Lipidomic Analysis for Lipid Peroxidation-Derived Aldehydes Using

Xiaodong Wen , Lacy Reynolds , Rohit S. Mulik , Soo Young Kim , Tim Van .... Alexandre P. Blanchard , Zemin Yao , Steffany A.L. Bennett , Daniel Figey...
0 downloads 0 Views 173KB Size
Download PDF
Chem. Res. Toxicol. 2007, 20, 99-107

99

Lipidomic Analysis for Lipid Peroxidation-Derived Aldehydes Using Gas Chromatography-Mass Spectrometry Yoshichika Kawai,* Sayaka Takeda, and Junji Terao Department of Food Science, Graduate School of Nutrition and Biosciences, The UniVersity of Tokushima, Tokushima 770-8503, Japan ReceiVed August 20, 2006

A lipidomic method is described for the measurement of lipid peroxidation-derived aldehydes using gas chromatography/electron ionization/mass spectrometry with selected ion monitoring (GC/EI/MSSIM). Aldehydes in the samples were converted into their pentafluorobenzyl (PFB)-oximes using PFBhydroxylamine, and other functional groups such as the hydroxyl groups were further derivatized into the trimethylsilyl ethers. The PFB-oxime derivatives could be comprehensively detected by the SIM of m/z 181, which is a characteristic fragment ion of the PFB-oxime derivatives. At the same time, each aldehyde was classified into five groups (alkanals, 2-alkenals, 2,4-alkadienals, 2-hydroxyalkanals, and 4-hydroxy-2-alkenals) by SIM of the structure-specific fragment ions. Determination of the 4-hydroxy2-alkenals was also performed by confirmation of their higher reactivity with the sulfhydryl group. On the basis of the mass spectrometric characterization, we have identified at least 33 aldehydes formed upon the FeII-mediated decomposition of the arachidonic acid-, linoleic acid-, and docosahexaenoic acidhydroperoxides in Vitro. We then applied this system to the in ViVo samples and successfully observed the increase in aldehydes in the liver of mice intraperitoneally injected with bromobenzene, an experimental animal model for lipid peroxidation. Using this comprehensive analysis, unique differences in the formation between each aldehyde could be observed. This method is useful for simultaneously monitoring the lipid peroxidation-derived aldehydes formed under oxidative stress in ViVo. Introduction Polyunsaturated fatty acids and their esters are known to produce lipid peroxidation products under oxidative stress (1). During the lipid peroxidation reaction, lipid hydroperoxides (LOOH1) are formed as the primary products. In the presence of transition metals (such as iron) or ascorbic acid (2), LOOH generates LOO• and/or LO• that induce autocatalytic lipid peroxidation reactions. The LOO•/LO• are also decomposed by cleavage of adjacent C-C bonds to generate aldehydic compounds as the end products. It has been reported that lipid peroxidation products, especially aldehydic compounds, are implicated in the pathogenesis of various diseases, including atherosclerosis, cancer, and neurodegenerative diseases (3, 4). Therefore, lipid peroxidation products are recognized as useful markers for lipid peroxidation processes in ViVo. The lipid peroxidation reaction generates various types of aldehydes, and their bioactivities differ from each other. Although aldehyde groups basically react with amino groups to generate the Schiff’s base adducts, the R,β-unsaturated aldehydes (-CHdCH-CHdO), such as acrolein and crotonaldehyde, are also adducted with both amino groups and sulfhy* To whom correspondence should be addressed. Tel: 81-88-633-9592. Fax: 81-88-633-7089. E-mail: [email protected]. 1 Abbreviations: LOOH, lipid hydroperoxides; HNE, 4-hydroxy-2nonenal; ONE, 4-oxo-2-nonenal; HHE, 4-hydroxy-2-hexenal; GC/EI/MS, gas chromatography/electron ionization/mass spectrometry; SIM, selected ion monitoring; PFB, pentafluorobenzyl, TMS, trimethylsilyl; PFBHA, O-(2,3,4,5,6-pentafluorobenzyl)-hydroxylamine; PAPC, 1-palmitoyl-2arachidonyl-phosphatidylcholine; BSTFA, N,O-bis(trimethylsilyl)trifluoroacetamide; MTBSTFA, N-methyl-N-(t-butyldimethylsilyl)trifluoroacetamide; MAOOH, methyl arachidonate hydroperoxides; MLOOH, methyl linoleate hydroperoxides; MDOOH, methyl docosahexaenoate hydroperoxides; NAC, NR-acetyl-cysteine; NAL, NR-acetyl-leucine; TBDMS, tbutyldimethylsilyl; MDA, malondialdehyde.

dryl groups via Michael addition. It is well known that 4-hydroxy-2-nonenal (HNE) is far more reactive with sulfhydryl groups than other R,β-unsaturated aldehydes (5). HNE is also a specific inducer for cyclooxygenase-2 in several cell lines (6, 7). It has recently been reported that 4-oxo-2-nonenal (ONE), a novel lipid peroxidation-derived aldehyde, is more reactive with sulfhydryl groups than HNE. ONE has also been reported to be more reactive with DNA than other aldehydes (8) and specifically activate p53-dependent apoptosis in a human neuronal cell line (9). In addition, the lipophilicity/hydrophilicity of aldehydes that affects the interaction with biomolecules depends on their structures including the chain lengths. It has been reported that 4-hydroxy-2,6-dodecadienal, a long chain (C12) 4-hydroxy-2-alkenal analogue, is more reactive toward phosphatidylethanolamine than HNE (C-9) and 4-hydroxy-2hexenal (HHE, C-6) (10). In addition, the pattern for the formation of aldehydes may depend on their original polyunsaturated fatty acids. These observations suggest the significance of evaluating various aldehydes for not only estimating the total lipid peroxidation status but also understanding the specific distributions in tissues, their polyunsaturated fatty acid sources, and subsequent contributions to tissue damage under oxidative stress. In this study, we developed an analytical method for comprehensively evaluating aldehydes formed during the peroxidation of three polyunsaturated fatty acids (arachidonic acid, linoleic acid, and docosahexaenoic acid) by using gas chromatography/electron ionization/mass spectrometry with a selected ion monitoring system (GC/EI/MS-SIM). The aldehyde groups were derivatized into their pentafluorobenzyl (PFB)-oximes, which universally exhibit a fragment ion at m/z 181, and the polar functional groups such as the hydroxyl groups were further

