Bivariate Correlation Analysis of the Chemometric Profiles of Chinese Wild Salvia miltiorrhiza Based on UPLC-Qqq-MS and Antioxidant Activities

To better understand the mechanisms underlying the pharmacological actions of Salvia miltiorrhiza, correlation between the chemical profiles and in vitro antioxidant activities in 50 batches of wild S. miltiorrhiza samples was analyzed. Our ultra-performance liquid chromatography–tandem mass spectrometry analysis detected twelve phenolic acids and five tanshinones and obtained various chemical profiles from different origins. In a principal component analysis (PCA) and cluster analysis, the tanshinones cryptotanshinone, tanshinone IIA and dihydrotanshinone I exhibited higher weights in PC1, whereas the phenolic acids danshensu, salvianolic acids A and B and lithospermic acid were highly loaded in PC2. All components could be optimized as markers of different locations and might be suitable for S. miltiorrhiza quality analyses. Additionally, the DPPH and ABTS assays used to comprehensively evaluate antioxidant activities indicated large variations, with mean DPPH and ABTS scavenging potencies of 32.24 and 23.39 μg/mL, respectively, among S. miltiorrhiza extract solutions. Notably, samples that exceeded the mean IC50 values had higher phenolic acid contents. A correlation analysis indicated a strong correlation between the antioxidant activities and phenolic acid contents. Caffeic acid, danshensu, rosmarinic acid, lithospermic acid and salvianolic acid B were major contributors to antioxidant activity. In conclusion, phenolic compounds were the predominant antioxidant components in the investigated plant species. These plants may be sources of potent natural antioxidants and beneficial chemopreventive agents.


Introduction
Many human disorders, such as coronary heart disease, inflammation, diabetes, carcinogenesis and neurodegenerative diseases, may result from increased concentrations of free radicals produced by numerous physiological and biochemical processes within the body [1,2]. Plants (vegetables, fruits, medicinal herbs) contain a wide variety of free radical-scavenging molecules, such as phenolic acids, summarized in Table S1. The contents of the 17 investigated compounds varied among the tested S. miltiorrhiza samples, although the proportions of the constituents were quite similar among the samples.
In general, Salvianolic acid B (9) was the most abundant component, and it ranged in concentration from 22.57 to 46.63 mg/g. The concentration of salvianolic acid B met the standard required by the Pharmacopoeia of the People's Republic of China [22], except for samples S5, S11 and S38, which had salvianolic acid B concentrations of 25.13, 22.57 and 21.12 mg/g, respectively. The next most plentiful compounds were rosmarinic acid (7), lithospermic acid (8) and danshensu (1). The contents of protocatechualdehyde (3), cinnamic acid (11), protocatechuic acid (2), caffeic acid (4), ferulic acid (5) and isoferulic acid (6) were relatively low. However, these microconstituents could still play an important role in the Radix Salvia miltiorrhiza. The content of tanshinones (tanshinone IIA and cryptotanshinones) differed among the samples. The total content of cryptotanshinone, tanshinone IIA and tanshinone I ranged from 0.27 to 19.97 mg/g. The highest concentration was in sample S13 (Changzhi, China) and the lowest was in sample S45 (Hangzhou, China). According to the recent edition of the Chinese Pharmacopoeia (2015) [14], the total content of tanshinone I, tanshinone IIA and cryptotanshinone should not be less than 0.25%. More than 20 samples were not up to this standard, and most of these samples were from the southern provinces of Jiangxi, Hunan and Zhejiang. The coefficients of variation revealed that compounds from different locations were moderately variable, except for salvianolic acid C, tanshinone IIA and miltirone. The tanshinones compounds, including tanshinone IIA and cryptotanshinone, showed a larger range of variation than the phenolic acids. These results agree with earlier findings on S. miltiorrhiza samples from natural populations [23]. It was nonetheless surprising to find that the tanshinone content in samples from the south of China, such as the Jiangxi, Hunan and Zhejiang provinces, was lower than samples from the north.

Linear Relation Test
The antioxidant capabilities of different S. miltiorrhiza samples were determined using DPPH and ABTS assays, with Trolox as a positive control. Standard curves were obtained for both assays by measuring the ABTS and DPPH scavenging activities at different Trolox concentrations. The concentration-response curves for DPPH and ABTS are shown in Figure 3a,b, respectively. In the DPPH assay, the linear regression equation and coefficient of correlation for Trolox were y = 10.625x − 0.7075 and r 2 = 0.9974, respectively, where y is the % inhibition and x is the Trolox concentration. In the ABTS assay, the corresponding results were y = 16.129x − 1.1556 and r 2 = 0.9991, respectively.

