A novel bicyclic lactone and other polyphenols from the commercially important vegetable Anthriscus cerefolium

Garden chervil, Anthriscus cerefolium (L.) Hoffm. is an important herb commonly applied in Norwegian large-scale commercial kitchens. This species is a highly enriched source of phenolics, containing 1260 mg gallic acid equivalents (GAE) 100–1 g DM, however, the individual phenolic compounds have been scarcely characterized. Here we report on the qualitative and quantitative content of phenolics in garden chervil. The structure of the main phenolic compound was elucidated to be the previously undescribed compound 1,3-dicaffeoyl-5-malonyl-δ-quinide (1) by means of 1D- and 2D NMR and high-resolution mass spectrometry. The known flavones apigenin 7-O-β-(2″-apiofuranosylglucopyranoside) (= apiin) (2), apigenin 7-(2″-apiosyl-6″-malonylglucoside) (3) and luteolin 7-glucoside (4) were also identified. Compound 3 is reported for the first time from this plant species. The main phenolic compound, 1,3-dicaffeoyl-5-malonyl-δ-quinide, exhibited moderate cytotoxicity towards acute monocytic leukaemia cells (MOLM-13) and rat kidney epithelial cells (NRK) with EC50 between 400 and 600 µM.

Sample preparation and analysis. Sample preparations were performed as described by Slimestad et al. 14 . Fresh garden chervil, including leaves and stems, were lyophilized for 72 h using a freeze-dryer (CoolSafe 4 ScanVac, ScanLaf AS, Denmark). Dry matter contents were determined based on the sample weights prior to and after lyophilization. Dried plant material was minced in a bowl chopper and grinded (Bosch KM13, Slovenia). For determination of total phenolic content, radical scavenging capacity and UHPLC analysis, 200 mg of the sample was extracted with 10 mL methanol in a 20 mL test tube at ambient temperature for 48 h in the darkness. Samples were filtered through 0.4 μm syringe filters prior to analysis. Two parallel samples were analysed 14 . Fractionation and isolation. Fractionation and isolation of phenolics from garden chervil was based on extracts obtained from 500 g fresh plant material (160 g dry weight). The plant material was minced (about 20 mm), and extraction was performed two times for 24 h in the darkness by using 2 × 500 mL portions of methanol. The extract was filtered (folded filter quality 315, VWR Norway), concentrated to a volume of 100 mL on a rotavapor (Büchi, Switzerland), and partitioned against equal volumes of dichloromethane, in order to remove chlorophylls and lipophilic content. The water phase was further concentrated to a total volume of 50 mL, and the concentrated aqueous extract was applied to a bed of 0.5 kg Amberlite XAD-7 HP (Sigma) in a 5 × 60 cm open-top glass column, rinsed with 2 L distilled water and eluted by use of 2 L MeOH as mobile phase. The XAD-7 purified extract was finally concentrated to a volume of 50 mL.
Further purification was performed by size-exclusion chromatography by using a 5 × 100 cm open-top column filled with 500 g Sephadex LH20 (GE Healthcare, Norway). A step-gradient elution was used with increasing concentrations of methanol in the mobile phase (0, 20, 40, 60 and 80%; 0.1% TFA) 14 . Pure compounds were isolated by preparative HPLC. The HPLC instrument was equipped with a 250 × 20 mm, C18 Ascentis column. Two solvents were used for elution; mobile phase A (water-TFA 99.9:0.1 v/v) and mobile phase B (methanol-TFA 99.9:0.1 v/v). Portions of 200 µL were manually injected into the HPLC column and the collected fractions were transferred to HPLC vials for purity control using analytical HPLC 14 .
Total phenolic content. Total phenolic content was determined in accordance with the method of Price and Butler with stabilization of the Prussian Blue complex as described by Graham 12,13 and Slimestad et al. 14 . 100 μL of sample was diluted with 3 mL deionized water and mixed with 1 mL of a 0.1 M ferric chloride in 0.1 M hydrochloric acid solution together with 1 mL of 8 mM potassium ferricyanide solution. The reaction was allowed to run for fifteen minutes at ambient temperature. 5 mL of an acidic gum Arabic solution was added (1 g gum Arabic dissolved in 100 mL hot water. 10 mL of this solution was mixed with 10 mL 85% phosphoric acid and 30 mL water). Absorption at 700 nm was measured by use of an Agilent 8453 spectrophotometer (Agilent Technologies, Matriks, Norway). Samples were measured against a standard curve of gallic acid, and outputs are given as gallic acid equivalents, mg GAE g −114 .
Radical scavenging. The TEAC assay (Trolox Equivalent Antiradical Capacity) was carried out following the procedures previously described by Slimestad et al. 14 and Re et al. 15 . 2,2-azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) was dissolved in water to a 7 mM solution with potassium persulfate to a concentration of 2.45 mM. The solution was kept at ambient temperature for about 16 h. The ABTS ·+ solution was diluted with PBS (phosphate buffered saline: 100 mM KH 2 PO 4 -buffer, pH 7.4 and 150 mM NaCl), to an absorbance of 0.70 (± 0.02) at 734 nm. Samples were diluted so that, after the introduction of a 10 µL aliquot of each extract into the assay, they produced between 20 and 80% inhibition of the blank absorbance. After addition of 1.0 mL of diluted ABTS ·+ solution to 10 µL of extracts or trolox standards (final concentration 0-15 µM) in PBS, the absorbance reading was taken at 6 min. Appropriate PBS blanks were run in each assay. The final TEAC decolorization assay values were calculated against a standard curve of 0-10 mg Trolox in 100 mL methanol, and the percentage of inhibition of absorbance at 734 nm were expressed as mg TEAC 100 g −1 DW 14,15 .

