Quality Evaluation of Tetrastigmae Radix from Two Different Habitats Based on Simultaneous Determination of Multiple Bioactive Constituents Combined with Multivariate Statistical Analysis

Tetrastigmae Radix, also known as Sanyeqing (SYQ) in Chinese, is an important traditional Chinese medicine with a long history. Tetrastigma hemsleyanum Diels et Gilg mainly grows in the south of the Yangtze River and is widely distributed. The content of bioactive constituents in SYQ varies greatly in different habitats, and there are obvious differences in the content of bioactive constituents between southwestern SYQ (WS) and southeastern SYQ (ES). To distinguish and evaluate the quality of ES and WS, an analytical method based on ultrafast performance liquid chromatography coupled with triple quadrupole-linear ion trap mass spectrometry (UFLC-QTRAP-MS/MS) was established for the simultaneous determination of 60 constituents including 25 flavonoids, 9 phenolic acids, 15 amino acids, and 11 nucleosides in 47 samples from ES and WS. In addition, orthogonal partial least squares discriminant analysis (OPLS-DA), t-test, and gray correlation analysis (GRA) were used to discriminate and evaluate the ES and WS samples based on the contents of 60 constituents. The results showed that there were significant differences in the bioactive constituents between ES and WS, and ES was superior to WS in terms of quality evaluation. This study not only provides basic information for differentiating ES and WS but also provides a new perspective for the comprehensive evaluation and quality control of SYQ from two different habitats.

The quality of SYQ is affected by factors such as habitat [17], seed source [18], growth environment [19], and processing methods [20], and the appearance of the roots varies significantly, in which the accumulation of chemical constituents is also affected. Few studies have been conducted on these factors. Some studies have found that the habitat factor has a greater influence on the accumulation of constituents in SYQ [17], but they are not comprehensive enough.
At present, many analytical methods have been reported for quality assessment and control in SYQ, such as high-performance liquid chromatography (HPLC) [21][22][23], inductively coupled plasma-mass spectrometry (ICP-MS) [24], and liquid chromatographymass spectrometry (LC-MS) [4,[25][26][27]. More studies have focused on the quantitative analysis of total flavonoids, total phenolic acids, and polysaccharide content, and few have studied the quantitative analysis of phenolic acids, amino acids, and nucleosides of SYQ based on LC-MS. It is necessary to establish a method combining multivariate index constituents to distinguish SYQ from different habitats.
This study aimed to identify and evaluate the quality of southeastern SYQ (ES) and southwestern SYQ (WS) based on the simultaneous determination of multiple bioactive constituents in combination with multivariate statistical analysis. A reliable method based on ultrafast liquid chromatography coupled with triple quadrupole-linear ion trap tandem mass spectrometry (UFLC-QTRAP-MS/MS) for the simultaneous determination of 60 constituents in SYQ was developed. Orthogonal partial least squares discriminant analysis (OPLS-DA) and t-test were applied to distinguish and reveal the differential constituents of ES and WS. In addition, gray correlation analysis (GRA) was used to assess the quality of SYQ based on the correlation between the detected component contents and the samples. The established method can provide a basis for a comprehensive evaluation and quality control of SYQ from two different habitats, and it provides fundamental data to distinguish between ES and WS.

Optimization of UFLC Conditions
To obtain the best chromatographic conditions, UFLC chromatographic conditions such as column, mobile phase, and column temperature were optimized to achieve a higher separation effect and better peak shape of the target constituents in SYQ. The results showed that the separation capacity and sensitivity of the XBridge ® C18 column (4.6 mm × 100 mm, 3.5 µm) were relatively superior. In addition, five mobile phase systems (water-methanol, water-acetonitrile, water-methanol:acetonitrile (1:1), 0.1%, 0.4%, 0.8% formic acid water solution-methanol solution, and 0.1%, 0.4%, 0.8% formic acid water solution-acetonitrile solution), flow rates (0.3, 0.4, 0.5, 0.7, 0.8, 0.9, 1.0 mL/min), and column temperatures (25,30,35, 40 • C) were examined and compared. The expected separation was achieved by gradient elution with 0.4% formic acid as Eluent A and methanol as Eluent B at a flow rate of 0.8 mL/min under the column temperature of 30 • C.

