Next Article in Journal
Sanguinarine Induces Apoptosis in Papillary Thyroid Cancer Cells via Generation of Reactive Oxygen Species
Previous Article in Journal
Effect of Initial Conformation on the Starch Biopolymer Film Formation Studied by NMR
Open AccessArticle

Metabolomics Combined with Multivariate Statistical Analysis for Screening of Chemical Markers between Gentiana scabra and Gentiana rigescens

by Gaole Zhang 1,2,†, Yun Li 2,3,†, Wenlong Wei 2, Jiayuan Li 2, Haoju Li 2, Yong Huang 2 and De-an Guo 1,2,*
1
School of Pharmaceutical Sciences, Changchun University of Chinese Medicine, Changchun 130117, China
2
Shanghai Research Center for Modernization of Traditional Chinese Medicine, National Engineering Laboratory for TCM Standardization Technology, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Haike Road 501, Shanghai 201203, China
3
Department of Traditional Chinese Medicine, China Pharmaceutical University, Tongjiaxiang 24, Nanjing 210009, China
*
Author to whom correspondence should be addressed.
These authors contributed equally to this work.
Molecules 2020, 25(5), 1228; https://doi.org/10.3390/molecules25051228
Received: 4 February 2020 / Revised: 24 February 2020 / Accepted: 28 February 2020 / Published: 9 March 2020
(This article belongs to the Section Analytical Chemistry)

Abstract

Gentianae Radix et Rhizome (Longdan in Chinese, GRR) in Chinese Pharmacopoeia is derived from the dried roots and rhizomes of Gentiana scabra and G. rigescens, that have long been used for heat-clearing and damp-drying in the medicinal history of China. However, the characterization of the chemical components of two species and the screening of chemical markers still remain unsolved. In current research, the identification and characterization of chemical components of two species was performed using ultra-high-performance liquid chromatography (UHPLC) coupled with linear ion trap-Orbitrap (LTQ-Orbitrap) mass spectrometry. Subsequently, the chemical markers of two species were screened based on metabolomics and multivariate statistical analysis. In total, 87 chemical constituents were characterized in G. scabra (65 chemical constituents) and G. rigescens (51 chemical constituents), with 29 common chemical constituents being discovered. Thereafter, 11 differential characteristic components which could differentiate the two species were designated with orthogonal partial least squares discriminant analysis (OPLS-DA) and random forest (RF) iterative modeling. Finally, seven characteristic components identified as (+)-syringaresinol, lutonarin, trifloroside, 4-O-β-d-glu-trifloroside, 4″-O-β-d-glucopyranosy1-6′-O-(4-O-β-d-glucaffeoyl)-linearroside, macrophylloside a and scabraside were selected as the chemical markers for the recognition of two Gentiana species. It was implied that the results could distinguish the GRR derived from different botanical sources, and also be beneficial in the rational clinical use of GRR.
Keywords: Gentianae Radix et Rhizome; Gentiana scabra; Gentiana rigescens; metabolomics; UPLC/LTQ-Orbitrap-MS; chemical markers Gentianae Radix et Rhizome; Gentiana scabra; Gentiana rigescens; metabolomics; UPLC/LTQ-Orbitrap-MS; chemical markers

1. Introduction

Gentianae Radix et Rhizoma (GRR), named Longdan in Chinese, is a widely used as traditional Chinese medicine (TCM) for the treatment of icterepatitis, dermatophytes and herpes zoster [1]. As first recorded in Shennong Bencao Jing (Shennong’s Classic of Materia Medica) in Han Dynasty, GRR has been used as hepatoprotective and choleretic drug for thousands of years. It has also been described in various Chinese ancient medicinal monographs, such as Tujing Bencao in Song Dynasty and Bencao Gangmu in Ming Dynasty. It was mainly used for clearing liver and gallbladder dampness and heat, purging fire of the liver and gallbladder, relaxing tendons and relieving pain, etc. [2,3,4]. Modern pharmacological studies reported that it possessed anti-inflammatory, anti-oxidative and antiviral activities [5,6,7,8]. According to phytochemical research, the main constituents of GRR are iridoids, xanthones, flavonoids, triterpenoids, and others [4,9,10,11]. Arrays of studies demonstrated that iridoids and xanthones were the active components and had various remarkable pharmacological activities including hepatoprotective [12], anti-inflammatory [13,14,15], antioxidant [8,16,17] and immunomodulatory activities [17]. However, investigations into the chemical profiling of GRR and screening of chemical markers of GRR derived from the different sources remain inadequate.
It is well known that morphologic and microscopic identifications, high-performance thin-layer chromatography (HPTLC) technique, DNA barcodes and metabolomics could all be used for the interspecies identification [18]. However, the there is a high possibility of false identification by applying morphologic and microscopic techniques, due to the subjective judgement by the experimenters. Although the TLC and DNA barcoding techniques were more objective and accurate compared to other approaches, the results lacked integrated adequate chemical information, and could not screen chemical markers to identify and classify different species in an efficient and swift manner. Metabolomics revealed the global chemical profiling of TCM and the raw data could be further analyzed with multivariate statistical analysis for the screening of chemical markers among different species or sources [19].
Nowadays, ultra-high-performance liquid chromatography (UHPLC) coupled with high-resolution mass spectrometry (HRMS) has been widely applied for the chemical profiling of complex mixtures from natural products [20,21]. LC-HRMS could offer the potential chemical composition and infer the detailed structures of analytes on the basis of the measurement of accurate mass and generation of the MS/MS or MSn data [22,23,24]. Moreover, it could determine hundreds or even thousands of MS features using a single injection in a short time and with less consumption of organic solvents, and implement high-throughput data acquisition [25].
In this study, an untargeted metabolomics combined with univariate analysis, orthogonal partial least-squares-discriminant analysis (OPLS-DA), and random forest (RF) for the screening of chemical markers which could be applied to accurately distinguish G. scabra and G. rigescens. Furthermore, the characterization of chemical components of GRR was performed by UHPLC-LTQ-Orbitrap/MS. The common and specific chemical constituents of G. scabra and G. rigescens were compared and classified accordingly.

