*Corresponding Author:
Tian E Zhang
Department of Basic Medical Sciences, Chengdu University of Traditional Chinese Medicine, Chengdu, Wenjiang 611137, China
This article was originally published in a special issue, “Recent Developments in Biomedical Research and Pharmaceutical Sciences”
Indian J Pharm Sci 2022:84(4) Spl Issue “7-24”

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as the author is credited and the new creations are licensed under the identical terms


Purpose of this study was to screen the active compounds of Zhenwu decoction and their gene targets in the treatment of non-alcoholic fatty liver disease by network pharmacology and cytological validation test. The candidate compounds and related targets of Zhenwu decoction were obtained from traditional Chinese medicine system pharmacology database and PharmMapper. The non-alcoholic fatty liver disease-related genes were obtained from online Mendelian inheritance in man, GeneCards and DisGeNET databases. Molecular docking was carried out to simulate the binding affinities between potential core compounds and key target genes, and the chemical constituents of Zhenwu decoction were analyzed and identified by Ultra-Performance Liquid Chromatography-Quadrupole Time-of-Flight Mass Spectrometry technology. Therapeutic effect and mechanism of Zhenwu decoction on non-alcoholic fatty liver disease were validated using in vitro analysis. The protein-protein interaction network analysis identified the albumin, protein kinase B1, epidermal growth factor receptor, caspase 3 and peroxisome proliferator activated receptor gamma as the key target genes. Gene ontology annotation analysis showed that the Zhenwu decoctionnon- alcoholic fatty liver disease target genes were mainly involved in the response to steroid hormone and lipid catabolic process. The results of the Kyoto encyclopedia of genes and genomes pathway enrichment analysis were mainly related to the Forkhead box O signaling pathway and phosphatidylinositol 3-kinaseprotein kinase B signaling pathway. Paeoniflorgenone had the highest binding affinities for albumin, protein kinase B1, epidermal growth factor receptor, caspase 3 and peroxisome proliferator activated receptor gamma. The in vitro experiment showed that 19.5 mg/ml (p≤0.05) and 39 mg/ml (p≤0.01) of Zhenwu decoction could reduce lipid accumulation. Real-time quantitative polymerase chain reaction analyses revealed that Zhenwu decoction induced down-regulation of epidermal growth factor receptor messenger ribonucleic acid levels and up-regulation of heat shock protein 90 alpha family class a member 1, mitogen-activated protein kinase 1 and phosphatidylinositol 3-kinase messenger ribonucleic acid levels.


Non-alcoholic fatty liver disease, Zhenwu decoction, network pharmacology, molecular docking, cytological validation

Non-Alcoholic Fatty Liver Disease (NAFLD) is defined as the presence of hepatic steatosis determined by imaging or histology, excluding the secondary causes of fat accumulation in the liver[1]. NAFLD can be divided into the Non-Alcoholic Fatty Liver (NAFL) and Non-Alcoholic Steatohepatitis (NASH)[1]. NAFLD is usually associated with metabolic complications such as obesity, diabetes and dyslipidemia. It is the major cause of liver diseases worldwide with a significantly increasing burden. At present, the incidence of NAFLD is about 200/10 000 persons per year. It is likely to be the leading cause of End-Stage Liver Disease (ESLD), which affects both adults and children[2]. Over the past decade, chronic liver diseases, cardiovascular diseases and Type 2 Diabetes Mellitus (T2DM) have been focused on among the NAFLD-related chronic diseases. A recent meta-analysis showed that the mortality rate of NAFLD, which is mainly caused by liver-related and cardiovascular diseases, had increased by 57 %, while the risk of T2DM and NAFLD-related chronic kidney disease had nearly been tripled[3]. It is crucial to implement novel therapeutic interventions that can prevent or reverse the damaging effects of NAFLD[4].

The famous Chinese medicine formula, Zhenwu Decoction (ZWD), was first described in the “Treatise on Febrile Diseases” by Zhang Zhongjing. ZWD is composed of Aconiti Lateralis Radix Praeparata (Fuzi in Chinese, lateral radix of Aconitum carmichaelii Debx.), Zingiber officinale Roscoe (Shengjiang in Chinese, the rhizome of Zingiber officinale Rosc.), Poria cocos (Schw.) Wolf. (Fuling in Chinese, the sclerotium of Poria cocos (Schw.) Wolf.), Paeoniae Radix Alba (Bai Shao in Chinese, a radix of Paeonia lactiflora Pall.) and Atractylodis macrocephalae Rhizoma (Bazhu in Chinese, a radix of Atractylodes macrocephalae Koidz.) in the ratio of 9:9:9:9:6, respectively. ZWD has a clear hypolipidemic effect, lowering serum triglycerides, total cholesterol and raising high density lipoprotein levels[5].