10.1021/tx060199e CCC: $37.00 © 2007 American Chemical Society Published on Web 12/09/2006

100 Chem. Res. Toxicol., Vol. 20, No. 1, 2007 Scheme 1. GC/EI/MS-SIM Assay Procedures

derivatized into their trimethylsilyl (TMS) derivatives (Scheme 1). These derivatizations give the characteristic fragment ions of aldehydes and help us to identify/quantify these aldehydes in the GC/EI/MS. Much attention has recently been paid to the comprehensive analysis of lipid components (termed as lipidomics). By considering the implication of lipid peroxidation products in various diseases, our current study suggests the potential importance of targeting oxidized lipidomics under oxidative stress in ViVo.

Experimental Procedures Materials. O-(2,3,4,5,6-Pentafluorobenzyl)-hydroxylamine (PFBHA) hydrochloride and methyl arachidonate were purchased from the Sigma-Aldrich Co. (St. Louis, MO). Methyl docosahexaenoate, methyl eicosapentaenoate, and methyl γ-linolenate were obtained from the DOOSAN Serdary Research Laboratories (Kyungki, Korea). 1-Palmitoyl-2-arachidonyl-phosphatidylcholine (PAPC) was obtained from Avanti Polar Lipid, Inc. (Alabaster, AL). N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA), N-methyl-N-(t-butyldimethylsilyl)trifluoroacetamide (MTBSTFA), methyl linoleate, 2,4-decadienal, and bromobenzene were purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Tridecanal was purchased from Kanto Chemical Co., Inc. (Tokyo, Japan). 4-Hydroxy-2nonenal (HNE) was purchased from OXIS International, Inc. (Foster City, CA). Malondialdehyde was prepared by the acid hydrolysis of tetraethoxypropane. Glyoxal and 2,4-heptadienal were obtained from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Preparation of Lipid Hydroperoxides. The monohydroperoxides of methyl arachidonate and methyl docosahexaenoate were prepared by autoxidation at 37 °C for 72-120 h in the presence of 1% R-tocopherol (11). The methyl linoleate hydroperoxides were prepared by soybean lipoxygenase-catalyzed oxidation as previously described (12). The monohydroperoxides were purified by preparative layer chromatography (silica gel 60 F254, 10 × 20 cm, 2 mmthickness, Merck). The formation of isomeric monohydroperoxides was confirmed by a normal phase high-performance liquid chromatography (HPLC, LC-10AT, Shimadzu, Kyoto, Japan) with UV detection at 235 nm (SPD-10AV, Shimadzu). The samples were injected to a silica-gel column (TSK-gel Silica-60, 4.6 × 250 mm, 5 µm, Tosoh, Tokyo, Japan) equilibrated with n-hexane/isopropanol (99.5/0.5, v/v) at a flow rate of 2.0 mL/min. The concentrations of the lipid hydroperoxides were spectrophotometrically determined: methyl arachidonate hydroperoxides (MAOOH), 234 ) 27 000 (13); methyl linoleate hydroperoxides (MLOOH), 233 ) 28 000 (14); methyl docosahexaenoate hydroperoxides (MDOOH), 233 ) 25 200 (15). Decomposition of Lipid Hydroperoxides. The purified lipid hydroperoxides (150 nmol) were emulsified in 100 µL of 100 mM Tris-HCl buffer (pH 7.4) containing 1 mM FeSO4 and incubated at 37 °C for 4 h. The residual hydroperoxides were determined by normal phase HPLC analysis as described above. The free radical-

