Comparison of the Saponins in Three Processed American Ginseng Products by Ultra-High Performance Liquid Chromatography-Quadrupole Orbitrap Tandem Mass Spectrometry and Multivariate Statistical Analysis

A method with ultrahigh performance liquid chromatography Quadrupole-Orbitrap tandem mass spectrometry (UHPLC-Q-Orbitrap-MS/MS) was applied for the quality evaluation of different processing and drying of American ginseng, including natural drying (ND), steam drying (SD), and vacuum freeze-drying (VFD). A total of 51 saponins were successfully identified in three processed products. Three processed American ginseng products were well-differentiated in orthogonal partial least-squares discriminant analysis (OPLS-DA). The S-plot also identified the marker compounds in each product, while the major ginsenosides of ND (malonyl (M)-Rd, M-Rb1, Rg1), SD (20 (S)-Rg3, 20 (S)-Rg2), and VFD (M-Rd, M-Rb1) were found. The results indicate that the method by vacuum freeze-drying can retain the content of rare ginsenosides and malonyl-ginsenosides. The marker compounds selected will benefit the holistic evaluation of related American ginseng products.

Ginseng has three popular processes, including natural air drying, steaming drying, and vacuum freeze-drying, respectively [10][11][12][13][14][15][16][17]. e drying process affects the quality of ginseng products and changes the content of ginsenosides. e essence of the change in ginsenoside content is the transformation of ginsenosides during the drying process. ermal processing (natural air drying and steaming drying) converts saponins of larger molecular weight into saponins of smaller molecular weight. Ginsenoside M-Rb 1 , M-Rb 2 , M-Rc, Re, Rg 1 , Rb 1 , Ginsenoside Rb 2 , Rc, Rd, Re, Rg 1 , and Ginsenoside Rd, Rk 1 , Rg 5 , Rg 3 are the major ginsenosides of white ginseng, red ginseng, and black ginseng, respectively [15,18,19]. Nonthermal processing (vacuum freeze-drying) can keep the shape and color of ginseng consistent with its fresh state, containing more natural active ingredients. e characteristic components in vacuum freeze-drying ginseng were M-Re, M-Rb 1 , M-Rc, M-Rb 1 isomer, M-Rb 2 , M-Rb 3 and M-Rd isomer [11].
e LC-MS technique combines the high separation ability for complex samples with the high selectivity of highresolution mass spectrometry and the ability to provide information (molecular weight and structural) and is widely used to control quality standards for traditional Chinese medicine [20][21][22]. e composition of Chinese herbal medicines has been rapidly analyzed by UHPLC-Q-Orbitrap-MS/MS and the changes in their chemical composition before and after processing as an effective tool for identifying active ingredients with improved sensitivity and accuracy. e fragment information of tandem mass spectrometry can be used to identify the structure of compounds [23][24][25].
Multivariate analysis methods were applied to analyze whether differences existed between test samples and to determine which compounds were altered for quality evaluation of herbal medicines of different origins, parts, and processing methods [11,23,24,26,27]. Recent studies demonstrated that the UPLC-QTOF/MS is an optimal application for holistic evaluation of ginseng [18,28]. Principal component analysis (PCA) is distinguished from the compounds of different drying processes of Houttuyniae Herba [29]. e common processed products of American ginseng are dried American ginseng [30][31][32]. It is necessary to control the drying process in the processing of American ginseng [33][34][35]. In recent years, steamed American ginseng and vacuum freeze-dried American ginseng have appeared in the functional food market. However, the systematic comparison of ginsenoside conversions of natural drying (ND), steam drying (SD), and vacuum freeze-drying (VFD) has not yet been studied.
In this paper, UHPLC-Q-Orbitrap-MS/MS analysis combined with multivariate analysis approach was applied to evaluate the composition of ND, SD, and VFD. is study aims to explore the trends of transformation of ginsenosides and characterize and quantify the chemical ingredients in three processed American ginseng products to standardize the processing procedures reasonably.

Processed American Ginseng Samples.
e samples of fresh American ginseng were first cleaned and then processed by the experiment (steaming drying, natural air drying, and vacuum freeze-drying). e ND samples were produced by drying at 50°C. e SD samples were produced in a ginseng steaming cabinet at 100°C for 3 h and allowed to dry in an oven at 50°C. e VFD samples were produced by first prefreezing at −20°C for 12 h and then placing them in a vacuum freeze-dryer (Ningbo Xinzhi Biotechnology Co., Ltd., Zhejiang, China) to freeze-dry for 72 h, as shown in Figure 1.  A fine powder (0.1 g) was ultrasonically extracted with 70% methanol-water (5 mL) for 45 min. e extraction was filtered through a 0.22 μm syringe filter before analysis [36].

