Simultaneous Detection of Low and High Molecular Weight

Article pubs.acs.org/ac

Simultaneous Detection of Low and High Molecular Weight Carbonylated Compounds Derived from Lipid Peroxidation by Electrospray Ionization-Tandem Mass Spectrometry Ivana Milic,†,‡ Ralf Hoffmann,†,‡ and Maria Fedorova*,†,‡ †

Institute of Bioanalytical Chemistry, Faculty of Chemistry and Mineralogy and ‡Center for Biotechnology and Biomedicine, Universität Leipzig, Deutscher Platz 5, Leipzig 04103, Germany S Supporting Information *

ABSTRACT: Reactive oxygen species (ROS) and other oxidative agents such as free radicals can oxidize polyunsaturated fatty acids (PUFA) as well as PUFA in lipids. The oxidation products can undergo consecutive reactions including oxidative cleavages to yield a chemically diverse group of products, such as lipid peroxidation products (LPP). Among them are aldehydes and ketones (“reactive carbonyls”) that are strong electrophiles and thus can readily react with nucleophilic side chains of proteins, which can alter the protein structure, function, cellular distribution, and antigenicity. Here, we report a novel technique to specifically derivatize both low molecular and high molecular weight carbonylated LPP with 7-(diethylamino)coumarin-3-carbohydrazide (CHH) and analyze all compounds by electrospray ionization-mass spectrometry (ESI-MS) in positive ion mode. CHH-derivatized compounds were identified by specific neutral losses or fragment ions. The fragment ion spectra displayed additional signals that allowed unambiguous identification of the lipid, fatty acids, cleavage sites, and oxidative modifications. Oxidation of docosahexaenoic (DHA, 22:6), arachidonic (AA, 20:4), linoleic (LA, 18:2), and oleic acids (OA, 18:1) yielded 69 aliphatic carbonyls, whose structures were all deduced from the tandem mass spectra. When four phosphatidylcholine (PC) vesicles containing the aforementioned unsaturated fatty acids were oxidized, we were able to deduce the structures of 122 carbonylated compounds from the tandem mass spectra of a single shotgun analysis acquired within 15 min. The high sensitivity (LOD ∼ 1 nmol/L for 4-hydroxy-2-nonenal, HNE) and a linear range of more than 3 orders of magnitude (10 nmol/L to 10 μmol/L for HNE) will allow further studies on complex biological samples including plasma.

O

react with nucleophilic side chains of lysine, cysteine, and histidine residues to predominantly form 1,4-Michael or Schiff base adducts.8,9 Lipid peroxidation has a significant impact on cell homeostasis, since lipids are important structural and functional components of the cell. Biological membranes are complex structures consisting mostly (70−80%) of cholesterols and phospholipids.10 ROS mainly attack the PUFA in phospholipids (PL) at the sn-2 position, which disturbs membrane assembly, fluidity, and permeability but also alters ion transport mechanisms and inhibits metabolic processes.11 Moreover, oxidized PL and especially oxoLPP can be recognized as damage-associated molecular patterns by numerous pattern recognition receptors on immune cells, thus initiating inflammatory and immunogenic responses, which are characteristic for atherosclerosis, inflammatory arthritis, lupus, multiple sclerosis, and type 1 diabetes.12−14 Importantly, the presence of reactive carbonyl groups within oxidized lipid structures is necessary for activation of, for example, Toll-like receptors 2.15 Additionally, protein adducts

xidative stress is characterized by the overproduction of reactive oxygen species (ROS), which can damage virtually all biomolecules within cells and the extracellular matrix.1−3 ROS can introduce a variety of reversible and irreversible oxidative modifications, such as reactive carbonyls (i.e., aldehydes and ketones) on proteins, lipids, or nucleic acids, which are among the most harmful oxidation products.1,4 Their toxicity is well-known, and the extent of protein and nucleic acid carbonylation is often used to measure oxidative stress-derived injuries.5,6 Lipid-bound reactive carbonyls have received less attention, despite growing evidence of their biological significance. Lipids often contain polyunsaturated fatty acids (PUFA), which are very susceptible targets for free radicals due to the high reactivity of the hydrogens at the methylene carbon between two adjacent double bonds. Reactive oxidants can abstract these hydrogens leading to bond rearrangements and addition of molecular oxygen.7 The resulting lipid hydroperoxides can be oxidatively cleaved by ROS to yield carbonylated lipid peroxidation products (oxoLPP), such as alkenals, hydroxy/oxoalkenals, epoxyalkenals, and γ-ketoaldehydes. These oxidation products contain reactive carbonyl groups that are very strong electrophiles and thus can readily © 2012 American Chemical Society

