Identification of C-glycosyl flavones by high performance liquid chromatography electrospray ionization mass spectrometry and quantification of five main C-glycosyl flavones in Flickingeria fimbriata

Flickingeria fimbriata is commonly applied in China as a traditional Chinese medicine (TCM), however the quality control of it is incomplete. In this work, we aim to identify and quantify the structures of C-glycosyl flavones in F. fimbriata. High performance liquid chromatography-diode array detector (HPLC-DAD) and High performance liquid chromatography–electrospray ionization–multiple stage tandem mass spectrometry (HPLC–ESI–MSn) methods were combined to identify C-glycosyl flavones and determine their contents. Twenty acylated C-glycosyl flavones and ten non-acylated C-glycosyl flavones were identified for the first time in F. fimbriata on systematic MSn analysis via HPLC–ESI–MSn. The aglycones of all of these compounds were apigenin or chrysoeriol and were acylated with p-coumaric, ferulic, 3,4-dimethoxycinnamic or 3,4,5-trimethoxycinnamic acids. Furthermore, the quantification result suggest that two C-glycosyl flavones (vicenin-I and vicenin-III) with relative high contents were revealed to be more strongly acylated in F. fimbriata. The method is sufficiently precise, accurate, and sensitive for the qualitative and quantitative analysis of C-glycosyl flavones, which is expected to establish a standard for quality control and identification in this plant.


Introduction
Flickingeria fimbriata (Bl.) Hawkes is commonly used as a source of a valuable TCM called "Shihu", which is normal referred to Dendrobium genus such as Dendrobium officinale. And this medicinal plant is commonly used as "Shihu" in Guangdong, Guangxi and Hainan provinces. The main places of production of F. fimbriata were Guangdong, Guangxi and Sichuan provinces. Its efficacy in soothing lung irritation and relieving cough has been reported in Guangdong Chinese Materia Medicine Standards, which are exploited in the treatment of diseases including pneumonia, tuberculosis, bronchitis, asthma, and pleurisy [1]. Previous phytochemical studies on F. fimbriata mainly lied in the isolation and analysis of diterpenoids [2,3], phenanthrenes [4,5], sterols [6] and phenolic constituents [5]. However, quality control study in this medicinal plants is incomplete. Only trait morphological identification and microscopic identification methods were mentioned in Guangdong Chinese Materia Medicine Standards of this plants and it have no Open Access BMC Chemistry *Correspondence: huangyuechun218686@outlook.com † Yawen Wang and Zhiyun Liang contributed equally to this work 1 College of the First Clinical Medical, Guangzhou University of Chinese Medicine, Guangzhou 510405, China Full list of author information is available at the end of the article specific method to control quality. The previous study of F. fimbriata in our laboratory showed that 7-8 stable common peaks of flavonoids in the characteristic spectra were found using HPLC [7], five of which were characterized as non-acylated C-glycosyl flavones (vicenin-II, vicenin-I, schaftoside, isoschaftoside and vicenin-III) by ion-trap mass spectrometer. The other uncharacterized peaks still required further investigation to complete a quality control study. It was proved that flavone is a suitable compound for quality control study of F. fimbriata.