Optimization of Mass Spectrometry (MS) Conditions
The individual solutions of all standard compounds (about 100 ng/mL) were examined with the electrospray ionization (ESI) source in the positive and negative ion modes. After repeated experimental tests, amino acids and nucleosides showed good sensitivity and intensity in the positive ion mode, while flavonoids and phenolic acids were more suitable for detection in the negative ion mode. Uridine responded better in negative ion mode than in positive ion, and Orientin and Iso-orientin responded better in positive ion mode than in negative ion mode [28]. Therefore, both ESI+ and ESI− modes were used in this study. Although the retention times of some constituents were similar, they could be precisely quantified based on different precursor and product ion pairs. Table 1 lists the best details of the 60 constituents in terms of retention time (t R ), precursor and product ions, declustering potential (DP), and collision energy (CE). The MS spectra of 32 constituents in negative ion mode are shown in Figure S2, and the MS spectra of 28 constituents in positive ion mode are shown in Figure S3. Figure 1 shows the multiple reaction monitoring (MRM) for the 60 constituents.

Method Validation
All method validations of quantification were performed by the established UFLC-QTRAP-MS/MS method. The detailed results of each method validation are presented in Table 2. Each standard calibration curve was constructed by plotting the peak areas (Y) against the corresponding concentrations (X). All analytes showed good linearity with appropriate determination coefficients (r > 0.9989). The ranges of limits of detection and quantification (LODs and LOQs) were 0.03-13.59 ng/mL and 0.09-45.3 ng/mL, respectively. The relative standard deviations (RSDs) of intraday and interday variations ranged from 0.93% to 4.97% and 0.88% to 4.97%, respectively. The RSDs of the repeatability and stability were less than 4.98% and 4.99%, respectively. The overall recoveries varied from 96.1% to 101.76%, with RSDs < 4.87%. The slope ratio values of the matrix curve to the pure solution curve were between 0.92 and 1.05, indicating that the matrix effect on the ionization of analytes was not obvious under optimized conditions.  Table 1.).   Table 1.).

Quantitative Analysis of Samples
Sample information is shown in Figure 2. The validated analytical method of UFLC-QTRAP-MS/MS was successfully applied to simultaneously determine 60 constituents (25 flavonoids, 9 phenolic acids, 15 amino acids, and 11 nucleosides) in SYQ. The quantitative results of 60 constituents are presented in Table S1. The SYQ samples were all rich in amino acids, with total amino acid contents ranging from 360.04 to 2856.77 µg/g, accounting for more than 65% of the total analyte content in this study. In addition, the contents of Proline (10), Alanine (5), Phenylalanine (27), and Lysine (1) were relatively high. The total content of nucleosides ranged from 40.05 to 246.80 µg/g, with Adenosine (16), Hypoxanthine (14), Uridine (15), and Uracil (12) accounting for more than 86% of the total nucleoside content. The total content of phenolic acids was 4.33-134.78 µg/g, of which the content of Piceatannol (42) accounted for more than 58%. The total content of flavonoids was 26.2-2361.67 µg/g, of which Procyanidin B2 (30), Catechin (33), Nicotifiorin (53), Rutin (48), Isoquercitrin (49), Astragalin (52), Epicatechin (38) were relatively high, and 30 accounted for more than 64% of the total flavonoids. The content of 30, 33, and 38 in the ES sample was higher than that in the WS sample, while 52, 48, and 49 were on the contrary. Figure 3 shows that the total contents of amino acids, nucleosides, phenolic acids, and flavonoids in ES were significantly higher than those in WS.

Quantitative Analysis of Samples
Sample information is shown in Figure 2. The validated analytical method of UFLC-QTRAP-MS/MS was successfully applied to simultaneously determine 60 constituents (25 flavonoids, 9 phenolic acids, 15 amino acids, and 11 nucleosides) in SYQ. The quantitative results of 60 constituents are presented in Table S1. The SYQ samples were all rich in amino acids, with total amino acid contents ranging from 360.04 to 2856.77 μg/g, accounting for more than 65% of the total analyte content in this study. In addition, the contents of Proline (10), Alanine (5), Phenylalanine (27), and Lysine (1) were relatively high. The total content of nucleosides ranged from 40.05 to 246.80 μg/g, with Adenosine (16), Hypoxanthine (14), Uridine (15), and Uracil (12) accounting for more than 86% of the total nucleoside content. The total content of phenolic acids was 4.33-134.78 μg/g, of which the content of Piceatannol (42) accounted for more than 58%. The total content of flavonoids was 26.2-2361.67 μg/g, of which Procyanidin B2 (30), Catechin (33), Nicotifiorin (53), Rutin (48), Isoquercitrin (49), Astragalin (52), Epicatechin (38) were relatively high, and 30 accounted for more than 64% of the total flavonoids. The content of 30, 33, and 38 in the ES sample was higher than that in the WS sample, while 52, 48, and 49 were on the contrary. Figure 3 shows that the total contents of amino acids, nucleosides, phenolic acids, and flavonoids in ES were significantly higher than those in WS.  Table 4.

