Comprehensive comparison on antioxidant properties and UPLC-Q-TOF/MS-based metabolomics discrimination between Gentiana veitchiorum and G. szechenyii

2.1 Chemicals

2,2′-Azinobis-(3-ethylbenzthiazoline-6-sulphonate) (ABTS) and gallic acid were procured from Aladdin Industrial Corporation (Shanghai, China). 2,2-Diphenyl-1-picrylhydrazyl (DPPH) was supplied by Macklin Biochemical Co. Ltd. (Shanghai, China). LC-MS-grade acetonitrile and methanol were purchased from Fisher Chemicals (Pittsburg, USA). Methanol of HPLC grade, K₂S₂O₈, FeSO₄·7H₂O, 30 % H₂O₂, and salicylic acid were purchased from Kelong Co. Ltd. (Chengdu, China). Deionized water was purchased from Watsons (Beijing, China). All other reagents used were of analytic grade.

2.2 Herbal materials

A total of 20 batches of Bang Jian samples were collected (including self-collection, hospital collection, herbal market, and company purchase) in different regions of China, including 10 batches of G. veitchiorum (LYZ) and 10 batches of G. szechenyii (DH) (Table 1). Among them, the amount of self-collected samples is about 200 g, and the amount of samples collected by hospitals and purchased by the herbal market and companies is about 500 g. All the samples were authenticated by Prof. Songrong Qin and associate Prof. Zhiwei Zhang from Chongqing Institute of Traditional Chinese Medicine and Prof. Rui Gu from Chengdu University of Traditional Chinese Medicine. And the voucher specimens were deposited at the Institute of Ethnic Medicine, Chengdu University of Traditional Chinese Medicine, Sichuan, China. Prior to experiments, the flowers and flowering branches herbal materials were dried, powdered and passed through 4 mesh sieves. All the sample powders were stored in the dryer at room temperature.

2.3 Sample preparation

Raw G. veitchiorum and G. szechenyii powders (0.5 g) were ultrasonically extracted with 25 mL of methanol at 25 °C for 30 min (300 W, 40 kHz). The samples were added with methanol to compensate for weight lost during extraction (Fan et al., 2023). The supernatants were collected analyzed by UPLC-Q-TOF/MS after filtration by 0.22 μm membrane filter. For the measurement of antioxidant activity, the samples obtained after sonication were sequentially diluted with methanol to 4.0, 2.0, 1.0, 0.5, 0.25 and 0.125 mg/mL. All the samples were collected and stored at 4 °C until further analysis.

2.4 Determination of Antioxidant activity

2.5 UPLC-Q-TOF/MS metabolomics data acquisition

A quality control (QC) sample was prepared by mixing all tested samples (20 batches) in equal volumes and used to evaluate the system stability. The QC sample was repeated four times initially and during the sampling process every four tested samples.

All the samples were analyzed using SYNAPT G2 HDMS QTOF-MS (Waters Corporation, Milford, USA) and Water ACQUITY FTN UPLC system via MassLynx (version 4.1, Waters Corporation, USA). A flow rate of 0.2 mL/min was used to inject 8 μL sample solution into ACQUITY UPLC CSH C₁₈ column (2.1 mm × 100 mm, 1.7 μm) at 35°C. The mobile phase composition was 0.1 % formic acid solution in water (solvent A) and acetonitrile solution (solvent B). Each sample was eluted in 51 min at flow rate of 0.2 mL/min. The gradient elution was designed as follows: 0–7.5 min, 95–82 % A; 7.5–10.5 min, 82–80 % A; 10.5–15 min, 80–75 % A; 15–19.5 min, 75–71 % A; 19.5–27 min, 71–63 % A; 27–30 min, 63–60 % A; 30–34.5 min, 60–55 % A; 34.5–36 min, 55–52 % A; 36–40.5 min, 52–40 % A; 40.5–43.5 min, 40–30 % A; 43.5–45.75 min, 30–5% A; 45.75–46.5 min, 5–95 % A; 46.5–51 min, 95 % A. QTOF was operated in negative ion mode. The capillary and sample cone voltages were set to 2.0 kV and 40.0 V for negative ion mode. MS data were acquired using Continuum MSE model. The survey scan time was 0.3 s, and TOF mass range was 50–1200 Da. The precursors were fragmented using 15–35 eV for MS/MS evaluations with scan time of 1.0 s. A lockedspray infusion at the flow rate of 7 μL/min and locked mass of leucine enkephalin at the concentration of 0.2 ng/mL were employed to monitor the negative ion patterns (m/z 554.2615, [M − H]⁻) and to record accurate masses.