Flavonoids, as common and widespread secondary plant metabolites, distributed in all parts of plants. They present as glycosides in the vacuoles, leaves, stem, and roots of flowers [8]. Sugar substitution on the flavonoid skeleton may occur through hydroxyl groups, in the case of O-glycosides (O-glycosyl flavones), or directly to carbon atoms in the A ring in C-glycosides (C-glycosyl flavones) [9]. Contents and types of C-glycosyl flavones were ideal index for identifying plants from the same species for its high specificity [10]. Generally, the flavonoids classification depends on the nature of aglycones, sugars, and acylate groups. Some secondary plant metabolites occur in the form of acylated glycosyl flavones with benzoic acid and/or cinnamic acid moiety. The cinnamoyl groups including p-coumaroyl, feruloyl, 3,4-dimethoxycinnamoyl and 3,4,5-trimethoxycinnamoyl [11][12][13], and their differences lie in the number and/or position of hydroxy and methoxy substituents. Many compounds like diterpenoids and phenanthrenes in F. fimbriata are acylated with aromatic acids, such as trans-cinnamoyl acid [14] and methoxybenzoyl acid derivatives [3,5,14,15], suggesting that the aromatic acids could be synthesized in this plants. Moreover, O-methyltransferase (OMT) genes revealed the internal relations of cinnamic acids with different substituents, it was possible that OMTs might be associated with the formation of 3, 4-dimethoxycinnamate and 3, 4, 5-trimethoxycinnamate in biosynthesis of plants [16]. Additionally, acylated flavonoids have several health beneficial effects including anti-inflammatory [17,18] and antioxidant activity [19], and the acylation position on glucose is regard as a potential approach for the antioxidant and cytoprotective effects of flavonoid glycosides [20].
MS is important due to its applicability for analyzing herbal medicines. The application of electrospray ionization (ESI) enabled the analysis of flavonoid glycosides without derivatization [21]. Although distinction between glycosidic and aromatic acidic substituents of flavonoids is problematic, such as deoxyhexoside and coumaric acid, both of which lose a fragment of 146 Da [9], high-performance liquid chromatography (HPLC) combined with a diode array detector (DAD) could provide online UV spectrum for each individual peak in a chromatogram which displays different spectrums between the glycosidic and aromatic acid substituents of flavonoids. Additionally, HPLC-ESI-MS n , equipped with an ion trap (IT) mass analyzer can obtain a large number of fragmentation patterns and typical losses up to MS 4 [22] which could be used to identify many complex isomers of C-glycoside flavones [23]. By this way, the nature of aglycones and sugars as well as the position of sugar and acyl groups could be deduced in C-glycoside flavones. The determination by MS fragmentation of acylated O-glycosyl flavones is possible [24], and the O-glycosylation at 2′′ and at 6′′ positions could be deduced from it [25]. However, few systematic analysis of acylated C-glycosyl flavones in plants via HPLC-ESI/ MS n combining with HPLC-DAD method was afforded before this study.
To date, 20 acylated C-glycosyl flavones and 10 nonacylated C-glycosyl flavones were identified by HPLC-ESI-MS n and HPLC-DAD. These 30 compounds have not been reported yet in F. fimbriata. In addition, 12 batches of F. fimbriata were successfully quantitatively analyzed, which is expected to establish a standard for quality control and identification.
Twelve samples of natural medicinal parts of F. fimbriata were collected from different regions of China (Guangdong, Guangxi, and Sichuan provinces). Of these, 6 batches were from Guangdong province (No. FF1-FF6), 4 batches were from Guangxi province (No. FF7-FF10), and 2 batches were from Sichuan province (No. FF11-FF12) ( Table 1). The tested samples of F. fimbriata (12 batches) were authenticated by professor Yuechun Huang from The First Affiliated Hospital of Guangzhou University of Chinese Medicine, Guangzhou, China. The voucher specimens (No. FF20190701) were deposited in the School of Pharmaceutical Science, Guangzhou University of Chinese Medicine, Guangzhou.