OPLS-DA of Samples
Firstly, principal component analysis (PCA) was used to differentiate and assess the quality of ES and WS. Since Principal Component Analysis (PCA) could not clearly reflect the classification of the measured samples, it was not suitable to provide a basis for differentiating and evaluating the quality of ES and WS. Orthogonal partial least squares discriminant analysis (OPLS-DA) is a supervised latent structure discriminant analysis method that maximizes between-group variation and minimizes within-group separation. Orthogonal partial least squares discriminant analysis (OPLS-DA) is a supervised latent structure discriminant analysis method that maximizes between-group variation and minimizes within-group separation. The method maximizes group differences and minimizes within-group separation. Figure 4 shows the OPLS-DA score plot. ES and WS were divided into two groups, thus indicating significant differences in chemical composition between them. R2 describes the degree of fitting of the model. Q2 describes X's ability to predict Y. It is generally believed that Q2 greater than 0.5 indicates that the model has good reliability and predictability, and greater than 0.9 is excellent [29]. In this comparison, the statistical parameters of OPLS-DA R2X(cum), R2Y(cum), and Q2(cum) are 0.767, 0.931, and 0.895, respectively, indicating that the model has good repeatability and predictability. The variable importance of projection (VIP) is a vector summarizing the total importance of variables in explaining the model. If a variable's VIP > 1, it indicates that the variable contributes significantly to the classification of these samples. As shown in Figure 5, according to the VIP value, eight constituents were found to play a dominant role in the cluster, including Lysine (1), Histidine (2), Alanine (5), Proline (10), Leucine (25), Phenylalanine (27), Procyanidin B2 (30), and Catechin (33).

OPLS-DA of Samples
Firstly, principal component analysis (PCA) was used to differentiate and assess the quality of ES and WS. Since Principal Component Analysis (PCA) could not clearly reflect the classification of the measured samples, it was not suitable to provide a basis for differentiating and evaluating the quality of ES and WS. Orthogonal partial least squares discriminant analysis (OPLS-DA) is a supervised latent structure discriminant analysis method that maximizes between-group variation and minimizes within-group separation. Orthogonal partial least squares discriminant analysis (OPLS-DA) is a supervised latent structure discriminant analysis method that maximizes between-group variation and minimizes within-group separation. The method maximizes group differences and minimizes within-group separation. Figure 4 shows the OPLS-DA score plot. ES and WS were divided into two groups, thus indicating significant differences in chemical composition between them. R2 describes the degree of fitting of the model. Q2 describes X's ability to predict Y. It is generally believed that Q2 greater than 0.5 indicates that the model has good reliability and predictability, and greater than 0.9 is excellent [29]. In this comparison, the statistical parameters of OPLS-DA R2X(cum), R2Y(cum), and Q2(cum) are 0.767, 0.931, and 0.895, respectively, indicating that the model has good repeatability and predictability. The variable importance of projection (VIP) is a vector summarizing the total importance of variables in explaining the model. If a variable's VIP > 1, it indicates that the variable contributes significantly to the classification of these samples. As shown in Figure 5, according to the VIP value, eight constituents were found to play a dominant role in the cluster, including Lysine (1), Histidine (2), Alanine (5), Proline (10), Leucine (25), Phenylalanine (27), Procyanidin B2 (30), and Catechin (33).   Table 1. The eight constituents of VIP >1 are shown in the enlarged graph.

T-Test of Samples
T-test was used to analyze the contents of bioactive constituents detected to evaluate the changes of 60 constituents in ES and WS, and it was considered that the values with p values less than 0.05 had significant differences. As shown in Figure 6, more than half of    Table 1. The eight constituents of VIP >1 are shown in the enlarged graph.

T-Test of Samples
T-test was used to analyze the contents of bioactive constituents detected to evaluate the changes of 60 constituents in ES and WS, and it was considered that the values with p values less than 0.05 had significant differences. As shown in Figure 6, more than half of Figure 5. VIP for classification of ES and WS. The X-axis numbers denoted are the same as those in Table 1. The eight constituents of VIP >1 are shown in the enlarged graph.

