Capillary Electrophoresis of Cytochrome P-450 ... - ACS Publications

Oct 19, 2002 - The essential fatty acid arachidonate is oxidized by cyto- chrome P-450 epoxygenases to four epoxyeicosatrienoic acids (EETs): 14,15-, ...
0 downloads 0 Views 144KB Size
Download PDF
Anal. Chem. 2002, 74, 5859-5865

Capillary Electrophoresis of Cytochrome P-450 Epoxygenase Metabolites of Arachidonic Acid. 1. Resolution of Regioisomers Victoria A. VanderNoot†,‡ and Mike VanRollins*,§

Departments of Chemical & Biochemical Engineering and Internal Medicine, University of Iowa, Iowa City, Iowa 52242

The essential fatty acid arachidonate is oxidized by cytochrome P-450 epoxygenases to four epoxyeicosatrienoic acids (EETs): 14,15-, 11,12-, 8,9-, and 5,6-EETs. Each of the four EET regioisomers and their hydrolysis products (DHETs) has multiple paracrine and autocrine functions and may also potently dilate blood vessels and activate potassium channels. The present work describes a method to resolve EETs and DHETs by capillary electrophoresis (CE) using trimethyl-β-cyclodextrin and CH3CN as buffer additives. While stored at 25 °C, most of the EET and DHET regioisomers remained intact when suspended in alkaline vehicle. However, under these same conditions, 5,6-EET rapidly broke down to a lactone and was slowly converted to 5,6-DHET. When subjected to CE, the EET and DHET regioisomers were baseline resolved (R g 1.3); 10 pg of an EET or a DHET regioisomer was readily detectable at 194 nm. In addition, the UV spectra were regiospecific and identical to those obtained during HPLC except that an additional, weak absorption occurred at 235 nm. Together, the highsensitivity, high-resolution, and differential UV spectra permitted the identification and quantification of EETs in phospholipids isolated from murine liver. Thus, CE was successfully used for the trace analysis of eicosanoids. Arachidonate, an essential fatty acid, is the substrate for multiple cellular oxygenases, which generate products with varied and potent biological effects.1 Arachidonate is normally concentrated in cell membranes by being esterified to phosphatidylcholine and phosphatidylinositol. Upon release from cell membranes by receptor-activated phospholipases, arachidonate is rapidly oxygenated by three competing enzyme systems: cyclooxygenases, lipoxygenases, and cytochrome P-450 epoxygenases. Recently, much interest has focused on the third pathway. Cytochrome P-450 epoxygenases are heme-containing enzymes that are most concentrated in liver and generate four cis-epoxyeicosatrienoic acids (EETs): 14,15-, 11,12-, 8,9-, and 5,6-EETs.2 The four products * Corresponding author. Phone: (319) 356-7629. Fax: (319) 356-7893. E-mail: [email protected]. † Department of Chemical & Biochemical Engineering. ‡ Current address: Sandia National Laboratories, P.O. Box 969, Livermore, CA 9455-0969. § Department of Internal Medicine. (1) Funk, C. D. Science 2001, 294, 1871-1875. (2) Zeldin, D. C. J. Biol. Chem. 2001, 276, 36059-36062. 10.1021/ac025909+ CCC: $22.00 Published on Web 10/19/2002

