Chemical characterization of different parts of Forsythia suspensa and α-glucosidase and pancreatic lipase inhibitors screening based on UPLC-QTOF-MS/MS and plant metabolomics analysis

2.1 Plant material

14 batches of fruits of FS (G1-G14), 14 batches of flowers of FS (H1-H14) and 12 batches of leaves of FS (Y1-Y12) were collected in Hebei, Shanxi, and Henan provinces in northern China, and all the batches of different parts of FS samples were dried in the shade and stored in desiccator. The voucher specimens that identified by Associate Professor Guo Long have been deposited at Hebei Provincial Technology Innovation Center of Chinese Medicine Processing, Hebei University of Chinese Medicine. Sample information of different parts of FS are shown in Supplementary Table S1.

2.2 Instruments

The chemical characterization of different parts of FS were performed on Agilent 1290–6545 UPLC-QTOF-MS/MS system (Agilent Technologies, Santa Clara, USA). Chromatographic separation was performed on ACQUITY UPLC BEH C18 column (Waters, Milford, USA). The absorbance in α-glucosidase and pancreatic lipase inhibitory activities was measure by VICTOR Nivo multimode microplate reader (PerkinElmer, Waltham, USA). Ultrapure water was prepared by a Synergy water purification system (Millipore, Billerica, United States).

2.3 Materials and reagents

The reference standards of isochlorogenic acid B, rutin, quercetin, kaempferol-3-O-rutinoside, pinoresinol-4-O-glucoside, phillyrin, phillygenin were purchased from Chengdu Must Bio-Technoligy Co., Ltd. (Chengdu, China). α-Glucosidase, p-nitrophenyl-α-D-glucopyranoside (p-NPG), porcine pancreatic lipase (type II), 4-methylumbelliferyl oleate (4-MUO) were purchased from Sigma-Aldrich (St Louis, MO, United States). HPLC grade methanol, acetonitrile and formic acid were obtained from Fisher Scientific (Pittsburgh, PA, United States). Other chemicals and reagents were of analytical grade.

2.4 Extraction procedure

The samples of fruits, flowers and leaves of FS were crushed and passed through 40 mesh sieves. 0.1 g of sample powder was accurately weighed and mixed with 7 mL of 70 % (v/v) methanol, sonicated for 1 h, and adjusted to the initial weight by adding 70 % methanol (v/v) as needed. The sample was centrifuged at 13000 rpm/min for 10 min, 200 μL of the supernatant was removed and diluted 10 times, and 50 μL of isochlorogenic acid B solution (1.2 mg/mL) was added precisely as internal standard. Then, the supernatant was filtered through a 0.22 μm membrane before UPLC-QTOF-MS/MS analysis. For α-glucosidase and pancreatic lipase inhibition experiments, 600 μL of the above-extracted sample of different parts of Forsythia suspensa was taken out, the solvent was evaporated to remove the organic solvents and re-dissolved with equal volumes of phosphate buffer and Tris-HCl buffer, respectively.

2.5 Chemical profile of different parts of Forsythia suspensa by UPLC-QTOF-MS/MS

The UPLC-QTOF-MS/MS analysis of fruits, flowers and leaves of FS was carried out on an Agilent 1290 infinity UPLC system coupled with an Agilent 6545 quadrupole time-of-flight mass spectrometer system (Agilent Technologies, Santa Clara, USA). Chromatographic separation was performed on an ACQUITY UPLC BEH C18 column (2.1 mm × 100 mm, 1.7 µm, Waters, Milford, USA). The mobile phases were consisted of 0.1 % formic acid water (A) with methanol (B) at a flow rate of 0.3 mL/min. The linear gradient elution was optimized as follows: 0–5 min, 10 % B; 5–15 min, 10–15 % B; 15–20 min, 15 % B; 20–26 min, 15–20 % B; 26–32 min, 20–25 % B; 32––40 min, 25–50 % B. The column temperature was maintained at 30 °C, and the injection volume was 1 μL. The MS parameters were set as follows: drying gas (N₂) temperature, 320 °C; drying gas (N₂) flow rate, 10.0 L/min; sheath gas temperature, 350 °C; sheath gas flow (N₂) rate, 11 L/min; nebulizer gas pressure, 35 psi; fragmentor voltage, 135 V; capillary voltage, +4000 V; collision energy, 40 eV. Data acquisition was performed on Agilent MassHunter Workstation (Agilent Technologies, Santa Clara, CA, USA). In addition, the quality control (QC) samples were prepared by mixing equal volumes of each test solution and used to assess system stability and data reproducibility. The QC samples were alternated every 5 injections, and analyzed in a data-dependent MS tandem.