2.6 Compounds identification

Masslynx software (Waters Corporation, USA) was used to gather the raw data files from UPLC-Q-TOF/MS analysis. These files were imported into Progenesis QI 2.3 software (Waters Corporation, USA) for peak comparison, normalization, peak selection, deconvolution, compound identification, and statistical analysis. The compound ions not found in certain samples had default values of zero. Then, A self-established database of reported Gentiana compounds was established by consulting relevant literature and websites (CNKI, Massbank, ChemSpider, Web of Science, PubMed, SciFinder, etc). The databases were imported into the UNIFI platform attached to Waters ACQUITY UPLC and SYNAPT G2 HDMS systems for matching of target compounds. Total of 380 compounds were included in database. The compounds were firstly identified through reference substances database, while compounds without controls were referred to the aforementioned compounds database and published literatures with the following parameters: mass error was within 5 ppm, retention time, accurate m/z, and MS/MS data. Data from MassBank’s online database and literature references were employed as the criteria to make identifications of compounds.

2.7 Statistical analysis

The Progenesis QI 2.3 software processed data were exported to EZinfo 3.0, SIMCA 14.1, and Metaboanalyst website for Chemometrics statistical analysis, including principal component analysis (PCA), orthogonal partial least squares discriminant analysis (PLS-DA) and hierarchical cluster analysis (HCA) in Auto scaling mode. Screening for differential metabolites was based on univariate analysis, including fold change analysis, T-tests, and volcano plot. Data were analyzed using GraphPad Prism 6.0 software. All data are presented as mean ± stand deviation (SD), and each experiment was performed at least three times. IC₅₀ data were analyzed using Microsoft Excel and the IC₅₀ values were calculated using Graphpad Prism by converting the concentrations to their logarithmic value and then selecting non-linear inhibitor regression equation (log inhibitor) vs normalized response-variable slope equation. The correlation between the identified differential metabolites and the IC₅₀ values of antioxidant properties was further analyzed using correlation heatmaps in Metaboanalyst website. Significant differences between groups were determined using one-way ANOVA with Dunnet’s multiple comparisons test or unpaired t-test. P < 0.05 was considered difference significantly.

3.1 Antioxidant activities between G. veitchiorum and G. szechenyii

In order to assess the free radical scavenging capacity of G. veitchiorum and G. szechenyii, at least two different methods should generally be selected for determination. The antioxidant capacity of each extract cannot be determined by just one single method because each method uses different processes and principles. It is very necessary to use more than one type of method that contains more than one antioxidant approach. In the present study, DPPH, ABTS, and Hydroxyl radical have been established, utilizing different mechanisms to assess the antioxidant activity of the two species.

The results of the antioxidant activity of extracts of G. veitchiorum and G. szechenyii are shown in Fig. 2 and Table 2. In the DPPH assay, ascorbic acid showed an IC₅₀ of 0.039 ± 0.002 mg/mL. The IC₅₀ value of G. szechenyii was lower than that of G. veitchiorum, indicating that G. szechenyii exhibits more antioxidant activity. In addition, the DPPH radical scavenging rate gradually increased with increasing concentration until the concentration reached 1.0 mg/mL, the DPPH radical scavenging rate of the samples from G. veitchiorum and G. szechenyii was not significantly different from that at the same concentrations of Vc, and both of them could reach more than 95 %. The maximum removal rates of G. szechenyii and G. veitchiorum could reach 96.45 % and 94.21 %, respectively. However, from 0.125 mg/mL to 0.5 mg/mL, DPPH radical scavenging of G. veitchiorum was significantly lower than that of the G. szechenyii, and there was significant statistical difference (P < 0.05) (see Table 3).

Close

Fig. 2

The antioxidant activity of DH and LYZ with different concentration by DPPH, Hydroxyl, and ABTS method. DPPH radical scavenging activity (A), Hydroxyl radical scavenging activity (B), ABTS radical scavenging activity (C). *LYZ, DH, and Vc represented G. veitchiorum, G. szechenyii and positive control, respectively.

