Analysis of HETEs in human whole blood by chiral UHPLC-ECAPCI/HRMS

The biosynthesis of eicosanoids occurs enzymatically via lipoxygenases, cyclooxygenases, and cytochrome P450, or through nonenzymatic free radical reactions. The enzymatic routes are highly enantiospecific. Chiral separation and high-sensitivity detection methods are required to differentiate and quantify enantioselective HETEs in complex biological fluids. We report here a targeted chiral lipidomics analysis of human blood using ultra-HPLC-electron capture (EC) atmospheric pressure chemical ionization/high-resolution MS. Monitoring the high-resolution ions formed by the fragmentation of pentafluorobenzyl derivatives of oxidized lipids during the dissociative EC, followed by in-trap fragmentation, increased sensitivity by an order of magnitude when compared with the unit resolution MS. The 12(S)-HETE, 12(S)-hydroxy-(5Z,8E,10E)-heptadecatrienoic acid [12(S)-HHT], and 15(S)-HETE were the major hydroxylated nonesterified chiral lipids in serum. Stimulation of whole blood with zymosan and lipopolysaccharide (LPS) resulted in stimulus- and time-dependent effects. An acute exposure to zymosan induced ∼80% of the chiral plasma lipids, including 12(S)-HHT, 5(S)-HETE, 15(R)-HETE, and 15(S)-HETE, while a maximum response to LPS was achieved after a long-term stimulation. The reported method allows for a rapid quantification with high sensitivity and specificity of enantiospecific responses to in vitro stimulation or coagulation of human blood.


Sources of human plasma and serum
The clinical study was conducted in accordance with the Declaration of Helsinki. The clinical protocol (NCT02095288) was approved by the Institutional Review Board of the University of Pennsylvania and by the Advisory Council of the Center for Human Phenomic Science of the University of Pennsylvania. All the participants provided written informed consent, and were healthy, nonsmoking, and nonpregnant volunteers. The participants refrained from all the medications, including NSAIDs, for at least 2 weeks before a blood donation. Human whole blood was drawn by venipuncture. For plasma collection, whole blood was drawn with syringes containing 10 IU of sodium heparin and distributed in 500 l aliquots into 96-well deep sterile polypropylene plates. Zymosan (125 g/ml final concentration) or LPS (100 g/ml final concentration) was added to the blood in 20 l aliquots of PBS for single agonist stimulation or in 10 l aliquots each for a combination of stimuli, and incubated for 4 and 24 h at 37°C. Plates were covered with MicroClime-Environmental lids to minimize edge effects. After the incubation, blood was spun down at 3,000 g for 10 min at 4°C and plasma was removed for ECAPCI/ HRMS analysis. Blood samples, spun down in polypropylene tubes immediately after the blood draw, served as untreated controls. Whole blood from fifteen human volunteers was used for in vitro stimulation with LPS or zymosan (n = 15). Coincubation of LPS and zymosan was performed in whole blood from five independent subjects (n = 5). For serum preparation, 500 l aliquots of human whole blood were incubated in glass tubes at 37°C for 1 h and serum was isolated for UHPLC-ECAPCI/HRMS analysis. Nine genetically unrelated human volunteers provided blood for serum collection (n = 9).

Chiral eicosanoid extraction and synthesis of PFB derivatives
Plasma and serum samples (200 l) were spiked with 1 ng of the stable isotope-labeled internal standard for 15(S)-HETE ([ 2 H 8 ]-15(S)-HETE) in 900 l of acetonitrile. Then, plasma and sera were incubated with 1% formic acid at room temperature for 15 min. After that, samples were sonicated for 1 min and the supernatants were transferred to Phree cartridges for phospholipid and protein removal. The samples were eluted with a slight vacuum (<20 kPa) and dried under a gentle stream of nitrogen at the ambient temperature. The PFB derivatives were prepared by dissolving the residues from extracted plasma and sera in 100 l of diisopropylethylamine in acetonitrile (1:19, v/v) followed by the addition of 50 l of PFB-Br in acetonitrile (1:9, v/v) and the incubation of the solution at 60°C for 30 min. The solution was evaporated to dryness under a nitrogen stream at room temperature and redissolved in 100 l of hexane/ethanol (97:3, v/v). A 5 l aliquot of each sample was injected for chiral UHPLC-HRMS analysis.