© 2002 American Chemical Society

differ in the epoxido position (Figure 1). Specific for each tissue, cytochrome P-450 epoxygenase isozymes generate the four EET regioisomers in varying proportions. Moreover, the epoxide ring in each regioisomer is cleaved by epoxide hydrolases to form a threo vicinal diol, dihydroxyeicosatrienoic acid (DHET; Figure 1). Because tissue-specific isoforms of epoxide hydrolases occur, the profiles of DHET regioisomers are also believed to be tissue specific. Thus, identifying the EET and DHET profiles in tissues will help establish the extent to which P-450 epoxygenases utilize released arachidonate and will clarify which EETs and DHETs are available. Outside the cardiovascular system, EETs and DHETs have multiple paracrine and autocrine functions.2 Within the cardiovascular system, they exhibit a wide range of effects that include potent vasodilation3 and inhibition of platelet aggregation.4,5 Interestingly, EETs also circulate in the blood and are excreted into the urine.6,7 An increased EET secretion into urine is associated with hypertension and atherosclerosis.3,6 In addition to disease, trauma to the vascular endothelium, ischemia, and high cholesterol diets stimulate the synthesis and release of EETs from the blood vessel wall. Moreover, physiological increases in plasma levels are likely because EETs act as endothelial-dependent hyperpolarizing factors (EDHFs).8 In this role, EETs and DHETs are esterified to phospholipids in vascular endothelial cells9 and may be released to act as potent arteriolar dilators and activators of large-conductance, calcium-activated potassium channels.8,10,11 Thus, EET levels may be elevated by physiological and pathological stimuli. Results on solid tissues indicate that EET regioisomer and enantiomer contents vary greatly and that increases in EET concentrations may be regioisomer specific.2,12 However, the regio(3) Roman, R. J. Physiol. Rev. 2002, 82, 131-185. (4) Fitzpatrick, F. A.; Ennis, M. D.; Baze, M. E.; Wynalda, M. A.; McGee, J. E.; Liggett, W. F. J. Biol. Chem. 1986, 261, 5334-5338. (5) VanRollins, M. J. Pharmacol. Exp. Ther. 1995, 274, 798-804. (6) Catella, F.; Lawson, J. A.; Fitzgerald, D. J.; FitzGerald, G. A. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 5893-5897. (7) Karara, A.; Wei, S.; Spady, D.; Swift, L.; Capdevila, J. H.; Falck, J. R. Biochem. Biophys. Res. Commun. 1992, 182, 1320-1325. (8) Zhang, Y. D.; Oltman, C. L.; Lu, T.; Lee, H. C.; Dellsperger, K. C.; VanRollins, M. Am. J. Physiol. Heart Circ. Physiol. 2001, 280, H2430-H2440. (9) VanRollins, M.; Kaduce, T. L.; Fang, X.; Knapp, H. R.; Spector, A. A. J. Biol. Chem. 1996, 271, 14001-14009. (10) Oltman, C.; Weintraub, N. L.; VanRollins, M.; Dellsperger, K. Circ. Res. 1998, 83, 932-939. (11) Lu, T.; Katakam, P. V. G.; VanRollins, M.; Weintraub, N. L.; Spector, A. A.; Lee, H. C. J. Physiol. (London) 2001, 534, 651-667. (12) Dumoulin, M.; Salvail, D.; Gaudreault, S. B.; Cadieux, A.; Rousseau, E. Am. J. Physiol. Lung Cell. Mol. Physiol. 1998, 19, L423-L431.

Analytical Chemistry, Vol. 74, No. 22, November 15, 2002 5859

Figure 1. Structures of the 14,15-EET and 14,15-DHET arising from cytochrome P450-catalyzed oxygenation of the essential fatty acid arachidonate. Initial epoxygenations at other double bonds (see numbering) and subsequent hydrolysis generates the 11,12-, 8,9-, and 5,6-EETs and corresponding DHETs.

and stereochemical composition of vascular EETs is unknown, in part because EET levels in cardiovascular tissues are much lower than in solid organs;2 the concentration of total unesterified EETs in rat and human plasma concentrations is only ∼1.0 nM.7 Thus, sensitive methods to identify EET/DHET profiles in vascular tissues are in great demand. The present studies focus on resolving EET and DHET regioisomers (part 1) and 14,15EET enantiomers (part 2). No simple method is available to quantitate EET and DHET regioisomers. The identification and quantitation of EET regioisomers generally requires HPLC, derivatization, and GC/MS.13-15 Because GC is unable to resolve three of the four EET regioisomers,13 gas chromatography/mass spectrometry (GC/MS) analysis requires that EETs first be resolved by sequential reversedand normal-phase high-performance liquid chromatography (HPLC).16 The combination of multiple chromatographic steps and (13) VanRollins, M.; Knapp, H. R. J. Lipid Res. 1995, 36, 952-966. (14) Karara, A.; Dishman, E.; Blair, I.; Falck, J. R.; Capdevila, J. H. J. Biol. Chem. 1989, 264, 19822-19827. (15) Bernstrom, K.; Kayganich, K.; Murphy, R. C. Anal. Biochem. 1991, 198, 203-211.