Clinical studies have proved the better therapeutic potential of ZWD in combination with Sini powder on T2DM overlapping NAFLD (T2DM-NAFLD) as compared to the control group but their mechanism of action is still not clear[6]. ZWD could regulate energy metabolism, reduce oxidative stress by lowering serum lysophospholipid levels in rats[7].

Network pharmacology adopts the “multi-compound, multi-target and multi-pathway” method and involves system biology, bioinformatics and pharmacology. It is usually used to study the systematic effects of traditional Chinese medicine and is considered to be a novel strategy for drug discovery[8] for significantly improving the success rate of drug development, reducing the cost and predicting the side effects of drugs[9]. This study was the first to elucidate the mechanisms of ZWD in NAFLD treatment with a network pharmacology approach. The flowchart of the study is presented in fig. 1.


Fig. 1: Flowchart of the study design

Materials and Methods

Extraction and screening of the active compounds of ZWD:

Chinese names of the five traditional Chinese herbs present in ZWD were used to search the active compounds of each medicine in the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) (http://tcmspw.com/tcmsp.php). TCMSP contains Chinese medicinal herbs, chemicals, targets and drug-target networks. The active compounds in ZWD were obtained using Oral Bioavailability (OB)≥30 % and Drug-Likeness, (DL)≥0.18 as screening conditions.

Screening of the targets and core compounds of ZWD:

The screened active compounds of ZWD were submitted into the PharmMapper database (http://lilab-ecust.cn/ pharmmapper/index.html) to obtain the corresponding targets using “Norm Fit≥0.90” as screening criteria. The corresponding gene names of the target proteins were obtained from through UniProt database (https://www.uniprot.org/). Duplicate gene names were then removed. The compound-target network of ZWD was constructed and visualized using Cytoscape 3.8.2, and the core compounds of ZWD were determined.

Collection of known therapeutic targets acting on the NAFLD and candidate genes:

The therapeutic target genes of NAFLD were obtained from three databases, including GeneCards (https://www.genecards.org/), Online Mendelian Inheritance in Man (OMIM) (http://www.omim.org/) and DisGeNET (https://www.disgenet.org/). The search term “nonalcoholic fatty liver” was used in all three databases and the species were limited to “Homo sapiens”. The therapeutic targets were collected and duplicate targets were removed. NAFLD-related protein targets were imported into the UniProt database. Then, the obtained genes were submitted to Venny 2.1 (https://bioinfogp.cnb.csic.es/tools/venny/) to draw a Venn diagram. The intersected genes in the two datasets were selected as candidate genes.

Construction of the Protein-Protein Interaction (PPI) network:

Search Tool for the Retrieval of Interacting Genes/ Proteins (STRING) database (https://string-db.org/) was used to study the association of the retrieved active ingredient targets of ZWD and NAFLD target genes. The confidence score of relevance was set to ≥0.4 as the critical value to obtain the PPI network. The PPI network was visualized using Cytoscape 3.8.2 and the key genes were identified by degree, betweenness and closeness.

Enrichment analyses of Gene Ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway:

The gene targets of ZWD-NAFLD were subjected to GO annotation and KEGG pathway enrichment analysis using Metascape (https://metascape.org/). The target species was set to Homo sapiens. The results were presented using an R software package.

Construction of the compound-target-pathway network:

The active compound of ZWD, selected gene targets and related pathways were submitted to Cytoscape 3.8.2 in order to build a compound-target-pathway network. The graphic network represented the connections and interactions among the target genes, proteins or molecules and pathways. The value of each node represented the number of connections between the node and other nodes and the greater the degree, the more important the node.

Molecular docking between the main active components of ZWD and core proteins:

In order to further verify the reliability of this study, the top five targets in the PPI network and the top 10 core compounds of ZWD were selected for molecular docking. First, the Spatial Data File (SDF) files of the core compounds of ZWD were downloaded from the PubChem database (https://www.uniprot.org/) and then converted to mol2 format files using Open Babel 2.4.1 software. The best three Dimensional (3D) protein structures of the main protein targets were obtained from Protein Data Bank (PDB) and saved in PDB format. Then, the downloaded protein structures and active compounds of ZWD were opened in PyMOL to remove the water molecules and ligands. The active compounds and protein targets were then imported to AutoDock software for docking. The smaller the value of docking binding force, the more stable the compound binds to the protein and the more likely it is to interact. Finally, the PDB file was imported to LigPlot software to find the interaction between small molecules and large proteins.