Received: August 15, 2012 Accepted: November 27, 2012 Published: November 27, 2012 156

dx.doi.org/10.1021/ac302356z | Anal. Chem. 2013, 85, 156−162

Analytical Chemistry

Article

sonicated for 15 min. The vesicles (1.5 mmol/L) were oxidized with CuSO4 (75 μmol/L) and ascorbic acid (150 μmol/L) at 37 °C on an orbital shaker for 72 h. Oxidized samples were analyzed immediately or stored at −20 °C. Derivatization with CHH. Oxidized PUFA (10 mmol/L; 5 μL) or PC vesicles (1.5 mmol/L; 5 μL) were individually derivatized with an equally concentrated solution of CHH (50 μL; 50% aqueous acetonitrile) at 37 °C on an orbital shaker for 1 h. Derivatized samples were used directly for analysis or stored at −20 °C. Mass Spectrometry. Samples were diluted (to 10 pmol/ μL) in a mixture of methanol and chloroform (2:1, v/v) containing ammonium formate (5 mmol/L) and analyzed by direct infusion using Nanospray TYP II emitters (BioMedical Instruments, Zöllnits, Germany) on an ESI-LTQ-Orbitrap XL (Thermo Fisher Scientific GmbH, Bremen, Germany) operating in positive ion mode. The transfer capillary temperature was 200 °C, the tube lens voltage was 120 V, and the ion spray voltage was 1.7 kV. MS spectra were acquired in Fourier transform-mass spectrometry (FT-MS) scan mode with a target mass resolution of 100 000 at m/z 400. CID fragmentation experiments used an isolation width of 1−1.5 u. Collision induced dissociation (CID) normalized collision energy (nCE) was optimized with HNE-CHH and CHH-1palmitoyl-2-(9-oxo)-nonanoyl-GPC within the range of 0− 100%. The optimized nCE value (35%) was used for all further experiments. A data-dependent acquisition (DDA) cycle consisted of a FT-MS survey scan followed by consecutive CID fragmentations of the five most abundant ions in the LTQ using an m/z range from 300 to 800 for derivatized PUFA. Analyzed m/z values were excluded from DDA for a period of 300 s using the maximal size of the dynamic exclusion list (500 m/z values). The acquisition period was 15 min. The significantly higher sample complexity in case of PC vesicles demanded an optimized shotgun acquisition method to detect, identify, and structurally characterize all derivatized carbonyls. Thus we employed gas-phase fractionation (GPF), which is a well-established approach in proteomics, to analyze the same sample several times by DDA using consecutive narrow m/z ranges. This allows the analysis of more compounds in each mass range in the same period of time, which provides access to less intense signals that often represent modified species that would be otherwise missed. Gas-phase fractionation (GPF) was applied for PC vesicles using five segments representing five consecutive DDA experiments (top 5) within five different m/z ranges (280− 450, 450−600, 600−750, 750−900, and 900−1100) with each segment being acquired for 3 min. The isolation width was set to 1 u in the first two segments and 1.5 u in the following three segments. Analyzed m/z values were excluded from DDA for only 45 s allowing the collection of at least four tandem spectra for each compound. The total analysis time was 15 min. As the ion trap requires a certain offset to isolate the highest number of precursor ions,24 one representative signal was selected for each m/z range in GPF mode to optimize the isolation offset. Interestingly, the isolation offset depended on the m/z value (Figure S6 in the Supporting Information). In the lowest m/z range (280−450), the highest signal intensity was obtained for an offset of zero. The optimum isolation offset increased stepwise by 0.1 u for each of the subsequent GPF segments, reaching 0.4 u for the upper mass range (m/z 900−1100). For each segment a specific offset for precursor ion isolation in the

with oxoLPP, such as hydroxynonenal (HNE) and malondialdehyde (MDA) were shown to activate a Th17 cell response, leading to the production of autoantibodies directed to HNEand MDA-modified host molecules.16 Despite much interest in the field of oxidative lipidomics, only very few oxoLPP structures and products have been identified, which are most commonly detected after derivatization with thiobarbituric acid.17 This method, however, suffers from low sensitivities and specificities. Its combination with HPLC improves sensitivity but still does not provide any structural information.18 More recently, advances in mass spectrometry have allowed the identification of lipids and lipid peroxidation products in complex mixtures with high sensitivities in short periods of time. Domingues’ group was able to identify numerous PL-bound LPP species such as oxygen adducts and oxygen-driven cleavage products by tandem mass spectrometry.19−22 Low molecular weight aldehydes, however, were not detected. These low molecular weight oxoLPP, such as malondialdehyde, acrolein, and crotonaldehyde, are typically analyzed by gas chromatography/mass spectrometry (GC/MS) after derivatization.23 To the best of our knowledge, no method has been reported so far that simultaneously detects low (volatile aliphatic) and high (lipidesterified) molecular weight aldehydes/ketones. Here we present a novel approach to analyze both classes of oxoLPP by electrospray ionization-tandem mass spectrometry (ESI-MS/MS) after derivatization of aldehyde and keto groups with 7-(diethylamino)coumarin-3-carbohydrazide (CHH). Its hydrophobicity allowed the consecutive extraction of all CHHderivatized oxoLPP, and its superior ionization efficiency in positive ion mode ESI-MS/MS permitted a good sensitivity and linear range of more than 3 orders of magnitude. Thus, 122 compounds were identified from a mixture of four in vitrooxidized PUFA and phosphatidylcholines (PC).

