Integration of pharmacodynamics, metabolomics and network pharmacology to elucidate the effect of Prunella vulgaris seed oil in the treatment of hyperlipidemia

2.1 Plant material and preparation of PVSO

Prunella vulgaris L. was purchased from Hunan Yafei Traditional Chinese Medicine Slices Co., Ltd. (No. 211201). The plant species identification was confirmed by Professor Limei Lin (College of Pharmacy, Hunan University of Chinese Medicine). Supercritical fluid extraction using CO₂ was performed using a laboratory-scale supercritical fluid system (Hangzhou Huali Pump Co., Ltd) to produce PVSO. The extraction conditions were carried out with slight modifications according to Lee et al. (2021). The extraction was performed at 40 ℃, 25 MPa of extraction kettle, 50 ℃, 5 MPa of separation kettle, and constant CO₂ flow rate of 15 L/h for 2 h. Then the obtained PVSO was weighted and stored at −80 ℃ until further analysis. The extract yield of PVSO from Prunella vulgaris seeds was 16.63 %. The extraction yield (% w/w) was calculated using the following equation:

2.2 UHPLC-QE-MS and GC-TOF-MS analysis

1. UHPLC-QE-MS.

Metabolite extraction: PVSO (100 mg) was mixed with 500 μL of −40 ℃ precooled extraction solvent (methanol: water = 4:1, including internal standard 1000:10). The solution was then homogenized (JXFSTPRP-24, Shanghai Jingxin Technology Co., Ltd.) for 4 min under 45 Hz, and sonicated for 1 h in ice-water bath (YM-080S, Shenzhen Fangao Microelectronics Co., Ltd.). After holding at 40 ℃ for 1 h, the extract was centrifuged at 12000 rpm (Heraeus Fresco17, Thermo Fisher Scientific) for 15 min at 4 ℃. Then, 100 μL of supernatant was taken for injection analysis after 0.22 μM microporous membrane filtration.

Instrumental analysis (Zhao et al., 2017): An UHPLC system (Vanquish, Thermo Fisher Scientific) with a Waters UPLC BEH C18 column (1.7 μm, 2.1 × 100 mm) was used to perform HPLC separation. The mobile phase consisted of 0.1 % formic acid in water (A) and 0.1 % formic acid in acetonitrile (B). Gradient elution was set as follows: from 0 to 11 min, 85–25 % A; from 11 to 12 min, 25–2 % A; from 12 to 14 min, 2–2 % A; from 14 to 14.1 min, 2–85 % A; from 14.1 to 16 min, 85–85 % A. The flow rate was 0.5 mL/min, and the sample injection volume was 5 μL. An Orbitrap Exploris 120 mass spectrometer coupled with Xcalibur software (Thermo Fisher Scientific) was employed to obtain the MS and MS/MS data based on the IDA acquisition mode. The ionization parameters were set as follows: sheath gas flow rate-35 Arb; aux gas flow rate-15 Arb; ion transfer tube and vaporizer temperature-350 ℃; collision energy-16/32/48 eV (NCE mode); spray voltage-5.5 kV (positive) or −4 kV (negative).

• 2)

  GC-TOF-MS.

Metabolite extraction: 50 ± 1 mg of sample was weighed into a 2 mL tube and spiked with 450 μL of precooled extraction mixture (methanol/chloroform (v:v) = 3:1) with 10 μL of adonitol (0.5 mg/mL stock) and 10 μL of L-2-chlorophenylalanine (1 mg/mL stock) as internal standards. Solutions were vortexed for 30 s and homogenized with a ball mill at 45 Hz for 4 min, followed by ultrasonication in ice water for 5 min. After centrifugation (12000 rpm, 4 ℃, 15 min), 300 μL of supernatant was transferred to a fresh tube. Sixty microliters of each sample were removed and combined together to prepare the QC (quality control) sample. After evaporation in a vacuum concentrator, 50 μL of methoxyamination hydrochloride (20 mg/mL in pyridine) was added, incubated at 80 ℃ for 30 min, and then derivatized by 70 μL of BSTFA reagent (1 % TMCS, v/v) at 70 ℃ for 1.5 h. After cooling to room temperature, 5 μL of FAMEs (in chloroform) was added to the QC sample. All samples were then analyzed by an Agilent 7890 gas chromatograph coupled with a time-of-flight mass spectrometer (GC-TOF-MS).

Instrumental analysis (Kind et al., 2009): The separations were performed with a DB-5MS capillary column. Helium was used as carrier gas at a flow rate of 1 mL/min, and the injection volume was 1 μL in splitless mode. The initial temperature was kept at 50 ℃ for 1 min. After that the temperature was raised to 310 ℃ at a ramping rate of 10 ℃/min, and then held at 310 ℃ for 8 min. The injection, transfer line, and ion source temperatures were 280, 280 and 250 ℃, respectively. The energy was −70 eV in electron impact mode. The mass spectrometry data were acquired in full-scan mode with a m/z range of 50–500 at a rate of 12.5 spectra per second after a solvent delay of 6.25 min. Chroma TOF (V 4.3x, LECO) software was used for raw data analysis, and the LECO-Fiehn Rtx5 database was used for metabolite identification by matching the mass spectrum and retention index.

2.3 Pancreatic lipase inhibition assay

Triglycerides are decomposed into glycerol and fatty acids by pancreatic lipase, so they can be titrated with sodium hydroxide (phenolphthalein as an indicator) to calculate the generation of fatty acids and reflect the activity of pancreatic lipase. The pancreatic lipase inhibitory activity of PVSO was determined according to the method described by Wang et al. (2014) with some modifications. Briefly, 1 mL of homogenized olive oil (substrate) and various volumes of PVSO (0, 5, 10, 15, 20, and 25 μL) were dissolved in 5 mL of 0.025 mol/L sodium phosphate buffer (pH 7.04, containing 0.1 M NaCl) and maintained at 40 ℃ for 5 min. 1 mL of pancreatic lipase (30,000 U/g, purchased from Shanghai Ruiyong Biotechnology Co., Ltd) solution (240 U/mL) was added to this reaction mixture, and then incubated at 40 ℃ for 5 min. After that, 15 mL of 75 % ethanol was immediately added to terminate the enzyme reaction. With phenolphthalein as the indicator and the reaction mixture without PVSO as the blank control, the released fatty acid was titrated with 0.025 mol/L sodium hydroxide to obtain the molar mass. The amount of the enzyme that released 1 mmol free fatty acid per minute under standard measurement conditions was defined as a “lipase unit”. The experiment was conducted in triplicate, with the data presented as the means ± SEMs. The inhibitory rate of a certain PVSO on pancreatic lipase activity was expressed as follows:

2.4 Cholesterol esterase inhibition assay

The cholesterol esterase inhibition activity of PVSO was determined spectrophotometrically based on the method of Long et al. (2022). Cholesterol esterase (48 U/mg, purchased from Yuanye Biotechnology Co., Ltd., Shanghai, China) was predissolved in high-purity water (20 μg/mL), and 100 μL of substrate PNPB (purchased from Shanghai Ruiyong Biotechnology Co., Ltd) was predissolved in 2800 μL of acetonitrile. NaCl (0.2925 g) and sodium taurocholate (0.1387 g) were dissolved in 50 mL (0.1 mol/L, pH 7.4) of sodium phosphate buffer solution. Then, 50 μL porcine cholesterol esterase (0.96 U/mL) was mixed into 1 mL of the above solution, and then PVSO at different volumes (0, 5, 10, 15, 20, 25 μL) was added and incubated at room temperature for 5 min. An ultraviolet visible spectrophotometer was used to measure the absorbance at 405 nm after coloring with 10 μL of PNPB. The experiment was performed in triplicate, and data are presented as the means ± SEMs. The percentage inhibition was calculated as

2.5 Fluorescence measurements

The effects of PVSO on the fluorescence spectra of pancreatic lipase and cholesterol esterase was determined according to the method of Martinez-Gonzalez et al. (2017), with some modifications. Briefly, 2 mL of the pancreatic lipase or 1.5 mL of the cholesterol esterase solution (final concentration is 0.2 mg/mL) and different volumes (0, 5,10, 15, 20 μL) of PVSO were mixed for 1 min and incubated at 37 ℃ for 1 h. Then the mixture was centrifuged at 5000 rpm for 15 min. A volume of 1 mL of supernatant was diluted to 3 mL with the phosphate buffer, and a Shimadzu F-6000 fluorescence spectrometer (Shimadzu, Japan) was used to measure the effects of different volumes of PVSO on the fluorescence spectrum of pancreatic lipase or cholesterol esterase. The fluorescence spectrum parameters were as follows: excitation wavelength-280 nm, emission wavelength-300 to 420 nm, slit width of excitation band and emission light-5 nm, data interval-1 nm, and scanning speed-6000 nm/min. Fluorescence quenching was determined using the Stern-Volmer equation (Gan et al., 2013):

2.6 Determination of cholesterol micellization solubility

Cholesterol micelles were prepared according to the method described by Zhu et al. (2022a). Cholesterol (7.70 mg, Sigma) and oleic acid (14.10 mg, Shanghai Macklin Biochemical Technology Co., Ltd) were dissolved in 2 mL methanol and dried with a nitrogen blowing apparatus. Then, 53.77 mg sodium taurocholate (purchased from Shanghai Ruiyong Biotechnology Co., Ltd) and 10 mL sodium phosphate buffer solution (pH 7.4) were added. After 30 min of ultrasound (400 W, 20 kHz) treatment, the micelles were incubated at 37 ℃ for 24 h.

Then, various amounts of PVSO (0, 2, 4, 6, 8, 10 μL) were added into the 1 mL micellar solution. The micellar solution without PVSO was used as a substrate blank to calculate the recovery. After shaking and mixing, the mixtures were placed at 37 ℃ for 2 h and then centrifuged at 15000 rpm for 30 min. The cholesterol content in the supernatant was determined by a total cholesterol kit (Nanjing Jiancheng Bioengineering Institute, Jiangsu, China) (Zhang et al., 2012). The inhibition of micelle cholesterol solubility was calculated using the following equation:

2.7 Transmission electron microscopy (TEM)

The study was carried out according to the method described by Vermeer et al. (2008), with some modifications. Complex particles are formed from PVSO and micellar. To observe all the particles formed, micelles were generated using the standard procedure without centrifugation. To prevent micellar particles from adhering to obtain high-resolution imaging, the noncentrifugal mixed micellar suspension was diluted 10 times, and then 10 μL of suspension was applied to 200 mesh porous carbon copper mesh. To obtain a thin liquid membrane, the liquid membrane was dried after removing excess liquid with filter paper, and TEM (HT7800, HITACHI) imaging was observed.

2.8 Animals

Forty-eight male C57BL/6J mice (SPF, 20 ± 2 g) were purchased from Hunan Slack Jingda Experimental Animal Co., Ltd. [Changsha, China, Certificate number: SYXK (Xiang) 2019–0004].

After 1 week of accommodation, the mice were randomly divided into six groups: (1) normal control (CON) group and (2) PVSO control (PVSO-C) group, were fed with a standard normal chow diet (10 kcal% fat, 20 kcal% carbohydrate and 20 kcal% protein); (3) model control (HFD) group, (4) low-dose PVSO (PVSO-L) group, (5) middle-dose PVSO (PVSO-M) group, and (6) high-dose PVSO (PVSO-H) group, were fed with a high fat diet (20 kcal% fat, 44 kcal% carbohydrate and 22 kcal% protein). Meanwhile, mice in the (4–6) group were supplemented with 200, 500, and 800 mg/kg body weight PVSO solution (dissolved in 0.5 % CMC-Na) by gavage once a day; mice in the (1 and 3) group were supplemented with 0.5 % CMC-Na solution by gavage once a day; and mice in the (2) group were supplemented with 800 mg/kg body weight PVSO solution (dissolved in 0.5 % CMC-Na) by gavage once a day. An overview of the experimental design for all groups is shown in Fig. 3A.

The experiment lasted for 10 weeks. At the end of the experimental period, the mice were fasted overnight and anesthetized with pentobarbital sodium. Then, blood samples were collected and centrifuged to obtain serum. Parts of fresh liver tissues were also collected and immediately fixed in 4 % paraformaldehyde for pathological section observation, and the rest were stored at −80 ℃ for further assays. The control diet (XT310) and the high-fat diet (XT304) were purchased from Jiangsu Xietong Pharmaceutical Bioengineering Co., Ltd. (Nanjing, China). All the diets were stored at −20 ℃ throughout this study until needed. All animal experiments were approved by the Ethics Animal Care Committee of Hunan University of Chinese Medicine.

