gas at 2.7 mtorr in the Rf-only quadrupole. LC/MRM/MS analysis was conducted using PFB derivatives. The following MRM transitions were monitored: heptanone-etheno-dGuo-PFB (m/z 584 3 468) and [ 15 N 5 ]-labeled heptanone-etheno-dGuo-PFB (m/z 589 3 473).
LC/UV chromatography for DNA bases was conducted using gradient-2 on a Hitachi L-6200A Intelligent Pump equipped with a Hitachi L4000 UV detector (Hitachi, San Jose, CA). The separation employed a Phenomenex Synergi 4-m polar reverse phase column (250 ϫ 4.6-mm inner diameter, 4 m). Solvent A was 5 mM ammonium acetate in water, and solvent B was 5 mM ammonium acetate in acetonitrile. Gradient 2 was as follows: 6% B at 0 min, 6% B at 3 min, 20% B at 9 min, 20% B at 10 min, 80% B at 12 min, 80% B at 15 min, 6% B at 17 min, and 6% B at 25 min. The flow rate was 1 ml/min. The separation was performed at ambient temperature using a linear gradient.
LC/APCI/MS analysis of DNA adducts using gradient 3 was performed on a Waters Alliance 2690 HPLC system (Waters Corp.). The separation employed an XTerra C18 column (250 ϫ 4.6-mm inner diameter, 5 m). Solvent A was 5 mM ammonium acetate in water, and solvent B was 5 mM ammonium acetate in acetonitrile. Gradient-3 was as follows: 6% B at 0 min, 6% B at 3 min, 20% B at 9 min, 20% B at 13 min, 40% B at 22 min, 70% B at 27 min, 80% B at 28 min, 80% B at 32 min, and 6% B at 34 min. The flow rate was 1.0 ml/min. The separation was performed at ambient temperature using a linear gradient. Incubations of Intact RIES Cells with Vitamin C-RIES cells were cultured in RPMI supplemented with 10% FBS, 2 mM glutamine, 100,000 units/liter penicillin, and 100,000 units/liter streptomycin until almost confluent. The medium was removed and replaced with medium containing increasing amounts of vitamin C (0.00, 0.03, 0.1, 0.5, 1.0, 5.0 mM in water). Cells were then incubated for 24-h at 37°C.

Incubations of Intact RIES Cells with NS-398 and Aspirin-Rat
Extraction of DNA from Intact RIES Cells-DNA isolation from 8 plates of RIES cells (total 8 ϫ 10 6 cells) was performed using a Wako DNA extraction WB kit (Wako Chemicals, Richmond, VA). Briefly, cells were lysed after a 24-h incubation and then treated with an enzyme reaction solution and protease to liberate the DNA from the nucleus. This was followed by the addition of aqueous sodium iodide extraction followed by several wash steps to purify the DNA. Typically 300 -500 g of DNA was obtained. It was hydrolyzed as described below, and normal bases were quantified by LC/UV using gradient 2.
Hydrolysis of DNA and Isolation of Heptanone-Etheno-dGuo-The DNA extracted from 8 plates of RIES cell was combined and dissolved in 10 mM MOPS containing 100 mM NaCl (pH 7.0, 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, nuclease P1 (15.5 units) was added along with 25 mM ZnCl 2 , and the incubation continued for a further 2 h at 37°C. Alkaline phosphatase (30 units) in 0.4 M MOPS (pH 7.8) was then added, and the incubation conducted for 1 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 2. 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 pre-washed 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). Heptanone-etheno-dGuo adducts were eluted with an acetonitrile/water mixture (6 ml, 1:1 v/v). The eluates were evaporated to dryness under nitrogen.
Incubation of Intact RIES Cells with Calcium Ionophore A23187-RIES cells were cultured in RPMI supplanted with 10% FBS, 2 mM glutamine, 100,000 units/liter penicillin, and 100,000 units/liter streptomycin until almost confluent. The medium was removed and replaced with medium containing 0.1% FBS and calcium ionophore A23187 (1 M final concentration). Cells were then incubated for 1 h at 37°C.