2.6 Metabolomic analysis

Peak filtering, matching, calibration and normalization of metabolomics data were performed on Agilent Mass Profiler Professional B.06 software (Agilent Technologies, Santa Clara, CA, USA). Compound abundance was log 2 transformed, and normalized by the parent nuclear ion ([M + H]⁺ 517.1633) of the internal standard (isochlorogenic acid B). Then, the metabolomic data was exported into SIMCA 14.1 (Umetrics, Malmo, Sweden) for PCA and OPLS-DA.

2.7 α-Glucosidase and pancreatic lipase inhibitory activities in vitro

To measure the α-glucosidase inhibition of different parts of FS, the experimental method was slightly adapted from literature (Chang et al., 2022). Briefly, 20 μL of FS samples were mixed with 75 μL of phosphate buffer (0.1 M, pH 6.9) and 65 μL of α-glucosidase solution (1 U/mL) in a 96-well plate. Pre-incubation for 15 min at 37℃, then the reaction was started by addition of 30 μL of p-NPG (2 mM). After incubation for 20 min at 37 ℃, the reaction was stopped by adding 50 μL of Na₂CO₃ (0.2 M). The absorbance (A_sample) was measure by a microplate reader at 405 nm. The background sample (A_background) was prepared by replacing the α-glucosidase solution with the same volume of phosphate buffer. The control sample (A_control) was using phosphate buffer instead of FS sample solution. The blank sample (A_blank) was prepared by adding phosphate buffer instead of α-glucosidase solution and FS sample solution. All the samples were analyzed in triplicate with five different concentrations, and the α-glucosidase inhibition (%) was calculated as follows:

α-glucosidase inhibition (%) = 1 - (A_sample – A_background)/(A_control – A_blank) × 100.

The inhibitions of different parts of FS on pancreatic lipase were assessed according to literatures (Chang et al., 2021). The FS samples, pancreatic lipase, and 4-MUO were prepared in Tris-HCl buffer solution (pH 8.0). 25 µL of FS samples and 25 µL of pancreatic lipase solution (1 mg/mL) were added into a black bottom 96-well plate. Pre-incubation for 10 min at 37 °C, the reaction was started by addition of 50 µL of 4-MUO solution (1 mM). After incubation for 20 min at 37℃, the reaction was stopped by adding 100 μL of 0.1 M citrate buffer solution (PH 4.2). The amount of 4-methylumbelliferone released by the pancreatic lipase was measured with a fluorometric microplate reader with an excitation wavelength at 355 nm and an emission wavelength at 460 nm. The background sample (A_background) was prepared by replacing the pancreatic lipase solution with the same volume of buffer solution. The control sample (A_control) was prepared by adding buffer solution instead of the FS sample. The blank sample (A_blank) was prepared by adding buffer solution instead of pancreatic lipase solution and FS sample solution. All the samples were analyzed in triplicate with five different concentrations, and the pancreatic lipase inhibition (%) was calculated as follows: $$pancreaticlipaseinhibition\left(\%\right)=1-\left(A_{\mathit{sample}}-A_{\mathit{background}}\right)/\left(A_{\mathit{control}}-A_{\mathit{blank}}\right)\times 100$$

The α-glucosidase and pancreatic lipase inhibitory activities of different parts of FS samples were expressed as the concentration of sample with 50 % reduction in enzyme activity (IC₅₀), which was obtained by non-linear regression analysis of α-glucosidase and pancreatic lipase inhibition (%) versus sample concentration (mg/mL extracts equivalents) curves using GraphPad Prism version 9.0 (GraphPad Software, San Diego, CA, USA).

The α-glucosidase and pancreatic lipase inhibitory activities of selected potential α-glucosidase and pancreatic lipase inhibitors (rutin, quercetin, kaempferol-3-O-rutinoside, pinoresinol-4-O-glucoside, phillyrin) were determined and the IC₅₀ values were also calculated.

2.8 Pearson correlation analysis

Pearson correlation analysis was employed to explore the correlation between the characteristic metabolites and α-glucosidase and pancreatic lipase inhibitory activities of fruits, flowers and leaves of FS. Taking the Pearson correlation coefficient as an index, the abundance of the characteristic metabolites in different parts of FS samples was set as one set of independent variables, and α-glucosidase or pancreatic lipase inhibitory activities (IC₅₀ values) was set as the other set of independent variables. Then, Pearson correlation coefficients between the characteristic metabolites and IC₅₀ values were calculated by SPSS 22.0 statistics software (SPSS Inc., Chicago, IL, United States).