■

MATERIALS AND METHODS Chemicals. Copper(II) sulfate was purchased from Fluka Chemie GmbH (Buchs, Switzerland). HNE (>95% GC) was obtained from Enzo Life Science (Lausen, Switzerland). Ascorbic acid, tert-butyl methyl ether (MTBE), CHH and oleic, linoleic, arachidonic, docosahexaenoic, and formic acids were supplied by Sigma-Aldrich GmbH (Taufkirchen, Germany). 1-Palmitoyl-2-oleoyl-sn-phosphatidylcholine (POPC), 1-palmitoyl-2-linoleoyl-sn-phosphatidylcholine (PLPC), 1-stearyl-2-arachidonyl-sn-phosphatidylcholine (SAPC), and 1-palmitoyl-2-docosadehanoyldocosahexaenoylsn-phosphatidylcholine (PDPC) were purchased from Avanti Polar Lipids, Inc. (Alabaster, Alabama). Acetonitrile and methanol (ULC/MS grade) were obtained from Biosolve BV (Valkenswaard, Netherlands). Ethanol was from Carl Roth GmbH & Co. KG, and chloroform was from Merck KGaA (Darmstadt, Germany). Oxidation of PUFA and PC Vesicles. Oleic, linoleic, arachidonic, or docosahexaenoic acids (10 mmol/L) were incubated in a mixture of ethanol and water (10:90; v/v) containing CuSO4 (0.5 mmol/L) and ascorbic acid (1 mmol/ L) at 37 °C on an orbital shaker for 24, 48, and 72 h. Oxidized samples were analyzed immediately or stored at −20 °C. To prepare a solution of PC vesicles (6 mmol/L), POPC, PLPC, SAPC, and PDPC (0.1625 mg each) were mixed in chloroform (365 μL) in a flat-bottomed borosilicate glass vial, slowly dried by a stream of nitrogen, rehydrated in aqueous ammonium bicarbonate (3 mmol/L, pH 7.6; 143 μL), and 157

dx.doi.org/10.1021/ac302356z | Anal. Chem. 2013, 85, 156−162

Analytical Chemistry

Article

Figure 1. LTQ-CID mass spectrum of HNE-CHH (precursor m/z 414.2); the club symbol indicates fragment ions derived from the CHH group; ★ indicates fragment ions derived from the HNE moiety.

compounds in sample mixtures. The derivatization efficiency, MS behavior, and limit of detection were evaluated for HNE, which is commonly used to study carbonyl-containing lipid peroxides, in positive ion mode. For HNE-CHH (m/z 414.24), the limit of detection was 1 nmol/L (Figure S-2 in the Supporting Information) and the linear range was more than 3 orders of magnitude from 10 nmol/L to 10 μmol/L, with a correlation coefficient (R2) of 0.9941. The tandem mass spectrum recorded for HNE-CHH under CID conditions in the linear ion trap displayed three reporter ions, i.e., the base peak at m/z 244.0 (Figure 1) and two minor signals at m/z 261.9 and 276.0, which confirmed the CHH derivatization. The most intense HNE-specific fragment ion at m/z 396.1 indicated loss of water from the C-4 position, which carries the hydroxyl group. This elimination most likely generates a double bond between C-4 and C-5, which can be cleaved as indicated by the signal at m/z 326.0. The signals at m/z 340.1 and 354.3 can be explained by alternative bond cleavages between C-5 and C-6 and between C-6 and C-7, respectively, as CID favorably cleaves alkenyl chains adjacent to sp2-hybridized carbon atoms. In this respect, the signals at m/z 302.2 and 314.1 suggest another double bond between carbon atoms C-2 and C-3. Overall, the HNE-specific fragments provided sufficient structural information for locating the double bond and the hydroxyl group. The detection of structurally important ions was further improved by optimizing the collision energies for precursor (m/z 414.2), reporter (m/z 244.0), and structurally significant fragment ions (e.g., m/z 396.1, 326.0, 302.2) (Figure S-3A in the Supporting Information). The optimal nCE was 35%, which was used for all consecutive studies. Detection of oxoLPP in oxidized PUFA. Docosahexaenoic (DHA, 22:6), arachidonic (AA, 20:4), linoleic (LA, 18:2), and oleic (OA, 18:1) acids, which are the four most abundant PUFA in mammalian cells,31,32 were oxidized and analyzed either directly or after CHH derivatization by shotgun ESI-MS in positive ion mode. The mass spectra of the underivatized oxidized PUFA (Figure S-4A in the Supporting Information) did not contain any signals for the expected carbonylated LPP, most likely due to their high volatilities (i.e., loss during sample preparation) and low ionization efficiencies. These compounds were detected in the derivatized samples (Figure S-4B in the Supporting Information). On the basis of