Preparation of sample extraction
The air-dried and smashed F. fimbriata (0.5 g) samples from each batch material were accurately weighted and extracted with 50 mL of methanol, after being weighted with a vessel, in the case of the volatilization of methanol then refluxed for 4 h at 90 °C using a Jie Rui Er HH-4 constant temperature water bath (Jiang Su Jie Rui Er electric Co., Ltd., Jiang Su, China). The extractions were removed and cooled down. The extraction was weighted again, and methanol was added into the vessel to compensate for the lost weight. Of the filtrated extraction, 25 mL was accurately transferred into an evaporation pan. The resultant concentrated extractions were transferred to a 2 mL volumetric flask and diluted to the indicated volume (2 mL). The obtained extract was filtrated through a 0.22 μm pore-size nylon filter for MS analysis and 0.45 μm poresize nylon filters for quantitative analysis.

HPLC-ESI-MS n and HPLC-DAD analysis condition
Analysis was performed on an HPLC system equipped with a vacuum degasser, quaternary pump, auto-sampler, and ultraviolet detector (Thermo Separation Products Inc., Riviera Beach, FL, USA) and coupled with a Thermo Finnigan LCQ FLEET (Thermo Finnigan, Riviera Beach FL, USA) ion trap mass spectrometer, equipped with an electrospray ionization interface in negative ion mode. Chromatographic separations were carried out on a Kromasil 100-5 C 18 column (250 mm × 4.6 mm, 5 μm, Akzo Noble, Sweden), maintained at 35 °C. The mobile phases were acetonitrile (A) and 0.1% (v/v) formic acid (B), at a flow rate of 0.8 mL/min. The gradient elution program was 0-10 min, 14% A; 10-20 min, 14-16% A; 20-45 min, 16-22% A; and 45-80 min, 22-40% A, with an elution gradient. The injection volume was 5 μL each time. The detection wavelength was set to 340 nm. The optimized MS conditions were as follows: full-scan mode between m/z 50 and 1000, spray voltage 3.0 kV, capillary voltage fixed at − 35.0 V, capillary temperature 350 °C, sheath gas flow rate of 30 (arbitrary units), and auxiliary gas flow rate of 10 (arbitrary units). The data acquisition and the system control were performed using a Finnigan Xcalibur 2.0 advanced chromatography workstation (Thermo Quest Corporation, San Jose, CA, USA). HPLC-DAD analysis was performed using an Agilent 1100 system (Agilent, USA). The conditions were as same as the chromatographic separation method in HPLC-ESI-MS n .

HPLC quantitative analysis condition
Quantitative analysis was performed using an Agilent 1100 system (Agilent, USA). Chromatographic

Identification of chemical compounds
The extract of F. fimbriata was analyzed by HPLC-ESI-MS n and HPLC-DAD. The UV chromatogram at 340 nm is shown in Fig. 1a, and its total ion chromatograms (TICs) are shown in Fig. 1b. Flavonoids typically exhibit two major absorption bands in the ultraviolet region: Band I in the 320-385 nm region, representing B-ring absorption, and Band II in the 250-285 nm range, representing A-ring absorption [25]. These UV data are in accordance with C-glycosyl apigenin and chrysoeriol, respectively [11,26]. The majority of flavonoids with cinnamoyl acid have a UV spectrum with an intense Band I at approximately 330 nm and a small Band II at approximately 270 nm (Fig. 1c), as a result of the overlapped UV spectra [27]. C-glycosyl flavones, with the characteristic saccharides substitution directly attached to aglycone in ring A through a C-C bond, all show substituents in position 6 (C-6) and/or 8 (C-8) of the aglycone moiety [8]. Apart from glucose, monosaccharides including xylose, arabinose, and rhamnose are ubiquitous in plants [28]. Due to the cross-ring cleavages of the flavonoid saccharide residue, characteristic ions [Ag-H+42] − and [Ag-H+72] − were observed in the MS 4 spectra for the  [24,29]. In F. fimbriata, combining the loss of mass with the previously reported results, and considering the high contents of vicenin-II, vicenin-I, schaftoside, isoschaftoside and vicenin-III, the xylose, arabinose, and glucose moieties were found to be involved in glycosylation. The major fragmentation pathways concern the cross-ring cleavages of the saccharide residue and the loss of water molecules. In negative mode, the characteristic ions of sugars in C-glycosyl flavones lost 120 Da and 90 Da in the hexose substituents, and 90 Da and 60 Da in pentose substituents by crossing cleavages, respectively [30].
The acyl group types were identified by neutral losses, which are characteristics of the acyl group or the acylated glycosyl residue. The acylation of p-coumaroyl and feruloyl in the hydroxyl of the C-glucosylation sugar showed higher polarity when compared to the acylation of 3,4-dimethoxycinnamoyl and 3,4,5-trimethoxycinnamoyl. These four types of acyl groups all belong to the derivatives of trans-cinnamoyl, but differ in the number of hydroxy and methoxy substituents. Characteristic acyl-related product ions [M-H-Acyl] − and acid-related product ions [M-H-Acid] − were observed in the former two types in the CID MS 2 spectra, whereas in the latter  (26) two, only [M-H-Acid] − was be detected. That is to say, the lower polarity acylated-C-glycosyl flavones are without the loss of the radical acyl group neutral fragments. In the CID MS 2 spectra, the neutral fragment losses are 146 Da and 164 Da for p-coumaroyl (Fig. 3), and 176 Da and 194 Da for feruloyl (Fig. 4) in the hydroxyl of the C-glucosylation sugar, respectively [31], however, they are only 208 Da for 3,4-dimethoxycinnamoyl (Fig. 5) and only 238 Da for 3,4,5-trimethoxycinnamoyl (Fig. 6). Finally, the acylated C-glycosyl flavones we found are all acylated with p-coumaroyl, feruloyl, 3,4-dimethoxycinnamoyl or 3,4,5-trimethoxycinnamoyl on the hydroxyl in this work, and the majority of them are isomers.

