Chemical Discrimination of Astragalus mongholicus and Astragalus membranaceus Based on Metabolomics Using UHPLC-ESI-Q-TOF-MS/MS Approach

Astragalus mongholicus (MG) and Astragalus membranaceus (MJ), both generally known as Huangqi in China, are two perennial herbals widely used in variety diseases. However, there were still some differences in the chemical ingredients between MG and MJ. In this paper, metabolomics combined with the ultra-high performance liquid chromatography coupled with electrospray ionization/quadrupole time-of-flight mass spectrometry (UHPLC-ESI-Q-TOF-MS/MS) was employed to contrastively analyze the chemical constituents between MG and MJ. As a result, principal component analysis showed that MG and MJ were separated clearly. A total of 53 chemical markers were successfully identified for the discrimination of MG and MJ. Of them, the contents of 36 components including Astragaloside I~III, Astragaloside IV, Agroastragaloside I, etc. in MJ were significantly higher than those in MG. On the contrary, the contents of 17 other components including coumaric acid, formononetin, sophoricoside, etc. in MG were obviously higher than those in MJ. The results showed that the distinctive constituents in MG and MJ were remarkable, and MJ may own stronger pharmacological activities than MG. In a word, MG and MJ may be treated as two different herbs. This paper demonstrated that metabolomics was a vitally credible technology to rapidly screen the characteristic chemical composition of traditional Chinese medicine.


Introduction
Traditional Chinese medicine (TCM), which originated in ancient China, has played a positive role in Chinese people's health for thousands of years. Besides, with its continuation and development, TCM has been extensively applied for diseases' prevention and treatment not only in China, but also in some Western countries. Astragalus mongholicus (locally known as menggu huangqi, MG) and Astragalus membranaceus (locally known as mojia huangqi, MJ), two leguminous plants belonging to the same family of TCM named Huangqi in Chinese Pharmacopoeia, were generally used to improve immunity [1], cardiotonic [2], anti-hypertension [3,4], anti-cancer [4], anti-virus [5], and anti-diabetes [6], among others [7][8][9][10]. MG and MJ share great similarities in morphology, chemical constituent, and gene sequence [11,12]. The taxonomy controversy of MG and MJ has lasted several decades. As the discrimination of TCM played a critical role in pharmacological and clinical effects, it was urgent to MG was considered as a variation of MJ, as the main constituents of MG and MJ were similar, generally including flavonoids, saponins, polysaccharides, and so on [13]. However, there were still many differences in the detailed chemical composition between the two herbs.
In this paper, in order to solve the taxonomy uncertainty and explore the discrimination of MG and MJ, metabolomics method based on ultra-high performance liquid chromatography coupled with electrospray ionization/quadrupole time-of-flight mass spectrometry (UHPLC-ESI-Q-TOF-MS/MS) technology was employed to dramatically distinguish MG and MJ. The experimental detail procedures are shown in Figure 1, which was also provided as supplementary materials ( Figure S1). Our present study indicated that metabolomics was a powerful approach for discriminating different Chinese herbs. The obtained results not only provided chemical information for their quality assessment, but also offered rigorous evidence for further pharmacological study of MG and MJ.

Selection of Extraction Methods
Before the formal analysis of MG and MJ, extraction methods were optimized in order to gain satisfying chromatographic peaks. By testing different extraction solvents (50% methanol, 75% methanol, and 100% methanol) and diverse extraction times (0.5 h, 1 h, 1.5 h, 2 h), the method of extracting for 1 h with 75% methanol showed the large number of detectable peaks.

Optimization of UHPLC-ESI-Q-TOF-MS/MS Conditions
The employed UHPLC-ESI-Q-TOF-MS/MS was a powerful technology, which could offer integrate information of TCM. Therefore, the conditions of UHPLC-ESI-Q-TOF-MS/MS were very

Selection of Extraction Methods
Before the formal analysis of MG and MJ, extraction methods were optimized in order to gain satisfying chromatographic peaks. By testing different extraction solvents (50% methanol, 75% methanol, and 100% methanol) and diverse extraction times (0.5 h, 1 h, 1.5 h, 2 h), the method of extracting for 1 h with 75% methanol showed the large number of detectable peaks.

