Anal. Chem. 1997, 69, 2143-2146
Simultaneous Solid Phase Extraction, Derivatization, and Gas Chromatographic Mass Spectrometric Quantification of Thromboxane and Prostacyclin Metabolites, Prostaglandins, and Isoprostanes in Urine Joachim Wu 1 bert, Elke Reder, Andrea Kaser, Peter C. Weber, and Reinhard L. Lorenz*
Institute for Prophylaxis and Epidemiology of Cardiovascular Disease, University of Munich, Pettenkoferstrasse 9, 80336 Munich, Germany
The current analytical methods for the various prostanoids require a separate and extended sample workup, derivatization, and gas chromatographic/mass spectrometric detection of each compound. Therefore, we developed and validated a rapid method for the common purification, derivatization, and GC/MS determination of 11-dehydrothromboxane B2, 2,3-dinor-6-keto-PGF1a, PGF2a, PGE2, PGD2, and isoprostanes in urine. A single reversed-phase solid-phase extraction step and modified reaction conditions yielded excellent sample purification at high recoveries and efficient derivatization for all compounds in one vial. The method allows, for the first time, the simultaneous quantification of these index metabolites of systemic thromboxane and prostacyclin synthesis, renal prostaglandin formation, and nonenzymatic in vivo lipid peroxidation in a single GC/MS run with high sensitivity and precision. Prostanoids, a diverse group of lipid mediators with potent, multiple biological actions, are synthesized from arachidonic acid in a highly regulated manner and a cell-specific pattern. They are involved in physiologic functions like hemostasis, thromboresistence, and vascular tone, contribute to pathologic processes like inflammation, atherosclerosis, and cell proliferation, and are the target of many therapeutic interventions. Besides the tightly controlled enzymatic conversion to prostanoids, arachidonic acid may also be transformed to structurally related isoprostanes by nonenzymatic in vivo lipid peroxidation, a key step in atherogenesis. To understand the role of the functionally diverse arachidonic acid products in these conditions requires the sensitive and specific detection of nanomolar levels of structurally closely related metabolites in complex biological matrices. GC/MS methods are best suited and have been described for the index metabolites of prostacyclin1-5 and thromboxane6-12 formation, classical prosta(1) Falardeau, P.; Oates, J. A.; Brash, A. R. Anal. Biochem. 1981, 115, 359367. (2) Vesterqvist, O.; Gre´en, K. Prostaglandins 1984, 28, 139-154. (3) Fischer, C.; Meese, C. O. Biochem. Mass. Spectrom. 1985, 12, 399-404. (4) Ferretti, A.; Flanagan, V. P. J. Chromatogr. 1993, 622, 109-115. (5) Daniel, V. C.; Minton, T. A.; Brown, N. J.; Nadeau, J. H.; Morrow, J. D. J. Chromatogr. 1994, 653, 117-122. (6) Vesterqvist, O.; Gre´en, K.; Lincoln, F. H.; Sebek, O. K. Thromb. Res. 1983, 33, 39-49. S0003-2700(96)01143-2 CCC: $14.00