Ultra-Performance Liquid Chromatography- Quadrupole Time-of-Flight Mass Spectrometry (UPLC-QTOF/MS) analysis of ZWD:

Standard preparation: 0.3 ml, 0.6 ml, 1.2 ml of ZWD were diluted into 1.2 ml, 0.9 ml, 0.3 ml of 50 % methanol solution, ultrasonic treatment for 30 min and then filtered with 0.22 μm microporous membrane. UPLC analysis conditions were as follows. The results were performed on a liquid chromatography-mass spectrometer (SNAPT XS) equipped with ACQUITY UPLC CSH C18 (100 mm×2.1 mm, 1.7 μm) at 40°. The mobile phase consisted of 0.1 % formic acid aqueous solution (A) and acetonitrile (B). The optimized gradient program was 0-5 min, 95 %-90 % A; 5-8 min, 85 % A; 8-15 min, 85 %-83 % A; 15-18 min, 83 %-80 % A; 18-22 min, 80 %-70 % A; 22-28 min, 70 %-5 % A; 28-31 min, 5 %-5 % A; 31-35 min, 5 %-95 % A. The flow rate was 0.2 ml/min and the injection volume was 2 μl.

MS analysis conditions were as follows. MS analysis was performed on a liquid chromatographymass spectrometer (SNAPT XS) equipped with an Electrospray Ionization (ESI) source. The capillary voltage is set to 2000 V, the source temperature is 200°, the desolvation temperature is 450°, the cone gas flow is 50 l/h and the desolvation gas flow is 600 l/h. The sample infusion flow rate was 5 μ/min. The full scan quality range of the scan is m/z 50-1200. Data is obtained in continuous mode.

Preparation of ZWD:

The dried raw herbs of ZWD were purchased from the outpatient department of Chengdu University of traditional Chinese medicine (Chengdu, China). Zingiber officinale Roscoe was homemade (fresh Chengdu-origin ginger was washed and sliced). Aconiti Lateralis Radix Praeparata, Paeoniae radix Alba, Poria cocos (Schw.) Wolf., Zingiber officinale Roscoe and Atractylodes macrocephala Koidz. (Production batch number: 21090104, 21081906, 21101061, C038210701, respectively) were mixed at the proportion of 9:9:9:9:6, that is 42 g soaked in water (500 ml) for 30 min and then boiled for 30 min. The residues were filtered and concentrated to 0.78 g/ml and after centrifugation, the supernatant was collected and five concentration gradients (9.75 mg/ml, 19.5 mg/ml, 39 mg/ml, 78 mg/ ml and 156 mg/ml) were prepared, which were then stored at 4°.

Cell cultures and treatments:

Liver hepatocellular carcinoma cell line (HepG2) cells were cultured in 90 % high glucose sugar Dulbecco’s Modified Eagle Medium (DMEM, Gibco), containing 10 % Fetal Bovine Serum (FBS, Gibco), 1 % penicillin and streptomycin. The cells were maintained at 37° under 5 % Carbon dioxide (CO2). Free Fatty Acids (FFA) are a mixture of oleic acid and palmitic acid at 2:1. The cells were treated with different concentrations of FFAs and cell viability was observed.

The HepG2 cell viability was determined using Cell Counting Kit-8 (CCK-8). For this purpose, HepG2 cells in the logarithmic phase (2×103 cells per well) were inoculated into 96-well plates (100 μl/well) and treated with the different concentrations of ZWD (9.75 mg/ml, 19.5 mg/ml, 39 mg/ml, 78 mg/ml, 156 mg/ml) for 24 h. Then, the bottom solution was removed and 10 μl of CCK-8 solution was added to 100 μl complete medium. The CCK-8 solution was added to each well and the plate was incubated at 37° for 2 h. The absorbance was measured at 450 nm using a microplate reader.

