Chemometrics integrated with in silico pharmacology to reveal antioxidative and anti-inflammatory markers of dandelion for its quality control

Identification of phenolic constituents in dandelion

HPLC-DAD-MS/MS was applied to analyze phenolic components in dandelion. Due to the special conjugated system, phenolic compounds have characteristic UV absorbance spectra [26]. Such a property is a core clue to recognize phenolic compounds from others. Together with MS/MS spectra, a total of 22 phenolic components were tentatively identified, including 9 phenylpropionic acids and 13 flavonoids (Fig. 1). Representative UV absorbance spectra and base peak chromatograms are shown in Additional file 1: Fig. S1. Detailed information on the identified peaks, including the retention time, molecular formula, ion type, detected m/z, and fragment ions, are summarized in Additional file 1: Table S1.

Fig. 1figure 1

Chemical structures of 22 phenolic compounds in dandelion

Identification of phenylpropionic acids and their derivatives

Among the nine phenylpropionic acids, seven of them were positively identified as caffeic acid (peak 5), caffeoyltartaric acids (peaks 1 and 7) and caffeoylquinic acids (peaks 3, 4, 11 and 14) by comparing with reference standards. These compounds were generalized by quinic or tartaric acid with several caffeic acids, thus, the predominant fragment ions were produced by the successive loss of caffeic acid and tartaric acid/quinic acid moieties. For example, cichoric acid (peak 7) could lose caffeic acid moiety to generate the fragment ion of m/z 293.0270, and then lose tartaric acid to produce m/z 179.0320; isochlorogenic acid A (peak 11) produced the fragment ions at m/z 353.0836 and m/z 191.0526 by the stepwise loss of caffeic acid and quinic acid moieties (Fig. 2a). In addition, peak 2, showing a precursor ion of m/z 137.0220 and fragment ions of m/z 119.0122, m/z 108.0208 and m/z 92.0255, was putatively identified as p-hydroxybenzoic acid according to the literature [27]. Similarly, Peaks 8 was putatively identified as p-coumaric acid [28].

Fig. 2figure 2

The proposed fragmentation pathway of representative compounds in the negative MS/MS spectrum. a peak 7 (cichoric acid), b peak 11 (isochlorogenic acid A), c peak 9 (luteolin-7-O-β-d-gentiobioside)

Identification of flavonoids

Flavonoids in dandelion had a strong UV absorbance band between 300 and 390 nm, indicating they involved cinnamoyl structure and belonged to flavone or flavonol [26]. Given the MS/MS fragmentation, flavonoids could generate deprotonated ions [M-H]− in negative ion mode and the prominent fragments were produced through C-ring (central three-carbon chain) cleavage, successive losses of substituents and/or glycose moieties. Among the 13 detected flavonoids, five were positively identified as luteolin (peak 18), apigenin (peak 20), luteolin-7-O-β-d-glucoside (peak 12), quercetin (peak 17) and kaempferol (peak 19) by comparing with reference standards. The other 8 compounds were tentatively annotated as the glycosides or derivatives of luteolin, kaemplerol and apigenin. Take peak 9 for an example, it exhibited an [M-H]− ion at m/z 609.1461 and generated a base peak at m/z 285.0374 (Fig. 2c), indicating that peak 9 was a derivative of luteolin. The neutral loss of 324.0987 (C12H20O10) corresponded to a substitute of disaccharide, thus, it was putatively identified as luteolin-7-O-β-d-gentiobioside (Fig. 2c) [29].

Antioxidant activity of dandelion

Two approaches of DPPH and FRAP were applied to evaluate the antioxidant activity of dandelion, which was represented by the of radical scavenging (DPPH) or electron transfer rate (FRAP). Gallic acid with good antioxidant activity was used as positive control drug. As shown in Fig. 3a, dandelion scavenged radicals in a concentration-dependent manner in the range of 62.5–2000 µg/mL, which was a prerequisite to unclose active components through chemical-efficacy relationship. Then the concentration of 500 µg/mL was chosen for the activity evaluation of 31 batches of dandelion. The antioxidative capacity ranged from 1.9 to 71.05% and 28.99–114.10% based on DPPH and FRAP, respectively (Fig. 3d).