2.9 Biochemical analysis

Total cholesterol (TC), total triglyceride (TG), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C) were measured with commercial kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China).

According to the manufacturers' instructions, frozen liver tissues were accurately weighed and homogenized with anhydrous ethanol or saline (1:9 (w/v)). Afterward, the liver homogenate was centrifuged (10000 rpm, 4 ℃) for 10 min, and the supernatant was taken for the experiment. The protein concentration in the liver was quantified by the Bicinchoninic Acid (BCA) Protein Colorimetric Assay Kit (Elabscience, Wuhan, China). Then, lipid-related indicators (TC and TG) and oxidative stress parameters in the liver, such as superoxide dismutase (SOD), malondialdehyde (MDA) and glutathione peroxidase (GSH-Px), were analyzed by assay kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Oxidative stress parameters were normalized to the respective protein concentrations.

2.10 Histopathological examination

According to a previous study (Ta et al., 2023), fresh liver tissues fixed in 4 % paraformaldehyde were embedded in paraffin and sectioned, followed by staining with hematoxylin and eosin (H&E) for histopathologic analysis. Liver slides were observed under a microscope (Nikon ECLPSE-Ci).

2.11 Untargeted metabolomics

Referring to the method of Zhou et al. (2019), untargeted metabolomics techniques were used to study the changes in serum metabolites. Briefly, serum samples were mixed with cold methanol/acetonitrile (1:1, v/v) to remove the protein and centrifuged at 14,000 g for 20 min at 4 ℃. The vacuum-dried supernatants were redissolved in 100 µL acetonitrile/water (1:1, v/v) solvent and centrifuged (14000 g, 4 ℃) for 15 min, and then the supernatant was injected.

Ultrahigh-performance liquid chromatography (Vanquish UHPLC, Thermo) coupled to a Q Exactive mass spectrometer (Orbitrap MS, Thermo) was performed for analysis. Briefly, a 2.1 mm × 100 mm ACQUITY UPLC BEH Amide 1.7 µm column (Waters, Ireland) was used for hydrophilic interaction liquid chromatography (HILIC) separation. After sum normalization, the processed data were analyzed by the R package (ropls). The distributions of all samples were observed, and a general overview of changes in metabolic patterns was obtained through principal component analysis (PCA). Variables related to the separation were identified, and the variable influence on projection (VIP) was calculated through orthogonal partial least squares-discriminate analysis (OPLS-DA). Metabolites with VIP > 1 and p values < 0.05 (Student’s t test) were defined as significantly different metabolites (DEMs).

2.12 RNA isolation and real-time quantitative PCR analysis in liver

Total RNA was extracted from frozen liver tissue using TRIzol reagent (Sangon Biotech, Shanghai, China), and reverse transcription was performed according to the manufacturer's instructions for the RevertAid RT Reverse Transcription Kit (Thermo Scientific, USA). The mRNA levels were quantified using a real-time PCR system (QuantStudio^TM1 Plus System, ABI/Thermo, USA) with 2X SG Fast qPCR Master Mix (High Rox) (Sangon Biotech, Shanghai, China) as previously described (Liu et al., 2021). The PCR conditions were as follows: 95 ℃ for 3 min for initiation, followed by 45 cycles of 15 s at 95 ℃ and 30 s at 60 ℃. The primers of the target genes are shown in Table 1. Relative expression of the target gene was normalized to the expression of β-actin by the 2^-ΔΔCt method.

2.13 Integrative analysis of the network pharmacology and metabolomics

Integrated metabolomics and network pharmacology analysis was performed as described previously (Zhang et al., 2022b).

The potential targets related to PVSO: 262 ingredients of PVSO (Similarity > 80 %) identified by LC-MS/MS and GC–MS/MS were searched through SwissADME (https://www.swissadme.ch/index.php) with the SMILES obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov) to select the ingredients with high GI absorption and at least two terms of druglikeness were YES. Since then, 200 ingredients meeting the above criteria have been considered to be potential active ingredients. Then, the potential active ingredients were imported into Swiss Target Prediction (https://www.swisstargetprediction.ch) to identified the potential target proteins under the condition of “Probability*>0.1″ and deleting duplicate targets.

The potential targets related to hyperlipidemia: The targets related to hyperlipidemia were obtained from Online Mendelian Inheritance in Man database (OMIM, https://www.ncbi.nlm.nih.gov/omim) and GeneCards (https://www.genecards.org/). Finally, the common targets shared by the two databases were retained, while the other targets were removed.

The potential targets related to DEMs: Based on metabolomics, the obtained differential metabolites between HFD and PVSO groups were input into HMDB database to obtain the SMILE format of substances and input into Swiss Target Prediction database to obtain metabolite targets.

Construction of regulation network diagram: Venn diagram of potential targets from PVSO active ingredients, serum metabolomics, and hyperlipidemia disease was drawn to find out intersection targets (overlapped genes) by using an online network drawing tool (https://bioinformatics.psb.ugent.be/webtools/Venn/). The protein–protein interaction (PPI) network diagram was obtained by importing the intersection targets into String (https://string-db.org/Version 10.5) website and visualizing with Cytoscape 3.8.2 software. The nodes in the network are ranked according to the Degree by using the CytoHubba plug-in of Cytoscape 3.8.2, and the top 10 targets were selected as potential core targets. Furthermore, the intersection targets were imported into the Metpascape website (http//metascape.org/) for GO function and KEGG pathway enrichment analysis. The pathway with p < 0.05 was screened. Potential active ingredients corresponding to 17 core targets from the PVSO were selected as active ingredients, and DEMs corresponding to 17 core targets were selected as target DEMs. The intersection targets, active ingredients, target DEMs, and KEGG pathway were imported into Cytascape 3.8.2 software to build an “Active ingredients-Targets-KEGG pathway-Hyperlipidemia” network. The top 5 ingredients were selected as the key ingredients based on the degree value and relative contents, and 5 key targets were obtained at the same time. Then, molecular docking analysis was conducted between these key ingredients and targets.

2.14 Molecular docking

The crystal structures were collected from the RCSB PDB (https://www.rcsb.org/). The pre-processed target receptors and ligands were processed in the way of ultra-precise molecular docking based on the grid algorithm with all functional residues retained in the grid. The structure of the ligands remained flexible to produce different conformations and the calculations were performed in OPLS force fields. Schrodinger Maestro 11.9 (https://www.schrodinger.com/) was used for all computational calculations.

