- Email: [email protected]
A multiplex quantitation method for eicosanoids and platelet-activating factor using column-switching reversed-phase liquid chromatography–tandem mass spectrometry
ANALYTICAL BIOCHEMISTRY Analytical Biochemistry 342 (2005) 134–143 www.elsevier.com/locate/yabio
A multiplex quantitation method for eicosanoids and platelet-activating factor using column-switching reversed-phase liquid chromatography–tandem mass spectrometry 夽 Yoshihiro Kita, Toshie Takahashi, Naonori Uozumi, Takao Shimizu ¤ Department of Biochemistry and Molecular Biology, Faculty of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan Received 4 March 2005 Available online 20 April 2005
Abstract Eicosanoids and platelet-activating factor (PAF) are phospholipid-derived lipid mediators produced by various tissues and cells through a cascade pathway. For a comprehensive analysis of these lipid mediators, a simultaneous quantitation method with sensitivity and reliability is necessary. This article details a development of column-switching reversed-phase liquid chromatography–tandem mass spectrometry for multiplex quantitation of eicosanoids and PAF. The adsorptive nature of lipids caused signiWcant loss of signal in a conventional column-switching conWguration. The use of an online-dilution method allowed use of 100% methanol as a sample solvent, which prevented sample adsorption to contacting surfaces. Addition of 0.2% formic acid to the sample solvent was required for the successful introduction of LTC4 to the trapping column and minimizing its carryover. The optimized method provided rapid analysis of 14 lipid mediators with a throughput of 96 samples/24 h, lower limits of quantitation of 5 pg on column, and linear calibration ranges up to 2000–5000 pg. The system was highly compatible with solid-phase-extracted samples, as methanoleluted fractions were directly injected without reconstitution. The analysis of lipid mediator production of macrophage-like RAW264.7 cells demonstrated that the cell-based assay can be performed in a 96-well format, suitable for metabolomics analyses and/or screening strategies. 2005 Elsevier Inc. All rights reserved. Keywords: Lipid mediators; Quantitation; Column-switching; LC-MS/MS
Eicosanoids, such as prostaglandins (PGs),1 leukotrienes (LTs), and hydroxyeicosatetraenoic acids (HETEs), and platelet-activating factor (PAF) are bioactive lipid 夽 This work was supported in part by Grants-in-Aid from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. * Corresponding author. Fax: +81 3 3817 8732. E-mail address: [email protected] (T. Shimizu). 1 Abbreviations used: PG, prostaglandin; LT, leukotriene; HETE, hydroxyeicosatetraenoic acid; PAF, platelet-activating factor; COX, cyclooxygenase; LO, lipoxygenase; ELISA, enzyme-linked immunosorbent assay; RIA, radioimmunoassay; ESI, electrospray ionization; MeOH, methanol; EDTA, ethylenediaminetetraacetic acid; MeCN, acetonitrile; DMSO, dimethyl sulfoxide; HBSS, Hanks’ balanced salt solution; LPS, lipopolysaccharide; FBS, fetal bovine serum; IS, internal standard; SRM, selected reaction monitoring; CID, collision-induced dissociation; PEEK, poly-ether-ether-ketone.
0003-2697/$ - see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2005.03.048
mediators produced by various tissues and cells in vivo, playing vital roles in physiological and pathophysiological conditions [1,2]. These mediators are produced through a cascade pathway (Fig. 1), in which several key enzymes such as phospholipase A2, cyclooxygenases (COXs), and lipoxygenases (LOs) function as common regulators in combination with speciWc terminal synthases that produce speciWc lipid mediator molecular species [3–5]. Various biological stimuli that elicit intracellular signals can activate or modulate this cascade pathway and cause lipid mediator production, the proWle of which may diVer by cell types and/or their activation states. For detailed understanding of biological processes involving lipid mediators, a comprehensive analysis tool for lipid mediators is necessary.
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
135
Fig. 1. Structures and production pathways of lipid mediators. The structures and production pathways of target lipid mediators for quantitation are presented. Names of enzymes involved in the pathway are indicated along the path. Since PGI2 and thromboxane (Tx)A2 are unstable, stable metabolites 6-keto-PGF1 and TxB2 are measured instead. LO, lipoxygenase; LTA4H, LTA4 hydrolase; LTCS, LTC4 synthase; 5-HpETE, 5-hydroperoxyeicosatetraenoic acid.
An enzyme-linked immunosorbent assay (ELISA) [6,7], and a radioimmunoassay (RIA) [8] have been the most widely used immunological techniques for the quantitation of small amounts of lipid mediators. A major limitation of these assays is incompatibility with multiplex analysis; they require much sample when multiple target compounds are quantitated. Although a Xow cytometric immunoassay that use multicolor-labeled beads [9,10] enabled multiplex quantitation, it still requires speciWc antibodies for each target. The antibodies, especially when raised against lipids, sometimes do not provide speciWcity suYcient for the discrimination of structurally similar but biologically diVerent lipid metabolites. Mass spectrometry (MS)-based quantitation techniques such as GC-MS (GC-MS/MS) and LC-MS (LC-MS/MS) have advantages in speciWcity over immunoassays. Furthermore, they are well suited to multiplex analysis, which minimizes the sample amount required.
Recent advances in electrospray ionization (ESI) technology provided an interface between LC and MS, which paved the way for the analysis of bioactive lipids without derivatization [11]. The present report describes development of a multiplex quantitation method that covers eicosanoids and PAF using a column-switching reversed-phase LC-MS/MS technique. The columnswitching method was adopted for the eVective sample introduction to the analytical column [12]. This strategy provides substantial improvements in the sensitivity of the system as compared with previously reported methods that use direct sample injection to analytical column [13,14], as a large volume of sample can be introduced. The present report also describes the necessity of an online-dilution technique for sample introduction in combination with the column-switching system. The use of the column-switching method for the analysis of eicosanoids has been described [15]. The present optimized
136
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
system provides better recovery and sensitivity to perform a cell-based assay in a 96-well culture plate and a throughput to process 96 samples/24 h.
Materials and methods Chemicals HPLC-grade methanol (MeOH), formic acid, ethanol, petroleum ether, and ethylenediaminetetraacetic acid (EDTA) were obtained from Wako (Osaka, Japan). MS-grade acetonitrile (MeCN) was obtained from Kanto (Tokyo, Japan). All eicosanoids and PAF (PAFC16; 1-O-hexadecyl-2-O-acetyl-sn-glycero-3-phosphocholine) reference compounds including deuterium-labeled internal standards were obtained from Cayman (Ann Arbor, MI). Dimethyl sulfoxide (DMSO), RPMI-1640 medium, Hanks’ balanced salt solution (HBSS), and lipopolysaccharide (LPS; from Escherichia coli, Serotype 0111:B4) were obtained from Sigma (St. Louis, MO). Fetal bovine serum (FBS) was obtained from Equitech Bio (Kerrville, TX). A23187 was obtained from Calbiochem (San Diego, CA). Preparation of stock solutions and standards A stock solution of eicosanoids and PAF was prepared as a mixture that contains each compound at a concentration of 1 g/ml in MeOH and stored in a glass vial at ¡80 °C (calibration-master). Deuterium-labeled eicosanoids and PAF (6-keto-PGF1-d4, TxB2-d4, PGF2-d4, PGE2-d4, PGD2-d4, PGB2-d4, LTB4-d4, 5(S)HETE-d8, 12(S)-HETE-d8, 15(S)-HETE-d8, and PAFC16-d4) were mixed at concentrations of 5–50 ng/ml in MeOH, depending on the sensitivity of each compound, and used as an internal standard (IS) solution. The calibration-master solution was further diluted by MeOH:formic acid (100:0.2) to prepare the following standard solutions: 5, 20, 50, 200, 500, 2000, and 5000 pg/ 47.5 l. These solutions were mixed with IS solution at a ratio of 19:1 to prepare IS-containing calibration samples of 5–5000 pg/50 l. Accordingly, these calibration samples contained 5–50 pg/50 l of deuterium-labeled compounds. For the standard addition and recovery experiments, standard samples without IS were prepared independently. Cell culture, stimulation, sample collection, and pretreatment Murine macrophage-like RAW264.7 cells were maintained in RPMI-1640 culture medium supplemented with 10% FBS at 37 °C, 5% CO2 in air. Cells were seeded onto 96-well culture plates at a density of 1 £ 105 cells/ well. Two hours later, cells were treated with or without