2. Result and Discussion

2.1. Optimization of Chromatographic Conditions and Sample Extraction

In order to obtain good separation, the different chromatographic columns, gradient elution program, column temperature and flow rate were optimized in detail. The six different columns including Eclipse Plus C18, Extent C18, BEH C18, Kinetex C18, Zorbax SB C18 and HSS T3 were used to compare the separation capacity, the detailed column information was listed in Table S1. As a result, the HSS T3 column showed a better retention capacity and column efficiency for the separation than the other columns. Subsequently, the flow rate (0.3, 0.4, and 0.5 mL/min) and column temperature (25, 30, 35, and 40 ℃) were compared under the optimized gradient elution program, and 0.4 mL/min at 35 ℃ was selected for separation (Figure 1). The injection volume was 2 μL with no solvent effect. In addition, QC samples of GRR were collected in both positive and negative ion modes, and more useful information was prevailed in negative mode (Figure 2).
Different extract solvents (methanol, ethanol and acetonitrile) were compared according to the response intensity of four standards (longanic acid, gentiopicroside, sweroside and trifloroside), and the result of the extraction of methanol solvent was better than others. The solvent ratio was evaluated as above, and the extraction efficiency of 50% methanol–water (v/v) was optimal (Figure 3).

2.2. Characterization of Chemical Components by UPLC-LTQ-Orbitrap/MS

2.2.1. Deductive Fragmentation Pathway of Iridoids

Iridoids belong to the highly oxygenated monoterpene; its parent nucleus was a five-carbon cyclopenta pyranoid skeletal structure. In the simple iridoids, C-1 semi-acetal hydroxyl was active, and always linked with β-d-glucose to form glycosides, mostly mono-glycosides.
Some iridoids, known as the seco-iridoids, such as 7,8-secoderivative, were formed by the cleavage of cyclopentane ring at the bond of C-7 and C-8. In the seco-iridoids, it could be calssified into four categories: gentiopicrins, swerosides, swertiamarinss and dicarboxylic acids. The structure of gentiopicrin was characterized by the formation of double bonds between C-3 and C-4 and C5 and C-6 positions, based on the seco-iridoid parent nucleus. To differentiate from gentiopicrin, sweroside glycoside’s skeletal structure only has one double bond in C-3 and C-4. Swertiamarin glycoside has the hydroxyl substituents at C-5 on the basis of the sweroside parent nucleus structure. Both sweroside and swertiamarin had a glycosyl–acetylation structure. The structure of rindoside, one compound of swertiamarin group, was taken as an example for summarizing the cleavage pattern: de-saccharification occurs first and then deacetyl group as shown in Figure 4 [26].

2.2.2. Deductive Fragmentation Pathway of Flavonoids

Due to the stable conjugated structure of flavonoids, cleavage often occurred on the sugar moiety attached to the A ring, such as homoorientin shown in Figure 5. The homoorientin generated its [M-H] ion at m/z 447.15; it produced [M-H-H2O] ion at m/z 430.09, or generated a series of ion at m/z 358.06 [M-H-C3H6O3], m/z 328.05[M-H-C3H6O3-OH]and m/z 299.06 [M-H-C3H6O3-OH-CH2O]. The proposed fragmentation pathway was similar with deductive homoorientin, therefore, the other 13 flavonoid compounds were tentatively characterized [27].

2.2.3. Deductive Fragmentation Pathway of Xanthones

The fragment ion formation process of mangiferin and mechanism of mass spectrometry fragmentation are shown in Figure 6. Mangiferin displayed [M-H] ion at m/z 421.08, then generated two kinds of ion: m/z 259.02 [M-H-Glu] or m/z 332.05 [M-H-C3H6O3], and the major ion m/z 332.05 [M-H-C3H6O3] successively produced m/z 302.04 [M-H-C3H6O3-CH2O] and 272.03 [M-H-C3H6O3-CH2O]-[28].
According to the cleavage regularity and detected chemical data information, in total, 87 compounds, including 54 iridoids, 13 flavonoids, two xanthone, four triterpenoids and five other components were characterized in G. scabra (65 chemical constituents) and Gentiana rigescens (51 chemical constituents), among which 29 common chemical constituents were shown in Table 1 and Figure 7.

2.3. Screening of Chemical Markers by Metabolomics and Chemometrics

2.3.1. Univariate Analysis for the Screening of Differential Metabolites

Compared with multivariate statistics, univariate analysis focuses more on independent changes in the levels of metabolites. After the data filtering with 80% and 15% rules, the 366 stable metabolic characteristics were obtained. Student’s t-test was applied to figure out the features which had statistical difference (p-value < 0.05) between the two groups. The results showed that 283 features with p-value < 0.05 were screened out, and the heat map (Figure 8) illustrated the differences in 283 features between the two Gentiana species. It can be seen clearly that the 283 screened features accumulated in the two Gentiana species show great differences; about half of the metabolites were upregulated in the G. scabra group compared with their counterparts in the G. rigescens group.

2.3.2. Data Visualization and Experimental Stability Evaluation

Principal component analysis (PCA) was carried out to provide an unbiased visual representation of the sample distributions in the extracted principal components (PCs) space. As a result, after the Pareto scaling, the first seven PCs of the PCA-X model explained 90.6% of the variation in the original dataset (R2X(cum) = 0.906); 74.4% of the variation in original data was predicted by the model according to 7-fold cross validation (Q2(cum) = 0.744), which means that this PCA-X model could well represent the variation information of the original data (Figure 9a). The sample distribution of two Gentiana species and QC samples in the score plot (PC1 vs. PC2) was provided in Figure 9b. The sample distribution of QC samples was closely clustered, showing that the data collection was relatively stable during the entire experiment. These test samples from different species are well separated, indicating that the differences in metabolic characteristics between two Gentiana species were obvious. Therefore, for the next step, the decision combination of OPLS-DA and RF were further used in search of the chemical markers which could be used to successfully distinguish two Gentiana species.