2.15 Statistical analysis

Origin 8.5 software was used for linear fitting, and GraphPad Prism 8.0 was used for statistical analysis. Two-way analysis of variance (ANOVA) followed by multiple comparisons test and one-way analysis of variance (ANOVA) followed by Dunnett's multiple comparisons test or Tukey's multiple comparison test were used to compare multiple groups (Gosis et al., 2022). Values are presented as the mean ± standard error of the mean (SEM), and p < 0.05 was considered statistically significant.

3.1 Compounds in PVSO

3.1.1

3.1.1 UPLC-MS/MS

PVSO detected 21,514 peaks in the positive (ESI + ) ion mode (Fig. 1A), of which 272 compounds were identified (Table S1, Supplementary Materials). The peak area normalization method was used to calculate the relative percentage content of each component. PVSO in the positive ion mode mainly involves eight categories (Table 2), including terpenoids, fatty acids, phenylpropanoids, alkaloids, phenols, quinones, aromaticity, and flavonoids. The main components of PVSO were (10E,12E)-9-hydroxyoctadeca-10,12-dienoic acid (1.97 %), alpha-linolenic acid (0.23 %), 11,17,21-trihydroxypregn-4-ene-3,20-dione (0.22 %), lucidenic acid M (0.19 %), glyceryl linolenate (0.17 %), dihydroactinidiolide (0.13 %), grandisin (0.10 %), bryonolic acid (0.10 %), loliolide (0.09 %), betulin (0.09 %), etc.

Close

Fig. 1

Total ion chromatogram (TIC) of PVSO acquired by LS-MS and GS-MS. The relative standard deviation (RSD) ≤ 30 % was used to select 21,514 metabolites, ion numbers in the positive (ESI + ) mode (B) and 20,675 in the negative (ESI-) mode (A). Total ion chromatogram (TIC) of PVSO acquired by GS-MS (C).

Export to PPT

Table 2 Positive ion mode.

Class	Number	Peak area	Relative content (Accounting for all peaks)	Relative content (Accounting for identified peaks)
Terpenoids	100	3.7142 × 109	1.38 %	25.39 %
Fatty acids	4	6.5421 × 108	0.24 %	4.47 %
Phenylpropanoids	16	4.8385 × 108	0.18 %	3.31 %
Alkaloids	15	4.4178 × 108	0.16 %	3.02 %
Phenols	11	2.9401 × 108	0.11 %	2.01 %
Quinones	3	1.6863 × 108	0.06 %	1.15 %
Aromaticity	4	9.2493 × 107	0.03 %	0.63 %
Flavonoids	13	8.7691 × 107	0.03 %	0.60 %
Others	106	4.6343 × 108	3.24 %	59.14 %
Identified peaks	272	1.4627 × 1010	5.45 %
Unknown peaks	21,242	2.5394 × 1011	94.55 %
All peaks	21,514	2.6857 × 1011

A total of 20,675 peaks were detected according to the negative (ESI-) MS/MS spectrum (Fig. 1B), of which 79 compounds were identified (Table S2, Supplementary Materials). The peak area normalization method was used to calculate the relative percentage content of each component. PVSO in the negative ion mode mainly involves seven categories (Table 3), including fatty acids, terpenoids, phenylpropanoids, phenols, flavonoids, aromaticity, and alkaloids. The main components of PVSO were 9-hydroxy-10,12,15-octadecatrienoic acid (1.66 %), 9E,11E-linoleic acid (0.36 %), azelaic acid (0.27 %), 13-HODE (0.26 %), honokiol (0.14 %), 12,13-EODE (0.14 %), cholic acid (0.11 %), trans- vaccenic acid (0.10 %), etc.

Table 3 Negative ion mode.

Class	Number	Peak area	Relative content (Accounting for all peaks)	Relative content (Accounting for identified peaks)
Fatty acids	11	7.5158 × 108	0.69 %	16.06 %
Terpenoids	21	4.8510 × 108	0.44 %	10.36 %
Phenylpropanoids	11	4.1490 × 108	0.38 %	8.86 %
Phenols	8	1.8767 × 108	0.17 %	4.01 %
Flavonoids	7	4.7724 × 107	0.04 %	1.02 %
Aromaticity	4	4.2863 × 107	0.04 %	0.92 %
Alkaloids	2	1.0804 × 107	0.01 %	0.23 %
Others	15	2.7406 × 109	2.51 %	58.54 %
Identified peaks	79	4.6812 × 109	4.29 %
Unknown peaks	20,596	1.0455 × 1011	95.71 %
All peaks	20,675	1.0923 × 1011

3.2 Effect of PVSO on the activity of pancreatic lipase and cholesterol esterase

The inhibitory effect of PVSO on pancreatic lipase and cholesterol esterase activities is presented in Fig. 2A and B. PVSO inhibited the activities of pancreatic lipase and cholesterol esterase in a dose-dependent manner, with IC50 values of 1.68 and 3.964 mg/mL, respectively. The inhibitory effect of PVSO on pancreatic lipase is stronger than that on cholesterol esterase.

Close

Fig. 2

The lipid-lowering activity of PVSO in a simulated intestinal environment. (A) The inhibition rate of PVSO on pancreatic lipase at different concentrations (0, 0.59, 1.18, 1.77, 2.36, 2.95 mg/mL). (B) Inhibition rate of PVSO on cholesterol esterase at different concentrations (0, 3.91, 7.82, 11.73, 15.64, 19.55 mg/mL). (C) The pancreatic lipase of PVSO treated with different concentrations (0, 0.69, 1.38, 2.06, 2.75 mg/mL) was analyzed by fluorescence spectroscopy. (D) The cholesterol esterase of PVSO treated with different concentrations (0, 0.92, 1.83, 2.75, 3.67 mg/mL) was analyzed by fluorescence spectroscopy. (E) Fluorescence spectra of pancreatic lipase treated with different concentrations of PVSO and Stern-Volmer plot of PVSO fluorescence quenching of pancreatic lipase (λem = 340 nm). (F) Fluorescence spectra of cholesterol esterase treated with different concentrations of PVSO and Stern-Volmer plot of PVSO on fluorescence quenching of cholesterol esterase (λem = 330 nm). (G) PVSO (0, 1.65, 3.29, 4.94, 6.59, and 8.23 mg/mL) was added to a bile salt micellar solution containing cholesterol in vitro to form precipitates. When different concentrations of PVSO were added to the bile salt micelle solution, it was observed that the solution immediately became turbid. (H) Effect of PVSO on the solubility of cholesterol micelles. The results are expressed as the mean ± SEM deviation, with different letters indicating significant differences (p < 0.05). (I) TEM. The left is blank cholesterol micelle, while the right is cholesterol compound micelle containing 8 mg/mL PVSO.

