Analysis of vitamin E metabolites including carboxychromanols and sulfated derivatives using LC/MS/MS.

Tocopherols and tocotrienols are metabolized via hydroxylation and oxidation of their hydrophobic side chain to generate 13′-hydroxychromanols (13′-OHs) and various carboxychromanols, which can be further metabolized by conjugation including sulfation. Recent studies indicate that long-chain carboxychromanols, especially 13′-carboxychromanol (13′-COOH), appear to be more bioactive than tocopherols in anti-inflammatory and anticancer actions. To understand the potential contribution of metabolites to vitamin E-mediated effects, an accurate assay is needed to evaluate bioavailability of these metabolites. Here we describe an LC/MS/MS assay for quantifying vitamin E metabolites using negative polarity ESI. This assay includes a reliable sample extraction procedure with efficacy of ≥ 89% and interday/intraday variation of 3–11% for major metabolites. To ensure accurate quantification, short-chain, long-chain, and sulfated carboxychromanols are included as external/internal standards. Using this assay, we observed that sulfated carboxychromanols are the primary metabolites in the plasma of rodents fed with γ-tocopherol or δ-tocopherol. Although plasma levels of 13′-COOHs and 13′-OHs are low, high concentrations of these compounds are found in feces. Our study demonstrates an LC/MS/MS assay for quantitation of sulfated and unconjugated vitamin E metabolites, and this assay will be useful for evaluating the role of these metabolites in vivo.

Plasma samples or cell culture media were extracted by a solvent mixture containing 6 vol of working methanol (containing 0.2 mg/ml ascorbic acid) and 12 vol of hexane via vigorous vortexing for 1 min. ␦ TE-13 ′ -COOH (100 pmol) was added as an internal standard (IS). After centrifugation at >12,000 rpm for 2 min, the upper hexane layer was collected in a new tube with preadded butylated hydroxytoluene (BHT; 0.1 mg). The methanol layer (90-95%) was transferred into a clean tube, and the residual pellet was extracted one more time with 4 vol of working methanol. After vortexing and centrifugation, the combined methanol layers were dried under nitrogen. Both dried methanol-and hexane-extracted samples were resuspended in working methanol before being analyzed by HPLC or LC/MS/MS. During the extraction procedure, samples were protected from light. Before LC/MS/MS analysis, ␣ -CMBHC (1 or 5 M) was added as an additional IS.

Extraction of vitamin E metabolites from feces and urine
Approximately 30 mg of fecal samples was homogenized in 2 ml of methanol with ascorbate (0.2 mg/ml). After centrifugation, 1.5 ml of methanol layer was dried and resuspended in 200 l methanol, which was then diluted 10 times with addition of synthesized ␦ TE-13 ′ -COOH (1 µM) as IS before being analyzed by LC/MS/MS. One hundred microliters of urine samples was added with ␦ TE-13 ′ -COOH (1 M) and extracted by 500 l of working methanol (0.2 mg/ml ascorbic acid). The extraction was repeated one more time with 200 l of working methanol. The combined methanol was dried under N 2 . This procedure yielded >95% recovery of ␦ TE-13 ′ -COOH.

Analysis of vitamin E metabolites and vitamin E forms by HPLC with fl uorescent detection
As previously described ( 4 ), vitamin E metabolites, tocopherols, and tocotrienols were separated on a 150 × 4.6 mm, 5 m Supelcosil™ LC-18-DB column using HPLC and detected by a Shimadzu RF-10AXL spectrofl uorometric detector (Columbia, MD) with the excitation and emission wavelength at 292 nm and 327 nm, respectively. This method was used in experiments in studying the extraction effi cacy and accuracy of metabolites and vitamin E forms.