2.3.3. Screening of Differential Metabolic Characteristics

A supervised OPLS-DA approach was used to preliminarily investigate the metabolite characteristics that showed the greatest differences between the two Gentiana species. As a result, after the Pareto scaling, the OPLS-DA model described 86.5% of the variation in X (R2X (cum) = 0.865); 98.9% in response Y (R2Y(cum) = 0.989) and 97.5% in response Y was predicted by the model according to 7-fold cross validation (Q2(cum) = 0.975), with one predictive and six orthogonal (1 + 6) components. The high value of those parameters demonstrated that the OPLS-DA model presented a good classification and prediction ability to distinguish two class. In Figure 10a, the sample distribution in the score plot (predictive component 1 vs. X side orthogonal component 1) of two Gentiana species were well separated, and the interclass samples heterogeneity of G. scabra were larger than G. rigescens, which indicated that the chemical variation in different batches of G. scabra was larger than that of G. rigescens. Subsequently, permutation tests (n = 200) were performed to validate the model performance. As shown in Figure 10b, the values of R2 = (0.0, 0.394) and Q2 = (0.0, −0.664) of category 1 and R2 = (0.0, 0.385) and Q2 = (0.0, −0.672) of category 2 indicated the OPLS-DA model in the present study having no risk of overfitting. In total, 47 variables (VIP > 1, Figure 10c) in the OPLS-DA model were selected as the differential metabolic characteristics, since these variables have an important identification capability, which is also consistent with the result shown in S-plot (Figure 10d). These differential metabolic characteristics were helpful to clarify the chemical differences in two Gentiana species.
Although 47 differential metabolic characteristics selected by OPLS-DA could successfully identify two Gentiana species, many metabolic characteristics were not easy to detect in practice. Therefore, these differential metabolic characteristics need to be further refined to select a set of robust features labeled as chemical markers after characterization. Furthermore, the model performance has been greatly affected by the data scaling method [29], which may interfere with the correct selection of important variables in the metabolomic research. By contrast, RF was not very sensitive to the choice of data scaling method [30], and the RF model was further adopted, with 47 differential metabolic characteristics. The parameters of RF were set for 500 trees, and the mtry was the default value (square root of the number of variables). RF was modeled through 100 iterations; the variable importance of each modeling process was ranked, then the ranked variable ID was stored for further analysis.
In Figure 11a, according to the sorted table of the variable importance of 100 RF model iterations, the cumulative number of each variable in top Nth was obtained. Each line represented a metabolic characteristic; it could be clearly seen that the variables were divided into two parts. The cumulative number of 36 variables in the left side reached 100 around top 35th, while the remaining 11 variables were situated in the right side due to their relatively low feature importance. Then, 36 variables which showed a relatively larger feature importance in 100 iteration modeling were selected. Subsequently, the frequency of the variable as the most important parameter in the 100 iterative modeling was calculated. In Figure 11b, the variable with the highest frequency appeared nine times, indicating that this metabolic characteristic was of major importance in differentiating between the two Gentiana species. Furthermore, one variable appeared six times, four variables five times and five variables four times, and showed great discriminating power. In the results, 11 metabolic characteristics that appeared more than four times were selected. Finally, based on 11 selected metabolic characteristics, RF model (500 trees) was established to distinguish the two Gentiana species. Results showed that the out-of-bag estimate of error rate of RF model is 0, and the area under the curve (AUC) of receiver operating characteristic (ROC) curve of the RF model was 1.00, indicating that the two Gentiana species could be correctly identified using 11 selected metabolic characteristics.

2.3.4. Identification of Chemical Markers

The 11 characteristic markers were putatively identified according to the retention time, accurate mass, MS2 and fragmentation pattern of standards. Finally, the seven markers were identified as (+)syringaresinol, lutonarin, trifloroside,4-β-d-glu-trifloroside, 4″-O-β-d-glucopyranosy1-6′-O-(4-O-β-d-glucaffeoyl) linearroside, macrophylloside A and scabraside respectively, and four markers remained unidentified (Table 2).

3. Experimental

3.1. Materials and Reagents

The standard compounds, gentiopicroside, swertiamarin, sweroside, loganin, homoorientin, mangiferin, were purchased from Shanghai Nature-Standard Biotech Co., Ltd (Shanghai, China); 6’-O-β-d-gentiopicroside from Shanghai Yuanye Bio-Technology Co., Ltd (Shanghai, China); secologanoside from Chengdu Priifa Technology Development Co. Ltd (Chengdu, China); loganic acid, trifloroside, rindoside and 6’-O-β-d-glu-loganic acid from Chengdu DeSiTe Biological Technology Co., Ltd (Chengdu, China).
HPLC-grade acetonitrile and formic acid used in the mobile phase were purchased from Merck KGaA (Darmstadt, Germany) and Tedia Company Inc. (Fairfield., OH, USA), respectively. Ultrapure water (18.2 MΩ cm at 25 °C) was in-lab prepared by a Millipore Alpha-Q water purification system (Millipore, Bedford, MA, USA). HPLC-grade methanol for the sample preparation was purchased from Sino Pharm Chemical Reagent Co., Ltd. (Shanghai, China).

3.2. Sample and Standards Preparation

The roots and rhizome of GS and GR were collected from Liaoning Province in September and Yunnan Province in China in November, 2019. Fifty-four batches of GS samples were collected from six areas of Qingyuan, particularly in Liaoning Province (the largest base for the cultivation of GS in China.) Thirty-four batches of GR samples were collected from four areas of Yunxian in Yunnan province. The collection information was summarized in Table 3. These samples were all recorded according to their resources. All of them were identified by Professor Jinglong Zhang, who is the expert in medicinal botany in Changchun University of Chinese Medicine. The plant specimens were stored in Shanghai Research Center for Modernization of Traditional Chinese Medicine, National Engineering Laboratory for TCM Standardization Technology, Shanghai Institute of Materia Medica, Chinese Academy of Sciences.
An aliquot of 0.250 g of accurately weighted fine powder of GRR was initially immersed in 10 mL 50% aqueous methanol (v/v) and extracted on a water bath at 40 °C with ultrasound (1130 W, 37 kHz) assistance for 30 min. The supernatant was transferred into a 2 mL centrifuge tube. After being centrifuged at 14,000 rpm in 10 min, the supernatant was filtered through a 0.22 μm PTFE microporous membrane (Agilent Technologies, Santa Clara, CA, USA) to prepare the test solutions. The standard solutions of gentiopicroside, swertiamarin, sweroside, loganin, homoorientin, mangiferin, 6’-O-β-d-gentiopicroside, secologanoside, loganic acid, trifloroside, rindoside and 6’-O-β-d-glu-loganic acid were prepared in methanol at the appropriate concentration.

