Anal. Chem. 2007, 79, 7087-7096

Comprehensive Analysis of Vitamin E Constituents in Human Plasma by Liquid Chromatography-Mass Spectrometry Korne´l Nagy,*,† Marie-Claude Courtet-Compondu,† Birgit Holst,‡ and Martin Kussmann†

Functional Genomics Group and Nutrient Bioavailability Group, BioAnalytical Science Department, Nestle´ Research Centre, Nestec Ltd., Lausanne, Switzerland

The present paper describes the development and validation of a normal-phase liquid chromatography-mass spectrometry (NP-HPLC-MS) method for the screening and quantification of vitamin E constituents in human plasma and food matrixes. Liquid-liquid extraction combined with isotope dilution was applied to extract the lipophilic target analytes. Baseline separation of r-tocopherylacetate, r-tocopherol, r-tocotrienol, r-tocopherylquinone, β-tocopherol, γ-tocopherol, β-tocotrienol, γ-tocotrienol, δ-tocopherol, and δ-tocotrienol was achieved utilizing a normal-phase amine column operated with n-hexane and 1,4-dioxane as solvents. Detection was achieved by positive-ion atmospheric-pressure chemical ionization (APCI). Key features of the method are lower limits of detection, 3-51 nmoles/L; lower limits of quantification, 8-168 nmoles/L; linearity coefficients, 0.9778-0.9989; linear ranges, 0.01-29 µmol/L; recoveries, 53-92%; accuracies, 99-103%; intraday precisions, 2-17%; interday precisions, 5-18%; and suppression values, 0-29%. Fragmentation of tocopherols was studied by tandem mass spectrometry, and a fragmentation scheme for tocotrienols/tocopherols is postulated. Neutral-loss and precursor-ion scan experiments were performed for targeted discovery of oxidation products of tocopherols in human blood and fish oil, the latter being an important food component. The presented data suggest that this method will help to expand the number of quantified/discovered vitamin E constituents detected in food products and analyzed during human/animal trials in order to give a more comprehensive picture to nutritionists about the fate of vitamin E. Vitamin E is an essential micronutrient. The term “vitamin E” encompasses eight naturally occurring homologues, i.e., a group of lipid-soluble, chain-breaking antioxidants that include the wellknown tocopherols and tocotrienols, collectively also called vitamers; see compounds 1-8 in Figure 1. The basic R-, β-, γ-, and δ-vitamers differ in the number and position of methyl groups * To whom correspondence should be addressed. Korne´l Nagy, Nestle´ Research Centre, Vers-Chez-les-Blanc, 1000 Lausanne, Switzerland. E-mail: [email protected]. Phone: + 41 21 785 8290. Fax: + 41 21 785 9486. † Functional Genomics Group. ‡ Nutrient Bioavailability Group. 10.1021/ac0708689 CCC: $37.00 Published on Web 08/16/2007

© 2007 American Chemical Society

on the chroman ring. Tocotrienols differ from tocopherols by having three double bonds in the isoprenoid side chain. More detailed chemical structures and the nomenclature of vitamin E constituents are described elsewhere.1,2 Vitamin E (R-tocopherol in particular) is the major and most potent lipid-soluble antioxidant in vivo.3-5 It functions as the major radical scavenging antioxidant in lipoproteins and efficiently interrupts the chain propagation of lipid oxidation, thus protecting polyunsaturated fatty acids and low-density lipoproteins from oxidation.6 Vitamin E is also associated with the prevention of coronary heart disease, atherosclerosis, cancer, diabetes, Parkinson’s disease, Alzheimer’s disease, impaired immune function,7,8 and ischemic heart disease in cross-cultural epidemiology.9 Furthermore, it is involved in the inhibition of cell proliferation in vascular smooth muscle,10 protein kinase C activity,11 or regulation of gene expression.12 Recent evidence also suggests that vitamin E (concretely R-tocopherol) has further roles in cellular signaling, which are independent of its antioxidant function.13 While most studies focus on R-tocopherol, other vitamers (such as γ-tocopherol) have also been associated with the reduced incidence of coronary heart disease,14,15 prostate cancer,16 and peroxynitrite-dependent damage.17 Moreover, tocotrienols have been shown to possess significant antioxidant (1) Ball, G. F. M. Vitamin E. In Bioavailability and Analysis of Vitamins in Foods, 1st ed.; Chapman and Hall: London, U.K., 1998; pp 195-239. (2) Ruperez, F. J.; Martin, D.; Herrera, E.; Barbas, C. J. Chromatogr., A 2001, 935, 45-69. (3) Burton, G. W.; Joyce, A.; Ingold, K. U. Arch. Biochem. Biophys. 1983, 221, 281-90. (4) Ingold, K. U.; Webb, A. C.; Witter, D.; Burton, G. W.; Metcalfe, T. A.; Muller, D. P. Arch. Biochem. Biophys. 1987, 259, 224-25. (5) Hoppe, P. P.; Krennrich, G. Eur. J. Nutr. 2000, 39, 183-93. (6) Morrissey, P. A.; Sheehy, P. J. Proc. Nutr. Soc. 1999, 58, 459-68. (7) Bramley, P. M.; Elmadfa, I.; Kafatos, A.; Kelly, F. J.; Manios, Y.; Roxborough, H. E.; Schuch, W.; Cheehy, P. J. A.; Wagner, K.-H. J. Sci. Food Agric. 2000, 80, 913-38. (8) Traber, M. G.; Sies, H. Annu. Rev. Nutr. 1996, 16, 321-47. (9) Gey, K. F.; Puska, P.; Jordan, P.; Moser, U. K. Am. J. Clin. Nutr. 1991, 53, 326S-34S. (10) Tasinato, A.; Boscoboinik, D.; Bartoli, G. M.; Maroni, P.; Azzi, A. Proc. Natl. Acad. Sci. U.S.A 1995, 92, 12190-94. (11) Ricciarelli, R.; Tasinato, A.; Clement, S.; Ozer, N. K.; Boscoboinik, D.; Azzi, A. Biochem. J. 1998, 334 ( Part 1), 243-49. (12) Aratri, E.; Spycher, S. E.; Breyer, I.; Azzi, A. FEBS Lett. 1999, 447, 91-94. (13) Brigelius-Flohe, R.; Kelly, F. J.; Salonen, J. T.; Neuzil, J.; Zingg, J. M.; Azzi, A. Am. J. Clin. Nutr. 2002, 76, 703-16. (14) Kontush, A.; Spranger, T.; Reich, A.; Baum, K.; Beisiegel, U. Atherosclerosis 1999, 144, 117-22. (15) Ohrvall, M.; Sundlof, G.; Vessby, B. J. Intern. Med. 1996, 239, 111-17.