5860

Analytical Chemistry, Vol. 74, No. 22, November 15, 2002

incomplete derivatizations impact GC/MS sensitivity and detection limits. Recently, EETs have also been analyzed using microbore reversed-phase HPLC after EETs are esterified to a multiringed derivative for fluorescence measurements.17 However, HPLC resolution of the 8, 9- and 11, 12-EET ester derivatives is partial at best. Therefore, alternative separation techniques to GC and HPLC are desirable. Capillary electrophoresis (CE) is an attractive option to HPLC and GC. Regarding the ability to resolve and detect small molecules, CE typically has much higher efficiencies (N .100 000 theoretical plates/m) than HPLC (N ∼ 60 000 plates/m). The inherently high efficiency of CE contributes in part to the high mass sensitivity associated with CE; in contrast to the microgram amounts required for HPLC, only picogram quantities need be injected onto a CE column for detection purposes. (Up to now, publications on HPLC of underivatized EETs describe only the use of conventional 0.46-cm-i.d. columns; however, microbore and (16) Capdevila, J. H.; Dishman, E.; Karara, A.; Falck, J. R. Methods Enzymol. 1991, 206, 441-453. (17) Nithipatikom, K.; Pratt, P. F.; Campbell, W. B. Am. J. Physiol. Heart Circ. Physiol. 2000, 279, H857-H862.

capillary HPLC are more appropriate comparisons and are expected to exhibit mass sensitivities at least in the higher picogram to nanogram range.) Recently, capillary micellar electrokinetic chromatography18 and CE using indirect fluorescence19 were used to resolve free fatty acids, which are structurally similar to EETs. Unlike GC, CE can utilize buffer additives in the mobile phase to fine-tune selectivity. Cyclodextrins are perhaps the bestknown additives; although primarily used as chiral selectors (part 2), cyclodextrins are also effective in achiral resolutions.20 In the present study, we report that CE with trimethyl-β-cyclodextrin resolves picogram amounts of underivatized EETs and DHET regioisomers. Both the migration times and the UV spectra were used to identify EETs and DHETs. This method was used to determine endogenous EET levels in murine liver. EXPERIMENTAL SECTION Materials. Unsubstituted cyclodextrins (R, β, γ), plus (R,S)hydroxypropyl-, hydroxyethyl- (Astec, Whippany, NJ), and 2,6-diand 2,3,6-trimethylated β-cyclodextrins, plus hydroxypropyl- R- and γ-cyclodextrins (Astec; Aldrich, St. Louis, MO), were weighed out in a Teflon vial, capped, and mixed in a running buffer (CH3CN/ 50 mM sodium phosphate, pH 7.0), filtered (0.22-µm Teflon filters; MSI, Westbury, MA), and sonicated for 40 s before use. Unlabeled and [1-14C]-labeled EET/DHET regioisomers were synthesized as before.13,21 Capillary Electrophoresis. After EETs and DHETs were suspended in the sample vehicle (5 mM sodium phosphate, pH 8.0), all samples were transferred to plastic vials (BeckmanCoulter, Fullerton, CA) for immediate analysis or stored in screwcapped Teflon vials to eliminate adsorptive losses on glass. Separations were carried out using a Beckman 5500 P/ACE instrument equipped with a photodiode array detector. We used fused-silica capillaries with 50-µm-i.d. and 150-µm extended light paths (Agilent Technologies). To permit correct positioning of the light path in Beckman CE cartridges, interfering column collars were removed by soaking them in acetone for several hours. Each column was cut to length (97 cm ) Lt), and positioned over the cartridge slit using a microscope (90 cm ) Leff). To eliminate carryover between samples and reduce baseline noise, 3-4 mm of polyimide covering was removed from each end of the columns using a resistive heating stripper (Scientific Resources, Inc., Eatontown, NJ). Samples were made up in a low ionic strength (“stacking”) solution (5 mM sodium phosphate, pH 8.0). Injections were made by applying 0.5 psig pressure for 25 s. Separations were done at 20 °C and at 20 kV (205 V/cm under normal polarity). Resolution and Identification of Regioisomers. Screening of the capacity of various β-cyclodextrins to resolve EET and DHET regioisomers was done using a running buffer composed of 50 mM sodium phosphate (pH 7) and 15 mM cyclodextrin. Resolution was optimized by varying the CH3CN content of the running buffer from 0 to 40% (v/v). Stability during CE Analysis. Immediately after the ethanol vehicle was removed with a nitrogen stream, a mixture of 5,6(18) Collet, J.; Gareil, P. J. Chromatogr., A 1997, 792, 165-177. (19) Wang, T. L.; Wei, H. P.; Li, S. F. Y. Electrophoresis 1998, 19, 2187-2192. (20) Luong, J. H. T.; Nguyen, A. L. J. Chromatogr., A 1997, 792, 431-444. (21) VanRollins, M.; Kaduce, T. L.; Knapp, H. R.; Spector, A. A. J. Lipid Res. 1993, 34, 1931-1942.