3.3. Instrumentation and Condition

Chromatographic separation was executed on an Ultimate 3000 UHPLC system (Thermo Fisher Scientific, San Jose, CA, USA). The separation was performed on a Waters (Milford, MA, USA) HSS T3 column (2.1×100 mm, 1.7 μm) and maintained at 30 °C. The GRR samples were eluted with a mobile phase consisting of 0.1% formic acid water (A) and acetonitrile (B) in the following gradient program: 0–4 min, 10–20% B; 4–12 min, 12–27% B; 12–30 min, 27–53% B; 30–31 min, 53–100% B; 31–25 min, 100% B. The flow rate was set at 0.4 mL/min. The injection volume was 2 μL.
An LTQ-Orbitrap Velos Pro hybrid mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA) was used for accurate mass measurements and data collection in negative mode, and with ESI-source-operated. The ESI source parameters were set as follows: ion spray voltage 2.7 kV, capillary temperature 320 °C, source heater temperature 200 °C, sheath gas (N2) 15 arbitrary units, auxiliary gas (N2) eight arbitrary units, and sweep gas (N2) two arbitrary units. The Orbitrap analyzer scanned the mass range from m/z 50 to 1345 with a resolution of 30,000 (FWHM defined at m/z 400) for MS. The MS data were recorded in profile and centroid formats, respectively. The average acquisition time required to finish a scan circle (containing four scan events) was 1.8 s. Default values were used for most other acquisition parameters.

3.4. Data Processing and Multivariate Analysis

Progenesis QI (Waters, Milford, USA) was used to process the raw data acquired from the UPLC-LTQ-Orbitrap/MS. The feature detection, precursor ions fusion, and retention time correction operated by Progenesis QI to get a peak table (5528) including retention time, m/z, and normalized abundance of all samples. Regarding the data filtering, firstly, features with a relative standard deviation (RSD) greater than 15% in QC samples were removed because these features were unstable during the data collection of the entire experiment. Secondly, the remaining features were filtered according to the 80% rule, whereas features present in at least 80% of samples in one group were allowed to remain.
In data analysis, univariate and multivariate statistical methods were both introduced to contribute the complementary advantages [31]. Firstly, univariate data analysis, i.e., Student’s t-test, was used to figure out the features which had statistical difference (the p-value of a Student’s t-test of < 0.05) between the two groups. Secondly, a visual representation of the sample distributions for two Gentiana species and QC samples were provided by principal component analysis (PCA). Thirdly, OPLS-DA analysis was carried out to distinguish two Gentiana species, and the metabolic characteristics with VIP > 1 were chosen as differential metabolic characteristics. Finally, the RF analysis was iteratively modeled 100 times to deeply investigate the permutation accuracy importance of these differential metabolic characteristics, and the chemical markers were selected to differentiate two Gentiana species.

3.5. Software

The raw data of UPLC-LTQ-Orbitrap-MS were processed by Progenesis QI. Univariate analysis of Student’s t-test was calculated by Excel, heat map was formed by heatmap package (version 4.6-14) in an R environment (version 3.4.3). Multivariate analysis of PCA and OPLS-DA was processed by software SIMCA-P+ 13.0 (Umetrics AB, Umeå, Sweden). An RF model was established by random Forest package (version 4.6-14) in R environment (version 3.4.3).

4. Conclusions

A systematic chemical characterization method was developed to provide an effective and scientific basis for the quality control of GRR, as well as to carry out an integrated platform based on plant metabolomics and chemometrics for the characterization and classification of two officinal Gentiana species. In the present study, 87 components were identified in GRR by UHPLC-LTQ-Orbitrap/MS, including 54 iridoids, 13 flavonoids, two xanthone, and four triterpenoids. Then, after the data visualization and classification analysis of whole metabolite profiles operated by PCA and OPLS-DA, two Gentiana species of GRR were clearly separated and correctly identified, and 47 differential metabolic characteristics were selected. Subsequently, 11 differential features were further selected based on the 100 RF iterative modeling. After matching with authentic compounds, seven differential components were selected as chemical markers for the recognition of two Gentiana species. The results indicated that metabolomics combined with chemometrics is a powerful tool to differentiate closely related species and could be used as essential data for quality control of traditional Chinese medicines.

Supplementary Materials

The following are available online. Table S1: The columns screened for separation of GRR.

Author Contributions

Conceptualization, G.Z. and Y.L.; Data curation, G.Z., Y.L. and H.L.; Formal analysis, G.Z. and L.Y.; Funding acquisition, G.Z. and Y.L.; Methodology, W.W.; Project administration, D.-a.G.; Software, Y.H.; Supervision, D.-a.G.; Validation, J.L.; Writing—original draft, G.Z. and L.Y.; Writing—review and editing, W.W. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (No. 81530095, 81673591), Strategic Priority Research Program of the Chinese Academy of Sciences (No. XDA12020348). National Standardization of Traditional Chinese Medicines Project (ZYBZH-K-LN-01). Science and Technology Commission Foundation of Shanghai (No.15DZ0502800).

Conflicts of Interest

The authors declare no conflict of interest.