Linear Relation Test
The antioxidant capabilities of different S. miltiorrhiza samples were determined using DPPH and ABTS assays, with Trolox as a positive control. Standard curves were obtained for both assays by measuring the ABTS and DPPH scavenging activities at different Trolox concentrations. The concentrationresponse curves for DPPH and ABTS are shown in Figure 3a,b, respectively. In the DPPH assay, the linear regression equation and coefficient of correlation for Trolox were y = 10.625x − 0.7075 and r 2 = 0.9974, respectively, where y is the % inhibition and x is the Trolox concentration. In the ABTS assay, the corresponding results were y = 16.129x − 1.1556 and r 2 = 0.9991, respectively.

Evaluation of the Antioxidant Capacities of S. miltiorrhiza Extracts Using DPPH and ABTS Assays
The DPPH and ABTS scavenging effects of 50 samples of S. miltiorrhiza from different places were examined at different concentrations. The IC 50 values shown in Table S2 indicate extremely large variations in the antioxidant activities of different S. miltiorrhiza samples. As with Trolox, the extract solutions exhibited antioxidant activity. As shown in Table S2, the r 2 value for the efficiency curve of each sample exceeded 0.99, indicating a good relationship between the sample and the DPPH and ABTS inhibition rates. Here, a lower IC 50 value implies higher antioxidant activity. and ABTS assays, with Trolox as a positive control. Standard curves were obtained for both assays by measuring the ABTS and DPPH scavenging activities at different Trolox concentrations. The concentration-response curves for DPPH and ABTS are shown in Figure 3a,b, respectively. In the DPPH assay, the linear regression equation and coefficient of correlation for Trolox were y = 10.625x − 0.7075 and r 2 = 0.9974, respectively, where y is the % inhibition and x is the Trolox concentration. In the ABTS assay, the corresponding results were y = 16.129x − 1.1556 and r 2 = 0.9991, respectively.  All samples exhibited strong, concentration-dependent antioxidant potentials against DPPH free radicals at 16.5449-85.2250 µg/mL. S8 most strongly inhibited DPPH (IC 50 = 16.5449 µg/mL), whereas S11 exhibited the weakest inhibition (IC 50 = 85.2250 µg/mL), a five-fold difference in antioxidant activity. The results of the ABTS assay paralleled those of the DPPH assay, with IC 50 values of 12.7269-56.9827 µg/mL. The best ABTS free radical scavenging activities were exerted by samples S8 and S15 (IC 50 = 12.7269 and 13.3907 µg/mL, respectively), whereas samples S38 and S11 exhibited lower activities (IC 50 = 48.4268 and 56.9827 µg/mL, respectively). The S. miltiorrhiza extract solutions had mean DPPH and ABTS scavenging potencies of 32.2369 and 23.3933 µg/mL, respectively. A comparable evaluation of S. miltiorrhiza by Matkowski et al. yielded a mean DPPH IC 50 value of 17.14 µg/mL [24], or slightly lower than the value obtained in this investigation. This discrepancy may be attributable to different extraction methods and sample collection locations.
Notably, samples that exceeded the mean IC 50 values had higher phenolic acid contents. For instance, samples S8, S33 and S15 exhibited the strongest free radical-scavenging capabilities in both assays and had the highest total phenolic acid contents, whereas S11 and S38 possessed lower antioxidant capacities and lower phenolic acid contents. In addition, differences in altitude and environment influenced the sample contents. For example, the contents varied among four samples (S25, S26, S27, S28) from Ji'an (Jiangxi province), with maximum and minimum IC 50 values of 33.8957 and 21.3639 µg/mL, respectively, in the DPPH assay. The antioxidant activities of these samples may have varied because they were collected from different habitats, such as wooded areas, fields, hillsides and roadsides.

PCA
A PCA was performed to evaluate the level of homogeneity in the quality of danshen collected from different locations in China. The PCA procedure facilitated the determination of the most important variables, or those representing a majority of the total information. As salvianolic acid C was not detected in some samples, only the contents of the other 16 investigated components were analyzed using SPSS. In this analysis, the first three principal components (F1-3) explained 63.888% of total variability among the 16 variables in the original data set ( Table 2). The tanshinones cryptotanshinone, tanshinone IIA and dihydrotanshinone I were more heavily weighted in F1, whereas the phenolic acids danshensu, salvianolic acids A and B and lithospermic acid were highly loaded in F2. Thus, Cryptotanshinone, tanshinone IIA, dihydrotanshinone I, danshensu, salvianolic acids A and B and lithospermic acid could be optimised as location markers, and all might be suitable for evaluating the quality of danshen. Additionally, cryptotanshinone, tanshinone IIA and dihydrotanshinone I exhibited obvious positive phase loading in F1, indicating that the F1 increased with increasing concentrations of these tanshinones. The opposite finding was observed for danshensu and salvianolic acids A and B, consistent with earlier analyses of S. miltiorrhiza samples [25,26].
An SPSS-based analysis of the investigated components in the PCA score plots (Figure 4) showed that S13 (Changzhi, China), which had the highest tanshinone concentration (28.4 mg/g), was distinct from the other samples. Although S5, S11, S38, S49 and S50, which had low total phenolic acid contents relative to the other samples, were considered to have clustered into one group, the remaining samples were clustered into another group, indicating that the proportions of the constituents were quite similar within this group. Overall, these results were similar to those obtained from the UPLC-Qqq-MS chemical profiles. To elucidate the relationships between the chemical profiles and antioxidant activities, a Pearson correlation analysis was subsequently introduced.