Optimization of UHPLC-ESI-Q-TOF-MS/MS Conditions
The employed UHPLC-ESI-Q-TOF-MS/MS was a powerful technology, which could offer integrate information of TCM. Therefore, the conditions of UHPLC-ESI-Q-TOF-MS/MS were very important for the plant metabalomics analysis. In our paper, the mobile phase was optimized and 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) showed a satisfying result. Besides, two columns including ACQUITY™ BEH C 18 column (100 mm × 2.1 mm i.d., 1.8 µm) and ACQUITY™ UPLC HSS T 3 column (100 mm × 2.1 mm i.d., 1.8 µm) were explored during the analysis, and the latter one showed much more approving data. Furthermore, in order to obtain good sensitivity and resolution, the flow rate was set at 0.4 mL/min and the injection volume was selected at 3 µL. Besides, MS conditions, including ion spray voltage; ESI heater temperature; and pressures of GS1, GS2, and CUR were optimized respectively so as to achieve better separation and stronger response for most peaks from MG and MJ. Anything else, in order to accelerate the efficiency of metabalomics analysis, a characteristic acquire mode IDA was employed to gain the MS/MS information of ions when acquiring MS information. On this basis, the top 10 ions during every 0.08 s accumulation time interval were the most meaningful ions matched IDA criteria. Besides, the dynamic background subtraction method was used to differentiate the background and matrix-related MS/MS ions intelligently. On the basis of the above conditions, the majority of base peak chromatograms (BPCs) were separated well ( Figure 2), indicating that the established method was suitable for our analysis. important for the plant metabalomics analysis. In our paper, the mobile phase was optimized and 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) showed a satisfying result. Besides, two columns including ACQUITY™ BEH C18 column (100 mm × 2.1 mm i.d., 1.8 µm) and ACQUITY™ UPLC HSS T3 column (100 mm × 2.1 mm i.d., 1.8 µm) were explored during the analysis, and the latter one showed much more approving data. Furthermore, in order to obtain good sensitivity and resolution, the flow rate was set at 0.4 mL/min and the injection volume was selected at 3 µL. Besides, MS conditions, including ion spray voltage; ESI heater temperature; and pressures of GS1, GS2, and CUR were optimized respectively so as to achieve better separation and stronger response for most peaks from MG and MJ. Anything else, in order to accelerate the efficiency of metabalomics analysis, a characteristic acquire mode IDA was employed to gain the MS/MS information of ions when acquiring MS information. On this basis, the top 10 ions during every 0.08 s accumulation time interval were the most meaningful ions matched IDA criteria. Besides, the dynamic background subtraction method was used to differentiate the background and matrixrelated MS/MS ions intelligently. On the basis of the above conditions, the majority of base peak chromatograms (BPCs) were separated well ( Figure 2), indicating that the established method was suitable for our analysis.

Characteristic Multivariate Data Analysis of Metabalomics
PCA and OPLS-DA analysis were used in the multivariate data analysis. PCA is an unsupervised pattern recognition that could effectively find out the most important information in data by reducing the original complex data. It could be used for variable reduction and separation into different classes. In this paper, the PCA model was used to identify the difference of metabolites from the MG group and MJ group. The PCA score maps were shown in Figure 2, which was also be provided as supplementary materials ( Figure S2). OPLS-DA was used to discriminate between the MG group and MJ group. As shown in Figure 3, which was also provided as supplementary materials

Characteristic Multivariate Data Analysis of Metabalomics
PCA and OPLS-DA analysis were used in the multivariate data analysis. PCA is an unsupervised pattern recognition that could effectively find out the most important information in data by reducing the original complex data. It could be used for variable reduction and separation into different classes. In this paper, the PCA model was used to identify the difference of metabolites from the MG group and MJ group. The PCA score maps were shown in Figure 2, which was also be provided as supplementary materials ( Figure S2). OPLS-DA was used to discriminate between the MG group and MJ group. As shown in Figure 3, which was also provided as supplementary materials ( Figure S3), the MG group and MJ group were clearly divided into two regions, indicating that there were significant chemical differences between the MG group and MJ group, and the established metabolomics method could characterize the chemical characteristics successfully. VIP was a common method for evaluating the contribution of variables in OPLS-DA. It was generally believed that the ions whose value ≥1 owned statistical significance and indicated the characteristics of study object. As shown in the VIP-plot in Figure 3, ions whose VIP value ≥1 were selected and regarded as the most significant differential chemical markers of the MG group and MJ group.
Molecules 2019, 24, x 4 of 12 ( Figure S3), the MG group and MJ group were clearly divided into two regions, indicating that there were significant chemical differences between the MG group and MJ group, and the established metabolomics method could characterize the chemical characteristics successfully. VIP was a common method for evaluating the contribution of variables in OPLS-DA. It was generally believed that the ions whose value ≥1 owned statistical significance and indicated the characteristics of study object. As shown in the VIP-plot in Figure 3, ions whose VIP value ≥1 were selected and regarded as the most significant differential chemical markers of the MG group and MJ group. .