LC/MS/MS
The LC/MS/MS analysis was done with an Agilent 1200 LC system coupled to an Agilent 6460 QQQ mass spectrometer equipped with a jet stream ESI source (Santa Clara, CA). The chromatography utilized an Atlantis dC18 column (2.1 × 150 mm, 3 µm) from Waters Corporation (Milford, MA). Buffer A consisted of acetonitrileethanol-water (165:135:700, v/v/v), and buffer B was acetonitrileethanol-water (539:441:20, v/v/v), both of which contained 10 mM ammonium acetate with acetic acid to adjust pH to 4-4.3. The LC gradient was as follows: time 0 min, 0% B; time 1 min, 0% their role in vitamin E-mediated benefi cial effects remains unclear. One of the reasons is that there is limited information regarding the bioavailability of these metabolites in tissues. To this end, a reliable and accurate analytical method is needed to evaluate whether these metabolites can be generated at suffi cient amounts in the body.
Although we have previously developed an HPLC-fl uorescent assay for analyzing various vitamin E metabolites ( 4 ), this method is not suitable for simultaneously quantifying metabolites as a result of supplementation of a mixture of vitamin E forms. Previously published LC/MS assays are not ideal because they do not directly measure sulfated carboxychromanols and did not have suitable standards for extraction and quantifi cation ( 19,20 ). Furthermore, an assay using GC/MS, despite being highly sensitive, is not capable of directly detecting conjugated metabolites ( 9 ). In this study, we describe a sensitive LC/MS/MS assay that simultaneously analyzes hydroxychromanols, carboxychromanols, and sulfated metabolites. To ensure accurate quantitation, our assay includes an improved sample processing procedure, and uses various carboxychromanols and a sulfated carboxychromanol as internal and external standards. We then applied this assay to quantify bioavailability of vitamin E metabolites in plasma, urine, and fecal samples from rodents supplemented with ␥ T or ␦ T.

Materials
␣ T (99%), ␥ T (97%-99%), and ␦ T (97%) were purchased from Sigma (St. Louis, MO). ␥ -CEHC ( у 98%), ␣ -CEHC, and (±)-␣ T-5 ′ -COOH ( ␣ -CMBHC) were from Cayman Chemicals (Ann Arbor, MI). ␦ T-13 ′ -COOH and ␦ TE-13 ′ -COOH, which are metabolites from ␦ T and ␦ TE, respectively, were synthesized according to a published procedure ( 21 ). Tissue culture reagents were from Invitrogen (Rockville, MD). All other chemicals were purchased from Sigma. housed in Purdue Life Science Animal Facility for a week for adaptation before experiments and then randomly grouped by body weight match. Rats were administered with ␥ T or ␦ T at 100 mg/kg body weight by gavage using tocopherol-stripped corn oil (0.5 ml) as the vehicle (n = 3 in each group). Control animals received 0.5 ml of tocopherol-stripped corn oil. Six hours later, animals were euthanized, and plasma, liver, and other tissues were collected. In another study, rats were given ␥ T at 50 mg/kg by gavage, and urine was collected for 7 h. The urine samples were then aliquoted and frozen at Ϫ 80°C until use.
In the study of metabolite formation in response to ␥ T-or ␦ Tsupplemented diets, male Balb/c mice (5-6 weeks) were obtained from Harlan (Indianapolis, IN) and single-housed under controlled temperature with unrestricted access to diets and water. After a week of acclimatization, mice were randomly divided into control (AIN-93G diet) and ␥ T-or ␦ T-supplemented (0.1% diet) group. These mice were subjected to induction of colon tumorigenesis by azoxymethane/dextran sodium sulfate (AOM/ DSS) as previously described ( 12 ). When the study was terminated, mice were on supplemented diets for more than 150 days ( 12 ). During euthanasia, plasma and feces were collected.

Statistical analysis
In the study of extraction effi cacy, ANOVA was used to calculate intraday and interday variances. Student's t -test was used in B; time 30 min, 99% B; time 40 min, 99% B; time 43 min, 0% B; time 48 min, 0% B. The fl ow rate was 0.3 ml/min with a total run time of 48 min. Multiple reaction monitoring was used to analyze each compound . Negative polarity ESI was used with the following source conditions: gas temperature, 325°C; gas fl ow, 10 liters per min; nebulizer pressure, 30 psi; sheath gas temperature, 250°C; sheath gas fl ow, 7 liters per min; capillary voltage, 4,000 V; nozzle voltage, 1,500 V; and an electron multiplier voltage of Ϫ 300 V. All data were evaluated with Agilent MassHunter Qualitative Analysis software, version B.06.00.