© 1997 American Chemical Society
glandins,13 and isoprostanes.14-16 These methods usually require a separate, extensive sample workup for a single prostaglandin metabolite. A few methods have been tried to accomplish the analysis of more than one prostanoid or their metabolites,17-21 but these usually include time-consuming chromatographic steps requiring the splitting of the sample or the use of tandem mass spectrometry (GC/MS/MS). The aim of this study was to speed up and unify in one vial the sample purification and derivatization for the index metabolites from different arachidonic acid pathways and quantify them in a single GC/MS run. EXPERIMENTAL SECTION Instrumentation. A Vac Elut SPS 24 workstation (ICT, Bad Homburg, Germany), a TSQ 70 mass spectrometer (Finnigan MAT, Bremen, Germany), a Finnigan MAT A200S autosampler, a Varian 3400 gas chromatograph (Varian, Palo Alto, CA) equipped with a DB5MS capillary column (30 m × 0.25 mm i.d. × 0.25 µm film thickness, J&W Scientific, Folsom, CA), and a Gerstel CIS3 cold injection system (Gerstel, Mu¨hlheim, Germany), were used. Reagents. Standard 11-dehydrothromboxane B2, [2H4]11dehydrothromboxane B2, and [2H4]prostaglandins D2, E2, and F2a (7) Lawson, J. A.; Brash, A. R.; Doran, J.; FitzGerald, G. A. Anal. Biochem. 1985, 150, 463-470. (8) Chiabrando, C.; Benigni, A.; Piccinelli, A.; Carminati, C.; Cozzi, E.; Remuzzi, G.; Fanelli, R. Anal. Biochem. 1987, 163, 255-262. (9) Schweer, H.; Meese, C. O.; Fu ¨ rst, O.; Ku ¨ hl, P. G.; Seyberth, H. W. Anal. Biochem. 1987, 164, 156-163. (10) Uedelhoven, W. M.; Meese, C. O.; Weber, P. C. J. Chromatogr. 1989, 497, 1-16. (11) Lorenz, R.; Helmer, P.; Uedelhoven, W.; Zimmer, B.; Weber, P. C. Prostaglandins 1989, 38, 157-170. (12) Ferretti, A.; Flanagan V. P.; Maida E. J. Prostaglandins, Leukotrienes Essential Fatty Acids 1992, 46, 271-275. (13) Rosenfeld, J. M.; Moharir, Y.; Hill, R. Anal. Chem. 1991, 63, 1536-1541. (14) Morrow, J. D.; Harris, T. M.; Roberts, L. J., II. Anal. Biochem. 1990, 184, 1-10. (15) Pratico, D.; Lawson, J. A.; FitzGerald, G. A. J. Biol. Chem. 1995, 270, 98009808. (16) Nourooz-Zadeh, J.; Gopaul, N. K.; Barrow, S.; Mallet, A. I.; A ¨ nggard, E. E. J. Chromatogr. 1995, 667, 199-208. (17) Schweer, H.; Kammer, J.; Seyberth, H. W. J. Chromatogr. 1985, 338, 273280. (18) Chiabrando, C.; Pinciroli, V.; Campoleoni, A.; Benigni, A.; Piccinelli, A.; Fanelli, R. J. Chromatogr. 1989, 495, 1-11. (19) Blair, I. A.; Prakash, C.; Phillips, M. A.; Dougherty, R. M.; Iacono, J. M. Am. J. Clin. Nutr. 1993, 57, 154-160. (20) Schweer, H.; Watzer, B.; Seyberth, H. W. J. Chromatogr. B 1994, 652, 221227. (21) Fauler, J.; Tsikas, D.; Mayatepek, E.; Keppler, D.; Fro ¨lich, J. C. Pediatr. Res. 1994, 36, 449-455.