Kawai et al. mediated peroxidation of the polyunsaturated fatty acid methyl esters or PAPC was performed by incubating the emulsified lipids in 100 mM Tris-HCl buffer (pH 7.4) in the presence of 50 µM FeSO4/500 µM ascorbic acid at 37 °C for 24 h. After the incubation period, an equal volume of 2 mM ethylenediamine tetraacetic acid/ 20 µM 2,6-di-tert-butyl-p-cresol was added to the reaction mixtures and stored at - 30 °C until analysis. The effect of the addition of the sulfhydryl group on the detection of the 4-hydroxy-2-alkenals was examined upon incubation of the samples with 10 mM NRacetyl-cysteine (NAC) or NR-acetyl-leucine (NAL) at 37 °C for 15 min prior to the derivatization steps. GC/EI/MS Analysis of Aldehydes. For the PFB-derivatization of aldehydes, the samples (200 µL) were mixed with a 50 mM PFBHA methanolic solution (100 µL) and stored at room temperature for 2 h. The PFB-oxime derivatives were extracted twice with 500 µL of n-hexane, and the organic layers were dried under N2 gas and in a vacuum desiccator. For the trimethylsilyl (TMS)derivatization of other functions in the PFB-oximes, 50 µL of BSTFA was added to the samples and incubated at room temperature for 1 h. In the case of the t-butyldimethylsilyl (TBDMS)derivatization, 60 µL of MTBSTFA/dimethyl formamide (1/1, v/v) was added instead of BSTFA and incubated at 70 °C for 1 h to obtain maximal yields of the derivatives. Decomposition of samples during incubation at 70 °C could not be observed. The PFB/ TBDMS-derivatives were extracted twice with 500 µL of n-hexane, dried under N2 gas and in a vacuum desiccator, and then dissolved in 50 µL of n-hexane. The derivatization/extraction steps were preliminarily examined, and a suitable procedure, described above, was then determined. GC/EI/MS was performed using an AQP-5050A mass spectrometer (Shimadzu Co., Kyoto, Japan) equipped with an SPB-1 fused silica capillary column (30 m × 0.25 mm, 10 µm film thickness; Supelco, Bellefonte, PA) in an electron ionization mode of 70 eV. The carrier gas, helium, was applied at a flow rate of 1.0 mL/min. The column oven was held at 50 °C for 2 min before being raised to 290 °C at the rate of 10 °C/min and then held constant for 5 min. The molecular separator temperature and ion source temperature were 250 and 260 °C, respectively. The derivatized samples (1 µL) were injected into the GC/EI/MS system in the splitless mode. Animal Experiments. Six-week-old female C57BL/6J mice were purchased from Japan SLC, Inc. (Shizuoka, Japan) and housed under standard experimental conditions (24 °C, 12 h light and 12 h dark cycle). They were divided into six groups (n ) 3) and fed the MF diet (Oriental Yeast Co., Ltd., Tokyo, Japan) for one week. The animals were then starved overnight before receiving, by oral administration, bromobenzene (7.5 mmol/kg body weight) dissolved in corn oil (100 µL/mice). The control group (0 h) was treated with corn oil alone and immediately sacrificed. We confirmed that lipid peroxidation levels in the liver of mice treated with corn oil were scarcely changed after 24 h. They were killed at 0, 1, 2, 4, 8, and 24 h after poisoning. Liver homogenates (10%) were prepared in ice-cold phosphate buffered saline (pH 7.4) containing 0.4 mM ethylenediamine tetraacetic acid and 20 µM 2,6-di-tert-butyl-pcresol. The protein contents of the homogenates were determined using a BCA protein assay reagent kit (Pierce). For the GC/EI/MS analysis, tridecanal (2.5 nmol) as an internal standard was added to the homogenates (0.4 mL) and then derivatized with a 50 mM PFBHA methanolic solution (200 µL). After incubation at room temperature for 2 h, a drop of sulfuric acid was added and then extracted twice with 500 µL of n-hexane. The following derivatization procedures were preformed as described previously. From these derivatization/extraction steps, we confirmed that at least 95% of tridecanal could be recovered.

Results Preparation and Decomposition of Lipid Hydroperoxides. To establish the procedure for simultaneously analyzing aldehydes during lipid peroxidation in ViVo, the monohydroperoxides of arachidonic acid, linoleic acid, and docosahexaenoic acid,

Lipidomic Analysis of Aldehydes by GC/EI/MS

Figure 1. Possible fragmentation patterns of five aldehyde classes.

the major endogenous polyunsaturated fatty acids, were prepared and then decomposed in phosphate buffer in the presence of FeII. The monohydroperoxides of these polyunsaturated fatty acid methyl esters were prepared by autoxidation in the presence of R-tocopherol. In this oxidation system, the formation of approximately equal amounts of the monohydroperoxide isomers is expected (11). We confirmed the formation of the isomeric monohydroperoxides by thin-layer chromatography and normalphase HPLC (Supporting information, Figure S1A). The timeand FeII dose-dependent decomposition of hydroperoxides was monitored by HPLC-UV. The representative data for the methyl arachidonate hydroperoxide (MAOOH) are shown in the Supporting information (Figure S1B). On the basis of these experiments, the incubation time and FeII concentration were determined to be 4 h and 1 mM, respectively. Comprehensive Detection of PFB-Oxime Derivatives by GC/EI/MS-SIM. For the GC/EI/MS analysis, the aldehydes in the reaction mixtures were first converted into their PFBoxime derivatives. To derivatize other functional groups, such as hydroxyl groups, the samples were further treated with BSTFA to form TMS derivatives. The PFB and PFB/TMS derivatives were then injected into the GC/EI/MS system with selected ion monitoring (SIM). Because the PFB derivatives commonly exhibit a characteristic fragment ion at m/z 181, we comprehensively monitored all aldehydes in the SIM at m/z 181. At the same time, the aldehydes were classified into five groups (alkanals, 2-alkenals, 2,4-alkadienals, 2-hydroxyalkanals, and 4-hydroxy-2-alkenals) by the SIM of aldehyde class-specific fragment ions. The fragment patterns are summarized in Figure 1. Aldehydes that could not be classified to these five groups were classified into the others group. The representative GC/ EI/MS-SIM profiles for the decomposed mixture of MAOOH are shown in Figure 2. The 16 aldehydes from MAOOH could be classified into the 5 groups. Similarly, the classifications of 11 and 14 aldehydes were determined upon the decomposition of methyl linoleate hydroperoxide (MLOOH) and methyl docosahexaenoate hydroperoxide (MDOOH), respectively (Supporting information, Figure S2). A typical exception for the classification into the five groups was glyoxal, a well-known lipid peroxidation product. The di-PFB derivative of glyoxal, the mass spectrum of which (tR ) 16.5 min) exhibited an expected molecular ion at m/z 448, was detected upon the decomposition of all three lipid hydroperoxides (Figure 3A). The di-PFB derivative of malondialdehyde (MDA) generates a characteristic fragment ion at m/z 250 (tR ) 16.8, 17.0 min),