Fig. 3figure 3

The bioactivity and chemical profile of dandelion. a the antioxidant activity represented by radical scavenging rates of dandelion at different concentrations and gallic acid (GA, 1.6 µg/mL); b the anti-inflammatory activity represented by LPS-induced NO production of dandelion and dexamethasone (DXMS, 2 µg/mL) and c their cytotoxicity on RAW 264.7 cells at different concentrations; d antioxidant activity of different dandelion samples at 500 µg/mL; e anti-inflammatory activity of different dandelion samples at 100 µg/mL; f the correlation heatmap of chemicals and active capacity of dandelion samples. DE: dandelion extract. Data is expressed as mean ± SEM of three experiments (n = 3) performed in triplicate, one-way ANOVA, ####p < 0.0001, LPS (1 µg/mL) vs. control group; ***p < 0.001, ****p < 0.0001, other groups vs. LPS (1 µg/mL)

Anti-inflammatory activity of dandelion

At first, the cytotoxicity of dandelion sample on RAW 264.7 cells was measured by CCK-8 assay. RAW 264.7 cells treated with dexamethasone were utilized as positive control. As shown in Fig. 3c, the dandelion had no toxicity to RAW 264.7 cells in the range of 12.5–400 µg/mL and significantly promoted cell proliferation in the range of 100–200 µg/mL. The secretion of NO in RAW 264.7 cells was extremely low (0.67 µM) without stimulus, while LPS induced a significant increase of NO level (43.72 µM). When treated with a series of concentration of dandelion (12.5–400 µg/mL), the secretion of NO was dose-dependently reversed and the inhibition rate varied over the range of 6.02–97.83% (Fig. 3b). Then the concentration of 100 µg/mL was applied to evaluate the anti-inflammatory capability of 31 batches of dandelion, which ranged from 22.99 to 60.36% (Fig. 3e). Furthermore, Fig. 3f showed diverse Pearson distances among the chemicals and antioxidant/anti-inflammatory activities of dandelion, which indicated that the potential active components could be discovered by chemicals-efficacy relationship.

Correlations between the phenolic components and bioactivities

GRA was primarily used to assess the correlation of each phenolic component annotated in dandelion with antioxidant and anti-inflammatory activities. The GRD values of 22 phenolic compounds were over 0.9 (Table 2), indicating their high correlation with corresponding activities. The results agreed with the previous view that phenolic compounds are the major components responsible for antioxidant and anti-inflammatory activities [30, 31].

Table 2 The grey relational grade of phytochemical components and bioactivities Screening of antioxidant and anti-inflammatory candidate markers of dandelion

PLSR was utilized to construct the relationship between 22 phenolic compounds and corresponding activities for the screening of efficacy-related markers.

For antioxidant capacity, the parameters of R2Y (cum) & Q2 (cum) in the PLSR models about DPPH and FRAP were 0.904 & 0.753 and 0.861 & 0.656, respectively, indicating a satisfactory explanation and prediction ability. Permutation tests verified the models not overfitted (Additional file 1: Fig. S2). Following the application of PLSR models, the regression coefficients were obtained, which could reflect the positive or negative contribution of each peak to the activity. And higher coefficient values implied the variables more important for the activity. VIP value was another indispensable parameter that was usually employed to screen the important variables contributing to the bioactivity. Overall, variables with coefficient > 0.1 & VIP > 1.0 were signed as candidate bioactive compounds. Due to the different sensitivities, the important components screened by DPPH or FRAP were combined to reduce the false negative results. As shown in Table 3, peaks 1, 3, 5, 7 and 18 were identified to be the potential components responsible for antioxidant activity of dandelion.