Analytical Chemistry, Vol. 69, No. 11, June 1, 1997 2143
were purchased from Cayman Chemical Co. (Ann Arbor, MI), prostaglandins E2, D2, and F2a from Cascade Biochem Limited (Reading, U.K.), and 2,3-dinor-6-keto-prostaglandin F1a and [2H3]2,3dinor-6-ketoprostaglandin F1a from Biomol (Hamburg, Germany). Methoxyamine hydrochloride was from Regis Technologies Inc. (Morton Grove, IL), pentafluorobenzyl bromide and N,N-diisopropylethylamine from Sigma-Aldrich (Deisenhofen, Germany), and Chromabond C18ec and Chromabond SiOH cartridges from Macherey-Nagel (Du¨ren, Germany). Procedures. Urine samples were always collected on butylated hydroxytoluene (BHT, final concentration >0.01%), immediately frozen, and stored at -20 °C until analysis. Usually, 3.0 mL aliquots of urine were spiked with 2.0 ng of [2H4]11dehydrothromboxane B2, 0.2 ng of [2H3]2,3-dinor-6-keto-prostaglandin F1a, and 2.0 ng of [2H4]prostaglandins E2, D2, and F2a as internal standards in ethanolic solution. To this was added 0.5 mL of 2-propanol. As tetradeuterated standards of isoprostanes are not commercially available at present, [2H4]prostaglandin F2a served as a substitute internal standard for its stereoisomeric isoprostanes. The sample was brought to pH 3 with 1 N formic acid and equilibrated at room temperature for 30 min. A Chromabond C18ec cartridge (octadecylsilane, endcapped, 1000 mg) was preconditioned with 12 mL of methanol, 6 mL of H2O, and 6 mL of 0.05 N formic acid, loaded with the sample, and washed with 8 mL of 1 N formic acid/acetonitrile (3:1 v/v) and 4 mL of water. The cartridge was dried by blowing nitrogen through it for 15 min and then eluted with 4 mL of methanol into a derivatization vial. The eluate was dried under nitrogen, and the residue was rinsed to the bottom of the tube with 500 µL of tert-butyl methyl ether and dried with nitrogen. The sample was redissolved in 100 µL of ethyl acetate/concentrated formic acid (9:1 v/v), activated at 45 °C for 30 min, and dried under nitrogen. Derivatization. The sample was then reacted with 50 µL of 0.5 g of methoxyamine hydrochloride in 9.5 mL of N,N-dimethylformamide at 45 °C for 30 min and dried under nitrogen. The sample was redissolved in 50 µL of acetonitrile, 20 µL of N,Ndiisopropylethylamine, and 20 µL of 1 g of pentafluorobenzyl bromide in 3 mL of acetonitrile, allowed to react at 45 °C for 25 min, and dried under nitrogen. The sample was taken up again in 50 µL of bis(trimethylsilyl)trifluoroacetamide (BSTFA), incubated at 45 °C for 2 h, and left at room temperature overnight. The excess solvent was removed under nitrogen, 1 mL of pentane was added, and the derivatization tube was vortex mixed for 2 s. The supernatant was transferred to a new vial, dried under nitrogen, and resolved in 1 mL of hexane. A Chromabond SiOH cartridge (100 mg) was preconditioned with 2 mL of dichloromethane, 2 mL of dichloromethane/hexane (50:50 v/v), and 2 mL of hexane. The sample was then applied, washed with 2 mL of hexane, 2 mL of dichloromethane/hexane (50:50 v/v), and 2 mL of dichloromethane, eluted with 2 mL of dichloromethane/ methanol (100:1, v/v), and dried under nitrogen. After repeat silylation with 50 µL of BSTFA at 45 °C for 2 h, the completely derivatized eicosanoid preparation was dried under nitrogen, taken up in 50 µL of heptane, sealed in vials, and stored at -20 °C until GC/MS/NICI analysis. The procedure is summarized in Table 1. GC/MS/NICI Analysis. A 5 µL aliquot of the sample was injected using the septumless Gerstel CIS3 cold injection system kept at 60 °C for 6 s (solvent split) and then heated to 300 °C at 10 °C/s (splitless). The DB5MS capillary column was directly 2144 Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
Table 1. Summary of the Purification and Derivatization Procedure for 11-Dehydrothromboxane B2, 2,3-Dinor-6-ketoprostaglandin F1a, Prostaglandins, and Isoprostanes 1. 2. 3. 4. 5. 6.
7. 8. 9.
10. 11. 12. 13. 14.