EET (30 µg), 8,9-EET (20 µg), 11,12-EET (24 µg) and 14,15-EET(19 µg) was suspended in 900 µL of 5.0 mM sodium phosphate (pH 8.0) for 1.0 min. A 300-µL aliquot of the EET mixture was transferred to a 500-µL plastic vial and positioned in the sample rack for injection every 1.5 h. The remaining 600 µL of the EET mixture was stored at 25°C in a sealed Teflon vial for injections at 16 h and 5 days. In a similar fashion, a mixture of DHETs (14-17 ng/µL) was prepared and repeatedly injected over a 17-h period. Quantitation Studies. EETs or DHETs of varying dilutions were mixed and pressure injected (0.5 psig) for 25 s (17.5 nL) into a 97 cm × 50 µm i.d. capillary (90-cm effective length). Separations were done at 20 °C using normal polarity, 20 kV (205 V/cm), and a running buffer of 50 mM sodium phosphate (pH 7.0) containing CH3CN (60:40, v/v) and 15 mM trimethyl-βcyclodextrin. Absorbance was monitored at 194 ( 2 nm in an extended light path (150 µm) capillary fitted with a 100 × 800 µm slit width. Ultraviolet spectra from 190 to 300 nm were simultaneously collected using a photodiode array detector. To minimize evaporative effects, all EET dilutions were prepared using Hamilton syringes just prior to injection. The final concentrations of EET mixtures were assessed by liquid scintillation techniques. Retention times and integrated peak areas (velocity adjusted) were normalized to that of arachidonate. Isolation of Murine Phospholipid EETs. Lipids from mouse liver were extractively isolated, separated into neutral and phospholipid fractions by column chromatography, and saponified.9 After internal standards [1-14C]-8,9-, -11,12-, and -14,15-EETs (43 mCi/µg) were added to the phospholipid fractions (n ) 4), 93 ( 6% of the applied radioactivity was recovered in the organic extract following saponification. A retention time marker, the methyl ester of 5,6-DHET (5,6-DHET-ME), was added to the saponified phospholipids, and the mixture was injected onto an octadecasilyl column. EETs and DHETs were resolved in CH3CN/H2O (pH 2.3) (46:54, v/v), flowing at 1.5 mL/min and 1250 psig. HPLC fractions I and II contained 0.7 ( 0.1% (n ) 4) of the radioactivity from [1-14C]EET standards; thus, little DHET was generated during EET processing. Fractions III and IV (total EETs) were collected and pooled for analysis by CE. RESULTS AND DISCUSSION Resolution Using Trimethyl-β-cyclodextrin. Multiple cyclodextrin and CH3CN concentrations were explored as buffer additives. To resolve the four EET regioisomers, we screened unsubstituted R-, β-, and γ-cyclodextrins, and a series of substituted cyclodextrins, which included hydroxypropyl-, hydroxyethyl-, di-and trimethyl-β-cyclodextrins, and hydroxypropyl- R- and γ-cyclodextrins. Each was tested over a range of CH3CN concentrations from 0 to 20% (v/v). The best candidate was a lipophilic, permethylated (trimethyl)-β-cyclodextrin, which partially resolved the EET regioisomers if suspended in 20% CH3CN. When the [CH3CN] was increased to 40% (v/v), all four EET regioisomers were baseline resolved (R ) 1.4) in the presence of 15 mM trimethyl-β-cyclodextrin (Figure 2A). Keeping the [CH3CN] at 40% but decreasing the [cyclodextrin] reduced both the column capacity to resolve EETs and the dynamic range for EET quantitations. In contrast, keeping the [trimethyl-β-cyclodextrin] at 15 mM but lowering the [CH3CN] reduced EET migration times, but also failed to fully resolve the EET regioisomers. Analytical Chemistry, Vol. 74, No. 22, November 15, 2002

5861

Figure 2. Electropherograms and UV spectra of EET and DHET standards. (A) EET (16-33 ng/µL) or (B) DHET (14-17 ng/µL) standards were suspended in 5 mM sodium phosphate (pH 8.0), and 17.5 nL was injected by applying 0.5 psig for 25 s. Individual eicosanoids were resolved at 20 °C in 50 mM sodium phosphate (pH 7.0)/CH3CN (60:40, v/v) containing 15 mM trimethyl-β-cyclodextrin by applying 20 kV. EET and DHET absorptions at 194 nm and UV spectra (C, D) were monitored using a photodiode array detector.