U(H)PLC-MS.
The qualitative and relative quantitative contents of individual flavonoids and other phenolic compounds was determined by using an Agilent 1290 Infinity II instrument equipped with a 6120 quadrupole mass detector 14 . Separation was achieved with a Zorbax Eclipse XDB-C8 column (2.1 × 100 mm, 1.8 μm, Agilent Technologies). Water with 0.02% HCOOH (solvent A) and acetonitrile (solvent B) were used for gradient elution with the following time program (% B in A): from 0 to 10 (in 1 min), from 10 to 25 (in 25 min), from 25 to 95 (2 min), from 95 to 0 (1 min), and finally isocratic recondition for 1 min. Flow was set to 0.300 mL/min (max back pressure 540 bar), and injections of 5 μL was used. UV-detection was performed at 280, 320 and 360 nm Cytotoxicity assays. The cell lines used to study cytotoxicity were: Molm-13 16 17 . The cytotoxicity measurements were performed as described earlier 18 . In brief, the compound was dissolved in DMSO, and further diluted in DMSO. Using a pipette robot (Mosquito High Volume, SPT Labtech), 1 L was transferred to 96-well plates to create identical plates for testing on different cells. Cell suspension was then added, and the cells incubated for 72 h before assessment of viability using the WST-1 cell proliferation assay (Roce Applied Sciences). Some experiments were also performed with dilution series performed in culture medium to exclude the effect of DMSO on cell viability. After recording the WST-1 signal, the cells were fixed, DNA stained using Hoechst 33342, and cell death confirmed by microscopic evaluation of nuclear and surface morphology 18 .

Results and discussion
The dry matter content of chervil, Anthriscus cerefolium, was found to be 32.3% which is higher compared to the dry matter content of parsley, Petroselinum crispum, at 24.7% provided by the same grower 14 . The total phenolic content was found to be 1260 mg gallic acid equivalents (GAE) 100 g −1 , quite similar to that of parsley (1270 mg GAE 100 g −1 ) 14 . The antiradical potential was found to be 50 ± 5 mg TEAC 100 g −1 , similar to the level detected in parsley, 54 mg TEAC 100 g −114 and comparable to previous studies of chervil 11 .
A related compound with similar central bicyclic δ-quinide ring system to that of 1, namely 3,5-O-dicaffeoylepi-δ-quinide, has previously been reported to occur in processed coffee beans, where the identification was solely based on mass spectrometry 19 . Recently, Stojkovic et al. reported the presence of several derivatives of quinic acid acylated with caffeoyl and malonyl moieties, which were identified by hyphenated high resolution mass spectrometry (UHPLC-MS) 7 , however, none of these compounds were lactones.
Previously, extracts of chervil have exhibited weak to moderate cytotoxic activity towards cancer cell lines including gliblastoma cells, where EC 50 of 765 µg/mL was observed 7,20 , however, the cytotoxicity of pure compounds isolated from chervil has not been reported in current literature. The cytotoxicity of the main phenolic compound 1,3-dicaffeoyl-5-malonyl-δ-quinide (1) towards MOLM13, OCI-AML3 and MV4-11 acute leukaemia cells and towards normal cell lines (NRK cells and H9C2) (Fig. 2 shows results on MOLM13 and NRK cells after 24 and 72 h incubation). 1,3-dicaffeoyl-5-malonyl-δ-quinide (1) exhibited moderate toxicity towards all cell lines tested with EC 50 values above 1000 µM at 24 h incubation, and between 400 and 600 µM at 72 h incubation ( Fig. 2 and data not shown). The compound did not show selective cytotoxicity towards any of the leukemia cell lines compared to the non-cancerous cell lines.