Instrument and Condition.
A UPLC system (Ultimate 3000) was used for separations. A Supelco C 18 column (3.0 × 50 mm, 2.7 μm) was used at 35°C for separation and eluted by the mobile phases (solvent A and B) were acetonitrile and water containing 0.1% formic acid, separately. e gradient elution program with a 0.5 mL/min flow rate   International Journal of Analytical Chemistry 5

Analysis of ree Processed American Ginseng Products by
Orbitrap-MS/MS was applied for determination [23,37,38]. e identification of the extracts of ND, SD, and VFD samples in the negative ion mode is shown in Figure 2. e compounds were separated distinctly in 30 min by UPLC. e distinct samples of several peaks demonstrated the different intensities at 15-25 min. e ratio of Rg 1 and Re was changed with steaming treatment, and the content of Rg 1 in SD samples was significantly decreased at 10 min. Meanwhile, the content of minor ginsenosides was increased at 18-22 min. A slice of studies performed showed that the major ginsenosides (Re and Rg 1 ) were converted to minor ginsenosides (Rh 4 , Rk 1 , Rk 3 , 20 (S)-Rg 3 , Rg 5 , and 20 (R)-Rg 3 ) by heat treatment [39]. e content of malonyl-ginsenosides in VFD samples was higher than in SD samples. ese results demonstrated the change in ginsenoside composition of ND, SD, and VFD samples. e Q-Orbitrap-MS accurately measured the compounds' mass values [24]. Meanwhile, the fragmentation pattern of standards was compared to identified ginsenosides from three processed American ginseng products.  Figure 3(a), fragment ion at m/z 475 indicated the successive losses of two glucose residues. In Figure 3(b), the successive losses of four glucose residues were observed from fragment ion at m/z 459. e losses of two glucose residues and a β-d-glucuronic acid (176 Da) were observed from fragment ion at m/z 455 in Figure 3(c). In Figure 3(d), the loss of a malonyl residue was observed from malonylginsenoside at m/z 1107, and the fragment ion at m/z 459 was similarly produced by successive loss of four glucose residues. In Figure 3(e), the loss of a rhamnose residue (146 Da) was observed from fragment ion at m/z 653. In total, 51 ginsenosides were identified from ND, SD, and VFD by comparing standards and literature records of the tandem MS spectra (Table 1).