Analytical Chemistry, Vol. 79, No. 18, September 15, 2007 7087

Figure 1. Structures of natural vitamin E constituents and major oxidation products. Note, that one group of compounds possesses closed chroman ring (R-tocopherol) and the other group contains open chroman ring (R-tocopherylquinone). This latter feature determines strongly the CID fragmentation of these analytes.

properties.18,19 Details on physiology, bioavailability, and potency of vitamin E are available in several excellent reviews.5,7,8,17,20 Vitamin E intake is for most people far from optimal;21 thus, many clinical studies investigate the (enhanced or compromised) bioavailability and metabolism of this micronutrient. On the other hand, the bioavailability and health benefits of vitamin E are still not fully understood.22,23 The aforementioned facts make vitamin E an important scientific topic and motivate the development of effective and comprehensive analytical methods for its accurate determination in biological fluids and food products. (16) Huang, H. Y.; Alberg, A. J.; Norkus, E. P.; Hoffman, S. C.; Comstock, G. W.; Helzlsouer, K. J. Am. J. Epidemiol. 2003, 157, 335-44. (17) Brigelius-Flohe R; Traber, M. G. FASEB J. 1999, 13, 1145-55. (18) Packer, L. In Nutrition, Lipids, Health and Disease; Ong, A. S., Niki, E., Packer, L., Eds.; American Oil Chemists’ Society (AOCS): Champaign, IL, 1995. (19) Qureshi, N.; Qureshi, A. A. In Vitamin E in Health and Disease; Packer, L., Fuchs, J., Eds.; Dekker: New York, 1993. (20) FAO/WHO (Food and Agriculture Organization of the United Nations/ World Health Organization) Human Vitamin and Mineral Requirements; Report No. 121-131; 2002. (21) Bruno, R. S.; Leonard, S. W.; Park, S.; Zhao, Y. Y.; Traber, M. G. Am. J. Clin. Nutr. 2006, 83, 299-304. (22) Proteggente, A. R.; Turner, R.; Majewicz, J.; Rimbach, G.; Minihane, A. M.; Kramer, K.; Lodge, J. K. J. Nutr. 2005, 135, 1063-69. (23) Lodge, J. K.; Traber, M. G.; Elsner, A.; Brigelius-Flohe, R. J. Lipid Res. 2000, 41, 148-54.

7088

Analytical Chemistry, Vol. 79, No. 18, September 15, 2007

A large variety of techniques have been published for the analysis of vitamin E.2 Recovery of vitamin E from biological samples is usually done by applying liquid-liquid extraction,24,25 but solid-phase extraction26-28 was also utilized to recover tocopherols from different matrixes including plasma. Chromatographic separation of vitamin E constituents can be performed either by gas chromatography, liquid chromatography, or capillary electrophoresis. For gas chromatographic analysis, the sample is usually converted into trimethyl-silyl derivatives26,29-32 to increase volatility and prevent degradation of analytes. Liquid chromatographic analysis of vitamin E is performed often by both reversed(24) Finckh, B.; Kontush, A.; Commentz, J.; Hubner, C.; Burdelski, M.; Kohlschutter, A. Methods Enzymol. 1999, 299, 341-48. (25) Podda, M.; Weber, C.; Traber, M. G.; Milbradt, R.; Packer, L. Methods Enzymol. 1999, 299, 330-41. (26) Lechner, M.; Reiter, B.; Lorbeer, E. J. Chromatogr., A 1999, 857, 231-38. (27) Heudi, O.; Trisconi, M. J.; Blake, C. J. J. Chromatogr., A 2004, 1022, 11523. (28) Bonvehi, J. S.; Coll, F. V.; Rius, I. A. J. AOAC Int. 2000, 83, 627-34. (29) Liebler, D. C.; Burr, J. A.; Ham, A. J. Methods Enzymol. 1999, 299, 30918. (30) Pyka, A.; Sliwiok, J. J. Chromatogr., A 2001, 935, 71-76. (31) Mottier, P.; Gremaud, E.; Guy, P. A.; Turesky, R. J. Anal. Biochem. 2002, 301, 128-35. (32) Liebler, D. C.; Burr, J. A.; Philips, L.; Ham, A. J. Anal. Biochem. 1996, 236, 27-34.

