Localization analysis of essential oils in perilla herb (Perilla frutescens var. crispa) using derivatized mass spectrometry imaging

Abstract The localization of essential oils, including flavor components, in perilla herb (Perilla frutescens var. crispa) were visually determined using matrix‐assisted laser desorption/ionization (MALDI) mass spectrometry (MS) imaging. The surface of a perilla leaf was peeled using a cyanoacrylate adhesion compound and contained oil glands that retained their morphology and chemical properties. We imaged the three essential oils perillaldehyde, β‐caryophyllene, and rosmarinic acid (RA). Perillaldehyde was derivatized using glycine to prevent evaporation and allow its detection and imaging while localized in oil glands. β‐caryophyllene also localized in the oil glands and not in the epidermis region. RA was detected throughout the leaf, including the oil glands. Quantitative data for the three essential oils were obtained by gas chromatography‐ or liquid chromatography‐MS. The concentrations of perillaldehyde, β‐caryophyllene, and RA were 12.6 ± 0.62, 0.27 ± 0.02, and 0.16 ± 0.02 [mg/g] in the paste sample of perilla herb. Peeling using a cyanoacrylate adhesion compound, and derivatization of a target such as an aroma component have great potential for mass spectrometry imaging for multiple essential oils.

selectively stain essential oils. The construction of antibodies for small molecules is difficult, which limits immunostaining methods for spatial visualization. Gas chromatography (GC)-MS analysis provides quantitative information and is suitable for volatile organic compounds (VOCs) such as aromatic compounds but provides no visual spatial information. Therefore, information on the localization of these essential oils is lacking and thus the development of selective visualization methods is required not only for fundamental plant science (Shiono & Taira, 2020) but also in fields such as food science (Cho et al., 2020;Enomoto et al., 2018) and omics science (Chen et al., 2019). Mass spectrometry imaging (MSI), and typically matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS), enables the direct mapping and imaging of biomolecules present in tissue sections (Stoeckli et al., 2001).
Tissue cross-sections are mounted on an electrically conductive glass slide, sprayed with an organic matrix, and irradiated with laser shots. Each spot irradiated by the laser becomes a pixel in the final image. Target-specific markers such as antibodies are not required, and MSI enables the simultaneous detection of multiple analytes in a single section of animal tissue (Fujii, 2018;Hase, 2019;Tatsuta, 2017) or plant tissue (Shikano, 2020;Shiono, 2017;Taira et al., 2015). We solved the difficulty of making sections of thin samples such as leaves by applying cyanoacrylate adhesive to the leaf surface, then peeling the adhesive off the leaf surface with indium tin oxide (ITO) glass ( Figure 1). In the present study, focusing on essential oils, we imaged three target compounds (PA (boil- Generally, organic compounds with a boiling point below 523 [K] are defined as a VOC. Imaging VOCs under vacuum is problematic due to their volatility, and thus, we developed a method of amino acid derivatization by nucleophilic addition reaction of amino group of glycine. The aldehyde group of PA reacts with the primary amine of glycine, increasing its BP value above that of a VOC. We compared the localizations and concentrations of these three essential oils in fresh perilla and perilla paste by MALDI-MSI and GC-and liquid chromatography (LC)-MS.

| Sample preparation
Standard (-)-perillaldehyde was purchased from Tokyo Chemical Industry. β-caryophyllene was purchased from Wako. Rosmarinic acid was purchased from Sigma-Aldrich. Perilla herb (Perilla frutescens) was purchased at a supermarket and stored at 4°C prior to extraction and MSI analysis. The sample was also ground in a mortar with a pestle to provide a paste for GC-and LC-MS analysis.

| Solid phase micro extraction (SPME) fiber
An SPME fiber (length 10 mm) coated with 50/30 mm divinylbenzene/carbon WR/polydimethylsiloxane (DVB/CAR/PDMS) phase (Restek) was used to extract volatile compounds. The fibers were conditioned before use and thermally cleaned between analyses by inserting the fibers into the injector port of a gas chromatography system set at 270°C for 30 min in a stream of helium.