100 ng/ml LPS for 12 h (LPS priming). After washing three times with HBSS, cells were stimulated with 2 M A23187 or vehicle (0.1% DMSO) in 100 l HBSS for 30 min. Lipid mediator production was terminated by adding equal amount (100 l) of MeOH:formic acid (100:0.2). After the addition of 10 l IS to each well, entire volumes (»200 l/well) were transferred to 1-ml glass collection vials arranged in a standard 96-well (12 £ 8) format using a multichannel pipette. Then, another 100 l of MeOH:formic acid (100:0.2) was added to the well and mixed for 1 h at room temperature. After the addition of 100 l water:formic acid (100:0.03) to each well, the resultant solutions (»200 l/well) were transferred to collection vials (Wnal volume, »400 l/ sample). The samples were submitted to solid phase extraction using an Oasis HLB 96-well cartridge cluster (5 mg; Waters, Milford, MA). BrieXy, samples were diluted with water:formic acid (100:0.03) to give a Wnal MeOH concentration of »20% by volume, applied to preconditioned cartridges, and washed serially with water:formic acid (100:0.03), water:ethanol:formic acid (90:10:0.03), and petroleum ether. Samples were eluted into glass vials with 200 l of MeOH:formic acid (100:0.2) by centrifugation (100g, 1 min) and stored at 4 °C in the autosampler while waiting for injection. The volumes of eluted fraction were 170–180 l, 50 l of which was injected for analysis. Equipment and conditions Four LC10AD pumps (Shimadzu, Kyoto, Japan), a 3033 autosampler (Shiseido, Tokyo, Japan), and an electrically controlled six-port switching valve (Valco, Houston, TX) were conWgured as shown in Fig. 2. Three diVerent conWgurations were used for the optimization and validation of the method. ConWguration A is a simple HPLC system with autosampler, used for the direct injection of a small volume (5 l) of sample. ConWgurations B and C are column-switching systems with diVerent sample introduction methods. ConWguration B is a conventional column-switching system, where an autosampler is located at the downstream of pumps C and D. ConWguration C is a column-switching system with online-dilution Xow path, where an autosampler is located at the downstream of pump D and samples injected are diluted by mobile phase C before the trapping column. The compositions of mobile phases A, B, C, and D were water, MeCN:formic acid (100:0.1), water:formic acid (100:0.1), and MeOH:formic acid (100:0.2), respectively. For the carryover experiment of LTC4, formic acid was removed from mobile phase D for comparison. Opti-Guard Mini C18 (1 £ 15 mm; Optimize Technologies, Oregon City, OR) and Capcellpak C18 MGS3 (1 £ 100 mm; Shiseido, Tokyo, Japan) were used as a trapping column and a separating column, respectively. In conWgurations B and C, during the
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
137
pressure, 2.0–2.2 mTorr (argon gas). Capillary voltages, tube lens voltages, and collision energies were optimized for each compound. For selected reaction monitoring (SRM) mode operations, Q1 and Q3 resolutions were set to »1.5 Da (by full-width at half-maximum). For the peak detection, integration, and quantitative calculation of MS data, the Xcalibur 1.2 software package (Thermo Electron) was used. Calibration curve Calibration curves were calculated by the IS method using peak area ratios of target compounds against IS. Compounds without corresponding commercially available deuterium-labeled counterparts were calibrated by either PGB2-d4 (for LTC4 and LTD4) or 15-HETE-d8 (for 8-HETE and 11-HETE) as IS. Linear regression with 1/x weighting was used for the calculation of coeYcient of determination (R2). Calibration ranges with R2 > 0.99 were considered ranges with suYcient linearity. Accuracy and precision
Fig. 2. LC-MS/MS system conWgurations. (ConWguration A) An ordinary binary solvent system. (ConWguration B) A conventional columnswitching system. Pumps C and D are used to generate isocratic Xow with mobile phase C:mobile phase D ratios of 80:20, 70:30, and 60:40. (ConWguration C) An online-dilution column-switching system. In this system, an autosampler Xow path is Wlled with mobile phase D. Samples introduced are diluted by mobile phase C in the tee connector before the trapping column. The Xow path between the tee connector and the trapping column is minimized to avoid unwanted sample adsorptions. In all conWgurations, poly-ether-ether-ketone materials were used for sample-contacting surfaces unless stated otherwise.
trapping phase (0–3.0 min), 50 l of the sample injected from autosampler was directly introduced (conWguration B) or introduced after dilution with mobile phase C (conWguration C). Samples were introduced to the trapping column at a total Xow rate of 500 l/min. Then, the valve was switched to introduce sample to the separating column (3.0 min). Separation of samples was conducted by a step gradient with mobile phase A:mobile phase B ratios of 64:37 (0–6.0 min), 55:45 (6.0–7.0 min), and 35:65 (7.0 min–next injection; typically 14.5 min) at a Xow rate of 120 l/min and a column temperature of 46 °C. MS data acquisition was started at 3.0 min, which corresponds to 0 min in the chromatograms. A TSQ-7000 triple-stage quadrupole MS instrument equipped with an ESI interface (Thermo Electron, Waltham, MA) was used. Output of LC was directly introduced to ESI without splitting. Source conditions were as follows: negative ESI mode; spray voltage, 3000 V; heated-capillary temperature, 270 °C; sheathe gas (N2) pressure, 80 psi (auxiliary gas not used); collision cell
Accuracy and precision for calibration samples was determined by recalculation of each calibration level to calibration curve generated. Recovery rates calculated as ([calculated amount]/[nominal amount]) £ 100 and coeYcient of variation (CV) for multiple injections (triplicate) were used for the evaluation of accuracy and precision, respectively. Accuracy and precision of biological samples were determined by standard addition-and-recovery test. RAW264.7 cell samples with four diVerent treatments (with regard to LPS priming and A23187 stimulation; see Cell culture, stimulation, sample collection, and pretreatment) were cleaned without adding IS and pooled. These four samples that contained diVerent levels of eicosanoids and PAF were used as biological samples to which standards of known amounts were added. For each sample, 0.1 volume of 10£ concentrated standards of IS-only, IS + 50 pg/200 l, and IS + 500 pg/200 l were added and quantitated (triplicate determinations for each level) using the calibration curve made by standards without biological matrices. Recovery rates calculated as ([calculated amount for IS + 50 (or 500) pg/ 200 l addition] ¡ [calculated amount for IS-only addition])/[nominal amount, e.g., 50 (or 500) pg/200 l] £ 100 and coeYcient of variation calculated for the four diVerent biological samples were used for the determination of accuracy and precision, respectively. Validation of sample pretreatment procedure Consistency of sample pretreatment procedure was validated by the recovery of standard samples. Standard mixtures (500 pg/well in 50 l MeOH, without IS) were
138
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
loaded onto wells of a 96-well culture plate that contained 100 l/well of HBSS, and subsequently 10 l of IS was added to each well. After performing cleanup procedure as described above, samples were quantitated. The net recovery rates of target compounds during solidphase extraction were 80% or more for all target compounds (data not shown).
Results and discussion Detection of eicosanoids and PAF Conditions for the detection of eicosanoids and PAF by MS in SRM mode are summarized in Table 1. All eicosanoids were ionized as deprotonated ionic species ([M ¡ H]¡) in the negative ESI mode, when water, MeCN, and small amounts (0.03–0.08%) of formic acid were used as a spraying solvent. PAF was ionized as a formic acid adduct ([M + HCOOH-H]¡) in this solvent. These molecular-related ions (shown as Q1 values in Table 1) were used as precursor ions for collisioninduced dissociation (CID). As typical fragmentation patterns, CID spectra of PGE2 and PAF are shown in Fig. 3. Dehydrated fragment ions such as m/z 333 and 315 shown in Fig. 3A were avoided for SRM due to the lack of speciWcity. For the detection of PAF, the acetyl fragment (m/z 59) was used for SRM for the structural Table 1 SRM conditions Compounds
Q1a (m/z)
Q3b (m/z)
Collision energy (eV)
6-keto-PGF1 TxB2 PGF2 PGE2 PGD2 LTB4 LTC4 LTD4 5-HETE 8-HETE 11-HETE 12-HETE 15-HETE PAF-C16
369 369 353 351 351 335 624 495 319 319 319 319 319 568
245 195 193 271 271 195 272 177 115 155 167 179 219 59
32 18 25 18 18 20 26 20 20 18 20 16 16 30
6-keto-PGF1-d4 TxB2-d4 PGF2-d4 PGE2-d4 PGD2-d4 PGB2-d4 LTB4-d4 5-HETE-d8 12-HETE-d8 15-HETE-d8 PAF-C16-d4
373 373 357 355 355 337 339 327 327 327 572
249 199 197 275 275 239 197 116 184 226 59
32 18 25 18 18 22 20 20 16 16 30
a b
First quadrupole. Third quadrupole.
Fig. 3. CID spectra of PGE2 and PAF-C16. (A) CID spectrum of PGE2. A fragment ion of m/z 271 corresponding to [M-2H2O-CO2-H]¡ ionic species is selected for SRM transition. (B) CID spectrum of PAF. An acetyl fragment ion, m/z 59, is used for SRM transition. The intensity of molecular-related ion, m/z 568.3, is multiplied (£300) for visualization.
discrimination from lysophosphatidylcholine (Fig. 3B and [16]). Similarly, fragments characteristic for each compound were selected for SRM (Table 1). Results of three diVerent isocratic separations in reversed-phase LC are shown in Fig. 4. Retentions of PGs were weaker than those of LTs, HETEs, and PAF in the C18 column, as it required low MeCN concentration of 37% (Fig. 4A). In this condition, PGE2 and PGD2 were successfully resolved, and this allowed using the same SRM transition (353 ! 271) for the detection of these PGs. Among HETE isomers, 8-HETE and 12-HETE were not resolved in the C18 column (Fig. 4C). Although CID spectra of these HETE isomers suggested a suYcient selectivity of SRM transitions selected (data not shown), potential cross talk was evaluated in the more sensitive SRM mode. When commercially available 8-HETE and 12-HETE were used, cross talk was approximately 0.1% for 8-HETE to 12-HETE SRM and 0.3–0.4% for 12-HETE to 8-HETE SRM (Fig. 5). These values were constant for the injections of 500–5000 pg on column, although cross talk signals by the injection of 500 pg of 8-HETE on column were under the detection limit (data not shown). It was diYcult to trace the origin of this cross talk, since it can be caused either by a minor but true fragmentation or by impurities in the standard compounds. The cross talk was negligible for 12-HETE quantitation, since an injection of 5000 pg of 8-HETE (upper limit of quantitation) on column may cause a signal that is equivalent to »3 pg of 12HETE, which is below the lower limit of quantitation. However, an injection of 5000 pg of 12-HETE on column may cause a signal that is equivalent to 20–30 pg of 8-HETE. Therefore, compensations are required for the quantitation of 8-HETE, especially when samples contain large amounts (near the upper limit of quantitation) of 12-HETE.
