- Email: [email protected]
Rapid baseline-separation of all eight tocopherols and tocotrienols by reversed-phase liquid-chromatography with a solid-core pentafluorophenyl column and their sensitive quantification in plasma and liver
Journal of Chromatography A, 1243 (2012) 39–46
Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Rapid baseline-separation of all eight tocopherols and tocotrienols by reversed-phase liquid-chromatography with a solid-core pentafluorophenyl column and their sensitive quantification in plasma and liver Nadine Grebenstein, Jan Frank ∗ Institute of Biological Chemistry and Nutrition, University of Hohenheim, D-70599 Stuttgart, Germany
a r t i c l e
i n f o
Article history: Received 27 February 2012 Received in revised form 11 April 2012 Accepted 12 April 2012 Available online 20 April 2012 Keywords: Baseline separation Pentafluorophenyl (PFP) Reversed-phase liquid chromatography Tocopherols Tocotrienols
a b s t r a c t Of the eight natural vitamin E congeners (␣-, -, ␥-, and ␦-tocopherol and ␣-, -, ␥-, and ␦-tocotrienol), the non-␣-tocopherol congeners have unique biological properties that may contribute to human health. Their study in vivo has been complicated by the lack of a simple analytical method that completely resolves and sensitively detects all eight natural tocopherols and tocotrienols in biological matrices. We thus developed and validated (according to the FDA guidelines for bioanalytical method validation) the first reversed-phase liquid chromatographic method for the baseline-separation and quantification of all eight tocopherols and tocotrienols. Analytes were extracted from human plasma or mouse liver and separated on a Phenomenex Kinetex PFP column (2.6 m, 150 × 4.6 mm) by elution with methanol:water (85:15, vol/vol) at a flow rate of 0.8 mL/min. The developed RP-LC method used a solid-core pentafluorophenyl stationary phase and achieved baseline separation of all eight vitamin E congeners within 15 min at a backpressure of 23 MPa, which is suitable for most conventional HPLC systems. The method was fast, linear, accurate, and precise with detection limits of 27–156 pg and good recoveries (82–122%) for all analytes. In conclusion, we developed and validated the first RP-LC method for baseline resolution of all eight tocopherols and tocotrienols extracted from plasma and liver, which should be useful for the quantification of individual vitamin E congeners in large epidemiological studies and randomized controlled trials. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Of the eight natural substances exerting vitamin E activity, namely ␣-, -, ␥-, and ␦-tocopherol and ␣-, -, ␥-, and ␦-tocotrienol (Fig. 1), ␣-tocopherol has been studied the most [1,2] since the discovery of vitamin E in 1922 [3]. The non-␣-tocopherol vitamin E congeners, however, have unique biological properties that may contribute to human health. The tocotrienols, in particular, have attracted the attention of nutritionists and medical researchers as chemopreventive [4] and neuroprotective [5] dietary agents. The in vivo-study of non-␣-tocopherol congeners of vitamin E has been hampered by the lack of a fast analytical method that completely resolves and sensitively detects all eight natural tocopherols (T) and tocotrienols (T3 ) in biological matrices. The separation of the - and ␥-congeners (Fig. 1), which differ in the
∗ Corresponding author at: Institute of Biological Chemistry and Nutrition, University of Hohenheim, Garbenstr. 30, D-70599 Stuttgart, Germany. Tel.: +49 711 459 24459; fax: +49 711 459 24540. E-mail address: [email protected] (J. Frank). URL: http://www.nutrition-research.de (J. Frank). 0021-9673/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2012.04.042
position of only one methyl group on the chromanol ring [2], has been the major challenge in liquid chromatography. Therefore, only the ␥-congeners or the sum of - and ␥-congeners is often reported [5]. In the past, baseline-separation of all eight T and T3 has only been achieved by normal-phase (NP-LC) but, to the best of our knowledge, not by reversed-phase liquid chromatography (RP-LC) [6,7]. RP-LC, however, has certain advantages over NP-LC, such as better compatibility with the predominantly aqueous biological sample matrices, the use of less harmful solvents, and better stability and durability of the analytical columns. Chromatographic separation of up to six T and T3 has been achieved by RP-LC using gradient elution, C30 -stationary phases, and/or two columns arranged in series; the - and ␥-congeners, however, remained unresolved [7–11]. Even the most recent publication, describing the use of a novel RP-LC octadecylsilane stationary phase, reports only partial resolution of the - and ␥-tocopherols and -tocotrienols and a total runtime of 62 min [12]. By using a bonded pentafluorophenyl (PFP) phase, Richheimer and co-workers [13] achieved nearly complete separation of T and ␥T by RP-LC. Their method, however, did not include the T3 and required about 20 min for the analysis of the four T alone [13]. More recently,
40
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
Fig. 1. Chemical structures and numbers and positions of the methyl groups substituted at the chromanol head of tocopherols and tocotrienols.
solid-core PFP phases have been developed and are now commercially available. Given the previous superior performance of the PFP stationary phase in RP-LC resolution of the - and ␥-congeners, we set out to develop and validate a rapid and sensitive analytical method for the reversed-phase baseline-separation of all eight vitamin E congeners on a solid-core PFP stationary phase and their sensitive and rapid quantification in plasma and liver. In order to allow the adaptation of the developed method in many laboratories, we used routine instrumentation available in most analytical laboratories, namely a standard (>15 year-old) HPLC system coupled with a fluorescence detector. 2. Materials and methods 2.1. Chemicals and stock solutions Methanol, n-hexane, glacial acetic acid, and butylated hydroxytoluene (BHT) were purchased from Carl Roth. Individual stock solutions (10 mmol/L) of authentic ␣-, -, ␥-, or ␦-tocopherol and ␣, -, ␥-, or ␦-tocotrienol (Sigma-Aldrich) were prepared in ethanol and concentrations confirmed photometrically. Identical aliquots of the eight individual stock solutions were combined to prepare a standard solution containing all eight T and T3 at a concentration of 1.25 mmol/L each. The ␣T, ␥T, ␣T3 , T3 , ␥T3 , and ␦T3 standards were ≥97% pure, T was ≥99% and ␦T ≥95.5% pure. 2.2. Sample preparation Liver tissue (200 mg) was placed in a glass tube with screw-cap (on ice) and 2 mL of 1% ascorbic acid in ethanol (wt/vol), 900 L H2 O, and 300 L saturated potassium hydroxide were added. Samples were saponified at 70 ◦ C in a shaking water bath for 30 min and cooled on ice. Twenty-five L ethanolic butylated hydroxytoluene (1 mg BHT/mL ethanol), 1 mL H2 O, 300 L glacial acetic acid, and 2 mL n-hexane were added, test tubes closed, and mixed for 1 min by hand inversion. Plasma samples were processed without saponification and therefore 100 L plasma were added to a glass tube with screwcap (on ice) and 2 mL of 1% ascorbic acid in ethanol (wt/vol), 900 L H2 O, 25 L ethanolic BHT, and 2 mL n-hexane were added, the test tube closed, and mixed for 1 min by hand inversion. Both liver and plasma samples were centrifuged at low speed to facilitate phase separation and an aliquot of the supernatant (1.5 mL) was transferred to a fresh test tube. The extraction was repeated with an additional 2 mL n-hexane, 1-min hand-inversion, low-speed centrifugation (1000 rpm, 3 min), and transfer of 1.5 mL supernatant. The combined extracts were dried under vacuum with a centrifugal evaporator (Savant SpeedVac), the residues
resuspended in 100 L methanol/H2 O (85:15, vol/vol), and 10 L injected into the HPLC system. 2.3. Liquid chromatography The chromatographic system consisted of a JASCO autosampler AS-950, pump PU-980, ternary-gradient unit LG-980-02, 3-line degasser DG-1580-53, and a fluorescence detector 3120-FP interfaced to a personal computer with an LC-Net II/ADC module (JASCO Labor- & Datentechnik). HPLC vials in the autosampler were cooled with a custom-made aluminium sample rack that was constantly purged with 6 ◦ C-cold water using a Haake Fisons K15/DC3 thermostat. Tocopherols and tocotrienols were separated on a Phenomenex KinetexTM PFP column (2.6 m, 150 × 4.6 mm; Phenomenex) using methanol/H2 O (85:15, vol/vol) as eluent at a flow rate of 0.8 mL/min. The fluorescence detector was operated at an excitation wavelength of 296 nm and emission wavelength of 325 nm. Peaks were recorded and integrated using the chromatography software ChromPass II (version 1.8.6.1; JASCO). T and T3 were quantified against external standard curves using authentic compounds (Sigma-Aldrich). 2.4. Method validation The analytical method was validated in terms of selectivity, linearity, limit of detection (LOD), lower and upper limits of quantification (LLOQ and ULOQ, respectively), analyte stability, recovery, accuracy, and precision according to the guidelines for bioanalytical method validation of the Center for Drug Evaluation and Research of the U.S. Food and Drug Administration [14]. Spiked samples for method validation (at low, medium, and high concentrations; 0.625, 3.125, and 6.25 mol/L, respectively) were prepared by directly adding appropriate aliquots of the mixed standard solution to human plasma or liver samples (mashed with a pestle). 2.4.1. Selectivity To confirm the absence of co-eluting or interfering peaks, blank and spiked plasma samples from five different species (human, pig, rat, mouse, and guinea pig) and liver samples from three different species (rat, mouse, and guinea pig) were extracted and injected into the HPLC system. 2.4.2. Linearity The linearity of the detector response for standard solutions was tested on eight consecutive days by injection of 10 L from equimolar mixtures of all eight vitamin E congeners at concentrations of 0.025, 0.0625, 0.125, 0.625, 1.25, 3.125, 6.25, 12.5, and 25 mol/L each. The linearity of the detector response for T and T3 added at
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
41
Fig. 2. Representative chromatograms of a calibration standard (A), of blank plasma (human, pig, rat, mouse, and guinea pig) and liver (rat, mouse, and guinea pig) samples (overlay; B), and of human plasma and mouse liver samples (C) spiked with medium concentrations (3.125 mol/L each) of ␣-, -, ␥-, and ␦-tocopherols and -tocotrienols and analysed using the developed protocol.