Chiral UHPLC-ECAPCI/HRMS
Normal-phase chiral chromatography was performed using an UltiMate 3000 binary UPLC equipped with a refrigerated Scheme 1. Eicosanoid formation in humans. HPETE, hydroperoxyeicosatetraenoic acid. autosampler (6°C) and a column heater (35°C). Gradient elution was performed in the linear mode with some modification of a previously described method (7). A Chiralpak AD-H column (250 × 4.6 mm i.d., 5 m; Daicel Chemical Industries, Ltd., Tokyo, Japan) was employed with a flow rate of 1 ml/min. Solvent A was hexanes and solvent B was 2-propanol/methanol (5/5, v/v). The linear gradient was as follows: 2% B at 0 min, 2% B at 3 min, 8% B at 11 min, 8% B at 13 min, 50% B at 14 min, 50% B at 18 min, and 2% B at 18.5 min with an equilibration step for the next 2.5 min. The column effluent was diverted to waste before 3 min and after 13 min. MS was conducted on a Thermo QExactive HF HRMS. The mass spectrometer was equipped with an APCI source operating in negative electron capturing ion mode. The operating conditions were as follows: vaporizer temperature, 450°C; heated capillary temperature, 320°C; and corona discharge needle, set at 30 A. The sheath gas (nitrogen) and auxiliary gas (nitrogen) pressures were 40 psi and 10 (arbitrary units), respectively. The S lens was 60. The QE HF was alternating between full scan (m/z 100-600) at a resolution of 30,000 and parallel reaction monitoring (PRM) at 120,000 resolutions with a precursor isolation window of m/z 2 with normalized collision energy 20. The molecular [M  ] precursor was 319.23 for all the HETEs and 327.27 for [ 2 H 8 ]-15(S)-HETE that was used as an internal standard. Quantification was done based on the most intense product ion for each of the HETEs with ±1.5 ppm. The analysis of serum TxB 2 was performed as previously described (6).

Chiral data analysis
All data analyses were performed using Xcalibur software version 2.0 SR2 (Thermo Electron Corporation) from raw mass spectral data. Calibration standard samples were prepared with charcoal-stripped FBS. Calibration samples were spiked with the authentic standards of (±)5-HETE, (±)9-HETE, 11(S)-HETE, 11(R)-HETE, (±)12-HETE, 12(S)-HHT, (±)15-HETE, and 20-HETE in the amounts of 0, 0.1, 0.25, 0.5, 1, 2.5, 5, and 10 ng, and 1 ng of the internal standard [ 2 H 8 ]-15(S)-HETE. Lipids were extracted, purified, derivatized, and analyzed as described above for the analytical samples. Calibration curves were plotted using a linear regression of the peak area ratio of analytes against the internal standard. Concentrations of the chiral products were calculated by interpolation from the calculated regression lines. Data were normalized to sample volume and expressed as lipid concentration in nanograms per milliliter.

Statistics
Statistical analyses were performed using GraphPad Prism software version 5.0 for Mac OS X. Data represent the mean ± SEM of 9 donors for serum and plasma from untreated blood and of 5-15 donors for plasma from stimulated blood. The statistical significance of the differences between various groups was sought by unpaired two-tailed t-tests for comparison of serum and untreated plasma, and by paired two-tailed t-tests for the enantiomers in serum. To analyze the differences between stimulation groups in plasma per analyte, one-way ANOVA was used followed by Dunn's multiple comparison test. To compare the means of lipids between serum and plasma for all analytes per donor, two-way ANOVA was used. To analyze the differences between stimulation conditions for all analytes per time point, two-way ANOVA was used followed by Dunn's test with Bonferroni P-value adjustment. P < 0.05 was considered statistically significant. Plots with lipids expressed in nanograms per milliliter were built using GraphPad Prism software. Principal component analysis (PCA) and visualizations were performed using custom R scripts.