Abbreviation

Traditional Chinese Medicine, TCM; Gentiana scabra, GS; Gentiana rigescens, GR; Gentianae Radix et Rhizome (Longdan), GRR; Orthogonal Partial Least Squares Discriminant Analysis, OPLS-DA; Random Forest, RF; Principal Component Analysis, PCA; Variable Importance for the Projection, VIP; Quality Control, QC: Linear Ion Trap-Orbitrap Mass Spectrometer, LTQ-Orbitrap MS; Ultra-High-Performance Liquid Chromatography, UHPLC; High Resolution Mass Spectrometry, HRMS; Base Peak Ion, BPI; Principal Components, PCs; Electron Spray Ionization, ESI; Relative Standard Deviation, RSD; High Performance Thin Layer Chromatography, HPTLC; Thin Layer Chromatography, TLC;

References

  1. Wang, Y.M.; Xu, M.; Wang, D.; Zhu, H.T.; Yang, C.R.; Zhang, Y.J. Review on “Long-Dan”, one of the traditional Chinese medicinal herbs recorded in Chinese pharmacopoeia. Nat. Prod. Bioprod. 2012, 2, 1–10. [Google Scholar] [CrossRef]
  2. State Pharmacopoeia Commission of the PRC. Pharamacopoeia of the People’s Republic of China; People’s Medical Publishing House: Beijing, China, 2015.
  3. Lian, L.H.; Wu, Y.L.; Bai, T.; Jiang, S.; Yang, Y.; Nan, J.X. Hepatoprotective effects of Gentiana manshurica Kitagawa. Acta Pharmacol. Sin. 2013, 34, 129. [Google Scholar]
  4. Yang, J.L.; Liu, L.L.; Shi, Y.P. Phytochemicals and biological activities of Gentiana species. Nat. Prod. Commun. 2010, 5, 1934578X1000500432. [Google Scholar] [CrossRef]
  5. Suh, H.W.; Lee, K.B.; Kim, K.S.; Yang, H.J.; Choi, E.K.; Shin, M.H.; Park, Y.S.; Na, Y.C.; Ahn, K.S.; Jang, Y.P.; et al. A bitter herbal medicine Gentiana scabra root extract stimulates glucagon-like peptide-1 secretion and regulates blood glucose in db/db mouse. J. Ethnopharmacol. 2015, 172, 219–226. [Google Scholar] [CrossRef] [PubMed]
  6. Yang, B.; Kim, S.; Kim, J.H.; Lim, C.; Kim, H.; Cho, S. Gentiana scabra Bge roots alleviates skin lesions of contact dermatitis in mice. J. Ethnopharmacol. 2019, 233, 141–147. [Google Scholar] [CrossRef]
  7. Wang, K.; Coyle, M.E.; Mansu, S.; Zhang, A.L.; Xue, C.C. Gentiana scabra Bge formula for herpes zoster: Biological actions of key herbs and systematic review of efficacy and safety. Phytother. Res. 2017, 31, 375–386. [Google Scholar] [CrossRef]
  8. Wang, Z.Y.; Wang, C.Y.; Su, T.T.; Zhang, J. Antioxidant and immunological activities of polysaccharides from Gentiana scabra Bge roots. Carbohydr. Polym. 2014, 112, 114–118. [Google Scholar] [CrossRef]
  9. Singh, A. Phytochemicals of Gentianaceae a review of pharmacological properties. Inter. J. Pharm. Sci. Nanotech. 2008, 1, 33–36. [Google Scholar]
  10. Pan, Y.; Zhao, Y.L.; Zhang, J.; Li, W.Y.; Wang, Y.Z. Phytochemistry and pharmacological activities of the genus Gentiana (Gentianaceae): Review. Phytother. Res. PTR 2016, 13, 107–150. [Google Scholar]
  11. Mirzaee, F.; Hosseini, A.; Jouybari, H.B.; Davoodi, A.; Azadbakht, M. Medicinal, biological and phytochemical properties of Gentiana species. J. Trad. Comple. Med. 2017, 7, 400–408. [Google Scholar] [CrossRef] [PubMed]
  12. Wang, A.Y.; Lian, L.H.; Jiang, Y.Z.; Wu, Y.L.; Nan, J.X. Gentiana manshurica Kitagawa prevents acetaminophen-induced acute hepatic injury in mice via inhibiting JNK/ERK MAPK pathway. World J. Gastroenter. 2010, 16, 384–391. [Google Scholar] [CrossRef] [PubMed]
  13. Li, W.; Zhou, W.; Kim, S.; Koo, J.E.; Kim, Y.; Koh, Y.S.; Shim, S.H.; Ma, J.Y.; Kim, Y.H. Three new secoiridoid glycosides from the rhizomes and roots of Gentiana scabra and their anti-inflammatory activities. Nat. Prod. Res. 2015, 29, 1920–1927. [Google Scholar] [CrossRef] [PubMed]
  14. He, Y.M.; Zhu, S.; Ge, Y.W.; Kazuma, K.; Zou, K.; Cai, S.Q.; Komatsu, K. The anti-inflammatory secoiridoid glycosides from Gentianae Scabrae Radix: The root and rhizome of Gentiana scabra. J. Nat. Med.-Tokyo 2015, 69, 303–312. [Google Scholar] [CrossRef] [PubMed]
  15. Choi, R.-Y.; Nam, S.-J.; Lee, H.-I.; Lee, J.; Leutou, A.S.; Ri Ham, J.; Lee, M.-K. Gentiopicroside isolated from Gentiana scabra Bge inhibits adipogenesis in 3T3-L1 cells and reduces body weight in diet-induced obese mice. Bioorg. Med. Chem. Let. 2019, 29, 1699–1704. [Google Scholar] [CrossRef]
  16. Cheng, Z.; Zhang, Y.; Song, H.; Zhou, H.; Zhong, F.; Hu, H.; Feng, Y. Extraction optimization, characterization and antioxidant activity of polysaccharide from Gentiana scabra Bge. Inter. J. Biol. Macromol. 2016, 93, 369–380. [Google Scholar] [CrossRef] [PubMed]