Export to PPT

The advantage of ABTS over DPPH is that it is not affected by pH, and the method is applicable to antioxidants of lipophilic or hydrophilic nature. As can be seen in Fig. 2B, both species showed some inhibition of ABTS free radicals. Similarly with the increment of sample concentration, the free radical scavenging rate of both the samples gradually increased. When the sample concentration was 4.0 mg/mL, the free radical scavenging rate of G. veitchiorum and G. szechenyii were 71.70 % and 77.29 % respectively. Moreover, the scavenging rate of G. szechenyii was significantly higher than that of G. veitchiorum at all concentrations except 0.125 mg/mL. Further, it can also be seen from Table 2 that the average IC₅₀ values of G. szechenyii and G. veitchiorum were 0.80 and 1.25 mg/mL, respectively. Obviously, G. szechenyii is superior. Generally, the DPPH assay depends on both Single electron transfer (SET) and Hydrogen atom transfer (HAT) while ABTS depends only on SET (Thu et al., 2019). Consequently, the ABTS assay may be considered confirmatory in comparison to the DPPH assay.

Hydroxyl radical scavenging capacity is an important indicator of the antioxidant capacity of a substance. It mainly reflects the ability of the sample to scavenge Hydroxyl radicals by measuring the decrease in absorbance of o-diazophene, a sample inhibitor of color development·H₂O₂/Fe²⁺ generates Hydroxyl radicals through Fenton reaction, which oxidizes Fe²⁺ to Fe³⁺ in o-diazofibe-Fe²⁺ aqueous solution, resulting in a decrease in absorbance at 536 nm. The degree of decrease in absorbance at 536 nm by the sample reflects the ability of the sample to scavenge Hydroxyl radicals. From Table 2, it can also be seen that the IC₅₀ value of G. veitchiorum is nearly twice that of G. szechenyii. In addition, it can also be seen from Fig. 2C that the scavenging rate of Hydroxyl radicals was dose-dependent with the increase of sample concentration. The maximum 78.62 % inhibition was reached at 4.0 mg/mL of G. szechenyii. The inhibition rate of G. szechenyii at different concentrations was significantly higher than that of G. veitchiorum, which was similar to the results of the previous assay.

3.2 Metabolic profiles of G. veitchiorum and G. szechenyii

A UPLC-Q-TOF/MS method was developed and successfully utilized for the characterization and identification of the chemical constituents of G. veitchiorum and G. szechenyii. With the established method, all the samples of both G. veitchiorum (10 batches) and G. szechenyii (10 batches) were chromatographically analyzed. The base peak chromatograms (BPI) of all the samples were illustrated in Fig. 3A and B. G. veitchiorum and G. szechenyii chemically differed and thus had different metabolite levels. G. szechenyii displayed more discernible peaks compared to G. veitchiorum.

Close

Fig. 3

Chemometric analysis for species discrimination of LYZ and DH. (A and B) The UPLC-Q-TOF/MS chromatograms of LYZ (LYZ1-10) and DH (DH1-10), respectively; (C) LYZ and DH were discriminated by PCA on negative HRMS spectra (D) the PLS-DA analysis of LYZ and DH; (E and F) the 100 X permutation tests in negative ion mode.

Export to PPT

QC sample is frequently employed to assess the stability and reproducibility of the UPLC-Q-TOF/MS method (Li et al., 2023). Besides, the equilibration of the system was further assessed by the dispersion of QC samples distributed in the figure of the principal component analysis (PCA). The PCA plot (Fig. 3C) showed that all the QC samples were found to tightly cluster together and closely to coordinate point of origin indicating the reproducibility of the analytical system. In order to analyze the differences in chemical composition between the LYZ and DH, an untargeted metabolomics analysis of the two types was performed (Fig. 3C, D, E, F). The clustering heat map of the metabolites also clearly showed the differences between LYZ and DH as well as the similarities between the individual varieties.

Unsupervised PCA was performed after pre-processing negative modes metabolomic data for attaining preliminary understanding of metabolic variations in samples and degree of variations between the samples within groups. As illustrated in the PCA score plot, the results of indicated that the first two principal components were extracted as 55.4 % and 8.0 %, respectively (Fig. 3C). It was found that samples from the same group clustered while of different groups were separated. G. veitchiorum and G. szechenyii were distributed in different quadrants on left and right sides of y-axis. There were significant differences between G. veitchiorum and G. szechenyii.