2.9 Molecular docking analysis

To provide deep insight into the interaction effects between the screened inhibitors and enzymes, the molecular docking analysis was carried out based on the method described in previous literatures (Darwish et al., 2022). The crystal structures of α-glucosidase (PDB ID: 3A4A) and pancreatic lipase (PDB ID: 1LPB) were obtained from the RCSB Protein Data Bank (https://www.rcsb.org). The receptors were imported into Maestro 13.1 and preprocessed by Protein Preparation Wizard in the Schrodinger software to remove unnecessary water molecules, followed by hydrogen bond network optimization. The LigPrep module in the Schrodinger software was used to process ligands to generate possible three-dimensional (3D) structures. The Glide Grid Generation Wizard was used to generate docking grids. The docking grids was set as a 20 Å × 20 Å× 20 Å square pocket. The enzyme activity site coordinates between the ligand and 1LPB receptor protein was X: 11.12, Y: 20.24, Z: 48.25. The enzyme activity site coordinates with 3A4A receptor protein was X: 13.5, Y: −10.3, Z: 17.57. Molecular docking was performed under the Ligand Docking wizard, using SP and XP modes. The docking number output was set to 10, and the number of ligands docking energy optimization cycles was set to 100. Using MM-GBSA technology provided by Glide module and Prime module, the ligand binding energies and ligand strain energies could be calculated for each small molecule ligand and two proteins (1LPB and 3A4A), and the contribution value of hydrogen bond and the stability of binding could be shown. The final result was based on XP Gscore as the evaluation criterion. PYMOL was further used for visual analysis of possible binding interactions between inhibitors and enzymes.

3.1 Chemical profile of different parts of Forsythia suspensa by UPLC-QTOF-MS/MS

For the comprehensive characterization of secondary metabolites in different parts of FS, the crude extracts of fruits, flowers and leaves of FS were analyzed by UPLC-QTOF-MS/MS. To achieve a maximum sensitivity for secondary metabolites, the effects of the ionisation parameters including ionisation mode, nebulizer gas pressure, electrospray voltage of the ion source and collision energy were investigated. The detection signal in positive ion mode was better than that in negative ion mode because of sensitivity and selectivity. Under the optimized chromatographic condition, most of the secondary metabolites in different parts of FS were well separated, and the typical total ion chromatograms of fruits, flowers and leaves of FS are displayed in Fig. 2. Combining the retention time, mass-to-charge ratio, fragmentation patterns, previous literatures and databases, the secondary metabolites were identified (Chen et al., 2017; Zhang et al., 2020; Li et al.,2022; Zhou et al.,2022). A total of 74 secondary metabolites, including 15 phenylethanoid glycosides, 20 lignans, 10 cyclohexanol derivatives, 11 organic acids, 9 flavonoids, 3 triterpenes, and 10 other constituents were identified or tentatively identified in three different parts of FS. The identification information, such as retention time, molecular formula, molecular ions and MS/MS fragments is summarized in Table 1, and the chemical structures of the 74 secondary metabolites are shown in Supplementary Fig. S1.

Close

Fig. 2

The typical total ion chromatograms of fruits (A), flowers (B) and leaves (C) of Forsythia suspensa by UPLC-QTOF-MS/MS. The peak numbers are consistent with the compound numbers presented in Table 1.

Export to PPT

Phenylethanoid glycosides are one of the main bioactive compounds in FS. Forsythoside A was taken as an example to explore the fragmentation patterns of phenylethanoid glycosides. In the MS/MS spectrum (Supplementary Fig. S2), forsythoside A produced a protonated ion [M + H]⁺ at m/z 625.1893. Due to the loss of caffeoyl and H₂O, forsythoside A exhibited fragment ions at m/z 463.1652 [M + H-C₉H₆O₃]⁺ and 445.1489 [M + H-C₉H₆O₃-H₂O]⁺. Caffeic acid-related fragment ions, such as m/z 181.0481 [C₉H₈O₄ + H]⁺, 163.0325 [C₉H₈O₄ + H-H₂O]⁺, and 137.0566 [C₉H₈O₄ + H-CO₂]⁺ were also observed in the MS/MS spectra of phenylethanoid glycosides. According to the characteristic fragmentation patterns of phenylethanoid glycosides, a total of 15 phenylethanoid glycosides, including hydroxytyrosol glucoside (8), forsythoside E (14), R-suspensaside A (31), S-suspensaside A (32), calceolarioside A (34), calceolarioside C (36), forsythoside I (40), lianqiaoxinoside C (43), forsythoside H (47), forsythoside B (48), forsythoside A (49), calceolarioside B (51), suspensaside A (55), acteoside (57) and forsythenside K (61) were identified in different parts of FS.