Chem. Res. Toxicol., Vol. 20, No. 1, 2007 101

identical to the key ion of the 2-alkenals; therefore, we included MDA among the group of the 2-alkenals (data not shown). Identification of Aldehydes. Identification of each aldehyde was performed upon analysis in the full scanning mode. For example, the mass spectra for hexanal and 4-hydroxy-2-nonenal (HNE) are shown in Figure 3B and 3C. The ion at m/z 295 (tR ) 10.7 min) showed the expected molecular ion for the PFBhexanal derivative. An ion at m/z 114 [M - 181] was also observed. In addition, the mass spectrum and retention time were totally consistent with those of the authentic hexanal (data not shown). The ions at m/z 242 [M - 181] and m/z 226 [M OPFB] (tR ) 10.7 min) showed the expected fragment ions for the PFB/TMS-HNE derivative. The mass spectrum and retention time were totally consistent with those of the authentic HNE (data not shown). The derivatives of hexanal, HNE, and several other aldehydes were detected as a pair of peaks due to the generation of the syn- and anti-PFB-oxime derivatives. When the authentic compounds were available, we identified the peaks of the aldehydes using a similar strategy. If the authentic compounds were not available, a structural characterization was performed by indirect determination as to whether an aldehyde contained the CH3- or COOCH3-terminal structures that originated from the parent polyunsaturated fatty acid compared with the formation from other structurally related polyunsaturated fatty acids. The representative examples are shown in Figure 4. The CH3- and COOCH3-terminal structures of arachidonic acid are analogous with those of γ-linolenic acid and eicosapentaenoic acid, respectively. Among the four alkanals (peaks 1-4) detected upon the peroxidation of methyl arachidonate at m/z 239, peaks 1 and 2 were identified to be pentanal and hexanal, respectively, on the basis of a comparison with authentic aldehydes. These two aldehydes were also detected upon the peroxidation of methyl γ-linolenate. These two peaks were also detected from the peroxidized methyl linoleate (data not shown). These results indicated that pentanal and hexanal were the ω-6-derived (CH3-terminal-containing) products. Peaks 3 and 4 could be detected in the peroxidized methyl eicosapentaenoate but not in the peroxidized γ-linolenate, suggesting that these aldehydes were the COOCH3-containing aldehydes. In addition to this information, the expected molecular ions of peak 3 (m/z 311) and peak 4 (m/z 325) were detected (Supporting information, Figure S3), and these peaks were finally identified to be the methyl esters of 4-oxo-butanoate and 5-oxovaleroate, respectively. Similarly, among the three peaks detected at m/z 250 (for the 2-alkenals and MDA), peak 1′ was presumed to be an ω-6-derived product, and peak 2′ was a COOCH3-containing aldehyde. In addition, the expected molecular ions of peak 1′ (m/z 321) and peak 2′ (m/z 351) were detected (Supporting information, Figure S3). On the basis of this information, peaks 1′ and 2′ were identified to be 2-octenal and methyl 7-oxo-5-heptenoate, respectively. Peak 3′, which was detected upon the peroxidation of all the polyunsaturated fatty acids, was identified as MDA. Characterization of 4-Hydroxy-2-alkenals. Upon the decomposition of MAOOH, we detected four 4-hydroxy-2-alkenals (HNE and peaks a-c) by GC/EI/MS-SIM at m/z 352 (Figure 5A). The formation of HNE and peak b was also observed from the peroxidized 1-palmitoyl-2-arachidonyl-phosphatidylcholine (PAPC), suggesting that these two 4-hydroxy-2-alkenals were ω-6-derived (CH3-terminal-containing) aldehydes. Similarly, the formation of these two aldehydes was also confirmed upon the peroxidation of ω-6 methyl γ-linolenate (Supporting information, Figure S4). Peaks a and c were detected in the peroxidized methyl eicosapentaenoate (Supporting information, Figure S4),

102 Chem. Res. Toxicol., Vol. 20, No. 1, 2007

Kawai et al.

Figure 2. Representative GC/EI/MS-SIM profiles for aldehydes formed upon the decomposition of MAOOH. All of the PFB-oxime derivatives (i.e., all aldehydes) were monitored by SIM at m/z 181. At the same time, each aldehyde observed at m/z 181 was also classified into one of the five groups by SIM of the ions specific to the aldehyde classes.

suggesting that these two 4-hydroxy-2-alkenals were the COOCH3-containing aldehydes. It is generally known that 4-hydroxy-2-alkenals are highly reactive with sulfhydryl groups compared to other aldehydes (16). When the decomposed MAOOH samples were treated with NAC or NAL (as control)

for 15 min prior to derivatization, the peak ratio (NAC treatment/ NAL treatment) of the 4-hydroxy-2-alkenals was significantly decreased compared to that of the other aldehydes (Figure 5B). These results showed that the four peaks are indeed derived from 4-hydroxy-2-alkenals. The identification of peaks a-c was

Lipidomic Analysis of Aldehydes by GC/EI/MS

Chem. Res. Toxicol., Vol. 20, No. 1, 2007 103

Figure 3. Mass spectra for glyoxal (A), hexanal (B), and HNE (C). The SIM chromatograms of glyoxal (at m/z 181 and m/z 448), hexanal (at m/z 181 and m/z 239), and HNE (at m/z 181 and m/z 352) are illustrated in the inset.

