Comprehensive analyses of oxidized phospholipids using a measured MS/MS spectra library[S]

Oxidized phospholipids (OxPLs) are widely held to be associated with various diseases, such as arteriosclerosis, diabetes, and cancer. To characterize the structure-specific behavior of OxPLs and their physiological relevance, we developed a comprehensive analytical method by establishing a measured MS/MS spectra library of OxPLs. Biogenic OxPLs were prepared by the addition of specific oxidized fatty acids to cultured cells, where they were incorporated into cellular phospholipids, and untargeted lipidomics by LC-quadrupole/TOF-MS was applied to collect MS/MS spectra for the OxPLs. Based on the measured MS/MS spectra for about 400 molecular species of the biogenic OxPLs, we developed a broad-targeted lipidomics system using triple quadrupole MS. Separation precision of structural isomers was optimized by multiple reaction monitoring analysis and this system enabled us to detect OxPLs at levels as low as 10 fmol. When applied to biological samples, i.e., mouse peritoneal macrophages, this system enabled us to monitor a series of OxPLs endogenously produced in a 12/15-lipoxygenase-dependent manner. This advanced analytical method will be useful to elucidate the structure-specific behavior of OxPLs and their physiological relevance in vivo.

lipidomic workflow that can monitor low abundant and unknown molecular species, including OxPLs (14,15). In this procedure, untargeted lipidomics was first performed by high-resolution MS, and all detected ions were putatively identified using an online database based on m/z values without fragmentation. Then, the structural identification of selected ions was performed based on MS/MS spectra and MRM mode was applied for validation. They showed the presence of more than 100 molecular species of OxPLs by applying this procedure to the activated human platelets. However, the structure of oxidized fatty acyl chains in many molecular species of the OxPLs were not determined because of the low abundance of OxPLs generated by human platelets, and the diagnostic fragments of oxidized fatty acyl chains were hardly acquired. In this study, we first aimed to develop a precise MS/MS spectra library for OxPLs using a series of biogenic materials. Untargeted lipidomics was applied to collect MS/MS spectra for biogenic OxPLs prepared by the addition of oxidized fatty acids to HEK293 cells, where they were incorporated into cellular PLs. This procedure made it easier to identify the precise OxPL structures based on MS/MS spectra, because oxidized fatty acid was selectively incorporated into cellular PLs that produce selective OxPL molecular species. By using these MS/MS spectra for biogenic OxPLs, we successfully optimized MRM conditions and developed a broad-targeted lipidomics system to monitor about 400 molecular species of OxPLs simultaneously. This system will be useful to determine the physiological relevance of OxPLs in health and diseases.

Preparation of biogenic OxPLs
HEK293 cells were cultured in DMEM supplemented with 10% FCS (Gibco, Carlsbad, CA) and 100 U/ml penicillin (Gibco), 100 g/ml streptomycin (Gibco), and 2 mM L-glutamine (Gibco) in a 37°C incubator with 5% CO 2 in air. For preparation of OxPLs, HEK293 cells (1.0 × 10 6 cells) were plated on a 6 cm dish and incubated with 10 M of oxidized fatty acids for 1 h. Cells were harvested in ice-cold methanol and were extracted by solid-phase extraction using a monospin C18 column (GL Sciences, Tokyo, Japan). Briefly, the samples were applied to a monospin C18 column preconditioned with 300 l methanol and 300 l water and then washed with 300 l water and 300 l hexane followed by the elution with 250 l isopropanol. The phosphorus content of the extracted lipids was quantified by the method of Bartlett (16); the extracted lipids were reconstituted to 1 mM phosphorus with chloroform:methanol (1:2) and then stored at 80°C until use.