Sample Preparation spike 3 mL of urine with deuterated standards add 2-propanol, acidify to pH 3, equilibrate 30 min Solid-Phase Extraction on C18ec precondition cartridge, load sample wash with 1 N formic acid/CH3CN (3:1) and water and dry cartridge elute with 4 mL of methanol, dry sample under N2 redissolve in ethyl acetate/concentrated CHOOH for 30 min and dry again Derivatization react with methoxyamine hydrochloride in DMF for 30 min and dry sample react with (PFB)Br and diIPEA in CH3CN for 25 min and dry sample add BSTFA, heat for 2 h at 45 °C, and leave at room temperature overnight Removal of Reactants dry sample, transfer in pentane to new vial, dry, take up in 1 mL of hexane precondition SiOH cartridge, apply sample wash with hexane, CH2Cl2/hexane, CH2Cl2 elute sample with 2 mL of CH2Cl2/MeOH (100:1) and dry repeat silylation with BSTFA for 2 h at 45 °C, dry, store in heptane
connected to the ion source. The initial column temperature of 100 °C was maintained for 2 min. The column was then heated to 250 °C in 7 min, then to 300 °C at 2 °C/min and kept there for 10 min. The transfer line and ion source were maintained at 300 and 150 °C, respectively. Helium was used as the carrier gas at a flow rate of 1 mL/min. The mass spectrometer was operated in the negative ion, chemical ionization mode, utilizing methane as the reagent gas (manifold pressure 10-5 Torr). The electron energy was 70 eV, and the filament current was 0.2 mA. For simultaneous determination of 11-dehydrothromboxane B2, 2,3-dinor-6-ketoprostaglandin F1a, prostaglandins D2, E2, F2a, and isoprostanes, the range from m/z 510 to 590 was scanned in 500 ms from 20 to 35 min after start. The masses used for quantification were m/z 586 (589) for 2,3-dinor-6-ketoprostaglandin F1a and its trideuterated analogue, respectively, m/z 569 (573) for prostaglandin F2a, its tetradeuterated analogue, and its stereoisomeric isoprostanes 8-epi-, 9-β-, and 11-β-PGF2a, m/z 524 (528) for prostaglandins D2 and E2, and m/z 511 (515) for 11-dehydrothromboxane B2. The stereochemistry, labels, and masses of the completely derivatized compounds are shown in Figure 1. The retention times of the various prostanoids were determined with authentic standards and confirmed by full NICI mass spectra. The derivatives of 2,3-dinor-6-ketoprostaglandin F1a, PGE2, and PGD2 elute as double peaks due to the formation of syn-anti isomers during the conversion of the keto group to the methoxylamine derivatives. The areas of both peaks were added for quantification of these compounds. Typical retention times were 22:57 and 23:18 min for 2,3-dinor-6-ketoprostaglandin F1a, 24:02 for 9-β-, 24:08 for 8-epi-, and 24:39 for 11-β-PGF2a, 24:59 for PGF2a, 25:28 and 26:07 for PGD2, 25:33 and 26:32 for PGE2, and 31:44 for 11-dehydrothromboxane B2. Tetra- and trideuterated analogues eluted usually 3-5 s before the unlabeled compounds. RESULTS AND DISCUSSION During method development, several steps were found to be essential for optimal results. Prior to extraction from urine,
Figure 1. Stereochemical configuration of the methoxime (dNOCH3)- trimethylsilyl ether (-OTMS)-pentafluorobenzyl ester (-PFB) derivatives of the prostanoids analyzed, location of deuterium label, and masses of the M - PFB- fragments of the deuterated internal standards and the endogenous compounds scanned for quantification. From top to bottom: (I) 2,3-dinor-6-keto-PGF1a, the major metabolite of systemic prostacyclin (PGI2-M), (II) 11-dehydrothromboxane B2, the major metabolite of systemic thromboxane, (III) prostaglandin F2a and its stereoisomeric counterpart 8-epi-prostaglandin F2a, index product of nonenzymatic arachidonic acid peroxidation and (IV) prostaglandin E2 and its regioisomeric counterpart, prostaglandin D2.