HR-MS/MS analysis of eicosanoids-PFB by ECAPCI
PFB derivatives of eicosanoids were analyzed after chromatographic separation under negative mode using APCI. All HETE isomers had the same molecular ion, so only one PRM was used to determine the HRMS/MS data. Figure 2 shows examples of the HRMS/MS spectra for some HETEs. The most intense product ions for each PFB derivative are listed in Table 1. All the HETEs and 12(S)-HHT showed MS/HRMS like those reported from lowresolution instruments (data not shown) (8)(9)(10)(11)(12).

Chiral separation of eicosanoids-PFB
Characteristic HRMS product ions were selected for each of the HETE isomers based on the most intense product ion. The chiral separation was modified from a previous report (7), shortening the gradient to reduce the time for analysis. Switching to the UPLC system, the enantiomers for 5-HETE were baseline separated with 0.3 min between 5(R)-HETE (11.0 min) and 5(S)-HETE (11.3 min). This separation is challenging to achieve with a normal HPLC system, but on the UPLC system, the retention times exhibit less than 0.1 min shifts after running more than 300 samples.
The chiral LC-ECAPCI-MS/HRMS profile of human serum and plasma from unstimulated whole blood (Fig. 3) identified most of the chiral HETEs, but due to the high concentration of 12(S)-HETE, 12(R)-HETE was only detected in plasma, but not in serum. The chromatogram showed that 15(R)-HETE [retention time (rt) 9.6 min] was present in a 1:3 ratio to 15(S)-HETE (rt 11.1 min) in serum, but the ratio was closer to 1:1 in human plasma. The 11(R)-HETE (rt 8.9 min) was the major enantiomer present in serum, but in the plasma the enantiomers exhibited less enantiospecificity with the 11(S)-HETE (rt 9.8 min), showing about the same intensity. The 8(R)-HETE (rt 9.2 mi) and 8(S)-HETE (rt 9.8 min) showed no enantioselectivity in serum or plasma, and the same was true about 9(R)-HETE (rt 9.4 mi) and 9(S)-HETE (rt 9.6 min). Both serum and plasma from unstimulated whole blood showed 20-HETE (rt 9.8 min).
Quantitation was performed based on the MS/HRMS data using a 3 ppm (±1.5 ppm) window with qualifying peaks from the HR full scan data. Reinjection of samples after storage in the autosampler for 24 and 48 h resulted in the calculated amounts of the eicosanoids and signal intensity within a 5% coefficient of variation (data not shown).