| HS-SPME/GC-MS analysis
Headspace SPME that is a clean-up of procedure was used to extract headspace volatiles from the samples. The Perilla frutescens sample was ground in a mortar with a pestle to obtain a paste, and then, 100 mg of paste was placed in a 20 ml headspace vial fitted with a silicone septum. Cyclohexanone (10 µl, 1,000 μg/ml in methanol) as an internal standard was added. After equilibration for at least 10 min, SPME sampling was performed by exposing the fiber for 30 min in the headspace of the sampling vial at 40°C. The SPME device was placed into a GC Ltd.). The column temperature was programmed to hold at 40°C for F I G U R E 1 Schematic illustration of perilla herb leaf and the peeling method using an adhesion compound 5 min, subsequently raised to 250°C at 5°C/min, and held for 15 min.
The injector temperature was set to 250°C. The flow rate of the carrier gas (helium) was 1.0 ml/min. The MS detector was operated in electron impact ionization mode at 70 eV. The analysis was performed in the SCAN mode in the 30-550 m/z range. Tentative identification of constituents was based on comparison of the retention time and mass fragmentation with pure standards, and on computer matching with commercial mass spectra libraries (NIST & WILEY) and a home-made library based on pure compounds, which were analyzed under identical conditions. Furthermore, the volatile compounds in raw Perilla frutescens samples with intact oil glands were analyzed. Briefly, raw Perilla frutescens sample (100 mg) was placed in a 20 ml headspace vial fitted with a silicone septum. After the addition of an internal standard, the headspace gas was analyzed by HS-SPME/GC-MS under the same analytical conditions as described above. Perillaldehyde content in the Perilla frutescens sample was determined by preparing calibration curves of perillaldehyde, with cyclohexanone as an internal standard.

| Preparation of perilla herb peeling sections for mass spectrometry imaging (MSI)
Cyanoacrylate-type adhesive compound (Aron Alpha, TOAGOSEI) was dropped on backed perilla herb. A glass slide coated with ITO (Bruker Daltonics) was pressed on the sample to transfer epidermal tissue.
The release paper film was pressed on the peeled epidermal tissue to crush the oil glands. Optical images of the tissue were obtained using a microscope scanner (Nanozoomer, Hamamatsu Photonics, Shizuoka, Japan) before analysis by MALDI-MSI (Supporting information).

| MALDI-MSI
A 10 mg/ml solution of the matrix α-cyano-4-hydroxycinnamic acid (CHCA, Nacalai Tesque) was suspended in 6 ml of acetonitrile/water/ trifluoroacetic acid (50/49.9/0.1 v/v) and sprayed on the perilla herb tissue sections on the ITO-coated glass slides (Bruker Daltonics) using an automated pneumatic sprayer (TM-Sprayer, HTX Tech). Ten passes were sprayed using the following conditions: flow rate, 120 µl/min; air flow, 10 psi; nozzle speed, and 1,100 mm/min. Ionization and imaging of the essential oils were confirmed with a MALDI-TOF-MS (rapifleX, Bruker Daltonics). Tandem MS spectra of the sections were obtained using a collision-induced dissociation method. Precursor ion (obtained using 1000× shots) and fragment ion (obtained using 4000× shots) signals were integrated using fl-

| Quantitative analysis for target molecules in perilla herb
PA and βC standards were separated using the same gradient conditions by GC-MS (Nexis, GC-2030, Shimadzu). We selected two different ions that corresponded to PA and βC (m/z 67.0 and 93.0).
The PA, βC, and RA contents (mg/g sample) were calculated from the peak areas in the chromatograms and by using calibration curves.
The concentrations of PA and βC in perilla paste were 12.6 ± 0.62 and 0.27 ± 0.02 [mg/g] sample and in fresh perilla were 0.20 ± 0.07 and 0.03 ± 0.006 [mg/g] sample, respectively. The concentrations of PA and βC in the paste sample were 63 and 9 times that in the fresh sample, showing that essential oils are mostly stored in oil glands.
We therefore pressed a peeled perilla herb section to image essential oils. For RA, we used UPLC-MS since RA is not a VOC. The concentration of RA was 0.16 ± 0.02 [mg/g] sample (Table 1).   is difficult because the signal intensity depends on many factors (e.g., ionization efficiency, extraction efficiency from the tissue, and sample preparation), mass spectrometry imaging is a powerful tool for the direct visualization of fruit compounds and biomolecules in biological tissues (Taira, 2008;Tatsuta, 2017), and may have applications in agriculture relevant to commercial food products.

ACK N OWLED G EM ENTS
This work was partly supported by the Japan Science and Technology (JST) Grants-in-A-STEP (VP30118067678 to S.T.).