PLS-DA is often used as a supervised pattern recognition method to correlate the biological activities of medicinal plants with spectroscopic data. PLS-DA analysis was made to maximize the distinction between different species and screen differential compounds between G. veitchiorum and G. szechenyii. As illustrated in Fig. 3D, The PLS-DA results showed that component 1 and component 2 explained the 58.1 % and 8.0 % change, respectively. The Q² and R² values were used to assess the modeling and predictive abilities of the PLS-DA model, respectively. The values of Q² and R²Y are closer to 1, the more reliable and stable the model is. The cross- validation [R²Y (cum) = 0.938, Q² (cum) = 0.905] also suggested the good predictive capability and the significant explanation for effective species discrimination of G. veitchiorum and G. szechenyii. Moreover, through the permutation test it can be seen that the values of Q² and R² are both smaller than the original values, Q² intersects and is negative with the Y-axis, which indicates that the model has an overfitting phenomenon, and the PLS-DA model is more reliable and can be used to screen differential metabolites.

As shown in Fig. 4A, the result of the heatmap of species correlation indicated a significantly low inter-species correlation with a strong intra-species correlation between these two species. In addition, the grouping trend was confirmed by HCA, which could be seen from the well-clustered groups of G. veitchiorum and G. szechenyii (Fig. 4D). All these results further suggested the significant difference in chemical compositions between G. veitchiorum and G. szechenyii.

Close

Fig. 4

Pattern recognition for visualization of species discrepancy of LYZ and DH. (A) the heatmap of visualization for the discrepancy between LYZ and DH; (B) the HCA discrimination of LYZ and DH.

Export to PPT

3.3 Compounds identification of G. Veitchiorum and G. Szechenyii metabolites

Representative BPI (Base peak chromatograms) for G. veitchiorum and G. szechenyii were shown in Fig. 5A and B, respectively. Obviously, it could be seen that G. veitchiorum and G. szechenyii have different chemical profiles and therefore have variability in their chemical composition. The compounds were further identified by processing the raw data collected by Waters mass spectrometry using UNIFI software and comparing the databases (including the self-constructed library and Commercial library Waters Traditional Medicine). The compounds were identified based on their accurate mass, retention time, and mass spectrometry (MS) fragmentation patterns, with a mass error of 5 ppm as the screening criterion. In addition, on the basis of the previous research of our research group, the fragment ions of the mixed Reference substances were compared after the determination of UPLC-Q-TOF/MS to identify the compounds (Fan et al., 2023). Finally, a total of 56 compounds were identified from the two species, as shown in Table 2. Among them, 48 were identified from the G. szechenyii and 43 from G. veitchiorum. There were 35 components common to both species, 13 components unique to G. szechenyii and 8 components unique to G. veitchiorum. The discovered compounds could be classified into 10 categories based on their chemical classification attribution data, which included mainly flavonoids, iridoids, triterpenoids, saccharides. As previously reported in the literature, iridoids constituents and flavonoid constituents are the main components of G. szechenyii and G. veitchiorum (Bao et al., 2023; Fu et al., 2018; Li, 2017; Li et al., 2016; Zou et al., 2010). Similar to the results of previous studies, and as can be seen from Fig. 5C and D, the compounds identified from G. szechenyii and G. veitchiorum were similar in type, but the relative contents of the major components including flavonoids, iridoids, and triterpenoids varied considerably. Specifically, G. veitchiorum has a higher proportion of flavonoids and triterpenoids, while G. szechenyii has a higher proportion of iridoids. The reason for this result may be related to the relatively high content of characteristic components specific to each variety.

Close

Fig. 5

The base peak chromatograms (BPI) of the representative samples of LYZ-1 (A) and DH-1 (B); the compositional features in LYZ (C) and DH (D).