Identification of Chemical Markers between MG and MJ
On the basis of the VIP results, the candidate ions between MG and MJ were identified tentatively. Take the identification of Calycosin for example. In positive mode, the ion (RT = 12.88 min and [M -H]-= 283.0688) detected in the MG group was calculated to be C16H12O5 based on the elemental composition and fractional isotope abundance. Besides, the degree of unsaturation was calculated as 14, indicating that it may be a compound with one or two benzene rings. The main MS/MS fragments ions were m/z 267, m/z 239, m/z 211, and m/z 91, meaning that the fragments may be C15H7O5 -, C14H7O4 -, C13H7O3 -, and C6H3O -, respectively. Finally, after being contrasted with the MS/MS fragments database, the ion was finally confirmed to be Calycosin. The corresponding mass spectrums and related structures were shown in Figure 4, which was also provided as supplementary materials ( Figure S4). According to the above-mentioned analysis method, a total of 53 chemical markers discriminating MG and MJ were successfully identified, including 27 candidate ions in positive mode and 26 candidate ions in negative mode. The detailed information of identified components was shown in Table 1, the related MS/MS maps were also provide as supplementary materials (peak 1-53).

Identification of Chemical Markers between MG and MJ
On the basis of the VIP results, the candidate ions between MG and MJ were identified tentatively. Take the identification of Calycosin for example. In positive mode, the ion (RT = 12.88 min and [M − H] − = 283.0688) detected in the MG group was calculated to be C 16 H 12 O 5 based on the elemental composition and fractional isotope abundance. Besides, the degree of unsaturation was calculated as 14, indicating that it may be a compound with one or two benzene rings. The main MS/MS fragments ions were m/z 267, m/z 239, m/z 211, and m/z 91, meaning that the fragments may be C 15 Figure 4, which was also provided as supplementary materials ( Figure S4). According to the above-mentioned analysis method, a total of 53 chemical markers discriminating MG and MJ were successfully identified, including 27 candidate ions in positive mode and 26 candidate ions in negative mode. The detailed information of identified components was shown in Table 1, the related MS/MS maps were also provide as supplementary materials (peak 1-53).

Relative Intensity Comparison of Chemical Markers between MG and MJ
Metabolomics was a high-throughput method to analyze the different metabolic components in plant extracts. It could clarify the primary and secondary metabolites on different levels and provide a reliable basis for the quality control of Chinese herbs. On the basis of the metabolomics analysis, heat maps were employed to show the relative intensity comparison of chemical markers between MG and MJ. Ulteriorly, a relative content comparison between MG and MJ was also conducted in order to reveal the content of chemical markers intuitively. The detailed results were shown in Figure 5, which was also be provided as supplementary materials ( Figure S5).
As shown in Figure

Relative Intensity Comparison of Chemical Markers between MG and MJ
Metabolomics was a high-throughput method to analyze the different metabolic components in plant extracts. It could clarify the primary and secondary metabolites on different levels and provide a reliable basis for the quality control of Chinese herbs. On the basis of the metabolomics analysis, heat maps were employed to show the relative intensity comparison of chemical markers between MG and MJ. Ulteriorly, a relative content comparison between MG and MJ was also conducted in order to reveal the content of chemical markers intuitively. The detailed results were shown in Figure 5, which was also be provided as supplementary materials ( Figure S5).  Traditionally, MG and MJ were used interchangeably in clinical because they were considered to share the equal efficacy. However, MG and MJ were two species of Astragalus usually planted in different areas. According to great different growing locations and climate conditions, numerous chemical wispy differences were inevitable between MG and MJ. The identified chemical components showed entirely different activities in MG and MJ. Among the chemical components, 68% of them were higher in MJ than in MG. Besides, four primary saponins constituents including Astragaloside I, Astragaloside II, Astragaloside III, and especially Astragaloside IV with anti-cancer [28], anti-oxidant [29], anti-inflammatory [30], immune regulation [31], and metabolic regulation activities [32], among others [33][34][35], in MJ were distinctly higher than that in MG. The result offered evidence that MJ may own stronger pharmacological activity and wider application than MG. Moreover, even if MG and MJ were applied in the same disease, they may be attributed to different constituents and mechanisms.

