UPLC‒MS characterization of chemical constituents from WDD

The high-resolution MS data of WDD were promptly collected using the UPLC-Q-TOF-MSE technique. The base peak intensity (BPI) chromatograms of WDD in positive and negative ion modes are shown in Fig. 1. The BPI chromatograms of 6 single herbs of WDD are shown in Additional file 1: Figs. S1-S6.

Fig. 1

Bask peak ion chromatograms of WDD in negative and positive ESI modes

Fragmentation pattern study of WDD

The MS data were processed and analysed by UNIFI screening technology, and the fragment information was then automatically matched. Following further verification, a total of 107 compounds were confirmed in WDD, including 54 flavonoids, 14 triterpenoids, 10 organic acids, 7 alkaloids, 7 coumarins, and 15 other types. The comprehensive MS information of these compounds is summarized in Table 3. Furthermore, chemical structures were validated using precise mass, MSE data, and relevant literature.

Table 3 Chromatographic and mass spectral data of the 107 compounds analysed by UPLC-QTOF-MS/MSIdentification of flavonoids

Flavonoids and their glycosides are abundant in WDD and are widely present in plant material. In this study, the matching of mass spectral data with the UNIFI analytical platform authenticated a total of 54 flavonoids, including 16 flavones, 7 flavonols, 26 flavanones, 1 isoflavanone, 3 chalcones and 1 flavanol.

The glycosidic linkages joined by oxygen atoms in flavonoid glycosides may be cleaved in both positive and negative ion modes, with the majority of them characterized by neutral losses, for example, 162 Da (Glc), 146 Da (Rha) and 132 Da (Api) [33]. The cross-ring cleavages of flavone C-glycosides of saccharidic residues produced the primary product ions. Therefore, it was simple to lose C2H4O2 (60 Da), C3H6O3 (90 Da), and C4H8O4 (120 Da) groups from the precursor ions [7]. Compound 60 revealed a quasimolecular [M + H]+ ion at m/z 611.1879 (C28H34O15). The fragment ions at m/z 465.1396 ([M + H-Rha]+), 449.1448 ([M + H-Rha-O]+), 431.1339 ([M + H-Rha-O-H2O]+), 303.0868 ([M + H-Rha-Glc]+), 285.0761 ([M + H-Rha-Glc-H2O]+) were produced by removing a molecule of rhamnose and a molecule of glucose, respectively. Therefore, it was recognized as neohesperidin by analysing the reference substance, and its fragmentation behaviour is shown in Additional file 1: Fig. S7. Compound 24 exhibited a parent ion [M-H]− at m/z 623.1612, which generated fragmentation ions at m/z 533.1279 ([M-H-C3H6O3]−), 503.1146 ([M-H-C4H8O4]−), 443.1023 ([M-H-2C3H6O3]−), 413.0898 ([M-H-C4H8O4-C3H6O3]−), and 383.0763 ([M-H-C4H8O4-C3H6O3-CO]−) by the elimination of the CO moiety. These results indicated that the fragmentation patterns were comparable to those of diosmetin 6,8-di-C-glucoside. The mass spectrum and possible fragmentation pathways of diosmetin 6,8-di-C-glucoside in negative ion mode are shown in Additional file 1: Fig. S8.

It is widely known that the RDA fragmentation processes as well as losses of small molecules and radicals, including CH3, CO, and CO2, are the primary MS behaviours of flavone aglycones. Furthermore, neutral loss of CH4 (16 Da) was produced in the presence of an ortho-methoxyl substituent group. Compound 104 displayed a [M + H]+ ion at m/z 433.1489 (C22H24O9) with diagnostic ions at m/z 417.1170 and 403.1018 via the loss of 16.0313 Da (CH4) and 30.0470 Da (2CH3), respectively. Moreover, fragment ions at m/z 388.0773 ([M + H-3CH3]+), 385.0914 ([M + H-2CH3-H2O]+), 373.0511 ([M + H-2CH3-CO]+), and 360.0847 ([M + H-3CH3-CO]+) were also observed. Thus, this compound was identified as 3,5,6,7,8,3′,4′-heptamethoxyflavone, and fragmentation patterns are shown in Additional file 1: Fig. S9.