Animal studies
All the animal studies were approved by Purdue Animal Care and Use Committee. In the study of metabolite formation in response to a single gavage of ␥ T or ␦ T, male Wistar rats (230-250 g) were purchased from Charles River (San Diego, CA). Rats were in the negative mode gave much stronger signals than other ionization methods.
We then optimized the ESI conditions by varying fragmentation transitions and collision energies for ␥ -CEHC, ␣ -CEHC, 5 ′ -COOH, ␦ TE-13 ′ -COOH, and ␦ T-13 ′ -COOH. To optimize ionization conditions of other metabolites including sulfated counterparts, we used conditioned media that contain sulfated carboxychromanols produced by incubation of A549 cells with ␥ -tocotrienol ( 4,5 ). Typical fragmentations of carboxychromanols are shown in Fig. 3 . Major fragmentations of various vitamin E metabolites were caused by loss of the side chain along with two carbons in the nonaromatic part of the chromanol ( Fig. 3 ). An additional loss of -CH2 at the 4-position was observed for ␣ -CMBHC. Sulfated metabolites were similarly fragmented with a signature loss of the sulfate group ( Fig. 3 ).
To maximize signals, we chose the transition showing the highest intensity under varied collision energies for each analyte for quantifying carboxychromanols. The parameters for each carboxychromanols and vitamin E forms are summarized in Table 2 .
It should be mentioned that besides sulfated metabolites, we attempted to directly detect glucuronidated carboxychromanols using full scan mode on a time-of-fl ight instrument or monitoring M+176 ions and fragments by product ion scanning, precursor loss scanning, and multiple reaction monitor. However, we did not observe significant formation of glucuronidated metabolites in the plasma of rats gavaged with tocopherols.

Quantifi cation of sulfated carboxychromanols
Conjugated carboxychromanols have often been quantifi ed after being converted to unconjugated counterparts by sulfatase/glucuronidase ( 22 ). However, this approach could not reveal the nature of conjugation and may be inappropriate when sample size is small. Further, our unpublished data indicate that many sulfatase or glucuronidases were not able to effectively deconjugate long-chain carboxychromanols such as sulfated-13 ′ -carboxychromanol (13 ′ S; unpublished observations). In order to directly quantify sulfated metabolites, we purifi ed ␥ TE-9 ′ S from conditioned media from cells incubated with ␥ TE ( 16 ) (Materials and Methods). The concentration of stock ␥ TE-9 ′ S was then determined by HPLC with UV-visible detection after being converted to ␥ TE-9 ′ -COOH with the statistical analyses for comparison of controls with tocopherolsupplemented groups. All results are expressed as mean ± SD.

Optimization of sample processing procedure
In our previously published HPLC method, vitamin E metabolites were extracted into ethylacetate after sample solutions were acidifi ed to pH <4. This protocol has been shown to effectively extract short-or medium-length carboxychromanols such as CEHC and 9 ′ -COOH ( 4 ). However, extraction effi ciency for 13 ′ -COOHs by this procedure was not optimal, especially for plasma or serum samples (Q. Jiang, unpublished observations). Another weakness is that vitamin E forms and their metabolites need to be extracted separately by hexane/methanol, which requires additional samples. For these reasons, we decided to modify and optimize the extraction procedure aiming to effectively extract vitamin E forms and all the metabolites simultaneously from one (30-50 µl) biological sample.
To optimize and evaluate extraction effi cacy, we used FBS spiked with vitamin E forms and metabolites including ␥ T, ␦ T, ␥ -CEHC, ␣ -CEHC, ␣ -CMBHC, ␦ TE-13 ′ -COOH, and ␦ T-13 ′ -COOH. The optimized extraction procedure included an extraction step with a solvent mixture containing 6 vol of methanol and 12 vol of hexane. After centrifugation, we collected both layers and performed an additional extraction of the pellet with 4 vol methanol (details in Materials and Methods). The hexane layer contains tocopherols and tocotrienols, and the combined methanol layer contains carboxychromanols and hydroxychromanols. This procedure yielded high extraction effi ciency and accuracy, as indicated in Table 1 based on HPLC-fl uorescent analyses.