Plant Materials, Reagents and Chemicals
The radix of MG (age of 2) and MJ (age of 2) were both collected from the medical plants garden in Qiqihar Medical University and identified by doctor Jicheng Liu from Qiqihar Medical University.
Methanol and acetonitrile (HPLC grade) were bought from Merck company (Darmstadt, Germany). Formic acid (HPLC grade) was bought from Thermo Fisher Scientific (Pittsburgh, PA, USA). Distilled water was purified by a Milli-Q pure-ultrapure water system (Millipore, Bedford, MA, USA). Other reagents and chemicals were of analytical grade.

Preparation of MG Extraction and MJ Extraction for UHPLC-ESI-Q-TOF-MS/MS Analysis
A total of 1 g of dried MG and MJ powder was respectively steeped in 15-fold volume of 75% methanol in a reflux devices and extracted twice, each extraction lasting for 1 h. Then, the supernatant was obtained and filtered through a 0.22 µm membrane. Finally, a 3 µL aliquot of MG and MJ sample solution was injected for UHPLC-ESI-Q-TOF-MS/MS analysis.

Conditions of Analysis Platform
An ultra-high performance liquid chromatography system LC-30A (Shimadzu Corporation, Kyoto, Japan) coupled with a mass spectrometer Triple TOF 4600 system (AB Sciex Corporation, Redwood city, CA, USA) equipped with an electrospray ionization (ESI) source was employed in the analysis. The parameters of liquid chromatography and mass spectrometry were set as follows.
Mass spectrometry conditions: In positive mode, ionspray voltage floating (ISVF) was set at 5500 V; ESI heater temperature was maintained at 600 • C; nebulizer gas (GS 1), auxiliary gas (GS 2), and curtain gas (CUR) were set at 55 psi, 55 psi, and 30 psi, respectively; declustering potential (DP) and collision energy (CE) were set at 100 V and 10 V. The accumulation time of TOFMS was set at 0.15 s. In negative mode, ISVF was set at −4500 V; other parameters were the same as in positive mode. During the positive and negative acquire mode, MS and MS/MS data were acquired simultaneously via an information-dependent acquisition (IDA) mode. Under the IDA-MS/MS experiment, CE was set at 40 ± 20 V, and at most, 10 candidate ions were monitored per cycle. Besides, a dynamic background subtract mode, which distinctly differentiated the background and matrix-related MS/MS ions from the endogenous or exogenous components, was conducted during the analysis. Finally, a continuous calibration was proceeded every four injections during the analysis period. were selected for the negative ion mode. After peak alignment, peak matching, and peak extraction, a three-dimensional matrix including retention time, m/z, and peak area information was obtained successfully. Next, it was exported into EZinfo 3.0 software for further data analysis, after pareto scaling transformation, principal component analysis (PCA), and orthogonal partial least squares discriminant analysis (OPLS-DA) analysis were conducted to further recognize the pattern of the data. Then, variable importance in the projection plots (VIP-plots) were constructed to choose the differential variables. At last, interest variables whose VIP ≥1 were listed to the constructed OPLS-DA and the differential ions between MG and MJ were finally screened out.

Identification of Potential Chemical Markers
The above-mentioned differential ions were transferred back to Progenesis QI software and marked with tag I, meanwhile, ions whose p-value (as measured by Student's t-test) less than 0.05 were marked with tag II. Then, tag I and tag II were further subjected to matching with some databases such as SDF database created by the information of Astragalus plant in the SciFinder database and MS/MS database (Natural product database from AB Sciex, Metlin database, and National Institute of Standards and Technology database). On the basis of the characteristic fragments and the fragmentation patterns confirmed by MS/MS, the list of identified potential chemical markers, including compound's formulas, molecular weights, compound's names, Chemical Abstracts Service numbers, and mass errors, were finally obtained.

Supplementary Materials:
The following are available online. Figure S1: The experimental procedures for chemical discrimination of MG and MJ. Figure S2: The base peak chromatograms in positive mode and negative mode. Figure S3: The VIP and PCA plots between MG and MJ. Figure S4: The detail MS and MS/MS information of Calycosin. Figure