performed by a determination of their molecular weights on the basis of TBDMS-derivatization, which exhibits the characteristic ion [M - 57]. The samples were first derivatized with PFBHA followed by derivatization with MTBSTFA instead of BSTFA. The expected [M - 57] ions of peak a (m/z 438), peak b (m/z 448), and peak c (m/z 478) were detected (Supporting information, Figure S5). On the basis of this information, peaks

a-c were finally identified to be 8-oxo-5-hydroxy-6-octenoate methyl ester, 4-hydroxy-2,6-dodecadienal, and 11-oxo-8-hydroxy-5,9-undecadienoate methyl ester (Figure 5C), respectively. Application of Aldehyde Lipidomics to in ViWo Samples. As described above, the aldehydes formed during the peroxidation of arachidonic acid, linoleic acid, and docosahexaenoic acid were characterized by the GC/EI/MS-SIM system, and at

104 Chem. Res. Toxicol., Vol. 20, No. 1, 2007

Kawai et al.

Figure 4. Determination of the partial structures of oxidized arachidonate-derived aldehydes in comparison with the formation from the analogous polyunsaturated fatty acids. (A) Chemical structures (methyl ester forms) of arachidonic acid (AA), γ-linolenic acid (γ-LNA), and eicosapentaenoic acid (EPA). The CH3- and COOCH3-terminal structures of AA are analogous to those of γ-LNA and EPA, respectively. (B) SIM chromatograms of alkanals (m/z 239, left) and 2-alkenals (m/z 250, right) formed during the peroxidation of AA (top), γ-LNA (middle), and EPA (bottom). The formation of peak 2′ in oxidized EPA is also illustrated in the inset. Peak allocation is as follows: 1, pentanal; 2, hexanal; 3, methyl-4-oxo-butanoate; 4, methyl-5-oxo-valeroate; 1′, 2-octenal; 2′, methyl-7-oxo-5-heptenoate; and 3′, malondialdehyde.

least 33 aldehydes were identified. A list of these aldehydes is shown in Table 1. This list includes the CH3-terminal aldehydes (14 compounds), the COOCH3-terminal aldehydes (17 compounds), and the non-terminal aldehydes (2 compounds). The structures and mass spectra information of all aldehydes identified in this study are summarized in the Supporting information (Figure S6). Unfortunately, several aldehydes that were expected to form, including acrolein, crotonaldehyde, and 4-oxo-2-nonenal, could not be detected under our experimental conditions. To apply this lipidomic method to in ViVo samples, we first developed an internal standard to semiquantitatively analyze the aldehydes. Tridecanal was used as the internal standard because it is not a lipid peroxidation product, and the PFB-oxime derivative could be clearly separated from the other aldehydes. The tridecanal-PFB oxime derivative eluted at tR ) 17.9 min and exhibited an alkanal-derived fragment ion at m/z 239 as well as m/z 181. Standard curves with authentic aldehydes using tridecanal as the internal standard were developed for the semiquantitative study. Standards run through the entire procedure for GC-derivatization gave standard curves with an R2 of at least 0.93 (data not shown). We then analyzed the liver samples of mice intraperitoneally injected (7.5 mmol/kg body weight) with bromobenzene, which is an animal model that strongly induces lipid peroxidation in the liver (17). Under this experimental condition, increased thiobarbituric acid-reactive substances (TBARS), a general marker for lipid peroxidation, were observed (0 h, 24.8 nmol/g tissue; 24 h, 90.3 nmol/g tissue). The homogenates were mixed with the internal standard,

Figure 5. Characterization of 4-hydroxy-2-alkenals formed during the peroxidation of arachidonate. (A) SIM chromatograms for 4-hydroxy2-alkenals (m/z 352) of decomposed MAOOH (upper) and oxidized PAPC (lower). (B) Reactivity of the 4-hydroxy-2-alkenals with the sulfhydryl group. Decomposed MAOOH samples were treated with NR-acetyl-cysteine (NAC, treated) or NR-acetyl-leucine (NAL, control) for 15 min prior to derivatization. The peak ratio (NAC treatment/ NAL treatment) of the 4-hydroxy-2-alkenals in SIM at m/z 352 was calculated. (C) Chemical structures of four 4-hydroxy-2-alkenals identified in the decomposed MAOOH.

derivatized, and then analyzed by GC/EI/MS-SIM. Figure 6 shows the representative SIM chromatogram of the liver samples of mice injected with bromobenzene. We could simultaneously detect nine major aldehydes by SIM of each aldehyde-specific mass. Interestingly, HHE, an ω-3-derived aldehyde, was clearly detected in the liver samples, suggesting the peroxidation reaction of ω-3 polyunsaturated fatty acids such as DHA. Alternatively, we could observe the time-dependent changes in these aldehydes by expressing the peak area of each aldehyde as the ratio to that of tridecanal (Figure 7). Unexpectedly, the increases in HNE and MDA were relatively slight compared to those of the other aldehydes. In contrast, significant increases in hexanal, HHE, glyoxal, 2-hydroxyheptantal, and 2-hydroxybutanal (∼3.5 fold) were time-dependently observed. These aldehydes, except for glyoxal, then decreased after 8 h. HHE reached a maximum after 2 h, which was earlier than the other aldehydes, because of the higher susceptibility of the ω-3 polyunsaturated fatty acids, especially DHA. We confirmed that the liver of mice contained 6-8% DHA of the total fatty acids (data not shown). We calculated that the basal levels of hexanal,