  17. Wang, C.; Wang, Y.; Zhang, J.; Wang, Z. Optimization for the extraction of polysaccharides from Gentiana scabra Bge and their antioxidant in vitro and anti-tumor activity in vivo. J. Taiwan Inst. Chem. Eng. 2014, 45, 1126–1132. [Google Scholar] [CrossRef]
  18. Xu, Y.; Li, Y.; Maffucci, K.G.; Huang, L.; Zeng, R. Analytical methods of phytochemicals from the genus Gentiana. Molecules 2017, 22, 2080. [Google Scholar] [CrossRef]
  19. Zhang, W.; Jiang, H.; Yang, J.; Song, G.; Wen, D.; Liu, W.; Jin, M.; Wang, Q.; Du, Y.; Sun, Q.; et al. A high-throughput metabolomics approach for the comprehensive differentiation of four Pulsatilla Adans herbs combined with a nontargeted bidirectional screen for rapid identification of triterpenoid saponins. Anal. Bioanal. Chem. 2019, 411, 2071–2088. [Google Scholar] [CrossRef]
  20. Theodoridis, G.A.; Gika, H.G.; Want, E.J.; Wilson, I.D. Liquid chromatography-mass spectrometry based global metabolite profiling: A review. Anal. Chim. Acta 2012, 711, 7–16. [Google Scholar] [CrossRef]
  21. Ma, S.; Chowdhury, S.K. Data acquisition and data mining techniques for metabolite identification using LC coupled to high-resolution MS. Bioanalysis 2013, 5, 1285–1297. [Google Scholar] [CrossRef]
  22. Lv, X.; Sun, J.Z.; Xu, S.Z.; Cai, Q.; Liu, Y.Q. Rapid characterization and identification of chemical constituents in Gentiana radix before and after wine-processed by UHPLC-LTQ-Orbitrap MS(n). Molecules 2018, 23, 3222. [Google Scholar] [CrossRef] [PubMed]
  23. Saez, V.; Riquelme, S.; Baer, D.V.; Vallverdu-Queralt, A. Phenolic profile of Grape Canes: Novel compounds identified by LC-ESI-LTQ-Orbitrap-MS. Molecules 2019, 24, 3163. [Google Scholar] [CrossRef] [PubMed]
  24. Pan, H.; Yang, W.; Zhang, Y.; Yang, M.; Feng, R.; Wu, W.; Guo, D. An integrated strategy for the systematic characterization and discovery of new indole alkaloids from Uncaria rhynchophylla by UHPLC/DAD/LTQ-Orbitrap-MS. Anal. Bioanal. Chem. 2015, 407, 6057–6070. [Google Scholar] [CrossRef] [PubMed]
  25. Blazenovic, I.; Kind, T.; Ji, J.; Fiehn, O. Software tools and approaches for compound identification of LC-MS/MS data in metabolomics. Metabolites 2018, 8, 31. [Google Scholar] [CrossRef]
  26. Dinda, B. Pharmacology and Applications of Naturally Occurring Iridoids; Springer: Basel, Switzerland, 2019. [Google Scholar]
  27. Suying, L.; Fengrui, S.; Zhiqiang, L. Mass Spectrometry Analysis of Traditional Chinese Medicine; Science Press: Beijing, China, 2012. [Google Scholar]
  28. Qin, J.P.; Deng, J.G.; Feng, Y.Q.; Feng, X. Applications of organic mass spectrometry in structure identification of an impurity compound in prepared mangiferin extracted from Mangifera indica L. leaves. J. Chin. Mass Spectrom. Soc. 2008, 29, 218–225. [Google Scholar]
  29. Tugizimana, F.; Steenkamp, P.A.; Piater, L.A.; Dubery, I.A. A Conversation on data mining strategies in LC-MS untargeted metabolomics: Pre-processing and pre-treatment steps. Metabolites 2016, 6, 40. [Google Scholar] [CrossRef]
  30. Gromski, P.S.; Xu, Y.; Hollywood, K.A.; Turner, M.L.; Goodacre, R. The influence of scaling metabolomics data on model classification accuracy. Metabolomics 2014, 11, 684–695. [Google Scholar] [CrossRef]
  31. Pezzatti, J.; Boccard, J.; Codesido, S.; Gagnebin, Y.; Joshi, A.; Picard, D.; González-Ruiz, V.; Rudaz, S. Implementation of liquid chromatography–high resolution mass spectrometry methods for untargeted metabolomic analyses of biological samples: A tutorial. Anal. Chim. Acta 2020. [Google Scholar] [CrossRef]
Sample Availability: Not available.
Figure 1. The typical Base Peak ion (BPI) of Gentianae Radix et Rhizome (GRR) chromatographic separation.
Figure 1. The typical Base Peak ion (BPI) of Gentianae Radix et Rhizome (GRR) chromatographic separation.
Molecules 25 01228 g001
Figure 2. The typical BPI of GRR in positive and negative ion mode.
Figure 2. The typical BPI of GRR in positive and negative ion mode.
Molecules 25 01228 g002
Figure 3. Optimization of sample extraction conditions: (A) Different extraction solvent, (B) Different methanol ratio.
Figure 3. Optimization of sample extraction conditions: (A) Different extraction solvent, (B) Different methanol ratio.
Molecules 25 01228 g003
Figure 4. Deductive fragmentation pathway of rindoside.
Figure 4. Deductive fragmentation pathway of rindoside.
Molecules 25 01228 g004
Figure 5. Deductive fragmentation pathway of homoorientin.
Figure 5. Deductive fragmentation pathway of homoorientin.
