Targeted lipidomics using electron capture atmospheric pressure chemical ionization mass spectrometry


  • Seon Hwa Lee,

    1. Center for Cancer Pharmacology, University of Pennsylvania School of Medicine, 1254 BRB II/III, 421 Curie Boulevard Philadelphia, PA 19104-6160, USA
    Search for more papers by this author
  • Michelle V. Williams,

    1. Center for Cancer Pharmacology, University of Pennsylvania School of Medicine, 1254 BRB II/III, 421 Curie Boulevard Philadelphia, PA 19104-6160, USA
    Search for more papers by this author
  • Raymond N. DuBois,

    1. Division of Gastroenterology, Department of Medicine, Vanderbilt University Medical Center, 1161 21st Ave. South, Nashville, TN 37232-2279, USA
    Search for more papers by this author
  • Ian A. Blair

    Corresponding author
    1. Center for Cancer Pharmacology, University of Pennsylvania School of Medicine, 1254 BRB II/III, 421 Curie Boulevard Philadelphia, PA 19104-6160, USA
    • Center for Cancer Pharmacology, University of Pennsylvania School of Medicine, 1254 BRB II/III, 421 Curie Boulevard, Philadelphia, PA 19104-6160, USA.
    Search for more papers by this author


There is an increasing need to be able to conduct quantitative lipidomics analyses as a complement to proteomics studies. The highest specificity for proteomics analysis can be obtained using methodology based on electrospray ionization (ESI) or atmospheric pressure chemical ionization (APCI) coupled with liquid chromatography/tandem mass spectrometry (LC/MS/MS). For lipidomics analysis it is often necessary to be able to separate enantiomers and regioisomers. This can be very challenging when using methodology based on conventional reversed-phase chromatography. Normal-phase chromatography using chiral columns can provide dramatic improvements in the resolution of enantiomers and regioisomers. However, conventional ESI- and APCI-MS/MS has limited sensitivity, which makes it difficult to conduct studies in cell culture systems where only trace amounts of non-esterified bioactive lipids are present. The use of electron capture APCI-MS/MS overcomes this problem. Enantiomers and regioisomers of diverse bioactive lipids can be quantified using stable isotope dilution methodology coupled with normal-phase chiral chromatography and electron capture APCI-MS/MS. This methodology has allowed a lipidomics profile from rat epithelial cells maintained in culture to be delineated and allowed the effect of a non-selective lipoxygenase inhibitor to be assessed. Copyright © 2003 John Wiley & Sons, Ltd.

The concept of profiling protein production (proteomics) in cells, tissues, and biological fluids has become widely accepted as a means to help unravel the underlying biochemical mechanisms involved in the pathogenesis of disease1 and for the identification of novel drug targets.2 Liquid chromatography/electrospray ionization/tandem mass spectrometry (LC/ESI-MS/MS) has become an essential component of proteomics research.3,4 As the field of proteomics has become more mature, quantitative LC/MS-based proteomics methodology has assumed an increasing importance.5 Further advances in understanding pathophysiological changes that occur in cells and tissues requires complementary profiling of lipid-derived metabolic products. This has led to the development of MS-based lipidomics methodology.6,7 However, this methodology is not suitable for the separation of complex mixtures of isomeric non-esterified lipids that are present in trace amounts (pmol/mL) in biological fluids.

The ability to resolve enantiomeric, regioisomeric, and stereoisomeric bioactive lipids is particularly important for cyclooxygenase (COX)-, lipoxygenase (LOX)-derived bioactive lipids as well as those arising non-enzymatically from reactive oxygen species (ROS) (Scheme 1). For example, 13(S)-hydroperoxy-(Z,E)-9,11-octadecadienoic acid [13(S)-(9Z,11E)-HPODE] is an important metabolite of 15-LOX-18 and COX-2,9 whereas its racemic stereoisomer, 13(R,S)-(9E,11E)-HPODE, is formed non-enzymatically during lipid peroxidation10 and its regioisomer, 9(R)-(10E,12Z)-HPODE, is formed by COX-2.9 From a mechanistic perspective it is important to be able to distinguish these isomeric compounds with high specificity and sensitivity. Typically, resolution of such molecules has been conducted using normal-phase chiral chromatography.11 Unfortunately, the sensitivity of MS/MS analysis using ESI methodology and normal-phase solvents is rather poor. Sensitivity can be enhanced by the use of atmospheric pressure chemical ionization mass spectrometry (APCI-MS).12 This methodology has quite limited applicability for analyzing non-esterified bioactive lipids, which are normally present in relatively low concentrations in biological fluids.