Lignans are also one of the major active constituents of FS. The MS/MS fragmentation behavior of lignans was investigated using phillyrin. As shown in Supplementary Fig. S3, phillyrin showed a strong protonated ion [M + Na]⁺ at m/z 557.1974 with the molecular formula C₂₇H₃₄O₁₁. The main fragment ions of phillyrin appeared at m/z 395.1477 [M + Na-C₆H₁₀O₅]⁺ and m/z 380.1183[M + Na-C₆H₁₀O₅-CH₃]⁺, which were due to the loss of the glucose (C₆H₁₀O₅) and CH₃. A total of 20 lignans including 8-hydroxypinoresinol-4-glucoside (28), isolariciresinol-4-O-glucoside (35), 8-hydroxypinoresinol-4′-glucoside (37), lariciresinol-9-O-glucoside (42), pinoresinol-4-O-glucoside (54), epi-pinoresinol-4-O-glucoside (56), epi-pinoresinol-4′-O-glucoside (58), agastinol (59), demethylphillyrin (60), (+)-pinoresinol monomethyl ether-β-D-glucoside (63), suspenoidside B (64), matairesinoside (65), phillyrin (66), arctiin (67), arctigenin (68), (+)-pinoresinol (69), (-)-pinoresinol (71), epipinoresinol (72), matairesinol (73) and phillygenin (74) were identified from the different parts of FS samples.

Forsythenside B is a typical cyclohexanol derivative in FS. Forsythenside B exhibited the protonated ion [M + Na]⁺ at m/z 489.1355 (C₂₂H₂₆O₁₁). Its fragment ions were at m/z 339.1818 [M + Na-C₈H₆O₃]⁺, 321.0945 [M + Na-C₈H₆O₃-H₂O]⁺, 177.0521 [M + Na-C₈H₆O₃-C₆H₁₀O₅]⁺ and 159.0521 [M + Na-C₈H₆O₃-H₂O-C₆H₁₀O₅]⁺, which were typically characterized by the loss of sugar moieties and H₂O (Supplementary Fig. S4). Thus, 10 cyclohexanol derivatives such as rengynic acid-4-O-β-D-glucoside (1), galiridoside (7), 6′-methoxypolygoacetophenoside (9), vanilloloside (12), salidroside (13), forsythenside B (16), rengyoside D (26), rengyoside C (27), forsythenside A (30) and forsythenside H (52) were characterized in different parts of FS.

Organic acids contained in FS are mainly caffeoylquinic acids or its derivatives. Caffeoylquinic acids can present distinct fragments of caffeic acid and quinic acid, such as m/z 181 and 193 in positive ion mode. For example, chlorogenic acid showed the precursor ion [M + H]⁺ at m/z 355.1022 (C₁₆H₁₈O₉), and the fragment ions at m/z 181.0342 [caffeoyl + H]⁺, 163.0386 [M + H-C₉H₆O₃-H₂O]⁺, and 145.0283 [M + H-C₉H₆O₃-2H₂O]⁺ (Supplementary Fig. S5). Based on the fragmentation patterns of caffeoylquinic acids, 11 organic acids including vanillic acid (2), quinic acid (4), caffeic acid 2-(1 naphthyl) bethyl ester (6), chlorogenic acid (15), 4-O-β-D-glucosyl-4-coumaric acid (17), caffeic acid (19), (3S)-3-hydroxydecanoic acid (20), 1-O-feruloyl-β-D-glucose (22), p-coumaroyl quinic acid (23), 7-epi-12-hydroxyjasmonic acid glucoside (25) and 3,4-diethoxybenzoic acid (29) were identified or tentatively identified.

For flavonoids, rutin could be an example. Rutin exhibited the precursor ion [M + H]⁺ at m/z 611.1629 (C₂₇H₃₀O₁₆) in MS/MS spectrum (Supplementary Fig. S6), and produce ions at m/z 465.1045 [M + H-C₆H₁₀O₄]⁺, 303.0516 [M + H-C₆H₁₀O₅-C₆H₁₀O₄]⁺, 302.0388 [M + H-C₆H₁₀O₅-C₆H₁₀O₄-H]⁺, 273.0405 [M + H-C₆H₁₀O₅-C₆H₁₀O₄-CH₂O]⁺ by loss of glucose, rhamnose and CH₂O. In addition, the RDA cleavage-related ions at m/z 153.0218 [RDA]⁺ were also observed. A total of 9 flavonoids, such as phlorizin (3), hesperetin 5-O-glucoside (24), quercetin (39), quercitrin (39), rutin (41), hyperoside (44), hesperidin (45), kaempferol-3-O-rutinoside (46), kaempferol (53) have been identified in FS.