Identification of triterpenoids

A total of 14 triterpenoids were identified in WDD, including 8 pentacyclic triterpenoids from GRR and 6 limonoids from AFI and CRP.

Triterpenoid saponins in GRR were made up of oleanane-type triterpene sapogenins and saccharide groups via the hydroxyl group at the C-3 position, with glucose and glucuronic acid being the most frequent saccharides. As a result, the saccharide groups in the structures were identified using the neutral losses of the sugar moiety. Compound 96 was positively identified as glycyrrhizic acid using a reference standard. The MS fragmentation pattern of glycyrrhizic acid was studied in depth to contribute to the characterization of these pentacyclic triterpenoids (Additional file 1: Fig. S10). Glycyrrhizic acid gave an [M + H]+ ion at m/z 823.4116, along with five major fragment ions at m/z 647.3798 ([M + H-GluA]+), 471.3476 ([M + H-2GluA]+), 453.3371 ([M + H-2GluA-H2O]+), 435.3252 ([M + H-2GluA-2H2O]+), and 407.3302 ([M + H-2GluA-2H2O-CO]+) observed in the high-energy MSE spectra.

Limonoids were created by removing four terminal carbons from the side chain of an apotirucallane or apoeuphane skeleton and then cyclizing them to generate the 17-furan ring [7]. Additionally, some compounds lost complicated groups, such as CO, CO2, H2, and others. Compound 73 gave a hydrogenated molecule [M + H]+ at m/z 455.2051 and produced predominant fragment ions at m/z 437.1988 ([M + H-H2O]+), 411.2189 ([M + H-CO2]+), 409.2007 ([M + H–CO-H2O]+), 393.2066 ([M + H-CO2- H2O]+), 391.1943 ([M + H–CO-2H2O]+), and 349.1822 ([M + H–CO-2H2O- C3H6]+) in positive ion mode. It was identified as obacunone (Additional file 1: Fig. S11).

Identification of organic acids

A total of 10 organic acids were found in positive ion mode and originated from five of six Chinese medicines, except BCT. Organic acids, which are mainly extracted with polar solvents, are the abundant chemical constituents of Pinellia rhizoma species [34]. [M-CH3]+, [M-H2O]+ and [M-HCOOH]+ in the mass spectra of organic acids showed the presence of a polyhydroxy molecule comprising carboxylic acid groups. Compound 53 produced an [M + H]+ ion at m/z 154.0210 with the chemical Formula C7H6O4. The main ions emerged at 111.0438 ([M + H-CO2]+) and 93.0710 ([M + H-CO2-H2O]+), which corresponds to the usual structure of organic acids. Thus, Compound 53 was considered to be 3,4-dihydroxybenzoic acid (Additional file 1: Fig. S12).

Identification of alkaloids

Seven alkaloids in WDD were derived from five of six Chinese medicines, except ZRR. According to the literature, the principal fragment patterns of alkaloids are neutral losses, such as methyl radicals, hydrogen radicals, and CO, caused by the serial cleavage of substituted methoxyl or methylenedioxyl groups on the A- and D-rings [35]. Component 3 exhibited a quasimolecular ion at m/z 168.1018 (C9H13NO2). The fragment of synephrine was suggested by the diagnostic ions at m/z 150.0901 ([M + H-H2O]+), 134.0593 ([M + H-H2O-CH4]+), 119.0499 ([M + H-H2O- NH2CH3]+), and 107.0489 ([M + H-H2O-CHNHCH3]+) in the spectrum (Additional file 1: Fig. S13).

Identification of coumarins

Seven coumarins were identified in WDD, mostly from AFI and GRR. The basic fragmentation mechanism of coumarins involves the losses of OH, CH3, CO and CO2 [36]. The precursor ion [M + H]+ at m/z 261.1112 (C15H16O4), which was recognized as Compound 46, was confirmed to be meranzin by the fragment ions at m/z 243.0992 ([M + H-H2O]+), 189.0536 ([M + H-C4H8O]+), 131.0483 ([M + H-C4H8O-CO-CH2O]+), 103.0537 ([M + H-C4H8O-CH2O-2CO]+) (Additional file 1: Fig. S14).