Scheme 1.

Formation of bioactive lipids.

The corona discharge used to generate ions under conventional APCI conditions also provides a rich source of gas-phase electrons. The pioneering work of Horning et al. suggested that this occurs by displacement of electrons from the nitrogen sheath gas.13 We realized that, under APCI conditions, suitable analytes would undergo electron capture in the gas phase in a similar manner to that observed for electron capture negative chemical ionization (ECNCI) in gas chromatography (GC)/MS studies.14 This technique (electron capture APCI-MS) provides an increase in sensitivity of two orders of magnitude when compared with conventional negative ion APCI methodology of underivatized analytes.15 It is relatively simple to derivatize bioactive lipids with an electron-capturing group such as the pentafluorobenzyl (PFB) moiety before LC analysis. Furthermore, the PFB derivative has been used previously in many ultrasensitive GC/ECNCI-MS studies (with methane or ammonia used as the reagent gas) because it undergoes efficient dissociative electron capture in the gas phase. This results in the formation of negative ions through dissociative electron capture and subsequent loss of a PFB radical. An identical process is observed for lipid-PFB derivatives under LC/electron capture APCI-MS conditions when nitrogen is used as the sheath gas (Fig. 1). Paradoxically, loss of the PFB moiety results in a tremendous increase in sensitivity when compared with the intact derivative detected in the positive APCI mode.15 More importantly, electron capture APCI provides similar or even enhanced sensitivity when normal-phase solvents are used instead of reversed-phase solvents. In contrast, the PFB derivatives give relatively weak signals under conventional positive APCI conditions when normal-phase solvents are used. Therefore, the electron capture methodology provides an excellent means to analyze lipids when chromatography is performed on chiral columns using normal-phase solvents. We report the use of stable isotope dilution chiral LC/electron capture APCI-MS/MS for quantitative targeted lipidomics analysis of fatty acids, hydroxy fatty acids, prostaglandins (PGs) and isoPGs (iPs) and the application of this methodology for mechanistic studies in rat intestinal epithelial (RIE) cells.

Figure 1.

Mechanism for electron capture APCI analysis of 13(S)-HODE-PFB ester.


Linoleic acid (LA), arachidonic acid (AA), 13(R)-hydroxy-9Z,11E-octadecadienoic acid [13(R)-HODE], 13(S)-hydroxy-9Z,11E-octadecadienoic acid [13(S)-HODE], 9(R)-hydroxy-10E,12Z-octadecadienoic acid [9(R)-HODE], 9(S)-hydroxy-10E,12Z-octadecadienoic acid [9(S)-HODE], 11(R)-hydroxy-5Z,8Z,12E,14Z-eicosatetraenoic acid [11(R)-HETE], 11(S)-hydroxy-5Z,8Z,12E,14Z-eicosatetraenoic acid [11(S)-HETE], 12(R)-hydroxy-5Z,8Z,10E,14Z-eicosatetraenoic acid [12(R)-HETE], 12(S)-hydroxy-5Z,8Z,10E,14Z-eicosatetraenoic acid [12(S)-HETE], 15(R)-hydroxy-5Z,8Z,11Z,13E-eicosatetraenoic acid [15(R)-HETE], 15(S)-hydroxy-5Z,8Z,11Z,13E-eicosatetraenoic acid [15(S)-HETE], 9-oxo-11α,15S-dihydroxyprosta-5Z,13E-dien-1-oic acid (PGE2), 9-oxo-11β,15S-dihydroxyprosta-5Z,13E-dien-1-oic acid (11β-PGE2), 9-oxo-11α,15S-dihydroxy-(8β)-prosta-5Z,13E-dien-1-oic acid (8-iso PGE2), 9α,15S-dihydroxy-11-oxo-prosta-5Z,13E-dien-1-oic acid (PGD2), 9α,11α,15S-trihydroxyprosta-5Z,13E-dien-1-oic acid (PGF2α), 9α,11β,15S-trihydroxyprosta-5Z,13E-dien-1-oic acid (11β-PGF2α), 9α,11α,15S-trihydroxy-(8β)-prosta-5Z,13E-dien-1-oic acid (8-iso PGF2α; iPF2α-III), 9-oxo-11α,15S-dihydroxyprosta-5Z,13E-dien-1-oic-3,3,4,4-2H4 acid ([2H4]-PGE2) and 9α,11α,15S-trihydroxyprosta-5Z,13E-dien-1-oic-3,3,4,4-2H4 acid ([2H4]-PGF2α) were purchased from Cayman Chemical Co. (Ann Arbor, MI, USA). Diisopropylethylamine (DIPE) and 2,3,4,5,6-pentafluorobenzyl bromide (PFB-Br) was purchased from Sigma-Aldrich (St. Louis, MO, USA). Nordihydroguaiaretic acid (NDGA) was purchased from TCI America (Tokyo, Japan). DMEM and FBS were supplied by GIBCO (Gland Island, NY, USA). HPLC-grade hexane, methanol and isopropanol were obtained from Fisher Scientific Co. (Fair Lawn, NJ, USA). ACS-grade ethanol was obtained from Pharmco (Brookfield, CT, USA). Gases were supplied by BOC Gases (Lebanon, NJ, USA).