Incubations of WT RIE Cells with NS-398 -Cells were cultured in RPMI supplanted with 10% FBS, 2 mM glutamine, 100,000 units/liter penicillin, and 100,000 units/liter streptomycin until almost confluent. The medium was removed and replaced with medium containing 0.1% FBS or 0.1% FBS and NS-398 (110 M). Cells were then incubated for 24 h at 37°C.
Western Blot Analysis of RIES Cells during 24 h of Incubation-Cells were harvested at different time points and washed twice with ice-cold phosphate-buffered saline. They were suspended in lysis buffer containing 50 mM Tris (pH 8.0), 5 mM EDTA, 150 mM NaCl, 0.5% Nonidet P-40, 0.5 mM phenylmethanesulfonyl fluoride, and 1ϫ protease mixture for mammalian tissue (Sigma-Aldrich). The cells were lysed on ice and centrifuged at 10,000 ϫ g for 5 min. The supernatant was collected, and the protein concentration was determined by the Quick Start Bradford protein assay kit (Bio-Rad). Cellular protein (10 g) was loaded on pre-cast 8% NuPAGE Tris-Gly gels (Invitrogen) and then transferred to 0.45 mM nitrocellulose membranes (Invitrogen). The blots were blocked with 5% nonfat milk in Tris-buffered saline (Tris 100 mM (pH 7.5), NaCl 150 mM) containing 0.1% Tween 20 and then incubated with COX-2 antibody (Santa Cruz Biotechnology; Santa Cruz, CA) followed by reaction with secondary antibody, anti-goat (Santa Cruz Biotechnology). Protein bands were visualized with the ECL reagent (Amersham Biosciences).
RT-PCR Analysis of 15-LOX-2 in RIES Cells-Total RNA was extracted from the RIES cell culture samples (Ambion, Austin, TX) according to the manufacturer's instructions. The cells were harvested when cultures had reached full confluence. Conventional RT-PCR reactions were conducted on 2 g of total RNA. For 15-LOX-2, the primers used were 5Ј-CCT-GGT-CCA-ATA-TGT-TAC-CAT-3Ј (forward) and 5Ј-CCC-TGG-CTC-CTT-GCT-TAG-CAG-3Ј (reverse), which gives a 234-bp amplified product. Thirty cycles of reaction at 56°C for 30 s, 72°C for 30 s, and 95°C for 30 s were carried out on a thermal cycler (PerkinElmer Life Sciences). Wistar rat skin cDNA was used as a positive control for the RT-PCR reaction. It has been demonstrated previously that the 15-LOX-2 mRNA is expressed in this tissue. 2  Targeted Lipidomics Analysis from Intact RIES Cells-Standard curves were constructed for 15(S)-and 15(R)-HETE and PGE 2 on three separate days in the range 0.1-5 pmol/10 6 cells, and quality control samples were analyzed (see the supplemental material). Standard curves were constructed for all of the analytes in the range 6.7 pg/ml (0.02 pmol/ml) to 667 pg/ml (2.0 pmol/ml) (1 ml ϭ 10 5 cells). Typical regression lines for 11(R)-HETE, 15(R)-HETE, 15(S)-HETE, and PGE 2 were y ϭ 301.83x ϩ 132.54 (r 2 ϭ 0.9972), y ϭ 0.0028x ϩ 0.0285 (r 2 ϭ 0.9980), y ϭ 0.0049x ϩ 0.0279 (r 2 ϭ 0.9999), and y ϭ 0.0042x ϩ 0.0067 (r 2 ϭ 0.9990), respectively. The assay was validated by demonstrating for replicate quality control samples a precision of better than Ϯ15% and accuracy between 85 and 115% on three separate days. The limit of detection for 15(S)-and 15(R)-HODE were 0.15 pmol/10 6 cells and for PGE 2 was 0.10 pmol/ 10 6 cells (precision was better than Ϯ20%, and accuracy was between 80 and 120% on three separate days). Quantitation of the lipids from three separate incubations of the RIES cells showed the presence of 0.50 Ϯ 0.05 and 0.99 Ϯ 0.07 pmol/10 6 cells of 15(S)-HETE and PGE 2 , respectively (Table I).