Export to PPT

3.4 Screening of different metabolites between G. Veitchiorum and G. Szechenyii

Compounds consistent with VIP > 1 were selected based on the identified compounds of G. szechenyii and G. veitchiorum selected by multifactorial analysis of VIP values based on the PLS-DA model. A total of 56 compounds of G. szechenyii and G. veitchiorum were filtered according to VIP > 1, FC ≥ 2 or ≤ 0.5, P < 0.05, of which 24 potential differential compounds, included 11 up-regulated and 13 down-regulated. These differentially metabolized substances mainly contain 10 flavonoids, 9 iridoids, 2 triterpenoids and three other compounds. Among them, 6′-O-β-D-glucopyranosyl gentiopicroside, gentiopicrin, swertiamain, kaempherol 3-O-glucoside were present only in G. veitchiorum. Syringic acid, sweroside, naringenin, gentizechenlioside A, szechenyin A, isomangiferin, depressine, and 1-Hydroxy-3,4,5-trimethoxy-9H-xanthen-9-one were present only in G. szechenyii. The remaining 12 components were common to both species and differed only in content. The relative contents of the specific 24 differential metabolite components are shown in Fig. 6E.

Close

Fig. 6

Screen for potential marker compounds for authentication of LYZ and DH. Potential marker compounds ranked by VIP values (VIP > 1) on negative ion HRMS spectra (A); Volcano plot combined results from Fold Change (FC) Analysis and T-tests, FC ≥ 2 or ≤ 0.5, significant changes: P < 0.05 on negative ion spectra (B); Heatmap of samples and markers ranked by unpaired t-test value on negative ion HRMS spectra respectively (C and D). Boxplot of the characteristic distribution of potential markers of components in LYZ and DH (E). Data are presented as Mean S.D. (n = 10).

Export to PPT

The heap map revealed that samples could be divided into two groups having big differences between two species. Correlation analyses between varieties and heat maps of the identified differential metabolites can be seen in Fig. 6C and D. Heat maps between varieties based on the differential metabolites screened, with each square representing the relative content of the component. Compared to the heat map comparison of chemical profiles, the color difference between the two varieties is greater, indicating a more pronounced difference in relative content (Fig. 6C). The clustering heatmaps of the differential metabolites also clearly showed the similarity between the biological replicates and the differences among the two species of mainstream of Bang Jian (Fig. 6D).

In order to further verify the feasibility and accuracy of the 24 differential metabolites for the identification of two species of G. szechenyii and G. veitchiorum. Therefore, the relative contents of the above 24 differential metabolites were normalized to establish a new chemometrics model. In addition, as shown in Fig. 7, PLS-DA, HCA analyses of the 24 differential metabolites screened above showed that the two species could be better distinguished. Moreover, the model showed relatively satisfactory results by replacement test (R²X = 0.7, R²Y = 0.991, Q² = 0.988).

Close

Fig. 7

Accuracy evaluation of potential marker compounds. (A) LYZ and DH were well discriminated at specie level by PLS-DA using 24 optimized potential makers; (B) LYZ and DH were well separated by HCA using 24 optimized potential makers.

Export to PPT

3.5 Relationship between different compounds and antioxidant activity

The correlation between the different components and the two species was analyzed to better understand the effect of different metabolites on the antioxidant activity of G. veitchiorum and G. szechenyii. Compounds with Pearson's correlation coefficient r > 0.5 and P < 0.05 were selected to have an effect on antioxidant activity.

As shown in Fig. 8, the 11 differential metabolites were negatively correlated with their IC₅₀ values (positively correlated with antioxidant activity), in which the 11 compounds were mainly flavonoids and iridoids, with 90.83 % of both constituents in G. szechenyii, and a total of 67.85 % in G. veitchiorum. Further studies on the antioxidant activity of the two varieties were carried out to explore the role of compounds in antioxidant activity. Based on the Pearson's correlation coefficient of r > 0.5 and P < 0.05 conditions, depressine (r = 0.73), szechenyin A (r = 0.66), syringic acid (r = 0.76), isoorientin-2′'-O-glucopyranoside (r = 0.76), isoscoparin 2″-O-glucoside (r = 0.74) and naringenin (r = 0.70) were screened as the most relevant compounds with antioxidant activity.

Close

Fig. 8

Correlation analysis of IC₅₀ values of DPPH, ABTS, Hydroxyl, and differential markers. (A) Each ellipse represents the correlation between the two attributes, and different colors represent the sizes of the correlation coefficients between the attributes. (B) Histograms of correlation between antioxidant activity (IC₅₀ values of DPPH, ABTS, Hydroxyl) and 24 differential metabolites.

