Mercapturic Acid Conjugates of 4-Hydroxy-2-nonenal and 4-Oxo-2-nonenal Metabolites Are in Vivo Markers of Oxidative Stress*

Oxidative stress-induced lipid peroxidation leads to the formation of cytotoxic and genotoxic 2-alkenals, such as 4-hydroxy-2-nonenal (HNE) and 4-oxo-2-nonenal (ONE). Lipid-derived reactive aldehydes are subject to phase-2 metabolism and are predominantly found as mercapturic acid (MA) conjugates in urine. This study shows evidence for the in vivo formation of ONE and its phase-1 metabolites, 4-oxo-2-nonen-1-ol (ONO) and 4-oxo-2-nonenoic acid (ONA). We have detected the MA conjugates of HNE, 1,4-dihydroxy-2-nonene (DHN), 4-hydroxy-2-nonenoic acid (HNA), the lactone of HNA, ONE, ONO, and ONA in rat urine by liquid chromatography-tandem mass spectrometry comparison with synthetic standards prepared in our laboratory. CCl4 treatment of rats, a widely accepted animal model of acute oxidative stress, resulted in a significant increase in the urinary levels of DHN-MA, HNA-MA lactone, ONE-MA, and ONA-MA. Our data suggest that conjugates of HNE and ONE metabolites have value as markers of in vivo oxidative stress and lipid peroxidation.

Lipid peroxidation (LPO) 2 products are breakdown products of fatty acids formed under conditions of oxidative stress. 4-Hydroxy-2-nonenal (HNE) is a well established LPO product that has been shown to contribute to the development and progression of age-related diseases such as Alzheimer and atherosclerosis (1)(2)(3)(4) in addition to being cytotoxic and genotoxic (5,6). The mechanism of formation for HNE from linoleic acid via 4-hydroperoxy-2-nonenal (HPNE) has been previously demonstrated (7). Once HNE is formed, it can be further metabolized by cytochrome P450, aldehyde dehydrogenase, aldo-keto reductase (AKR), and conjugated by glutathione S-transferase (GST). Certain isoforms of murine and human P450s (8) and aldehyde dehydrogenase (9) can catalyze the oxidation of HNE to 4-hydroxy-2-nonenoic acid (HNA). When conjugated, HNA can undergo spontaneous intramolecular condensation, resulting in lactone formation (10). Aldo-keto reductase 1B1 has been shown to reduce HNE to form 1,4-dihydroxy-2-nonene (DHN) (11). Glutathione (GSH) can form conjugates with HNE and other LPO products (6) via a Michael-type addition mediated by GST (12)(13)(14). The GSH can then be further metabolized in the liver and in the kidney to form mercapturic acid (MA), resulting in the conjugates shown in Fig. 1. A number of studies have examined HNE and its metabolites in vivo (10,(15)(16)(17)(18) and have demonstrated the formation of the MA conjugates HNE-MA, DHN-MA, HNA-MA, and HNA-MA lactone in vivo (10). In addition, histidine-1,4-dihydroxynonane (His-DHN) and His-HNA have been found in the urine of obese Zucker rats, a model of metabolic syndrome (19).
Not only can HPNE break down into HNE, but it can also form 4-oxo-2-nonenal (ONE) as shown in vitro (20) and in cultured cells (21). ONE can be reduced at the C-4 position by carbonyl reductase to form HNE (22), but it can also be reduced at the C-1 position by aldo-keto reductase to form 4-oxo-2nonen-1-ol (ONO) (21,23,24) or oxidized by aldehyde dehydrogenase (human aldehyde dehydrogenase 2) to form 4-oxo-2-nonenoic acid (ONA) (25). Again, GST can mediate conjugate formation, resulting in metabolism to the MA conjugates shown in Fig. 2. In view of these findings, ONE metabolites are expected to be formed in vivo as end products of LPO, but their presence in vivo has not been demonstrated.
Here we report that phase-2 metabolites of ONE, ONA, and ONO are formed after an acute oxidative stress insult in rats. Rats were exposed to CCl 4 , an established model of in vivo oxidative stress (26). MA conjugates of HNE and ONE metabolites were detected in the urine by LC-MS/MS comparison with synthetic standards. Our results demonstrate, for the first time that ONE metabolites, in addition to HNE metabolites, are formed in vivo as products of LPO. Currently, F 2 -isoprostanes are considered to be the most reliable marker of in vivo oxidative stress and LPO (27). The CCl 4 induced formation of the HPNE-derived LPO product conjugates show potential for these metabolites as additional or alternative markers of oxidative stress in animals and in humans.