LA (0.50 ng, 1.79 pmol), AA (0.50 ng, 1.64 pmol), 13(R)-HODE (0.50 ng, 1.69 pmol), 13(S)-HODE (0.50 ng, 1.69 pmol), 9(R)-HODE (0.50 ng, 1.69 pmol), 9(S)-HODE (0.50 ng, 1.69 pmol), 11(R)-HETE (0.50 ng, 1.56 pmol), 11(S)-HETE (0.50 ng, 1.56 pmol), 12(R)-HETE (0.50 ng, 1.56 pmol), 12(S)-HETE (0.50 ng, 1.56 pmol), 15(R)-HETE (0.50 ng, 1.56 pmol), 15(S)-HETE (0.50 ng, 1.56 pmol), PGE2 (1.00 ng, 2.84 pmol), 11β-PGE2 (1.00 ng, 2.84 pmol), 8-iso PGE2 (1.00 ng, 2.84 pmol), PGD2 (1.00 ng, 2.84 pmol), PGF2α (10.0 ng, 28.2 pmol), 11β-PGF2α (10.0 ng, 28.2 pmol), iPF2α-III (10.0 ng, 28.2 pmol), [2H4]-PGE2 (1.00 ng, 2.81 pmol) and [2H4]-PGF2α (10.0 ng, 27.9 pmol) in acetonitrile (100 μL) were treated with 100 μL of PFB-Br in acetonitrile (1:19, v/v) followed by 100 μL of DIPE in acetonitrile (1:9, v/v) and the solution was heated at 60°C for 60 min. The solution was allowed to cool, evaporated to dryness under nitrogen at room temperature, and redissolved in 100 μL of hexane/ethanol (97:3, v/v) for normal-phase chiral chromatography ready for LC/MS analysis.

Normal-phase chiral chromatography for LC/MS experiments was performed using a Waters Alliance 2690 HPLC system (Waters Corp., Milford, MA, USA). Gradient elution was performed in the linear mode. A Chiralpak AD-H column (250 × 4.6 mm i.d., 5 μm; Daicel Chemical Industries, Ltd., Tokyo, Japan) was employed with a flow rate of 1.0 mL/min. For system 1, solvent A was hexane and solvent B was ethanol/hexane (7:3, v/v). For system 2, solvent A was hexane and solvent B was isopropanol/ethanol (9:1, v/v). The gradient for systems 1 and 2 was as follows; 4% B at 0 min, 4% B at 15 min, 30% B at 17 min, 30% B at 20 min, and 4% B at 22 min. For system 3, solvent A was hexane and solvent B was methanol/isopropanol (1:1, v/v). The linear gradient was as follows: 2% B at 0 min, 2% B at 3 min, 4% B at 13 min, 30% B at 15 min, 30% B at 20 min, 50% B at 22 min, 50% B at 25 min, and 2% B at 27min. The separation was performed at 30°C.