Optimization of LC/MS/MS conditions
At the early stage of method development, we compared relative ionization effi cacy of carboxychromanols among different MS ionization approaches including ESI and atmospheric pressure chemical ionization and photoionization interfaces. In these studies, we used ␦ T-13 ′ -COOH and ␥ -CEHC as the model compounds and found that ESI Effi cacy (%)  97  89  96  98  95  107  101  Intraday CV%  9  4  7  4  3  7  6  Interday CV%  3  11  6  10  5  11  8 CV, coeffi cient of variation . Tocopherols and metabolites were spiked into FBS to yield ␥ -CEHC (2 µM), ␣ -CEHC (10 µM), ␣ -CMBHC (2 µM), ␦ TE-13 ′ -COOH (1 µM), ␦ T-13 ′ -COOH (0.4 µM), ␥ T (0.4 µM), and ␦ T (0.4 µM). One hundred microliters of spiked FBS was extracted by 600 l working methanol (with 0.2 mg/ml ascorbic acid) and 1.2 ml hexane via vortexing vigorously for 1 min. After centrifugation for 2 min at 10,000 rpm, the upper hexane layer was transferred into a tube with preadded 10 µl BHT (10 mg/ml). The methanol layer was transferred into a clean tube, and the pellet was extracted with 400 µl working methanol for the second time. The two methanol extractions were combined and dried under N 2 . The dried methanol and hexane-extracted samples were resuspended in 100 µl working methanol before being analyzed by HPLC with fl uorescent detection. Data are expressed as mean ± SD (based on >4 independent experiments). carboxychromanols can vary from time to time, it is necessary to quantify these compounds by internal and external standards that are analyzed along with biological samples. Specifi cally, in the subsequent analysis of biological samples, we used ␦ T-13 ′ -COOH and ␦ TE-13 ′ -COOH to quantify 13 ′ -OHs and long-chain carboxychromanols, ␥ TE-9 ′ S for sulfated long-chain metabolites, and ␥ -CEHC or ␣ -CEHC for short-chain carboxychromanols. In these studies, ␦ TE-13 ′ -COOH and ␣ -CMBHC were used as IS and the rest as external standards.
We noticed that ␥ TE-derived long-chain metabolites bearing different numbers of double bonds were coeluted under the current LC conditions. For instance, ␥ TE-11 ′ S containing one (481.2 → 149) and two (479.2 → 149) double bonds have the same retention time ( Table 2 ). Because these coeluted metabolites differ in one double bond, 13 C isotopic correction will be necessary for proper quantifi cation of these compounds ( 23 ).

Vitamin E metabolites detected in the plasma and feces of mice fed ␥ T-or ␦ T-supplemented diets
Using the established LC/MS/MS assay, we found that mice fed a diet supplemented with ␥ T or ␦ T had low but detectable amounts of CEHC (3 ′ -COOH) and sulfated long-chain carboxychromanols as well as 13 ′ -COOH in the plasma, whereas none of these metabolites were detectable sulfatase digestion. We have subsequently used ␥ TE-9 ′ S as an external standard to quantify sulfated long-chain carboxychromanols.