LTQ was used (0, 0.1, 0.2, 0.3, and 0.4 Da). Recorded spectra were analyzed with Xcalibur (version 2.0.7, Thermo Fisher). The tandem mass spectra recorded for the derivatized oxoLPP were interpreted manually.

■

RESULTS AND DISCUSSION Derivatization of Reactive Carbonyls with CHH. Lipid peroxidation products containing reactive carbonyls (oxoLPP), i.e., aldehydes and ketones, represent important pro-inflammatory and immunogenic mediators in many human disorders.13,14 An increasing amount of evidence indicates that oxoLPP might be promising biomarker candidates for early diagnosis and evaluation of disease onset.25,26 However, analysis by mass spectrometry, which is the most versatile technique for discovering biomarkers, is challenged by the complexity of oxoLPP and their low content in biological samples. These carbonyl compounds have generally low ionization efficiencies in ESI and MALDI due to their low proton affinities and relatively high hydrophobicities.27 Chemical derivatization with reagents providing high proton affinities could enhance their ionization in positive ion mode and thus provide access to all oxoLPP. The derivatization usually relies on hydrazides or hydrazines, such as biotin hydrazide or dinitrophenylhydrazine (DNPH).28−30 DNPH derivatization of oxidized phospholipids (PL) gives access to apolar lipid-esterified carbonylated compounds (or high molecular weight oxoLPP), but the water-soluble (or low molecular weight) oxoLPP are lost.29 In order to overcome the aforementioned limitations, with oxoLPP we have tested an alternative reagent containing CHH, assuming that the tertiary amine would improve the ionization efficiency and that the hydrazide would react rapidly and specifically with aldehydes and ketones. Additionally, the hydrophobicity and relatively high mass of CHH were expected to enable the simultaneous extraction of both short aliphatic and apolar high molecular weight carbonyls with organic solvents. The CHH reagent was well ionized in positive ESI-MS and displayed a dominant fragment ion at m/z 244.2 in CID, corresponding to a neutral loss of hydrazine (−N2H4; −32 u), besides two weaker signals at m/z 261.8 and 218.1 (Figure S-1 in the Supporting Information). These fragment ions were used as reporter ions for identification of accordingly derivatized 158

dx.doi.org/10.1021/ac302356z | Anal. Chem. 2013, 85, 156−162

Analytical Chemistry

Article

Table 1. Number of oxoLPP Identified after Oxidation of Different PUFA and PC Vesicles in Vitroa

a

type of carbonyl

oleic acid

linoleic acid

arachidonic acid

docosahexaenoic acid

lipid vesicles

small aldehyde dicarbonyl hydroxy-alkanal/alkenal oxocarboxylic acid PC-bound oxoLPP total

7 3 4 3

8 5 11 19

8 5 13 35

9 3 15 32

17

43

61

59

12 5 14 19 36 86

Detailed information is provided in Tables S1 and S2 in the Supporting Information.

Figure 2. LTQ-CID mass spectra of CHH-propenal (A, precursor m/z 316.2); CHH-glyoxal (B, m/z 573.2); CHH-5-oxo-3-pentenoic acid (C, m/z 374.2); and CHH-3-hydroxy-6-oxo-4-hexenoic acid (D, m/z 402.2) produced by PUFA oxidation. The club symbol indicates fragment ions derived from the CHH group; ★ indicates fragment ions derived from the oxoLPP moiety.