It has been widely reported that secondary metabolites of FS were dominated by phenylethanoid glycosides, lignans and cyclohexanol derivatives. However, various organic acids and flavonoids have been also detected in different parts of FS. Based on the chemical characterization results, it could be noted that the chemical profiles of secondary metabolites in fruits, flowers and leaves were similar, but there were still certain different constituents among the different parts of FS samples (Dong et al., 2017). Four phenylethanoid glycosides, including calceolarioside C (36), forsythoside H (47), forsythoside B (48), calceolarioside B (51), and four lignans, including agastinol (59), suspenoidside B (64), epipinoresinol (72), matairesinol (73) were only contained in fruits of FS. Several metabolites, such as phlorizin (3), galiridoside (7), vanilloloside (12), 3,4-diethoxybenzoic acid (29), arctigenin (68) were only detected in flowers of FS. The leaves of FS also contained some unique secondary metabolites, such as rengyoside D (26) and rengyoside C (27), two cyclohexanol derivatives.

3.2 Chemical comparison of different parts of Forsythia suspensa by metabolomics analysis

To further explore the distribution of secondary metabolites and characterize differential metabolites among the fruits, leaves and flowers of FS, plant metabolomics analysis was performed on UPLC-QTOF-MS/MS data. Multivariate statistical analysis including PCA and OPLS-DA were employed to compare and investigate the secondary metabolites difference of the three parts of FS.

PCA is an unsupervised multivariate statistical method that can examine correlations among multiple variables, and enable data simplification and visualization (Chen et al., 2022). In order to fully understand the aggregation and dispersion differences among fruits, leaves and flowers of FS, PCA was performed on the relative abundance of 74 secondary metabolites identified by UPLC-QTOF-MS/MS. The results of PCA display the metabolic differences and similarities among the different samples. As shown in Fig. 3A, the green dots, red dots, blue dots and yellow dots represented the fruits, leaves, flowers and QC samples, respectively. The R²X (Chang et al., 2022), R²X (Chang et al., 2021) and Q² parameters of the PCA model were 0.326, 0.216 and 0.914, which indicated a receivable classification and prediction ability of the model. The score plot of PCA showed that the QC samples were tightly clustered in the center as one cluster, which indicated that the analytic system was stable. 40 batches of different parts of FS samples could be clearly divided into three clusters corresponding to fruits, leaves and flowers of FS. It could be noted that the fruits and flowers samples were relatively concentrated, which indicated relatively smaller within-group differences of the samples. While the leaves samples were relatively evacuated, indicating larger within-group differences. In general, the difference among the fruits, leaves and flowers of FS samples were obvious and differentiated, and the PCA results indicated that the secondary metabolites among the fruits, leaves, flowers of FS were significant different.

Close

Fig. 3

The score plots of PCA (A) and OPLS-DA (B) of fruits, flowers and leaves of Forsythia suspensa samples. Fruits, G1-G14; Flowers, H1-H14; Leaves, Y1-Y12; Quality control, QC1-QC8.

Export to PPT

OPLS-DA is a supervised multivariate statistical method which focuses on the differentiation between groups, and a reliable tool to screen of differential metabolites (Jin et al., 2021). The OPLS-DA model could achieve effective separation of samples with little differences, and identify the characteristic variables for distinction. Thus, OPLS-DA was further employed to compare the secondary metabolites difference and find the potential differential metabolites among the fruits, leaves and flowers of FS. The OPLS-DA results showed that the R²Y and Q² of the established model were 0.994 and 0.957, and there was no overfitting or outliers when the outputs of the permutation test (N = 200) and the Hotelling T2 test (using 95 % and 99 % confidence limits) were examined, which indicating the OPLS-DA model had a great classification and prediction ability. The score plot of OPLS-DA (Fig. 3B) showed that the three different parts of FS samples were clearly concentrated in three regions, which was consistent with the results of PCA. To select the differential metabolites, the secondary metabolites identified in different parts of FS were further screened based on VIP values. The VIP values represent the differences of the variables, and variables could be regard as important roles for the differentiation when the VIP values were more than 1.5 (Li et al., 2024). Thus, the VIP values of the secondary metabolites were further calculated to select the potential differential metabolites among the different parts of FS samples. Finally, a total of 29 secondary metabolites were screened out as the potential differential metabolites based on the condition of VIP > 1.5, P < 0.05 (Supplementary Table S2). The results showed that the 29 potential differential metabolites including 5 phenylethanol glycosides, 8 lignans, 5 cyclohexanol derivatives, 4 flavonoids, 4 organic acids, 1 triterpene, and 2 other components were significant to effectively discriminate the fruits, leaves and flowers of FS samples, and these characteristic metabolites could be recognized as chemical markers for discrimination and difference of the three different parts of FS.

