Rapid Analysis of α-Tocopherol and Its Oxidation Products Using Supercritical Carbon Dioxide and Proton Transfer Reaction Ionization Mass Spectrometry

We have developed a rapid and sensitive analytical method for α-tocopherol and its oxidative products by combining online hyphenation of supercritical fluid extraction-supercritical fluid chromatography (SFC) with proton transfer reaction (PTR) ionization mass spectrometry (MS). α-Tocopherol is a well-known antioxidant that plays a vital role in the antioxidant defense system in plant cells. However, studies on the cellular mechanisms of α-tocopherol have been limited owing to the lack of a rapid analytical method, which limits the comparison of plant cells incubated in various conditions. Additionally, complex sample preparation and long chromatography separation times are required. Moreover, the majority of the involved molecules are a combination of isomers, which must be separated before applying tandem MS. α-Tocopherol produces the α-tocopheroxyl radical in the first step of its antioxidant function; another ion with the same mass may also be generated from the source. SFC separation effectively distinguished the observed ions from their oxidative products in the sample and those produced during the ionization reaction process. This method enabled the measurement of α-tocopherol and its oxidative products such as α-tocopheroxyl radical and α-tocopheryl quinone in approximately 3 min per sample, including the time required for sample preparation.


INTRODUCTION
Since its development by Lindinger et al. 1) in the late 1990s, proton transfer reaction mass spectrometry (PTR-MS) has been used as an e ective method for measuring volatile organic compounds (VOCs) in gaseous samples. It has been used in various applications involving the measurement of VOCs in volcanic gases and forest res, metabolite analysis of breath, identi cation of microorganisms, etc. [2][3][4][5][6] Recently, we have invented the simpli ed ion source inspired by PTR-MS, which uses a mixture of water-positive and negative ions as reagent ions. ose ions react with analytes so that the proton transfer reaction occurs in protonated and deprotonated directions, which is slightly di erent from the traditional PTR-MS. Although it is essentially a water ion chemical ionization, we have called it proton transfer reaction (PTR) ionization mass spectrometry (MS). 7,8) Supercritical uid extraction (SFE) is a rapid and safe method for extracting lipophilic compounds from complex sample matrices. When applicable, SFE can also be used to extract polar molecules employing an entrainer. SFE was developed by Zosel in the 1970s 9) and has been applied in the industrial-scale extraction of food and ne chemical industries because carbon dioxide, the most common uid for SFE, is safe to use in foods. A supercritical uid selectively extracts compounds depending on the temperature, pressure, and entrainer (such as ethanol) that can transport various molecules to another vessel, making it suitable for industrial-scale extraction and analytical-scale sample preparation.
Because supercritical uids have a lower viscosity than liquids and an almost identical density to liquids, supercritical uid chromatography (SFC) exhibits a higher separation e ciency than liquid chromatography (LC). us, SFC has also been used for the optical peak resolution (Rs) of racemic compounds 10) and steroids. 11) In addition to separation e ciency, the selectivity (α) of SFC can change the pressure and temperature of the mobile phase, which gives more exibility to optimize the peak resolution of racemic compounds than LC. Unique applications such as bioactive lipids in aqueous solution have been reported 12) by combining SFE and SFC. Recently, noteworthy advances in uid control technology have a orded new developments such as ultrafast separation using SFC. 13,14) Plant cells have several enzymatic and nonenzymatic self-defense mechanisms against oxidation. α-Tocopherol (α-T), in particular, is well known to prevent lipid peroxidation. [15][16][17] e α-T oxidation reaction scheme reported by Kumar et al. 18) and Tang et al. 19) is summarized in Fig. 1. e log P values shown in the gure were calculated using a function on the rdkit.Chem Crippen module. 20) Kumar et al. investigated the formation of α-tocopheroxyl radical and α-T-hydroperoxide by thylakoid membrane and rose bengal as the source of singlet oxygen ( 1 O 2 ). ey used a combination of analytical instruments, including LC with a photodiode array (PDA) detector, uorescence spectroscopy, and electron paramagnetic resonance spectroscopy to detect α-T hydroperoxide, α-T radicals, and 1 O 2 . A crucial function of α-T is quenching 1 O 2 before the oxidation of fatty acids. Tang et al. reported the detailed chemical structures of α-T oxidization products by monitoring changes in ultraviolet absorption spectra using LC-PDA detection and analyzing the corresponding chromatographic peaks via quadrupole time-of-ight (TOF) mass spectrometer. Although the details of the self-defense system of plant cells against 1 O 2 using α-T are recently being determined by combining several analytical methods, the analytical process involves complex procedures and requires more than 30 min of chromatographic run time. Accordingly, a more rapid and sensitive method for investigating the defense system in individual cells is required. Additionally, because ambient autooxidation occurs during sample preparation, sample processing must be completed as quickly as possible to avoid possible artifacts. As previously reported, 7,8) SFE-PTR MS can also detect nonvolatile small molecules with high sensitivity, at least up to m/z 1000. e addition of chromatography separation to SFE-PTR MS improves the analytical performance, such as isomer peak resolution and better detection sensitivity. We further evaluation of our PTR ow tube prototype revealed approximately two seconds of longitudinal dispersion, which is practically applicable to the SFC and a ords an overall narrow peak width. In this study, we have investigated a preparation-free rapid analysis of α-T and its oxidative products using SFE/SFC-PTR MS.