Identification of other types

This group includes certain chemicals with fewer species and lower concentrations. The mass spectra data detected from the MassLynx workstation were compared with UNIFI software, and the results were confirmed by a literature study. A total of 15 compounds were deduced. Component 103 displayed a molecular ion [M + H]+ at m/z 295.0240 (C17H26O4). The characteristic ions at m/z 277.1777 ([M + H-H2O]+), 259.1012 ([M + H-2H2O]+), 179.0635 ([C10H11O3]+), 177.1246 ([C11H13O2]+), 137.0593 ([C8H9O2]+) were generated. Therefore, the component was unequivocally identified as 6-gingerol with the reference material (Additional file 1: Fig. S15).

Multivariate statistical analysis

In terms of QTOF-MS, it was discovered that positive ion mode may provide more sensitive and accurate mass spectra in terms of UPLC-QTOF-MS. Furthermore, the positive ion mode made it simpler to validate molecular ions in the identification of each signal. As a result, the UPLC-QTOF-MS data for the multivariate statistical analysis were collected in positive ion mode.

Few technical and analytical mistakes in UPLC-QTOF-MS-based metabolomics can prevent interference with multivariate statistical analysis to generate trustworthy and high-quality results. The QC sample was also evaluated in tandem with the WDD samples to ensure system stability. The data quality was assessed by comparing all QC samples.

Samples of the WDD material reference and its commercial formulations were imported into progenesis QI for principal component analysis (PCA). The scatterplot of PCA is shown in Fig. 2. QC samples are spread around the origin and are closely aggregated. The findings suggest that the substantial differences identified by multivariate statistical analysis between material reference and commercial preparation were more likely to be the consequence of composition changes rather than artefacts resulting from technical faults. R2 (cum) is a popular metric for evaluating the quality of a PCA model, with values near 1.0 indicating strong fitness and predictive performance. In this study, R2X (cum) is 0.9, showing that the developed PCA model has acceptable fitness and prediction. The 23 samples were almost evenly split into two groups, indicating a difference in quality between the WDD material reference and its commercial preparations.

Fig. 2

PCA score plot of material reference and commercial preparation

Modern pharmacological studies have shown that WDD mainly has lipid-lowering, anti-inflammatory, anti-schizophrenia, protection against cell damage and other effects. Eleven bioactive ingredients with major pharmacological effects were screened from WDD. Liquiritigenin, liquiritin, glycyrrhizic acid, naringin and tangeretin have anti-inflammatory effects by decreasing the synthesis of IL-6, TNF-α and VEGF [37,38,39,40,41,42]. Eriocitrin and liquiritin regulate the expression of Nrf2 and NF-κB, thereby downregulating inflammation and oxidative stress [38, 43]. Synephrine, liquiritigenin, hesperidin, neohesperidin and tangeretin possess lipid-lowering effects by regulating the insulin receptor (IR) and suppressing adipocyte differentiation [44,45,46,47,48]. Liquiritigenin has the ability to cure immunological dysfunction by increasing cAMP synthesis in several cell lines and controlling immune cell death [40, 49]. Adenosine alleviated amyloid β-protein25-35-induced brain damage by preventing apoptosis and oxidative stress [50]. Naringenin may have anti-coronavirus disease 2019 (COVID-19) effects by inhibiting the COVID-19 major protease 3-chymotrypsin-like protease (3CLpro) and decreasing angiotensin converting enzyme receptor activation [51].

The contents of 11 components in the WDD material reference and its commercial prescriptions were normalized by a heatmap (Fig. 3). Clearly, WDD samples in the material reference gathered into one cluster according to the hierarchical cluster analysis. There was no clear aggregation trend among the commercial WDD, but there was an obvious correlation with the material reference, that is, CP8. Overall, among the 11 components analysed, the contents of neohesperidin and naringin in the WDD material reference were higher than those in its commercial formulas, adenosine and synephrine in the WDD material reference were lower than those in its commercial formulas, while the contents of the other components were generally consistent. Generally, the differences between the WDD material reference and its commercial prescriptions may be caused by the differences in the original plant, doses, processing methods of Chinese medicines and extraction methods. Therefore, the plant origin, dose, processing method of medicines and extraction process should be regulated to obtain WDD preparations with consistent quality in each batch.