Correlation Analysis
A Pearson correlation analysis was used to evaluate the spectrum-effect relationships between the contents of 17 compounds and the IC50 values from the DPPH and ABTS free radical-scavenging assays. The results are shown in Table 3.

Correlation Analysis
A Pearson correlation analysis was used to evaluate the spectrum-effect relationships between the contents of 17 compounds and the IC 50 values from the DPPH and ABTS free radical-scavenging assays. The results are shown in Table 3. The correlation coefficients revealed close correlations of the antioxidant activities of methanolic S. miltiorrhiza extracts with the chemical compounds. The caffeic acid, danshensu, rosmarinic acid, lithospermic acid and salvianolic acid B, as well as total phenolic acids, exhibited obvious negative correlations with the IC 50 values (p < 0.05). These negative correlations were most significant for salvianolic acid B and total phenolic acids (p < 0.01). A smaller IC 50 value corresponds to stronger DPPH and ABTS free radical-scavenging abilities and, consequently, stronger antioxidant activity. The above results indicate that higher caffeic acid, danshensu, rosmarinic acid, lithospermic acid, salvianolic acid B and total phenolic acid contents corresponded to stronger antioxidant activities of S. miltiorrhiza. It was reported that the antioxidant capacity of the root of Salvia miltiorrhiza correlated with the total polyphenol and the hydroxycinnamic acid [24], including caffeic acid, danshensu, rosmarinic acid, lithospermic acid, salvianolic acid B etc. (such as the list in Figure 1). The results described herein highlight that the phenolic acid contents in S. miltiorrhiza were consistent with the assessed antioxidant tests, and that phenolic acids strongly contributed to the antioxidant activities of the S. miltiorrhiza samples analysed in this study. These findings were consistent with previous findings that the antioxidant capacity depends significantly on the contents of danshensu, salvianolic acid B and total phenolic acids [12,27]. Therefore, S. miltiorrhiza might be a source of potent natural antioxidants and beneficial chemo-preventive agents. By contrast, the result in this research indicated that the tanshinones content correlated positively with the IC 50 values of DPPH and ABTS, corresponded to lower antioxidant activities. Specifically, dihydrotanshinone I exhibited a significant positive correlation with the IC 50 values (p < 0.05) (as shown in Table 3). However, up to now, there are no reports providing evidence that tanshinones have a direct relationship with antioxidant activity. To find the reason, we calculated the intercorrelation coefficients between effective compounds shown in Table S4. It was indicated that tanshinones content correlated negatively with total phenolics, especially, dihydrotanshinone I exhibited a significant negative correlation with caffeic acid, rosmarinic acid, salvianolic acid B, and total phenolics, respectively. Thus, the general significant positive correlation of tanshinones with the IC 50 values may be caused by the opposite content of phenolic acids in S. miltiorrhiza extracts, which were already proved to have strong antioxidant activities. In addition, a strong positive correlation was observed between the DPPH and ABTS assays, with a Pearson correlation coefficient of r = 0.980 (p < 0.01). This finding agrees with the results of the previous study, which determined a correlation coefficient of r = 0.949 between the DPPH and ABTS assays [28].