Targeted Lipidomics Analysis of Intact RIES Cell
Targeted Lipidomics Analysis of RIES Cells in the Presence of Vitamin C-When the cells were treated with vitamin C (0.1 mM), LC/APCI/MRM/MS analysis of the cell culture medium showed a similar lipidomics profile (Fig. 3) as the RIES supernatants without vitamin C (Fig. 1). Quantitation (n ϭ 3) revealed the presence of 15(S)-HETE (0.39 Ϯ 0.05 pmol/10 6 cells) and PGE 2 (0.95 Ϯ 0.09 pmol/10 6 cells) (Table I). Therefore, a modest reduction of 15(S)-HETE was observed in the vitamin C-treated cells when compared with the untreated cells. There was no effect of vitamin C on PGE 2 biosynthesis.  (Table I) and to 0.25 Ϯ 0.01 and 0.50 Ϯ 0.01 pmol/10 6 cells, respectively, in the presence of vitamin C (Table I). At a concentration of 110 M NS-398, PGE 2 and 15(S)-HETE concentrations were below the lower limit of detection in the absence (Fig. 4, a and e) or presence of vitamin C (Fig. 5, a and e).
Targeted Lipidomics Analysis of Intact RIES Cells in the Presence of Aspirin-At a concentration of 200 M aspirin 15(S)-HETE and PGE 2 concentrations were below the lower limit of detection of the assay. However, aspirin induced the formation of 15(R)-HETE at a concentration of 0.69 Ϯ 0.01 pmol/10 6 cells. Vitamin C did not have any effect on aspirinmediated 15(R)-HETE formation (Fig. 6a) or on the inhibition of 15(S)-HETE and PGE 2 biosynthesis (Fig. 6, a and e).
Analysis of Heptanone-Etheno-dGuo from Intact RIES Cells-DNA was isolated from RIES cells after 24 h of incubation in media containing 0.1% FBS. The DNA was subjected to enzyme hydrolysis in the presence of 15 N 5 -labeled heptanoneetheno-dGuo internal standard. A portion of hydrolysate was removed for the quantitation of normal bases, by which the amount of DNA was determined. The endogenous DNA adducts

Cox-2-mediated DNA Damage
were isolated using solid phase extraction from the remaining hydrolysate. The heptanone-etheno-dGuo adduct and its 15 N 5 internal standard were then converted to PFB derivatives to improve their positive APCI ionization characteristics (Fig. 7A). LC/MS/MS in the MRM mode was conducted on heptanoneetheno-dGuo-PFB (m/z 584 3 m/z 468) and 15 N 5 -labeled heptanone-etheno-dGuo-PFB (m/z 589 3 m/z 473). A typical regression line for heptanone-etheno-dGuo over the range of 0.05-5 pg/ml was y ϭ 0.5012ϫ Ϫ 0.0041 (r 2 ϭ 0.9993). The assay was validated by demonstrating for replicate quality control samples a precision of better than Ϯ15% and accuracy between 85 and 115% on three separate days. The limit of detection was 0.05 pg/ml (precision was better than Ϯ20%, and accuracy was between 80 and 120% on three separate days). When the authentic standard of heptanone-etheno-dGuo was analyzed with its 15 N 5 internal standard, they eluted at the same retention time of 24.6 min (Fig. 7A). A typical chromatogram for the RIES cell-derived DNA adduct and its internal standard is shown in Fig. 7B. The DNA adducts were quantified using a calibration curve and normalized by the amount of DNA that was extracted from the RIES
Analysis of Heptanone-Etheno-dGuo from Intact RIES Cells Incubated with Vitamin C-RIES cells were treated with increasing amounts of vitamin C. As described above, after 24 h of incubation, the DNA was extracted and hydrolyzed. Endogenous DNA adducts were then isolated and converted to PFB derivatives for LC/APCI/MS analysis. A dose-dependent increase in DNA adduct formation was observed in the presence of vitamin C. At a concentration of 0.1 mM vitamin C, the level of heptanone-etheno-dGuo was increased over 3-fold to 6.6 adducts/10 7 normal bases from the basal level of 2.1 adducts/ 10 7 normal bases. Maximal adduct formation was observed with 1 mM vitamin C where the DNA adduct levels were 7.7 adducts/10 7 normal bases (Fig. 8).