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
139
Fig. 5. Cross talk of 8-HETE and 12-HETE SRM. Simultaneous SRM of 8-HETE (319 ! 155) and 12-HETE (319 ! 179) were performed after injecting 5000 pg of 8-HETE and 12-HETE on column. Isocratic separation with mobile phase A:mobile phase B ratio of 35:65 was used. Values on arrows indicate cross talk ratios.
Fig. 6. Simultaneous analysis of 14 lipid mediators by step gradient separation. A typical chromatogram after injecting 50 pg on column. Separation was achieved within 10 min. Fig. 4. Isocratic separation of lipid mediators. (A) Isocratic separation of PGs and LTD4 with mobile phase A:mobile phase B ratio of 63:37. (B) Isocratic separation of LTs with mobile phase A:mobile phase B ratio of 55:45. (C) Isocratic separation of HETEs and PAF-C16 with mobile phase A:mobile phase B ratio of 35:65.
ConWguration of column-switching system Isocratic conditions as shown in Fig. 4 were combined to achieve three-step gradient separation. Fig. 6 shows a representative chromatogram. In this separation mode, 14 lipid mediators were analyzed within 10 min, with less than 15 min cycle time which includes 3 min of sample trapping procedure (throughput of 96 samples/24 h). When a conventional column-switching conWguration (Fig. 2, conWguration B) in which samples were introduced isocratically with 20, 30, or 40% MeOH sample solvent was used, any conditions tested satisWed the sensitivity for simultaneous analysis of PGs, LTs, HETEs, and
PAF (Fig. 7A). In this conWguration, increase in the MeOH concentration from 20 to 30 or 40% caused decrease in the sensitivities of PGs, whereas it caused signiWcant improvements in the sensitivities of LTs, HETEs, and PAF. We hypothesized that the decrease in the sensitivities for PGs in the higher concentration of MeOH may result from the loss of retention by the trapping column, and the poor sensitivities for LTs, HETEs, and PAF in the lower concentration of MeOH may result from the sample adsorption to contacting surfaces before the trapping column. In addition, a signiWcant decrease of signals was found in water-rich sample solvent during storage in glass autosampler vials within 24 h (data not shown), suggesting that the adsorptive nature of lipids may disturb the consistency of quantitation. To circumvent the problem, online dilution of samples dissolved in MeOH:formic acid (100:0.2) was examined (Fig. 2, conWguration C). In this conWguration,
140
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
Optimization for LTC4 analysis As shown in Figs. 4B and 6, LTC4 showed signiWcant peak tailing. It was aVected by formic acid concentrations in the mobile phase; an increase in formic acid concentration of mobile phase B ameliorated the peak tailing of LTC4 (data not shown). The concentration of formic acid in mobile phase B (MeCN:formic acid, 100:0.1) was determined by the peak shape of LTC4 and sensitivities of other compounds. Sensitivity and carryover for LTC4 were analyzed. As shown in Fig. 8, addition of 0.2% formic acid to both sample solvent (MeOH) and mobile phase D was required to achieve sensitivity with acceptable carryover values. When no formic acid was used, up to 40% decrease in the signal intensity was observed (Fig. 8A). This seemed not to be due to the loss of retention in the trapping column; rather it was due to sample adsorption to the Xow path as carryover values signiWcantly increased (Fig. 8B). Interestingly, the addition of formic acid only to mobile phase D improved LTC4 signal, but unacceptable carryover (»3%) occurred. In addition, when stainless steel tubing instead of PEEK tubing was used for the introduction of samples without formic acid, larger reduction of signals and longlasting carryover were observed (data not shown). This result suggests that stainless steel tubing including autosampler sampling needles are less compatible with LTC4 analysis. With regard to target compounds other than LTC4, the eVects of formic acid were negligible, although 0.1% formic acid in mobile phase C was required to
Fig. 7. Optimization of trapping condition in two diVerent columnswitching systems. (A) EVects of MeOH concentrations on sensitivities in conWguration B. MeOH concentrations of sample solvent were adjusted to those of mobile phase. (B) EVects of MeOH concentrations on sensitivities in conWguration C. Sample solvent composition was Wxed to 100% MeOH and dilution ratios were changed to achieve Wnal MeOH concentrations indicated. Data are normalized for each compound and are expressed as mean § SD; the average values for 20% MeOH in conWguration C are set to 100.
samples were diluted by water:formic acid (100:0.1) using a tee connector located adjacent to the trapping column. As shown in Fig. 7B, trapping at MeOH concentration of 20% (Wvefold dilution) provided the best sensitivity for all 14 compounds. Increase in the MeOH concentration to 30 or 40% did not improve sensitivity; rather it caused signiWcant decrease in the sensitivities of PGs and PAF. Trapping eYciencies of PGs and LTD4 as determined by comparison with the direct injection method (Fig. 2, conWguration A) were nearly 100% (Table 2), suggesting that further decrease in MeOH concentration is not necessary.
Fig. 8. EVects of formic acid on sensitivity and carryover of LTC4. Signal intensities upon injection of 5000 pg LTC4 (A) and carryover after injection of 5000 pg LTC4 (B) are shown. The eVects of 0.2% formic acid addition on the sample solvent and/or mobile phase D are compared. Data are expressed as mean § SD.
Table 2 Trapping eYciencies for prostaglandins and LTD4a
ConWguration A ConWguration C EYciencyb (%) a
6-keto-PGF1
TxB2
Area
CV (%)
Area
CV (%)
Area
CV (%)
Area
CV (%)
Area
CV (%)
Area
CV (%)
29193 29545 101.2
4.5 5.1
141694 147346 104.0
5.0 5.1
25966 26787 103.2
6.8 5.2
148293 148276 100.0
2.9 3.3
127354 126723 99.5
2.6 2.9
46049 46017 99.9
4.8 2.5
PGF2
PGE2
LTD4
Samples prepared as 50 pg/5 l in MeOH:water (80:20) were injected by a volume of 5 l and analyzed by isocratic condition (37% mobile phase
B). b
PGD2
DeWned as [Area for ConWguration C]/[Area for ConWguration A] £ 100.
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
141
Using the optimized conditions as described above, the sensitivity and linearity were examined. As shown in Table 3, lower limits of quantitation of 5 pg were achieved for all 14 target compounds examined and linear calibration ranges up to 5000 pg, excepting PGD2 whose linear calibration range was 5–2000 pg, were achieved with R2 > 0.99 in the linear regression with 1/x weighting.
RAW264.7 was selected for this purpose. The cells cultured in a 96-well culture plate at a density of 1 £ 105 cells/well were stimulated with calcium ionophore A23187, and lipid mediator productions within 30 min were measured. To compare diVerent cell states, the cells pretreated with LPS which reportedly induce COX-2 expression [17] were also analyzed (Fig. 9). When not primed with LPS, the cells produced LTC4 as a major product, which was readily stimulated by A23187. LPS priming shifted the cells to produce huge amounts of PGD2. PAF accumulation was also found after LPS pretreatment.
Solid-phase extraction of lipid mediators
Accuracy and precision in biological samples
Sample pretreatment procedure for cell-based assay in a 96-well format was designed and validated. As the online-dilution system allowed the use of 100% MeOH as sample solvent, MeOH:formic acid (100:0.2)-eluted fractions were directly analyzed without drying and reconstitution. As shown in Table 4, standards added to 96-well polystyrene culture plate were consistently recovered.
