UHPLC–MS/MS Analysis on Flavonoids Composition in Astragalus membranaceus and Their Antioxidant Activity

Astragalus membranaceus is a valuable medicinal plant species widely distributed in Asia. Its root is the main medicinal tissue rich in methoxylated flavonoids. Origin can highly influence the chemical composition and bioactivity. To characterize the principal chemicals influenced by origin and provide more information about their antioxidant profile, the extracts of A. membranaceus roots from four origins were analysed by UHPLC-MS/MS. Thirty-four flavonoids, including thirteen methoxylated flavonoids, fifteen flavonoid glycosides and six flavonols, were identified. By principal component analysis, eighteen identified compounds were considered to be principal compounds. They could be used to differentiate A. membranaceus from Shanxi, Inner Mongolia, Heilongjiang and Gansu. The antioxidant activity was analysed by ORAC assay, DPPH radical scavenging activity assay and cell antioxidant activity assay. ‘Inner Mongolia’ extract showed the highest antioxidant activity. These results were helpful to understand how origin influenced the quality of A. membranaceus.


Introduction
Astragalus membranaceus (Fabaceae) is a well-known traditional Chinese herbal medicine, which is mainly distributed in Asian regions [1]. The root of A. membranaceus is the medicinal tissue widely used due to their beneficial effects to lung health, which has been further proved to exhibit immunomodulatory, antioxidant, antiperspirant, antidiarrheal, and antidiabetic activities [2]. The bioactive compounds in the root of A. membranaceus are complicated. Previous research has revealed that the main bioactivities depend on the presence of non-volatile components, such as polysaccharides, saponins, and phenolics [3,4].
Flavonoids are a natural bioactive compound with C6-C3-C6 skeleton. They are commonly found in plant species, and possess good bioactivities which have been applied in nutraceutical, medicinal, and cosmetic products [5]. Many different subclasses of flavonoids have been described from A. membranaceus including flavone, flavonol, flavanone, flavanonol, chalcone, aurone, isoflavone, and pterocarpan [6]. Phenolics, including flavonoids, have been judged as the marker for quality evaluation and standardization of A. membranaceus and its processed products [7]. Methoxylated flavonoids and their glycosides, such as calycosin and calycosin-7-O-β-D-glucoside, were the major bioactive constituents of A. membranaceus due to their superior bioactivities [8,9]. After glycosylation and methylation, the reactivity and solubility of the flavonoids usually improved and thereby so did their absorption and bioactivity [10]. Many methoxyl flavones and flavonoid glycosides have been proved the superior antioxidant activities [11,12]. Due to the plant origin and extremely low toxicity, they have become the hotspots of natural antioxidants drug discovery and development [13,14].
The accumulation of bioactive compounds depends on geographical location and growing environment, including climate, soil, and fertilizer [15]. It is highly correlated to the health benefits and pharmaceutical activities. Different origins usually lead to different phytochemical profiles, which further relate to the quality of A. membranaceus. However, relevant information regarding this topic is limited. Therefore, in the present study, A. membranaceus from four production origins (Gansu, Shanxi, Inner Mongolia, and Heilongjiang provinces of China) were collected and the flavonoids were identified by ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS). Afterward, the antioxidant activities were investigated by ORAC assay, DPPH radical scavenging activity assay and CAA assay. These results help to understand the qualities of A. membranaceus from different origins.

Extraction
The bioactive compounds were extracted as explained previously by Wu et al. [16] with some modifications. Five hundred milligrams of A. membranaceus root powder were extracted with 5 mL of methanol for one week in dark at room temperature. The extract was centrifuged at 9000× g for 20 min. Then the supernatants were collected and concentrated at 45 • C by a vacuum rotary evaporator (Eyela N-1100 V-W, Tokyo Ri-kakikai Co. Ltd., Tokyo, Japan). The aliquots were stored at 4 • C before use. The extraction was conducted in triplicate.