Lipidomic Analysis of Aldehydes by GC/EI/MS

Chem. Res. Toxicol., Vol. 20, No. 1, 2007 105

Table 1. Summary of the Aldehydes Identified in This Study classes

aldehyde namesa

fatty acid sourcesb

alkanals

pentanal hexanal 4-oxo-butanoate 5-oxo-valeroate 8-oxo-octanoate 9-oxo-nonanoate

AA, LA AA, LA AA, DHA AA LA LA

2-alkenals

2-pentenal 2-octenal 6-oxo-4-hexenoate 7-oxo-5-heptenoate

DHA AA, LA DHA AA

2,4-alkadienals

2,4-heptadienal 2,4-decadienal 8-oxo-4,6-octadienoate 9-oxo-5,7-nonadienoate

DHA AA, LA DHA AA

2-hydroxyalkanals

2-hydroxybutanal 2-hydroxyheptanal 5-oxo-4-hydroxypentanoate 6-oxo-5-hydroxyhexanoate 10-oxo-9-hydroxydecanoate

DHA AA, LA DHA AA LA

4-hydroxy-2-nonenals 4-hydroxy-2-hexenal 4-hydroxy-2-nonenal 4-hydroxy-2,6-nonadienal 7-oxo-4-hydroxy-5-heptenoate 8-oxo-5-hydroxy-6-octenoate 4-hydroxy-2,6-dodecadienal 4-hydroxy-2,6,9-dodecatrienal 10-oxo-7-hydroxy-4,8-decadienoate 11-oxo-8-hydroxy-5,9-undecadienoate 12-oxo-9-hydroxy-10-dodecenoate 13-oxo-10-hydroxy-4,7,11-tridecatrienoate

DHA AA, LA DHA DHA AA AA DHA DHA AA LA DHA

others

AA, LA, DHA AA, LA, DHA AA

glyoxal malondialdehyde 4,5-epoxy-2-decenal

a

Aldehydes expressed as “-ate” were all identified as their methyl esters. b Abbreviations: AA, arachidonic acid; LA, linoleic acid; DHA, docosahexaenoic acid. The methyl esters of these fatty acids were used in the peroxidation reactions in Vitro.

HNE, MDA, glyoxal, and 2,4-decadienal were 6.2, 0.18, 7.8, 11.8, and 12.1 nmol/g liver, respectively.

Discussion The lipid peroxidation reaction generates lipid hydroperoxides as the primary products, and decomposition leads to the formation of a variety of aldehydes as end products. Although aldehydes are generally recognized to be reactive with biomolecules containing sulfhydryl and amino groups and implicated in the cell death and/or mutations, it has been reported that the bioactivities of each aldehyde may differ. For example, HNE, among the various lipid peroxidation products, specifically induces the cyclooxygenase-2 expression (6). In contrast to the higher reactivity with sulfhydryl groups and the cytotoxicity of HNE, its mutagenic potential is relatively weak, probably due to its lower reactivity to DNA (8, 18). Glyoxal and 4-oxo-2nonenal (ONE) have been reported to be far more reactive with the 2′-deoxynucleosides than other aldehydes (8). The lipid peroxidation levels in ViVo are generally evaluated by (i) an overall estimation as TBARS or conjugated dienes and (ii) a measurement of specific lipid peroxidation markers such as F2isoprostanes. In addition to the accepted approaches for evaluating lipid peroxidation, we developed a method for the comprehensive analysis of lipid peroxidation-derived aldehydes using the GC/EI/MS-SIM system. Although the HPLC-UV analysis of the 2,4-dinitrophenylhydradine (DNPH)-derivatized aldehydes is also an alternative method for the comprehensive analysis of aldehydes, the retention times of several aldehyde-DNPH

Figure 6. Representative SIM chromatograms of the liver of C57BL/ 6J mice injected with bromobenzene (7.5 mmol/kg body weight, 24 h after injection). Aldehydes were comprehensively detected in the SIM of the ions characteristic for each aldehyde class. The chromatogram for 2-octenal is illustrated in a ×50 magnification versus that of MDA. Abbreviations: 2HH, 2-hydroxyheptanal; 2HB, 2-hydroxybutanal; DDE, 2,4-decadienal; GO, glyoxal.

adducts overlap each other. GC might provide a higher resolution than HPLC. Indeed, we clearly separated most of the aldehydes on the basis of our analytical conditions (Figure 3). The PFB-oxime derivatization produces the characteristic fragmentation of m/z 181 upon GC/EI/MS analysis of all of the aldehydes. In addition to the m/z 181 fragment ion, the PFB or PFB/TMS derivatives of the aldehydes exhibited the characteristic fragment ions specific to their aldehyde classes (Figure 2). On the basis of these mass spectrometric results of the derivatives, we have classified the aldehydes into five major groups (alkanals, 2-alkenals, 2,4-alkadienals, 2-hydroxyalkanals, and 4-hydroxy-2-alkenals) by SIM. We have finally identified at least 33 aldehydes formed upon the peroxidation of arachidonic acid, linoleic acid, and DHA. However, several major lipid-aldehydes including acrolein, crotonadehyde, and ONE could not be detected using our system. The PFB derivatives of acrolein and crotonaldehyde may elute in the front of the analytical program, and then the ions are masked by an excess of the nonspecific ions that originated from the derivatizing reagents and/or residual lipids in the samples. It may also be important to measure ONE because of its higher reactivity and bioactivities (8, 9, 19). However, at least using our current conditions, we could not identify the formation of ONE both in the peroxidized fatty acids and in the livers of the bromobenzene-injected mice. This may be due to the instability of ONE during the lipid peroxidation (9) and the derivatization efficiency of the two carbonyl groups of ONE. We are now developing improved procedures for detecting these aldehydes. In comparison with the GC/MS analysis with the EI mode, negative ion-chemical ionization (NICI) generally has the advantage of the higher sensitivity to detect PFB derivatives. To obtain more detailed information for lipid peroxidation in ViVo, similar lipidomic analyses using GC/NICI/MS may also be required in the future.