Molecules 25 01228 g005
Figure 6. Deductive fragmentation pathway of mangiferin.
Figure 6. Deductive fragmentation pathway of mangiferin.
Molecules 25 01228 g006
Figure 7. Characterization of chemical constituents of two species of GRR by UHPLC -LTQ-Orbitrap/MS in negative ion mode.
Figure 7. Characterization of chemical constituents of two species of GRR by UHPLC -LTQ-Orbitrap/MS in negative ion mode.
Molecules 25 01228 g007
Figure 8. Heat map formed by 283 features (p-value < 0.05) in Gentiana scabra and Gentiana rigescens.
Figure 8. Heat map formed by 283 features (p-value < 0.05) in Gentiana scabra and Gentiana rigescens.
Molecules 25 01228 g008
Figure 9. (a) Accumulated R2X and Q2 with different numbers of PCs. (b) score plot of PCA-X model for data visualization and experimental stability evaluation of two Gentiana species and QC samples.
Figure 9. (a) Accumulated R2X and Q2 with different numbers of PCs. (b) score plot of PCA-X model for data visualization and experimental stability evaluation of two Gentiana species and QC samples.
Molecules 25 01228 g009
Figure 10. (a) Score plot of OPLS-DA model for differentiating two Gentiana species. (b) Permutation plot of two classes at 200 times of permutations. (c) Variable importance for the projection (VIP) of each variable. (d) S-plot of each variable.
Figure 10. (a) Score plot of OPLS-DA model for differentiating two Gentiana species. (b) Permutation plot of two classes at 200 times of permutations. (c) Variable importance for the projection (VIP) of each variable. (d) S-plot of each variable.
Molecules 25 01228 g010
Figure 11. (a) Cumulative number of each variable in the top Nth of 100 RF iterative modeling. (b) Frequency of the variable as the most important parameter in 100 RF iterative modeling.
Figure 11. (a) Cumulative number of each variable in the top Nth of 100 RF iterative modeling. (b) Frequency of the variable as the most important parameter in 100 RF iterative modeling.
Molecules 25 01228 g011
Table 1. Tentative characterization of chemical constituents of two species of GRR by UHPLC-LTQ-Orbitrap/MS in negative ion mode.
Table 1. Tentative characterization of chemical constituents of two species of GRR by UHPLC-LTQ-Orbitrap/MS in negative ion mode.
NoNameFormulam/zAdductionFragment IonRTScabraRigescens
Iridoid
1Secologanic acidC16H22O10373.11[M-H]179/108/204/2828.40 +
2Longanic acid *C16H24O10375.13[M-H]371/1792.35+
3GentiolactoneC10H12O5211.10[M-H]1677.50++
4Gentiopicroside*C16H20O9401.11[M-H+HCOOH]179/355/149/1195.22++
5Sweroside *C16H22O9403.12[M-H+HCOOH]357/195/179/1255.63++
68-epikingsideC17H24O11403.12[M-H]371/223/1797.49+
7CaryptosideC17H26O11405.14[M-H+HCOOH]1792.26++
8Swertiamarin *C16H22O10419.12[M-H+HCOOH]179/355/211/1193.85++
9Loganin *C17H26O10435.22[M-H+HCOOH]389.227.17 +
10SecoxyloganinC17H24O11449.13[M-H+HCOOH]179/241/359/4032.66+
11MorronisideC17H26O11451.14[M-H+HCOOH]405/243/179/2.94+
122′-O-(2,3-dihyrben)-gentiopicrosideC20H28O14491.14[M-H]167/323/4594.59++
132′-O-(2,3-dihyben)-swertamairnC23H26O13509.13[M-H]153/297/315/36710.53 +
143′-O-(2,3-dihyben)-swertamairnC23H26O13509.22[M-H]153/517/411/3216.11 +
15Deglu-noneacetylate-rindoisideC23H26O13509.22[M-H]153/297/315/3679.53 +
16Rigenolide AC25H28O12519.15[M-H]30711.20 +
174-glu+D97RTC26H34O11521.20[M-H]359/3299.44 +
18LacriciresinolC26H34O11521.20[M-H]473/355/375/41515.26 +
196′-O-d-glu swertiamarinC25H28O13535.14[M-H]409/491/153/33914.39+
206′-O-d-glu-Loganic acidC22H34O15537.18[M-H]2132.05++
21Dideacetylate-deglu-rindosideC25H28O14551.14[M-H]491/409/50912.34 +
22GentianasideC22H36O13553.13[M-H+HCOOH]5075.03+
236-O-d-glu-gentiopicrosideC22H30O14563.16[M-H+HCOOH]341/5175.69++
24Tortoside BC28H38O13581.15[M-H]401/357/313/2691.47 +
25GentiabavarosideC26H30O15581.16[M-H]401/357/313/2211.47+
266-O-d-glu-swertianmarinC22H32O15581.17[M-H+HCOOH]341/535/517/1792.83+
27Deacetylate-deglu-rindosideC27H30O15593.15[M-H]451/531/55119.54++
28Deglu-triflorosideC29H32O15619.17[M-H]57724.5++
29Deglu-gelidosideC29H32O16635.16[M-H]551/451/59323.19++
30GentiotriflorosideC29H36O17655.19[M-H]315/493/5297.41++
31Deglu-scabrasideC34H34O15681.18[M-H]639/475/15328.45++
32Gentrigeoside A C36H60O12683.40[M-H]640/62228.46+
332,3-deacetyl-triflorosideC32H40O17695.18[M-H]31513.78+
346′-O-ace-3-O-glu-2-hy-swerosideC31H38O18697.17[M-H]315/535/571/31514.25+
35Trideacetylate-triflorosideC31H38O18697.18[M-H+HCOOH]505/651/313/2698.78 +
36Scabran G3C28H40O19725.21[M-H]341/383/503/6792.55++