Using the RAW264.7 cell samples, accuracy and precision of the system were examined. Accuracy determined by the standard addition of 50 pg/200 l and 500 pg/200 l (mimicking the addition of 50 pg/well and 500 pg/well in the 96-well plate) to pooled, pretreated samples that contain diVerent levels of intrinsic eicosanoids and PAF and precision determined by repetitive injections are shown in Table 5. Among the lipid mediators, LTC4 showed signiWcantly biased recovery value of »30% when calibration curve was made with standard samples without the biological matrix. The most probable explanation for the positive bias in biological samples may be the matrix eVects that prevent LTC4 adsorption. Since
achieve suYcient trapping when mobile phase D did not contain formic acid (data not shown). Sensitivity and linearity
Simultaneous quantitation of lipid mediators in RAW264.7 cells The system was examined for use in cell-based biological assays. A murine macrophage-like cell line Table 3 Linearity, accuracy, and precision for reference compoundsa Compounds
Linearity (R2) Calibration Levels 5 pg
20 pg
50 pg
200 pg
500 pg
2000 pg
5000 pg
Recovery CV Recovery CV Recovery CV Recovery CV Recovery CV Recovery CV Recovery CV (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) 6-keto-PGF1 TxB2 PGF2 PGE2 PGD2b LTB4 LTC4 LTD4 5-HETE 8-HETE 11-HETE 12-HETE 15-HETE PAF-C16 a b
0.9989 0.9991 0.9987 0.9993 0.9971 0.9998 0.9991 0.9960 0.9981 0.9993 0.9978 0.9975 0.9993 0.9995
100.8 100.6 105.8 97.5 95.3 97.4 100.7 96.8 98.9 99.2 108.1 95.4 114.5 82.1
5.0 12.0 11.9 9.5 13.1 8.5 6.8 7.3 10.8 15.1 7.0 13.0 9.6 12.1
96.6 92.6 99.4 102.0 103.6 100.4 111.3 112.0 117.2 93.4 90.7 104.7 91.2 97.0
5.9 7.2 10.7 6.4 5.6 4.2 9.3 3.2 17.6 11.9 2.2 12.7 4.1 5.1
104.8 99.0 108.0 99.3 108.9 100.7 108.4 122.8 106.7 95.3 92.6 107.6 96.4 104.0
7.2 2.7 2.7 2.1 2.1 0.4 6.0 0.9 10.4 4.0 2.3 2.4 9.1 4.0
99.7 95.9 101.9 103.9 108.5 96.5 104.9 119.2 104.4 97.1 91.3 103.9 97.8 98.1
0.8 0.8 0.9 1.5 2.5 0.6 1.7 1.6 4.2 6.1 4.9 2.1 4.2 3.6
102.2 95.0 103.6 104.4 106.7 101.7 106.6 117.5 112.0 99.4 94.4 111.6 99.8 102.4
3.2 101.4 2.2 97.3 1.9 101.0 0.8 99.0 1.8 97.2 1.3 97.9 2.5 99.8 0.4 99.7 1.7 99.2 1.3 99.1 1.6 96.0 4.2 102.3 1.1 98.2 2.5 98.3
1.3 4.0 5.6 1.4 3.7 0.7 4.3 2.5 1.4 2.2 3.3 3.1 3.1 3.4
99.2 101.8 99.1 99.8 (80.4) 100.8 99.1 97.4 98.8 100.6 102.6 97.7 100.9 100.5
4.5 0.6 3.2 3.2 (1.6) 0.6 0.4 0.2 3.0 3.0 3.6 4.3 2.3 1.1
Triplicate determinations were performed for each calibration level. Linear regression of PGD2 was applied within the range of 5–2000 pg.
Table 4 Validation of solid-phase extractiona 6-keto-PGF1 TxB2 PGF2 PGE2 PGD2 LTB4 LTC4 LTD4 5-HETE 8-HETE 11-HETE 12-HETE 15-HETE PAF-C16 Recovery (%) 105.5 CV (%) 6.8 a
106.7 104.9 5.8 8.5
104.6 110.0 3.0 9.2
106.4 105.9 98.2 14.3 17.2 6.8
103.5 2.4
112.9 17.5
96.9 21.8
105.1 6.1
88.4 17.8
96.0 5.2
Standard mixtures placed on the 96-well plate (500 pg/well, quadruplicate) were added with IS and quantitated after solid-phase extraction.
142
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
metabolomics analysis and/or screening strategy can be performed. The present strategy can be extended for the quantitation of other eicosanoids and their metabolites, as the system is applicable to lipid metabolites with diverse hydrophobicity. The system can also be applied to other biological samples such as tissue extracts, although the sample pretreatment method should be properly developed and validated. The present method will be an essential tool for the comprehensive analysis of lipid mediators.
Note Fig. 9. Lipid mediator production of RAW264.7 cells. RAW264.7 cells seeded onto 96-well plates at a density of 1 £ 105 cells/well were treated as described under Materials and methods and lipid mediators were measured simultaneously. Data are expressed as mean § SD.
A related manuscript by Kita et al. that describes an application of the present method to the comprenhensive analysis of macrophages can be found in Ref. [18].
Table 5 Accuracy and precision for biological samplesa
References
Compounds
6-keto-PGF1 TxB2 PGF2 PGE2 PGD2 LTB4 LTC4 LTD4 5-HETE 8-HETE 11-HETE 12-HETE 15-HETE PAF-C16
+50 pg/200 l
+500 pg/200 l
Recovery (%)
CV (%)
Recovery (%)
CV (%)
97.8 98.3 94.1 97.2 108.0 99.6 127.6 111.3 89.7 97.5 100.4 100.8 89.4 98.6
3.9 3.0 8.5 7.0 14.9 5.8 8.6 26.6 3.5 11.1 11.2 9.7 8.1 7.1
101.0 101.7 97.6 99.1 93.3 98.4 133.6 106.4 89.2 99.9 100.9 103.6 88.8 106.3
4.7 6.7 4.3 4.8 5.7 5.2 11.2 16.4 2.9 9.7 9.4 4.7 10.1 5.2
a
Data are shown as averages of samples containing diVerent levels of intrinsic compounds (see Materials and methods and Fig. 9).
deuterium-labeled PGB2 is used as IS for LTC4, the eVect cannot be compensated by IS. However, in the cell-based assay where sample matrices are relatively uniform, assay consistency may not be disturbed. Actually, the conserved precision of LTC4 as shown in Table 5 suggests the uniformity of the bias within the samples that contain diVerent levels of LTC4 (less than 10 pg/50 l to more than 100 pg/50 l; see also Fig. 9).
Conclusions A multiplex quantitation method for 14 lipid mediators that covers eicosanoids and PAF was developed. They are primary enzymic products from the common phospholipid precursors. The system was successfully applied to the cell-based analysis of a macrophage-like cell line in a 96-well plate format, demonstrating that
[1] C.D. Funk, Prostaglandins and leukotrienes: advances in eicosanoid biology, Science 294 (2001) 1871–1875. [2] S.M. Prescott, G.A. Zimmerman, D.M. StaVorini, T.M. McIntyre, Platelet-activating factor and related lipid mediators, Annu. Rev. Biochem. 69 (2000) 419–445. [3] D.A. Six, E.A. Dennis, The expanding superfamily of phospholipase A2 enzymes: classiWcation and characterization, Biochim. Biophys. Acta 1488 (2000) 1–19. [4] D.L. Simmons, R.M. Botting, T. Hla, Cyclooxygenase isozymes: the biology of prostaglandin synthesis and inhibition, Pharmacol. Rev. 56 (2004) 387–437. [5] M. Peters-Golden, T.G. Brock, 5-lipoxygenase and FLAP, Prostaglandins Leukot. Essent. Fatty Acids 69 (2003) 99–109. [6] F. Shono, K. Yokota, K. Horie, S. Yamamoto, K. Yamashita, K. Watanabe, H. Miyazaki, A heterologous enzyme immunoassay of prostaglandin E2 using a stable enzyme-labeled hapten mimic, Anal. Biochem. 168 (1988) 284–291. [7] M. Reinke, Monitoring thromboxane in body Xuids: a speciWc ELISA for 11-dehydrothromboxane B2 using a monoclonal antibody, Am. J. Physiol. 262 (1992) E658–E662. [8] H. Gershman, E. Powers, L. Levine, H. Van Vunakis, Radioimmunoassay of prostaglandins, angiotensin, digoxin, morphine and adenosine-3⬘,5⬘-cyclic-monophosphate with nitrocellulose membranes, Prostaglandins 1 (1972) 407–423. [9] R.T. Carson, D.A. Vignali, Simultaneous quantitation of 15 cytokines using a multiplexed Xow cytometric assay, J. Immunol. Methods 227 (1999) 41–52. [10] K.L. Kellar, J.P. Douglass, Multiplexed microsphere-based Xow cytometric immunoassays for human cytokines, J. Immunol. Methods 279 (2003) 277–285. [11] C.M. Whitehouse, R.N. Dreyer, M. Yamashita, J.B. Fenn, Electrospray interface for liquid chromatographs and mass spectrometers, Anal. Chem. 57 (1985) 675–679. [12] P. Campins-Falco, R. Herraez-Hernandez, A. Sevillano-Cabeza, Column-switching techniques for high-performance liquid chromatography of drugs in biological samples, J. Chromatogr. 619 (1993) 177–190. [13] T. Boyle, V. Lancaster, R. Hunt, P. Gemski, M. Jett, Method for simultaneous isolation and quantitation of platelet activating factor and multiple arachidonate metabolites from small samples: analysis of eVects of Staphylococcus aureus enterotoxin B in mice, Anal. Biochem. 216 (1994) 373–382.
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143 [14] A. Margalit, K.L. DuYn, P.C. Isakson, Rapid quantitation of a large scope of eicosanoids in two models of inXammation: development of an electrospray and tandem mass spectrometry method and application to biological studies, Anal. Biochem. 235 (1996) 73–81. [15] W.S. Powell, Precolumn extraction and reversed-phase high-pressure liquid chromatography of prostaglandins and leukotrienes, Anal. Biochem. 164 (1987) 117–131. [16] K.A. Harrison, K.L. Clay, R.C. Murphy, Negative ion electrospray and tandem mass spectrometric analysis of platelet activat-
143
ing factor (PAF) (1-hexadecyl-2-acetyl-glycerophosphocholine), J. Mass Spectrom. 34 (1999) 330–335. [17] M. Caivano, P. Cohen, Role of mitogen-activated protein kinase cascades in mediating lipopolysaccharide-stimulated induction of cyclooxygenase-2 and IL-1 beta in RAW264 macrophages, J. Immunol. 164 (2000) 3018–3025. [18] Y. Kita, T. Takahashi, N. Uozumi, L. Nallan, M.H. Gelb, T. Shimizu, Pathway-oriented proWling of lipid mediators in macrophages, Biochem. Biophys. Res. Commun. 330 (2005) 898– 906.