phase30,31,33-38 and normal-phase liquid chromatography.30,39-44 Reversed-phase separation of tocopherols/tocotrienols is compatible with electrospray mass spectrometry and is easy to perform, but it cannot resolve all isomers.1,25,30,43,44 This can be a problem, particularly if the β and γ isomers must be quantified individually, as these vitamers have the same mass and hence their mass spectrometric distinction is not feasible. Normal-phase liquid chromatography is able to separate all the vitamers including the β and γ isomers with the same masses, but this type of chromatography can be coupled only to atmospheric-pressure chemical ionization since the applied solvents (including n-hexane) are not compatible with electrospray process. Detection of vitamin E constituents can be performed with a wide variety of techniques including fluorescence detection,36,38,40,42,45,46 electrochemical detection,24,34,47-50 ultraviolet absorbance (UV) detection,30,35,46 flame ionization detection,26 and mass spectrometry.27,29,31-33,37,39,51-53 Mass spectrometry is being deployed increasingly, as this is the only technique that can distinguish between isotope-labeled homologue compounds and thus allows the application of isotope-labeled internal standards.32 For gas chromatographic coupling, usually electron ionization mass spectrometry is used,29,31,32 while for liquid chromatographic coupling, electrospray31,37 or atmospheric-pressure chemical ionization27,33,39,51-53 mass spectrometry is employed. To date, mass spectrometry has only been applied to detect vitamin E constituents with known masses, whereas one big advantage of tandem mass spectrometry would be to detect (33) Leonard, S. W.; Good, C. K.; Gugger, E. T.; Traber, M. G. Am. J. Clin. Nutr. 2004, 79, 86-92. (34) Leray, C.; Andriamampandry, M. D.; Freund, M.; Gachet, C.; Cazenave, J. P. J. Lipid Res. 1998, 39, 2099-105. (35) Gimeno, E.; Castellote, A. I.; Lamuela-Raventos, R. M.; de la Torre-Boronat, M. C.; Lopez-Sabater, M. C. J. Chromatogr., B: Biomed. Sci. Appl. 2001, 758, 315-22. (36) Ruperez, F. J.; Barbas, C.; Castro, M.; Herrera, E. J. Chromatogr., A 1999, 839, 93-99. (37) Hall, W. L.; Jeanes, Y. M.; Pugh, J.; Lodge, J. K. Rapid Commun. Mass Spectrom. 2003, 17, 2797-803. (38) Huo, J. Z.; Nelis, H. J.; Lavens, P.; Sorgeloos, P.; De Leenheer, A. P. Anal. Biochem. 1996, 242, 123-28. (39) Kalman, A.; Mujahid, C.; Mottier, P.; Heudi, O. Rapid Commun. Mass Spectrom. 2003, 17, 723-27. (40) Kamal-Eldi, A.; Gorgen, S.; Pettersson, J.; Lampi, A. M. J. Chromatogr., A 2000, 881, 217-27. (41) Gawlik, M. T.; Gawlik, M. B.; Go´rka, A.; Brandys, J. Acta Chromatogr. 2003, 13, 185-95. (42) Panfili, G.; Fratianni, A.; Irano, M. J. Agric. Food Chem. 2003, 51, 394044. (43) Schuep, W.; Rettenmaier, R. Methods Enzymol. 1994, 234, 294-302. (44) Kramer, J. K.; Fouchard, R. C.; Kallury, K. M. Methods Enzymol. 1999, 299, 318-29. (45) Cayuela, J. M.; Garrido, M. D.; Banon, S. J.; Ros, J. M. J. Agric. Food Chem. 2003, 51, 1120-24. (46) Casal, S.; Macedo, B.; Oliveira, M. B. J. Chromatogr., B: Biomed. Sci. Appl. 2001, 763, 1-8. (47) Finckh, B.; Kontush, A.; Commentz, J.; Hubner, C.; Burdelski, M.; Kohlschutter, A. Anal. Biochem. 1995, 232, 210-16. (48) Takeda, H.; Shibuya, T.; Yanagawa, K.; Kanoh, H.; Takasaki, M. J. Chromatogr., A 1996, 722, 287-94. (49) Vatassery, G. T.; Smith, W. E.; Quach, H. T. Anal. Biochem. 1993, 214, 426-30. (50) Lang, J. K.; Gohil, K.; Packer, L. Anal. Biochem. 1986, 157, 106-16. (51) Vaule, H.; Leonard, S. W.; Traber, M. G. Free Radical Biol. Med. 2004, 36, 456-63. (52) Lauridsen, C.; Leonard, S. W.; Griffin, D. A.; Liebler, D. C.; McClure, T. D.; Traber, M. G. Anal. Biochem. 2001, 289, 89-95. (53) Andreoli, R.; Manini, P.; Poli, D.; Bergamaschi, E.; Mutti, A.; Niessen, W. M. Anal. Bioanal. Chem. 2004, 378, 987-94.