Oil Red O (ORO) staining:

The cells were washed with Phosphate Buffered Saline (PBS) solution and fixed with 4 % paraformaldehyde for 30 min. The ORO storage solution (0.5 %) was prepared with 0.25 g oil red dry powder and 50 ml isopropanol and filtered through a 0.22 μm filter. The ORO working solution was prepared by mixing the ORO storage solution and deionized water at a ratio of 3:2. The fixed HepG2 cells were stained with an ORO working solution for 30 min at 37° and then immediately washed with PBS for 3 times. The qualitative analysis of lipids and quantification of lipid accumulation under the microscope were based on the Optical Density (OD) value of the target ORO.

Ribonucleic Acid (RNA) isolation and Real-Time quantitative Polymerase Chain Reaction (RTqPCR):

RT-qPCR was performed to measure the hepatic cell expression of the predicted NAFLD-related genes. Total RNA was collected with the RNA isolation kit following the instructions, complementary Deoxyribonucleic Acid (cDNA) synthesis kit with genomic DNA (gDNA) eraser was used to carry out the reverse transcription reactions. RT-qPCR was performed to measure the relative expression of messenger RNA (mRNA), using SYBR Green RT-PCR master mix. The reaction conditions were 95°, 3 min; 95°, 10 s; 55°, 20 s and 72°, 30 s, for 40 cycles. Beta-actin was used as a control and the 2-ΔΔCt method was conducted for the data analysis. The primer sequence of target genes was synthesized by Invitrogen Biotech Co., Ltd (Table 1). The targets Epidermal Growth Factor Receptor (EGFR), Peroxisome Proliferator Activated Receptor Gamma (PPARG), Heat Shock Protein 90 Alpha Family Class A Member 1 (HSP90AA1) and Mitogen-Activated Protein Kinase 1 (MAPK1) formed the PPI network analysis and Phosphatidylinositol 3-Kinase (PI3K) formed the PI3K-Protein Kinase B (AKT) signaling pathway.

Gene name Amplicon length Sequences

Table 1: The Nucleotide Sequences of The Primer Pairs Used for Quantitative Gene Expression

Statistical analysis:

All the data were expressed as mean±Standard Error (SE). One-way Analysis of Variance (ANOVA) in GraphPad Prism v8.3.0 was used to test and analyze the data. The difference with p<0.05 was considered statistically significant.

Results and Discussion

According to the screening criteria of OB≥30 % and DL≥0.18, a total of 61 active compounds were obtained for the five medicinal herbs of ZWD. After removing the duplicate data, a total of 59 compounds were obtained of which 20 were from Fuzi, 12 from Bai Shao, 7 from Baizhu, 15 from Fuling and 5 from Sheng Jiang. After removing the duplicate data and screening the target genes using PharmMapper and through the standardized transformation of Uniprot, a total of 133 target genes were obtained.

The compound-target network was constructed to show the complex compounds and multiple target genes and the correlations between them. There were 303 nodes and 824 edges in the compound-target network of ZWD. It was suggested that the compounds of ZWD might play a synergistic effect on these target genes and might have pharmacological effects on NAFLD and other diseases. The highest connectivity was that of BSH5 with 55 target genes with an OB value of 53.87 %; this might be considered as the potential key active compound as shown in fig. 2.


Fig. 2:Compound-target network
Note: The pink square represents the herbs in ZWD, the circle represents the active compounds of each herb, the A1 and B1 circles represent the active compounds common in different herbs and the blue diamond shape represents the related goals

As shown by the compound-target network, some compounds play a crucial role in the network, thereby proving to be a great potential in the treatment of NAFLD, such as paeoniflorin, dihydrocapsaicin and poricoic acid B. Zhang found that paeoniflorin could prevent NAFLD by restoring the serum Alanine Aminotransferase (ALT), Aspartate Aminotransferase (AST), Serum Total Cholesterol (TC), Triglyceride (TG), High-Density Lipoprotein (HDL) and Low- Density Lipoprotein (LDL) levels. Paeoniflorin could alleviate the infiltration of liver fats induced by a high-fat diet through reducing steatosis, inflammation, balloon degeneration and necrosis. Suggested mechanisms may relate to the protection on cardiovascular system as body weight and hyperlipidemia reduced, inflammation blocked and lipid deposition inhibited[8]. In addition, paeoniflorin could also significantly reduce serum insulin and glucagon levels, enhance insulin sensitivity, restore serum lipid spectrum and alleviate liver steatosis[9]. The dietary dihydrocapsaicin could also reduce blood lipid levels and improve cholesterol metabolism in hyperlipidemia rats[10]. It is worth mentioning that the three compounds in Poria cocos Wolf, including poricoic acid, pachymic acid and ergosterol, could significantly reduce the TG level in HepG2 cells treated with FFAs[11].