were detected here as well. For several carbonylated species, the acid specificity also was confirmed, e.g., 2-pentanal was present only in the oxidized DHA samples. To the best of our knowledge, however, the number of oxoLPP presented here is higher than in previous MS studies. Furthermore, peroxidation of OA yielded predominantly small aldehydes, while the peroxidation of LA, AA, and DHA produced mostly hydroxyalkenals and oxocarboxylic acids. The CHH reporter ion at m/z 244 represented the base peak in all tandem mass spectra acquired for CHH-oxoLPP. Derivatized carboxyl groups were not observed. All product ion spectra also displayed compound-specific fragmentation patterns, as exemplified for four different types of oxoLPP in Figure 2. Propanal (exact mass 58.08 amu), a saturated aliphatic aldehyde, was detected after CHH derivatization at m/z 316.16. Because of the low number of carbon atoms and the lack of double bonds, propanal-CHH yielded a rather poor product ion spectrum with only a few CHH-specific signals (Figure 2A).

the high mass accuracy of the MS scan and the product ion spectra acquired for the five most intense signals present in each MS survey scan, a total of 69 aliphatic carbonyls were unambiguously identified in the oxidized OA, LA, AA, and DHA samples (Table S-1 in the Supporting Information). These carbonyls can be divided into four groups according to their structure: small aldehydes (e.g., propanal), dicarbonyl compounds (e.g., glyoxal), hydroxyalkanals/hydroxyalkenals (e.g., 4-hydroxy-2-nonenal), and oxocarboxylic acids (e.g., oxononenoic acid) (Table 1). Peroxidation of oleic acid (one double bond) produced only 17 oxoLPP, whereas the number of identified aldehydes increased with the number of additional double bonds to 43, 61, and 59 for LA, AA, and DHA, respectively. These data were in good agreement with previously published results, where the number of detected carbonyls increased with the number of double bounds in PUFA. Kawai et al.23 identified 33 carbonylated compounds in a mixture of LA, AA, and DHA by GC/MS, of which 26 (75%) 159

dx.doi.org/10.1021/ac302356z | Anal. Chem. 2013, 85, 156−162

Analytical Chemistry

Article

Figure 3. LTQ-CID mass spectra of CHH-1-palmitoyl-2-(8-oxooctanoyl)-GPC (A, precursor m/z 893.5); CHH-1-palmitoyl-2-(9-oxononanoyl)GPC (B, m/z 907.6); CHH-1-palmitoyl-2-(3-hydroxy-9-oxononanoic acid)-GPC (C, m/z 923.6); and CHH-1-palmitoyl-2-(3-hydroxy-6,7-epoxy-12oxododecanoic acid)-GPC (D, m/z 979.6) produced by oxidation of PC vesicles. The club symbol indicates fragment ions derived from the CHH group; ★ indicates fragment ions derived from the oxoLPP moiety.

also showed the characteristic losses of 18 and 46 m/z units (m/z 384.1 and 356.0, respectively). The signal at m/z 302.0 assigns the double bond between C-2 and C-3. The 3-hydroxy group was significantly cleaved under CID conditions, providing an additional signal (m/z 366.1) that corresponds to the loss of a second water molecule from the precursor. As discussed for hydroxyalkenals, the fragmentation pattern revealed the position of the hydroxyl group by inducing a bond cleavage between C-4 and C-5 (m/z 343.9) followed by a loss of water (m/z 326.0). Detection of oxoLPP in Oxidized PC Vesicles. On the basis of the data obtained for oxoLPP generated by oxidation of PUFA, we extended our analytical approach to oxidized phosphatidylcholines (PC), which represent the dominant phospholipids in mammalian cell membranes. PC vesicles containing lipids with the four most abundant PUFA, i.e., PLPC (16:0/18:2), POPC (16:0/18:1), SAPC (18:0/20:4), and PDPC (16:0/22:6), displayed in ESI-MS the expected protonated/sodiated signals at m/z 758.57/780.55, 760.58/ 782.56, 806.57/828.55, and 810.60/832.58, respectively (Figure S-5A in the Supporting Information). The signals at m/z 820.49, 822.50, 868.49, and 872.52 were assigned as [M + Zn]+ species based on the mass accuracy (error ± 5 ppm). The Zn2+-

Glyoxal is a dialdehyde with a very similar mass (exact mass 58.00 amu) and one of the end products of lipid peroxidation. Because of the presence of two carbonyl groups, double derivatization was observed (m/z 573.24), which was confirmed by the signals at m/z 297.9 and 312.1 (Figure 2B), corresponding to the loss of CHH−CO−NHN + H2 and CHH−CO−NH + H2, respectively. Additionally, fragments that retain the CHH group can be observed at m/z 276.1 and 261.0. Other dialdehyde compounds, such as malondialdehyde and oxobutenal, showed similarly characteristic derivatization and fragmentation patterns (Table S-1 in the Supporting Information). Many products were oxocarboxylic acids, which have attracted little attention in earlier studies. The tandem mass spectrum of CHH-derivatized 5-oxo-3-pentenoic acid (m/z 374.17, Figure 2C) showed a relatively intense signal at m/z 356.0, corresponding to the loss of water. The signal at m/z 327.9 indicated a loss of formic acid, which appears to be characteristic for carboxylic acids and was not observed for other oxoLPP. The signals at m/z 302.0 and 314.1 indicated a double bond between carbon atoms C-3 and C-4, as described above for HNE (Figure 1). Another typical example was 3hydroxy-6-oxo-4-hexenoic acid (m/z 402.16; Figure 2D), which 160