Measurement of OxPL levels in primary mouse peritoneal macrophages
Male C57BL/6J mice (11 weeks) were obtained from CLEA Japan, Inc. (Tokyo, Japan). The 12/15-LOX knockout mice on a C57BL/6 background were from the Jackson Laboratory (Bar Harbor, ME). Peritoneal cells were obtained by lavage and macrophages were isolated by adherence to plastic culture plates in RPMI1640 medium supplemented with 10% FCS for 2 h at 37°C. Adherent cells were harvested and lipids were extracted by solidphase extraction, as described above. As internal standards, PC

Untargeted lipidomics
Untargeted lipidomics was performed using an ACQUITY UPLC system (Waters, Milford, MA) coupled with a quadrupole/ TOF MS (TripleTOF 5600 + and TripleTOF 6600; Sciex, Framingham, MA). LC separation was performed using a reverse-phase column [ACQUITY UPLC BEH peptide C18 (50 × 1.7 mm inner diameter, 2.1 m particle size; Waters)] with a gradient elution of mobile phase A [methanol/acetonitrile/water (1:1:3, v/v/v) containing 5 mM ammonium acetate (Wako Chemicals, Osaka, Japan) and 10 nM EDTA (Dojindo, Kumamoto, Japan)] and mobile phase B (100% isopropanol containing 5 mM ammonium acetate and 10 nM EDTA); the composition was produced by mixing those solvents. The LC gradient consisted of a holding solvent (A/B:100/0) for 1 min, then linearly converting to solvent (A/B:60/40) for 4 min, linearly converting to solvent (A/B:36/64) for 2.5 min, and holding for 4.5 min, then linearly converting to solvent (A/B:17.5/82.5) for 0.5 min, linearly converting to solvent (A/B:5/95) for 1 min followed by returning to solvent (A/B:100/0) and holding for 5 min for re-equilibration. The injection volume was 2 l, the flow rate was 0.300 ml/min, and column temperature was 45°C. Information-dependent acquisition (IDA) mode was applied to confirm each of the biogenic OxPL structures. The source conditions were as follows: temperature, 300°C; curtain gas, 25 psi; ion source gas 1 and 2 at 80 and 40 psi; and ion spray voltage floating at 5.5 kV. The acquisition conditions were as follows: the accumulation time for full scan was 100 ms for scanning a mass range from m/z 75 to m/z 1,250. The accumulation time for each IDA experiment was 50 ms, and collision energies (CEs) were set to 3560 eV with a CE spread of 15 eV in high-resolution mode. IDA criteria were as follows: 10 most intense ions with an intensity threshold above 100 cps, isotope exclusion was set to 1.5 Dam, and an exclusion time of 10 s was set.

Broad-targeted analysis
Broad-targeted analysis was performed using an ACQUITY UPLC system coupled with a triple quadrupole (tripleQ) MS (QTRAP 6500; Sciex). LC separation was performed using a reverse-phase column [ACQUITY UPLC HSS T3 (50 × 2.1 mm inner diameter, 1.8 m particle size; Waters)] with a gradient elution of mobile phase A [methanol/acetonitrile/water (1:1:3, v/v/v) containing 50 mM ammonium acetate and 10 nM EDTA] and mobile phase B (100% isopropanol containing 50 mM ammonium acetate and 10 nM EDTA); the composition was produced by mixing those solvents. The LC gradient consisted of holding solvent (A/B:100/0) for 1 min, then linearly converting to solvent (A/B:50/50) for 4 min, linearly converting to solvent (A/B:36/64) for 7 min, then linearly converting to solvent (A/B:5/95) for 1 min and holding for 1 min followed by returning to solvent (A/B:100/0) and holding for 5 min for re-equilibration. The injection volume was 3.5 l, the flow rate was 0.350 ml/min, and column temperature was 50°C. MRM mode was applied to detection of OxPLs in biological samples. Selected MRM transitions and CEs are described in Table 1 and supplemental Table S1. For quantification, OxPL standard solutions corresponding to 10, 20, 50, 100, 200, and 500 nM were prepared to acquire calibration curves for concentration and efficiency of ionization. One microliter of those solutions was injected and measured as described above. Calibration curves were obtained from the concentrations and the area of intensity of each OxPL.