Mass spectrometry was conducted with a Thermo Finnigan TSQ 7000 triple-stage quadrupole mass spectrometer (Thermo Finnigan, San Jose, CA, USA) equipped with an APCI source in negative ion mode. The TSQ 7000 operating conditions were as follows: vaporizer temperature, 500°C; heated capillary temperature, 230°C; corona discharge needle, set at 16 μA. The sheath gas (nitrogen) and auxiliary gas (nitrogen) pressures were 40 psi and 10 (arbitrary units), respectively. Collision-induced dissociation (CID) was performed using argon as the collision gas at 2.7 mTorr in the second (rf-only) quadrupole. For full-scan and selected reaction monitoring (SRM) analyses, unit resolution was maintained for both precursor and product ions. LC/SRM-MS analysis was conducted using 50 pg of each compound as its PFB derivative. The following SRM transitions were monitored: LA-PFB, m/z 279 → 261 (collision energy, 20 eV); AA-PFB, m/z 303 → 259 (collision energy, 15 eV); 12-HETE-PFB, m/z 319 → 179 (collision energy, 16 eV); 15-HETE-PFB, m/z 319 → 219 (collision energy, 15 eV); 11-HETE-PFB, m/z 319 → 167 (collision energy, 18 eV); 13-HODE-PFB, m/z 295 → 195 (collision energy, 20 eV); 9-HODE-PFB, m/z 295 → 171 (collision energy, 20 eV); PGE2-PFB, m/z 351 → 271 (collision energy, 20 eV); [2H4]-PGE2-PFB, m/z 355 → 275 (collision energy, 20 eV); PGF2α-PFB, m/z 353 → 309 (collision energy, 22 eV); and [2H4]-PGF2α-PFB, m/z 357 → 313 (collision energy, 22 eV).

Rat intestinal epithelial cells transfected with COX-2 (RIES cells) were obtained from R. N. DuBois. Cells were cultured in DMEM supplemented with 10% FBS, 2 mM glutamine, 100 000 units/L penicillin, and 100 000 units/L streptomycin until almost confluent. The medium was then removed and replaced with medium containing 0.1% FBS or 0.1% FBS and NDGA (10, 20, 50 μM in ethanol). A portion of cell culture medium (3 mL) was transferred into a glass tube after 24 h of incubation. To quantify lipid metabolites, tubes containing cell culture medium (3 mL) were spiked with the following amounts of 14 authentic standards for lipid metabolites; 0.2, 0.5, 1.0, 2.0, 5.0, 10.0, and 20.0 ng. The mixture of internal standard ([2H4]-PGE2; 2 ng and [2H4]-PGF2α; 20 ng) was added to sample and standard solutions. The solutions were then adjusted to pH 3 with 2.5 N hydrochloric acid. Lipids were extracted with diethyl ether (4 mL × 2) and the organic layer was then evaporated to dryness under nitrogen. Analysis of the PFB derivatives (prepared as described above) by LC/electron capture APCI-multiple reaction monitoring (MRM)-MS analysis was conducted on a 20-μL aliquot of this solution. Calibration curves were obtained with a linear regression analysis of peak area ratios of analytes against internal standards (PGE2 and PGF2α). Concentrations of bioactive lipids in the medium were calculated by interpolation from the calculated regressions lines.


Electron capture APCI analysis of lipid PFB derivatives

PFB derivatives of LA, AA, LA-derived, and AA-derived bioactive lipids were analyzed under negative APCI conditions. Dissociative electron capture occurred for all analytes to form intense [M–PFB] ions at m/z 279 (LA), 303 (AA), 319 (12-, 15-, 11-HETE), 295 (13-, 9-HODE), 351 (PGE2, 11β-PGE2, 8-iso PGE2, PGD2), and 353 (PGF2α, 11β-PGF2α, iPF2α-III). Analysis of lipid-PFB derivatives by LC/electron capture APCI-MS using gradient system 1, 2 or 3 revealed that the derivatives had excellent chromatographic properties under normal-phase conditions.