endogenous and deuterated prostanoids had to be equilibrated and the δ-lactone ring of 11-dehydrothromboxane B2 closed by acidifying the sample for 30 min at pH 3. After application of the samples to the preconditioned C18 cartridges and washing with formic acid/acetonitrile and water, it was critical to completely dry the cartridges with nitrogen prior to elution of the sample. Before derivatization, acidic activation of the sample for 30 min at 45 °C in ethyl acetate/formic acid was found to be essential. Derivatization was then performed, modified according to reaction conditions previously described to yield the methoxime-trimethylsilyl ether-pentafluorobenzyl ester derivatives.17 After PFB esterification, the transfer to a new vial and the modified solvents and exact volumes used for the final silica extraction were critical for the purity of the sample and quantitative recovery of the derivatives. The conditions for GC and SIM mass spectrometry were optimized with pure standards to yield well-separated, dominating molecular peaks (M - PFB) of the compounds of interest and their deuterated internal standards. With the optimized extraction and derivatization procedure described above, sufficiently clean baselines at the m/z tracings of the endogenous compounds and
Figure 2. Representative SIM tracings of three of the prostanoids analysed. Top panel: 2,3-dinor-6-keto-PGF1a (m/z 586) and the trideuterated internal standard (m/z 589) elute as double peaks of the racemic methoxime derivatives. Second panel: 11-dehydro-TXB2 (m/z 511) and its tetradeuterated analogue (m/z 515). Bottom panel: For the stereoisomers 9-β-PGF2a, 8-epi-PGF2a, 11-β-PGF2a, and PGF2a (indicated by arrows from left to right at m/z 569), tetradeuterated PGF2a (m/z 573) served as internal standard.
the deuterated standards and well-defined peaks were obtained. An acceptable signal-to-noise ratio was maintained down to the low picogram range of endogenous concentrations, even in the SIM mode, avoiding the loss of sensitivity inherent to the GC/ MS/MS mode. The method does, therefore, not necessitate a triple-stage GC/MS/MS equipment. Representative original tracings of a urine sample with low concentrations of 11-dehydrothromboxane B2, 2,3-dinor-6-keto-PGF1a, PGF2a, and isoprostanes are shown in Figure 2. The deuterated internal standard peaks of 2,3-dinor-6-ketoPGF1a (m/z 589), 11-dehydrothromboxane B2 (m/z 515), and prostaglandin PGF2a (m/z 573) and prostaglandins represent 20, 200, and 200 pg on colum, respectively. The peaks of endogenous 2,3-dinor-6-keto-PGF1a (m/z 586), 11-dehydrothromboxane B2 (m/z 511), 9-β-, 8-epi-, and 11-β-PGF2a and prostaglandin PGF2a (m/z 569) correspond to 9, 60, 210, 120, 200, and 680 pg of each prostanoid on column, respectively. If the limit of detection was set to 3 times above background noise, it was found to be 11, 23, and 36 pg/mL for 2,3-dinor-6-keto-PGF1a, 11-dehydrothromboxane B2, and prostaglandin PGF2a. The peaks detectable on the m/z 569 tracing at 24:02, 24:08, and 24:39 min, as indicated by arrows, were shown to represent the 9-β-, 8-epi-, and 11-β-stereoisomers of PGF2a from nonenzymatic in vivo arachidonic acid peroxidation14 Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
2145
by exact coelution from GC with authentic standards and characteristic molecular spectra. As no deuterated internal standards of isoprostanes were available, only 8-epi-PGF2a (exerting potent biologic activities)22 and PGF2a were routinely integrated using tetradeuterated PGF2a as substitute internal standard. This approach has previously been validated.16 Meanwhile, we prefer to use an OV-1701 (Ohio Valley) column, which at the same GC conditions completely separates the diverse isoprostane isomeres. In addition to PGF2a and F2 isoprostanes, also PGE2 and PGD2 can effectively be purified, derivatized, and quantified by this procedure. Whereas the ample PGF2a and PGE2 excretion reflects renal synthesis of these prostaglandins, PGD2 