these concentrations to human plasma and mouse liver was investigated on three independent days.
amounts of the respective analytes that could be quantified with the required accuracy and precision (see below).
2.4.3. Limits of detection and quantification The limit of detection (LOD) was defined as the amount of the respective analyte injected into the HPLC system that could be reliable discerned from the background noise (ca. 3 times the background signal). The lower and upper limits of quantification (LLOQ and ULOQ, respectively) were defined as the lowest and highest
2.4.4. Stability Stock solution-stability was tested by comparing the detector response after injection of an appropriate dilution of the T and T3 stock solution with that of the same solution after storage at room temperature or −20 ◦ C, respectively, for 24 h in five replicate analyses. Short-term stabilities of the analytes in the sample
27 (69) 92 (231) 31 (77) 41 (99) 135 (330) 45 (110) 50 (123) 168 (409) 56 (136) 156 (366) 519 (1221) 173 (407) 86 (214) 287 (714) 96 (238) 48 (115) 160 (385) 53 (128) a
y = peak area [mV min], x = concentration [mol/L].
84 (202) 281 (674 94 (225) Limits of detection and quantification 147 (342) LOD, pg (fmol) 491 (1140) LLOQ, pg (fmol) 164 (380) ULOQ, ng (pmol)
0.9992 ± 0.0006 y = 16.180x − 0.591 0.9987 ± 0.0007 y = 12.080x − 1.082 0.9991 ± 0.0005 y = 10.070x − 1.478 0.9968 ± 0.0012 y = 4.188x − 2.044 0.9997 ± 0.0002 y = 15.690x − 0.681 0.9988 ± 0.0008 y = 8.888x − 1.246 0.9960 ± 0.0024 y = 6.959x − 4.469 Mouse liver [r2 ] Slope and y-intercepta
0.9983 ± 0.0004 y = 15.600x − 2.411
0.9946 ± 0.0013 y = 16.190x − 0.002 0.9980 ± 0.0013 y = 11.770x − 0.311 0.9976 ± 0.0019 y = 9.688x − 0.489 0.9967 ± 0.0010 y = 3.788x − 1.022 0.9999 ± 0.0001 y = 16.850x − 0.018 0.9979 ± 0.0014 y = 8.446x − 0.435 0.9872 ± 0.0086 y = 4.599x − 2.570 Human plasma [r2 ] Slope and y-intercepta
0.9924 ± 0.0014 y = 14.690x − 0.988
0.9991 ± 0.0004 y = 16.178x − 0.591 0.9989 ± 0.0007 y = 12.071x − 1.037 0.9994 ± 0.0003 y = 10.067x − 1.478 0.9976 ± 0.0009 y = 4.188x − 2.044 0.9996 ± 0.0004 y = 15.702x − 0.726 0.9990 ± 0.0007 y = 15.599x − 2.411 0.9991 ± 0.0005 y = 8.885x − 1.227 0.9974 ± 0.0009 y = 6.959x − 4.469
␥  ␣ ␦ ␥ 
Given the lack of a fast and sensitive reversed-phase liquid chromatographic method for the separation and quantification of all eight natural vitamin E congeners in biological tissues, we set out to develop and validate such a procedure. We further intended to make use of an approach that could be adapted by many researchers and laboratories without the need to invest in new and expensive instruments. We therefore modified existing and simple liquid–liquid extraction protocols [15–17] and made use of a more than 15 y-old HPLC system coupled with a fluorescence detector, equipment that is available in most analytical laboratories. In order to achieve complete separation of all T and T3 , especially the respective - and ␥-congeners, we used a stationary phase that represents a further-development from the previously most successful RP-column used for the separation of - and ␥-tocopherols, namely a pentafluorophenyl (PFP) column [13]. The novel PFP stationary phase used here has a solid core and a porous shell and offers reduced diffusion of the analytes through the particles, better masstransfer, and thus better resolution at a lower backpressure than conventional stationary phases. The reduced backpressure allows the use of small particle sizes, in this case 2.6 m particle diameter, in combination with comparably long columns (here 150 mm) together with conventional HPLC systems instead of the more expensive ultra-high-performance liquid chromatography systems usually required. The developed chromatographic procedure operated at a backpressure of 23 MPa, which is feasible for standard HPLC systems, and achieved baseline-separation of all eight T and T3 in less than 15 minutes (Fig. 2), which is shorter than some of the fastest NP-LC techniques reported [6,7]. The flow rate was set to 0.8 mL/min as this resulted in baseline separation of all eight analytes and separated the first peak from the front of poorly retained compounds at the beginning of the chromatogram (Figs. 2 and 4).
␣
3. Results and discussion
Linearity Standard solutions [r2 ] Slope and y-intercepta
2.4.6. Accuracy and precision Intraday accuracy and precision were determined by analysing five human plasma and five mouse liver samples that were spiked with low (0.625 mol/L), medium (3.125 mol/L), and high (6.25 mol/L) concentrations, respectively, and nine standard mixtures (ranging from 0.025 to 25 mol/L for each vitamin E congener). Concentrations of T and T3 were calculated and compared to the known concentrations in the samples. Accordingly, interday precision and accuracy were determined in human plasma and mouse liver at low, medium, and high concentrations on three independent days.
Tocotrienols
2.4.5. Recovery The recovery of the analytes was quantified by five independent analyses of human plasma and mouse liver samples, respectively, spiked with tocopherols and tocotrienols at low, medium, and high (0.625, 3.125, and 6.250 mol/L, respectively) concentrations and by comparison of the detector responses with those of standards containing identical amounts of the analytes.
Tocopherols
matrices were evaluated in triplicate in human plasma and mouse liver samples spiked with low (0.625 mol/L) or high (6.25 mol/L) concentrations of T and T3 after storage at room temperature for 24 h. Freeze-and-thaw stability was investigated in the spiked (low and high) plasma and liver samples by comparing detector responses after analysis of the freshly prepared samples with that after storage at −80 ◦ C for 24 h, thawing to room temperature, and repeated freezing to −80 ◦ C, which was repeated thrice (three freeze–thaw-cycles). Post-preparative stability was assessed by repeated injection from the same vial of four separate extracts from spiked (medium analyte concentrations) plasma and liver samples at 0, 6, 12, 18, and 24 h.
␦
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46 Table 1 Linearity [r2 ± SD], limit of detection (LOD) and lower (LLOQ) and upper limits of quantification (ULOQ) of the developed reversed-phase HPLC method for the quantification of tocopherols and tocotrienols (given as absolute amount injected dissolved in 10 L).