Fig. 3

Heatmap of the 11 chemical constituents of WDD prepared in the laboratory and its 8 commercial preparations. MR1-MR15 were 15 batches of WDD material reference. CP1-CP8 are commercial preparations of WDD

Quantitative analysis of 11 components in WDD

According to this qualitative study, the primary bioactive constituents in WDD include flavonoids, triterpenoids, alkaloids, and phenolics. Then, naringin, hesperidin, neohesperidin, liquiritin, glycyrrhizic acid, adenosine, liquiritigenin, tangeretin, eriocitrin, naringenin and synephrine were quantified. The chromatograms of 4 compounds in positive ion MRM mode and 7 compounds in negative ion MRM mode are shown in Additional file 1: Figs. S16 and S17.

Methodological verification

The calibration curve was constructed by diluting the sample solution into a specific gradient. Six concentration levels of standard stock solutions were diluted. The correlation coefficients (r) of the 11 components ranged from 0.9925 to 0.9994, indicating that the calibration curves were trustworthy for quantitative analysis. The limit of quantification (LOQ) for each component was determined by injecting a series of dilute solutions of known concentration at a signal-to-noise ratio (S/N) of 10. The LOQs ranged from 0.001 to 450.000 ng/mL (Additional file 1: Table S2).

The interday fluctuations, which were used to assess the precision of the devised approach, were studied by detecting the 11 analytes in 6 duplicates on a single day and repeating the tests. Six samples were prepared for the repeatability test. The sample solution was examined at various time periods (0, 2, 4, 8, 12, 24 h) to assess the sample's stability (Additional file 1: Table S3).

A recovery test was performed to confirm accuracy by introducing three concentration levels (low, medium, and high) of the mixed standard references into a certain quantity of sample solution. The recoveries ranged between 97.39% and 111.15%, with RSDs less than 8.86%, as detailed in Additional file 1: Table S4. All of these findings revealed that the developed procedures were sufficiently linear, precise, repeatable, stable, and recoverable for WDD sample quantification.

Sample analysis

The established analytical approach was effectively used for the simultaneous determination of 11 typical components in the WDD material reference and its commercial preparations, as shown in Fig. 4 and Table 4. It was clear that naringin, neohesperidin and glycyrrhizic acid were the major constituents of the WDD material reference samples among the studied chemicals in the laboratory-made samples. Furthermore, naringin was the most abundant component in a daily dose. Modern research has shown that naringin and neohesperidin possess higher antioxidant and anti-inflammatory activities as well as effects on bone regeneration, metabolic syndrome, oxidative stress, genetic damage and central nervous system (CNS) diseases [52, 53]. Glycyrrhizic acid is the principal bioactive ingredient with antiviral, anti-inflammatory and hepatoprotective effects [54]. Therefore, these components play an important role in WDD in the treatment of cardiovascular diseases, hyperglycaemia and dyslipidaemia. The laboratory-made WDD material reference has relative standard deviation (RSD) values of contents in the range of 11.78–54.80%. However, commercial samples with RSD values in the range of 44.89–150.50% clearly showed significant differences in the concentrations of each identified component. In terms of the content of 11 components in the commercial preparation, only CP8 is close to the material reference by comparing the RSD 24.74–59.39%, which is in agreement with the statistical analysis of the heatmap. Other commercial preparations are more different from the material reference. In addition to liquirtigenin and glycyrrhizic acid of WDD, the contents of the other 9 compounds to be prepared in the laboratory were higher than those in commercial preparations. In summary, the contents of the components examined in the commercial WDD samples were highly diverse, indicating considerable variances in their quality.

Fig. 4

Contents of 11 representative components in the WDD material reference and its commercial preparations. (Due to the large difference in the content of 11 components, which do not fit in one vertical coordinate, they are divided into two parts. The figure above shows the contents of synephrine, glycyrrhizic acid, liquiritin, naringin, neohesperidin and hesperidin. The figure below shows the content of naringenin, eriocitrin, tangeretin, liquirtigenin and adenosine)

Table 4 Contents of 11 representative components in the WDD material reference and its commercial preparations (μg/g, n = 3)