106 Chem. Res. Toxicol., Vol. 20, No. 1, 2007

Figure 7. Time-dependent changes in aldehydes in the liver of mice injected with bromobenzene (n ) 3). The ratios of the peak areas of the analyte to the internal standard at 0 h (control mice) were defined as 1.00, and the increasing ratios (∼fold) were then calculated. In the group of 24 h, only two data points were used for calculating because, unfortunately, one of the mice in the group died before 24 h. The amounts of aldehydes in the control liver were as follows: hexanal ) 6.2, HNE ) 0.18, MDA ) 7.8, glyoxal ) 11.8, and 2,4-decadienal ) 12.1 nmol/g liver. The authentic compounds for HHE and 2-hydroxyalkanals were not available in this study.

In this study, we did a comprehensive analysis of the liver samples of mice injected with bromobenzene. It has been reported that bromobenzene hepatotoxicity represents a good model for the study of in ViVo lipid peroxidation (17). We have successfully detected nine major aldehydes in the liver of bromobenzene-treated mice and semiquantitatively monitored the time-dependent changes. We observed unique differences in the formation of these aldehydes. It is of interest that we also detected slight increases in HNE and MDA in the liver samples. This may be, at least in part, due to the higher reactivities of HNE and MDA with biomolecules such as proteins. Protein-bound HNE and MDA have indeed been detected in various in ViVo samples (20, 21). In contrast to HNE, we observed a significant increase in HHE, an ω-3-derived 4-hydroxy-2-alkenal. It has been reported that the covalent binding capacity of the 4-hydroxy-2-alkenals to the head group of the phosphatidylethanolamine depends on hydrophobicity (4hydroxy-2,6-decadienal > HNE > HHE) (10). Little attention has been paid to the potential bioactivities of other 4-hydroxy-

Kawai et al.

2-alkenals except for the well-known HNE and HHE. Our GC/EI/MS-SIM system, which contains 11 types of 4-hydroxy2-alkenals, may, therefore, be a useful method for evaluating the formation of various 4-hydroxy-2-alkenal analogues. Our results also indicate that glyoxal may be a stable marker for lipid peroxidation even 24 h after treatment with bromobenzene (Figure 7). Thus, our results indicate the importance of monitoring not only a specific aldehyde or the total amount of aldehydes (such as TBARS) but also all individual aldehydes. Among the 33 aldehydes identified in this study, 17 were COOCH3-terminal aldehydes. Most of the carboxyl groups in the fatty acids were generally thought to be esterified with phospholipids, triacylglycerol, and/or cholesteryl esters in ViVo; therefore, most of the 17 aldehydes, if formed in ViVo, might be esterified with lipids as ester-core aldehydes. The processes for evaluating these ester-core aldehydes such as transmethylation with sodium methylate were not included in this study. In view of the potential importance of ester-core aldehydes in ViVo, the development of procedures for detecting ester-core aldehydes in our GC/EI/MS-SIM system is also required for future investigations. Applications of this GC/EI/MS-SIM system to other tissue samples are also required for future investigations. In this study, we focused on the peroxidation of three major polyunsaturated fatty acids. Although arachidonic acid is widely distributed in various tissues, linoleic acid and DHA are enriched in the lipoproteins (1) and brain (22), respectively. We have also shown the fatty acid source(s) of aldehydes detected by GC-EI/MSSIM (Table 1). MDA and glyoxal are classified as universal markers because they were commonly formed upon the peroxidation of three polyunsaturated fatty acids. In contrast, several aldehydes were specifically generated from only one of three polyunsaturated fatty acids, for example, HHE (from DHA), 5-oxo-valeroate (from arachidonate), and 9-oxo-nonanoate (from linoleate). Thus, our analytical system enables us to characterize the peroxidation of a specific polyunsaturated fatty acid in tissues. In summary, we have developed an analytical method for comprehensively evaluating the lipid peroxidation-derived aldehydes using GC/EI/MS-SIM. The lipid peroxidation products have gained attention as possible biomarkers of the risk for cardiovascular diseases, cancer, and neurodegenerative diseases associated with oxidative stress. Our approach should contribute to a better understanding of the oxidative status in ViVo and the cytotoxic/genotoxic mechanisms associated with lipid peroxidation products. Acknowledgment. This work was supported by a Grantin-Aid for Exploratory Research from the Ministry of Education, Culture, Sports, Science, and Technology and by the Human Nutritional Science on Stress Control’s 21st Century Center of Excellence Program (COE). Supporting Information Available: Preparation and decomposition of lipid hydroperoxides, supporting mass spectra for identifying aldehydes, and a summarized table for the aldehydes identified in this study. This material is available free of charge via the Internet at http://pubs.acs.org.

References (1) Esterbauer, H., Dieber-Rotheneder, M., Waeg, G., Striegl, G., and Jurgens, G. (1990) Biochemical, structural, and functional properties of oxidized low-density lipoprotein. Chem. Res. Toxicol. 3, 77-92. (2) Lee, S. H., Oe, T., and Blair, I. A. (2001) Vitamin C-induced decomposition of lipid hydroperoxides to endogenous genotoxins. Science 292, 2083-2086.