42
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
43
Table 2 Stability [% of nominal concentration] of tocopherols and tocotrienols in stock solutions or added at low (0.625 mol/L) or high concentrations (6.25 mol/L) to human plasma and mouse liver and stored at room temperature (RT) for 24 h, and during three freeze-and-thaw cycles (−80 ◦ C/room temperature). Tocopherols ␣
Tocotrienols 
␥
␦
Stock solution stability, 24 h [% nominal concentration] 96.3 100.5 RT −20 ◦ C 98.8 98.4
100.5 98.3
101.6 98.8
98.4 96.8
98.3 96.4
97.4 95.8
101.2 98.8
Human plasma Short-term stability, 24 h, RT [% nominal concentration] 97.8 98.7 Low 98.7 High 106.6
101.5 111.8
98.4 97.8
96.9 107.9
98.3 106.2
103.6 105.7
94.7 101.1
69.6 75.4
74.1 77.0
71.2 79.4
70.1 73.0
74.2 73.6
71.5 59.3
108.2 105.5
107.2 104.4
98.2 95.6
95.1 106.5
95.1 102.0
117.8 94.8
75.2 72.1
70.1 71.1
69.8 70.6
77.7 74.7
81.1 73.2
76.6 74.1
Freeze-and-thaw stability [% nominal concentration] 88.9 74.7 Low 87.1 76.8 High Mouse liver Short-term stability, 24 h, RT [% nominal concentration] 104.6 92.5 Low 104.6 106.5 High Freeze-and-thaw stability [% nominal concentration] 74.6 Low 78.3 High 81.9 76.9
Increasing the flow rate above 0.8 mL/min resulted in a loss of baseline resolution of the - and ␥-congeners. We decided against the use of an internal standard, because the resolution of eight analytes was already considered challenging and the addition of a ninth compound with similar polarity and physicochemical properties would have interfered with the separation of the analytes and might have substantially increased the total runtime of our method. Therefore, we favoured the use of authentic T and T3 as external standards, as they were required for peak identification either way. To the best of our knowledge, this is the first time complete baseline-resolution of all eight T and T3 , including the respective - and ␥-congeners, has been achieved on a RP-stationary phase. The only paper reporting partial resolution of the eight natural vitamin E congeners on a RP-column required gradient elution and a total runtime of a little over 1 h [12]. Hence, our method offers a significant (4-times) faster analysis as well as a simpler and less error-prone isocratic elution. We validated the newly developed method according to the guidelines for bioanalytical method validation of the Center for Drug Evaluation and Research of the U.S. Food and Drug Administration [14]. 3.1. Method validation 3.1.1. Selectivity Injection of plasma samples from five different species (human, pig, rat, mouse, guinea pig) and liver samples from three species (rat, mouse, and guinea pig) revealed no peaks interfering with the eight T and T3 under the selected chromatographic conditions (Fig. 2). Unfortunately, due to the widespread presence of ␣T and ␥T in human and animal diets, entirely vitamin E-free plasma or liver samples were not available. Therefore, minor ␣T and ␥T peaks are present in the blank samples; their identities could be confirmed by spiking experiments. Both plasma and liver samples contained a number of unidentified peaks; one compound eluted directly after ␥T3 and one shortly before ␦T, both, however, were separated from the analytes and thus did not interfere with the analysis (Fig. 2). 3.1.2. Linearity and limits of detection and quantification The detector responses for T and T3 standard solutions as well as extracts from human plasma or mouse liver were linear with a
␣

␥
␦
r2 > 0.99 for all eight analytes except for ␣T extracted from human plasma (r2 > 0.98; Table 1). Detector responses were linear over a wide range starting at low picogram (LOD) and up to low nanogram amounts (ULOQ; Table 1). The developed method was very sensitive with limits of detection for all vitamin E congeners in the range of 27–156 pg per injection (Table 1), which is an order of magnitude lower than the LOD (5–15 ng) reported for another recently developed RP-LC method with fluorescence detection [12]. The LOD of the ␣-congeners were generally higher due to their lower fluorescence intensity compared to the -, ␥-, and ␦-congeners, their concentrations in humans [1,5,18,19] and animals [20–22], however, are usually higher. The LOD and LOQ reported here refer to the sensitivity of the instrument and are given as the amount of analyte required with each injection (pg per injection). The respective values for plasma and liver samples can then be estimated by adjusting the given LOD and LOQ for the respective recoveries of the analytes (Table 3). 3.1.3. Stability Stock solutions stored at room temperature or −20 ◦ C for 24 h remained stable with <5% degradation observed (Table 2). The short-term (bench-top) stabilities of all congeners added at low (0.625 mol/L) or high (6.25 mol/L) concentrations to human plasma or mouse liver and stored for 24 h at room temperature were good with <5.3% maximum degradation observed in plasma and <7.5% in liver samples (Table 2). T and T3 added to plasma or liver were less resistant to repeated freezing (−80 ◦ C) and thawing (to room temperature) with typical losses of 10–25% after three freeze–thaw-cycles and maximum degradation of 40.7% observed for ␦T3 in plasma and 30.2% for ␣T3 in liver (Table 2). For accurate quantification of vitamin E it is thus advisable to thaw frozen samples only once and avoid re-freezing and re-thawing. In order to assess post-preparative stability, vitamin E was extracted from human plasma or mouse liver and the prepared samples were left in a chilled autosampler (ca. 6 ◦ C) for typical HPLC residence times. All eight congeners extracted from plasma or liver remained stable over 24 h (Fig. 3). 3.1.4. Recovery The intraday recoveries for T and T3 extracted from human plasma or mouse liver were generally very good and within the limits set by the FDA (≤15% deviation from the expected value
44
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
Fig. 3. Post-preparative stability was assessed by comparing the peak areas [mV min], after 0, 6, 12, 18, or 24 h residence time in the autosampler, of tocopherols (T) and tocotrienols (T3 ) added to human plasma or mouse liver samples (3.125 mol/L each), respectively, and extracted using the developed protocol.
at medium and high and ≤20% at low concentrations) [14], with few exceptions. The recovery of ␣T3 from human plasma at the medium concentration and of ␣T and ␦T3 from mouse liver at low and medium concentrations, respectively, exceeded these limits by 2–3% (Table 3). The interday recoveries too were within the FDA limits, with the only exception of T added at high concentrations to plasma, which was 16% lower than the known concentration and thus exceeded the limits by 1% (Table 3).
Intraday precision was very good; only for ␣T added at medium concentrations to plasma and at high concentrations to liver were the FDA limits not met. Interday precision was excellent with no values outside of the expected range (Table 3). 3.1.6. Versatility The developed RP-LC method offers outstanding versatility and can be adapted for other biological matrices and detection systems. For example, we were able to separate and quantify the concentrations of T and T3 in brains of guinea pigs fed a T3 -rich rice bran extract (Fig. 4A; guinea pig brain tissue was kindly donated by Dr. Gunter P. Eckert, University of Frankfurt, Germany) using the method developed for liver samples. As expected, ␣T was the major congener in the brain, while the remaining T, ␣T3 , and ␦T3 were present at low concentrations and T3 in trace amounts. T and ␥T were baseline-separated in the guinea pig brain samples (Fig. 4A). With slight modifications to the sample preparation, T and T3 could also be quantified in plant foods such as amaranth, almonds, and hazelnuts (Fig. 4B). Because of the significantly higher
3.1.5. Accuracy and precision The intraday accuracies were generally good and largely within the FDA limits at low and high concentrations (exception: T in plasma). At medium concentrations, however, the intraday accuracies of ␣T and ␦T added to plasma exceeded the set limits significantly, while ␥T, ␣T3 , and ␦T3 exceeded them to a small degree (Table 3). The interday accuracies, on the other hand, were much better and only two congeners added at high concentrations to plasma (T) or liver (T3 ) marginally exceeded the FDA limits (Table 3).