Chemicals
α-T, hexamethoxyphosphazine (P321), hexakis (2,2-di uoroethoxy) phosphazene (P621), ethanol, and acetonitrile (LC/MS grade) were purchased from FUJIFILM Wako Pure Chemicals Corporation, Osaka, Japan. Water was obtained from a Milli-Q puri cation system (Merck, Germany). Cylinders of general-grade helium and carbon dioxide (siphon type) (Iwatani Industrial Gases Corp, Osaka, Japan) were used. e purchased α-T was stored in the dark at −80°C until use. A 10-µM α-T solution was prepared in 1 mL acetonitrile and used without further puri cation a er incubation at room temperature (25°C) in Eppendorf sample tubes for a given period.

Mass spectrometer
With a minor modi cation, a JMS-T100 LP (AccuTOF) TOF mass spectrometer (JEOL, Tokyo, Japan) was used with an Acqiris Model U5303A (3.2 GS·s −1 12-bit digitizer, Geneva, Switzerland) as the data acquisition system. Data acquisition was performed using an open-source so ware "QtPlatz" (https://github.com/qtplatz/qtplatz) with a modied eld programmable gate array con guration for acquiring peak detection (PKD) and waveform averaging (AVG) simultaneously. 21) e PKD histogram and AVG waveform were collected every 200 ms using U5303A. Figure 2 shows the block diagram of the PTR ion source, comprising a ow tube with a corona discharge electrode. Before entering the ow tube, helium was allowed to ow through a 250-mL solvent-reservoir bottle, containing about 10 mL of ultrapure water (Millipore, MA, US). e helium ow rate was set to 70 mL·min −1 , which was controlled via a mass ow controller (8500MC-0-1-2, KOFLOC Corp., Kyoto, Japan). An AccuTOF vacuum system maintained pressure in a range of 90 to 105 Pa in the ow tube. A PS350 high-voltage power supply (Stanford Research Systems, Inc., Sunnyvale, CA, US) was used as a corona discharge power supply. e voltage for the discharge electrode was set to −2.0 kV, generating a current of 80-100 µA. Figure 3 shows the hydraulics of a splitless injection. Liquid carbon dioxide from the cylinder was precooled to −5°C in an ethanol/dry-ice bath and delivered at a rate of 0.5 mL·min −1 at 25 MPa using an LCPackings UltiMate Micropump ( ermo Scienti c, US), equilibrated at 40°C in an oven (Agilent 1100 Series ermostatted Column Compartment; Agilent, CA, US) and connected to an SFC column (InertSustain). As a pressure restrictor, the outlet of the column was connected to a PTR ow tube that passed through a 20-µm i.d. × 21-mm length fused silica capillary (GL Science, Tokyo, Japan). e sample dissolved in acetonitrile was applied to a stainless steel frit, and acetonitrile was allowed to evaporate. e frit with the sample was then attached to an inline lter (ACQUITY Column In-Line Filter (Waters, US)) and introduced into the SFC ow by switching the valve (Rheodyne 7000, Rheodyne, USA) position. Figure 4 shows the hydraulics of an SFE/SFC split injection, in which SFE and SFC have separate carbon dioxide delivery systems. In the SFE hydraulics, liquid carbon dioxide was precooled to −5°C in an ethanol/dry-ice bath and delivered at a rate of 200 µL·min −1 at a pressure of 25 MPa using an LCPackings UltiMate Micropump. Carbon dioxide from the pump was equilibrated at 40°C in an oven and then passed through Rheodyne 7000 switching valves V1 and V2. Carbon dioxide was discharged using a pressure restrictor (20-µm i.d. × 250-mm length).