A multiplex quantitation method for eicosanoids and platelet-activating factor using column-switching reversed-phase liquid chromatography–tandem mass spectrometry 夽 Yoshihiro Kita, Toshie Takahashi, Naonori Uozumi, Takao Shimizu ¤ Department of Biochemistry and Molecular Biology, Faculty of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan Received 4 March 2005 Available online 20 April 2005
Abstract Eicosanoids and platelet-activating factor (PAF) are phospholipid-derived lipid mediators produced by various tissues and cells through a cascade pathway. For a comprehensive analysis of these lipid mediators, a simultaneous quantitation method with sensitivity and reliability is necessary. This article details a development of column-switching reversed-phase liquid chromatography–tandem mass spectrometry for multiplex quantitation of eicosanoids and PAF. The adsorptive nature of lipids caused signiWcant loss of signal in a conventional column-switching conWguration. The use of an online-dilution method allowed use of 100% methanol as a sample solvent, which prevented sample adsorption to contacting surfaces. Addition of 0.2% formic acid to the sample solvent was required for the successful introduction of LTC4 to the trapping column and minimizing its carryover. The optimized method provided rapid analysis of 14 lipid mediators with a throughput of 96 samples/24 h, lower limits of quantitation of 5 pg on column, and linear calibration ranges up to 2000–5000 pg. The system was highly compatible with solid-phase-extracted samples, as methanoleluted fractions were directly injected without reconstitution. The analysis of lipid mediator production of macrophage-like RAW264.7 cells demonstrated that the cell-based assay can be performed in a 96-well format, suitable for metabolomics analyses and/or screening strategies. 2005 Elsevier Inc. All rights reserved. Keywords: Lipid mediators; Quantitation; Column-switching; LC-MS/MS
Eicosanoids, such as prostaglandins (PGs),1 leukotrienes (LTs), and hydroxyeicosatetraenoic acids (HETEs), and platelet-activating factor (PAF) are bioactive lipid 夽 This work was supported in part by Grants-in-Aid from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. * Corresponding author. Fax: +81 3 3817 8732. E-mail address: [email protected] (T. Shimizu). 1 Abbreviations used: PG, prostaglandin; LT, leukotriene; HETE, hydroxyeicosatetraenoic acid; PAF, platelet-activating factor; COX, cyclooxygenase; LO, lipoxygenase; ELISA, enzyme-linked immunosorbent assay; RIA, radioimmunoassay; ESI, electrospray ionization; MeOH, methanol; EDTA, ethylenediaminetetraacetic acid; MeCN, acetonitrile; DMSO, dimethyl sulfoxide; HBSS, Hanks’ balanced salt solution; LPS, lipopolysaccharide; FBS, fetal bovine serum; IS, internal standard; SRM, selected reaction monitoring; CID, collision-induced dissociation; PEEK, poly-ether-ether-ketone.
0003-2697/$ - see front matter 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2005.03.048
mediators produced by various tissues and cells in vivo, playing vital roles in physiological and pathophysiological conditions [1,2]. These mediators are produced through a cascade pathway (Fig. 1), in which several key enzymes such as phospholipase A2, cyclooxygenases (COXs), and lipoxygenases (LOs) function as common regulators in combination with speciWc terminal synthases that produce speciWc lipid mediator molecular species [3–5]. Various biological stimuli that elicit intracellular signals can activate or modulate this cascade pathway and cause lipid mediator production, the proWle of which may diVer by cell types and/or their activation states. For detailed understanding of biological processes involving lipid mediators, a comprehensive analysis tool for lipid mediators is necessary.
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
135
Fig. 1. Structures and production pathways of lipid mediators. The structures and production pathways of target lipid mediators for quantitation are presented. Names of enzymes involved in the pathway are indicated along the path. Since PGI2 and thromboxane (Tx)A2 are unstable, stable metabolites 6-keto-PGF1 and TxB2 are measured instead. LO, lipoxygenase; LTA4H, LTA4 hydrolase; LTCS, LTC4 synthase; 5-HpETE, 5-hydroperoxyeicosatetraenoic acid.
An enzyme-linked immunosorbent assay (ELISA) [6,7], and a radioimmunoassay (RIA) [8] have been the most widely used immunological techniques for the quantitation of small amounts of lipid mediators. A major limitation of these assays is incompatibility with multiplex analysis; they require much sample when multiple target compounds are quantitated. Although a Xow cytometric immunoassay that use multicolor-labeled beads [9,10] enabled multiplex quantitation, it still requires speciWc antibodies for each target. The antibodies, especially when raised against lipids, sometimes do not provide speciWcity suYcient for the discrimination of structurally similar but biologically diVerent lipid metabolites. Mass spectrometry (MS)-based quantitation techniques such as GC-MS (GC-MS/MS) and LC-MS (LC-MS/MS) have advantages in speciWcity over immunoassays. Furthermore, they are well suited to multiplex analysis, which minimizes the sample amount required.
Recent advances in electrospray ionization (ESI) technology provided an interface between LC and MS, which paved the way for the analysis of bioactive lipids without derivatization [11]. The present report describes development of a multiplex quantitation method that covers eicosanoids and PAF using a column-switching reversed-phase LC-MS/MS technique. The columnswitching method was adopted for the eVective sample introduction to the analytical column [12]. This strategy provides substantial improvements in the sensitivity of the system as compared with previously reported methods that use direct sample injection to analytical column [13,14], as a large volume of sample can be introduced. The present report also describes the necessity of an online-dilution technique for sample introduction in combination with the column-switching system. The use of the column-switching method for the analysis of eicosanoids has been described [15]. The present optimized
136
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
system provides better recovery and sensitivity to perform a cell-based assay in a 96-well culture plate and a throughput to process 96 samples/24 h.
Materials and methods Chemicals HPLC-grade methanol (MeOH), formic acid, ethanol, petroleum ether, and ethylenediaminetetraacetic acid (EDTA) were obtained from Wako (Osaka, Japan). MS-grade acetonitrile (MeCN) was obtained from Kanto (Tokyo, Japan). All eicosanoids and PAF (PAFC16; 1-O-hexadecyl-2-O-acetyl-sn-glycero-3-phosphocholine) reference compounds including deuterium-labeled internal standards were obtained from Cayman (Ann Arbor, MI). Dimethyl sulfoxide (DMSO), RPMI-1640 medium, Hanks’ balanced salt solution (HBSS), and lipopolysaccharide (LPS; from Escherichia coli, Serotype 0111:B4) were obtained from Sigma (St. Louis, MO). Fetal bovine serum (FBS) was obtained from Equitech Bio (Kerrville, TX). A23187 was obtained from Calbiochem (San Diego, CA). Preparation of stock solutions and standards A stock solution of eicosanoids and PAF was prepared as a mixture that contains each compound at a concentration of 1 g/ml in MeOH and stored in a glass vial at ¡80 °C (calibration-master). Deuterium-labeled eicosanoids and PAF (6-keto-PGF1-d4, TxB2-d4, PGF2-d4, PGE2-d4, PGD2-d4, PGB2-d4, LTB4-d4, 5(S)HETE-d8, 12(S)-HETE-d8, 15(S)-HETE-d8, and PAFC16-d4) were mixed at concentrations of 5–50 ng/ml in MeOH, depending on the sensitivity of each compound, and used as an internal standard (IS) solution. The calibration-master solution was further diluted by MeOH:formic acid (100:0.2) to prepare the following standard solutions: 5, 20, 50, 200, 500, 2000, and 5000 pg/ 47.5 l. These solutions were mixed with IS solution at a ratio of 19:1 to prepare IS-containing calibration samples of 5–5000 pg/50 l. Accordingly, these calibration samples contained 5–50 pg/50 l of deuterium-labeled compounds. For the standard addition and recovery experiments, standard samples without IS were prepared independently. Cell culture, stimulation, sample collection, and pretreatment Murine macrophage-like RAW264.7 cells were maintained in RPMI-1640 culture medium supplemented with 10% FBS at 37 °C, 5% CO2 in air. Cells were seeded onto 96-well culture plates at a density of 1 £ 105 cells/ well. Two hours later, cells were treated with or without