Table 3 Intra- and interday recoveries, accuracies, and precisions of tocopherols and tocotrienols extracted from human plasma or mouse liver spiked at low, medium, or high concentrations (0.625, 3.125, or 6.250 mol/L, respectively; n = 5 for each concentration). Tocopherols ␣  Human plasma
Tocotrienols
Tocopherols
Tocotrienols
␥
␦
␣

␥
␦
␣  Mouse liver
␥
␦
␣

␥
␦
Intraday recovery [%] 101 Low 101 Medium 91 High
86 107 91
97 116 92
86 104 102
91 82 101
96 87 111
103 88 92
107 99 101
122 96 88
97 94 90
109 97 94
92 87 95
95 106 109
98 110 114
99 111 101
103 117 95
Interday recovery [%] 86 Low 95 Medium 93 High
86 90 84
98 89 92
96 86 93
100 90 92
90 96 96
93 93 91
100 88 98
83 98 95
83 97 86
100 102 93
84 89 92
93 89 99
86 97 95
96 99 93
99 91 98
Intraday accuracy [bias%] 19.3 19.3 Low 26.7 8.2 Medium 11.3 18.7 High
1.7 15.7 8.3
8.3 18.4 9.3
12.2 16.5 11.2
7.4 10.1 10.5
12.2 11.0 6.6
0.2 15.3 7.4
20.0 8.1 5.4
19.2 9.3 14.3
2.8 8.2 9.4
19.8 15.3 10.4
13.3 15.8 2.0
20.4 10.7 11.8
11.6 7.6 10.3
11.4 15.1 7.2
Interday accuracy [bias%] 13.6 14.1 Low 4.8 10.5 Medium 7.5 16.2 High
2.5 10.8 7.6
3.7 13.9 7.2
0.0 9.7 8.0
9.9 4.6 4.5
6.8 7.0 8.6
0.3 12.1 2.1
17.3 2.5 5.1
16.7 2.9 13.7
0.4 1.7 7.0
15.9 11.0 8.4
7.4 11.5 1.1
14.0 3.3 15.5
4.1 1.1 7.5
0.6 9.1 1.7
Intraday precision [% CV] 8.8 Low 12.9 Medium 17.9 9.0 14.5 3.8 High
12.3 6.7 3.3
8.6 6.0 3.1
17.4 10.9 6.3
17.6 6.9 10.3
7.2 6.0 8.1
10.8 5.1 8.5
4.9 15.0 17.0
6.0 10.3 6.7
5.0 12.3 3.5
5.1 7.1 3.3
12.5 5.7 11.1
8.5 10.2 14.3
11.9 11.9 4.3
15.4 9.3 8.9
Interday precision [% CV] 4.8 Low 9.2 5.7 3.8 Medium 13.9 2.9 High
9.4 3.5 3.3
8.2 1.9 2.3
9.1 8.4 6.3
6.0 1.5 8.9
1.1 4.4 4.5
8.9 4.3 6.4
4.4 8.9 14.5
6.5 8.2 7.4
4.1 2.1 1.6
2.2 2.9 1.2
13.1 2.6 8.6
1.4 5.2 14.9
4.0 11.3 3.4
10.0 7.1 7.0
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
45
Fig. 4. Representative chromatograms obtained by injection of extracts from (A) brain tissue of a guinea pig fed a tocotrienol-rich rice bran extract or a low calibration standard and (B) amaranth, almonds, or hazelnuts. Note that the retention times vary slightly because the samples were analysed on different columns at different times.
concentrations of T and T3 in the food samples, only 10 mg of amaranth, almond, or hazelnut were used (compared to the 200 mg liver tissue required). Amaranth samples were used without further processing, while almonds and hazelnuts were roughly ground prior to extraction. The food samples were saponified (similar to liver samples) and a third extraction step was added for complete extraction of the analytes. To confirm complete extraction, a fourth hexane extraction was performed and the dried and re-dissolved (in mobile phase) residue was injected into the HPLC, and no T or T3 were detected. For plasma and liver samples, on the other hand, two extraction steps sufficed for complete extraction of the analytes. Due to the high fat content of almonds and hazelnuts, it was necessary to add twice as much saturated potassium hydroxide solution and water (600 and 1800 L, respectively) to the samples in order to facilitate removal of the saponifiable lipids. Accordingly, the volume of glacial acetic acid required for acidification of the samples after saponification was doubled to 600 L. The samples, however, were still rather dirty and further sample clean up or prior lipid extraction might be required for lipid-rich food samples. In the amaranth samples, an unidentified peak eluted just before ␦T, which might require slight adjustments to the water content of the mobile phase, if the method is to be adapted for routine analyses of amaranth or comparable plant food samples. Nevertheless, the - and ␥-tocopherols present in the almond and hazelnut samples were still resolved by our newly developed method (Fig. 4B). For our mobile phase, we chose a methanol:water ratio of 85:15 as the best compromise between resolution and speed of analysis for all analysed sample matrices (Figs. 2 and 4). As mentioned before, if unexpected interfering peaks are observed, increasing the water content of the mobile phase will lead to longer retention times and even better resolution of the analytes, and allows the adaptation of the chromatographic procedure to many other sample matrices, including additional animal and human tissue
samples or plant and food materials. Such changes, however, require appropriate validation of the modified method. In our hands, both fluorescence and coulometric electrochemical detection (with an ESA 8-channel CoulArray) were compatible with the developed method. In order to obtain the required conductivity of the mobile phase for electrochemical detection, 30 mmol/L lithium perchloride were added and baseline separation of all eight vitamin E congeners was achieved (not shown). Our RP-LC method should also be compatible with ultraviolet, photodiode array, and mass detectors, although we did not confirm this yet. 4. Conclusions In the past, the concentrations in biological tissues of - and ␥-tocopherols and - and ␥-tocotrienols, respectively, could only be reported as their respective sums, not the concentrations of the individual compounds. With the development of this rapid, sensitive, accurate, precise, and – importantly – versatile reversed-phase liquid chromatographic method it is now possible to determine the concentrations of all eight natural vitamin E congeners in biological tissues even in large epidemiological studies, randomized controlled trials, and animal experiments using a simple, fast, and reproducible liquid–liquid extraction. The developed method should thus be useful for researchers wishing to investigate the biological availabilities, pharmacokinetics, and/or biological activities of the individual vitamin E congeners and in particular of the respective - and ␥-congeners of tocopherols and tocotrienols. Acknowledgment The authors gratefully acknowledge financial support by the German Research Foundation (DFG) by means of grant no. 2478/4-1.
46
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13]
J. Frank, G. Rimbach, Aktuel. Ernaehr. Med. 34 (2009) 131. A. Kamal-Eldin, L.Å. Appelqvist, Lipids 31 (1996) 671. H.M. Evans, K.S. Bishop, Science 56 (1922) 650. C.K. Sen, S. Khanna, S. Roy, Mol. Aspects Med. 28 (2007) 692. J. Frank, X.W. Chin, C. Schrader, G.P. Eckert, G. Rimbach, Ageing Res. Rev. 11 (2012) 163. A. Kamal-Eldin, S. Gorgen, J. Pettersson, A.-M. Lampi, J. Chromatogr. A 881 (2000) 217. S.L. Abidi, J. Chromatogr. A 881 (2000) 197. B.L. Lee, A.L. New, C.N. Ong, Clin. Chem. 49 (2003) 2056. J. Gruszka, J. Kruk, Chromatographia 66 (2007) 909. M.N. Irakli, V.F. Samanidou, I.N. Papadoyannis, J. Sep. Sci. 34 (2011) 1375. M.H. Chen, C.J. Bergman, J. Food Compos. Anal. 18 (2005) 139. M.N. Irakli, V.F. Samanidou, I.N. Papadoyannis, J. Agric. Food Chem. 60 (2012) 2076. S.L. Richheimer, M.C. Kent, M.W. Bernart, J. Chromatogr. A 677 (1994) 75.
[14] Center for Drug Evaluation and Research (CDER) at the Food and Drug Administration (2001). http://www.fda.gov/downloads/Drugs/GuidanceCompliance RegulatoryInformation/Guidances/UCM070107.pdf. [15] M. Podda, C. Weber, M.G. Traber, L. Packer, J. Lipid Res. 37 (1996) 893. [16] K. Augustin, R. Blank, C. Boesch-Saadatmandi, J. Frank, S. Wolffram, G. Rimbach, J. Anim. Physiol. Anim. Nutr. 92 (2008) 705. [17] S. Gaedicke, X. Zhang, P. Huebbe, C. Boesch-Saadatmandi, Y. Lou, I. Wiswedel, A. Gardemann, J. Frank, G. Rimbach, Br. J. Nutr. 102 (2009) 398. [18] J. Frank, T.W. George, J.K. Lodge, A.M. Rodriguez-Mateos, J.P. Spencer, A.M. Minihane, G. Rimbach, J. Nutr. 139 (2009) 58. [19] J. Frank, S. Lee, S.W. Leonard, J.K. Atkinson, A. Kamal-Eldin, M.G. Traber, Am. J. Clin. Nutr. 87 (2008) 1723. [20] J. Frank, A. Budek, T. Lundh, R.S. Parker, J.E. Swanson, C.F. Lourenco, B. Gago, J. Laranjinha, B. Vessby, A. Kamal-Eldin, J. Lipid Res. 47 (2006) 2718. [21] J. Frank, C. Eliasson, D. Leroy-Nivard, A. Budek, T. Lundh, B. Vessby, P. Åman, A. Kamal-Eldin, Br. J. Nutr. 92 (2004) 169. [22] J. Frank, T. Lundh, R.S. Parker, J.E. Swanson, B. Vessby, A. Kamal-Eldin, J. Nutr. 133 (2003) 3195.
Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Rapid baseline-separation of all eight tocopherols and tocotrienols by reversed-phase liquid-chromatography with a solid-core pentafluorophenyl column and their sensitive quantification in plasma and liver Nadine Grebenstein, Jan Frank ∗ Institute of Biological Chemistry and Nutrition, University of Hohenheim, D-70599 Stuttgart, Germany
a r t i c l e
i n f o
Article history: Received 27 February 2012 Received in revised form 11 April 2012 Accepted 12 April 2012 Available online 20 April 2012 Keywords: Baseline separation Pentafluorophenyl (PFP) Reversed-phase liquid chromatography Tocopherols Tocotrienols
a b s t r a c t Of the eight natural vitamin E congeners (␣-, -, ␥-, and ␦-tocopherol and ␣-, -, ␥-, and ␦-tocotrienol), the non-␣-tocopherol congeners have unique biological properties that may contribute to human health. Their study in vivo has been complicated by the lack of a simple analytical method that completely resolves and sensitively detects all eight natural tocopherols and tocotrienols in biological matrices. We thus developed and validated (according to the FDA guidelines for bioanalytical method validation) the first reversed-phase liquid chromatographic method for the baseline-separation and quantification of all eight tocopherols and tocotrienols. Analytes were extracted from human plasma or mouse liver and separated on a Phenomenex Kinetex PFP column (2.6 m, 150 × 4.6 mm) by elution with methanol:water (85:15, vol/vol) at a flow rate of 0.8 mL/min. The developed RP-LC method used a solid-core pentafluorophenyl stationary phase and achieved baseline separation of all eight vitamin E congeners within 15 min at a backpressure of 23 MPa, which is suitable for most conventional HPLC systems. The method was fast, linear, accurate, and precise with detection limits of 27–156 pg and good recoveries (82–122%) for all analytes. In conclusion, we developed and validated the first RP-LC method for baseline resolution of all eight tocopherols and tocotrienols extracted from plasma and liver, which should be useful for the quantification of individual vitamin E congeners in large epidemiological studies and randomized controlled trials. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Of the eight natural substances exerting vitamin E activity, namely ␣-, -, ␥-, and ␦-tocopherol and ␣-, -, ␥-, and ␦-tocotrienol (Fig. 1), ␣-tocopherol has been studied the most [1,2] since the discovery of vitamin E in 1922 [3]. The non-␣-tocopherol vitamin E congeners, however, have unique biological properties that may contribute to human health. The tocotrienols, in particular, have attracted the attention of nutritionists and medical researchers as chemopreventive [4] and neuroprotective [5] dietary agents. The in vivo-study of non-␣-tocopherol congeners of vitamin E has been hampered by the lack of a fast analytical method that completely resolves and sensitively detects all eight natural tocopherols (T) and tocotrienols (T3 ) in biological matrices. The separation of the - and ␥-congeners (Fig. 1), which differ in the
∗ Corresponding author at: Institute of Biological Chemistry and Nutrition, University of Hohenheim, Garbenstr. 30, D-70599 Stuttgart, Germany. Tel.: +49 711 459 24459; fax: +49 711 459 24540. E-mail address: [email protected] (J. Frank). URL: http://www.nutrition-research.de (J. Frank). 0021-9673/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2012.04.042
position of only one methyl group on the chromanol ring [2], has been the major challenge in liquid chromatography. Therefore, only the ␥-congeners or the sum of - and ␥-congeners is often reported [5]. In the past, baseline-separation of all eight T and T3 has only been achieved by normal-phase (NP-LC) but, to the best of our knowledge, not by reversed-phase liquid chromatography (RP-LC) [6,7]. RP-LC, however, has certain advantages over NP-LC, such as better compatibility with the predominantly aqueous biological sample matrices, the use of less harmful solvents, and better stability and durability of the analytical columns. Chromatographic separation of up to six T and T3 has been achieved by RP-LC using gradient elution, C30 -stationary phases, and/or two columns arranged in series; the - and ␥-congeners, however, remained unresolved [7–11]. Even the most recent publication, describing the use of a novel RP-LC octadecylsilane stationary phase, reports only partial resolution of the - and ␥-tocopherols and -tocotrienols and a total runtime of 62 min [12]. By using a bonded pentafluorophenyl (PFP) phase, Richheimer and co-workers [13] achieved nearly complete separation of T and ␥T by RP-LC. Their method, however, did not include the T3 and required about 20 min for the analysis of the four T alone [13]. More recently,
40
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
Fig. 1. Chemical structures and numbers and positions of the methyl groups substituted at the chromanol head of tocopherols and tocotrienols.
solid-core PFP phases have been developed and are now commercially available. Given the previous superior performance of the PFP stationary phase in RP-LC resolution of the - and ␥-congeners, we set out to develop and validate a rapid and sensitive analytical method for the reversed-phase baseline-separation of all eight vitamin E congeners on a solid-core PFP stationary phase and their sensitive and rapid quantification in plasma and liver. In order to allow the adaptation of the developed method in many laboratories, we used routine instrumentation available in most analytical laboratories, namely a standard (>15 year-old) HPLC system coupled with a fluorescence detector. 2. Materials and methods 2.1. Chemicals and stock solutions Methanol, n-hexane, glacial acetic acid, and butylated hydroxytoluene (BHT) were purchased from Carl Roth. Individual stock solutions (10 mmol/L) of authentic ␣-, -, ␥-, or ␦-tocopherol and ␣, -, ␥-, or ␦-tocotrienol (Sigma-Aldrich) were prepared in ethanol and concentrations confirmed photometrically. Identical aliquots of the eight individual stock solutions were combined to prepare a standard solution containing all eight T and T3 at a concentration of 1.25 mmol/L each. The ␣T, ␥T, ␣T3 , T3 , ␥T3 , and ␦T3 standards were ≥97% pure, T was ≥99% and ␦T ≥95.5% pure. 2.2. Sample preparation Liver tissue (200 mg) was placed in a glass tube with screw-cap (on ice) and 2 mL of 1% ascorbic acid in ethanol (wt/vol), 900 L H2 O, and 300 L saturated potassium hydroxide were added. Samples were saponified at 70 ◦ C in a shaking water bath for 30 min and cooled on ice. Twenty-five L ethanolic butylated hydroxytoluene (1 mg BHT/mL ethanol), 1 mL H2 O, 300 L glacial acetic acid, and 2 mL n-hexane were added, test tubes closed, and mixed for 1 min by hand inversion. Plasma samples were processed without saponification and therefore 100 L plasma were added to a glass tube with screwcap (on ice) and 2 mL of 1% ascorbic acid in ethanol (wt/vol), 900 L H2 O, 25 L ethanolic BHT, and 2 mL n-hexane were added, the test tube closed, and mixed for 1 min by hand inversion. Both liver and plasma samples were centrifuged at low speed to facilitate phase separation and an aliquot of the supernatant (1.5 mL) was transferred to a fresh test tube. The extraction was repeated with an additional 2 mL n-hexane, 1-min hand-inversion, low-speed centrifugation (1000 rpm, 3 min), and transfer of 1.5 mL supernatant. The combined extracts were dried under vacuum with a centrifugal evaporator (Savant SpeedVac), the residues
resuspended in 100 L methanol/H2 O (85:15, vol/vol), and 10 L injected into the HPLC system. 2.3. Liquid chromatography The chromatographic system consisted of a JASCO autosampler AS-950, pump PU-980, ternary-gradient unit LG-980-02, 3-line degasser DG-1580-53, and a fluorescence detector 3120-FP interfaced to a personal computer with an LC-Net II/ADC module (JASCO Labor- & Datentechnik). HPLC vials in the autosampler were cooled with a custom-made aluminium sample rack that was constantly purged with 6 ◦ C-cold water using a Haake Fisons K15/DC3 thermostat. Tocopherols and tocotrienols were separated on a Phenomenex KinetexTM PFP column (2.6 m, 150 × 4.6 mm; Phenomenex) using methanol/H2 O (85:15, vol/vol) as eluent at a flow rate of 0.8 mL/min. The fluorescence detector was operated at an excitation wavelength of 296 nm and emission wavelength of 325 nm. Peaks were recorded and integrated using the chromatography software ChromPass II (version 1.8.6.1; JASCO). T and T3 were quantified against external standard curves using authentic compounds (Sigma-Aldrich). 2.4. Method validation The analytical method was validated in terms of selectivity, linearity, limit of detection (LOD), lower and upper limits of quantification (LLOQ and ULOQ, respectively), analyte stability, recovery, accuracy, and precision according to the guidelines for bioanalytical method validation of the Center for Drug Evaluation and Research of the U.S. Food and Drug Administration [14]. Spiked samples for method validation (at low, medium, and high concentrations; 0.625, 3.125, and 6.25 mol/L, respectively) were prepared by directly adding appropriate aliquots of the mixed standard solution to human plasma or liver samples (mashed with a pestle). 2.4.1. Selectivity To confirm the absence of co-eluting or interfering peaks, blank and spiked plasma samples from five different species (human, pig, rat, mouse, and guinea pig) and liver samples from three different species (rat, mouse, and guinea pig) were extracted and injected into the HPLC system. 2.4.2. Linearity The linearity of the detector response for standard solutions was tested on eight consecutive days by injection of 10 L from equimolar mixtures of all eight vitamin E congeners at concentrations of 0.025, 0.0625, 0.125, 0.625, 1.25, 3.125, 6.25, 12.5, and 25 mol/L each. The linearity of the detector response for T and T3 added at
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
41
Fig. 2. Representative chromatograms of a calibration standard (A), of blank plasma (human, pig, rat, mouse, and guinea pig) and liver (rat, mouse, and guinea pig) samples (overlay; B), and of human plasma and mouse liver samples (C) spiked with medium concentrations (3.125 mol/L each) of ␣-, -, ␥-, and ␦-tocopherols and -tocotrienols and analysed using the developed protocol.