unexpected or new metabolites that are likely to be included in vitamin E metabolism. In this regard, tandem mass spectrometric fragmentation pathways of vitamin E constituents allow selective screening for vitamin E metabolites or oxidation products with partially known structures. The main motivation for this work was to extend the capacity of mass spectrometry to generate a more comprehensive picture on the fate of vitamin E constituents in body fluids and food matrixes, such as human plasma or fish oil. Accordingly, the aim of the present study was to develop and validate (1) an HPLCMS method for quantification of all natural vitamin E constituents and oxidation products in human plasma and (2) provide tandem mass spectrometric approaches for the targeted discovery of vitamin E metabolites and oxidation products with unknown masses, but partially defined structures. For these purposes, isotope dilution-based extraction, normal-phase separation, and single- and tandem-stage mass spectrometric detection of tocopherols, tocopherylacetates, tocotrienols, and tocopherolquinones were optimized and validated. EXPERIMENTAL SECTION Chemicals. HPLC grade water, ethanol, 1,4-dioxane, n-hexane, BHT (2,6-di-tert-butyl-4-methylphenol), racemic mixtures (all-rac) of R-tocopherol, β-tocopherol, γ-tocopherol, and δ-tocopherol were obtained from VWR International AG, Dietikon, Switzerland. Allrac R-tocopherylacetate (2,5,7,8-tetramethyl-2-(4′,8′,12′-trimethyltridecyl)-6-chromanol-acetate) and all-rac 2H9-R-tocopherylacetate (2-methyl-5,7,8-tri(methyl-2H3)-2-(4′,8′,12′-trimethyltridecyl)-6-chromanol-acetate were obtained from Sigma-Aldrich Chemie GmbH, Buchs, Switzerland. All-rac R-tocopherylquinone was purchased from Juro Supply GmbH, Lucerne, Switzerland. All-rac 2H9-Rtocopherol (2-methyl-5,7,8-tri(methyl-2H3)-2-(4′,8′,12′-trimethyltridecyl)-6-chromanol) was obtained from Chemaphor Inc., Ottawa, Canada. Stereochemically pure D-R-tocotrienol, D-β-tocotrienol, D-γtocotrienol, and D-δ-tocotrienol were obtained from Davos Life Science Pte Ltd., Biopolis, Singapore. D-2H6-R-Tocopherylacetate (2,8-dimethyl-5,7-di(methyl-2H3)-2-(4′,8′,12′-trimethyltridecyl)-6chromanol-acetate) and D-2H6-R-tocopherol (2,8-dimethyl-5,7-di(methyl-2H3)-2-(4′,8′,12′-trimethyltridecyl)-6-chromanol) were provided by Orphachem, Clermont-Ferrand, France. Standard Solutions. (1) To check sensitivity and chromatographic resolution, a mixture of 10 analytes was measured before each sample queue at a concentration of 1 µg/mL in n-hexane/ 1,4-dioxane 99.9:0.1. (2) As internal standards, 2H9-R-tocopherylacetate (400 ng/mL) and 2H9-R-tocopherol (400 ng/mL) were utilized and dissolved in pure ethanol. (3) For spiking purposes, a solution containing 240 ng/mL R-tocopherolacetate, 1200 ng/ mL R-tocopherol, 1200 ng/mL 2H6-R-tocopherol, 800 ng/mL R-tocopherylquinone, 2400 ng/mL β-tocopherol, 2400 ng/mL γ-tocopherol, 2400 ng/mL δ-tocopherol, 2400 ng/mL R-tocotrienol, 3200 ng/mL β-tocotrienol, 3200 ng/mL γ-tocotrienol, and 4000 ng/mL δ-tocotrienol was prepared in pure ethanol. This stock solution was diluted according to the desired spike levels. Sample Preparation. Frozen plasma samples were received in polypropylene tubes. Samples were thawed and sonicated in an ultrasonic bath for 5 min. Samples were vortexed for 10 s. An amount of 100 µL of plasma was transferred into a Pyrex 15 glass tube (15 mm diameter, 100 mm length) containing 900 µL of water. Amber glass was used to protect the analytes from light. An Analytical Chemistry, Vol. 79, No. 18, September 15, 2007

7089

Table 1. Composition, Flow Rate, and Timing of the Gradient Applied for the Normal-Phase Separation of Vitamin E Constituentsa time [min] 0 3 4 15 16 17 25

solvent A [%] 100 95 80 60 0 100 100

solvent B [%] 0 5 20 40 100 0 0

flow rate [µL/min] 500 500 300 300 300 300 500

curve value 1 6 6 6 6 6 1

a Solvent A was n-hexane, solvent B was 1,4-dioxane. Note, that at the beginning and end of the gradient, the flow rate was speeded up in order to reduce analysis time.

amount of 500 µL of internal standard solution (400 ng/mL 2H9R-tocopherylacetate and 2H9-R-tocopherol dissolved in ethanol) and 500 µL of pure ethanol was added, and the tube was vortexed for 10 s and sonicated for 2 min. An amount of 4 mL of the butylated hydroxytoluene (BHT) solution (5 mg/100 mL in n-hexane) was added into the tube, and liquid-liquid extraction was performed in an IKA VXR Vibrax basic shaker (IKA-2819000) at 2200 rpm speed for 5 min. The BHT was added to the organic solvent in order to minimize oxidation of target analytes. After liquid-liquid extraction, the tubes were centrifuged at 4000 rpm (2900g) for 1 min in a Sigma 3-16K centrifuge, then 3.9 mL of the organic layer was transferred into another Pyrex 15 tube. Liquid-liquid extraction was repeated in a second step, and the resulting organic layers were unified. The aqueous phases were discarded. The organic phases were evaporated using nitrogen flow, and the residuum were dissolved in 200 µL of n-hexane/1,4-dioxane 99.9:0.1. An amount of 50 µL of this was injected into the HPLC-MS system. Fish oil produced for infant formulations was used to mimic oxidation processes of tocopherols during the heat-treatment step of oil-processing. An amount of 90 µL of fish oil was mixed with 10 µL of 2H6-R-tocopherol and incubated at 150 °C for 10 min. Then the mixture was frozen to stop chemical reactions. Before analysis, 1 mL of a mixture of n-hexane/1,4-dioxane 99.9:0.1 was added to the sample, and 10 µL was injected for each HPLC-MS analysis. Liquid Chromatography. HPLC separation of free tocopherols, tocopherolacetate, tocotrienols, and metabolites was achieved on a propyl-amine column (3 µm particle size, 2 mm × 250 mm, Varian A2014250X020) using a Waters Acquity UPLC system. Mobile phase A was 100% n-hexane, mobile phase B was n-hexane/1,4-dioxane 1:1. The mobile phase gradient is summarized in Table 1. Mass Spectrometry. Mass spectrometric detection was performed on a Micromass Quattro LC triple quadrupole tandem mass spectrometer in the positive atmospheric pressure chemical ionization (APCI) mode at unit resolution. The corona current was 10 µA, cone voltage was 30 V, extractor voltage was 3 V, source temperature was 150 °C, and vaporizer temperature was 500 °C. For tandem mass spectrometric experiments, a collision energy of 30 eV was applied. All other lens and gas pressure parameters were optimized for maximum sensitivity. 7090 Analytical Chemistry, Vol. 79, No. 18, September 15, 2007