Determination of Total Phenolics and Flavonoids Contents
The content of total phenolics was determined by using Folin-Ciocalteu colorimetric protocol [17]. Briefly, an aliquot (1 mL) of the extract or standard solution of gallic acid (0, 20, 60, 100, 150, 200, 300, 400, 500, and 600 µg/mL) was mixed with 0.4 mL of distilled water and incubated for 6 min with 0.1 mL of Folin-Ciocalteu phenol reagent. Then 1.0 mL of Na 2 CO 3 (7%, w/v) and 0.8 mL of distilled water were added to the mixture. After incubation for 90 min at room temperature, the absorbance was measured at 760 nm by using a multi-mode microplate reader (Spark, Tecan Group Ltd., Männedorf, Switzerland). The total phenolics contents were finally expressed as milligram of gallic acid equivalents per 100 g dry weight of A. membranaceus (mg GAE/100 g DW).

UHPLC and UHPLC-MS/MS Analyses
UHPLC analysis was applied to isolate the bioactive compounds presented in A. membranaceus. A Thermo Scientific Ultimate 3000 UHPLC system coupled with an Orbitrap Elite (Thermo Fisher Scientific, Waltham, MA, USA) was used. A C18 column (3.0 × 100 mm, 1.8 µm, Agilent, Santa Clara, CA, USA) was equipped. The samples were eluted with a gradient system consisting of solvent A (0.1% formic acid in water, v/v) and solvent B (0.1% formic acid in acetonitrile, v/v). A gradient elution program was conducted as follows: 0-3 min, 3% solvent B; 3-24 min, from 3% to 95% solvent B; 24-29 min, 95% solvent B; 29-30 min, from 95% to 3% solvent B; 30-35 min, 3% solvent B. The flow rate was 0.4 ml/min and the injection volume was 2 µL. The column temperature was maintained at 35 • C. Electron spray ionization of the analytes was used in positive mode. Mass spectra were collected in the full-scan mode in a mass range of 100-1000 Da. The MS/MS spectra were obtained with collision energy of 25 eV.

Determination of Oxygen Radical Absorbance Capacity (ORAC)
The oxygen radical absorbance capacity assay was conducted by following the protocol of literature [19]. The fluorescein sodium salt and ABAP were dissolved in 75 mM PBS buffer (pH 7.4), respectively. A total 20 µL of sample solution or positive control or blank were added in 96-well microplate (Corning Scientific, Corning, NY, USA) and incubated for 30 min at 37 • C. Then, 200 µL of fluorescein sodium salt (0.96 µM) were added. After incubating for 20 min, 20 µL of ABAP (119 mM) were added to the mixture and measured immediately at 37 • C. The fluorescence generation was monitored every 1.5 min for 60 cycles by using a multi-mode microplate reader (excitation wavelength of 480 nm, and emission wavelength of 520 nm). The Trolox equiv (TE) solutions (6.25, 12.5, 25, and 50 µM) were used as the positive control, and 75 mM phosphate (pH 7.4) was used as the blank. ORAC values were expressed as µmol of Trolox equivalents per gram of A. membranaceus methyl extract (µmol TE/g) on the basis of the regression equation between Trolox concentration and the net area under the curve (AUC).

Determination of DPPH Radical Scavenging Activity
The DPPH radical scavenging activity of extract was evaluated by using the method described previously [20]. The DPPH in methanol (100 µM) was prepared freshly, and ascorbic acid (AA) was prepared as the positive control. A total 180 µL of DPPH in methanol were added into 20 µL of sample solutions and incubated for 30 min in dark at room temperature. Afterward, the absorbance was measured at 517 nm immediately. The vehicle control was set up by using methanol instead of the sample. The DPPH scavenging activity was calculated according to the following equation: Scavenging activity (%) = (1 − absorbance of sample/absorbance of control) × 100, and the IC 50 value was calculated on the scavenging activity against DPPH radicals and expressed as µmol of ascorbic acid per gram of A. membranaceus extract (µmol AA/g).
The cytotoxicity assay was performed by the MTS staining method. Briefly, HepG2 cells were seeded in a 96-well microplate with a concentration of 4 × 10 4 cells/well. After incubation for 24 h, the cells were collected by removing the growth medium from each well and were washed by PBS buffer. The sample with various concentrations (0, 25, 50, 100, 200 µg/mL) was used to treat the cells for 24 h. MTS solution (50 µg/mL) was added to each well and incubated for 4 h. The absorbance was measured at 570 nm.