Export to PPT

Close

Fig. 3

Effect of PVSO supplementation on body and liver weights in HFD-fed mice. (A) Animal experimental design. (B) Body weight changes, (C) final body weights, (D) weight gain, (E) liver weights, and (F) liver index (liver: body weight). Data are represented as the means ± SEMs (n = 8). The differences were evaluated using two-way analysis of variance (ANOVA) followed by multiple comparisons test (B) or one-way ANOVA with Tukey’s multiple comparisons test (C-F) and denoted as follows: ^#p < 0.05, ^##p < 0.01, and ^###p < 0.001, compared with the CON group; *p < 0.05, ^**p < 0.01, ^***p < 0.001, compared with the HFD group.

Export to PPT

3.3 Fluorescence quenching measurements

The fluorescent amino acid residues contained in proteins are attributed to the endogenous fluorescence of proteins. These amino acid residues (especially Trp residues) contain conjugated double bonds or benzene ring structures, which can absorb and emit fluorescence (Huang et al., 2020). The interaction of chromophores and quenching agents may lead to a decrease in fluorescence quantum yield, and changes in the tertiary structure of the protein may lead to changes in intrinsic fluorescence emission (Su et al., 2016b; Sun et al., 2016). Therefore, in the absence and presence of PVSO, fluorescence spectroscopy can be used to detect the fluorescence characteristics of proteins, reflecting changes in protein fluorophores and determining structural changes in proteins.

As shown in Fig. 2C and D, the relative fluorescence intensity of pancreatic lipase and cholesterol esterase treated with PVSO decreased with increasing PVSO concentration, which indicates that PVSO can quench the endogenous fluorescence of pancreatic lipase and cholesterol esterase. We speculate that this is because PVSO interacts with pancreatic lipase or cholesterol esterase to form a complex that can quench the enzyme, resulting in changes in the tertiary structure of these enzymes. However, PVSO did not affect the maximum wavelength of fluorescence emission, reflecting that PVSO could not cause changes in the microenvironment of Trp residues in pancreatic lipase or cholesterol esterase. According to literature reports, the fluorescence quenching effect of proanthocyanidin and astaxanthin on pancreatic lipase was also not generated by changing the microenvironment of Trp in pancreatic lipase (Du et al., 2018). As shown in Fig. 2E and F, the value of Ksv for pancreatic lipase (0.09 mL/mg) was lower than that for cholesterol esterase (0.10 mL/mg), indicating that PVSO was more effective in quenching the intrinsic fluorescence of cholesterol esterase, with more affinity.

3.4 The inhibitory effect of PVSO on cholesterol micellization solubility

Approximately 50 % of cholesterol in the intestine is absorbed, while the rest is excreted through feces (Jesch et al., 2017). Cholesterol absorption in the small intestine is closely related to the concentration of total and low-density lipoprotein cholesterol in the plasma; thus, the efficiency of cholesterol absorption in the small intestine has attracted great interest (Jesch et al., 2017). Dietary cholesterol must be incorporated into bile salt micelles before absorption. Therefore, inhibiting cholesterol solubility in micelles and subsequently inhibiting its absorption can effectively reduce cholesterol. For example, green tea catechins, especially those with galloyl moieties, can reduce the micellar solubility of cholesterol through their hydrophobic domains (Ogawa et al., 2016). Therefore, interfering with the formation of cholesterol micelles in the intestine may contribute to the prevention of hyperlipidemia.

In the present study, when PVSO at different concentrations was added to the bile salt micelle solution, it was observed that the solution immediately became turbid (Fig. 2G). PVSO reduced the solubility of cholesterol in bile salt micelle solution in a dose-dependent manner. The highest inhibition rate on cholesterol micelle solubility was 48.29 ± 1.08 %, with 6.60 mg/mL of PVSO (Fig. 2H).

3.5 TEM

As shown in Fig. 2I, the particle size of the cholesterol micelle was small, but after adding PVSO, a composite micelle with a larger particle size was formed between cholesterol and PVSO. Based on Fig. 2G and H, we speculate that the inhibition of cholesterol micelle solubility by PVSO may be due to the interaction between PVSO and cholesterol leading to an increase in the particle size of the complex, as well as the formation of a coprecipitated complex of PVSO and cholesterol in the micelle solution (Zhang et al., 2020).

3.6 PVSO decreased body weight and liver index in HFD-treated mice

As shown in Fig. 3B-F, HFD-fed mice showed significantly higher final body weights and liver weights (p < 0.05) but no significant difference in weight gain and liver index compared with mice in the CON group. PVSO supplementation markedly decreased final body weights and weight gain, as well as liver weights, in a dose-dependent manner compared with the HFD group (p < 0.05), but did not alter the liver index. Furthermore, during the whole experiment, no mouse deaths and no significant differences in the final body weights, weight gain, and liver index were found in the PVSO-C group. These results indicated that PVSO could ameliorate body weight and liver weight gain in mice induced by a HFD diet without side effects.

3.7 PVSO decreased lipid accumulation in HFD-treated mice

Parameters relevant to lipid metabolism in both serum and liver were evaluated (Fig. 4A-E). Specifically, serum TC and LDL-C levels increased in mice fed the HFD compared to the CON group, while HDL-C significantly decreased (p < 0.05). However, medium-dose and high-dose PVSO remarkably reversed the changes in TC content in comparison with the HFD group (p < 0.05). Among them, the TC level distinctly decreased by 17.13 % and 13.99 % in the PVSO-M and PVSO-H groups, respectively. Moreover, the contents of TG and LDL-C in the PVSO-M and PVSO-H groups also showed a downward trend. Liver TC and TG levels of mice in the HFD group were higher than those in the CON group, which were significantly decreased with 500 mg/kg PVSO supplementation (p < 0.05; Fig. 4F, G). The 800 mg/kg PVSO intake also significantly reduced the content of TC (p < 0.05) and showed a decrease in the TG content of mice compared with the HFD group (Fig. 4F, G).