MS/MS analysis of lipid-PFB derivatives

Collision-induced dissociation (CID) and tandem mass spectrometry (MS/MS) analysis was performed on the [M–PFB] ions from each analyte. MS/MS spectra of monohydroxy unsaturated fatty acids (HODEs and HETEs) showed intense and structurally significant product ions, which involved the cleavage of a carbon–carbon bond adjacent to the hydroxyl group (α-hydroxy cleavage) with transfer of a proton to the unsaturated site. In the case of prostaglandins, CID of [M–PFB] ions revealed structurally specific α-hydroxy cleavage. However, the most abundant ions were formed by loss of water and/or carbon dioxide.


MS/MS analyses of m/z 295 ([M–PFB]) revealed two product ions at m/z 277 and 195 for 13-HODE and m/z 277 and 171 for 9-HODE. The most abundant ion (m/z 277) was due to the loss of water. The product ion at m/z 195 from 13-HODE corresponded to α-hydroxy cleavage at the C-12 and C-13 vinylic position with an additional proton (Fig. 2(A)). α-Hydroxy cleavage at the C-9 and C-10 bond of 9-HODE was observed at m/z 171, which involved loss of a proton (Fig. 2(B)). The MS and MS/MS spectra for the R- and S-isomers were identical.

Figure 2.

Product ion spectra from CID of [M−PFB] at m/z 295 for PFB derivatives of HODEs or m/z 319 for HETEs: (A) 13-HODE; (B) 9-HODE; (C) 12-HETE; (D) 15-HETE; and (E) 11-HETE.


CID of m/z 319, the [M–PFB] ion for 12-, 15- and 11-HETE, gave rise to common product ions at m/z 301 (–H2O), 275 (–CO2) and 257 (–[H2O+CO2]) in all cases. Characteristic product ions for each compound resulted from α-hydroxy cleavage at the vinylic and/or allylic position with transfer of a proton. The most intense product ion from CID of 12-HETE [M–PFB] was observed at m/z 179, which originated from α-hydroxy cleavage at the vinylic position, C-11 and C-12 bond, and contained an additional proton (Fig. 2(C)). In addition, the CID spectrum of 12-HETE [M–PFB] showed another characteristic odd-electron ion at m/z 208 corresponding to α-hydroxy cleavage of the C-12 and C-13 bond at the allylic position. MS/MS analysis of [M–PFB] from 15-HETE revealed a characteristic product ion at m/z 219 resulting from α-hydroxy cleavage of the C-14 and C-15 bond at the vinylic position (Fig. 2(D)). The ion at m/z 175 was formed by loss of CO2 from this α-hydroxy cleavage ion. In contrast to 12- or 15-HETE, α-hydroxy cleavage of 11-HETE [M–PFB] at an allylic position, C-10 and C-11 bond, produced a most intense characteristic ion at m/z 167 that contained an additional proton (Fig. 2(E)). The ion at m/z 195 was observed with much less abundance, which was from the cleavage of C-11 and C-12 bond at a vinylic position. The MS and MS/MS spectra for the R- and S-isomers were identical.

PGEs and PGD2

The series of prostaglandin E (PGEs; PGE2, 11β-PGE2, 8-isoPGE2) and PGD2 are isomeric and gave very similar product ion spectra. Intense product ions resulted from the losses of one and two molecules of water at m/z 333 and 315, respectively. The most abundant ion was observed at m/z 271 corresponding to the loss of carbon dioxide as well as the loss of two molecules of water. The ions at m/z 233 and 189 were formed from the α-hydroxy cleavage of the C-14 and C-15 bond at the vinylic position with an additional loss of H2O and H2O+CO2, respectively (Fig. 3).

Figure 3.

Electron capture APCI-MS/MS analysis of PGE2 isomers.


The series of prostaglandin F2 derivatives (PGFs; PGF2α, 11β-PGF2α, iPF2α-III) are also isomeric. CID spectra of PGFs showed the most intense product ion at m/z 309, which resulted from the loss of CO2. Other characteristic ions for PGFs were observed at m/z 335, 291, 273, 253, 235, and 209 with less abundance, corresponding to –H2O, –[H2O+CO2], –[2H2O+CO2], α-hydroxy cleavage of the C-14 and C-15 bond at the vinylic position with a transferred proton, additional loss of H2O from m/z 253, and additional loss of CO2 from m/z 253, respectively (Fig. 4).