SFE/SFC split injection
In the SFC hydraulics, liquid carbon dioxide was precooled to −5°C in an ethanol/dry-ice bath and delivered at a rate of 1.0 mL·min −1 at 30 MPa using a PU-980 pump (JASCO, Tokyo, Japan). Carbon dioxide from the pump was equilibrated at 40°C in an oven, and then the supercritical carbon dioxide was connected to the separation column via valve V2. e mobile phase from the column was connected to a pressure restrictor in the PTR ow tube. is method used either an InertSustain or an L-column3 column for SFC separation. SFC hydraulics shared V2 with SFE hydraulics, in which 1 µL of uid was divided and introduced into the column. e carbon dioxide ow rate, temperature, pressure, and timing of the split ow at the beginning of SFE may a ect the composition and amounts of the sample to be injected.
All pump heads for carbon dioxide delivery were cooled to approximately 8°C using a Peltier module (TES112705, Hebei, China).

Mass calibration
Mass calibration was performed using sodium tri uoroacetate, an electrospray ionization (ESI) source for m/z between 159 and 703, and third-order polynomials. e PTR ion source was then attached by replaced with the ESI source.

SFE/SFC analysis of α α-T incubate
e α-T incubates obtained a er 0, 1, and 7 days of incubation at ambient temperature (≈25°C) were analyzed using SFE/SFC splitless injection for all the 5 pmol of α-T. , and (C) shows a single chromatographic peak with the highest intensity. e chromatographic peaks shown in Fig. 5 are supposed to be a mixture of the α-T and its oxidative products listed in Fig. 1 and maybe molecules uncovered yet. e relative intensity of the chromatographic peak at m/z 430.381 drastically changes during the incubation period. Although the chromatogram for m/z 429.373 shows a broad peak on all three samples, the relative retention time to the chromatographic peak for m/z 431.388 appears to decrease.
As shown in Table 1, the relative peak area of chromatograms for m/z 430.381 compared with 431.388 increased by 1.45 and 2.20 folds for 1 and 7 days of incubation, respectively. Although the relative peak area of chromatograms for m/z 430.381 changed considerably, no apparent changes in chromatographic properties, such as peak resolution or retention factor, were observed. Figure 1 suggests that in the protonated form, several reported oxidative products could exhibit peaks at m/z 430.381 and 431.388; the obtained chromatographic peak comprises a couple of coelutes. To resolve these molecules, a better chromatography separation technique is required. e volume of an extraction vessel is approximately one microliter, which is small enough to form a narrow peak bandwidth that is compatible with SFC. However, chromatographic peaks were obtained with a peak width of 2.3-6.8 s (Table 1), which were derived from the SFE pro le with additional molecule-speci c extraction delay. Each molecule is injected into the column with its own bandwidth by slightly separated SFE pro les, which results in each molecule showing a few seconds of peak width. Split injection is a simple and e ective method for introducing a narrow sample bandwidth to SFC to improve peak resolution by limiting sample bandwidth as well as timing synchronization.

Determination of split timing
As described in Section 3.1, SFE extraction pro les show a peak with a width of a few seconds. Finding a peak apex point is optimal for splitting an SFE ow; however, apex timing is dependent on the molecule and SFE conditions. erefore, experimentally determining optimal timing for the given sample matrix is necessary. Figure 6 shows a series of sample injections splitting SFE ow into SFC hydraulics without using a separation column. An extracted ion pro le monitored at m/z 431.388 provided the best split timing of 20 s, with the most intense peak. However, a pro le monitored at m/z 430.381 and 429.373 produced the most intense peak at 10-s split timing. Because there is no considerable di erence between the peak intensity of m/z 431.388 at 10 and 20 s, we use the 10-s split timing for further experiments. Figure 7 shows a chromatogram of 10 pmol of the α-T sample using split injection, corresponding to Fig. 5 (B), which is a 1-day incubated α-T sample. e extracted ion chromatogram for m/z 431.388 shows a single peak with a retention time of 15.1 s and a width of 2.7 s, which is half the peak width obtained using splitless injections. A chromatogram for m/z 430.381 exhibited a peak with the same retention time m/z 431.388; however, it also showed an unresolved shoulder peak at 11.2 s. Sample amounts applied to the frit for split injection were 10 pmol, which was double the amounts of splitless injection (5 pmol) chromatogram; however, peak intensities obtained with the split injection were approximately 30 times higher than those obtained with splitless injection. It led us to discover the other oxidative product ions of α-T, including those with m/z values of 445.368, 463.378, and 447.383, as described in Fig. 1 and the literature. 19)