Close

Fig. 4

Effect of PVSO supplementation on the lipid profiles of HFD-fed mice. (A-D) The content of serum TC, TG, HDL-C, and LDL-C. (E) The ratio of LDL-C/HDL-C. (F-G) The content of hepatic TC and TG. (H) Histological assessment of the liver. The liver sections were stained with hematoxylin and eosin, and green, yellow and black arrows indicate hepatocyte cytoplasmic vacuolation, fatty changes, and inflammatory infiltration, respectively. Data are expressed as the means ± SEMs (n = 5–8). The differences were assessed by one-way ANOVA with Dunnett's multiple comparisons test and denoted as follows: ^#p < 0.05, ^##p < 0.01, and ^###p < 0.001, compared with CON; *p < 0.05, ^**p < 0.01, ^***p < 0.001, compared with HFD.

Export to PPT

Hepatic lipid accumulation was evaluated by H&E staining of liver sections in mice to determine whether PVSO reduced hepatic steatosis induced by a high-fat diet. The results showed that HFD supplementation led to cellular vacuolization, fatty changes, and inflammatory cell infiltration, while PVSO supplementation improved the alteration of liver structure in a dose-dependent manner (Fig. 4H).

3.8 PVSO improved the oxidative stress parameters and relative mRNA expression levels of liver tissues in HFD-fed mice

Enhanced lipid peroxidation levels may be one of the main pathogenic factors of HFD-induced oxidative stress. As a landmark product of oxidative stress, MDA indicates the process of lipid peroxidation in the liver, which accumulates in the body and is toxic to cells (Khafaga et al., 2017; Ahmed et al., 2020). SOD is an important part of the oxidation protection system, which can alleviate oxidative stress by catalyzing the dismutation of superoxide radicals. GSH is also the most important cellular antioxidant. The reduced activity of antioxidant enzymes (especially SOD) will increase the level of lipid peroxidation and serve as an indicator of oxidative stress in different tissues. This study further investigated the effect of PVSO on oxidative stress by measuring liver SOD and GSH-PX activities, as well as MDA levels, in each group of mice. As depicted in Fig. 5A-C, HFD significantly increased the content of MDA and decreased the activity of SOD and GSH-Px in the liver compared with the CON group (p < 0.05). After supplementation with PVSO, the level of MDA was decreased by 44.35 %, 50.98 %, and 41.95 % in the PVSO-L, PVSO-M, and PVSO-H groups, respectively. Moreover, PVSO significantly increased the activity of SOD and GSH-Px in the livers of mice (p < 0.05). These results indicated that PVSO alleviated oxidative injury induced by a high-fat diet and enhanced the antioxidant capacity of mouse liver tissue.

Close

Fig. 5

Effect of PVSO on HFD-induced alterations in oxidative stress marker GSH-Px (A), SOD (B), and MDA (C) levels and relative mRNA expression levels (D-E) of liver samples. Data are expressed as the means ± SEMs (n = 8), and differences were assessed by one-way ANOVA with Dunnett's multiple comparisons test and denoted as follows: ^#p < 0.05, ^##p < 0.01, and ^###p < 0.001, compared with CON; *p < 0.05, **p < 0.01, ^***p < 0.001, compared with HFD. (D-E) The relative mRNA expression levels related to oxidative stress in the liver. Data are expressed as the means ± SEMs (n = 3), and differences were assessed by one-way ANOVA with Tukey’s multiple comparisons test and denoted as follows: ^#p < 0.05, and ##p < 0.01, compared with CON. SOD1: superoxide dismutase 1; GPX1: glutathione peroxidase 1.

Export to PPT

To further understand the relieving effect of PVSO on oxidative stress in HFD mice, we evaluated the expression of oxidative stress-related genes. HFD significantly reduced the expression of SOD1 and GPX1 genes in the liver compared with the CON group (p < 0.05; Fig. 5D, E). In contrast, the relative mRNA levels of GPX1 were slightly recovered in the mouse livers with 500 mg/kg PVSO intake (p > 0.05; Fig. 5E).

3.9 Effects of PVSO on the serum metabolic profile

The above results indicated that the middle dose of PVSO exhibited significant therapeutic effects in improving oxidative stress and lipid metabolism disorders. Therefore, the PVSO-M group was selected for subsequent untargeted metabolomics analysis.

The PCA plots showed that the CON group and HFD group could be clearly separated, indicating significant changes in metabolites in mouse serum. In addition, the HFD group and PVSO group can also be clearly separated, indicating that there are also significant differences in metabolites between the two groups (Fig. 6A). To identify the differential metabolites, an OPLS-DA model was used. The OPLS-DA score plot shows a significant separation between the CON and HFD groups, as with the HFD and PVSO groups. To avoid overfitting, we performed OPLS-DA model verification (Permutation test n = 200). The coefficient of determination (R²) and predictive power (Q²) of the model were evaluated. These values between the HFD and CON groups were R² = 0.99 and Q² = 0.707. Similarly, the values between the PVSO and HFD groups were R² = 0.987 and Q² = 0.685 (Fig. 6B, C). These results indicated that the model does not exhibit overfitting and has good stability and predictive ability.

Close

Fig. 6

Multivariate statistical analysis of serum untargeted metabolomics. (A) PCA score plots among the CON, HFD, and PVSO groups. (B-C) Score plots of OPLS-DA. (D) Volcano plots. Fold changes (FC) > 1.2 indicated upregulation, and FC < 0.833 indicated downregulation. (E) Venn plot. (F) Hierarchical clustering heatmap of 36 significantly upregulated or downregulated common DEMs. Each row in the figure represents a differential metabolite, and each column represents a set of samples. Red represents a relatively high expression level of metabolites, while blue represents a relatively low expression level of metabolites. Metabolites with similar expression patterns are clustered in the same cluster on the left.