100 ng/ml LPS for 12 h (LPS priming). After washing three times with HBSS, cells were stimulated with 2 M A23187 or vehicle (0.1% DMSO) in 100 l HBSS for 30 min. Lipid mediator production was terminated by adding equal amount (100 l) of MeOH:formic acid (100:0.2). After the addition of 10 l IS to each well, entire volumes (»200 l/well) were transferred to 1-ml glass collection vials arranged in a standard 96-well (12 £ 8) format using a multichannel pipette. Then, another 100 l of MeOH:formic acid (100:0.2) was added to the well and mixed for 1 h at room temperature. After the addition of 100 l water:formic acid (100:0.03) to each well, the resultant solutions (»200 l/well) were transferred to collection vials (Wnal volume, »400 l/ sample). The samples were submitted to solid phase extraction using an Oasis HLB 96-well cartridge cluster (5 mg; Waters, Milford, MA). BrieXy, samples were diluted with water:formic acid (100:0.03) to give a Wnal MeOH concentration of »20% by volume, applied to preconditioned cartridges, and washed serially with water:formic acid (100:0.03), water:ethanol:formic acid (90:10:0.03), and petroleum ether. Samples were eluted into glass vials with 200 l of MeOH:formic acid (100:0.2) by centrifugation (100g, 1 min) and stored at 4 °C in the autosampler while waiting for injection. The volumes of eluted fraction were 170–180 l, 50 l of which was injected for analysis. Equipment and conditions Four LC10AD pumps (Shimadzu, Kyoto, Japan), a 3033 autosampler (Shiseido, Tokyo, Japan), and an electrically controlled six-port switching valve (Valco, Houston, TX) were conWgured as shown in Fig. 2. Three diVerent conWgurations were used for the optimization and validation of the method. ConWguration A is a simple HPLC system with autosampler, used for the direct injection of a small volume (5 l) of sample. ConWgurations B and C are column-switching systems with diVerent sample introduction methods. ConWguration B is a conventional column-switching system, where an autosampler is located at the downstream of pumps C and D. ConWguration C is a column-switching system with online-dilution Xow path, where an autosampler is located at the downstream of pump D and samples injected are diluted by mobile phase C before the trapping column. The compositions of mobile phases A, B, C, and D were water, MeCN:formic acid (100:0.1), water:formic acid (100:0.1), and MeOH:formic acid (100:0.2), respectively. For the carryover experiment of LTC4, formic acid was removed from mobile phase D for comparison. Opti-Guard Mini C18 (1 £ 15 mm; Optimize Technologies, Oregon City, OR) and Capcellpak C18 MGS3 (1 £ 100 mm; Shiseido, Tokyo, Japan) were used as a trapping column and a separating column, respectively. In conWgurations B and C, during the
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
137
pressure, 2.0–2.2 mTorr (argon gas). Capillary voltages, tube lens voltages, and collision energies were optimized for each compound. For selected reaction monitoring (SRM) mode operations, Q1 and Q3 resolutions were set to »1.5 Da (by full-width at half-maximum). For the peak detection, integration, and quantitative calculation of MS data, the Xcalibur 1.2 software package (Thermo Electron) was used. Calibration curve Calibration curves were calculated by the IS method using peak area ratios of target compounds against IS. Compounds without corresponding commercially available deuterium-labeled counterparts were calibrated by either PGB2-d4 (for LTC4 and LTD4) or 15-HETE-d8 (for 8-HETE and 11-HETE) as IS. Linear regression with 1/x weighting was used for the calculation of coeYcient of determination (R2). Calibration ranges with R2 > 0.99 were considered ranges with suYcient linearity. Accuracy and precision
Fig. 2. LC-MS/MS system conWgurations. (ConWguration A) An ordinary binary solvent system. (ConWguration B) A conventional columnswitching system. Pumps C and D are used to generate isocratic Xow with mobile phase C:mobile phase D ratios of 80:20, 70:30, and 60:40. (ConWguration C) An online-dilution column-switching system. In this system, an autosampler Xow path is Wlled with mobile phase D. Samples introduced are diluted by mobile phase C in the tee connector before the trapping column. The Xow path between the tee connector and the trapping column is minimized to avoid unwanted sample adsorptions. In all conWgurations, poly-ether-ether-ketone materials were used for sample-contacting surfaces unless stated otherwise.
trapping phase (0–3.0 min), 50 l of the sample injected from autosampler was directly introduced (conWguration B) or introduced after dilution with mobile phase C (conWguration C). Samples were introduced to the trapping column at a total Xow rate of 500 l/min. Then, the valve was switched to introduce sample to the separating column (3.0 min). Separation of samples was conducted by a step gradient with mobile phase A:mobile phase B ratios of 64:37 (0–6.0 min), 55:45 (6.0–7.0 min), and 35:65 (7.0 min–next injection; typically 14.5 min) at a Xow rate of 120 l/min and a column temperature of 46 °C. MS data acquisition was started at 3.0 min, which corresponds to 0 min in the chromatograms. A TSQ-7000 triple-stage quadrupole MS instrument equipped with an ESI interface (Thermo Electron, Waltham, MA) was used. Output of LC was directly introduced to ESI without splitting. Source conditions were as follows: negative ESI mode; spray voltage, 3000 V; heated-capillary temperature, 270 °C; sheathe gas (N2) pressure, 80 psi (auxiliary gas not used); collision cell
Accuracy and precision for calibration samples was determined by recalculation of each calibration level to calibration curve generated. Recovery rates calculated as ([calculated amount]/[nominal amount]) £ 100 and coeYcient of variation (CV) for multiple injections (triplicate) were used for the evaluation of accuracy and precision, respectively. Accuracy and precision of biological samples were determined by standard addition-and-recovery test. RAW264.7 cell samples with four diVerent treatments (with regard to LPS priming and A23187 stimulation; see Cell culture, stimulation, sample collection, and pretreatment) were cleaned without adding IS and pooled. These four samples that contained diVerent levels of eicosanoids and PAF were used as biological samples to which standards of known amounts were added. For each sample, 0.1 volume of 10£ concentrated standards of IS-only, IS + 50 pg/200 l, and IS + 500 pg/200 l were added and quantitated (triplicate determinations for each level) using the calibration curve made by standards without biological matrices. Recovery rates calculated as ([calculated amount for IS + 50 (or 500) pg/ 200 l addition] ¡ [calculated amount for IS-only addition])/[nominal amount, e.g., 50 (or 500) pg/200 l] £ 100 and coeYcient of variation calculated for the four diVerent biological samples were used for the determination of accuracy and precision, respectively. Validation of sample pretreatment procedure Consistency of sample pretreatment procedure was validated by the recovery of standard samples. Standard mixtures (500 pg/well in 50 l MeOH, without IS) were
138
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
loaded onto wells of a 96-well culture plate that contained 100 l/well of HBSS, and subsequently 10 l of IS was added to each well. After performing cleanup procedure as described above, samples were quantitated. The net recovery rates of target compounds during solidphase extraction were 80% or more for all target compounds (data not shown).
Results and discussion Detection of eicosanoids and PAF Conditions for the detection of eicosanoids and PAF by MS in SRM mode are summarized in Table 1. All eicosanoids were ionized as deprotonated ionic species ([M ¡ H]¡) in the negative ESI mode, when water, MeCN, and small amounts (0.03–0.08%) of formic acid were used as a spraying solvent. PAF was ionized as a formic acid adduct ([M + HCOOH-H]¡) in this solvent. These molecular-related ions (shown as Q1 values in Table 1) were used as precursor ions for collisioninduced dissociation (CID). As typical fragmentation patterns, CID spectra of PGE2 and PAF are shown in Fig. 3. Dehydrated fragment ions such as m/z 333 and 315 shown in Fig. 3A were avoided for SRM due to the lack of speciWcity. For the detection of PAF, the acetyl fragment (m/z 59) was used for SRM for the structural Table 1 SRM conditions Compounds
Q1a (m/z)
Q3b (m/z)
Collision energy (eV)
6-keto-PGF1 TxB2 PGF2 PGE2 PGD2 LTB4 LTC4 LTD4 5-HETE 8-HETE 11-HETE 12-HETE 15-HETE PAF-C16
369 369 353 351 351 335 624 495 319 319 319 319 319 568
245 195 193 271 271 195 272 177 115 155 167 179 219 59
32 18 25 18 18 20 26 20 20 18 20 16 16 30
6-keto-PGF1-d4 TxB2-d4 PGF2-d4 PGE2-d4 PGD2-d4 PGB2-d4 LTB4-d4 5-HETE-d8 12-HETE-d8 15-HETE-d8 PAF-C16-d4
373 373 357 355 355 337 339 327 327 327 572
249 199 197 275 275 239 197 116 184 226 59
32 18 25 18 18 22 20 20 16 16 30
a b
First quadrupole. Third quadrupole.
Fig. 3. CID spectra of PGE2 and PAF-C16. (A) CID spectrum of PGE2. A fragment ion of m/z 271 corresponding to [M-2H2O-CO2-H]¡ ionic species is selected for SRM transition. (B) CID spectrum of PAF. An acetyl fragment ion, m/z 59, is used for SRM transition. The intensity of molecular-related ion, m/z 568.3, is multiplied (£300) for visualization.
discrimination from lysophosphatidylcholine (Fig. 3B and [16]). Similarly, fragments characteristic for each compound were selected for SRM (Table 1). Results of three diVerent isocratic separations in reversed-phase LC are shown in Fig. 4. Retentions of PGs were weaker than those of LTs, HETEs, and PAF in the C18 column, as it required low MeCN concentration of 37% (Fig. 4A). In this condition, PGE2 and PGD2 were successfully resolved, and this allowed using the same SRM transition (353 ! 271) for the detection of these PGs. Among HETE isomers, 8-HETE and 12-HETE were not resolved in the C18 column (Fig. 4C). Although CID spectra of these HETE isomers suggested a suYcient selectivity of SRM transitions selected (data not shown), potential cross talk was evaluated in the more sensitive SRM mode. When commercially available 8-HETE and 12-HETE were used, cross talk was approximately 0.1% for 8-HETE to 12-HETE SRM and 0.3–0.4% for 12-HETE to 8-HETE SRM (Fig. 5). These values were constant for the injections of 500–5000 pg on column, although cross talk signals by the injection of 500 pg of 8-HETE on column were under the detection limit (data not shown). It was diYcult to trace the origin of this cross talk, since it can be caused either by a minor but true fragmentation or by impurities in the standard compounds. The cross talk was negligible for 12-HETE quantitation, since an injection of 5000 pg of 8-HETE (upper limit of quantitation) on column may cause a signal that is equivalent to »3 pg of 12HETE, which is below the lower limit of quantitation. However, an injection of 5000 pg of 12-HETE on column may cause a signal that is equivalent to 20–30 pg of 8-HETE. Therefore, compensations are required for the quantitation of 8-HETE, especially when samples contain large amounts (near the upper limit of quantitation) of 12-HETE.