Figure 4.

Electron capture APCI-MS/MS analysis of PGF2α isomers.

Analysis of lipid-PFB derivatives by chiral phase LC/electron capture APCI-MRM-MS

Specific precursor and product ions were selected for each lipid-PFB derivative in order to perform LC/MRM-MS analysis (Table 1). The chiral separation of lipid-PFB derivatives was performed on Chiralpak AD-H, an amylose tris(3,5-dimetylphenyl carbamate) chiral stationary phase, with a mixture of hexane and ethanol or hexanes and isopropanol at 1 mL/min flow rate. When hexane/ethanol was used as the solvent system (gradient system 1, Fig. 5(A)), chiral separation was achieved for 12-, 15-, 11-HETE and 13-HODE. Using the hexane/isopropanol system (gradient system 2, Fig. 5(B)), chiral separation was accomplished except for 13-HODE enantiomers. The elution order of the R- and S-isomers was reversed compared with the hexane/ethanol system. However, by using a mixture of methanol and isopropanol (1:1, v/v) with hexane (gradient system 3, Fig. 6), it was possible to obtain baseline separation for all HETE and HODE enantiomers as well as PGE2 and PGF2α with their stable isotope internal standards [2H4]-PGE2 and [2H4]-PGF2α, respectively. Under these conditions, the retention times of the PGE2 isomers PGE2, PGD2, 11β-PGE2 and 8-iso PGE2 were 19.8, 20.3, 20.6, and 27.2 min, respectively. Retention times of the PGF2 isomers 11β-PGF2α, PGF2α, and iPF2α-III were 19.1, 20.9, and 24.1 min, respectively.

Table 1. Multiple reaction monitoring (MRM) transitions and LC retention times for targeted lipidomics analysis
AnalytesPrecursor M-PFB ion (m/z)Product ion (m/z)Retention time* (min)
  • *

    Obtained from gradient system 3 (Fig. 6).

Linoleic acid2792613.4
Arachidonic acid3032593.4
Figure 5.

Analysis of lipid-PFB derivatives using LC/electron capture APCI-MRM-MS with (A) the hexane/ethanol-based chiral HPLC system and (B) the hexane/isopropanol-based chiral HPLC system.

Figure 6.

Analysis of lipid-PFB derivatives using LC/electron capture APCI-MRM-MS using the hexane/methanol/isopropanol-based chiral HPLC system.

Analysis of lipid metabolites from RIES cells

The LC/electron capture APCI-MRM-MS profile for lipid metabolites from control RIES supernatants revealed the presence of 13(R)-HODE (0.56 nM; 15.5 min), 13(S)-HODE (0.48 nM; rt, 12.7 min), 15(S)-HETE (0.11 nM; rt, 16.0 min), and PGE2 (0.14 nM; rt, 19.7 min) (Fig. 7(A)). When the cells were treated with NDGA, there was a dramatic dose-dependent increase in the levels of PGE2 and PGF2α. With 20 μM NDGA, the PGE2 in the cell supernatant was increased from 0.14 to 0.70 nM and PGF2α was increased from <0.20 to 0.58 nM, (Fig. 7(B)). There was also a modest increase in 15(S)-HETE from 0.13 to 0.26 nM (Fig. 7(B)). At 50 μM NDGA, the PGE2 and PGF2α concentrations had increased to 0.87 and 0.67 nM, respectively. The 15(S)-HETE was not significantly increased at this higher concentration of NDGA. There were no significant changes in any of the other bioactive lipids present in the cell supernatant on NDGA treatment.

Figure 7.

Analysis of lipid-PFB derivatives extracted from COX-2-transfected RIE cell media (A) 24-h incubation and (B) 24-h incubation with 20 μM NDGA.