DISCUSSION
The analytical method reported here employs chiral UHPLC-ECAPCI/HRMS and allows comprehensive, rapid, and highly sensitive quantification of isomeric and enantiomeric eicosanoids in human whole blood. Combining chiral chromatography with HRMS, we detected differences in the variety of chiral species and the magnitude of response during in vitro activation of inflammatory versus coagulation pathways.
It is important to be able to distinguish the enantiomeric excess of the isomeric compounds with high specificity and sensitivity to segregate an enzymatic process (chiral products) from oxidative stress (racemic products). In a biological system, which has the enzymatic machinery on the one hand, and is prone to oxidative stress, on the other hand, both enzymatic and nonenzymatic oxidations will contribute to enantiomer formation to varied degrees. Typically, resolution of such molecules has been conducted using normal-phase chiral chromatography coupled with UV detection (13) or MS (14). Unfortunately, the sensitivity of MS analysis using conventional ESI methodology and normal-phase solvents is rather poor. This methodology has a quite limited applicability for analyzing nonesterified bioactive lipids such as 15-HETE. Sensitivity can be enhanced by the use of APCI-MS after derivatization with an electroncapturing group like PFB. This technique, called ECAPCI-MS, provides a substantial increase in sensitivity when compared with conventional negative ion APCI methodology Human whole blood was incubated at 37°C for 1 h and serum was removed for analysis of TxB 2 and chiral HETEs, as described in the Materials and Methods. Lipids from serum were compared with plasma lipids from untreated whole blood. Data are expressed as mean ± SEM, serum versus plasma, unpaired two-tailed t-test, n = 9. ND, not detected; ns, not significant. a P  0.01, (R) versus (S) enantiomers, paired two-tailed t-test, n = 9. b P  0.05, (R) versus (S) enantiomers, paired two-tailed t-test, n = 9. c P  0.0001, (R) versus (S) enantiomers, paired two-tailed t-test, n = 9.   5. Time-and stimulus-dependent effects of in vitro stimulation on chiral HETEs in human whole blood. Heparinized human whole blood was stimulated with 100 g/ml LPS and 125 g/ml zymosan (Zym), alone (n = 15) or in combination (L+Z, n = 5) for 4 and 24 h at 37°C. Plasma was removed for analysis of chiral HETEs as described in the Materials and methods, and significant fold differences over the vehicle (PBS) control were summarized on a log scale in a heat map (A). Red and blue boxes denote elevation and reduction, respectively. Unpaired two-tailed t-test, n = 5-15. Absolute quantities of lipids are summarized in Table 2. Effects of stimulating conditions on lipid production at 4 h (B) and 24 h (C). Shaded areas depict interquartile range of distribution of the lipid concentrations. D: PCA of plasma lipids after stimulation of whole blood for 4 h and 24 h and presented in three-dimentional loading plots. Each symbol represents an independent subject (n = 5-15). E: Venn diagrams showing the number of analytes common or unique to treatments after 4 h and 24 h of stimulation, relative to PBS control. The percentage of total analytes affected by a specific treatment is shown in boxes. of underivatized analytes (15). We reasoned that EC methodology could provide an excellent way to analyze chiral lipids by HRMS, analogous to using low-resolution LC-MS/ MS (16).
Previous studies showed that PFB derivatization increases sensitivity of detection for carboxy- (17) or phenolcontaining (18) molecules. These studies used triple quadrupole instruments with unit resolution in MRM modes. The MS/HRMS spectra for each isomeric HETE showed specific fragments, with the -cleavage next to the hydroxyl group being the most intense fragment, as reported before for methods involving triple quadrupole instruments (11,19,20).
We used only one HETE internal standard {[ 2 H 8 ]15(S)-HETE} in this report. However, deuterated internal standards are available for almost all the HETEs, and their usage will increase the accuracy and precision of the method. More rigorous analysis can be achieved by including additional heavy isotope internal standards into the sample preparation step, both for less polar eicosanoids, such as the oxo-ETEs, and for more polar products, such as the prostaglandins and leukotrienes. An extended targeted lipidomics method could also include metabolites from other pathways, like epoxyeicosatrienoic acids that result from the cytochrome P450 pathway (21).