Multivariate Statistical Analysis.
e statistical methods were applied to display the differences in ginsenosides intuitively. After data preprocessing of the ND, SD, and VFD samples, the dataset was conducted to discover the marker compound by multivariate statistical analysis. e distinct ginsenoside composition of samples from ND, SD, and VFD was characterized in detail. e OPLS-DA model was effectively used to observe three processed American ginseng products (Figure 4). An excellent prediction ability was considered to the parameters of SD vs. ND (R2 Y � 0.865, Q2 � 0.999), VFD vs. ND (R2 Y � 0.822, Q2 � 0.998), and VFD vs. SD (R2 Y � 0.848, Q2 � 1). e score plot was used to discriminate between the groups of two selected samples (ND vs. SD, ND vs. VFD, and SD vs. VFD). e samples of ND were compared to the other samples by OPLS-DA, separately. By comparing the S-plot of ND and SD, the components observed from the lower left quadrant and the upper right quadrant were elevated in SD and ND, respectively (Figure 4(b)). e marker components were elevated in SD (20(S)-Rg3, 20(S)-Rg2, and an unknown compound) and ND (malonyl (M)-Rb1, M-Rd). Four components (20(S)-Rg3, 20(S)-Rg2, M-Rb1, and M-Rd) were identified as marker compounds in SD and ND. By comparing the S-plot of ND and VFD, it was performed that four unknown compounds and two compounds (M-Rb1, Rg1) were elevated in VFD and ND, respectively (Figure 4(d)). e ginsenosides of M-Rb1 and Rg1 were identified as marker compounds in ND and VFD. By comparing SD and VFD, it was performed that two malonyl-ginsenosides (M-Rd, M-Rb1) and three compounds (20(S)-Rg3, 20(S)-Rg2, and an unknown compound) were elevated in VFD and SD, respectively (Figure 4(f)). Four ginsenosides of M-Rd, M-Rb1, 20(S)-Rg3, and 20(S)-Rg2 were identified as marker compounds in VFD and SD. e repeated emergence of marker compounds was in different groups, including the SD group (20(S)-Rg3 and 20(S)-Rg2) and the ND group (M-Rb1). Meanwhile, the marker compounds could display various pharmacological activities such as antitumor (20(S)-Rg3) [40], against cardiocerebrovascular diseases (20(S)-Rg2) and affect the central nervous system (M-Rb1) [41,42]. International Journal of Analytical Chemistry e VIP value was used to select the marker compounds [29]. e features with VIP values larger than 1 were highlighted as marker candidates. e compounds with the statistically significant difference (p < 0.05) were screened for marker compounds by ANOVA. To visualize the tendency of marker compounds, the intensities of 22 marker compounds were constructed as a heat map. e composition change was shown in different colors by the ND group, the SD group, and the VFD group ( Figure 5). Between the SD and ND groups, the contents of ginsenoside-Rb 1 , -Rh 1 , -Rg 3 , -Rg 2 , -Rg 4 , -Rh 4 , -F 2 , and -Rg 5 in SD samples were increased, and the contents of malonyl-ginsenoside-Rg 1 , -Re, -Rc, -Rb 1 , -Rb 2 , -Rd, and ginsenoside-Re, -Rc, -Rg 1 , -Rb 3 , -Rd, -Ro, 20-(S)-PF 11 , and 20-(R)-PF 11 in ND samples were significantly higher. Comparing the groups of ND and VFD, the contents of malonyl-ginsenoside-Rg 1 , -Re, -Rc, -Rb 1 , -Rb 2 , and -Rd in the VFD group were significantly increased. Meanwhile, the contents of ginsenoside-Re, -Rc, -Rg 1 , -Rb 3 , -Ro, -Rd, 20-(S)-PF 11 , and 20-(R)-PF 11 in the ND samples were significantly increased. e malonyl residue of malonyl-ginsenoside in ND and VFD was well preserved without heating treatment. Similar to ginseng, it was observed that the content of malonyl-ginsenosides in American ginseng was increased by a vacuum freeze-drying approach [11]. e thermal sensitivities of malonyl-ginsenosides were less affected by the vacuum freeze-drying technology.
e malonyl-ginsenoside of chemical transformation is related to heat and pressure treatment [24]. Meanwhile, the alteration of three processed American ginseng products is possibly related to the degree of enzymatic activity, moisture, and microstructure [43]. e content of moisture can result in a rapid transformation by promoting corresponding microbiological growth, enzymatic and nonenzymatic reactions in American ginseng. e active compounds are cleaved by enzymes with steam and high temperatures [9,44]. As a result, the content of ginsenoside Rb 1 in SD samples was higher than in VFD and ND samples. e content of ginsenoside Rg 1 in ND and VFD samples was higher than in SD samples; it probably converted into 20(S)-Rg 3 , 20(R)-Rg 3 , Rk 1 , and Rg 5 by heat treatment [39]. Malonyl-ginsenosides in VFD and ND samples were higher than in SD samples, and they were probably converted into malonic acid through decarboxylation by heat treatment. In summary, these factors all lead to the transformation of ginsenosides in the three processed products. e composition and content of components in various processed products can affect their pharmacological activity, and it is necessary to elucidate the change of ginsenosides.

Quantification of Ginsenosides from
ree Processed American Ginseng Products. e 12 compounds' EIC was identified by the reference compounds in Figure S1 and Table S1 (Supplementary material). e standard curve was used to determine the actual amount of the three processed American ginseng products in Figure S2 and Table S2 (Supplementary material). e quantification was based on 12 ginsenosides for products (n � 3) and the amount of ginsenosides was calculated in Table S3 (Supplementary material).
Previously, Huang et al. has reported the 59 ginsenosides of protopanaxadiol, the concentrations of ginsenosides Rk 1 , Rg 5 , Rh 1 , 20(R)-ginsenoside Rg 2 , and 20(R)-ginsenoside Rg 3 are highest in red American Ginseng, these results have been matched to ours [10]. For ginsenosides analysis, the content of ginsenosides Rg 1 , Re, Rb 1 , Rc, Rd, and 24(R)-pseudoginsenoside F 11 is the highest in ND. As our results, the concentrations of ginsenosides Rg 1 , Re, Rb 1 , Rc, Rd, and 24(R)-pseudo-ginsenoside F 11 are highest in ND, and this is matched to another paper [18]. Ginsenoside Rh 1 is not detected in ND and VFD.

Conclusions
A successful method was performed for the chemical components of three processed American ginseng products using UHPLC-Q-Orbitrap-MS/MS. e 5 ginsenosides as characteristic marker compounds could be applied to elucidate the composition of ND, SD, and VFD samples by multivariate statistical analysis. e results will be useful to visualize the tendency of marker compounds in manufacturing. Furthermore, this study effectively provided a means for assessing and controlling different processed American ginseng products.

Data Availability
e data used to support findings of this study are included within the article.

Conflicts of Interest
e authors declare no conflicts of interest.

Authors' Contributions
Yunlong Guo, Guangzhi Cai, and Xiaokang Liu conceived the experiments. Jiyu Gong, Xin Huang, and Shuying Liu supervised the study and provided financial support via projects. Yuxin Bai and Na Guo wrote the paper.