Method validation
The standard curve regressions were based on data from five concentrations of each standard solution. The peak areas and standard concentrations of each flavonoid compound were linearly fitted to a linear relation of Y = AX + B, where X represents the injection amount (μg) and Y represents the peak area, measured by HPLC. The correlation coefficients were also calculated. As listed in Table 7, all the calibration curves showed good linearity in the injection amount range (μg) (R 2 > 0.999). The precision RSDs of the 5 compounds were 0.84-1.97%. The values for repeatability were 0.75-2.19%. To confirm the stability, a standard solution mixed with methanol was analyzed at 0, 2, 4, 8, 12, and 24 h to evaluate the stability of the solution. The results showed that the stability RSD ranged from 1.11 to 2.12%. The results showed that the HPLC method for vicenin-II, vicenin-I, schaftoside, isoschaftoside, and vicenin-III had an average assay recovery between 100.55 and 102.68% and a good reproducibility RSD ranged 0.68-1.49%. All the data indicated that this method is satisfactory for the qualitative and quantitative analysis of F. fimbriata.

Sample quantitative analysis
The proposed HPLC method was applied to analyze the five main compounds in the 12 batches of F. fimbriata  Fig. 7. The contents of vicenin-I and vicenin-III in all 12 batches of samples were higher than the other three compounds, as shown in Fig. 8. These quantitative results were in accordance with the results from MS n analysis, which showed that there was a greater acyl group substitution in vicenin-I and vicenin-III as well as pentose substitution on the flavonoid, referred to as xylose, as opposed to arabinose. Vicenin-II, vicenin-I, schaftoside, isoschaftoside, and vicenin-III were all found to be higher in the samples FF2, FF3, FF4, FF6, and FF10, as shown in Table 8. The differences in contents are due to many factors, including the growth environment, harvesting time, and growing years. Due to the lack of samples from different origins, this phenomenon should be studied in the future.

Contents of five di-C-glycosyl flavones in Flickingeria fimbriata
Quantification was based on an external standard method using calibration curves fitted by linear regression analysis. The validated HPLC method was subsequently applied to the determination of 12 batches of F. fimbriata, and the quantitative analysis of the five main di-C-glycosyl flavones are summarized in Table 8.
The quality of the F. fimbriata extracts was assessed by determining their flavonoid content. The contents of the 5 main di-C-glycosyl flavones eluted in the order of vicenin-II, vicenin-I, schaftoside, isoschaftoside, then vicenin-III, and the dried herbal material contents of them were 0.137-0.748 mg/g, 0.388-2.019 mg/g, 0.116-0.683 mg/g, 0.125-0.635 mg/g, and 0.463-2.154 mg/g, respectively.

Conclusion
Flickingeria fimbriata benefited by its superiority of medicine food homology which is widely applied in health industry. However, only morphological identification and microscopic identification methods reported in Guangdong Chinese Materia Medicine Standards indicated a lack of quality control. In this study, 20 acylated C-glycosyl flavones and 10 non-acylated C-glycosyl flavones were characterized for the first time in F. fimbriata using HPLC-DAD and HPLC-ESI-MS n , which laid a foundation of improving the standard for quality control and identification of F. fimbriata. The methods allowed us to identify several important structural characteristics of C-glycosyl flavones including (1) the nature of aglycone (apigenin or chrysoeriol) (2) types of sugar units (glucose, arabinose, or xyloside), (3) glycosylation position (6-C or 8-C), (4) types of acyl groups (p-coumaroyl, feruloyl, 3,4-dimethoxycinnamoyl, or 3,4,5-trimethoxycinnamoyl), and (5) acylation position (2″-O or 6″-O). Additionally, we found the isomers of 6-C and 8-C-glycosyl flavones almost coexisted in F. fimbriata. The acylated flavones also had following characteristics in this plants, when acylation occurs at the 2″-O position, sugar substitution for 6-C-glucosylation is more likely. When acylation occurred at the 6″-O position, sugar substitution for 8-C-glucosylation is more likely. And the acyl groups in the compounds analyzed in this article were all substituted on glucose.