Table 2. List of Molecular Weight, Observed Molecular Ion (Q1), Fragment Ion (Q3), and Characteristic Neutral Loss of Each Standard Analyte under Positive APCI Conditionsa analyte

MW [Da]

Q1 m/z

Q3 m/z

characteristic mass loss [Da]

9-R-tocopherolacetate R-tocopherolacetate 2H -R-tocopherol 9 R-tocopherylquinone 2H -R-tocopherol 6 R-tocopherol R-tocotrienol β-tocopherol γ-tocopherol β-tocotrienol γ-tocotrienol δ-tocopherol δ-tocotrienol

481 472 439 446 436 430 424 416 416 410 410 402 396

482 473 440 429 437 431 425 417 417 411 411 403 397

216 207 174 165 171 165 165 151 151 151 151 137 137

266 266 266 264 266 266 260 266 266 260 260 266 260

2H

a Note, fragmentation of tocopherols and tocotrienols is dominated by the same fragmentation process (details in the text) leading to the characteristic ions m/z 165 (marker for R-vitamers), m/z 151 (marker for β- and γ-vitamers), and m/z 137 (marker for δ-vitamers).

RESULTS AND DISCUSSION Method Development. The mass spectrometric behavior of all analytes was studied using both positive- and negative-ion APCI. Electrospray ionization (ESI) was not deployed, because it is not compatible with normal-phase eluents such as n-hexane. Positiveion APCI showed approximately 10 times better sensitivity than negative-ion APCI; thus, the positive mode was used for further work. Under such conditions, all target analytes (except R-tocopherylquinone) yielded mass spectra dominated by the protonated molecular ion (see mass spectrum of R-tocotrienol as an example in Figure 2 A). In the mass spectrum of R-tocopherylquinone, the loss of water is dominating so that the molecular ion appears at m/z 429 instead of 447. Mass channels applied for selected-ion monitoring (SIM) and multiple-reaction monitoring (MRM, see tandem mass spectrometry section) experiments are summarized in Table 2. Several solvent combinations were tested for optimal separation of the target analytes including n-hexane, tetrahydrofurane, isopropanol, ethyl acetate, methyl-tert-butyl-ether, and 1,4-dioxane. The best separation of the most challenging β-γ isomer pairs was achieved with n-hexane and 1,4-dioxane (see Table 1 for details on the applied gradient). A typical chromatogram for a mixture of 10 analytes is shown in Figure 3A. The identities of the peaks are given Figure 1. Retention time of the first analyte (R-tocopherylacetate) is 4.5 min, which ensures a good separation from the non- or weakly retained interferences such as BHT (antioxidant, see sample preparation section), which elutes before 2 min, as confirmed by UV detection. Apart from the efficient separation of all compounds, it can be observed that for tocotrienols the signal intensity is decreased by a factor of 2-3 compared to tocopherols, the latter lacking any double bond in the phytil chain. In order to minimize the analysis time, the flow rate was increased in the beginning and end of the gradient to 500 µL/min to speed up separation and re-equilibration. On the other hand, when the compounds of interest eluted from the column, the flow rate was reduced to 300 µL/min to maximize efficacy of ion formation during the APCI process. While separa-

Figure 2. Part A depicts the singly stage mass spectrum of R-tocotrienol obtained by positive APCI. The spectrum is dominated by the protonated molecular ion. Isotope pattern of the detected peak is close to the theoretical one, as shown in the magnified part. At lower masses some fragmentation can be observed, which is in accordance with the product-ion spectrum of R-tocotrienol as shown in part B.

Figure 3. Separation of a standard mixture of vitamin E constituents (for composition see text) on an amine-modified normal phase column is shown using single stage (part A) and tandem mass spectrometric detection (part B). Baseline separation of the β-γ isomers of tocopherols and tocotrienols was achieved. Note that peak height (sensitivity) is not the same for the different analytes, but it is within 1 order of magnitude (all peaks correspond to 500 ng injected amount).

tion was performed using an Acquity UPLC system, it is straightforward to implement the presented method on conventional HPLC systems, too, since the observed pressure did not exceed 230 bar.