Preparation of purified OxPL standards by soybean 15-LOX
OxPL standards were generated in accordance with a previous method (10) .0) and 10 mM deoxycholate at room temperature. The oxidation reaction was monitored by its absorbance at 234 nm and was terminated at 30 min. Hydroperoxide intermediates were reduced with excess sodium borohydride and incubations were extracted by using a monospin C18 column as described above. To separate and isolate conversion products, reverse-phase HPLC was carried out by using a Waters XBridge C18 column (100 × 4.6 mm inner diameter, 5 m particle size; Waters) with mobile phase A [methanol/ water (1:1) containing 0.01% acetic acid] and mobile phase B (100% methanol) at 0.7 ml/min of flow rate for PC and PE, and an ACQUITY UPLC HSS T3 (75 × 3.5 mm inner diameter, 1.8 m particle size; Waters) with mobile phase A [methanol/acetonitrile/water (1:1:3, v/v/v) containing 50 mM ammonium acetate and 10 nM EDTA] and mobile phase B (100% isopropanol containing 50 mM ammonium acetate and 10 nM EDTA) at 0.6 ml/ min for PI, PG, and PS. The concentration of purified standards was assessed by LC-MS/MS after saponification with 0.2 M NaOH at 60°C for 30 min in 50% isopropanol.

Statistical analysis
Results are expressed as mean ± SE. Differences between two groups were tested by the Student's t-test. A significance level of P < 0.05 was used.

Construction of a measured MS/MS spectra library for OxPLs
To acquire MS/MS spectra for OxPLs, we devised a method to prepare various types of OxPLs by use of biogenic conversion from oxidized fatty acids incorporated into cellular PLs. Oxidized fatty acids, such as hydroxyl and epoxy-containing fatty acids, were added to HEK293 cells for 1 h, cells were harvested, and lipids were extracted. Lipid fractions were analyzed by LC-quadrupole/TOF (QTOF)-MS-based untargeted lipidomics to collect MS/ MS spectra for each of the biogenic OxPLs. This method provides automatic MS to MS/MS switching by setting the MS/MS trigger at a low threshold level of intensity and then information-rich MS/MS spectra with high resolution could be acquired in a nonbiased fashion (17)(18)(19). For example, many types of lipids, such as lyso-PLs, PLs, and sphingolipids, were readily detected in lipid extracts of HEK293 cells and the candidate signals for PLs containing 12-HETE were obtained in 12-HETE-treated cells, as determined by the presence of a fragment ion (m/z 319.2) corresponding to mono-oxidized AA (abbreviated as AA+O) (Fig. 1) ), and 12-HETE-specific fragment ion (m/z 179.1), which could distinguish it from other HETEs (Fig. 2). In total, 12 types of OxPC, 8 types of OxPE, 3 types of OxPI, and 2 types of OxPS were identified in extracts from 12-HETE-treated cells (Table 1). This procedure was also applied to other oxidized fatty acids, i.e., HETEs, HEPEs, HDoHEs, HODEs, EETs, EpETEs, and EpDPEs, in order to expand the MS/MS spectra library. Characteristic fragment ions for identification of those OxPLs were individually optimized and MS/MS data of 386 total molecular species were successfully acquired (Fig. 2, Table 1, supplemental Table S1). Full MS/MS spectra data are deposited in the "Computational MS-based metabolomics" section of the RIKEN PRIMe website (http:// prime.psc.riken.jp/Metabolomics_Software/MS-DIAL/ index.html).