Cellular Antioxidant Activity
The cellular antioxidant activity was analysed as previously reported [19]. Briefly, HepG2 cells were seeded at 6 × 10 4 cells/well in a black 96-well microplate and inoculated at 37 • C for 24 h. The growth medium was then removed and each well was washed with 100 µL of PBS buffer. Afterward, the wells were treated with 100 µL of WME medium containing different concentrations of samples (0, 3.125, 6.25, 12.5, 25, 50 µg/mL) and 50 µM 2,7-diacetate dichlorofluorescein (DCFH-DA) were added to the cells and incubated for 1 h (Trolox at 3.125, 6.25, 12.5, 25, and 50 mM were used as control). The treatment solution was removed, and the 96-well microplate was washed three times with PBS, followed by the addition of an oxidant-supplemented medium (HBSS with 10 mM HEPES, 600 µM ABAP). The 96-well microplates were transferred to a multimode microplate reader. The fluorescence was recorded every 5 min for 1 h at 37 • C with an emission wavelength of 538 nm and excitation wavelength of 485 nm. After subtraction of the blank from each sample measurement, the area under the curve for fluorescence versus time was integrated to calculate CAA value as follows: CAA (units) = 1 − ( SA/ CA). In this equation, SA is the integrated area under the fluorescence-time curve of samples, and SA represents the integrated area under the curve of the control. By referring to the regression equation of concentration and CAA value, the EC 50 values and quercetin equivalents were calculated.

Statistical Analysis
All the experiments were performed in triplicate and the results were expressed as mean ± standard deviation. The peak heights of chemicals in mass spectra were recorded and averaged over three replicates. Principal component analysis (PCA) was performed to obtain an overview of variations between data. Meanwhile, a supervised multivariate statistical method, orthogonal partial least square discriminant analysis (OPLS-DA) mode was built to find the analytes in the data with variable importance in the projection. Moreover, variable importance in projection (VIP) is an important index. The differential constituents were selected based on VIP values > 1 and p-values from Student's t-test < 0.05 [21]. Heatmap was applied to visualize the number of differential metabolites and to analyse their hierarchical cluster. The results were subjected to analysis of variance and differences between means by using ANOVA test followed by Tukey's test or Dunnett's test. In all the tests, the differences between results were regarded as significant at p < 0.05.
of phenolics varied between 135.23-197.40 mg GAE/g extract and in a decreasing order of 'Inner Mongolia' (197.40 ± 1.95 mg GAE/g extract), 'Heilongjiang' (164.93 ± 2.14 mg GAE/g extract), 'Shanxi' (155.62 ± 0.95 mg GAE/g extract), and 'Gansu' (135.24 ± 1.66 mg GAE/g extract), respectively. Among them, the 'Inner Mongolia' and 'Heilongjiang' samples presented significantly (p < 0.05) higher levels of phenolics than the other two origins. Meanwhile, the total flavonoids contents of A. membranaceus were significantly different among the four origins. It ranged from 52.27 to 112.75 mg CE/g extract. Among them, the 'Inner Mongolia' samples (112.75 ± 0.77 mg CE/g extract) presented significantly higher (p < 0.05) flavonoids contents than the other three origins, followed by 'Gansu' (67.22 ± 3.67 mg CE/g extract) and 'Shanxi' (69.12 ± 6.59 mg CE/g extract) samples. Unlike the trend of phenolic compounds, a relatively low flavonoids content was observed in 'Heilongjiang' sample (52.27 ± 4.63 mg CE/g extract), which indicated the presence of non-flavonoids phenolics. It is widely known that a diet rich in fruit and vegetables has a protective effect against cancer insurgence and development. In the presence of an intense stressing event, cells activate specific responses to counteract cell death or senescence, which is known to act as a key task in the onset of age-related pathologies and the loss of tissue homeostasis [22]. Phenolics are generally recognized to be responsible for antioxidant and anti-aging effects, as well other beneficial actions [23]. Previous research showed that the total phenolic content of A. membranaceus from Western Siberia ranged from 100-190 mg GAE/g extract [24], which is consistent with this study. The total phenolics and flavonoids contents of A. membranaceus from Shijiazhuang origin were 27.646 ± 0.11 mg GAE/g extract and 7.048 ± 0.87 mg RE/g extract, respectively [25]. The difference of total phenolics and flavonoids contents distinguished the A. membranaceus materials from different origins clearly.