Therefore, 40% CH3CN and 15 mM trimethyl-β-cyclodextrin were routinely used in all subsequent studies. The EETs appear to bind to trimethyl-β-cyclodextrin by hydrophobic interactions. In the absence of cyclodextrin, the anionic EETs migrated more slowly than the electroosmotic flow (EOF) due to electrostatic attractions at the capillary inlet or anode (data not shown). In the presence of cyclodextrin, EETs migrated more rapidly due to binding to the neutral cyclodextrin, which is transported solely by the EOF. By adding individual standards to the EET mixture, the migration order was determined to be 5,6-, 8,9-, 11,12-, and 14,15-EET (Figure 2A). The CE migration order was opposite that seen in reversed-phase HPLC where an octadecasilyl stationary phase and mixtures of CH3CN/H2O (pH 2.2) were employed. In reversed-phase HPLC, eicosanoids of greater hydrophobicity are retained longer by the column, slowing the overall migration toward the detector.22 Because the most hydrophobic EETs migrated fastest during CE, the mechanism of EET binding probably reflects hydrophobic interactions within the cavity of permethylated cyclodextrin.23-27 Like EETs, DHETs appeared to bind to trimethyl-β-cyclodextrin by hydrophobic bonds. Under the same trimethyl-β-cyclodextrin and CH3CN conditions employed for resolving EETs, the four DHET regioisomers were well resolved (R ) 1.4) (Figure (22) VanRollins, M.; Aveldano, M. I.; Sprecher, H. W.; Horrocks, L. A. Methods Enzymol. 1982, 86, 518-530. (23) Miertus, S.; Frecer, V.; Chiellini, E.; Chiellini, F.; Solaro, R.; Tomasi, J. J. Inclusion Phenom. 1998, 32, 23-46. (24) Davidenko, T. I. Polym. Sci. A 2000, 42, 502-506. (25) Lichtenthaler, F. W.; Immel, S. Starch-Starke 1996, 48, 145-154. (26) Saito, Y.; Ueda, H.; Abe, M.; Sato, T.; Christian, S. D. Colloids Surf. A-Physicochem. Eng. Aspects 1998, 135, 103-108. (27) So, T. S. K.; Huie, C. W. J. Chromatogr., A 2000, 872, 269-278.

5862 Analytical Chemistry, Vol. 74, No. 22, November 15, 2002

2B). Moreover, the order of elution was 5,6-, 8,9-, 11,12-, and 14,15-DHET and reflected regioisomer hydrophobicity, as determined by reversed-phase HPLC.9 Thus, like the parent EETs, the DHETs were separated and eluted in an order that suggested primarily hydrophobic interactions with a lipophilic β-cyclodextrin. UV Spectra. In addition to monitoring migration times, the use of an on-line photodiode detector allowed concurrent acquisitions of UV absorption spectra from EETs and DHETs. As expected, the presence of multiple, isolated double bonds in these eicosanoids caused maximal absorptions at 192-194 nm.22 Moreover, the UV spectra of 5,6-EET and 14,15-EET were indistinguishable from each other but differed from those of the 8,9- and 11,12EETs. Compared to 8,9- and 11,12-EET, the 5,6-EET and 14,15EET regioisomers had marked increases in absorption in the 195-218-nm range (Figure 2, panel C). The differences in the absorption spectra were not due to changes in the background contributions; background spectra taken immediately before and after the peaks were identical. In addition, essentially the same spectrum was obtained when either a high (380 pg) or low (10 pg) amount of a specific regioisomer was injected. Comparable differences in absorption spectra were observed for DHETs (Figure 2, panel D). Moreover, the same spectral differences were seen during the reversed-phase HPLC of EETs and DHETs. One possible explanation for the different UV spectra is the occurrence of internal “solvent effects” due to hydrogen bonding between the carboxyl and epoxido/diol groups, which would be favored either by structural proximity (5,6-EET and 5,6-DHET) or by arching of the carbon skeleton (14,15-EET and 14,15-DHET).28 In any case, (28) Corey, E. J.; Iguchi, S.; Albright, J. O.; De, B. Tetrahedron Lett. 1983, 24, 37-40.