dx.doi.org/10.1021/ac302356z | Anal. Chem. 2013, 85, 156−162

Analytical Chemistry

Article

water losses (−54 u) from the precursor ion, clearly indicating the presence of the hydroxyl group (Figure 3C). Additional fragment ions at m/z 566.16, 580.15, and 594.18 corresponded to neutral losses of −CHH−C6H12 (−357 u), −CHH−C5H10 (−343 u), and −CHH−C4H8 (−329 u), respectively. The relative intensities of these three signals localized the hydroxyl group at C-7 (counting from the derivatized carbonyl group). These observations and interpretations were confirmed, for example, by CHH-derivatized 1-palmitoyl-2-(3-hydroxy-6,7epoxy-12-oxododecanoic acid)-GPC (Figure 3D). The elemental composition of the oxoLPP moiety within the lipidbound carbonyl was determined from the high-resolution MS as C12H20O5, indicating two oxygen atoms to be present along the carbon chain. The signal at m/z 925.39 corresponded to a neutral loss of three water molecules, indicating only one hydroxyl group. Therefore, the second oxygen within the oxoLPP was assumed to form an epoxide. The signals at m/z 636.2, 650.1, 664.2, and 678.1 were assigned as neutral losses of CHH−C5H10, CHH−C4H8, CHH−C3H6, and CHH−C2H4. Considering their relative signal intensities, an oxygen should be located at position C-6 (counting from the derivatized carbonyl group). The peaks at m/z 566.1 and 580.2 can be explained by additional losses of 70 m/z units from the two most intense signals of the first ion series (m/z 636.2 and 650.1), corresponding to a loss of −CH2−CH−-(CH)2O. This would be consistent with the presence of an epoxy group between positions C-6 and C-7 (counting from the carboxy terminus). The relative signal intensities of m/z 566.1 and 580.2 also indicate a hydroxyl group at C-3 (counting from the carboxy terminus). Following these fragmentation rules we were able to propose likely structures for the other 36 lipidbound and 50 aliphatic oxoLPP present in the oxidized PC vesicles.

ions might come from glass vials used for the formation and storage of lipid vesicles.33 Peroxidation of the lipid vesicles (Cu(II) ions/ascorbic acid) decreased the signal intensities of native PC, whereas new signals indicating short-chain and longchain LPP appeared (Figure S-5B in the Supporting Information). Small aliphatic oxoLPP were not detected, while several carbonylated lipid species (high molecular weight) were observed (Figure S-5B in the Supporting Information, blue asterisk). CHH derivatization enhanced the ionization of both aliphatic and lipid bound carbonyl-containing LPP, giving access to both small, aliphatic, water-soluble, and large apolar lipid-esterified carbonylated species (Figure S-5C in the Supporting Information, red asterisk). To the best of our knowledge, this is the first time that both compound classes have been simultaneously detected in a mixture of oxidized lipids. Having optimized the isolation offset and nCE of each GPF segment (Figure S-3B,C in the Supporting Information), it was possible to identify 50 aliphatic and 36 lipid-bound CHHderivatized carbonyls in the oxidized PC vesicles (Table S-2 in the Supporting Information). The glycerophosphatidylcholine (GPC) and the additional saturated fatty acid influenced the fragmentation characteristics of the detected carbonyls significantly. Tandem mass spectra of PC-bound carbonyls (Figure 3) displayed neutral losses of 275 (CHH), 260 (CHH fragment C14H16N2O3), and 458 u (phosphorylcholine, −HPO4(CH2)2N(CH3)3 and CHH). These characteristic mass losses were present in the product ion spectra of all PC-bound oxoLPP and thus can serve as specific reporter ions. The presence of these neutral losses compensates for the loss of the aforementioned CHH reporter ion at m/z 244 caused by the low mass cutoff characteristic for an ion trap mass analyzer. Thus, there was no need to use high-energy collision-induced dissociation (HCD) or pulsed Q collision-induced dissociation (PQD) techniques to cover the lower mass region, which both suffer from low sensitivity in the LTQ-Orbitrap-XL instrument used in this study. Therefore, the high molecular weight oxoLPP can be analyzed on any tandem mass spectrometer that employs CID fragmentation. All GPC-bound carbonyls always showed two fragment ions corresponding to successive losses of two water molecules, even in the absence of aliphatic hydroxyl groups. The tandem mass spectra of the four PC-bound CHHoxoLPP displayed neutral losses of trimethylamine (59 u, −N(CH3)3) and phosphorylcholine (183 u, −HPO4(CH2)2N(CH3)3), and deesterification of the saturated fatty acid (e.g., 256/238 u for palmitic acid and 284/266 u for stearic acid, −RCOOH/RCO) (Figure 3). Ions at m/z 496/478 (for palmitoyl-oxoLPP-GPC) and 524/506 (for stearyl-oxoLPPGPC), corresponding to a loss of CHH-oxoLPP (−CHH− RoxCOOH/CHH−RoxCO), i.e., the unsaturated fatty acid in the original phospholipid, confirm additionally the presence of the CHH tag on the GPC-bound carbonyl. Similar to the derivatized low molecular weight PUFA-derived oxoLPP, the CID spectra provided all signals necessary for revealing the structure of the carbonylated species in the sn-2 position of the GPC. CHH-derivatized 1-palmitoyl-2-(8-oxooctanoyl)-GPC (Figure 3A) and 1-palmitoyl-2-(9-oxononanoyl)-GPC (Figure 3B), for example, showed the same neutral losses of 315 (−CHH−C3H6), 329 (−CHH−C4H8), and 343 m/z units (−CHH−C5H10) indicating fragmentation along the oxoLPP. Hydroxylated lipid-bound oxoLPP, such as 1-palmitoyl-2-(3hydroxy-9-oxononanoic acid)-GPC, showed three consecutive