Then, a clustering heatmap analysis was established to reveal the distribution of 29 screened differential metabolites. As shown in Fig. 4, 40 batches of different parts of FS samples were divided into three groups corresponding to fruits, leaves, flowers based on the relative abundance of the differential metabolites. It could obviously find that the distribution of 29 secondary metabolites in fruits, leaves, flowers of FS differed greatly, which indicated that the relative contents of these metabolites were different. The relative contents of forsythoside I (40), pinoresinol-4-O-glucoside (54), epipinoresinol (72) and matairesinol (73) were higher in fruits of FS compared to those of flowers and leaves. Compared to fruits and leaves, the relative contents of caffeic acid 2-(1-naphthyl) ethyl ester (6), galiridoside (7), vanilloloside (12), 2-[4-(3-hydroxypropyl)-2-methoxyphenoxy]-1,3-propanediol-1-glucoside (62) and (+)-pinoresinol monomethyl ether-β-D-glucoside (63) were higher in flowers of FS. While the relative contents of rengynic acid-4-O-β-D-glucoside (1), forsythoside E (14), forsythenside A (30), (Z)-3-hexenylvicianoside (33), 8-hydroxypinoresinol-4-glucoside (37), kaempferol-3-O-rutinoside (46), forsythoside A (49), phillyrin (66) and phillygenin (74) of leaves samples were much higher than those of the fruits and flowers samples.

Close

Fig. 4

Heatmap analysis of 29 different metabolites in fruits, flowers and leaves of Forsythia suspensa. Fruits, G1-G14; Flowers, H1-H14; Leaves, Y1-Y12.

Export to PPT

3.3 α-Glucosidase and pancreatic lipase inhibitory activities of different parts of Forsythia suspensa

The α-glucosidase inhibitors can reduce glucose in dietary carbohydrates and lower postprandial blood glucose, which are helpful in reducing post-prandial blood glucose in treating prediabetic conditions and delaying the progression of diabetes (Hasan et al., 2023). Pancreatic lipase can involve in the absorption of dietary lipids in the gastrointestinal tract, and the pancreatic lipase inhibitors could prevent the hydrolytic absorption of fat (Zeng et al., 2018). In this present work, the α-glucosidase and pancreatic lipase inhibitory activities of fruits, leaves and flowers of FS samples have been studied by α-glucosidase and pancreatic lipase inhibition assays in vitro. Acarbose and orlistat were used as positive controls in α-glucosidase and pancreatic lipase inhibition assays, respectively. The results showed that acarbose had great inhibitory effect on α-glucosidase with the IC₅₀ 0.019 ± 0.001 μmol/mL and orlistat also exhibited good inhibitory effect on pancreatic lipase with the IC₅₀ 0.086 ± 0.016 μmol/mL, which indicated the established α-glucosidase and pancreatic lipase inhibition assays were reliable. As shown in Fig. 5A, all the different parts of FS samples displayed good α-glucosidase inhibitory activities, but the IC₅₀ of fruits, leaves and flowers of FS samples were different. The average IC₅₀ values of the fruits, leaves and flowers of FS samples on α-glucosidase were 0.24 ± 0.06 mg/mL, 0.17 ± 0.04 mg/mL, and 0.41 ± 0.10 mg/mL, respectively (Fig. 5B). For pancreatic lipase, all the different parts of FS samples also displayed good inhibitory activities (Fig. 5C), and the average IC₅₀ values of the fruits, leaves and flowers of FS samples were 0.94 ± 0.23 mg/mL 0.56 ± 0.33 mg/mL, 0.65 ± 0.15 mg/mL, respectively (Fig. 5D). The α-glucosidase and pancreatic lipase inhibition results showed that the leaves of FS had the strongest α-glucosidase and pancreatic lipase inhibitory capacities among the three parts of FS samples, which the IC₅₀ values were 0.17 ± 0.04 mg/mL and 0.56 ± 0.33 mg/mL, respectively.