Method Validation. As pointed out in the introduction, quantification of vitamin E constituents in human plasma is increasingly requested when studying bioavailability and bioefficacy in humans. Accordingly, human plasma was used in the Analytical Chemistry, Vol. 79, No. 18, September 15, 2007

7091

Table 3. Results of Assessing the Performance of Quantitation Are Showna

analyte R-tocopherylacetate 2H -R-tocopherol 6

R-tocotrienol R-tocopherylquinone β-tocopherol γ-tocopherol β-tocotrienol γ-tocotrienol δ-tocopherol δ-tocotrienol

lower limit of detection [nmol /L plasma]

lower limit of quantitation [nmol /L plasma]

linear range [µmol /L plasma]

linearity coefficient (R2)

recovery [%]

3 14 28 9 29 29 39 39 30 51

8 46 94 31 96 96 130 130 99 168

0.01-2.5 0.06-14 0.09-28 0.04-9 0.5-29 1.6-11 0.14-29 0.12-29 0.18-10 0.17-10

0.9977 0.9960 0.9989 0.9988 0.9923 0.9778 0.9884 0.9930 0.9939 0.9919

92 92 69 87 64 53 63 69 58 63

accuracy [%]

intraday repeatibility at point 4 [RSD %] (n ) 5)

interday repeatibility at point 4 [RSD %] (n ) 6)

suppression [%]

103 103 101 na na na na na na 99

2 4 12 8 10 13 17 11 13 7

5 15 13 9 15 18 18 11 14 12

0 14 23 25 27 9 18 30 27 29

a Validation parameters were determined for each standard analyte except accuracy, where the matrix (human plasma) already contained some of the analytes; thus, the determination of accuracy for these was not applicable (na).

present study as a model matrix to validate the method by employing 9-fold deuterated compounds as internal standards. Prior to method validation, plasma samples were prepared and analyzed and the mass channels of both 2H9-R-tocopherol and 2H9R-tocopherylacetate were checked to exclude interferences from endogenous compounds. Lower limits of detection (LLODs; criterion, signal-to-noise >3) were determined by injecting the dilution series of mixtures of all analytes. Lower limits of quantification (LLOQs; criterion, signal-to-noise >10) were determined in the same manner. Both LLOD and LLOQ results are shown in Table 3. In accordance to Figure 3A, tocotrienols generally show a lower MS response than tocopherols. On the other hand, the analytes with more methyl groups on the chroman ring (such as R-tocopherol, see Figure 1) show better sensitivity than the ones with less methyl groups (δtocopherol, see Figure 1). Calibration curves were established applying the method of standard additions: the samples were spiked with known amounts of analytes prior to the sample preparation step. The spiking steps were adjusted for each analytes to be at the following levels: LLOD, LLOQ, 10 × LLOD, 33 × LLOD, 100 × LLOD, 333 × LLOD, 1000 × LLOD. Each calibration point was determined in triplicates (including the sample preparation step). Endogenous levels of each of the analytes were determined after plotting the ratio [analyte area/internal standard area] as a function of spiked amount of compounds according to the following equation:

endogenous level )

b a

where a is the slope of the calibration curve; b is the intercept of the calibration curve. Then calibration curves were corrected by plotting [analyte area/internal standard area] values as a function of the absolute analyte amounts (sum of the endogenous and the spiked analyte amounts). A typical calibration curve is shown in Figure 4 for R-tocotrienol. The curve comprises a concentration range of 3 orders of magnitude starting from the LLOD. Linearity coefficients and linear ranges for each of the analytes are shown in Table 3. While the linear range for several analytes was determined to exceed 2 orders of magnitude, for some analytes only narrower linear ranges (e.g., a factor of 50 for γ-tocopherol) 7092 Analytical Chemistry, Vol. 79, No. 18, September 15, 2007

Figure 4. Calibration curve is shown for R-tocotrienol obtained from human plasma. Each point includes a separate sample preparation process.

Figure 5. Residual plot for R-tocotrienol. The deviations from the abscissa illustrate the bending of the calibration curve, which is otherwise difficult to judge by eye. In this case slight S-type bending can be observed, which is typical when preparing calibration curves covering a wide concentration range.

could be obtained. This can be explained by the fact that the endogenous plasma concentrations were significantly higher than the LLOD, hindering the acquisition of calibration points in the lower concentration range. In addition, this demonstrates that the described technique is sensitive enough to detect and quantify these analytes below the endogenous levels. Residual plots were also prepared and an example is shown in Figure 5 for

Figure 6. In-source CID (part A) and tandem mass spectrum (part B) of γ-tocopherol obtained from human plasma. While the in-source CID spectrum is very rich in signals making the identification difficult, the product ion spectrum is clean and contains only the fragments originated from γ-tocopherol including the very strong marker ion at m/z 151.

R-tocotrienol. The plot illustrates in a normalized scale how the calibration points differ from the calibration trend line and it aids to recognize weak bending tendencies in the calibration (for wide concentration ranges usually “S-type” bending can be observed). To investigate the efficacy of the sample preparation, the recovery values were determined by comparing the analyte levels before and after the extraction process. To determine the analyte levels after sample preparation, again the method of standard additions was performed: samples were spiked in two steps in triplicate after the extraction procedure. The analyte concentrations were derived from the respective calibration curve and ultimately compared with the analyte levels before the extraction process. Differences were expressed in percentages and the recovered values are given in Table 3. As expected, the more apolar compounds such as R-tocopherylacetate or R-tocopherol (structure shown in Figure 1) showed better recovery, as their solubility was worse in the aqueous phase and better in the organic phase. The accuracy (trueness) of the method was investigated by comparing theoretical and experimentally measured analyte levels for R-tocopherylacetate, 2H6-R-tocopherol, R-tocotrienol, and δ-tocotrienol. Results are shown in Table 3. The accuracy for other analytes was not determined because the plasma matrix originally contained these analytes in detectable amounts. Tandem Mass Spectrometry of Vitamin E. Confirmation of analyte identities can be achieved by inducing fragmentation of the molecular ions. The two most popular ways to achieve this are in-source CID and tandem mass spectrometry. However, if the detection of vitamin E is required directly in human plasma, in-source CID-aided identification is not feasible, because the mass spectra are extremely signal-rich. As an example, an in-source CID induced mass spectrum of γ-tocopherol from human plasma is shown in Figure 6A. By contrast, the application of “true” tandem mass spectrometry (mass selection of parent ions precedes the