Synthesis of HNE and ONE Metabolites
HNE-HNE was synthesized from 3-(Z)-nonenol following a method adapted from Gardner et al. (28). Briefly, 3-(Z)-non-enol (2 mmol) was dissolved in 8 ml of CH 2 Cl 2 , and 3-chloroperoxybenzoic acid (2 mmol) was added. The reaction mixture was stirred for 1 h at room temperature and, after the addition of 8 ml of 10% aqueous NaHCO 3 , stirred for 45 min. The reaction mixture was washed with water and dried with 2 g of powdered molecular sieves. 4-Methylmorpholine N-oxide (3 mmol) was added, and the mixture was stirred under argon for 30 min. After the addition of tetrapropylammonium perruthenate (0.1 mmol), the mixture was stirred for 1 h under argon, filtered through silica gel, and rinsed with ethyl ether. Next, 16 ml of 1.3 M sodium hydroxide was added to the filtrate, and the solution was stirred vigorously for 15 min. The reaction mixture was washed with water, dried with anhydrous MgSO 4 , filtered, and concentrated in vacuo. The yield was 35%. Our adaptations to the method of Gardner et al. (28) gave a slightly lower yield; however, we found that the changes resulted in more consistent yields. Data are 1  LPO-MA Adducts-All adducts were synthesized in the same manner and characterized by LC-electrospray ionization-MS/MS (see Table 1). A 20 mM solution of MA was prepared in 0.1 M phosphate buffer, pH 8. To 50 l of this solution was added 450 l of the same phosphate buffer and 400 l of water. A 1 mM solution of the LPO product of interest was made up in ethanol, and 100 l was added to the MA solution. The reaction was stirred at 37°C for 2 h and then acidified to pH 3 with 1 N HCl. It was then extracted with ethyl acetate, 3 ϫ 1 ml, evaporated under nitrogen using a Zymark TurboVap LV (Caliper HPNE can then be reduced to HNE and further metabolized by aldehyde dehydrogenase (ALDH), aldo-keto reductase (AKR), and GST. These enzymes cause oxidization, reduction, or GSH conjugation, respectively. Once LPO-GSH conjugates have formed, the GSH is further metabolized to MA. The LPO-MA conjugates are analyzed in this study. It is important to note that once HNE has been reduced to DHN, it will no longer form a Michael-type conjugate with GSH due to the lack of ␣,␤-unsaturation. Also, HNA-MA is subject to intramolecular condensation, resulting in the formation of a lactone. FIGURE 2. Formation of LPO-MA conjugates from ONE. HPNE can eliminate a water molecule, resulting in the formation of ONE. Similarly to HNE, ONE undergoes oxidation by aldehyde dehydrogenase (ALDH), reduction by aldoketo reductase (AKR), and GSH conjugation by GST. Unlike HNE, ONE has two possible sites of conjugate formation, at the C-2 and C-3 positions. Thus, after reduction, the ONO formed retains its ␣,␤-unsaturation, allowing it to form GSH conjugates via this route. The MA conjugates were analyzed in this study and are shown here as the C-2 conjugates.
DHN-MA-Because DHN has no ␣,␤-unsaturation with which a Michael-type adduct can form, the DHN-MA adduct was made by first synthesizing HNE-MA as described above. The aldehyde moiety was then reduced with 10 l of a 5 M sodium borohydride solution in 1 N NaOH (31). The reduction mixture was stirred at room temperature for 30 min then acidified to pH 3 with 1 N HCl. It was then extracted with ethyl acetate, 3 ϫ 1 ml, evaporated under nitrogen, and reconstituted in 1 ml of 20:80 acetonitrile:water containing 0.1% formic acid.

Animal Treatment
The experimental protocol for the animal studies was approved by the Institutional Animal Care and Use Committee at Oregon State University. Male F344 rats (Harlan, Indianapolis, IN), weighing 280 -320 g, were housed in individual plastic cages covered with Hepa filter and allowed free access to standard animal chow and water ad libitum. After 1 week of acclimatization, the rats were transferred to metabolism cages. Six animals were divided into two groups of three, with one group receiving an intraperitoneal dose of 1 ml/kg CCl 4 (dissolved in corn oil), and the other group (control) receiving the vehicle alone. The CCl 4 dose of 1 ml/kg was chosen on the basis of literature reports (32,33). Urine was collected from the rats during a 24-h period after treatment.