Analysis of Heptanone-Etheno-dGuo from Intact RIES Cells Incubated with NS-398 -RIES cells were treated with two different concentrations of the selective COX-2 inhibitor NS-398 in the absence or in the presence of vitamin C (0.1 mM). NS-398 caused a dose-dependent decrease in heptanone-etheno-dGuo adducts in the absence of vitamin C (Fig. 9). At concentrations of 55 and 110 M NS-398, heptanone-etheno-dGuo levels were reduced to 1.0 and 0.7 adducts/10 7 normal bases, respectively. In the presence of vitamin C, the DNA adduct levels were 7.7, 4.8, and 1.8 adducts/10 7 normal bases for NS-398 concentrations of 0, 55, and 110 M, respectively (Fig. 9).
Analysis of Heptanone-Etheno-dGuo from Intact RIES Cells Incubated with Aspirin-At a concentration of 200 M aspirin, heptanone-etheno-dGuo levels were reduced to 0.5 adducts/10 7 normal bases. Vitamin C did not significantly affect the inhibition of adduct formation induced by 200 M aspirin.
Artifact Formation during Isolation and Analysis of Heptanone-Etheno-dGuo-The artifactual formation of heptanoneetheno-dGuo was examined by adding 15  Targeted Lipidomics Analysis of RIES Cell Lysates with Increasing Concentrations of AA-The RIES cell lysates were treated with increasing concentrations of AA for 30 min at 37°C, and the product profiles were analyzed by stable isotope dilution chiral LC/electron capture APCI/MS (see the supplemental material). 11(R)-HETE was the major product at all AA concentrations used (Fig. 10). The next most abundant metabolite was PGE 2 followed by 15(S)-HETE and 15(R)-HETE. At low AA concentrations there was only slightly more 11(R)-HETE than PGE 2 . However, as the concentration of AA increased there was a gradual increase in the ratio of 11(R)-HETE to PGE 2 . In contrast, the relative amounts of 11(R)-HETE and 15-HETEs were almost constant. The ratio of 15(S)-HETE (58%) to 15(R)-HETE (42%) was also constant from 1 to 100 M AA.
Targeted Lipidomics Analysis of Calcium Ionophore-treated Intact RIES Cells-Cells were treated with calcium ionophore A23187 for 1 h at 37°C, and the product profiles were analyzed by stable isotope dilution chiral LC/electron capture APCI/MS using the method described for the cell lysates. Analysis of PGE 2 in WT RIE Cells-Standard curves were constructed for PGE 2 on three separate days in the range 0.67 pg/ml (0.002 pmol/ml) to 66.7 pg/ml (0.2 pmol/ml) (see the supplemental material). The assay was validated by showing for replicate quality control samples a precision of better than Ϯ15% and accuracy between 85 and 115% on three separate days. The limit of detection for PGE 2 was 0.01 pmol/10 6 cells  Western Blot Analysis of RIES Cells during 24 h of Incubation-During 24 h of incubation the amount of COX-2 protein remained constant (Fig. 11A).