is, however, not a major prostanoid excreted unmetabolized in urine and was not routinely quantified. The overall recovery of the extraction procedure as estimated by comparison to direct derivatization of pure standards was over 80%, which shows enhanced sensitivity compared to methods requiring repeated TLC or HPLC purification steps, considerably reducing overall recovery.19 When six aliquots of a spot urine sample were analyzed in parallel, the intraassay coefficient of variance was below 7% for 2,3-dinor-6-keto-PGF1a, 11-dehydro-TXB2, and PGF2a and below 15% for 8-epi-PGF2a, PGD2, and PGE2. Precision was determined by carrying aliquots of a urine sample through the whole procedure on six different days. The interassay coefficient of variance was below 12% for 2,3-dinor-6-keto-PGF1a, 11-dehydro-TXB2, and PGF2a and below 18% for 8-epi-PGF2a, PGD2, and PGE2. As a blank urine for prostanoids was not available, samples from six different subjects were analyzed with and without addition of 200 or 2000 pg of standard 2,3-dinor-6-keto-PGF1a, 11-dehydroTXB2 and PGF2a, 8-epi-PGF2a, PGD2, and PGE2 to assess accuracy. The percent recovery of added standards was calculated as the difference in measured concentrations in the spiked and unspiked samples. Accuracy lay between 92 and 106% for 2,3-dinor-6-ketoPGF1a, 11-dehydro-TXB2 and PGF2a, 8-epi-PGF2a, and PGE2. The linearity of recovery was demonstrated by the assay in quadruplicate of aliquots of a biological sample spiked with increasing amounts of standard 2,3-dinor-6-keto-PGF1a, 11-dehydroTXB2, PGF2a, 8-epi-PGF2a, PGD2, and PGE2 ranging from 10 to 400 pg for 2,3-dinor-6-keto-PGF1a and from 100 to 4000 pg/sample for the other compounds. The regression curves of measured over added amounts of 2,3-dinor-6-keto-PGF1a, 11-dehydro-TXB2, PGF2a, and 8-epi-PGF2a are plotted in Figure 3. Linearity of recovery was evident for every compound over the whole range from low physiologic levels to elevated levels found in pathologic conditions. In five apparently healthy subjects, basal excretion of all prostanoids studied lay in the range previously described. Pharmacologic inhibition of cyclooxygenase by 325 mg of acetylsalicylic acid in five subjects suppressed as expected 11-dehydro-TXB2 excretion from 1150 ( 400 to 280 ( 100 ng/g creatinin after 24 h, whereas 8-epi-PGF2a excretion was not impaired (280 ( 80 vs 330 ( 120 pg/g creatinin). In conclusion, a very rapid method based on a simple reversedphase solid-phase extraction for the combined single-stage GC/ MS determination of 11-dehydro-TXB2, 2,3-dinor-6-keto-PGF1a, PGF2a, and isoprostanes was developed and validated. One or (22) Morrow, J. D.; Hill, K. E.; Burk, R. F.; Nammour, T. M.; Badr, K. F.; Roberts, L. J. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 9383-9387.
2146 Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
Figure 3. Linear recoveries of standards added in increasing concentrations (pg/3 mL) to aliquots from a pooled biological sample. Top panel: 2,3-dinor-6-keto-PGF1a. Second panel: 11-dehydro-TXB2. Bottom panel: PGF2a and its nonenzymatic stereoisomer 8-epiPGF2a. Means ( SEM, n ) 4.
more TLC or HPLC steps or a triple-stage GC/MS/MS equipment can be spared compared to published methods. Twenty-four samples can be processed in parallel. Thus, metabolites of five different enzymatic pathways and of nonenzymatic in vivo arachidonic acid peroxidation can be simultaneously monitored in large sample numbers. This will be helpful in elucidating the pathophysiologic and prognostic long-term bearing of minor elevations in biological active arachidonic acid oxidation products in less dramatic conditions than acute vascular catastrophies. ACKNOWLEDGMENT The expert technical assistance of I. Papperitz is gratefully acknowledged. This work was supported by Deutsche Forschungsgemeinschaft (We 681-7). Received for review November 12, 1996. Accepted March 11, 1997.X AC9611430 X
Abstract published in Advance ACS Abstracts, April 15, 1997.