Table 3 The potential antioxidant constituents of dandelion

Similarly, the PLSR model about anti-inflammatory activity was also constructed in Additional file 1: Fig. S2 (R2Y (cum), 0.619 & Q2 (cum), 0.507). The same standards were applied to screen out the potential anti-inflammatory components. As shown in Table 4, peaks 8, 9, 13, 17, 18, 19, 20 and 21 were considered to be responsible for the anti-inflammatory activity of dandelion.

Table 4 The potential anti-inflammatory constituents of dandelion Activity evaluations of candidate markers in vitroEvaluation of antioxidant activity

As shown in Fig. 4 and Additional file 1: Table S2, luteolin, cichoric acid, 5-O-caffeoylquinic acid, caffeic acid and caftaric acid presented good antioxidant activity and their IC50 values were 13.15 ± 0.52 µM, 14.17 ± 2.50 µM, 36.76 ± 0.85 µM, 19.63 ± 0.60 µM and 26.53 ± 0.40 µM, respectively. Take Trolox (or FeSO4) as a reference, the antioxidant capacities of the four compounds were signed as 6.46 ± 1.49 mmol Trolox/g (11.01 ± 1.49 mmol FeSO4/g), 3.93 ± 0.49 mmol Trolox/g (9.39 ± 0.87 mmol FeSO4/g), 2.54 ± 0.64 mmol Trolox/g (4.66 ± 1.20 mmol FeSO4/g), 8.25 ± 0.78 mmol Trolox/g (18.70 ± 3.96 mmol FeSO4/g)) and 3.94 ± 0.30 mmol Trolox/g (10.46 ± 1.27 mmol FeSO4/g).

Fig. 4figure 4

The bioactivity evaluation of the potential active compounds. The antioxidant capacity of five potential compounds (a) and gallic acid (b); the anti-inflammatory capacity of four candidate compounds (c–f) and dexamethasone (g)

Evaluation of anti-inflammatory activity

As shown in Table 4, flavonoids were the major constituents that were responsible for the anti-inflammatory activity of dandelion. These constituents possessed the same flavone skeleton with diverse substituents at positions of C3, C6, C7 and C3′ (Fig. 1). To investigate the structure-efficacy relationship of these flavones, four typical compounds with available standards were included to evaluate their anti-inflammatory activity. As shown in Fig. 4, luteolin performed excellent anti-inflammatory activity (IC50, 17.70 µM), which was comparable to that of apigenin (IC50, 15.61 µM) but far superior to kaempferol (IC50, 54.17 µM) and luteolin-7-O-β-d-glucoside (IC50, > 400 µM). Comparing their structural characteristics, it could be deduced that the chromonyl group was the anti-inflammatory pharmacophore.

Additionally, the cytotoxicity of the anti-inflammatory compounds was investigated (Additional file 1: Fig. S3). The four flavones showed no significant toxicity to RAW264.7 cells in the range of 6.25–25 µM. When the concentration was over 25 µM, kaempferol and luteolin-7-O-β-d-glucoside presented a mild dose-dependent toxicity.

In silico pharmacology evaluation of candidate quality markers

To be effective markers, the molecules should reach their targets in the body in sufficient concentration. SwissADME (http://www.swissadme.ch/), was applied to predict the pharmacokinetics parameters (GA and drug-likeness) of these candidate markers. Drug-likeness was evaluated by Lipinski, Ghose, Veber, Ega, and Muegge rules, which were developed by major pharmaceutical companies based on the structural or physicochemical inspections to exclude molecules with properties most probable incompatible with an acceptable pharmacokinetics profile [17]. Generally, when GA was “high” and over two parameters of drug-likeness were “yes”, the compounds were considered to be accessible to the body for functions. As shown in Additional file 1: Table S3, only peaks 18, 20, 19, 5, 8 and 21 conformed to the above conditions. Thus, they were theoretically reasonable to be markers for the quality control of dandelion. Moreover, although the GA of peak 7 was low, it could be detected in vivo [32] due to its high content (> 20 times more than the other constituents). Thus, peak 7 was exceptionally involved in the quality markers.