Quantifi cation of short-and long-chain metabolites in biological samples
We observed a broad range of linearity between LC/ MS/MS response and analyte concentrations (from 10 nM to 5 µM) for short-and long-chain metabolites including ␥ -CEHC, ␣ -CEHC, ␣ -CMBHC, and 13 ′ -COOHs. Under optimized ionization conditions, the detection limits for carboxychromanols were ‫ف‬ 0.1-0.4 pmol on column with a signal-to-noise ratio of >8. To test ionization effi cacy in extraction matrix, known amounts of ␥ -CEHC, ␣ -CMBHC, ␥ TE-9 ′ S, and ␦ T-13 ′ -COOH were added to the reconstituted solution of FBS or plasma that were extracted by hexane/methanol with ␦ TE-13 ′ -COOH as an IS. These samples were then analyzed under the optimized LC/MS conditions. We found that these compounds spiked in the plasma or FBS extracts showed similar ionization effi cacy/ intensity to those prepared in methanol, indicating that the extraction matrix does not have signifi cant impact on ionization. These data, therefore, justify quantitation of metabolites using external standards combined with ISs. Because relative ionization effi cacy among different  in mice fed the control AIN93G diet ( Table 3 ). Interestingly, in contrast to low concentrations of metabolites in the plasma, we found relatively high levels of unconjugated short-, mid-, and long-carboxychromanols in feces. Among fecal excretion metabolites, 13 ′ -COOH and 13 ′ -OH were the predominant vitamin E metabolites ( Table 4 ). that results in high sensitivity and excellent reproducibility. We have used multiple carboxychromanols as external standards and ISs to ensure accurate quantifi cation. In addition, our method has a 9 ′ S as a standard, which allows direct quantifi cation of sulfated metabolites without the need for enzyme digestion. As a result, our study, for the fi rst time, documented the level of sulfated CEHCs in the plasma.
The ability to directly quantify conjugated metabolites including sulfated derivatives allows simpler and more accurate quantitation of these compounds than traditional procedures with enzyme digestion to remove the sulfate and/or glucuronide group. Our unpublished data indicate that enzyme digestion with sulfatase/glucuronidase could not effectively remove the sulfate group from sulfated 13 ′ -COOH, although conjugated CEHC and 9 ′ -COOH can be effectively converted to their unconjugated counterparts via sulfatase/glucuronidase digestion ( 22 ). This is probably because sulfated long-chain metabolites are poor substrates for the enzymes tested (unpublished observations). As a result, the approach with enzyme digestion can only accurately quantify conjugated shortchain carboxychromanols, but not the long-chain analogs. Our present assay enables accurate and direct quantitation of 9 ′ S, 11 ′ S, and 13 ′ S using ␥ TE-9 ′ S as a standard. Because sulfated and glucuronidated metabolites may have different bioactivities, this method offers an opportunity to measure different types of conjugates and may therefore be useful for further evaluation of their roles in vivo. On the other hand, despite our effort to detect glucuronidated carboxychromanols (176+M), we did not observe

Metabolites detected in the plasma and urine after a single gavage of ␥ T and ␦ T
To evaluate the bioavailability of metabolites as a result of a high supplement dose of tocopherols, we gave rats a single gavage of ␥ T or ␦ T at 100 mg/kg body weight and collected plasma samples 6 h later. We found that sulfated CEHCs (SO 3 -CEHC), 9 ′ S, and 11 ′ S were among the major metabolites in the plasma of tocopherol-supplemented rats ( Table 5 ). Unlike analyses by HPLC with fl uorescent detection ( 4,5 ), the LC/MS/MS method allowed direct measurement of sulfated CEHCs. Compared with feeding on tocopherol-supplemented diet, oral gavage resulted in much elevated conjugated carboxychromanols in the rats' plasma at the time of sample collection (cf .  Tables 3 and 5 ).
In a separate animal study, we collected urine samples within 7 h after rats were gavaged with ␥ T at 50 mg/kg or corn oil (controls). We observed high excretion of sulfated ␥ -CEHC even in control rats (i.e., 16.4 ± 2.4 M) but could not detect ␥ -CEHC. Supplementation of ␥ T led to increase of sulfated ␥ -CEHC and unconjugated form to 78.59 ± 35.9 and 0.014 + 0.003 M, respectively.