these concentrations to human plasma and mouse liver was investigated on three independent days.
amounts of the respective analytes that could be quantified with the required accuracy and precision (see below).
2.4.3. Limits of detection and quantification The limit of detection (LOD) was defined as the amount of the respective analyte injected into the HPLC system that could be reliable discerned from the background noise (ca. 3 times the background signal). The lower and upper limits of quantification (LLOQ and ULOQ, respectively) were defined as the lowest and highest
2.4.4. Stability Stock solution-stability was tested by comparing the detector response after injection of an appropriate dilution of the T and T3 stock solution with that of the same solution after storage at room temperature or −20 ◦ C, respectively, for 24 h in five replicate analyses. Short-term stabilities of the analytes in the sample
27 (69) 92 (231) 31 (77) 41 (99) 135 (330) 45 (110) 50 (123) 168 (409) 56 (136) 156 (366) 519 (1221) 173 (407) 86 (214) 287 (714) 96 (238) 48 (115) 160 (385) 53 (128) a
y = peak area [mV min], x = concentration [mol/L].
84 (202) 281 (674 94 (225) Limits of detection and quantification 147 (342) LOD, pg (fmol) 491 (1140) LLOQ, pg (fmol) 164 (380) ULOQ, ng (pmol)
0.9992 ± 0.0006 y = 16.180x − 0.591 0.9987 ± 0.0007 y = 12.080x − 1.082 0.9991 ± 0.0005 y = 10.070x − 1.478 0.9968 ± 0.0012 y = 4.188x − 2.044 0.9997 ± 0.0002 y = 15.690x − 0.681 0.9988 ± 0.0008 y = 8.888x − 1.246 0.9960 ± 0.0024 y = 6.959x − 4.469 Mouse liver [r2 ] Slope and y-intercepta
0.9983 ± 0.0004 y = 15.600x − 2.411
0.9946 ± 0.0013 y = 16.190x − 0.002 0.9980 ± 0.0013 y = 11.770x − 0.311 0.9976 ± 0.0019 y = 9.688x − 0.489 0.9967 ± 0.0010 y = 3.788x − 1.022 0.9999 ± 0.0001 y = 16.850x − 0.018 0.9979 ± 0.0014 y = 8.446x − 0.435 0.9872 ± 0.0086 y = 4.599x − 2.570 Human plasma [r2 ] Slope and y-intercepta
0.9924 ± 0.0014 y = 14.690x − 0.988
0.9991 ± 0.0004 y = 16.178x − 0.591 0.9989 ± 0.0007 y = 12.071x − 1.037 0.9994 ± 0.0003 y = 10.067x − 1.478 0.9976 ± 0.0009 y = 4.188x − 2.044 0.9996 ± 0.0004 y = 15.702x − 0.726 0.9990 ± 0.0007 y = 15.599x − 2.411 0.9991 ± 0.0005 y = 8.885x − 1.227 0.9974 ± 0.0009 y = 6.959x − 4.469
␥  ␣ ␦ ␥ 
Given the lack of a fast and sensitive reversed-phase liquid chromatographic method for the separation and quantification of all eight natural vitamin E congeners in biological tissues, we set out to develop and validate such a procedure. We further intended to make use of an approach that could be adapted by many researchers and laboratories without the need to invest in new and expensive instruments. We therefore modified existing and simple liquid–liquid extraction protocols [15–17] and made use of a more than 15 y-old HPLC system coupled with a fluorescence detector, equipment that is available in most analytical laboratories. In order to achieve complete separation of all T and T3 , especially the respective - and ␥-congeners, we used a stationary phase that represents a further-development from the previously most successful RP-column used for the separation of - and ␥-tocopherols, namely a pentafluorophenyl (PFP) column [13]. The novel PFP stationary phase used here has a solid core and a porous shell and offers reduced diffusion of the analytes through the particles, better masstransfer, and thus better resolution at a lower backpressure than conventional stationary phases. The reduced backpressure allows the use of small particle sizes, in this case 2.6 m particle diameter, in combination with comparably long columns (here 150 mm) together with conventional HPLC systems instead of the more expensive ultra-high-performance liquid chromatography systems usually required. The developed chromatographic procedure operated at a backpressure of 23 MPa, which is feasible for standard HPLC systems, and achieved baseline-separation of all eight T and T3 in less than 15 minutes (Fig. 2), which is shorter than some of the fastest NP-LC techniques reported [6,7]. The flow rate was set to 0.8 mL/min as this resulted in baseline separation of all eight analytes and separated the first peak from the front of poorly retained compounds at the beginning of the chromatogram (Figs. 2 and 4).
␣
3. Results and discussion
Linearity Standard solutions [r2 ] Slope and y-intercepta
2.4.6. Accuracy and precision Intraday accuracy and precision were determined by analysing five human plasma and five mouse liver samples that were spiked with low (0.625 mol/L), medium (3.125 mol/L), and high (6.25 mol/L) concentrations, respectively, and nine standard mixtures (ranging from 0.025 to 25 mol/L for each vitamin E congener). Concentrations of T and T3 were calculated and compared to the known concentrations in the samples. Accordingly, interday precision and accuracy were determined in human plasma and mouse liver at low, medium, and high concentrations on three independent days.
Tocotrienols
2.4.5. Recovery The recovery of the analytes was quantified by five independent analyses of human plasma and mouse liver samples, respectively, spiked with tocopherols and tocotrienols at low, medium, and high (0.625, 3.125, and 6.250 mol/L, respectively) concentrations and by comparison of the detector responses with those of standards containing identical amounts of the analytes.
Tocopherols
matrices were evaluated in triplicate in human plasma and mouse liver samples spiked with low (0.625 mol/L) or high (6.25 mol/L) concentrations of T and T3 after storage at room temperature for 24 h. Freeze-and-thaw stability was investigated in the spiked (low and high) plasma and liver samples by comparing detector responses after analysis of the freshly prepared samples with that after storage at −80 ◦ C for 24 h, thawing to room temperature, and repeated freezing to −80 ◦ C, which was repeated thrice (three freeze–thaw-cycles). Post-preparative stability was assessed by repeated injection from the same vial of four separate extracts from spiked (medium analyte concentrations) plasma and liver samples at 0, 6, 12, 18, and 24 h.
␦
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46 Table 1 Linearity [r2 ± SD], limit of detection (LOD) and lower (LLOQ) and upper limits of quantification (ULOQ) of the developed reversed-phase HPLC method for the quantification of tocopherols and tocotrienols (given as absolute amount injected dissolved in 10 L).