RT-PCR Analysis of 15-LOX-2 in RIES Cells-
The expected 234-bp amplification product was found in Wistar rat skin cDNA, used as the positive control (Fig. 11B). However, no amplification of 15-LOX-2 was observed from either the 1-or 5-g samples obtained from RIES cell preparations (Fig. 11B). DISCUSSION A targeted chiral lipidomics analysis of the RIES cell media revealed that the major hydroxylated polyunsaturated fatty acid was 15(S)-HETE with a trace amount of 15(R)-HETE (Fig.  1). Retention times of the analytes relative to their deuterated internal standards were identical to those obtained from a standard mixture (Fig. 2). For example the relative retention times of 15(R)-HETE and 15(S)-HETE to the relevant 15(S)-[ 2 H 4 ]HETE internal standard were 0.79 Ϯ 0.01 and 0.99 Ϯ 0.01, respectively. As predicted for normal phase chromatography, 15(S)-HETE eluted with a slightly shorter retention time than the deuterated 15(S)-HETE analog.
These data suggested that the predominant lipid hydroperoxide formed by RIES cells was 15(S)-HPETE. No COX-2derived 13(S)-HODE was detected, which indicated that 13(S)-HPODE did not contribute to DNA adduct formation (Scheme II). As expected, PGE 2 was also secreted by the RIES cells into the media (Fig. 1) together with small amounts of PGF 2␣ (data not shown). Quantitatively, there were approximately two times as much PGE 2 secreted as 15(S)-HETE (Table I) (Table I). This implied that COX-2-mediated 15(S)-HPETE formation was also decreased. NS-398 at a concentration of 55 M caused a reduction in both 15(S)-HETE and PGE 2 biosynthesis in the presence of 0.1 mM vitamin C (Table  I). 15(S)-HETE and PGE 2 formation was completely inhibited at 110 M NS-398 in the absence (Fig. 4) or presence of vitamin C (Fig. 5).
When the RIES cells were incubated with aspirin in the absence or presence of vitamin C, both PGE 2 and 15(S)-HETE biosynthesis were completely inhibited (Fig. 6). However, 15(R)-HETE was detected in amounts that were slightly higher than the original 15(S)-HETE that was observed in the absence of aspirin. This occurred in the absence or in the presence of vitamin C. Previous studies have required the use of exogenous AA either added to cells or to purified enzyme preparations to characterize this change in enantio-selectivity (Table II). Therefore, our new electron capture APCI methodology has confirmed that 15(R)-HETE and its presumed precursor 15(R)-HPETE are formed from endogenous AA in aspirin-treated cells that express the COX-2 enzyme. DNA was isolated from RIES cells that had been incubated for 24 h in media containing 0.1% FBS and then hydrolyzed in the presence of 15 N 5 -labeled-heptanone-etheno-dGuo internal standard. The ONE-derived heptanone-etheno-dGuo adduct and its 15 N 5 -labeled internal standard were then converted to PFB derivatives to improve their positive APCI ionization characteristics. LC/APCI/MS/MS in the MRM mode was then conducted to provide maximal sensitivity and specificity. In separate experiments it was shown that heptanone-etheno-dGuo was not generated as an artifact during the DNA isolation, hydrolysis, and derivatization procedure. A typical chromatogram for the RIES cell-derived DNA adduct and its stable isotope internal standard is shown in Fig. 7B. In the absence of vitamin C, heptanone-etheno-dGuo levels were 2.1 adducts/10 7 normal bases (Fig. 8). A dose-dependent increase in DNA adduct formation was observed in the presence of vitamin C. Maximal adduct formation was observed with 1 mM vitamin C where there was a more than 3-fold increase in DNA adduct levels (Fig. 8). The selective COX-2 inhibitor NS-398 caused a dose-dependent decrease in heptanone-etheno-dGuo adducts in the absence of vitamin C (Fig. 9). However, in the presence of vitamin C, NS-398 was unable to reduce the formation of DNA adduct to base-line levels (Fig. 9).