Figure 3. EET and DHET stability in 5 mM phosphate sample vehicle (pH 8.0) at 25 °C. Fresh mixtures of (A) EETs (20-33 ng/µL) or (B) DHETs (14-17 ng/µL) were added to the CE carousel and immediately analyzed as described in Figure 2. After repeated injections and analyses over a 9.5-h period, the original EET mixture (stored for 16 h or 5 days at 25 °C in a sealed Teflon vial) was similarly analyzed. DHET profiles were similarly assessed for up to 17 h.

the use of migration times and UV spectra helped differentiate members of the EET and DHET families. Interestingly, an additional minor absorption band centered at ∼235 nm was observed during the CE of EETs and DHETs (Figure 2, panels C and D). Although less prominent for EETs than for DHETs, the 235-nm band was completely absent during reversed-phase HPLC employing similar CH3CN/H2O-phosphate mixtures. The presence of a 235-nm band raises the possibility that oxirane rings and diols, variably positioned in the hydrophobic cavity, differentially interact with cyclodextrin heteroatoms. The 235-nm absorption band is likely due to inclusion compounds with the cyclodextrin, which are known to alter the spectra of the guest molecules;29 optical studies on inclusion compounds indicate that spectral changes do occur.30 Thus, the concomitant migration of EETs or DHETs with trimethyl-β-cyclodextrin resulted in an additional absorption peak at ∼235 nm that was absent during reversed-phase HPLC. Stability in the Sample Vehicle. In anticipation of generating calibration curves, we tested the chemical stability of EET and DHET standards positioned in the CE sample tray (25 °C). When suspended in an acidic medium, EETs are rapidly hydrolyzed to DHETs.9 Yet, in 5 mM sodium phosphate (pH 8.0), the 8,9-, 11,12-, and 14,15-EETs produced constant peak area ratios for 9.5 h (Figure 3A). When normalized to the 11,12-EET peak area, the 8,9-EET and 14,15-EET peak areas remained at 94.9 ( 0.7 (n ) 7) and 90.4 ( 0.6% relative standard deviation (RSD) over the 9.5-h period. Moreover, these ratios matched those of regioisomers stored at 25 °C in sealed Teflon vials for up to 5 days (Figure 3A, top two traces). However, with repeated injections over the 9.5-h period, the absolute areas generated by these compounds increased by as much as 61%. In contrast, samples left in sealed Teflon vials for 16 h or 5 days generated peak areas that matched those occurring immediately after suspension in the vehicle (29) Klein, C. T.; Viernstein, H.; Wolschann, P. Minerva Biotecnol. 2000, 12, 287-292. (30) Werner, T. C.; Forrestall, K. J.; McIntosh, S. L.; Pitha, J. Appl. Spectrosc. 2000, 54, 560-564.

(Figure 3A, top two traces). Thus, samples were concentrated due to evaporative losses. Perhaps more importantly, standards of 8,9-, 11,12-, and 14,15-EETs were stable enough to generate response curves for quantitation studies (see below). Unlike other EET regioisomers, 5,6-EET was chemically unstable at pH 8.0 and was hydrolyzed to a diol (compound X) within 1.5 h. Yet, the 5,6-DHET product increased in amount even when 5,6-EET was no longer detectable after 7.5-h suspension in the vehicle (Figure 3A). Suspecting that a δ-lactone of 5,6-EET was the direct precursor of 5,6-DHET,13 we added a 5,6-EET δ-lactone standard to the 5,6-EET sample and saw no additional peaks; apparently, the neutral 5,6-EET lactone was lost in the large refractive index distortions coinciding with the EOF. Thus, a vehicular pH of 8.0 favored a nucleophilic attack of C5 by the carboxyl anionic oxygen; moreover, formation of a stable sixmembered ring reinforced the rapid conversion of 5,6-EET to a δ-lactone,13 which was more slowly hydrolyzed to 5,6-DHET (weak nucleophilic attack at C5 by H2O). As with other DHETs, the newly formed 5,6-DHET remained intact at pH 8.0 because nucleophilic attack at C1 by diol oxygens is unlikely. Moreover, the predicted stability of the 5,6-DHET regioisomer was confirmed by a timecourse study (Figure 3B). When normalized to 8,9-DHET, the 5,6-DHET, 11,12-DHET, and 14,15-DHET peak areas were 89.5 ( 3.7, 80.6 ( 3.1, and 82.8 ( 3.5% RSD (n ) 6) over a 17-h period. Thus, CE was useful in assessing the chemical stability of EETs and DHETs. fast

slow

5,6-EET 98 6-hydroxy-5-lactone 98 5,6-DHET (stable) Reproducibility of Retention Times. Precise retention times are important for identifying endogenous arachidonate metabolites. The retention times of the three chemically stable EET regioisomers and arachidonate (52.96 ( 0.84 min; 1.6% RSD (n ) 6)) tended to decrease over time, probably because of evaporative losses of CH3CN from the running buffer (Figure 4A, traces A-F). However, the retention times of EETs relative to that of arachidonate were highly reproducible; the mean values and RSD (n ) Analytical Chemistry, Vol. 74, No. 22, November 15, 2002