Separation of α α-T oxidative products
Although split injection shows drastic improvement in column e ciency compared to splitless injection, extracted ion chromatograms for m/z 429.373 and 430.381 show an unresolved peak. e isomers, that are detected on the same m/z on the PTR MS must be resolved via chromatography to e ectively apply the tandem MS technique to select a precursor ion. We have further evaluated the peak resolution using L-column3 C18 with a 2.1-mm i.d. and a 100-mm length. Figure 8 shows a comparison of α-T chromatograms for the freshly prepared sample (bottom) and 21 days incubated sample (top) obtained using L-column3 with split injection and 10 pmol of α-T sample applied to the SFE frit. No chromatographic peaks were detected at m/z 429.373, 447.383, or 445.368 in the freshly prepared sample. We observed several chromatographic peaks in extracted ion chromatograms at m/z 429.373, 447.383, and 445.363, which correspond to the α-T oxidative products in addition to 430.381 and 431.388 on 21 days of incubated samples as shown in Fig. 8 (top). By contrast, no chromatographic peaks except for m/z 430.381 and 431.388 were observed from freshly prepared α-T samples shown in Fig. 8 (bottom), although chromatograms for m/z 429.373, 447.383, and 445.363 were not plotted for better visibility. e peak parameters for all the detected peaks are listed in Table 2, and the mass spectra obtained at the peak apex are presented in Fig. 9. An ion peak appeared at m/z 430.381, corresponding to the α-T-radical, was consistently observed alongside the protonated α-T ion at the same retention time of 82.3 s, without exception. erefore, it might have been generated during the PTR ionization process.
Two prominent peaks were observed for m/z 429.373 at 57.0 and 62.0 s, besides 82.5 s. ese two peaks were not shown on a freshly prepared sample (Fig. 8, bottom), thus they were considered as incubation products, although we do not have any information to assign them to (b), (c), or something else.
Two peaks observed for m/z 430.381 at 57.1 and 62.0 s may correspond to an isotope of m/z 429.373.
Two peaks with retention times of 46.5 and 56.9 s appeared on an m/z 447.383 chromatogram are supposed to be molecules (d) and (f). e log P value suggests that α-TQ (f) may exhibit a shorter retention time than molecule (d).
A peak observed at m/z 449.399 at 82.8 s listed in Table 2 bottom may correspond to α-tocopherylhydroquinone. One of the two peaks observed for m/z 479.373 may be ascribed to the molecule (g), and one of the peaks observed at m/z 463.378 may correspond to 5-formyl-γ-tocopherol+ OH, according to Tang et al. 19) All the peaks appearing on the chromatogram at the same m/z are separated by a peak resolution (Rs) of at least 0.86, making it possible to select precursor ions for tandem MS. We are working to reduce the instrumental dispersion at the PTR ow tube, which is currently approximately two seconds. We aim to minimize the dispersion that gives a minimum peak width of less than one second so that all the peaks listed here are fully resolved at Rs>1.5. e analysis of fat-soluble vitamins in the biological samples is not a trivial issue in liquid chromatography but it has been done well by using SFC. 22,23) e analysis of tocopherols in natural products such as wheat germ or vegetable oil by using SFE/SFC was also reported in past decades. 24) However, the analysis of α-T oxidative product has been studied since it was extremely complex and is a  short lifetime. Since it is a non-enzymatic reaction, the enzyme kinetics study is not applicable to studying plant selfdefense systems from 1 O 2 . Although several issues need to be addressed such as peak dispersion in the ion source, the present method is great potential to study plant physiology.

CONCLUSION
We have evaluated the SFE/SFC-PTR MS to separate α-T and its oxidative products that play a crucial role in the antioxidant defense system of a plant cell during photosynthesis. e mechanism and reaction scheme of α-T antioxidant function have been recently investigated using multiple analytical instruments, including chromatographic separation and complex sample processing. e combination of SFE with SFC and PTR MS has achieved almost the complete peak resolution of isomers reported in the literature in less than 2 min of chromatography by simply applying the sample to a stainless steel frit. Structure determination still requires the use of tandem MS to identify molecules; isomers must be resolved before ionization to select precursor ions. e present study demonstrated that SFE/SFC split injection can e ectively separate α-T oxidative products without requiring dedicated sample preparation techniques such as solid-phase extraction. is method achieved a fast chromatography separation rate with high detection sensitivity and exhibits excellent potential for analyzing complex sample matrices, including cells and tissues. It may help detect molecules from a single cell or small amounts of tissues.

Data Availability Statement
e spectrum data les of Fig. 9 are available in J-STAGE Data.