Cox-2-mediated DNA Damage
has previously been determined in recombinant enzymes or in intact cells after the addition of AA (Table II). There is no consensus on the enantio-selectivity of either COX-1 or COX-2. Hecker et al. (6) showed that ovine COX-1 produced an excess of 15(R)-HETE, whereas Thuresson et al. (7) showed that it produced an excess of 15(S)-HETE (Table II). There is one report that human COX-2 forms enantiomerically pure 15(R)-HETE (43). This study also reported that aspirin did not affect the chirality of the 15-HETE that was formed. Another study showed that both mouse and human COX-2 formed more 15(S)-HETE than 15(R)-HETE (8). These findings suggest that the chirality of COX-mediated 15-HETE formation may be dependent upon the incubation conditions that are used. There is general agreement that both COX-1 and COX-2 produce enantiomerically pure 11(R)-HETE (7,8,43). Furthermore, it is generally agreed that 15(R)-HETE is the major HETE that is formed after aspirin treatment of COX-2 (43-48) ( Table II). The initial step in 15(R)-HETE synthesis in acetylated COX-2 involves abstraction of the same pro-S hydrogen at C-13 of AA as occurs during normal PG biosynthesis (49). The change in oxygenation stereospecificity induced by aspirin treatment is proposed to arise because acetylated serine-516 in the human COX-2 (serine-530 in the mouse enzyme) forces a realignment of the -chain of AA. This unusual binding conformation appears to be responsible for oxygenation in the R configuration (43,(47)(48)(49).
No data exist on the relative amounts of HETE and PGE 2 formed by rat COX-2 or the chirality of rat COX-2-mediated HETE formation. To address these issues, HETEs and PGE 2 were quantified in RIES cell lysates treated with AA, and the chirality of the HETEs was determined. The presence of endogenous reducing co-factors in the lysates also helped to prevent the rapid inactivation of COX-2 that occurs in purified enzyme preparations 45-80 s after adding AA (50). 11(R)-HETE was the major metabolite at all AA concentrations used (Fig. 10). The next most abundant metabolite was PGE 2 followed by 15(S)-HETE and 15(R)-HETE. As the concentration of AA increased there was a gradual increase in the ratio of 11(R)-HETE to PGE 2 . The ratio of 15(S)-HETE (58%) to 15(R)-HETE (42%) was constant from 1 to 100 M AA. The preference for the 15(S) isomer was similar to that reported by Schneider et al. (8) for recombinant mouse and human COX-2, although the enantio-selectivity was somewhat lower (Table II).
There were some significant differences between the eicosanoids formed during incubations of intact RIES cells when compared with AA treatment of RIES cell lysates or with previous studies that examined AA treatment of recombinant COX-2 enzymes (Table II). The most surprising difference was the complete absence of 11(R)-HETE from the unstimulated incubations of intact RIES cells. This contrasts with our own work using cell lysates (Fig. 10) and with a number of studies using isolated enzymes showing that 11(R)-HETE is the major HETE generated by COXs (Table II). Interestingly, the enantio-selectivity observed for 15-HETE in the intact RIES cells was similar to that observed for recombinant mouse enzyme (Table II). In contrast, the ratio of PGE 2 to 15(S)-HETE (2:1) in intact RIES cells was similar to that observed at lower AA concentrations in the cell lysates. To further address this issue, eicosanoid production in the RIES cells from endogenous AA was induced using calcium ionophore A23187. This resulted in essentially the same pattern of metabolite formation, which occurred with AA treatment of the RIES cell lysates. It was particularly interesting that 11(R)-HETE was formed in concentrations that were similar to PGE 2 . There is one previous study of COX-2-mediated biosynthesis of 11-HETE in intact cells (51). Eicosanoids were analyzed by achiral LC/MS after vigorous activation of COX-2 by a combination of both ionophore A23187 and AA in human lung adenocarcinoma A549 cells. Under these conditions 11-HETE and PGE 2 were released into the cell supernatant at concentrations of 24 and 127 pmol/10 6 cells, respectively (ratio 1:5). Therefore, substantially more PGE 2 was formed than from simple ionophore A23187 treatment of the RIES cells, but significant amounts of 11-HETE were also formed.
There was a striking difference in basal 15-HETE production in intact RIES cells when compared with AA treatment of isolated recombinant enzymes or cells in which COX-2 was highly up-regulated (Table II). The absence of 11(R)-HETE and the relatively small amount of 15(R)-HETE formed by the unstimulated intact RIES cell incubations suggests that they or their hydroperoxide precursors can undergo further metabolism. In fact there are several reports describing the metabolism of other HETE and HPETE regioisomers (52,53).