Comparison of serum to untreated plasma revealed a robust production of chiral species during blood clotting. Anionic surfaces trigger the intrinsic pathway of coagulation through the auto-activation of the contact system factor XII (FXII) freely circulating in the bloodstream (22,23). In this study, we activated blood clotting in vitro using negatively charged glassware, and measured the maximum capacity of human blood to synthesize bioactive eicosanoids, particularly chiral HETEs, during coagulation. The 12(S)-HETE, 12(S)-HHT, and 15(S)-HETE were the most abundant hydroxylated chiral lipids in serum, which, together with TxB 2 , are attributable to platelet activation during coagulation (22,24,25). Platelets produce eicosanoids through the oxidation of arachidonic acid by 12-LOX and COX-1 enzymes, and nonenzymatic conversion of the PGH 2 substrate (25). Serum 12(S)-HETE is exclusively formed by the platelet-type, 12(S)-LOX (1,26), while 12(R)-HHT is generated nonenzymatically and through the enzymatic conversion of arachidonic acid by COX-1 and Tx synthase (25). Considering that 12(R)-HETE is synthesized by cytochrome P450 in the ocular system (27) and by 12(R)-LOX in the epidermis and tonsils (28,29), we hypothesize that low levels of serum 12(R)-HETE are formed nonenzymatically. The role of 12(S)-HETE in platelet biology is not fully understood, with studies reporting both activation (30)(31)(32) and inhibition (33,34) of platelet aggregation. The 12-HETE has been implicated in the pathophysiology of cancer, arteriosclerosis, and diabetes (26). The 12(S)-HHT was formed in comparable amounts to TxB 2 and, overall, accounts for roughly 30% of the arachidonate metabolites formed by activated platelets (35). Despite considerable production of 12(S)-HHT during blood clotting, its biological function is still unknown. Sources of serum 15(S)-HETE include eosinophil 15-LOX-1 and both COX isoforms, COX-1 from platelets and COX-2 from monocytes (1,14,36). However, 15(S)-HETE, as well as the R-enantiomer, originated primarily from platelet COX-1. The 15-HETE exerts pro-coagulant effects by increasing thrombin-induced activation and aggregation of platelets (34). Serum 11(R)-HETE can also be attributable to platelet COX-1 that generates 11-HETE only in the R-configuration (2). Finally, production of 5(S)-HETE through catalysis by 5-LOX indicates activation of neutrophils, which contribute to coagulation and thrombosis (37,38). Taken together, our method detected a wide variety of chiral HETEs in coagulated blood and confirmed the predominance of products formed enzymatically over those formed nonenzymatically.
Overall, using chiral chromatography and HRMS, we detected a much broader variety of chiral lipids than reported previously (6). The 12(S)-HETE was robustly and continuously produced in human whole blood, irrespective of the stimulating conditions. This is consistent with the observation that platelet 12(S)-LOX, unlike COX-1, continues to oxidize arachidonic acid over time, and that these enzymes use different pools of arachidonic acid and differentially rely on cytosolic phospholipase A 2 to release the arachidonate substrate (46). The 20-HETE is a potent vasoconstrictor made by vascular smooth muscle cells, among other cellular sources (2). In whole blood, it can be generated by stimulated neutrophils (47). Interestingly, plasma 20-HETE decreased after a long-term incubation with either LPS or zymosan. Activated platelets metabolize 20-HETE via 12(S)-LOX-and COX-1-dependent mechanisms (48) that may account for reduction in 20-HETE levels in stimulated Fig. 6. Enantioselective biosynthesis of 15-HETE compared with nonenzymatic formation of 9-HETE in stimulated human blood. Heparinized human whole blood was stimulated with 100 g/ml LPS and 125 g/ml zymosan (Zym), alone (n = 15) or in combination (LPS+Zym, n = 5), for 4 and 24 h at 37°C. Plasma was removed for analysis of 15-HETE (A) and 9-HETE (B) as described in the Materials and Methods. Data are expressed as mean ± SEM; *P  0.05, **P  0.01, ***P  0.001, ****P  0.0001 versus PBS; one-way ANOVA, Dunnett's test; n = 5-15. ns, not significant. whole blood. In addition, a decrease in 20-HETE may result from esterification into phospholipids of cellular membranes (49).
In conclusion, we have developed a method that allows for a rapid quantification with high sensitivity and specificity of enantiomeric lipids, particularly HETEs, in human whole blood. Although the precise origin of each chiral product is speculative, the method differentiates enzymatic versus nonenzymatic chiral species. This analytical method can also be readily applied to the interrogation of other classes of nonesterified oxidized lipids in various biological matrices (49).