Lipidomic Analysis of Aldehydes by GC/EI/MS (3) Gutteridge, J. M., and Halliwell, B. (1990) The measurement and mechanism of lipid peroxidation in biological systems. Trends Biochem. Sci. 15, 129-135. (4) Marnett, L. J., Riggins, J. N., and West, J. D. (2003) Endogenous generation of reactive oxidants and electrophiles and their reactions with DNA and protein. J. Clin. InVest. 111, 583-593. (5) Esterbauer, H., Schaur, R. J., and Zollner, H. (1991) Chemistry and biochemistry of 4-hydroxynonenal, malonaldehyde and related aldehydes. Free Radical Biol. Med. 11, 81-128. (6) Kumagai, T., Kawamoto, Y., Nakamura, Y., Hatayama, I., Satoh, K., Osawa, T., and Uchida, K. (2000) 4-Hydroxy-2-nonenal, the end product of lipid peroxidation, is a specific inducer of cyclooxygenase-2 gene expression. Biochem. Biophys. Res. Commun. 273, 437-441. (7) Kumagai, T., Matsukawa, N., Kaneko, Y., Kusumi, Y., Mitsumata, M., and Uchida, K. (2004) A lipid peroxidation-derived inflammatory mediator: identification of 4-hydroxy-2-nonenal as a potential inducer of cyclooxygenase-2 in macrophages. J. Biol. Chem. 279, 4838948396. (8) Kawai, Y., Uchida, K., and Osawa, T. (2004) 2′-Deoxycytidine in free nucleosides and double-stranded DNA as the major target of lipid peroxidation products. Free Radical Biol. Med. 36, 529-541. (9) Shibata, T., Iio, K., Kawai, Y., Shibata, N., Kawaguchi, M., Toi, S., Kobayashi, M., Kobayashi, M., Yamamoto, K., and Uchida, K. (2006) Identification of a lipid peroxidation product as a potential trigger of the p53 pathway. J. Biol. Chem. 281, 1196-1204. (10) Bacot, S., Bernoud-Hubac, N., Baddas, N., Chantegrel, B., Deshayes, C., Doutheau, A., Lagarde, M., and Guichardant, M. (2003) Covalent binding of hydroxy-alkenals 4-HDDE, 4-HHE, and 4-HNE to ethanolamine phospholipid subclasses. J. Lipid Res. 44, 917-926. (11) Yamagata, S., Murakami, H., Terao, J., and Matsushita, S. (1984) Decomposition of products of methyl arachidonate monohydroperoxides, Agric. Biol. Chem. 48, 101-109. (12) Kato, Y., Mori, Y., Makino, Y., Morimitsu, Y., Hiroi, S., Ishikawa, T., and Osawa, T. (1999) Formation of N-(hexanonyl)lysine in protein exposed to lipid hydroperoxide. A plausible marker for lipid hydroperoxide-derived protein modification. J. Biol. Chem. 274, 2040620414.

Chem. Res. Toxicol., Vol. 20, No. 1, 2007 107 (13) Porter, N. A., Lehman, L. S., Weber, B. A., and Smith, K. J. (1981) Unified mechanism for polyunsaturated fatty acid autoxidation. Competition of peroxy radical hydrogen atom abstraction, β-scission, and cyclization, J. Am. Chem. Soc. 130, 6447-6455. (14) Dix, T. A., and Marnett, L. J. (1985) Conversion of linoleic acid hydroperoxide to hydroxy, keto, epoxyhydroxy, and trihydroxy fatty acids by hematin. J. Biol. Chem. 260, 5351-5357. (15) Wu, G. S., and Rao, N. A. (1999) Activation of NADPH oxidase by docosahexaenoic acid hydroperoxide and its inhibition by a novel retinal pigment epithelial protein. InVest. Ophthalmol. Visual Sci. 40, 831-839. (16) Esterbauer, H., Ertl, A., and Scholz, N. (1976) The reaction of cysteine with R,β-unsaturated aldehydes. Tetrahedron 32, 285-289. (17) Benedetti, A., Pompella, A., Fulceri, R., Romani, A., and Comporti, M. (1986) Detection of 4-hydroxynonenal and other lipid peroxidation products in the liver of bromobenzene-poisoned mice. Biochim. Biophys. Acta 876, 658-666. (18) Chung, F. L., Chen, H. J., Guttenplan, J. B., Nishikawa, A., and Hard, G. C. (1993) 2,3-Epoxy-4-hydroxynonanal as a potential tumorinitiating agent of lipid peroxidation. Carcinogenesis 14, 2073-2077. (19) Doorn, J. A., and Petersen, D. R. (2002) Covalent modification of amino acid nucleophiles by the lipid peroxidation products 4-hydroxy2-nonenal and 4-oxo-2-nonenal. Chem. Res. Toxicol. 15, 1445-1450. (20) Toyokuni, S., Uchida, K., Okamoto, K., Hattori-Nakakuki, Y., Hiai, H., and Stadtman, E. R. (1994) Formation of 4-hydroxy-2-nonenalmodified proteins in the renal proximal tubules of rats treated with a renal carcinogen, ferric nitrilotriacetate. Proc. Natl. Acad. Sci. U.S.A. 91, 2616-2620. (21) Yamada, S., Kumazawa, S., Ishii, T., Nakayama, T., Itakura, K., Shibata, N., Kobayashi, M., Sakai, K., Osawa, T., and Uchida, K. (2002) Immunochemical detection of a lipofuscin-like fluorophore derived from malondialdehyde and lysine. J. Lipid Res. 42, 11871196. (22) Svennerholm, L. (1968) Distribution and fatty acid composition of phosphoglycerides in normal human brain. J. Lipid Res. 9, 570-579.

TX060199E