Identification of Bioactive Compounds
The UHPLC-MS/MS analysis of the methanolic extract was performed to characterize the bioactive compounds of A. membranaceus. The total ion chromatogram was presented in positive ion mode. After preliminary comparative analysis by retention time, MS/MS fragments, and the reported data in references or Compound Discoverer, 34 flavonoids were identified, including 13 methoxylated flavonoids, 15 flavonoid glycosides, and 6 flavonols. The compounds identified by UHPLC-MS/MS spectra are summarized in Table 1, and their putative fragmentation pathways are shown in Figure 2. The identification of methoxylated flavonoids and flavonoids glycosides were explained in detail as follows. It is widely known that a diet rich in fruit and vegetables has a protective effect against cancer insurgence and development. In the presence of an intense stressing event, cells activate specific responses to counteract cell death or senescence, which is known to act as a key task in the onset of age-related pathologies and the loss of tissue homeostasis [22]. Phenolics are generally recognized to be responsible for antioxidant and anti-aging effects, as well other beneficial actions [23]. Previous research showed that the total phenolic content of A. membranaceus from Western Siberia ranged from 100-190 mg GAE/g extract [24], which is consistent with this study. The total phenolics and flavonoids contents of A. membranaceus from Shijiazhuang origin were 27.646 ± 0.11 mg GAE/g extract and 7.048 ± 0.87 mg RE/g extract, respectively [25]. The difference of total phenolics and flavonoids contents distinguished the A. membranaceus materials from different origins clearly.

Identification of Bioactive Compounds
The UHPLC-MS/MS analysis of the methanolic extract was performed to characterize the bioactive compounds of A. membranaceus. The total ion chromatogram was presented in positive ion mode. After preliminary comparative analysis by retention time, MS/MS fragments, and the reported data in references or Compound Discoverer, 34 flavonoids were identified, including 13 methoxylated flavonoids, 15 flavonoid glycosides, and 6 flavonols. The compounds identified by UHPLC-MS/MS spectra are summarized in Table 1, and their putative fragmentation pathways are shown in Figure 2. The identification of methoxylated flavonoids and flavonoids glycosides were explained in detail as follows.

Flavonoid Glycosides
Flavonoid glycosides, including flavonoid C-glycosides and O-glycosides, were two common patterns distributed in plants with multiple bioactivities [29]. In this study, 15 flavonoid O-glycosides were found in A. membranaceus, while 12 of them were identified to bound one or two methyls. After comparing the quasi-molecular ions and fragment ions in MS/MS spectra with those of previously reported literature [30,31], they were identified as narcissin (1)