Export to PPT

Based on OPLS-DA results, differential metabolites (DEMs) were screened through VIP values (VIP > 1) and p values (p < 0.05). As a result, a total of 172 DEMs were screened between the HFD and CON groups, of which 75 and 97 DEMs were identified in POS and NEG ionization, respectively (Table S4, Supplementary Materials). Furthermore, we screened significantly upregulated or downregulated DEMs through fold changes (FC) > 1.2 or < 0.833. There were 52 significantly upregulated DEMs and 112 significantly downregulated DEMs in the HFD group compared to the CON group (Fig. 6D). After intervention with PVSO, 130 metabolites changed significantly, of which 53 and 77 DEMs were identified in POS and NEG ionization, respectively. Twenty-seven metabolites were significantly increased and 85 metabolites were significantly decreased compared with the HFD group (Fig. 6D; Table S5, Supplementary Materials). Forty-one metabolites were the common DEMs in the two comparison groups (Fig. 6E), of which 36 substances were significantly upregulated or downregulated (Fig. 6F). Among them, 29 substances exhibited opposite trends and were most likely potential biomarkers that have been screened. These biomarkers were divided into the following four categories: lipids and lipid-like molecules (52 %), organoheterocyclic compounds (21 %), organic acids and derivatives (7 %), and others (20 %), including amino acids (e.g., glycine and guanidoacetic acid), fatty acids (e.g., Fa 18:2 + 1o, 16-hydroxyhexadecanoic acid, and gamma-hydroxybutyric acid), indoles and derivatives (indole-3-pyruvic acid), and other endogenous compounds. Compared with the CON group, the levels of 16-hydroxyhexadecanoic acid, 4-(1-piperazinyl)-1h-indole, gamma-hydroxybutyric acid, glycine, guanidoacetic acid, tetrabenazine, etc., were significantly elevated, whereas the levels of indole-3-pyruvic acid, minoxidil, fenfluramine, hypoxanthine, xanthine, 7-methyluric acid, L-(+)-erythrulose, melibiose, etc., were significantly decreased in the serum of mice in the HFD group (p < 0.05). Furthermore, compared with the HFD group, PVSO significantly reversed the changes in the above metabolites (p < 0.05) (Table 5).

3.10 Metabolic pathway enrichment analysis

Furthermore, we selected the Kyoto Encyclopedia of Genes and Genomes (KEGG, https://www.genome.jp/kegg/pathway.html) database to screen differential metabolic pathways based on the conditions with pathway enrichment impact > 0.05 and p < 0.05. The results showed that the differential metabolic pathways between the CON and HFD groups included necroptosis, D-glutamine and D-glutamate metabolism, caffeine metabolism, glycine, serine and threonine metabolism, linoleic acid metabolism, etc. (Fig. 7). The differential metabolic pathways between the HFD and PVSO groups included choline metabolism in cancer, cholesterol metabolism, caffeine metabolism, glycine, serine and threonine metabolism, bile secretion, etc. (Fig. 7). Among them, ABC transporters, butanoate metabolism, caffeine metabolism, central carbon metabolism in cancer, glycine, serine and threonine metabolism, protein digestion and absorption were the common pathways between the CON and HFD groups, as well as between the HFD and PVSO groups. Hence, we focused on discussing these pathways in detail.

Close

Fig. 7

KEGG pathway analysis of DEMs. (A) Bubble diagram of the KEGG enrichment pathways. Each bubble in the bubble chart represents a metabolic pathway (select the top 20 with the highest significance based on p value). In topology analysis, the larger the abscissa and size of the bubble, the larger the influence factor. The vertical axis and color of bubbles represent the log10 p value for the enrichment analysis. The darker the color is, the smaller the p value, indicating a more significant degree of enrichment. The rich factor represents the proportion of the number of DEMs in the pathway to the number of metabolites annotated in the pathway. (B) Difference abundance score map of enriched metabolic pathways. The Y-axis represents the name of the difference path, and the X-axis coordinates represent the difference abundance score (DA score). A DA score of 1 indicates an upward trend in the expression of all identified metabolites in the pathway, while −1 indicates a downward trend in the expression of all identified metabolites in the pathway. The larger the point at the end of the line segment, the more metabolites there are. The depth of the line segments and point colors is proportional to the DA score value. The darker the red color is, the more inclined the overall expression of the pathway is to increase, while the darker the blue color is, the more inclined the overall expression of the pathway is to decrease.

Export to PPT

3.11 Association analysis

Spearman’s correlation analysis showed a correlation between the levels of DEMs in Table 5 and metabolic disorders, including body weights, liver weights, dyslipidemia, and oxidative stress. As shown in Fig. 8A, melibiose, 7-methyluric acid, L-(+)-erythrulose, Pc 38:6, fenfluramine, and hypoxanthine were significantly positively correlated with SOD and GSH-Px activities but negatively correlated with MDA levels (p < 0.05). Glycine, guanidinoacetic acid, gamma-hydroxybutyric acid, Fa 18:2 + 1o, Pi 36:3,2-linoleoyl-1-palmitoyl-sn-glycero-3-phosphoethanolamine, Pi 34:2, and Pi 36:4 showed the opposite association (p < 0.05). Glycine, guanidinoacetic acid, Pi 34:2, and Pi 36:4 were also significantly positively correlated with TC levels in serum and liver (p < 0.05). PC 36:8, fenfluramine, and hypoxanthine were markedly negatively correlated with body weight gain, as well as TC levels in serum and liver (p < 0.01). Indole-3-pyruvia acid was significantly positively correlated with SOD activity in the liver and HDL-C levels in serum (p < 0.01) but negatively correlated with TC and TG in the liver.

Close

Fig. 8

(A) Spearman’s correlation heatmap analysis of DEMs and metabolic syndrome-related indices. A positive correlation is in red, and a negative correlation is in blue. *p < 0.05 and **p < 0.01. (B) Metabolic network of the significantly altered metabolites associated with PVSO intervention.

Export to PPT

3.12 Integrative analysis of the network pharmacology and metabolomics

We constructed an interaction network based on metabolomics and network pharmacology to completely understand the mechanism of PVSO against hyperlipidemia. A total of 534 ingredients of PVSO were identified by LC-MS and GC–MS, 262 ingredients could be matched in SwissADME database, and 200 active ingredients of PVSO were obtained.

We searched “Related Targets” in the Swiss Target Prediction database to find target proteins corresponding to the 200 ingredients of PVSO. As a result, 932 targets were obtained. Prediction of hyperlipidemia disease related targets was conducted in OMIM, Gene Cards and Thermal Target Database databases, and 2040 potential targets were obtained. The DEMs between PVSO and HFD groups from serum metabolomics were input into Cytoscape’s Metscape plug-in to generate a compound reaction enzyme gene network, and 191 targets were obtained. The online mapping tool was used to map the target obtained from drug prediction with the target corresponding to the disease and serum metabolomics, and the Venn map was drawn to obtain 17 intersection targets (Fig. 9A).