Materials and Chemicals
The S. miltiorrhiza samples used in this study were collected from wild locations around China. The plants were collected from 50 different geographic regions, including 12 provinces of China, at different altitudes (100-1200 m) from June to September 2015, as shown in Table S3. High-performance liquid chromatography (HPLC)-grade acetonitrile and formic acid were obtained from E. Merck (Darnstadt, Germany). Other chemicals and solvents were of analytical grade. Deionised water was purified using a Milli-Q system (Millipore, Bedford, MA, USA). Tandem mass spectrometry analysis was performed on a Waters Xevo TQ mass spectrometer with an Electro Spray ionization source (Waters, Milford, MA, USA). Masslynx software ver. 4.1 (Waters Corporation, Milford, MA, USA) was used for data acquisition and analysis. The ionization mode was negative for phenolic acids and positive for tanshinones. Using MRM (multiple reaction monitoring), we selected the optimal conditions for the effective compounds. The desolvation gas flow was set at 800 L/h (L/hour) at a temperature of 500 • C, and the cone flow was 50 L/h. For the positive source, the capillary and cone voltages were set at 3.0 kV and 34 V, respectively. For the negative source, the capillary and cone voltages were 2.35 kV and 28 V, respectively. The retention times, precursor-to-product ion pairs, cone voltages and collision energies of the 17 standard compounds in S. miltiorrhiza are listed in Table 1.

Standard Solutions Preparation
To prepare standard stock solution, the 17 reference compounds were accurately weighed and transferred to 10 mL volumetric flasks to which 5% acetonitrile was added to the graduation mark. The solutions were stored at 4 • C prior to use.

Sample Preparation
S. miltiorrhiza samples were prepared according to previously published methods [29,30]. The dried roots were pulverized and sieved through No. 60 mesh (<0.250 mm). Each 0.2-g sample was accurately weighed and extracted in 20 mL of 70% methanol by ultrasonication for 60 min at room temperature, followed by centrifugation at 9000 rpm for 10 min. The supernatant was filtered through a 0.22-µm membrane before injection.

Preparation of Trolox Solution
Trolox was dissolved in methanol to a concentration of 1 mg/mL, and subsequently diluted to different concentrations (0.02, 0.05, 0.09, 0.12, 0.15, 0.20 mg/mL) to establish a standard calibration curve.

DPPH Assay
The antioxidant activities of 50 batches of S. miltiorrhiza samples were evaluated using DPPH free radical scavenging activity assays as previously reported, with some modifications [19,31,32]. A fresh DPPH stock solution was prepared by dissolving 394.32 mg DPPH in methanol to an absorbance of 0.700 ± 0.02 at 517 nm.
To ensure that extracts of all S. miltiorrhiza samples could reduce the stable free radical DPPH to the yellow-coloured DPPH, each sample solution was diluted to six different concentrations (range: 0.1-2.5 mg/mL). Extract solutions (0.1 mL) were mixed with 3 mL of a freshly prepared DPPH solution (0.1 mM in methanol). The mixture was shaken vigorously and kept in the dark at room temperature for 30 min, until stable absorbance values at 517 nm were obtained. The ability to scavenge DPPH radicals was calculated as a percentage according to the following equation: where A1 and A0 are the absorbance values of the sample and blank, respectively. The IC 50 values, or concentrations of samples required to scavenge 50% of the DPPH free radicals, were calculated using a nonlinear regression analysis. Trolox was used as a reference compound for evaluating free radical scavenging activities.

ABTS Assay
The ABTS free radical scavenging activity assay was performed as described previously, with slight modifications [28,33]. In this assay, the ABTS + radical cation is generated by reacting 7 mM ABTS and 2.45 mM potassium persulfate via incubation at room temperature in the dark for 12-16 h. The ABTS solution was diluted with methanol to an absorbance of 0.700 ± 0.020 at 734 nm.
Each sample solution was diluted to six concentrations (range: 0.1-2.5 mg/mL). Subsequently, 100 µL of each plant extract solution were mixed with 4 mL of ABTS solution. The reactive mixture was allowed to stand at room temperature for 6 min in the dark, after which the absorbance was recorded immediately at 734 nm. The percentage of inhibition of absorbance at 734 nm was calculated using Equation (1).

Data Analysis
All statistical analyses were performed using Excel (Microsoft Corp., Redmond, WA, USA) and SPSS 19.0 software packages (SPSS, Inc., Chicago, IL, USA).

Conclusions
In the present study, the chemical compounds and in vitro antioxidant activities were determined simultaneously in 50 batches of wild S. miltiorrhiza samples collected from different locations in China. Twelve phenolic acids and five tanshinones were detected and quantified to provide a comprehensive overview of the chemical composition of wild-grown danshen. Regarding this composition, a PCA indicated that crytotanshinone, tanshinone IIA and dihydrotanshinone I, danshensu, salvianolic acids A and B and lithospermic acid could be optimized as location markers. In addition, the antioxidant activities of plant extracts could be primarily attributed to phenolic compounds. In this study, the different antioxidant activities of danshen correlated with the main chemical compounds and antioxidant capacity relationships in danshen could provide a tool with which to evaluate differences in the internal qualities and antioxidant activities of wild S. miltiorrhiza samples, as well as a sound experimental foundation and model for related studies.