Multivariate Statistical Analysis
PCA analysis is an unsupervised method usually employed to determine patterns between multivariate samples. The PCA analysis showed a clear tendency of separation among A. membranaceus samples from four origins. Specifically, the first two principal components explained 77.9% of the total variance, in which the first principal component explained 51.2% and the second principal component explained 26.7%. The first principal component was represented by the compounds such as nicotiflorin, liquiritin, ononin, 7-hydroxy-2 -methoxy-4 ,5 -methylenedioxyisoflavane, calycosin 7-O-{6 -[-but-2-enoyl]}glucoside, methylinissolin, isoliquiritigenin, daidzein, and astragaisoflavane D. Most of them presented higher content in the 'Inner Mongolia' extract. The supervised discriminant OPLS-DA was performed to classify the samples from four regions and find out the differential compounds. A model with R 2 X of 0.997, R 2 Y of 0.998, and Q 2 of 0.99 was constructed. R 2 Y > 0.9 indicated an excellent fitted model, and Q2 > 0.9 suggested a good repeatability and predictability of the model. As shown in Figure 3A, no serious outlier was observed. The outlier is the plot out of the ellipse, which is defined as the Hotelling's T2 range 95% confidence [35]. The samples from different origins exhibited good separation. The 'Heilongjiang' sample and 'Inner Mongolia' sample were separated significantly in t [1] direction. 'Gansu' sample and 'Shanxi' sample located near the centre of the model plane and stayed close to each other, which indicated that their chemical compositions were similar. According to the searching rule of VIP value > 1 and p-value < 0.05, a total of 18 phenolics were regarded as principal metabolites marked in red in Figure 3B  The heat map was applied to demonstrate the variation of the identified compounds. As shown in Figure

ORAC Value and DPPH Radical Scavenging Activity
ORAC assay is a valid method to evaluate the antioxidant ability through monitoring the inhibition capacity against peroxyl radical [36]. The breakdown of ABAP can provide peroxyl radical and leads to subsequent oxidation. It can be monitored through fluorescent intensity change. As shown in Figure 4A, 'Inner Mongolia' (628.94 ± 3.30 µmol TE/g extract) showed significant (p < 0.05) higher ORAC values. The ORAC values of other three samples were in a decreasing order of 'Gansu' (553.18 ± 15.28 µmol TE/g extract), 'Shanxi' (522.48 ± 21.91 µmol TE/g extract), and 'Heilongjiang' (471.29 ± 8.61 µmol TE/g extract). It indicated that 'Inner Mongolia' and 'Gansu' extracts processed good peroxyl radical inhibition activities.

ORAC Value and DPPH Radical Scavenging Activity
ORAC assay is a valid method to evaluate the antioxidant ability through monitoring the inhibition capacity against peroxyl radical [36]. The breakdown of ABAP can provide peroxyl radical and leads to subsequent oxidation. It can be monitored through fluorescent intensity change. As shown in Figure 4A, 'Inner Mongolia' (628.94 ± 3.30 μmol TE/g extract) showed significant (p < 0.05) higher ORAC values. The ORAC values of other three samples were in a decreasing order of 'Gansu' (553.18 ± 15.28 μmol TE/g extract), 'Shanxi' (522.48 ± 21.91 μmol TE/g extract), and 'Heilongjiang' (471.29 ± 8.61 μmol TE/g extract). It indicated that 'Inner Mongolia' and 'Gansu' extracts processed good peroxyl radical inhibition activities. DPPH test is widely used to evaluate the antioxidant capacity of phenolics [37]. It is based on a stable nitrogen-centred free radical that is characterized by absorbance at 517 nm with a deep violet colour. In the presence of free radical scavenger, the absorbance of DPPH will decrease due to hydrogen donation from antioxidant with a dose-dependent behaviour [38]. As shown in Figure 4B, 'Heilongjiang' extract presented a higher DPPH radical scavenging activity with an IC 50 value of 8.10 ± 0.54 µmol AA/g extract, followed by 'Shanxi' extract (IC 50 of 6.63 ± 0.30 µmol AA/g extract) and 'Inner Mongolia' extract (IC 50 of 6.57 ± 0.40 µmol AA/g extract). The 'Gansu' extract possessed the lowest IC 50 value of 5.89 ± 0.36 µmol AA/g extract. This order was inconsistent with that of ORAC assay.
The occurrence of oxidation process is correlated with the existence of a surplus of free radicals, which are responsible for multiple diseases [39]. Peroxyl radicals are characterized as free radicals that predominate in the lipid oxidation of biological system [40]. Its inhibition plays an important role in disease prevention. Generally, the radical-trapping antioxidant activity of flavonoids is related to the hydrogen atom transfer to a peroxyl radical [41]. It can be used to explain the different radical-inhibitory activities of four origins samples. A. membranaceus from Lebanon possesses the lowest IC50 value of 102 ± 4.4 µg/mL [42]. Several bioactive compounds have been confirmed. For example, formononetin, calycosin, and calycosin-7-O-glucoside showed superior antioxidant activity and inhibited free radicals generated by DPPH in a dose-dependent manner [43]. Moreover, calycosin can enhance antioxidant enzymatic activities such as glutathione peroxidase, catalase, superoxide dismutase and attenuate H 2 O 2 -induced H 9 C 2 cell apoptosis rate in a dose-dependent manner as well [44]. Besides, the chemicals of formononetin from A. membranaceus have evidenced the capacity of inhibiting xanthine oxidase-induced cell injury significantly [45]. They were the major flavonoids in A. membranaceus.