Urine Samples
A volume of 0.5 ml of rat urine was acidified with 225 l of 1 N HCl to pH 3. To the urine was added 5 l of a 1 M solution of [9-2 H 3 ]HNE-MA as an internal standard. The samples were extracted with ethyl acetate, 3 ϫ 1 ml. The ethyl acetate layers were combined and evaporated under nitrogen. Samples were then reconstituted in 200 l of 20:80 acetonitrile:water containing 0.1% formic acid.
Urinary creatinine was measured using a Creatinine Assay Kit, catalog no. 500701 (Cayman Chemical). The assay was performed according to the manufacturer's directions. There was no significant difference between treated and control creatinine levels (p ϭ 0.9).

HPLC
A Shimadzu Prominence HPLC system (Shimadzu, Columbia, MD) consisting of four LC-20AD pumps, a DQU-20A 5 degasser, and an SIL-HTc autosampler equipped with switching valves was used for all chromatography. Two chromatographic systems were used, and unless otherwise stated, all data presented were recorded with system 1. For system 1, the HPLC column was a 250 ϫ 2-mm Synergi Max RP C 12 column (Phenomenex, Torrance, CA). The mobile phase consisted of solvent A, 0.1% (v/v) formic acid in water, and solvent B, acetonitrile containing 0.1% (v/v) formic acid. The flow rate was 0.2 ml/min. Separations were carried out using a linear solvent gradient from 20 to 50% B in 10 min, a linear gradient from 50 to 90% B over the next 2 min, held constant at 90% B for 7 min, returned to 20% B after 1 min, and equilibrated at 20% B for 5 min. System 2 consisted of a 50 ϫ 2.1-mm Inertsil ODS-3 column (Varian, Lake Forest, CA). The mobile phase consisted of solvent A, 0.1% (v/v) formic acid in water, and solvent B, methanol. The flow rate was 0.3 ml/min, and separations were carried out using a linear solvent gradient from 45 to 90% B over 5 min, held constant at 90% B for 1 min, then returned to 45% B and equilibrated for 3 min.

Mass Spectrometry
An Applied Biosystems MDS Sciex hybrid triple quadrupole/ linear ion trap mass spectrometer (4000 QTrap) equipped with a TurboV electrospray source (Concord) was used for these analyses. The TurboV source was maintained at 400°C. The ion-spray voltage was Ϫ4500 V, and the declustering potential was 40 V. Nitrogen was used as the source gas, curtain gas, and collision gas. Various scanning techniques, all run in negative ion mode, were used for the characterization and detection of LPO-MA products, including Q1, product ion scanning, and selected reaction monitoring (SRM). All SRM transitions and collision energies are shown in Table 1.

Data Analysis
Peak area analysis was performed using Analyst 1.4.1 (Applied Biosystems). Analyte peak areas were normalized for the internal standard (IS) peak area, a 2.5-fold sample concentration, and for the creatinine concentration in mg/ml. Thus, all data are represented as analyte peak area/(IS peak area ϫ 2.5 ϫ mg/ml creatinine). For these analyses, peak 3 (9.7 min, Fig. 3B) from the m/z 318 3 162 SRM transition was used to quantify ONO-MA, and peak 2 (9.3 min, Fig. 3A) from the m/z 318 3 171 SRM transition was used to quantify HNE-MA. Statistical comparisons were performed using GraphPad (San Diego, CA). Data are shown as mean Ϯ S.E.