fragmentation step) can reduce the interferences originating from coeluting compounds such as sterols and triacylglycerols and thus give a clean mass spectrum helpful for identification. As an example, Figure 6B shows the product ion spectrum of γ-tocopherol from human plasma. Like in the case of the standard compound, the spectrum is strongly dominated by the marker ion m/z 151, which corresponds to the chroman ring of the tocopherol (for details see later). Fragmentation characteristics of all analytes were studied using CID tandem mass spectrometry. Tocotrienols yielded product ion mass spectra dominated by the loss of 260 Da (see Figure 2B for R-tocotrienol), which corresponds to the opening of the ether bond and loss of the side chain (5E,9E)-2,6,10,14-tetramethylpentadeca1,5,9,13-tetraene. The proposed formation scheme for this process is shown in Figure 7, depicting the fragmentation pattern of R-tocotrienol. The same mechanism with a 6 Da heavier side chain is suggested for tocopherols revealing the loss of 266 Da, which was also experimentally observed. In this way, product ions at m/z 165 (for R), 151 (for β and γ), and 137 (for δ) can be used as selective marker ions for detecting/confirming tocopherols and tocotrienols. With dependence on the setup of biological trials and the required target analyte enrichment, chemical noise and endogenous interferences (especially fragment ions of major plasmacomponents) can deteriorate the detection limits and compromise the quantification process. To minimize interferences caused by nontarget analytes, the MS selectivity for the analysis can be enhanced through multiple-reaction monitoring (MRM). Characteristic parent and daughter ions for all analytes are summarized in Table 2. Selection of the mass channels from Table 2 and use of the parameters given in the Experimental Section, a chromatogram as shown in Figure 3B can be obtained. For identity of peaks refer to Figure 1. Analytical Chemistry, Vol. 79, No. 18, September 15, 2007

7093

Figure 7. Proposed fragmentation scheme for the formation of ion m/z 165, the most abundant fragment in the product ion spectrum of R-tocopherol and R-tocotrienol. The same mechanism is suggested for the formation of marker ions m/z 151 and m/z 137 for β-γ and δ isomers, respectively.

Targeted Discovery of Oxidation Products in Fish Oil. To discover metabolites and oxidation products (such as epoxytocopherone- or, tocopherolquinone-type compounds17,54) with partially known structures, but unknown masses, it is straightforward to apply compound-specific neutral-loss or precursor-ion scan experiments. This is illustrated below with another application of the present method in the nutritional field: the reaction of tocopherols with polyunsaturated fatty acids is studied. Fish oil, a rich source of polyunsaturated fatty acids, was incubated with 2H6-R-tocopherol to mimic natural oxidation processes of tocopherols (for details see Experimental Section). To discover tocopherol derivatives with a modified chroman ring, 266 Da neutral-loss scan experiments were performed. An example is shown in the case of Figure 8A, which depicts the 266 Da neutral-loss mass spectrum of two isomers of a new compound (not detected in heated fish oil in the absence of 2H6-R-tocopherol) detected at 7.8 and 8.3 min retention time, respectively. The identified molecular ion for both analytes is at m/z 453, which may correspond to 2H6-8a-hydroxytocopherone (for structure see Figure 1). Isomers of this compound contain one additional oxygen atom incorporated into the chroman ring compared to 2H6-R-tocopherol. The phytil chain and the ether bond are preserved, which explains why these compounds exhibit the 266 Da neutral-loss similarly to 2H6-R-tocopherol. The product ion spectrum of the hypothesized 2H6-8ahydroxy-tocopherone is shown in Figure 8B. The peak at m/z 171 suggests that such epoxy derivatives, besides following similar fragmentation of tocopherols (neutral loss of 266 Da), also tend to lose the incorporated oxygen atom from the chroman ring during the CID process. In this way, they can also efficiently be discovered by applying 282 Da neutral-loss scans or m/z 171 precursor-ion scans, as experimentally confirmed. To further support the identity of 2H6-8a-hydroxy-tocopherone (peak m/z (54) Liebler, D. C. Methods Enzymol. 1994, 234, 310-16.