■

CONCLUSIONS A new analytical approach based on CHH derivatization of oxoLPP derived from lipid peroxidation allowed us to identify 122 carbonylated compounds in oxidized PUFA and PC vesicles. This was accomplished by the high derivatization specificity of CHH as well as the superior properties of the CHH-derivatized compounds, i.e., their extraction efficiencies in organic solvents and ionization in positive ion-mode ESI-MS. Fragmentation by CID provided enough information to deduce the structures of most compounds, including the position of double bonds, hydroxyl groups, and epoxides for both the water-soluble, aliphatic (low molecular weight) and the apolar (high molecular weight) oxoLPP. Although applied here only to oxoLPP mixtures generated in vitro, the analytical technique should also be suited for complex biological samples like plasma and tissue extracts.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest. 161

dx.doi.org/10.1021/ac302356z | Anal. Chem. 2013, 85, 156−162

Analytical Chemistry

■

Article

(28) Uchiyama, S.; Inaba, Y.; Kunugita, N. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2011, 879, 1282−1289. (29) Milic, I.; Fedorova, M.; Teuber, K.; Schiller, J.; Hoffmann, R. Chem. Phys. Lipids 2012, 165, 186−196. (30) Nagy, K.; Pollreisz, F.; Takats, Z.; Vekey, K. Rapid Commun. Mass Spectrom. 2004, 18, 2473−2478. (31) Quehenberger, O.; Armando, A. M.; Brown, A. H.; Milne, S. B.; Myers, D. S.; Merrill, A. H.; Bandyopadhyay, S.; Jones, K. N.; Kelly, S.; Shaner, R. L.; Sullards, C. M.; Wang, E.; Murphy, R. C.; Barkley, R. M.; Leiker, T. J.; Raetz, C. R.; Guan, Z.; Laird, G. M.; Six, D. A.; Russell, D. W.; McDonald, J. G.; Subramaniam, S.; Fahy, E.; Dennis, E. A. J. Lipid Res. 2010, 51, 3299−3305. (32) Rapoport, S. I.; Chang, M. C.; Spector, A. A. J. Lipid Res. 2001, 42, 678−685. (33) Binder, H.; Arnold, K.; Ulrich, A. S.; Zschornig, O. Biophys. Chem. 2001, 90, 57−74.

ACKNOWLEDGMENTS Financial support from the European Regional Development Fund (ERFD, European Union and Free State Saxony) and the “Bundesministerium für Bildung and Forschung” (BMBF) are gratefully acknowledged. We thank Dr. Andrew Hagan for proofreading.