42
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
43
Table 2 Stability [% of nominal concentration] of tocopherols and tocotrienols in stock solutions or added at low (0.625 mol/L) or high concentrations (6.25 mol/L) to human plasma and mouse liver and stored at room temperature (RT) for 24 h, and during three freeze-and-thaw cycles (−80 ◦ C/room temperature). Tocopherols ␣
Tocotrienols 
␥
␦
Stock solution stability, 24 h [% nominal concentration] 96.3 100.5 RT −20 ◦ C 98.8 98.4
100.5 98.3
101.6 98.8
98.4 96.8
98.3 96.4
97.4 95.8
101.2 98.8
Human plasma Short-term stability, 24 h, RT [% nominal concentration] 97.8 98.7 Low 98.7 High 106.6
101.5 111.8
98.4 97.8
96.9 107.9
98.3 106.2
103.6 105.7
94.7 101.1
69.6 75.4
74.1 77.0
71.2 79.4
70.1 73.0
74.2 73.6
71.5 59.3
108.2 105.5
107.2 104.4
98.2 95.6
95.1 106.5
95.1 102.0
117.8 94.8
75.2 72.1
70.1 71.1
69.8 70.6
77.7 74.7
81.1 73.2
76.6 74.1
Freeze-and-thaw stability [% nominal concentration] 88.9 74.7 Low 87.1 76.8 High Mouse liver Short-term stability, 24 h, RT [% nominal concentration] 104.6 92.5 Low 104.6 106.5 High Freeze-and-thaw stability [% nominal concentration] 74.6 Low 78.3 High 81.9 76.9
Increasing the flow rate above 0.8 mL/min resulted in a loss of baseline resolution of the - and ␥-congeners. We decided against the use of an internal standard, because the resolution of eight analytes was already considered challenging and the addition of a ninth compound with similar polarity and physicochemical properties would have interfered with the separation of the analytes and might have substantially increased the total runtime of our method. Therefore, we favoured the use of authentic T and T3 as external standards, as they were required for peak identification either way. To the best of our knowledge, this is the first time complete baseline-resolution of all eight T and T3 , including the respective - and ␥-congeners, has been achieved on a RP-stationary phase. The only paper reporting partial resolution of the eight natural vitamin E congeners on a RP-column required gradient elution and a total runtime of a little over 1 h [12]. Hence, our method offers a significant (4-times) faster analysis as well as a simpler and less error-prone isocratic elution. We validated the newly developed method according to the guidelines for bioanalytical method validation of the Center for Drug Evaluation and Research of the U.S. Food and Drug Administration [14]. 3.1. Method validation 3.1.1. Selectivity Injection of plasma samples from five different species (human, pig, rat, mouse, guinea pig) and liver samples from three species (rat, mouse, and guinea pig) revealed no peaks interfering with the eight T and T3 under the selected chromatographic conditions (Fig. 2). Unfortunately, due to the widespread presence of ␣T and ␥T in human and animal diets, entirely vitamin E-free plasma or liver samples were not available. Therefore, minor ␣T and ␥T peaks are present in the blank samples; their identities could be confirmed by spiking experiments. Both plasma and liver samples contained a number of unidentified peaks; one compound eluted directly after ␥T3 and one shortly before ␦T, both, however, were separated from the analytes and thus did not interfere with the analysis (Fig. 2). 3.1.2. Linearity and limits of detection and quantification The detector responses for T and T3 standard solutions as well as extracts from human plasma or mouse liver were linear with a
␣

␥
␦
r2 > 0.99 for all eight analytes except for ␣T extracted from human plasma (r2 > 0.98; Table 1). Detector responses were linear over a wide range starting at low picogram (LOD) and up to low nanogram amounts (ULOQ; Table 1). The developed method was very sensitive with limits of detection for all vitamin E congeners in the range of 27–156 pg per injection (Table 1), which is an order of magnitude lower than the LOD (5–15 ng) reported for another recently developed RP-LC method with fluorescence detection [12]. The LOD of the ␣-congeners were generally higher due to their lower fluorescence intensity compared to the -, ␥-, and ␦-congeners, their concentrations in humans [1,5,18,19] and animals [20–22], however, are usually higher. The LOD and LOQ reported here refer to the sensitivity of the instrument and are given as the amount of analyte required with each injection (pg per injection). The respective values for plasma and liver samples can then be estimated by adjusting the given LOD and LOQ for the respective recoveries of the analytes (Table 3). 3.1.3. Stability Stock solutions stored at room temperature or −20 ◦ C for 24 h remained stable with <5% degradation observed (Table 2). The short-term (bench-top) stabilities of all congeners added at low (0.625 mol/L) or high (6.25 mol/L) concentrations to human plasma or mouse liver and stored for 24 h at room temperature were good with <5.3% maximum degradation observed in plasma and <7.5% in liver samples (Table 2). T and T3 added to plasma or liver were less resistant to repeated freezing (−80 ◦ C) and thawing (to room temperature) with typical losses of 10–25% after three freeze–thaw-cycles and maximum degradation of 40.7% observed for ␦T3 in plasma and 30.2% for ␣T3 in liver (Table 2). For accurate quantification of vitamin E it is thus advisable to thaw frozen samples only once and avoid re-freezing and re-thawing. In order to assess post-preparative stability, vitamin E was extracted from human plasma or mouse liver and the prepared samples were left in a chilled autosampler (ca. 6 ◦ C) for typical HPLC residence times. All eight congeners extracted from plasma or liver remained stable over 24 h (Fig. 3). 3.1.4. Recovery The intraday recoveries for T and T3 extracted from human plasma or mouse liver were generally very good and within the limits set by the FDA (≤15% deviation from the expected value
44
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
Fig. 3. Post-preparative stability was assessed by comparing the peak areas [mV min], after 0, 6, 12, 18, or 24 h residence time in the autosampler, of tocopherols (T) and tocotrienols (T3 ) added to human plasma or mouse liver samples (3.125 mol/L each), respectively, and extracted using the developed protocol.
at medium and high and ≤20% at low concentrations) [14], with few exceptions. The recovery of ␣T3 from human plasma at the medium concentration and of ␣T and ␦T3 from mouse liver at low and medium concentrations, respectively, exceeded these limits by 2–3% (Table 3). The interday recoveries too were within the FDA limits, with the only exception of T added at high concentrations to plasma, which was 16% lower than the known concentration and thus exceeded the limits by 1% (Table 3).
Intraday precision was very good; only for ␣T added at medium concentrations to plasma and at high concentrations to liver were the FDA limits not met. Interday precision was excellent with no values outside of the expected range (Table 3). 3.1.6. Versatility The developed RP-LC method offers outstanding versatility and can be adapted for other biological matrices and detection systems. For example, we were able to separate and quantify the concentrations of T and T3 in brains of guinea pigs fed a T3 -rich rice bran extract (Fig. 4A; guinea pig brain tissue was kindly donated by Dr. Gunter P. Eckert, University of Frankfurt, Germany) using the method developed for liver samples. As expected, ␣T was the major congener in the brain, while the remaining T, ␣T3 , and ␦T3 were present at low concentrations and T3 in trace amounts. T and ␥T were baseline-separated in the guinea pig brain samples (Fig. 4A). With slight modifications to the sample preparation, T and T3 could also be quantified in plant foods such as amaranth, almonds, and hazelnuts (Fig. 4B). Because of the significantly higher
3.1.5. Accuracy and precision The intraday accuracies were generally good and largely within the FDA limits at low and high concentrations (exception: T in plasma). At medium concentrations, however, the intraday accuracies of ␣T and ␦T added to plasma exceeded the set limits significantly, while ␥T, ␣T3 , and ␦T3 exceeded them to a small degree (Table 3). The interday accuracies, on the other hand, were much better and only two congeners added at high concentrations to plasma (T) or liver (T3 ) marginally exceeded the FDA limits (Table 3).