T-Test of Samples
T-test was used to analyze the contents of bioactive constituents detected to evaluate the changes of 60 constituents in ES and WS, and it was considered that the values with p values less than 0.05 had significant differences. As shown in Figure 6, more than half of the bioactive constituents in ES were higher (p < 0.05) than those in WS. The contents of Lysine, Histidine, Glycine, Serine, Alanine, Aspartic acid, Threonine, Cytidine, Guanosine, Isoleucine, Leucine, Phenylalanine, Procyanidin B2, Catechin, Procyanidin B1, Epicatechin, and Polydatin were significantly higher (p < 0.001) in ES than that in WS. However, the content of 2 -Deoxyadenosine was higher (p < 0.001) in WS than that in ES. In combination with OPLS-DA analysis, Procyanidin B2, Catechin, Lysine, Histidine, Alanine, Leucine, and Phenylalanine could be the most effective chemical markers to distinguish ES and WS.   Figure 6. Cont. Figure 6. The box plot of 60 constituents' contents in ES and WS (* p < 0.05, ** p < 0.01, *** p < 0.001). Figure 6. The box plot of 60 constituents' contents in ES and WS (* p < 0.05, ** p < 0.01, *** p < 0.001).

GRA of Samples
GRA is a measure of influence in gray systems theory that analyzes the uncertain relationship between a major factor and all other factors in a given system. Therefore, a comprehensive GRA evaluation of ES and WS was performed based on the content of 60 bioactive constituents. The GRA results including the gray comprehensive evaluation value (r i ) and the quality rankings are shown in Table 3. The r i indicates the relative correlation between component content and samples. The samples with higher relative correlation are of better quality.  Table 3 shows the grey comprehensive evaluation value (r i ) and the quality ranking. From this perspective, the overall quality of ES was significantly better than that of WS. SYQ produced in Zhejiang was of better quality relative to other provinces. In addition, among the southwestern habitat, SYQ produced in Guangxi was better and those produced in Yunnan and Guizhou were worse. The difference of r i varied widely, with a maximum value of 38.67%, which could well-distinguish the quality of the samples. However, there were differences in the quality of SYQ from different provinces, which could be attributed to latitude, altitude, and harvest time. In summary, as can be seen in Table 3, GRA can successfully assess the quality of SYQ based on the content of its multiple components.

Discussion
A highly efficient and reliable UFLC-QTRAP-MS/MS method was developed for the simultaneous determination of 25 flavonoids, 9 phenolic acids, 15 amino acids, and 11 nucleosides in 47 samples. The OPLS-DA, VIP values, and t-test indicated that there were significant differences in the bioactive constituents in ES and WS, such as procyanidin B2 (30), catechin (33), lysine (1), histidine (2), alanine (5), leucine (25), and phenylalanine (27), which can be considered to distinguish and control the quality of ES and WS (Figures 5 and 6). Compared with existing research methods [21][22][23][24], this method analyzes a large number of active constituents, including pharmacological and nutritional constituents. At the same time, the mass spectrometry detector uses the positive and negative ion mode for simultaneous determination, which allows precise determination of the molecular weight of the constituents and is more sensitive than conventional detectors such as UV detectors. This method can overcome the shortcomings of traditional detection methods and effectively reveal the complexity of sample composition, but it also has some limitations. Since the UFLC-QTRAP-MS/MS method is suitable for the determination of small molecules and constituents with low boiling points, there are difficulties in the identification of constituents with large molecular weights and volatile constituents. Though this drawback exists, the 60 constituents with small molecular weight and high boiling point detected in this experiment showed good response values in this method.
Amino acids are nutritional constituents, and the total amino acid content is higher than 50% in SYQ ( Figure 3). As the total amino acids in SYQ have some hepatic protective effects and are poorly studied [16], the UFLC-QTRAP-MS/MS method established in this study provides a basis for the amino acid content determination. From Figure 5, the VIP-value of constituents 2, 10, and 25 is close, with a range of 1.0-1.5. Meanwhile, for constituents 1, 5, 27, 30, and 33, VIP is > 2. Procyanidin B2 (30) and catechin (33) are flavonoids. The flavonoid constituents of SYQ have antitumor [5] and antioxidant [7] effects, and these constituents may provide a new idea for the study of the antitumor activity of SYQ from different origins. Based on the r i values of the samples (Table 3), the overall quality of ES was better than WS, indicating that different geographical regions can influence the accumulation of bioactive constituents. Meanwhile, the GRA data showed some differences in samples from the same habitat, possibly related to factors such as geographic environment and cultivation techniques. In conclusion, the UFLC-QTRAP-MS/MS method combined with multivariate statistical analysis can provide basic information for the identification and quality evaluation of SYQ from different habitats.