Close

Fig. 5

The inhibitory activities of fruits, flowers and leaves of Forsythia suspensa on α-glucosidase and pancreatic lipase. (A) The IC₅₀ values of different batches of fruits (G1-G14), flowers (H1-H14) and leaves (Y1-Y12) of Forsythia suspensa on α-glucosidase. (B) The average IC₅₀ values of fruits (G), flowers (F) and leaves (L) of Forsythia suspensa samples on α-glucosidase. (C) The IC₅₀ values of different batches of fruits (G1-G14), flowers (H1-H14) and leaves (Y1-Y12) of Forsythia suspensa samples on pancreatic lipase. (D) The average IC₅₀ values of fruits (G), flowers (F) and leaves (L) of Forsythia suspensa samples on pancreatic lipase. *P < 0.05, ** P < 0.01.

Export to PPT

The results of α-glucosidase and pancreatic lipase inhibition assays demonstrated that all the fruits, leaves and flowers of FS have significant medicinal potential in inhibiting α-glucosidase and pancreatic lipase, and the inhibitions of different parts of FS varied greatly. The different inhibitory effects of α-glucosidase inhibition and pancreatic lipase could be due to the presence of characteristic metabolites in different parts of FS samples. Further research is needed to investigated the relationships between the characteristic metabolites and α-glucosidase and pancreatic lipase inhibitory activities, and explore the potential α-glucosidase and pancreatic lipase inhibitory constituents.

3.4 Pearson correlation analysis

In order to further investigate the relevance between characteristic metabolites and enzyme inhibitory activities, Pearson correlation analysis was employed. Pearson correlation analysis is a multivariate statistical model, which is applied to extract factors that have the greatest impact on the outcome variables and maximize the relationships between the two sets of variables (Zhang et al., 2023). In this present work, the abundance of 29 differential metabolites and IC₅₀ of enzymes (α-glucosidase or pancreatic lipase) inhibitory capacities of fruits, leaves and flowers of FS samples were set as two groups of variables, and the Pearson correlation coefficients between these two groups were calculated. As shown in Fig. 6, among the 29 differential metabolites screened above, 15 metabolites were negatively correlated with α-glucosidase inhibition activity, and 21 metabolites were negatively correlated with pancreatic lipase inhibition activity. It could be observed (Table 2) that 12 characteristic metabolites including rengynic acid-4-O-β-D-glucoside (1), forsythoside E (14), 8-hydroxypinoresinol-4-glucoside (28), forsythenside A (30), R-suspensaside A (31), S-suspensaside A (32), quercitrin (38), rutin (41), kaempferol-3-O-rutinoside (46), pinoresinol-4-O-glucoside (54), phillyrin (66), phillygenin (74) were negatively correlated to both the two enzyme inhibitory activities, which indicated that these 12 metabolites had high inhibitory effects on both α-glucosidase and pancreatic lipase, and the inhibitory activities of the samples for both enzymes increased with the increase in the content of these components.

Close

Fig. 6

Pearson correlation analysis between 29 differential metabolites and α-glucosidase and pancreatic lipase inhibitory activities (IC₅₀ values). Red means positive correlation and blue means negative correlation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Export to PPT

Pearson correlation analysis results showed that some certain components in different parts of FS had significant contribution to α-glucosidase and pancreatic lipase inhibitory activities, and could be recognized as potential enzyme inhibitory ingredients. Finally, a total of 12 characteristic metabolites were screen out as potential α-glucosidase and pancreatic lipase inhibitors in different parts of FS.

3.5 Validation of α-glucosidase and pancreatic lipase inhibitors in vitro

According to the results of Pearson correlation analysis, 12 potential α-glucosidase and pancreatic lipase inhibitors have been explored in FS. In order to verify the reliability of the result, the α-glucosidase and pancreatic lipase inhibitory activities of 5 potential inhibitors including quercitrin, rutin, kaempferol-3-O-rutinoside, pinoresinol-4-O-glucoside and phillyrin were determined by α-glucosidase and pancreatic lipase inhibition assays in vitro. As shown in Fig. 7, all the five components displayed strong inhibitory effects on α-glucosidase and pancreatic lipase in concentration dependent manners. For α-glucosidase inhibitory activity, the IC₅₀ of quercitrin, rutin, kaempferol-3-O-rutinoside, pinoresinol-4-O-glucoside, and phillyrin were 0.162 ± 0.004 μmol/mL, 0.072 ± 0.004 μmol/mL, 0.101 ± 0.001 μmol/mL, 1.206 ± 0.082 μmol/mL, 0.515 ± 0.002 μmol/mL, respectively. For pancreatic lipase inhibitory activity, the IC₅₀ of quercitrin, rutin, kaempferol-3-O-rutinoside, pinoresinol-4-O-glucoside, and phillyrin were 0.215 ± 0.018 μmol/mL, 0.135 ± 0.009 μmol/mL, 0.105 ± 0.006 μmol/mL, 2.685 ± 0.197 μmol/mL, and 1.614 ± 0.114 μmol/mL, respectively. The above results indicated that the selected potential enzymes inhibitors have certain α-glucosidase and pancreatic lipase inhibition, which verified the results of Pearson correlation analysis.