5863

Figure 4. Electropherograms for quantitating (A) EET and (B) DHET regioisomers. Mixtures of [1-14C]-EET (39-48 ng/µL) or [1-14C]-DHET (7.8-8.1 ng/µL) standards were diluted up to 12-16-fold before arachidonate (10 or 20 ng/µL) was added as internal standard. The EET and DHET mixtures were subjected to electrophoresis as described in Figure 2. Tracings are labeled in order of injection and amount of dilution.

6) were 1.051 ( 0.2 (8,9-EET), 1.0739 ( 0.3 (11,12-EET), and 1.096 ( 0.3% (14,15-EET). The relative retention times of DHETs showed comparable improvements in migration time precision (Figure 4B, traces A-E); the mean values and RSD (n ) 5) were 1.0318 ( 0.07 (8,9-DHET), 1.0699 ( 0.14 (11,12-DHET), and 1.0886 ( 0.11% (14,15-DHET). Thus, compared to absolute values, the relative retention times had better than a 5-fold increase in precision and dramatically improved the ability to identify EET and DHET regioisomers by CE. Linearity of Calibration Plots. The three chemically stable EET regioisomers were baseline resolved (R > 1.32) over a 16fold range in concentration (Figure 4A). Labeled “1.00” in Figure 4A, a mixture of 8,9-EET (39.3 ng/µL), 11,12-EET (48.0 ng/µL), and 14,15-EET (36.9 ng/µL) was subjected to dilutions before arachidonate (20 ng/µL) was added. After EETs were resolved by CE, the peak area absorptions at 194 ( 2 nm were integrated and divided by the EET concentrations injected; using this technique, the regioisomers were shown to have similar absorptivity. Peak area responses (relative to arachidonate) were 0.0431 × ng of 8,9-EET/µL - 0.0173, 0.0390 × ng of 11,12-EET/µL 0.0362, and 0.0414 × ng of 14,15-EET/µL - 0.0308. Each calibration plot was linear (r2 ) 0.999) down to an injected mass of ∼40 pg and a signal-to-noise ratio of ∼50:1. Thus, underivatized EETs were readily quantitated in the low-picogram range, which is at least 10 000-fold lower than achieved by HPLC.13 The DHETs were similarly baseline resolved (R > 1.36) over a 12-fold range in concentration (Figure 4B). Labeled “1.00” in Figure 4B, a mixture of 8,9-DHET (8.1 ng/µL), 11,12-DHET (7.8 ng/µL), and 14,15-DHET (7.8 ng/µL) was diluted, and the internal standard arachidonate (10 ng/µL) was added. After mixtures at various concentrations were injected and resolved, peak area measurements indicated that the regioisomers had similar absorptivity. The integrated peak areas (relative to arachidonate) were 0.1222 × ng of 8,9-DHET/µL - 0.0005, 0.1077 × ng of 11,12-DHET/µL + 0.0098, and 0.1194 × ng of 14,15-DHET/µL + 0.0181. Perhaps more importantly, each DHET calibration plot was linear (r2 ) 0.996) down to 12-pg injections, which generated a signal-to-noise ratio of ∼22:1. Thus, underivatized DHETs and EETs in the low-picogram range were readily quantitated. 5864 Analytical Chemistry, Vol. 74, No. 22, November 15, 2002

Figure 5. Reversed-phase HPLC chromatograms of a saponified phospholipid extract from murine liver (lower trace) and of a reference standard mixture (upper trace). Liver lipids were extractively isolated, and the total phospholipids were isolated by column chromatography for saponification. After the methyl ester of 5,6-DHET (5,6-DHETME) was added as a retention time marker, the saponified phospholipids were injected onto an octadecasilyl column. EETs and DHETs were resolved in CH3CN/H2O (pH 2.3) (46:54, v/v), flowing at 1.5 mL/min and 1250 psig. Absorption at 194 nm was monitored. Roman numerals and heavy bars denote fractions with corresponding collection times.