7094 Analytical Chemistry, Vol. 79, No. 18, September 15, 2007

453), its accurate mass was determined employing high-resolution mass spectrometry (details not shown) and a mass deviation of only 3 ppm between theory and experiment was found. With the use of m/z 171 precursor ion scan experiments (specific for compounds with intact chroman ring), a compound with a molecular ion at m/z 435 at 8.14 min retention time was discovered, see Figure 9A. The product ion mass spectrum of this analyte is shown in Figure 9B. On the basis of the fragmentation pattern and the water loss characteristic for tocopherylquinones, this compound may correspond to 2H6-R-tocopherolquinone (for structure see Figure 1). Co-chromatography with nonlabeled R-tocopherylquinone revealed identical elution times and further supported the identity of 2H6-R-tocopherolquinone in the fish oil reaction mixture. Also in this case, the identity of 2H6-R-tocopherolquinone was further supported by determining its accurate mass using high-resolution mass spectrometry (details not shown), and a mass deviation of only 2 ppm from the theoretical value was found. In the same m/z 171 precursor ion scan run, two compounds (isomers) with m/z 451 were discovered at retention times of 9 and 9.2 min, respectively. These compounds may correspond to 2H -epoxy-R-tocopherolquinones (for structure see Figure 1), 6 assuming they exhibit the same characteristic water loss in the positive APCI mode, like observed for tocopherylquinone. This latter finding is supported by the ion chromatogram of m/z 469 (epoxy-R-tocopherylquinones without water loss), which contains two weak peaks at the same retention times 9 and 9.2 min, respectively. In our experience, tocopherylquinone and other analytes similar in structure exhibit this water loss only in positive APCI but not in negative APCI. To confirm the pseudomolecular mass of m/z 468 (without the water loss), the reaction mixture was analyzed in negative APCI. Two intense peaks confirmed the identity of 2H6-R-epoxy-tocopherolquinones in the ion chromatogram of m/z 468 at retention times 9 and 9.2 min, respectively. To further confirm the identity of 2H6-epoxy-R-tocopherylquinone isomers (peaks m/z 451 observed in positive mode and peaks m/z 468 in negative mode), their accurate mass was determined in both the positive and negative modes using high-resolution mass spectrometry (details not shown) and deviations of only 2 and 1 ppm between expected and measured masses were found, respectively. Targeted Discovery of Oxidation Products in Blood. In order to investigate whether the oxidation processes observed in vitro also occur in blood, the nonlabeled forms of R-tocopherol oxidation products were screened by employing m/z 165 precursor-ion scans and 266 Da neutral-loss scans. In these tandem mass spectra, R-tocopherol and R-tocopherylquinone were found to be the far most abundant compounds. However, in the m/z 165 precursor-ion spectrum, a minor peak at 8.5 min retention time was also observed with a molecular mass of m/z 445. This mass may correspond to epoxy-R-tocopherylquinones after water loss (for structure see Figure 1). To confirm this hypothesis, the peak was also detected by negative APCI. Just like in the case of 2H -epoxy-R-tocopherylquinone detailed above, the intense peak 6 at m/z 462 (without the water loss) indicated that the real mass of the detected compound is 462 Da and the observed ion at m/z 445 in positive mode is a result of water loss. The product ion spectrum of the ion at m/z 445 was found to be very similar to

Figure 8. Observed 266 Da neutral-loss mass spectrum (part A) and product-ion spectrum (part B) of 2H6-8a-hydroxy-tocopherone from the fish oil reaction mixture.

Figure 9. Observed m/z 171 precursor-ion spectrum (part A) and product-ion mass spectrum (part B) of 2H6-R-tocopherylquinone from the fish oil reaction mixture.

the pattern observed in the case of 2H6-epoxy-R-tocopherylquinone in fish oil, except the fact that the marker ion of the chroman ring was shifted from m/z 171 to m/z 165 according to the absence of the 6-fold deuteration. The identity of epoxy-R-tocopherylquinone was further supported by determining its accurate mass using high-resolution mass spectrometry (details not shown). Mass errors of 2 and 3 ppm were found in the positive and in the negative mode, respectively. Since the 2H9-homologues of epoxyR-tocopherylquinone cannot be detected in the samples, the

detected epoxy-R-tocopherylquinone is likely not an artifact introduced by the sample preparation procedure but an endogenous component of blood as a result of the spontaneous oxidation of R-tocopherol. To further consolidate this assumption, blood samples were incubated for 0, 1, and 16 h. Results are shown in Figure 10. While the absolute amount of R-tocopherol decreased with time from 22.3 to 17.4 µM (values in the range of the literature52), the amount of epoxy-R-tocopherylquinone increased by a factor of 2. When the ratio between epoxy-R-tocopherylAnalytical Chemistry, Vol. 79, No. 18, September 15, 2007

7095

Figure 10. Effect of incubation at 37 °C on the amount of R-tocopherol and epoxy-R-tocopherylquinone in blood. Absolute amount of R-tocopherol decreased from 22.3 to 17.4 µM during 16 h (see columns with diagonal stripes). In contrast, the amount of epoxy-R-tocopherylquinone increased by a factor of 2. This latter tendency is even more significant when plotting the ratio between epoxy-R-tocopherylquinone and R-tocopherol (columns with horizontal stripes).

quinone and R-tocopherol is plotted, the tendency is even more significant suggesting that spontaneous oxidation of R-tocopherol occurs when incubating blood at 37 °C. CONCLUSIONS This manuscript describes and validates a comprehensive HPLC-MS method combined with isotope dilution-based sample preparation for the quantitative analysis of all natural vitamin E constituents including tocotrienols and tocopherols in human plasma. In addition, CID fragmentation of vitamin E constituents is discussed and utilizing tandem mass spectrometric detection, the method was applied for the structure-specific discovery of oxidation products of vitamin E. Major achievements of the presented approach are the following. (i) Esterified and free tocopherols, quinones, tocotrienols including β-γ isomers could be separately quantified in the same HPLC-MS run without applying chemical derivatization. (ii) Analysis time is only 25 min including washing and re-equilibration of the column. (iii) Lower detection and quantification limits, recovery, accuracy, precision, linearity, and suppression effects were determined. (iv) Fragmentation characteristics of tocopherols and tocotrienols were studied and fragmentation schemes for the most intense daughter ions

7096

Analytical Chemistry, Vol. 79, No. 18, September 15, 2007

are proposed. Tandem mass spectrometric approaches including neutral-loss and precursor-ion scan experiments are suggested for the targeted discovery of vitamin E oxidation products and metabolites. To illustrate the latter, oxidation products were detected in human blood and in fish oil. The presented data suggest that this method will help to expand the number of quantified/discovered vitamin E constituents in food products and human/animal trials in order to give a more comprehensive picture to nutritionists about the fate of vitamin E. ACKNOWLEDGMENT The authors would like to thank Dr. A. Ross for providing the tocotrienol standards, Dr. J. Wang for providing the fish oil, Prof. G. Williamson for the biochemistry consultations, and Prof. M. G. Traber and Dr. S. Leonard for the useful information exchange on analytical questions.

Received for review April 27, 2007. Accepted July 12, 2007. AC0708689