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
139
Fig. 5. Cross talk of 8-HETE and 12-HETE SRM. Simultaneous SRM of 8-HETE (319 ! 155) and 12-HETE (319 ! 179) were performed after injecting 5000 pg of 8-HETE and 12-HETE on column. Isocratic separation with mobile phase A:mobile phase B ratio of 35:65 was used. Values on arrows indicate cross talk ratios.
Fig. 6. Simultaneous analysis of 14 lipid mediators by step gradient separation. A typical chromatogram after injecting 50 pg on column. Separation was achieved within 10 min. Fig. 4. Isocratic separation of lipid mediators. (A) Isocratic separation of PGs and LTD4 with mobile phase A:mobile phase B ratio of 63:37. (B) Isocratic separation of LTs with mobile phase A:mobile phase B ratio of 55:45. (C) Isocratic separation of HETEs and PAF-C16 with mobile phase A:mobile phase B ratio of 35:65.
ConWguration of column-switching system Isocratic conditions as shown in Fig. 4 were combined to achieve three-step gradient separation. Fig. 6 shows a representative chromatogram. In this separation mode, 14 lipid mediators were analyzed within 10 min, with less than 15 min cycle time which includes 3 min of sample trapping procedure (throughput of 96 samples/24 h). When a conventional column-switching conWguration (Fig. 2, conWguration B) in which samples were introduced isocratically with 20, 30, or 40% MeOH sample solvent was used, any conditions tested satisWed the sensitivity for simultaneous analysis of PGs, LTs, HETEs, and
PAF (Fig. 7A). In this conWguration, increase in the MeOH concentration from 20 to 30 or 40% caused decrease in the sensitivities of PGs, whereas it caused signiWcant improvements in the sensitivities of LTs, HETEs, and PAF. We hypothesized that the decrease in the sensitivities for PGs in the higher concentration of MeOH may result from the loss of retention by the trapping column, and the poor sensitivities for LTs, HETEs, and PAF in the lower concentration of MeOH may result from the sample adsorption to contacting surfaces before the trapping column. In addition, a signiWcant decrease of signals was found in water-rich sample solvent during storage in glass autosampler vials within 24 h (data not shown), suggesting that the adsorptive nature of lipids may disturb the consistency of quantitation. To circumvent the problem, online dilution of samples dissolved in MeOH:formic acid (100:0.2) was examined (Fig. 2, conWguration C). In this conWguration,
140
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
Optimization for LTC4 analysis As shown in Figs. 4B and 6, LTC4 showed signiWcant peak tailing. It was aVected by formic acid concentrations in the mobile phase; an increase in formic acid concentration of mobile phase B ameliorated the peak tailing of LTC4 (data not shown). The concentration of formic acid in mobile phase B (MeCN:formic acid, 100:0.1) was determined by the peak shape of LTC4 and sensitivities of other compounds. Sensitivity and carryover for LTC4 were analyzed. As shown in Fig. 8, addition of 0.2% formic acid to both sample solvent (MeOH) and mobile phase D was required to achieve sensitivity with acceptable carryover values. When no formic acid was used, up to 40% decrease in the signal intensity was observed (Fig. 8A). This seemed not to be due to the loss of retention in the trapping column; rather it was due to sample adsorption to the Xow path as carryover values signiWcantly increased (Fig. 8B). Interestingly, the addition of formic acid only to mobile phase D improved LTC4 signal, but unacceptable carryover (»3%) occurred. In addition, when stainless steel tubing instead of PEEK tubing was used for the introduction of samples without formic acid, larger reduction of signals and longlasting carryover were observed (data not shown). This result suggests that stainless steel tubing including autosampler sampling needles are less compatible with LTC4 analysis. With regard to target compounds other than LTC4, the eVects of formic acid were negligible, although 0.1% formic acid in mobile phase C was required to
Fig. 7. Optimization of trapping condition in two diVerent columnswitching systems. (A) EVects of MeOH concentrations on sensitivities in conWguration B. MeOH concentrations of sample solvent were adjusted to those of mobile phase. (B) EVects of MeOH concentrations on sensitivities in conWguration C. Sample solvent composition was Wxed to 100% MeOH and dilution ratios were changed to achieve Wnal MeOH concentrations indicated. Data are normalized for each compound and are expressed as mean § SD; the average values for 20% MeOH in conWguration C are set to 100.
samples were diluted by water:formic acid (100:0.1) using a tee connector located adjacent to the trapping column. As shown in Fig. 7B, trapping at MeOH concentration of 20% (Wvefold dilution) provided the best sensitivity for all 14 compounds. Increase in the MeOH concentration to 30 or 40% did not improve sensitivity; rather it caused signiWcant decrease in the sensitivities of PGs and PAF. Trapping eYciencies of PGs and LTD4 as determined by comparison with the direct injection method (Fig. 2, conWguration A) were nearly 100% (Table 2), suggesting that further decrease in MeOH concentration is not necessary.
Fig. 8. EVects of formic acid on sensitivity and carryover of LTC4. Signal intensities upon injection of 5000 pg LTC4 (A) and carryover after injection of 5000 pg LTC4 (B) are shown. The eVects of 0.2% formic acid addition on the sample solvent and/or mobile phase D are compared. Data are expressed as mean § SD.
Table 2 Trapping eYciencies for prostaglandins and LTD4a
ConWguration A ConWguration C EYciencyb (%) a
6-keto-PGF1
TxB2
Area
CV (%)
Area
CV (%)
Area
CV (%)
Area
CV (%)
Area
CV (%)
Area
CV (%)
29193 29545 101.2
4.5 5.1
141694 147346 104.0
5.0 5.1
25966 26787 103.2
6.8 5.2
148293 148276 100.0
2.9 3.3
127354 126723 99.5
2.6 2.9
46049 46017 99.9
4.8 2.5
PGF2
PGE2
LTD4
Samples prepared as 50 pg/5 l in MeOH:water (80:20) were injected by a volume of 5 l and analyzed by isocratic condition (37% mobile phase
B). b
PGD2
DeWned as [Area for ConWguration C]/[Area for ConWguration A] £ 100.
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
141
Using the optimized conditions as described above, the sensitivity and linearity were examined. As shown in Table 3, lower limits of quantitation of 5 pg were achieved for all 14 target compounds examined and linear calibration ranges up to 5000 pg, excepting PGD2 whose linear calibration range was 5–2000 pg, were achieved with R2 > 0.99 in the linear regression with 1/x weighting.
RAW264.7 was selected for this purpose. The cells cultured in a 96-well culture plate at a density of 1 £ 105 cells/well were stimulated with calcium ionophore A23187, and lipid mediator productions within 30 min were measured. To compare diVerent cell states, the cells pretreated with LPS which reportedly induce COX-2 expression [17] were also analyzed (Fig. 9). When not primed with LPS, the cells produced LTC4 as a major product, which was readily stimulated by A23187. LPS priming shifted the cells to produce huge amounts of PGD2. PAF accumulation was also found after LPS pretreatment.
Solid-phase extraction of lipid mediators
Accuracy and precision in biological samples
Sample pretreatment procedure for cell-based assay in a 96-well format was designed and validated. As the online-dilution system allowed the use of 100% MeOH as sample solvent, MeOH:formic acid (100:0.2)-eluted fractions were directly analyzed without drying and reconstitution. As shown in Table 4, standards added to 96-well polystyrene culture plate were consistently recovered.