372-deaceyl-triflorosideC33H38O19737.19[M-H]315/575/69316.1+
38Dedihydroxybenzoate-MacrophyllosideC33H40O19739.20[M-H]697/577/613/53515.76++
39Deacetylate-TriflorosideC33H40O19739.20[M-H]697/577/613/53514.7++
40Deacetylate-RindosideC33H40O20755.20[M-H]593/315/71313.92++
41Trifloroside *C35H42O20781.22[M-H]619/739/577/31518.44+
42Dideacetylate-MacophyllosideC36H40O20791.24[M-H]521/629/31515.11+
43Rindoside *C35H42O21797.21[M-H]315/493/635/75517.19++
44Deacetylate-scabrasideC38H42O19801.22[M-H]639/59718.41+
45Acetylate-triflorosideC37H44O21823.23[M-H]619/577/78121.59+
46Deacetylatemacrophylloside AC38H42O21833.21[M-H]671/697/535/31515.33++
47ScabrasideC40H44O20843.24[M-H]681/639/315/80121.66++
48Dideacetylate-4′-glu-triflorosideC37H48O23859.23[M-H]69719.31++
49Macrophylloside AC40H44O22875.22[M-H]739/577/53518,74++
50Deacetylate-4′-glu-triflorosideC39H50O24901.26[M-H]577/459/535/85913.36+
514-O-β-d-triflorosideC41H52O25943.27[M-H]459/619/901/57716.29++
52Acetylate-4′-glu-scabrasideC46H54O251005.29[M-H]963/681/639/47720.73+
534″-O-β-d-glucopyranosy1-6′-O-(4-O-β-d-glu-caffeoyl) linearrosideC46H56O251007.30[M-H]845/801/487/63911.59+
54Benzoxy-4″-O-β-d-glucopyranosy1-6′-O-(4-O-β-d-glucopyranosylcaffeoyl)linearroside C53H60O261111.33[M-H]845/487/639/80115.03+
Flavonoids
55IsovitexinC21H20O10431.10[M-H] 9.7++
56Isoorientin(Homoorientin) *C21H20O11447.15[M-H]327/357/4298.29++
57IsoscoparinC22H22O11507.17[M-H+COOH]4616.66 +
582-glu-isovitexinC27H30O15593.15[M-H]551/451/531/40919.54+
59IsosaponarinC27H30O16593.15[M-H]311/431/473/5037.68 +
60SaponarinC27H30O17593.15[M-H]36711.51 +
614′-glu-isoorientinC27H30O16609.14[M-H]4475.27 +
62LutonarinC27H30O16609.14[M-H]447/519/489/3276.87 +
63RutinC27H30O16609.14[M-H]447/519/489/3276.9 +
64KeampferolC15H10O6331.04[M-H+HCOOH]285/16522.50+
657-glu-isopyrenineC29H34O17653.17[M-H]31510.06++
66Hyperoside C21H20O12509.22[M-H+HCOOH]4638.95+
67LonicerinC27H30O15593.19[M-H]551/45117.06+
Miscellaneous Compounds
68Caffeic acidC9H8O4179.03[M-H]135/1099.25++
69Ferulic acidC10H10O4193.02[M-H]1493.77+
70Isoferulic acidC10H10O4193.02[M-H]1494.01+
71VanillolosideC13H16O9315.07[M-H]1531.6++
72Glu-2,3-dihydroxybenzoic acidC14H20O8315.11[M-H]187/297/253/14522.81+
73Glu-2-hydro-3-methobenC14H18O9329.09[M-H]1676.89++
74Methyl-3-(β-d-glucopyranosyl)-2-hydroxybenzoateC14H18O9329.23[M-H]209/31119.25+
75Glu-caffeic acid C15H18O9341.09[M-H]179/135/203/2391.97 +
76SyringinC17H24O9371.10[M-H]2494.7+
773-[(6-O-Arabinopyranosyl)-β-d-glucopyranosyloxy]oct-1-enC19H34O10421.20[M-H]289/13312.57+
78Methyl-3-[(6-O-β-d-glucopyranosyl)-β-d- glucopyranosyloxy]-2-hydroxybenzoateC23H24O12491.12[M-H]153/315/47512.31 +
79(+) SyringaresinolC24H30O8491.14[M-H+HCOOH]315/447/1539.94 +
80LonicerinC27H30O15593.19[M-H]551/451/16.68 +
Xanthones
81Mangiferin *C19H18O11421.08[M-H]403/331/3015.98+
82GentianabavarosideC26H30O15581.16[M-H]401/357/313/3.71+
Triterpenoids
83Gentrigeoside DC36H60O13745.40[M-H+HCOOH]69911.99 +
84Gentrigeoside CC42H70O17845.24[M-H]683/80321.77+
85Glu-Gentigeoside CC46H56O251007.31[M-H]845/801/639/48711.52+
86Gentrigeoside A (Dammarane)C36H60O12683.40[M-H]640/62228.46+
87Rha-GentrigeosideC46H54O271037.51[M-H+HCOOH]99113.6+
Annotation: + indicated the compound in the specie; * indicated compound compared with standards; blank indicated without the compound.
Table 2. Identification of chemical markers from two species of GRR.
Table 2. Identification of chemical markers from two species of GRR.
No.NameFormulaRTm/zAdductionFragment Ion
1(+)SyringaresinolC24H30O89.94491.14[M-H+HCOOH]315/447/153
2LutonarinC27H30O166.87609.14[M-H]447/519/489/327
3TriflorosideC35H42O2018.44781.22[M-H]619/739/577/315
44-β-d-glu-triflorosideC41H52O2516.29943.27[M-H]459/619/901/577
54″-O-β-d-glucopyranosy1-6′-O-(4-O-β-d-glucopyranosylcaffeoyl)linearrosideC46H56O2511.591007.30[M-H]845/323/487/639/801
6Macrophylloside AC40H44O2218.74875.22[M-H]739/577/535
7ScabrasideC40H44O2021.66843.24[M-H]681/639/315/801
8Unknown 25.12717.46
9Unknown 15.111105.32
10Unknown 19.00891.22
11Unknown 15.631053.27
Table 3. Sample collection information of GRR.
Table 3. Sample collection information of GRR.
SpecieslocationSample NumberSpeciesSpeciesSample Number
Gentiana scabraQY-wdz1–20Gentiana scabraYs-j52
Gentiana scabraQY--yem21–39Gentiana rigescensYx-hsc53–57
Gentiana scabraQY-40–49Gentiana rigescensYx-jfc58–62
Gentiana scabraYs-y50Gentiana rigescensYx-xh63–72
Gentiana scabraYs-a51Gentiana rigescensYx-th73–86
Back to TopTop