A previous study with cultured ovine tracheal epithelial cells in which COX-2 had been up-regulated by growth factors quantified the relative amounts of 15-HETE and PGE 2 produced by adding [ 3 H]AA to washed cell suspensions (44)  increased COX-2 activity in the latter two cell systems. Eicosanoid product profiles are highly dependent upon the length of time it takes for self-inactivation of the enzyme to occur. In isolated COX enzymes this takes place 45-80 s after AA addition, depending on the reducing co-factors that are present to protect the enzyme (50). For example, in studies with purified COX-1 from rat seminal vesicles (7), 15-HETE formation was maximal after 20 s, although PGE 2 concentrations increased for another 80 s. The ratio of 15-HETE to PGE 2 10 s after the addition of AA was 1:2, whereas after 100 s the ratio was 1:8 (7). In microsomes from COS-7 cells transiently infected with recombinant COX-2, the ratio of 15-HETE to PGE 2 formation was 1:16 after a 30-min incubation (46). This is similar to the ratio observed in the ovine tracheal epithelial cells in which COX-2 had been up-regulated (44). It is noteworthy that the 15-HETE to PGE 2 ratio was increased by a factor of 2.5 in freshly isolated ovine tracheal epithelial cells where there was an almost 4-fold lower capacity for PGE 2 biosynthesis compared with the cells with up-regulated COX-2 (44). Furthermore, in human nasal polyp epithelial cells, which naturally express COX-2 rather than COX-1, the ratio of 15-HETE to PGE 2 was close to 1:1 (55). In intact unstimulated RIES cells and ionophore A23187-treated RIES cells, we determined that the ratio of 15-HETE to PGE 2 was 1:2 and 1:1, respectively. This is consistent with the ratio we found in RIES cell lysates and with the ratio of 15-HETE to PGE 2 that was found in human nasal polyp epithelial cells (55). Therefore, when cellular COX-2 levels are closer to those normally found under physiological conditions, it appears that both 15-HETE and PGE 2 are major metabolites.
To completely inhibit rat COX-2 in cell culture systems, NS-398 concentrations in the range 50 -100 M have generally been required (57)(58)(59)(60). These concentrations are still well below the IC 50 for recombinant ovine COX-1. Using a more sensitive LC/electron capture APCI/MS method, no NS-398 inhibition of PGE 2 formation was observed in WT RIE cells treated with 110 M NS-398. This contrasts with complete inhibition observed in RIES cells, where more than five times as much PGE 2 was produced. Our findings are consistent with the rank order of potency described for the NS-398-mediated inhibition of eicosanoid biosynthesis by recombinant ovine COX-2 (56) and cellular rat COX-2 (57-60). These data taken together with the low PGE 2 biosynthesis by WT RIE cells, which only express COX-1, confirmed that the primary source of eicosanoids in the RIES cells was COX-2. In our earlier study with un-stimulated intact RIES cells (41), inhibition of 15(S)-HETE biosynthesis by NS-398 was not detected. However, no heavy isotope internal standard was used for 15-HETE. Seven heavy isotope internal standards were used in the present study including 15(S)-[ 2 H 8 ]HETE. Using this methodology it was unequivocally demonstrated that 15(S)-HETE is in fact a major COX-2 metabolite derived from RIES cells.