Export to PPT

As can be seen from Fig. 6E, the relative contents of all compounds were higher in G. szechenyii. Among them, depressine, syringic acid, naringenin and szechenyin A were only found in G. szechenyii, among them, depressine and szechenyin A were the characteristic components of G. szechenyii. Isoscoparin 2′'-O-glucoside and isoorientin-2′'-O-glucopyranoside, two flavonoids, were found in both varieties, but their contents in G. szechenyii were significantly higher than those in G. veitchiorum. It is possible that G. veitchiorum has some antioxidant properties mainly attributable to two flavonoids components. Therefore, the antioxidant activity of G. szechenyii should be higher than that of G. veitchiorum, which is basically consistent with the results of previous antioxidant activity experiments.

Meanwhile, the results of previous studies indicated that both flavonoids and iridoids are excellent natural antioxidants (Náthia-Neves et al., 2017; Tenuta et al., 2020; Wang et al., 2022). Both iridoids, depressine and szecheyin A, have some antioxidant activity and are structurally iridoids containing a benzoyl fragment, which may be related to the phenolic hydroxyl group in the structure and the ternary oxygen ring structure (Zhang et al., 2022a). Isoscoparin 2′'-O-glucoside and isoorientin-2′'-O- glucopyranoside are both flavonoids (Jung et al., 2013; Kim et al., 2018). Flavonoids are known to have strong antioxidant activity, protecting unsaturated fatty acids in cell membranes from oxidation (Li et al., 2016; Tadzhibaev et al., 1992; Yang et al., 2014). Besides, Naringenin also is a natural flavonoid showing strong anti-inflammatory and antioxidant activity, a natural antioxidant, free radical scavenger, anti-inflammatory, promoter of carbohydrate metabolism and immune system inhibitor (Wang et al., 2020a; Zhang et al., 2020; Zhang et al., 2022b). Naringenin was able to inhibit CYP19, CYP2C9, and CYP2C19 with an IC₅₀ value of less than 5 μM, and no inhibition of CYP2B6 or CYP2B6 was observed at concentrations of up to 10 μM. Significant inhibition of CYP2B6 or CYP2D6 was observed (Yen et al., 2015). It has been reported that the antioxidant activity of flavonoids is closely related to the number and location of hydroxyl groups. The structure contains 3–4 phenol hydroxyl substitutions, and the structure of 3 ', 4 '-catechol hydroxyl or methoxy group and hydroxyl group is present in the B ring (Heim et al., 2002). Similarly, in addition to these different metabolites, other flavonoids contained in G. veitchiorum and G. szechenyii have good antioxidant properties. Isoorientin, a carbonylated flavonoid, is a plant phenol produced by secondary metabolism of plants and is found in a variety of medicinal plants. It can directly scavenge free radicals and reduce metal ions, and its antioxidant mechanism mainly involves HAT and SET (Lefahal et al., 2022; Ziqubu et al., 2020). Kaempferol is a class of flavanols extracted from traditional Chinese medicine, which has various functions such as antioxidant, anticancer, diabetes prevention and protection of damaged cells. It has been reported that kaempferol has strong antioxidant activity and can inhibit the hemolysis of human erythrocytes induced by active oxygen by scavenging reactive oxygen species and protecting antioxidant enzyme activity (Shahbaz et al., 2023; Sharma et al., 2021).

In addition, syringic acid belongs to the phenolic acid group of compounds with strong antioxidant properties. Although it accounts for a small percentage of the two species, some studies have shown that syringic acid has the ability to scavenge free radicals and inhibit lipid peroxidation, which can protect cells from oxidative damage (Li et al., 2019; Ramorobi et al., 2022; Shimsa et al., 2023). It has been extensively studied for use in cancer, cardiovascular disease, and neurodegenerative diseases (Lozanova et al., 2022; Mihanfar et al., 2021; Ogut et al., 2022; Pei et al., 2021). Therefore, the results suggest that the differences in antioxidant activity between the two varieties of G. veitchiorum and G. szechenyii may be due to the differences in the contents of the six components in the two varieties. Overall, the observed difference in antioxidant activity may be attributed to the relative content difference between the differential metabolites, or it may be the result of synergistic action of these flavonoids, iridoids, phenolic acids, etc., resulting in the difference in antioxidant activity of the two varieties.