Close

Fig. 7

Inhibitory effects of quercitrin, rutin, kaempferol-3-O-rutinoside, pinoresinol-4-O-glucoside and phillyrin on α-glucosidase (A) and pancreatic lipase (B).

Export to PPT

3.6 Molecular docking analysis

In order to further verify the α-glucosidase and pancreatic lipase inhibitory activities of the five enzyme inhibitors mentioned above and predict the preferred binding sites, molecular docking models were constructed. The predicted binding modes of the five enzyme inhibitors docked into α-glucosidase are shown in Fig. 8. The Prime module of the Schrodinger software was used to calculate the free energy of binding between the five small molecule ligands and the α-glucosidase protein (3A4A). The optimal binding affinities of quercitrin, rutin, kaempferol-3-O-rutinoside, pinoresinol-4-O-glucoside, and phillyrin were −9.91 kcal/mol, −10.0544 kcal/mol, −10.4045 kcal/mol, −8.53959 kcal/mol and −9.29068 kcal/mol, respectively. The molecular docking results indicated that all the five small molecule compounds could bind α-glucosidase protein well, and kaempferol-3-O-rutinoside had the best binding effect, which was consistent with the trend of inhibitory capacities of the five compounds against α-glucosidase (IC₅₀ values). Kaempferol-3-O-rutinoside formed hydrogen bonds with LYS-156, SER-240, SER241, THR-310, ASP-307, ASP-352, ARG-315, GLU-277 residues, Pi-Pi interactions with residue TYR-158, and hydrophobic bonds with GLU-411 and TYR-158 residues of α-glucosidase protein. Similarly, the quercitrin, rutin, pinoresinol-4-O-glucoside and phillyrin also formed hydrogen bonds, ionic bonds and hydrophobic interactions with multiple residues of α-glucosidase protein. LYS-156 and GLU277 residues contributed the most to protein and ligand interactions.

Close

Fig. 8

The molecular docking analysis of five enzyme inhibitors with α-glucosidase protein receptor (3A4A). (A) quercitrin, (B) rutin, (C) kaempferol-3-O-rutinoside, (D) pinoresinol-4-O-glucoside, (E) phillyrin.

Export to PPT

The predicted binding modes of the five enzyme inhibitors docked into pancreatic lipase were also investigated (Fig. 9). The free energy of binding between the five small molecule ligands and the pancreatic lipase protein (1LPB) was calculated by the same method as above. The optimal binding affinity of quercitrin, rutin, kaempferol-3-O-rutinoside, pinoresinol-4-O-glucoside, and phillyrin and pancreatic lipase were −8.97 kcal/mol, −9.96 kcal/mol, −10.01 kcal/mol, −8.0428 kcal/mol and −9.56 kcal/mol, which indicated that these small molecule compounds and pancreatic lipase had relatively ideal potential activity effects. Among the five enzyme inhibitors, kaempferol-3-O-rutinoside had the best binding effect, which was basically consistent with the trend of the verification results of pancreatic lipase inhibitory activities in vitro. Kaempferol-3-O-rutinoside formed hydrogen bonds with PHE-77, ASP-79, ARG-256, THR-255, SER-152 residues and Pi-Pi interactions with PHE-215 residues of pancreatic lipase protein. The quercitrin, rutin, pinoresinol-4-O-glucoside and phillyrin also formed hydrogen bonds, ionic bonds and hydrophobic interactions with multiple residues of pancreatic lipase protein, such as PHE-215, ARG-256 and ASP-79.

Close

Fig. 9

The molecular docking analysis of five enzyme inhibitors with pancreatic lipase protein receptor (1LPB). (A) quercitrin, (B) rutin, (C) kaempferol-3-O-rutinoside, (D) pinoresinol-4-O-glucoside, (E) phillyrin.

Export to PPT

In summary, the molecular docking results indicated that the five potential enzyme inhibitors in FS mainly compete with the substrate for the active sites of α-glucosidase and pancreatic lipase through hydrogen bonding, hydrophobic forces and ionic bonding to achieve inhibition of enzyme activities.