Using the RAW264.7 cell samples, accuracy and precision of the system were examined. Accuracy determined by the standard addition of 50 pg/200 l and 500 pg/200 l (mimicking the addition of 50 pg/well and 500 pg/well in the 96-well plate) to pooled, pretreated samples that contain diVerent levels of intrinsic eicosanoids and PAF and precision determined by repetitive injections are shown in Table 5. Among the lipid mediators, LTC4 showed signiWcantly biased recovery value of »30% when calibration curve was made with standard samples without the biological matrix. The most probable explanation for the positive bias in biological samples may be the matrix eVects that prevent LTC4 adsorption. Since
achieve suYcient trapping when mobile phase D did not contain formic acid (data not shown). Sensitivity and linearity
Simultaneous quantitation of lipid mediators in RAW264.7 cells The system was examined for use in cell-based biological assays. A murine macrophage-like cell line Table 3 Linearity, accuracy, and precision for reference compoundsa Compounds
Linearity (R2) Calibration Levels 5 pg
20 pg
50 pg
200 pg
500 pg
2000 pg
5000 pg
Recovery CV Recovery CV Recovery CV Recovery CV Recovery CV Recovery CV Recovery CV (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) 6-keto-PGF1 TxB2 PGF2 PGE2 PGD2b LTB4 LTC4 LTD4 5-HETE 8-HETE 11-HETE 12-HETE 15-HETE PAF-C16 a b
0.9989 0.9991 0.9987 0.9993 0.9971 0.9998 0.9991 0.9960 0.9981 0.9993 0.9978 0.9975 0.9993 0.9995
100.8 100.6 105.8 97.5 95.3 97.4 100.7 96.8 98.9 99.2 108.1 95.4 114.5 82.1
5.0 12.0 11.9 9.5 13.1 8.5 6.8 7.3 10.8 15.1 7.0 13.0 9.6 12.1
96.6 92.6 99.4 102.0 103.6 100.4 111.3 112.0 117.2 93.4 90.7 104.7 91.2 97.0
5.9 7.2 10.7 6.4 5.6 4.2 9.3 3.2 17.6 11.9 2.2 12.7 4.1 5.1
104.8 99.0 108.0 99.3 108.9 100.7 108.4 122.8 106.7 95.3 92.6 107.6 96.4 104.0
7.2 2.7 2.7 2.1 2.1 0.4 6.0 0.9 10.4 4.0 2.3 2.4 9.1 4.0
99.7 95.9 101.9 103.9 108.5 96.5 104.9 119.2 104.4 97.1 91.3 103.9 97.8 98.1
0.8 0.8 0.9 1.5 2.5 0.6 1.7 1.6 4.2 6.1 4.9 2.1 4.2 3.6
102.2 95.0 103.6 104.4 106.7 101.7 106.6 117.5 112.0 99.4 94.4 111.6 99.8 102.4
3.2 101.4 2.2 97.3 1.9 101.0 0.8 99.0 1.8 97.2 1.3 97.9 2.5 99.8 0.4 99.7 1.7 99.2 1.3 99.1 1.6 96.0 4.2 102.3 1.1 98.2 2.5 98.3
1.3 4.0 5.6 1.4 3.7 0.7 4.3 2.5 1.4 2.2 3.3 3.1 3.1 3.4
99.2 101.8 99.1 99.8 (80.4) 100.8 99.1 97.4 98.8 100.6 102.6 97.7 100.9 100.5
4.5 0.6 3.2 3.2 (1.6) 0.6 0.4 0.2 3.0 3.0 3.6 4.3 2.3 1.1
Triplicate determinations were performed for each calibration level. Linear regression of PGD2 was applied within the range of 5–2000 pg.
Table 4 Validation of solid-phase extractiona 6-keto-PGF1 TxB2 PGF2 PGE2 PGD2 LTB4 LTC4 LTD4 5-HETE 8-HETE 11-HETE 12-HETE 15-HETE PAF-C16 Recovery (%) 105.5 CV (%) 6.8 a
106.7 104.9 5.8 8.5
104.6 110.0 3.0 9.2
106.4 105.9 98.2 14.3 17.2 6.8
103.5 2.4
112.9 17.5
96.9 21.8
105.1 6.1
88.4 17.8
96.0 5.2
Standard mixtures placed on the 96-well plate (500 pg/well, quadruplicate) were added with IS and quantitated after solid-phase extraction.
142
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143
metabolomics analysis and/or screening strategy can be performed. The present strategy can be extended for the quantitation of other eicosanoids and their metabolites, as the system is applicable to lipid metabolites with diverse hydrophobicity. The system can also be applied to other biological samples such as tissue extracts, although the sample pretreatment method should be properly developed and validated. The present method will be an essential tool for the comprehensive analysis of lipid mediators.
Note Fig. 9. Lipid mediator production of RAW264.7 cells. RAW264.7 cells seeded onto 96-well plates at a density of 1 £ 105 cells/well were treated as described under Materials and methods and lipid mediators were measured simultaneously. Data are expressed as mean § SD.
A related manuscript by Kita et al. that describes an application of the present method to the comprenhensive analysis of macrophages can be found in Ref. [18].
Table 5 Accuracy and precision for biological samplesa
References
Compounds
6-keto-PGF1 TxB2 PGF2 PGE2 PGD2 LTB4 LTC4 LTD4 5-HETE 8-HETE 11-HETE 12-HETE 15-HETE PAF-C16
+50 pg/200 l
+500 pg/200 l
Recovery (%)
CV (%)
Recovery (%)
CV (%)
97.8 98.3 94.1 97.2 108.0 99.6 127.6 111.3 89.7 97.5 100.4 100.8 89.4 98.6
3.9 3.0 8.5 7.0 14.9 5.8 8.6 26.6 3.5 11.1 11.2 9.7 8.1 7.1
101.0 101.7 97.6 99.1 93.3 98.4 133.6 106.4 89.2 99.9 100.9 103.6 88.8 106.3
4.7 6.7 4.3 4.8 5.7 5.2 11.2 16.4 2.9 9.7 9.4 4.7 10.1 5.2
a
Data are shown as averages of samples containing diVerent levels of intrinsic compounds (see Materials and methods and Fig. 9).
deuterium-labeled PGB2 is used as IS for LTC4, the eVect cannot be compensated by IS. However, in the cell-based assay where sample matrices are relatively uniform, assay consistency may not be disturbed. Actually, the conserved precision of LTC4 as shown in Table 5 suggests the uniformity of the bias within the samples that contain diVerent levels of LTC4 (less than 10 pg/50 l to more than 100 pg/50 l; see also Fig. 9).
Conclusions A multiplex quantitation method for 14 lipid mediators that covers eicosanoids and PAF was developed. They are primary enzymic products from the common phospholipid precursors. The system was successfully applied to the cell-based analysis of a macrophage-like cell line in a 96-well plate format, demonstrating that
[1] C.D. Funk, Prostaglandins and leukotrienes: advances in eicosanoid biology, Science 294 (2001) 1871–1875. [2] S.M. Prescott, G.A. Zimmerman, D.M. StaVorini, T.M. McIntyre, Platelet-activating factor and related lipid mediators, Annu. Rev. Biochem. 69 (2000) 419–445. [3] D.A. Six, E.A. Dennis, The expanding superfamily of phospholipase A2 enzymes: classiWcation and characterization, Biochim. Biophys. Acta 1488 (2000) 1–19. [4] D.L. Simmons, R.M. Botting, T. Hla, Cyclooxygenase isozymes: the biology of prostaglandin synthesis and inhibition, Pharmacol. Rev. 56 (2004) 387–437. [5] M. Peters-Golden, T.G. Brock, 5-lipoxygenase and FLAP, Prostaglandins Leukot. Essent. Fatty Acids 69 (2003) 99–109. [6] F. Shono, K. Yokota, K. Horie, S. Yamamoto, K. Yamashita, K. Watanabe, H. Miyazaki, A heterologous enzyme immunoassay of prostaglandin E2 using a stable enzyme-labeled hapten mimic, Anal. Biochem. 168 (1988) 284–291. [7] M. Reinke, Monitoring thromboxane in body Xuids: a speciWc ELISA for 11-dehydrothromboxane B2 using a monoclonal antibody, Am. J. Physiol. 262 (1992) E658–E662. [8] H. Gershman, E. Powers, L. Levine, H. Van Vunakis, Radioimmunoassay of prostaglandins, angiotensin, digoxin, morphine and adenosine-3⬘,5⬘-cyclic-monophosphate with nitrocellulose membranes, Prostaglandins 1 (1972) 407–423. [9] R.T. Carson, D.A. Vignali, Simultaneous quantitation of 15 cytokines using a multiplexed Xow cytometric assay, J. Immunol. Methods 227 (1999) 41–52. [10] K.L. Kellar, J.P. Douglass, Multiplexed microsphere-based Xow cytometric immunoassays for human cytokines, J. Immunol. Methods 279 (2003) 277–285. [11] C.M. Whitehouse, R.N. Dreyer, M. Yamashita, J.B. Fenn, Electrospray interface for liquid chromatographs and mass spectrometers, Anal. Chem. 57 (1985) 675–679. [12] P. Campins-Falco, R. Herraez-Hernandez, A. Sevillano-Cabeza, Column-switching techniques for high-performance liquid chromatography of drugs in biological samples, J. Chromatogr. 619 (1993) 177–190. [13] T. Boyle, V. Lancaster, R. Hunt, P. Gemski, M. Jett, Method for simultaneous isolation and quantitation of platelet activating factor and multiple arachidonate metabolites from small samples: analysis of eVects of Staphylococcus aureus enterotoxin B in mice, Anal. Biochem. 216 (1994) 373–382.
Multiplex quantitation of lipid mediators by LC-MS/MS / Y. Kita et al. / Anal. Biochem. 342 (2005) 134–143 [14] A. Margalit, K.L. DuYn, P.C. Isakson, Rapid quantitation of a large scope of eicosanoids in two models of inXammation: development of an electrospray and tandem mass spectrometry method and application to biological studies, Anal. Biochem. 235 (1996) 73–81. [15] W.S. Powell, Precolumn extraction and reversed-phase high-pressure liquid chromatography of prostaglandins and leukotrienes, Anal. Biochem. 164 (1987) 117–131. [16] K.A. Harrison, K.L. Clay, R.C. Murphy, Negative ion electrospray and tandem mass spectrometric analysis of platelet activat-
143
ing factor (PAF) (1-hexadecyl-2-acetyl-glycerophosphocholine), J. Mass Spectrom. 34 (1999) 330–335. [17] M. Caivano, P. Cohen, Role of mitogen-activated protein kinase cascades in mediating lipopolysaccharide-stimulated induction of cyclooxygenase-2 and IL-1 beta in RAW264 macrophages, J. Immunol. 164 (2000) 3018–3025. [18] Y. Kita, T. Takahashi, N. Uozumi, L. Nallan, M.H. Gelb, T. Shimizu, Pathway-oriented proWling of lipid mediators in macrophages, Biochem. Biophys. Res. Commun. 330 (2005) 898– 906.