DISCUSSION
We have developed a highly sensitive and specifi c LC/MS/ MS assay for simultaneously analyzing hydroxychromanols, carboxychromanols, and sulfated vitamin E metabolites. Our method is strengthened by optimized ionization of major metabolites and an improved extraction procedure Balb/c mice were fed AIN93G control diet or ␥ T-or ␦ T-supplemented diet (0.1% w/w) for 3 months, and these mice were subjected to AOM/DSS treatment (see Materials and Methods for details). Plasma samples were collected and extracted for LC/MS/MS analysis. "Low" indicates below or at the detection limit. Data are expressed as mean ± SD (n = 3-5 per group). * P < 0.05: difference between control and supplemented diets. Fecal samples were obtained from the same animal study described in Table 3 . "Low" indicates below or at detection limit. Data are expressed as mean ± SD (n = 3-5 per group). * P < 0.05, ** P < 0.01: difference between control and supplemented diets. compared with unmetabolized vitamins. ␥ -CEHC, but not tocopherols, has natriuretic activity ( 15 ) and is thought to be responsible for enhanced sodium excretion in response to supplementation of ␥ TE in rodents ( 24 ). We have demonstrated that 13 ′ -COOHs are competitive inhibitors of cyclooxygenases, whereas none of the tocopherols or sulfated long-chain metabolites directly inhibit the cyclooxygenase activity ( 16 ). 13 ′ -COOH derived from ␦ T inhibits 5-lipoxygenase activity and leukotriene formation in stimulated neutrophils. On the other hand, tocopherols have no effect on 5-lipoxygenase activity, and their suppression of leukotriene in white blood cells varies with specifi c stimuli ( 17 ). In addition, 13 ′ -COOHs induce apoptosis in liver hepatoma HpG2 cells more effectively than tocoherols ( 18 ). 13 ′ -OH and 13 ′ -COOH derived from ␣ T are found to have antiatherogenic activities ( 25 ). Given these bioactivities of long-chain metabolites, it is reasonable to assume that these compounds may play a role in vitamin E-mediated benefi cial effects in vivo ( 1 ), which warrants further investigation.
In summary, we have described an LC/MS/MS assay that simultaneously quantifi es hydroxychromanols, carboxychromanols, and sulfated vitamin E metabolites. This method should be useful for further evaluation of pharmacokinetics of vitamin E metabolite formation and their bioavailability in tissues and excretion, as well as for investigation of potential contributions of metabolites to vitamin E-mediated benefi cial effects in animals and humans.
any signifi cant amount of these metabolites in the plasma of rats gavaged with ␥ T. It remains to be determined whether glucuronidated carboxychromanols can be seen in other tissue or human samples.
Using the developed LC/MS/MS assay, we found some interesting aspects of vitamin E metabolite formation in response to supplementation of ␥ T or ␦ T. First, administration of ␥ T or ␦ T via a single gavage appeared to achieve much higher plasma concentrations of sulfated carboxychromanols than the steady concentrations achieved via chronic feeding with tocopherol-supplemented diets. This observation can be explained by the fact that animals consume a much smaller amount of ␥ T or ␦ T from supplemented diet at any given time than a single gavage by which tocopherols were given as a large boost . That relatively high plasma levels of sulfated metabolites were detected after gavage indicates that sulfation of long-chain or intermediate carboxychromanols takes place in parallel with ␤ -oxidation when a relatively large quantity of tocopherols is consumed. Second, large amounts of unconjugated metabolites, especially 13 ′ -COOH and 13 ′ -OH, which are very low in the plasma, are detected in feces. This observation is consistent with previous reports (12)(13)(14). The high level of carboxychromanols in feces suggests that most long-chain metabolites are not transported to the circulation after being generated in the liver, but rather are excreted via the bile. It has recently been estimated that up to 70% of vitamin E metabolites may be excreted to feces ( 13,14 ). Alternatively, it is also possible that high levels of these metabolites in feces are produced by gut fl ora, an intriguing possibility warranting further investigation. Regardless, our current results together with previous reports confi rm that 13 ′ -COOH and 13 ′ -OH are the major fecal excretion metabolites of vitamin E. In addition, although conjugated CEHCs have been reported in the urine ( 6-11 ), here we quantifi ed sulfated ␥ -CEHC. Interestingly, there are substantial amounts of SO 3 -␥ -CEHC even in the urine of control rats that were fed standard chow containing ‫ف‬ 9 mg/kg of ␥ T.
It has been demonstrated that specifi c vitamin E metabolites are more bioactive or have different activities  Wistar rats were given a single gavage of ␥ T or ␦ T in corn oil or corn oil alone (control). Six hours later, animals were euthanized, and plasma samples were obtained and stored at Ϫ 80°C. Detailed information on animal studies, sample processing, and the LC/MS/MS method are described in Materials and Methods. "Low" indicates below or at the detection limit. Data are expressed as mean ± SD (n = 3 per group). * P < 0.05, ** P < 0.01: difference between control and tocopherol supplementation.