A Western blot analysis revealed that COX-2 expression was maintained for 24 h after the media containing 10% FBS had been replaced by media containing only 0.1% FBS (Fig. 11A). The possibility that RIES cells contained a 15-LOX activity that could be responsible for the formation of 15-HETE was considered. Human reticulocyte 15-LOX-1 and rat leukocytetype 12/15-LOX are homologous genes in the two different species (61). Previous studies have demonstrated that 12-LOX is absent from the RIES cells. In fact the rat 12/15-LOX gene can be induced with sodium butyrate in WT RIE cells as demonstrated by the formation of 12-and 15-HETE (61). No 12-HETE was observed in butyrate-treated RIES cells. Similarly, 12-HETE was not formed in the present study of endogenous HETE formation by RIES cells. Therefore, RIES cells do not have any 12/15-LOX activity. In addition, RT-PCR was employed ( Fig. 11B) to show that 15-LOX-2 mRNA was not present in the RIES cells (10). These data taken together with the inhibition of 15(S)-HETE biosynthesis by aspirin and NS-398 in the RIES cells established that it was derived from COX-2 and not 15-LOX-1 or 15-LOX-2.
Holtzman et al. (44) showed that aspirin could convert COX-2 to a 15(R)-LOX in ovine tracheal epithelial cells. The amount of 15-HETE was increased more than 4-fold in aspirintreated cells when compared with untreated cells. However, 25% of the 15-HETE was accounted for by 15(S)-HETE. Therefore, the increase in 15(R)-HETE was actually 3.3-fold, which is similar to the 1.5-fold increase we observed in the aspirintreated intact RIES cells. The identification of 15(R)-HETE suggested that significant amounts of 15(R)-HPETE had also been formed. 15(R)-HPETE should give rise to exactly the same amount of heptanone-etheno-dGuo as its 15(S)-enantiomer (12). Therefore, it was somewhat surprising that no increase in heptanone-etheno adduct formation occurred in the aspirintreated RIES cells. In contrast, the expected aspirin-mediated inhibition of PGE 2 biosynthesis was observed both in the absence and presence of vitamin C (Fig. 6). This occurs through acetylation of serine 516 in the human enzyme (49), which inhibits formation of the hydroperoxide products PGG 2 and 11(R)-HPETE. Aspirin-mediated inhibition of these pathways could spare cellular reducing equivalents that would then be available to convert the resulting 15(R)-HPETE into 15(R)-HETE and so prevent the formation of heptanone-etheno adducts. In keeping with this possibility, we have shown recently that low M concentrations of lipid hydroperoxide-derived bifunctional electrophiles can rapidly reduce intracellular glutathione levels with concomitant formation of glutathione adducts (63). Therefore, aspirin appears to have an additional activity within the cell through acting in effect as an antioxidant that prevents COX-2-mediated formation of a lipid hydroperoxide (15(R)-HPETE), which can undergo homolytic decomposition to genotoxic bifunctional electrophiles.
Concentrations of vitamin C used in the present study are comparable to those found in human subjects (64). The efficiency with which vitamin C could induce lipid hydroperoxide decomposition to bifunctional electrophiles in vitro suggested that this process could induce significant levels of DNA damage in vivo (11). We have now demonstrated that this process can indeed occur when COX-2 is permanently expressed in RIES cells. The finding that vitamin C causes an increase in lipid hydroperoxide-mediated DNA adduct formation could help to explain why vitamin C has not been effective in cancer chemoprevention trials when used as a single agent (65). COX-2 is up-regulated in many tumor tissues (38), and a number of studies have been conducted to determine precisely how COX-2 mediates tumorigenesis. Current research suggests that it is involved cellular proliferation, angiogenesis, resistance to apoptosis, enhancing invasiveness, and modulation of immunosuppression (66). These biological activities are thought to result primarily through the formation of PGs such as PGE 2 . The present study provides evidence for an additional mechanism that may be involved in tumorigenesis; namely, the formation of etheno-DNA adducts that arise as a consequence of COX-2mediated lipid peroxidation (Scheme I). Under most circumstances lipid hydroperoxide-derived etheno-DNA adducts are repaired from the DNA. When there is a reduced capacity for DNA repair or when repair processes are overwhelmed, the etheno-DNA adducts could be responsible for increased mutagenesis and carcinogenesis (62). Therefore, inhibition of COX-2-mediated lipid hydroperoxide formation offers a potential therapeutic alternative to conventional COX-2 inhibitors in chemoprevention strategies for human populations.