Plant Materials
Plant materials were collected from nine provinces, including Zhejiang, Fujian, Jiangxi, Hunan, Hubei, Guangxi, Guizhou, Yunnan, and Chongqing. Table 4 shows the detailed geographic habitats of each sample. All the samples were authenticated by Professor Xunhong Liu (Nanjing University of Chinese Medicine, Nanjing, China) and were deposited in the laboratory of Chinese medicine identification, Nanjing University of Chinese Medicine. Detailed information is shown in Table 4. An API5500 triple quadrupole linear ion trap tandem mass spectrometer (AB SCIEX, Framingham, MA, USA) equipped with an electrospray ionization (ESI) source was used for detection. The operating parameters were as follows: ion source temperature, 550 • C; nebulizer gas (GS1) flow, 55 L/min; auxiliary gas (GS2) flow, 55 L/min; curtain gas (CUR) flow, 40 L/min; spray voltage (IS), 4500 V in the positive mode and −4500 V in the negative mode. Detection of analytes was performed in multiple-reaction mode (MRM).

Validation of the Method
The method was validated in terms of linearity, precision of intraday and interday, repeatability, stability, recovery, and matrix effect. Serial dilutions of mixed standards were used to establish the standard curves, and the linear regression equation, correlation coefficient, and linear range were calculated. The detection limit (LOD) and quantification limit (LOQ) for 33 constituents were calculated at the signal-to-noise ratios of 3 and 10, respectively. For intraday precision, the mixed standards solutions were injected for six replicates within one day, while for interday precision, the solutions were examined in triplicates for 3 consecutive days. To validate the repeatability, six samples of SYQ were accurately weighed and prepared independently according to the optimal conditions above and then analyzed. The same sample solution was taken and determined at 0, 2, 4, 8, 12, and 24 h according to the above chromatographic conditions to evaluate the stability. The recovery experiments were used to assess the accuracy of the method; standards at three different concentration levels, including low (80%), median (100%), and high (120%) were added to samples of known content. Each experiment was repeated three times, and the spiked samples were analyzed by UFLC-QTRAP-MS/MS to evaluate the recoveries. The recoveries were calculated by the formulae: recovery (%) = (detected amount − original amount)/spiked amount × 100%. The matrix effect refers to the enhancement or suppression of a chromatography signal by interference or coeluting constituents in the matrix. It was evaluated using a slope comparison method. In this way, the matrix effect was determined to be the ratio of the slope in a matrix-matched calibration curve to the slope in a solvent standard curve. The slope ratio close to 1.0 indicates that the matrix effect is weaker.

Multivariate Statistical Analysis
After data preprocessing, the global clustering trend of each group was observed by applying OPLS-DA and its distribution was visualized using SIMCA-P 13.0 software (Umetrics AB, Umea, Sweden). Further, statistical analysis of all detected component data was performed by t-test (SPSS 16.0 for Windows, IBM, Armonk, NY, USA) to detect differential constituents between ES and WS. Based on the results of quantitative measurements and t-tests, box line plots were created by OriginPro 2021b (OriginLab, Northampton, MA, USA) to obtain metabolite distribution maps and to analyze the differences between ES and WS. The quality of ES and WS samples was assessed based on the content of 60 active constituents using GRA, using Excel for Mac 2019 (Microsoft Corporation, Seattle, WA, USA). OriginPro 2021b (OriginLab, Northampton, MA, USA) was used to plot all histograms.

Conclusions
A reliable UFLC-QTRAP-MS/MS method was developed for the simultaneous determination of 60 constituents including flavonoids, phenolic acids, amino acids, and nucleosides in SYQ. Furthermore, multivariate statistical analyses such as OPLS-DA analysis, t-test, and GRA were applied to comprehensively analyze and evaluate different habitats of SYQ (ES and WS). The OPLS-DA analysis and t-test were applied to classify and identify SYQ from different habitats. It was found that ES and WS differed significantly and their classification was related to the differential constituents, such as procyanidin B2 (30), catechin (33), lysine (1), histidine (2), alanine (5), leucine (25), and phenylalanine (27), which could be used as chemical markers to distinguish ES between WS. In addition, the GRA results showed that ES was better in quality based on the content of 60 constituents.
These results suggest that the accumulation and quality of bioactive constituents in SYQ are influenced by different habitats. This research not only provides a foundation for distinguishing ES and WS but also for the comprehensive evaluation and quality control of SYQ in different habitats.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules27154813/s1, Table S1: Contents of 60 constituents in samples. Figure S1: Effects of ethanol concentration, solid-liquid ratio, and extraction time on extraction yields of four constituents. (extraction yield (%) = weight of analyte (mg)/weight of dried sample (g) × 100%). Figure S2: The MS spectra of 32 constituents in negative ion modes. Figure S3: The MS spectra of 28 constituents in positive ion modes.