Cellular Antioxidant Activity (CAA)
HepG2 cells line is a sensitive cell model in the determination of antioxidant biomarkers [46]. In the present work, the cytotoxic effects of A. membranaceus extract at different levels (25,50,100, and 200 µg/mL) against HepG2 cells were determined by MTS assay. From the results summarized in Figure 4C, no significant cytotoxicities were observed between extract-treated cells and untreated cells within concentration of 0-50 µg/mL. It indicated that this range could be used for cellular antioxidant activity assay.
The cellular antioxidant activities of A. membranaceus extracts were evaluated by CAA assay, and the results are shown in Figure 4D. It was obvious that the A. membranaceus extracts from four origins could protect HepG2 cells against peroxyl radicals with a dosedependent effect. Among them, 'Inner Mongolia' sample showed a higher CAA value than the others at 3.125 µg/mL, while 'Gansu' and 'Shanxi' extracts exhibited higher CAA values at 25 µg/mL. Trolox was used as positive control.
CAA assay is performed based on polarity, solubility, and molecular weight of the antioxidant and provides an important tool for the biological assessment of antioxidant activity [47]. In the CAA assay, the DCFH-DA is preloaded into the cell, treated with the intracellular peroxyl radicals generated from ABAP and therefore the fluorescence level is recorded [48]. Nuclear transcription factor, erythroid 2-like 2 (Nrf2) is a central conditioner of antioxidant response elements [49]. Phenolic compounds can protect HepG 2 cell against oxidative injury by promoting the Nrf2 translocation, which subsequently attenuates oxidative DNA damage, induce the expression of antioxidant enzymes, and reduce cellular ROS formulation [50][51][52]. The antioxidant activities of phenolics usually depend on the chemical construction of attached functional groups and their permutation. Previous research has mentioned that methoxyl and hydroxyl groups are directly related to the radical-inhibited ability [53]. When the same skeleton was presented, the presence of methoxyl usually brings an enhanced antioxidant activity for phenolics. However, sometimes this substitution by methoxyl can diminish the antioxidant activity [54]. The substitution pattern on the B-ring is important to the antioxidant activity of flavonoids.

Conclusions
As mentioned above, thirty-four flavonoids in A. membranaceus from different origins were identified by UHPLC-MS/MS. Eighteen identified compounds were considered to be significantly important based on OPLS-DA analysis. It characterized the samples from different origins, especially for 'Heilongjiang' and 'Inner Mongolia'. Through antioxidant activity evaluation, 'Inner Mongolia' extract presented significantly higher ORAC and CAA values (p < 0.05). It was consistent with its higher phenolics and flavonoids contents. Further investigation on in vivo antioxidant mechanisms of those flavonoids is worthy to be conducted in the future.