Development of a specific and highly sensitive analytical system for OxPLs
The abundance of OxPLs in biological samples is normally low and many of them are short-lived and structurally similar (9)(10)(11)(12)(13)(14)(15). Therefore, the development of a highly specific and sensitive method to monitor OxPL molecular species in vivo is important to understand their physiological and/or pathophysiological roles. MRM analysis with a particular pair of precursor and fragment ions at a specified retention time using LC tripleQ MS was applied to demonstrate  broad-targeted analysis of OxPLs. Based on the fragmentation patterns of MS/MS spectra library, MRM transitions from a targeted precursor ion to its specific fragment ion were carefully selected to avoid peak overlapping for individual OxPL molecular species. For example, structural differences of OxPLs, such as PC (16:0/5-HETE), PC (16:0/8-HETE), PC (16:0/12-HETE), and PC (16:0/ 15-HETE), were distinguished by the optimized MRM transitions, as depicted in Fig. 3A. In the case of OxPLs whose MS/MS spectra could not be acquired by untargeted lipidomics, we predicted fragmentation patterns for them to expand the measurement range as predictive MRM. LC gradient conditions were also optimized to separate similar OxPL structural isomers (Fig. 3B). The CE was optimized by determining the most sensitive value for each biogenic OxPL individually. The optimized MRM conditions for each OxPL molecular species are summarized in Table 1 and supplemental Table S1.

Preparation of purified OxPL standards for determination of the linear ranges and the lower limits of detection
To determine the linear ranges and the lower limits of detection in our analytical system, purified standard compounds were prepared using an in vitro LOX reaction. OxPLs containing 15-HETE as representative OxPL standards were prepared by in vitro oxidation of PC ( (20,21). The peak areas were acquired by OxPC, OxPE, OxPI, OxPS, and OxPG standard solutions corresponding to 10,20,50,100,200, and 500 nM (Fig. 4). The linear ranges were between 10 and 500 fmol for OxPC, OxPE, OxPI, and OxPG and between 50 and 500 fmol for OxPS. These results suggested that the lower limits of detection in our OxPL lipidomics system were 10 fmol for OxPC, OxPE, OxPI, and OxPG and 50 fmol for OxPS on column (19)(20)(21).

DISCUSSION
In this study, we first constructed a measured MS/MS spectra library by applying untargeted lipidomics to biogenic OxPLs prepared by the addition of oxidized fatty acids to cultured cells. Based on the MS/MS spectra for each OxPL molecular species, we developed a broad-targeted lipidomics system by MRM mode. We applied this system to mouse peritoneal macrophages and quantified a series of OxPLs endogenously produced in a 12/15-LOX-dependent manner. These results indicate that our lipidomics system will be applicable to the comprehensive analysis of OxPLs present in biological samples.
The previous analytical limitations were due to the lack of various types of OxPL standards. To solve this problem, we conceived the idea to prepare various types of OxPLs by adding different types of oxidized fatty acids to HEK293 cells, because these exogenous oxidized fatty acids would be incorporated into the cellular PLs by lyso-PL acyltransferase (22,23). In this study, a total of 29 types of oxidized fatty acids were added to HEK293 cells and about 400 molecular species of OxPLs were identified by untargeted lipidomics. Interestingly, some of these OxPLs had unique incorporation patterns for PLs. For example, 15-HETE and 18-HEPE were effectively incorporated into PI; whereas, 12-HETE was preferentially incorporated into PC. Those preferences for oxidized fatty acids may depend on the substrate specificities of the lyso-PL acyltransferases. Therefore, it might be possible to expand the biogenic MS/MS spectra library by manipulating these enzyme activities.
We also found that the LC retention time for OxPLs appeared to be affected by the hydroxyl and epoxy group position in the oxidized fatty acyl chains. For example, epoxy-containing OxPLs eluted later than hydroxyl-containing OxPLs, and OxPLs whose hydroxyl or epoxy group was located closer to the end of acyl chain eluted earlier than other OxPLs. Also, the optimal CE condition appeared to be dependent on the PL head group rather than on the molecular species of oxidized fatty acyl chains (Table 1,  supplemental Table S1).