Identification and Quantitation of EETs from Murine Liver. During reversed-phase HPLC, retention times and UV spectra also help to identify EETs and DHETs. However, except for 14,15-EET and 14,15-DHET, the chemically stable EETs and DHETs are not baseline-resolved (Figure 5, upper trace); moreover, the labile 5,6-EET and its δ-lactone comigrate with 8,9-EET and 5,6-DHET-ME, respectively (data not shown). Finally, absorptions from complex sample matrixes may interfere with the acquisition of representative UV spectra (Figure 5, lower trace). Interestingly, while no EET or DHET regioisomer absorbed at a wavelength greater than 222 nm, isomers that possessed centrally positioned epoxides and diols had decreased absorption in the 195-218-nm range (data not shown). Yet, despite regiospecific differences in UV spectra, the poor resolution of regioisomers and matrix interferences restrict the use of reversed-phase HPLC to identifying EETs and DHETs to simple samples such as tissue cultures.9

Figure 6. (A) Electropherograms (n ) 4) and (B) UV spectra of EETs isolated from murine liver phospholipids. The total EETs isolated from saponified phosholipids by HPLC (Figure 5) were subjected to CE as described in Figure 2.

Compared to reversed-phase HPLC alone, the combination of reversed-phase HPLC and CE is a much more powerful technique for identifying and quantitating EETs. As an illustration, HPLC and CE were used to measure the EET composition of mouse liver. More than 80% of the EETs present in mammalian liver occur esterified to cellular phospholipids.2 Therefore, lipids from mouse liver were first extractively isolated, separated into neutral and phospholipid fractions by column chromatography, and saponified.9 The saponified phospholipids, including the EETs and DHETs, were separated by HPLC using an octadecasilyl column (Figure 5, lower trace). As mentioned earlier (see Experimental Section), little EET was lost due to hydrolysis to DHET during sample processing; therefore, only fractions III and IVsrepresenting the total EETsswere collected and pooled for CE analysis. EET regioisomers isolated by HPLC from hepatic phospholipids were baseline-resolved and quantitated by CE (Figure 6). EETs were identified based on matches with standards in (A) retention times and (B) ultraviolet spectra. The UV spectra differentiated 14,15-EET from the other two regioisomers. Once baseline-resolved, EETs were readily quantitated (n ) 4) by integrating absorptions at 194 nm: 2997 ( 6.8 (14,15-EET), 3884 ( 6.3 (11,12-EET), and 4139 ( 6.8% RSD (8,9-EET). Based on linear regression curves generated with EET standards, the peak areas represented 4.75 (14,15-EET), 5.92 (11,12-EET), and 4.98 ng/µL (8,9-EET). Derived from EET specific radioactivities

and liquid scintillation assays, the internal standard contributions were 1.83 (14,15-EET), 1.84 (11,12-EET), and 1.68 ng/µL (8,9-EET). Thus, the ratio of liver EET to internal standard was 2.60 (14,15-EET), 3.22 (11,12-EET), and 2.96 (8,9-EET). Based on the amount of internal standard originally added, the EET concentration in mouse liver phospholipids was (µg/g of wet weight liver): 3.6 of 14,15-EET, 4.5 of 11,12-EET, and 3.8 of 8,9EET. The summed total for EETs was 11.9 µg, which was 10-fold more than the 1.0 µg reported for rat liver.14 Interestingly, little EET regioselectivity was detected in murine liver. In contrast, rat liver showed significant regioselectivity, even without phenobarbital stimulation.14 Determining whether the disparities reflect dietary, species, strain, or animal deserves further investigation. CONCLUSIONS EET and DHET regioisomers were completely resolved by CE using trimethyl-β-cyclodextrin; the migration orders reflected the EET and DHET hydrophobic properties. Identification of these eicosanoids was aided by regiospecific differences in UV spectra and characteristic migration times. Monitoring absorption at 194 nm permitted sensitive measurements of endogenous EET levels in murine liver. These CE techniques may also prove applicable to trace analysis of other polyunsaturated fatty acid products lacking conjugated double bonds: the isoprostanes and neuroprostanes. ACKNOWLEDGMENT Supported by the NIH RO1-HL-56670-02 (M.V.R.), PO1-HL49264 (M.V.R.), VA Merit Review Program (M.V.R.), and the American Heart Association, 96012380 (M.V.R.). Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under contract DE-AC04-94AL85000. The authors are deeply grateful to Professor Jonathan S. Dordick (Rensselaer Polytechnical Institute, Troy, NY) for graciously providing a 5500 P/ACE to perform these studies.

Received for review July 2, 2002. Accepted September 25, 2002. AC025909+

Analytical Chemistry, Vol. 74, No. 22, November 15, 2002

5865