RESULTS
Detection of the MA Conjugate of ONO-Using SRM transitions for the MA conjugate of HNE, three chromatographic peaks (8.9, 9.3, and 9.7 min, Fig. 3C) were observed in the urine of CCl 4 -dosed rats, of which only the first two corresponded with the two chromatographic peaks produced by a synthetic sample of HNE-MA (Fig. 3A). The product ion spectra of the chromatographically resolved HNE-MA isomers (first two peaks at 8.9 and 9.3 min) showed ␤-elimination fragment ions observed at m/z 189 and m/z 171 (Fig. 4). The third chromatographic peak at 9.7 min (urine), absent from the SRM chromatogram of HNE-MA, showed the greatest signal for the m/z 318 3 162 SRM transition, indicating release of mercapturate (m/z 162) from the molecular anion. This suggests that it favored retro-Michael (RM) cleavage over ␤-elimination upon collision-induced dissociation, presumably because it prefers the formation of an enolate (thus, setting the stage for RM) over formation of a cyclic hemiacetal or hemiketal (more resistant to RM). Because hemiketal or hemiacetal formation is less favorable for ketones than for aldehydes (34), we hypothesized that the third peak represented the MA conjugate of ONO, which is isobaric with HNE-MA. Synthetic ONO-MA indeed showed the expected retention time (9.7 min, Fig. 3B) and MS/MS spectrum (major fragment with m/z 162), which was taken as evidence for the excretion of ONO-MA in the urine (Fig. 3). ONO metabolites have not previously been reported as in vivo products of oxidative stress.
Distinction between the Isobaric Metabolites, HNA-MA Lactone, and ONE-MA-When first analyzed using chromatographic system 1, it was not possible to distinguish between the isobaric metabolites HNA-MA lactone and ONE-MA. Both compounds produced the same mercapturate fragment (m/z 162), and they co-eluted. A second chromatographic system was developed to separate the compounds (system 2). The nowresolved HNA-MA lactone differed from ONE-MA by the production of an additional fragment ion with m/z 143, which allowed the distinction between the two conjugates under the new chromatographic conditions (Fig. 5). Both MA conjugates were detected in rat urine (Fig. 5C).   (8.9, 9.3, and 9.7 min). The standards confirm that the third peak contains only ONO-MA. This third peak was used to represent ONO-MA for comparison between oxidatively stressed rats and control rats. the control animals (Fig. 7). Urinary concentrations of ONO-MA, HNE-MA, and HNA-MA were elevated in the CCl 4treated rats, but the increases were not significant at the p Ͻ 0.05 level. Statistical comparisons were performed using the Student's t test, and p values for each conjugate are shown in Table 1. All calculations included normalization for creatinine levels to account for variation in urine concentration between rats. There was no difference in the creatinine levels between the treatment and control groups (p ϭ 0.9).

DISCUSSION
Lipid hydroperoxides, specifically hydroperoxy octadecadienoic acids (HPODEs), may be formed enzymatically through the action of lipoxygenases (35,36) or non-enzymatically through the reaction of linoleic acid with reactive oxygen species (37). Degradation of lipid hydroperoxides yields a multitude of reactive aldehydes (38), but only HNE and, more recently, ONE have been studied extensively with regard to their detrimental effects in biological systems. HNE and ONE have been shown to cause adverse effects because of their ability to covalently bind to 2Ј-deoxyguanosine (39 -42), to cause protein cross-linking (43), and to induce aggregation of low density lipoproteins (44). ONE shows greater toxicity in cultured cells, presumably due to its reactivity as a bifunctional electrophile (45). Furthermore, aldo-keto reductase-mediated reduction of the aldehyde functionality renders HNE inactive as a Michael acceptor, whereas ONE retains its electrophilicity after conversion into ONO, a regioisomer of HNE (Fig. 8). The formation of ONO as a reactive metabolite of ONE has received little or no attention in the literature. Moreover, it was suggested by Blair that "ONO may contribute to the biological activities that have been ascribed previously to HNE" (24). Phase-2 metabolites of ONO and HNE are difficult to distinguish from one another by LC-MS/MS due to similar chromatographic behavior, lack of mass difference, and similar MS/MS spectra. We prepared the MA adducts of HNE and ONO by chemical methods and demonstrated that at least one (set of) ONO-MA isomer can be distinguished from HNE-MA and detected in rat urine on the basis of chromatographic retention times and MS/MS fragmentation (Fig. 3). The in vivo formation of ONO-MA has not previously been reported in the literature.