The formation of bioactive lipids from polyunsaturated (PUFA) fatty acids can occur by enzymatic pathways or through ROS-mediated free radical reactions (Scheme 1). AA- or LA-containing lipids release the free PUFAs when acted on by specific lipases.16 The free AA can then be converted to 15(S)-HPETE by 15-LOX-117 or 15-LOX-2,11 to 12(S)-HPETE by 12-LOX,17 or to 5(S)-HPETE by 5-LOX.18 COX-1 and -2 produce small amounts of 11(R)-HPETE and 15(S)-HPETE.16,19 The HPETEs undergo a two-electron reduction reaction mediated by glutathione peroxidases or glutathione transferases to give the corresponding HETEs together with a molecule of water.20 The major product of COX-1 and COX-2 is PGG2, which in turn is reduced to PGH2 by the peroxidase activity of the COX enzymes. PGH2 is then converted to PGs such as PGE2 and PGF2α by the action of various PG synthase enzymes.16 Free LA is a substrate for 15-LOX-1 and 15-LOX-2, which results in the formation of 13(S)-HPODE with high regio- and stereospecificity.11,17 LA is the preferred substrate compared with AA for 15-LOX-1,8 whereas AA is the preferred substrate for 15-LOX-2.8,17 COX-2 converts LA into a mixture of 9(R)-HPODE and 13(S)-HPODE.9 Interestingly, LA is a poor substrate for COX-1.21

15-LOX-1 (but not 15-LOX-2) can oxidize intact phospholipids20,22 and cholesteryl esters22,23 to give 13(S)-HPODE- and 15(S)-HPETE-containing lipids. The lipid hydroperoxides are subsequently reduced and released by specific lipases from the membrane lipids and appear as the free HODEs and HETEs.20 ROS-mediated lipid peroxidation of intact lipids can also occur, which results in the formation of 13-HPODE- and 15-HPETE-containing lipids.20 However, in contrast to the 15-LOX-1-derived products, they are racemic and contain a complex mixture of E and Z isomers. AA-containing lipids also undergo free-radical-mediated endoperoxide formation and conversion to iPs on intact lipids.24,25 Subsequent action of lipases results in the release of free iPs.26 This contrasts with COX-derived products, which are synthesized from free AA and are not stored as esterified lipids.16

The quantitative targeted lipidomics analysis is a valuable complementary mechanistic approach to conventional proteomics analysis of cellular proteins. It has proved difficult to conduct lipidomics analysis using GC/MS because of the thermal instability of many of the HETE analytes. Furthermore, the implementation of LC/MS-based methodology has been hampered by the difficulty in separating isomeric compounds using conventional reversed-phase chromatography. Thus, most investigators employ methodology that is based on LC/radioactivity detection of products from radiolabeled fatty acid substrates added to disrupted cells or to proteins isolated from cells.11,27 The ability to separate enantiomeric pairs of endogenously generated HETEs with high sensitivity using LC/MS is normally a formidable challenge. However, the electron capture APCI methodology simplifies this task considerably. Most bioactive lipids have a functional group that makes it possible to readily attach an electron-capturing derivative. There is a wealth of literature describing highly sensitive GC/ECNCI-MS studies using electron-capturing derivatives28 including, PGs,29 leukotrienes, iPs,30 epoxyeicosatrienoic acids,29 platelet activating factor,31 and thromboxanes,32 DNA-adducts33 and sterols.34 The derivatives used in these studies should all be amenable to use in combination with electron capture APCI. We chose to use the PFB derivative because it has proved to be an extremely robust derivative that has found an enormous number of GC/MS applications,33,35,36 since we first described its use for the analysis of endogenous prostaglandins.37 It is noteworthy the 2-nitro-4-trifluoromethylphenyl derivative38,39 has been used recently for the analysis of steroids by electron capture APCI and that nitroaromatic compounds40 and azasteroids41 can be analyzed by electron capture APCI without the need for derivatization.