The targets of the above 7 markers (peak 18, 20, 19, 5, 8, 21, and 7) were predicted based on their structures using SwissTargetPrediction web tool (http://www.swisstargetprediction.ch/). The obtained targets were annotated via DAVID and then a total of 49 targets related to oxidation and inflammation were discovered. The results were visualized in Fig. 5a. The antioxidative markers of caffeic acid (peak 5) and cichoric acid (peak 7) targeted ALOX5, CA3, FYN, MAPK1, APP, PTGS1 and AKR1B1, while the 5 anti-inflammatory markers targeted up to 47 unique genes. As shown in Fig. 5b, the anti-inflammatory markers had high overlapped targets: approximately 70% of the 47 unique genes were targeted by at least 3 markers, which could attribute to their similar structures. These genes mainly functioned in the biological processes of innate/adaptive immunity, inflammatory response, lipid and glucose metabolism, host-virus interaction, and apoptosis, which supported the claimed efficacy of dandelion such as swelling and pain of eye, mammary/pulmonary abscess, scrofula, and urinary infection [2, 7]. The top 20 pathways of KEGG functional enrichment analysis (Fig. 5c) included kinds of carcinomas, VEGF signaling pathway, platinum drug resistance, immune checkpoint pathway, B cell receptor signaling pathway and central carbon metabolism of cancer, etc. indicating the promising potential of dandelion for the treatment of cancers [32,33,34].

Fig. 5figure 5

In silico pharmacology. a the antioxidant and anti-inflammatory targets of candidate markers predicted by SwissTargetPrediction, b Venn plot of flavones with highly overlapped targets, and c KEGG pathway enrichment analysis based on the predicted targets

Application of marker compounds in the quality control of dandelion

As shown in Additional file 1: Fig. S4, 80% methanol, resulting in a better extraction efficiency of phenolic components, was used to prepare samples for quality control. The contents of p-coumaric acid, eupalitin, kaempferol and apigenin always fluctuated around the limit of quantitation, indicating their unsatisfying measurability. Given the excellent bioactivity (Fig. 4) and high overlapped targets (Fig. 5b), it was supposed that luteolin could represent the bioactivity of flavones in dandelion. Therefore, cichoric acid, caffeic acid and luteolin were considered as markers for the quality control of dandelion.

A simultaneous quantitation method was developed based on HPLC-UV. Methodology investigation was performed (Additional file 1: Table S4), showing good linearity with correlation coefficients (r2) higher than 0.9994 in a in a relatively wide concentration range. The contents of three markers in the frequently used dandelion species of TAM and TAO were determined. As shown in Fig. 6a, PCA analysis could obviously distinguish TAM from TAO. The content of cichoric acid and caffeic acid in TAM was significantly higher than that in TAO (Fig. 6d and e). Meanwhile, TAM showed much higher antioxidant activity than TAO (Fig. 6f), further demonstrating that cichoric acid and caffeic acid could be the antioxidant markers of dandelion. That was also supported by the activity-response spectrum of cichoric acid and caffeic acid (Fig. 6g). On the other hand, despite no statistic difference, the content of luteolin and anti-inflammatory activity showed similar tendency between TAM and TAO (Fig. 6e, f and h). Focusing on the effects of geographic regions, the samples belonging to the same species were analyzed. As shown in Fig. 6b and c, the samples from different regions dispersed in the PCA score plots, indicating the quality of dandelion was not susceptible to regions.

Fig. 6figure 6

The application for quality control of dandelion. a PCA score plot of TAM&TAO, b PCA score plot of TAM, c PCA score plot of TAO, d, e distribution of quality markers and f active capacity of dandelion from different species, g the response of cichoric acid and caffeic acid to antioxidant activity, h the response of luteolin to anti-inflammatory activity, i content of quality markers in dandelion

留言 (0)

沒有登入
gif