■

REFERENCES

(1) Sayre, L. M.; Perry, G.; Smith, M. A. Chem. Res. Toxicol. 2008, 21, 172−188. (2) Berlett, B. S.; Stadtman, E. R. J. Biol. Chem. 1997, 272, 20313− 20316. (3) Cadenas, E. Annu. Rev. Biochem. 1989, 58, 79−110. (4) Butterfield, D. A. Antioxid. Redox Signal. 2006, 8, 1971−1973. (5) Buss, I. H.; Winterbourn, C. C. Methods Mol. Biol. 2002, 186, 123−128. (6) Dalle-Donne, I.; Aldini, G.; Carini, M.; Colombo, R.; Rossi, R.; Milzani, A. J. Cell. Mol. Med. 2006, 10, 389−406. (7) Spiteller, P.; Kern, W.; Reiner, J.; Spiteller, G. Biochim. Biophys. Acta 2001, 1531, 188−208. (8) Zhu, X.; Tang, X.; Anderson, V. E.; Sayre, L. M. Chem. Res. Toxicol. 2009, 22, 1386−1397. (9) Doorn, J. A.; Petersen, D. R. Chem. Biol. Interact. 2003, 143−144, 93−100. (10) Zimmerberg, J.; Gawrisch, K. Nat. Chem. Biol. 2006, 2, 564− 567. (11) Catala, A. Chem. Phys. Lipids 2009, 157, 1−11. (12) Winyard, P. G.; Ryan, B.; Eggleton, P.; Nissim, A.; Taylor, E.; Lo Faro, M. L.; Burkholz, T.; Szabo-Taylor, K. E.; Fox, B.; Viner, N.; Haigh, R. C.; Benjamin, N.; Jones, A. M.; Whiteman, M. Biochem. Soc. Trans. 2011, 39, 1226−1232. (13) Fu, P.; Birukov, K. G. Transl. Res. 2009, 153, 166−176. (14) Wu, W.; Chen, Y.; d’Avignon, , A.; Hazen, S. L. Biochemistry 1999, 38, 3538−3548. (15) West, X. Z.; Malinin, N. L.; Merkulova, A. A.; Tischenko, M.; Kerr, B. A.; Borden, E. C.; Podrez, E. A.; Salomon, R. G.; Byzova, T. V. Nature 2010, 467, 972−976. (16) Duner, P.; To, F.; Berg, K.; Alm, R.; Bjorkbacka, H.; Engelbertsen, D.; Fredrikson, G. N.; Nilsson, J.; Bengtsson, E. Atherosclerosis 2010, 212, 457−465. (17) Rael, L. T.; Thomas, G. W.; Craun, M. L.; Curtis, C. G.; Bar-Or, R.; Bar-Or, D. J. Biochem. Mol. Biol. 2004, 37, 749−752. (18) Chirico, S.; Smith, C.; Marchant, C.; Mitchinson, M. J.; Halliwell, B. Free Radic. Res. Commun. 1993, 19, 51−57. (19) Reis, A.; Domingues, M. R.; Amado, F. M.; Ferrer-Correia, A. J.; Domingues, P. Biomed. Chromatogr. 2005, 19, 129−137. (20) Domingues, M. R.; Reis, A.; Domingues, P. Chem. Phys. Lipids 2008, 156, 1−12. (21) Reis, A.; Domingues, M. R.; Amado, F. M.; Ferrer-Correia, A. J.; Domingues, P. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2007, 855, 186−199. (22) Reis, A.; Domingues, P.; Ferrer-Correia, A. J.; Domingues, M. R. Free Radic. Res. 2007, 41, 432−443. (23) Kawai, Y.; Takeda, S.; Terao, J. Chem. Res. Toxicol. 2007, 20, 99−107. (24) Schuhmann, K.; Herzog, R.; Schwudke, D.; Metelmann-Strupat, W.; Bornstein, S. R.; Shevchenko, A. Anal. Chem. 2011, 83, 5480− 5487. (25) Samhan-Arias, A. K.; Ji, J.; Demidova, O. M.; Sparvero, L. J.; Feng, W.; Tyurin, V.; Tyurina, Y. Y.; Epperly, M. W.; Shvedova, A. A.; Greenberger, J. S.; Bayir, H.; Kagan, V. E.; Amoscato, A. A. Biochim. Biophys. Acta 2012, 1818, 2413−2423. (26) Guichardant, M.; Chen, P.; Liu, M.; Calzada, C.; Colas, R.; Vericel, E.; Lagarde, M. Chem. Phys. Lipids 2011, 164, 544−548. (27) Eggink, M.; Wijtmans, M.; Ekkebus, R.; Lingeman, H.; de Esch, I. J.; Kool, J.; Niessen, W. M.; Irth, H. Anal. Chem. 2008, 80, 9042− 9051. 162

dx.doi.org/10.1021/ac302356z | Anal. Chem. 2013, 85, 156−162