In Vivo Detection of 4-Oxo-2-nonenal Metabolites
tamic acid residue of the GSH-ONE adduct participates in cyclization, the bicyclic thiadiazabicyclo-ONE-glutathione metabolite cannot be further metabolized to an MA adduct of ONE or its derivatives. Our data indicate that ONE is produced in vivo and undergoes phase-2 metabolism to form ONE-MA (Fig. 5), which would compete with thiadiazabicyclo-ONE-glutathione formation. Furthermore, our data also indicate that ONE-MA is metabolized to the corresponding carboxylic acid, ONA-MA, based on LC-MS/MS comparison with a synthetic sample of ONA-MA (Fig. 6). At present it is not known whether the oxidation of the aldehyde functionality precedes or follows the conjugation reaction because ONA retains its ability to form a conjugate with GSH. Like ONO-MA, ONE-MA and ONA-MA have not previously been reported as in vivo metabolites.
To determine whether the formation of HNE and ONE increases under conditions of oxidative stress in vivo, we measured the MA conjugates of these LPO products in the urine of rats exposed to CCl 4 and in the urine of control animals. CCl 4 is known for its ability to cause LPO in the liver through the formation of the trichloromethyl radical (CCl 3 ⅐ ) via cytochrome P450-mediated homolytic cleavage of the carbon-chlorine bond (47)(48)(49). The CCl 3 ⅐ radical can form chloroform by abstracting a hydrogen from the bisallylic methylene functionality in polyunsaturated fatty acids, thereby forming lipid-based radicals that spontaneously react with molecular oxygen to form regioisomeric lipid hydroperoxides, e.g. 9-hydroperoxy-10,12-octadecadienoic acid (9-HPODE) and 13-hydroperoxy-9,11-octadecadienoic acid (13-HPODE) from linoleic acid. Alternatively or additionally, CCl 3 ⅐ may directly react with molecular oxygen to give trichloromethyl peroxy radicals (CCl 3 OO ⅐ ), a reactive species considered by some authors to produce HPODEs (47)(48)(49). Subsequently, these HPODEs are converted into carbon-carbon cleavage products, such as HPNE, HNE, and ONE (50). Regardless of the precise mechanism of CCl 4 -induced LPO, acute CCl 4 poisoning is an accepted model of in vivo oxidative stress and LPO. For instance, HNE-deoxyguanosine adducts have been shown to have significantly increased levels in F344 rats treated with CCl 4 as compared with control animals (39,42).
In our studies the urinary levels of the HNE metabolites DHN-MA and HNA-MA lactone and the levels of the ONE   metabolites ONE-MA and ONA-MA were significantly higher in the CCl 4 -treated rats compared with the control animals (Fig. 7). The urinary levels of HNE-MA and ONO-MA were higher in the CCl 4 -treated rats, but the difference was not statistically significant at p Ͻ 0.05 (p ϭ 0.18 and p ϭ 0.10, respectively). This is an unexpected finding because HNE is generally considered to be a marker of LPO and in vivo oxidative stress. It is conceivable that the lack of statistical difference is due to inter-individual variation in the metabolism of HNE conjugates to form DHN and HNA conjugates. Similarly, ONO-MA may be converted into DHN-MA by carbonyl reductase (22). Our semiquantitative data (normalized peak areas), however, suggest that oxidation of the aldehyde functionality in the HNE and ONE conjugates is the preferred metabolic pathway because HNA-MA lactone and ONA-MA are among the most abundant end products of HPNE metabolism (Fig. 7). The arachidonic acid-derived F 2␣ -isoprostanes are generally considered to be the most reliable markers of in vivo oxidative stress. The value of F 2␣ -isoprostanes as oxidative stress markers in humans is exemplified by studies of smokers (51,52), patients with renal failure (53), patients with coronary artery disease (54), and patients with lupus erythematosus (55). Kadiiska et al. (56) observed a significant increase in the urinary concentrations of F 2␣ -isoprostanes in rats injected intraperitoneally with 120 or 1200 mg/kg of CCl 4 compared with control animals. Sicilia et al. (57) reported elevated F 2␣ -isoprostanes levels in liver and kidney tissue of rats after an oral CCl 4 dose of 1 ml/kg. However, these authors observed no statistical difference between the urinary levels of F 2␣ -isoprostanes obtained from treated and control animals in the same study (57), which they attributed to kinetic differences between oral gavage and the intraperitoneal administration used by Kadiiska et al. (56). Our findings indicate that the end products of HPNE metabolism are elevated in urine obtained from CCl 4 -treated rats, which holds promise for these metabolites as additional or alternative markers of oxidative stress in animals and humans.