Close

Fig. 9

Establishment of PVSO candidate ingredient-target systematic network and enrichment analyses, as well as molecular docking studies. (A) Venn diagram of component and disease related targets. (B) GO enrichment analysis. (C) KEGG pathways bubble chart. (D) “Active ingredients-Targets-KEGG pathway-Hyperlipidemia” network diagram. The red circles and blue diamonds represent the proteins and active compounds, respectively. (E) Molecular docking energy heat map of each key ingredient and each key target.

Export to PPT

In the GO function enrichment analysis (Fig. 9B), a total of 21 GO entries were obtained, including 7 biological processes (BP), 7 motor functions (MF), and 7 cellular ingredients (CC). BP was significantly enriched in carnitine shuttle, fatty acid transmembrane transport, carnitine metabolic process, amino-acid betaine metabolic process, purine ribonucleoside metabolic process, etc. MF was mainly enriched in peroxidase activity, oxidoreductase activity, acting on peroxide as acceptor, etc. CC was mainly concentrated in mitochondrial membrane, lysosome, etc. The enrichment results of KEGG pathway showed that intersection targets were significantly enriched in 6 pathways (p＜0.05). The data showed that, PVSO mainly regulated pathways in Fatty acid degradation, Glycerophospholipid metabolism, Fatty acid metabolism, PPAR signaling pathway, AMPK signaling pathway, and Thermogenesis, indicating that the intersection targets may intervene hyperlipidemia mainly by regulating the above pathways (Fig. 9C). The top 10 targets selected by using the CytoHubba plug-in of Cytoscape 3.8.2 in the degree ranking were MPO, Crnitine palmitoyltransferase 1A (CPT1A), PTGS2, PTGS1, MGAM, ADA, XDH, CPT1B, ACHE, and CPT2. Moreover, 67 active ingredients and 31 DEMs were selected corresponding to 17 core targets. The intersection targets, active ingredients, target DEMs, and KEGG pathway were imported into Cytascape 3.8.2 software to build a “Active ingredients-Targets-KEGG pathway-Hyperlipidemia” network (Fig. 9D). The key ingredients linolenic acid (LA), α-linolenic acid (ALA), 11,17,21-trihydroxypregn-4-ene-3,20-dione, honokiol, lucidenic acid M were selected as the core ingredients based on the degree value and relative contents, and key targets ACHE, CPT1A, 3-Hydroxy-3-methylglutaryl coenzyme A reductase (HMGCR), LYPLA1, PLD2 were obtained at the same time. Then, molecular docking analysis was conducted between these 5 key ingredients and 5 key targets.

3.13 Docking results

Based on the PPI network, the key targets (ACHE, CPT1A, HMGCR, LYPLA1, PLD2) and key ingredients (LA, ALA, 11,17,21-trihydroxypregn-4-ene-3,20-dione, honokiol, lucidenic acid M) were selected for molecular docking. The lower the binding energy between the target protein and small molecule, the more stable the binding, and the greater the possibility of action. A lower affinity represents a more stable interaction, and an affinity lower than −4.25 kcal/mol represents a threshold for interaction. In the present study, small molecules could enter the active center of target proteins, and the average binding energy was −6.164 kcal/mol. Among them, the binding energies of 11,17,21-Trihydroxypregn-4-ene-3,20-dione, honokiol, and lucidic acid M with the five potential core targets were all less than −4.25 kcal/mol. In addition, the binding energies of linolenic acid and α-linolenic acid with HMGCR were also less than −4.25 kcal/mol. This indicates that these five active compounds may bind to five potential core targets (Fig. 9E). Through systematic analysis of PPI and molecular docking results, we found that honokiol was a candidate key ingredient of PVSO in the treatment of hyperlipidemia, and targets of LYPLA1, ACHE, as well as PLD2 ranked in the top three. Therefore, they might have a more direct relationship with honokiol. In order to further understand the binding mode of compound and protein, PyMOL software was used to visualize the compound protein interaction (each target selects the compound with the lowest binding energy).

As seen in Fig. 10 and Table 6, Honokiol bonded with the amino acid residues MET-180 and ASP-172 of LYPLA1 through hydrogen bonding. Lucidenic acid M formed five hydrogens bonded with residues LEU-509, LEU-512, TYR-514, TYR-533 and ARG-515 of HMGCR, while eight hydrogen bonds combined with amino acid residues ARG-777, LEU-438, and VAL-405. 11,17,21-trihydroxypregn-4-ene-3,20-dione formed four hydrogen bonds either with the residues GLN-413, TYR-238, PRO-235, and SER-240 of ACHE, or with the residues ARG-364, GLN-369, and VAL-370 of CPT1A, respectively. LA or ALA combined with amino acid residue ARG-178 or SER-612 of HMGCR through only one hydrogen bond.

Close

Fig. 10

The 3D interaction diagrams of key ingredients (LA, ALA, 11,17,21-trihydroxypregn-4-ene-3,20-dione, honokiol, lucidenic acid M) and key targets (ACHE, CPT1A, HMGCR, LYPLA1, PLD2).

Export to PPT

Table 6 Dock binding free energies (^ΔGb) and bonds of compounds and proteins.

Protein	PDB ID	Structure	Docking compound	ΔGb(kcal/mol)	Bonds formed between functional groups of compound and protein residues
					Functional groups	Protein residues	bond
HMGCR	1DQ9		LA	−5.7	C = O	ARG-178	H-bond
			ALA	−4.5	C = O	SER-612	H-bond
			Lucidenic acid M	−8.1	–OH	LEU-509	H-bond
					C = O and –OH	LEU-512	H-bond
					–OH	TYR-514	H-bond
					–OH	TYR-533	H-bond
					–OH	ARG-515	H-bond
ACHE	5HQ3		11,17,21-trihydroxypregn-4-ene-3,20-dione	−8.1	C = O	GLN-413	H-bond
					–OH	TYR-238	H-bond
					–OH	PRO-235	H-bond
					–OH	SER-240	H-bond
CPT1A	2LE3		11,17,21-trihydroxypregn-4-ene-3,20-dione	−8.1	–OH	ARG-364	H-bond
					–OH	GLN-369	H-bond
					–OH	VAL-370	H-bond
LYPLA1	6QGN		Honokiol	−8.9	–OH	MET-180	H-bond
					–OH	ASP-172	H-bond
PLD2	7SVP		Lucidenic acid M	−8.1	–OH	ARG-777	H-bond
					–OH	LEU-438	H-bond
					–OH	VAL-405	H-bond