Table 3 Intra- and interday recoveries, accuracies, and precisions of tocopherols and tocotrienols extracted from human plasma or mouse liver spiked at low, medium, or high concentrations (0.625, 3.125, or 6.250 mol/L, respectively; n = 5 for each concentration). Tocopherols ␣  Human plasma
Tocotrienols
Tocopherols
Tocotrienols
␥
␦
␣

␥
␦
␣  Mouse liver
␥
␦
␣

␥
␦
Intraday recovery [%] 101 Low 101 Medium 91 High
86 107 91
97 116 92
86 104 102
91 82 101
96 87 111
103 88 92
107 99 101
122 96 88
97 94 90
109 97 94
92 87 95
95 106 109
98 110 114
99 111 101
103 117 95
Interday recovery [%] 86 Low 95 Medium 93 High
86 90 84
98 89 92
96 86 93
100 90 92
90 96 96
93 93 91
100 88 98
83 98 95
83 97 86
100 102 93
84 89 92
93 89 99
86 97 95
96 99 93
99 91 98
Intraday accuracy [bias%] 19.3 19.3 Low 26.7 8.2 Medium 11.3 18.7 High
1.7 15.7 8.3
8.3 18.4 9.3
12.2 16.5 11.2
7.4 10.1 10.5
12.2 11.0 6.6
0.2 15.3 7.4
20.0 8.1 5.4
19.2 9.3 14.3
2.8 8.2 9.4
19.8 15.3 10.4
13.3 15.8 2.0
20.4 10.7 11.8
11.6 7.6 10.3
11.4 15.1 7.2
Interday accuracy [bias%] 13.6 14.1 Low 4.8 10.5 Medium 7.5 16.2 High
2.5 10.8 7.6
3.7 13.9 7.2
0.0 9.7 8.0
9.9 4.6 4.5
6.8 7.0 8.6
0.3 12.1 2.1
17.3 2.5 5.1
16.7 2.9 13.7
0.4 1.7 7.0
15.9 11.0 8.4
7.4 11.5 1.1
14.0 3.3 15.5
4.1 1.1 7.5
0.6 9.1 1.7
Intraday precision [% CV] 8.8 Low 12.9 Medium 17.9 9.0 14.5 3.8 High
12.3 6.7 3.3
8.6 6.0 3.1
17.4 10.9 6.3
17.6 6.9 10.3
7.2 6.0 8.1
10.8 5.1 8.5
4.9 15.0 17.0
6.0 10.3 6.7
5.0 12.3 3.5
5.1 7.1 3.3
12.5 5.7 11.1
8.5 10.2 14.3
11.9 11.9 4.3
15.4 9.3 8.9
Interday precision [% CV] 4.8 Low 9.2 5.7 3.8 Medium 13.9 2.9 High
9.4 3.5 3.3
8.2 1.9 2.3
9.1 8.4 6.3
6.0 1.5 8.9
1.1 4.4 4.5
8.9 4.3 6.4
4.4 8.9 14.5
6.5 8.2 7.4
4.1 2.1 1.6
2.2 2.9 1.2
13.1 2.6 8.6
1.4 5.2 14.9
4.0 11.3 3.4
10.0 7.1 7.0
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
45
Fig. 4. Representative chromatograms obtained by injection of extracts from (A) brain tissue of a guinea pig fed a tocotrienol-rich rice bran extract or a low calibration standard and (B) amaranth, almonds, or hazelnuts. Note that the retention times vary slightly because the samples were analysed on different columns at different times.
concentrations of T and T3 in the food samples, only 10 mg of amaranth, almond, or hazelnut were used (compared to the 200 mg liver tissue required). Amaranth samples were used without further processing, while almonds and hazelnuts were roughly ground prior to extraction. The food samples were saponified (similar to liver samples) and a third extraction step was added for complete extraction of the analytes. To confirm complete extraction, a fourth hexane extraction was performed and the dried and re-dissolved (in mobile phase) residue was injected into the HPLC, and no T or T3 were detected. For plasma and liver samples, on the other hand, two extraction steps sufficed for complete extraction of the analytes. Due to the high fat content of almonds and hazelnuts, it was necessary to add twice as much saturated potassium hydroxide solution and water (600 and 1800 L, respectively) to the samples in order to facilitate removal of the saponifiable lipids. Accordingly, the volume of glacial acetic acid required for acidification of the samples after saponification was doubled to 600 L. The samples, however, were still rather dirty and further sample clean up or prior lipid extraction might be required for lipid-rich food samples. In the amaranth samples, an unidentified peak eluted just before ␦T, which might require slight adjustments to the water content of the mobile phase, if the method is to be adapted for routine analyses of amaranth or comparable plant food samples. Nevertheless, the - and ␥-tocopherols present in the almond and hazelnut samples were still resolved by our newly developed method (Fig. 4B). For our mobile phase, we chose a methanol:water ratio of 85:15 as the best compromise between resolution and speed of analysis for all analysed sample matrices (Figs. 2 and 4). As mentioned before, if unexpected interfering peaks are observed, increasing the water content of the mobile phase will lead to longer retention times and even better resolution of the analytes, and allows the adaptation of the chromatographic procedure to many other sample matrices, including additional animal and human tissue
samples or plant and food materials. Such changes, however, require appropriate validation of the modified method. In our hands, both fluorescence and coulometric electrochemical detection (with an ESA 8-channel CoulArray) were compatible with the developed method. In order to obtain the required conductivity of the mobile phase for electrochemical detection, 30 mmol/L lithium perchloride were added and baseline separation of all eight vitamin E congeners was achieved (not shown). Our RP-LC method should also be compatible with ultraviolet, photodiode array, and mass detectors, although we did not confirm this yet. 4. Conclusions In the past, the concentrations in biological tissues of - and ␥-tocopherols and - and ␥-tocotrienols, respectively, could only be reported as their respective sums, not the concentrations of the individual compounds. With the development of this rapid, sensitive, accurate, precise, and – importantly – versatile reversed-phase liquid chromatographic method it is now possible to determine the concentrations of all eight natural vitamin E congeners in biological tissues even in large epidemiological studies, randomized controlled trials, and animal experiments using a simple, fast, and reproducible liquid–liquid extraction. The developed method should thus be useful for researchers wishing to investigate the biological availabilities, pharmacokinetics, and/or biological activities of the individual vitamin E congeners and in particular of the respective - and ␥-congeners of tocopherols and tocotrienols. Acknowledgment The authors gratefully acknowledge financial support by the German Research Foundation (DFG) by means of grant no. 2478/4-1.
46
N. Grebenstein, J. Frank / J. Chromatogr. A 1243 (2012) 39–46
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13]
J. Frank, G. Rimbach, Aktuel. Ernaehr. Med. 34 (2009) 131. A. Kamal-Eldin, L.Å. Appelqvist, Lipids 31 (1996) 671. H.M. Evans, K.S. Bishop, Science 56 (1922) 650. C.K. Sen, S. Khanna, S. Roy, Mol. Aspects Med. 28 (2007) 692. J. Frank, X.W. Chin, C. Schrader, G.P. Eckert, G. Rimbach, Ageing Res. Rev. 11 (2012) 163. A. Kamal-Eldin, S. Gorgen, J. Pettersson, A.-M. Lampi, J. Chromatogr. A 881 (2000) 217. S.L. Abidi, J. Chromatogr. A 881 (2000) 197. B.L. Lee, A.L. New, C.N. Ong, Clin. Chem. 49 (2003) 2056. J. Gruszka, J. Kruk, Chromatographia 66 (2007) 909. M.N. Irakli, V.F. Samanidou, I.N. Papadoyannis, J. Sep. Sci. 34 (2011) 1375. M.H. Chen, C.J. Bergman, J. Food Compos. Anal. 18 (2005) 139. M.N. Irakli, V.F. Samanidou, I.N. Papadoyannis, J. Agric. Food Chem. 60 (2012) 2076. S.L. Richheimer, M.C. Kent, M.W. Bernart, J. Chromatogr. A 677 (1994) 75.
[14] Center for Drug Evaluation and Research (CDER) at the Food and Drug Administration (2001). http://www.fda.gov/downloads/Drugs/GuidanceCompliance RegulatoryInformation/Guidances/UCM070107.pdf. [15] M. Podda, C. Weber, M.G. Traber, L. Packer, J. Lipid Res. 37 (1996) 893. [16] K. Augustin, R. Blank, C. Boesch-Saadatmandi, J. Frank, S. Wolffram, G. Rimbach, J. Anim. Physiol. Anim. Nutr. 92 (2008) 705. [17] S. Gaedicke, X. Zhang, P. Huebbe, C. Boesch-Saadatmandi, Y. Lou, I. Wiswedel, A. Gardemann, J. Frank, G. Rimbach, Br. J. Nutr. 102 (2009) 398. [18] J. Frank, T.W. George, J.K. Lodge, A.M. Rodriguez-Mateos, J.P. Spencer, A.M. Minihane, G. Rimbach, J. Nutr. 139 (2009) 58. [19] J. Frank, S. Lee, S.W. Leonard, J.K. Atkinson, A. Kamal-Eldin, M.G. Traber, Am. J. Clin. Nutr. 87 (2008) 1723. [20] J. Frank, A. Budek, T. Lundh, R.S. Parker, J.E. Swanson, C.F. Lourenco, B. Gago, J. Laranjinha, B. Vessby, A. Kamal-Eldin, J. Lipid Res. 47 (2006) 2718. [21] J. Frank, C. Eliasson, D. Leroy-Nivard, A. Budek, T. Lundh, B. Vessby, P. Åman, A. Kamal-Eldin, Br. J. Nutr. 92 (2004) 169. [22] J. Frank, T. Lundh, R.S. Parker, J.E. Swanson, B. Vessby, A. Kamal-Eldin, J. Nutr. 133 (2003) 3195.