The use of chiral chromatography with hexane/ethanol as solvent in combination with LC/electron capture APCI-MRM-MS for the PFB-ester derivatives of 9-HODE, 13-HODE, 11-HETE, 12-HETE, and 15-HETE resulted in separation of all the regioisomers. The PFB-esters of 9-HODE and 13-HODE were distinguished by the presence of product ions at m/z 171 and 195, respectively, whereas 11-HETE, 12-HETE, and 15-HETE were distinguished by specific product ions at m/z 167, 179, and 219, respectively (Fig. 5(A)). The R-enantiomers generally eluted ahead of the S-enantiomers. Unfortunately, the 12-HETE enantiomers did not separate to baseline and the 9-HODE enantiomers could not be separated using this system. With hexane/isopropanol as solvent the S-enantiomers tended to elute ahead of the R-enantiomers. The 9-HODE enantiomers were now separated but the 13-HODE and 15-HETE enantiomers were no longer resolved (Fig. 5(B)). By using a combination of methanol and isopropanol with hexane as solvent, all of the isomers and enantiomers were separated (Fig. 6). The isomeric PG derivatives such as PGD2 and PGE2, as well as PG and iP isomers such as PGF2α and iPF2α-III, were also resolved under these conditions. Quantitation was performed by use of two heavy isotope internal standards, [2H4]-PGE2 and [2H4]-PGF2α.

The need for targeted quantitative lipidomics is exemplified in our studies of endogenous bioactive lipids produced by RIE cells that have been transfected with COX-2 (RIES cells). Previous studies have suggested that there is no 15-LOX activity in the cells.27 However, we have observed lipid hydroperoxide-derived DNA-adducts12 that could have arisen from 15-HPETE. The targeted lipidomics approach has revealed that 15(S)-HETE is in fact an important endogenous metabolite of the COX-2-transfected RIE cells (Fig. 7(A)). As shown in Scheme 1, 15(S)-HETE is formed from a two-electron reduction of 15(S)-HPETE. Therefore, the DNA-adducts in the cells could have arisen from homolytic decomposition of the 15(S)-HPETE to bifunctional electrophiles.12 The 15(S)-HETE in the cell supernatant was not inhibited by treatment of the RIES cells with 400 μM of the specific COX-2 inhibitor NS-398 (data not shown). This ruled out the possibility that the 15(S)-HETE was a COX-2-derived product (Scheme 1). In view of the chiral nature of the 15-HETE (Fig. 7(A)), it appears that the cells possess a previously unrecognized 15-LOX activity. Surprisingly, 15(S)-HETE formation was not inhibited by the non-specific LOX inhibitor NDGA (Fig. 7(B)). In fact, there was a modest (almost two-fold) increase in 15(S)-HETE concentration in the cell supernatant when 20 μM NDGA was used. The 15-LOX-1 enzyme is known to be inhibited by NDGA so this suggested that another LOX activity was present in the cells. Interestingly, there was a substantial increase in PGE2 and PGF2α concentrations in the supernatant of RIES cells treated with 20 μM NDGA (Fig. 7(B)). This has stimulated additional proteomics analyses to identify the enzymes that are responsible for formation of 15(S)-HETE in the RIES cells under basal conditions and for the NDGA-mediated increase in PGE2 and PGF2α biosynthesis. 12(S)-HETE was detected in the RIES cell supernatant but was present in almost the same concentration as in the 0.1% fetal calf serum that was used in the cell incubation medium. Therefore, 12(S)-HETE was not a significant product of the RIES cells. The 13-HODE appeared to be primarily racemic and so it was most likely formed non-enzymatically.

The use of only two internal standards provides limited accuracy and precision for the procedure. More rigorous analyses can be performed when the most important analytes have been identified. This simply involves the incorporation of additional heavy isotope labeled internal standards into the analytical procedure. Furthermore, the assay can readily be modified to include bioactive lipids derived from other PUFAs such as eicosapentaenoic acid and docosahexaenoic acid. The targeted lipidomics approach can also be modified to include bioactive lipids from other pathways such as the cytochrome P-450-derived epoxyeicosatetraenoic acids, 5-LOX-derived leukotrienes, 5-, 12-, and 15-LOX-derived lipoxins, sphingomyelinase-derived ceramides, and ceramidase-derived sphingosines. We anticipate that this approach will be particularly useful when used in combination with the analysis of more abundant esterified lipids using conventional LC/MS methodology6 and with concomitant proteomics analysis.3


We acknowledge the support of NIH grants RO-1 CA91016, P01 CA77839, and RO DK47297.