A total of 3181 NAFLD-related target genes were obtained from OMIM, GeneCards and DisGeNET databases. After removing the duplicate data, a total of 2608 target genes were obtained as shown in Table 2. The intersection genes among the NAFLD-related and ZWD target genes were extracted, which were 67 target genes and were considered as a candidate gene. The intersection genes are shown in fig. 3.

Source Compound Degree Binding affinity/KJ/mol
BSH5 Paeoniflorin 56 -1.74 -1.63 -1.24 -2.36 -1.35
BSH2 Paeoniflorgenone 30 -4.42 -3.8 -4.55 -3.83 -3.67
BZ3 14-acetyl-12-senecioyl-2E,8Z,10E-atractylentriol 28 -1.74 -2.31 -2.82 -2.05 -1.55
SHJ5 Dihydrocapsaicin 22 -2.42 -2.26 -2.45 -3.22 -2.43
FZ17 Jesaconitine 20 0.88 0.26 -0.02 -1.64 0.74
BZ2 14-acetyl-12-senecioyl-2E,8E,10E-atractylentriol 20 -3.34 -1.83 -2.41 -2.59 -2.56
FUL12 Poricoic acid B 17 -1.56 -1.93 -1.84 -2.08 -1.77
FUL11 Poricoic acid A 17 -1.93 -1.35 -1.98 -2.14 -1.75
FUL3 7,9(11)-dehydropachymic acid 16 -2.5 -2.42 -3.03 -3.45 -2.46
SHJ2 6-methylgingediacetate 2 15 -0.36 -0.51 -1.82 -1.78 -1.04

Table 2: The Binding Energy Values Of Core Compounds Of Zwd And Core Targets


Fig. 3: Venn diagram of intersection genes among the ZWD target and NAFLD-related genes. There were 70 targets for ZWD and 2541 targets for disease and 67 intersection genes

A ZWD-NAFLD target PPI network was obtained, which contained 64 nodes and 718 edges (fig. 4). The nodes represented proteins and edges represented protein interactions. The average node degree value was 10.7 and the average local clustering coefficient was 0.627. In the PPI network, some nodes had a high degree, such as Albumin (ALB), Protein Kinase B1 (AKT1), EGFR, Caspase 3 (CASP3) and PPARG.


Fig. 4: PPI network for the intersection genes of ZWD target and NAFLD-related genes.
Note: Each node represents a protein target and each edge represents the interaction between two nodes. The PPI network diagram was arranged according to the degree of freedom (df). The more important the node is, the closer it is to the center

These target nodes were identified as key genes and might be potential therapeutic targets for the treatment of NAFLD by ZWD. Alkaline phosphatase, ALT, AST, ALB and total serum protein were significantly correlated with the severity of NAFLD[12]. Besides, there were significant differences in the ALB-binding function between healthy subjects and different severity groups of NAFLD[13]. Differential regulation of AKT1 and AKT2 is consistent with upregulation of Forkhead Box Protein O1 (FOXO1), may justify the paradoxical state of insulin resistance relative to the glucoregulatory pathway and augmented insulin sensitivity of the liporegulatory pathway, typical of steatosis and the metabolic syndrome in obese patients[14]. Liang et al. demonstrated that the EGFR was phosphorylated in the liver tissues of the highfat diet murine model of NAFLD. The inhibition of EGFR could prevent diet-induced lipid accumulation, oxidative stress, hepatic stellate cell activation and matrix deposition[15]. Hepatocyte death is an important feature of NASH. The CASP3 activation in hepatocytes played an important role in the NASH-related apoptosis and fibrosis[16]. Finally, a study reported PPARG was positively correlated with liver TG levels in male mice[17]. Therefore, ALB, AKT1, EGFR, CASP3 and PPARG might be the potential therapeutic targets for the treatment of NAFLD.

The GO annotations of the candidate genes were carried out in three categories, including Biological Process (BP), Molecular Function (MF) and Cell Composition (CC). The top 20 significantly-enriched terms in BP, MF and CC categories (adjusted p<0.01) are presented in fig. 5A-fig. 5C, respectively. A total of 793 entries were displayed in the BP enrichment analysis, among which, many BPs such as response to nutrient levels, response to a steroid hormone, cellular response to lipids and lipid catabolic process, were closely associated with the pathogenesis of NAFLD. The CC enrichment results showed 45 terms, including vesicle lumen, secretory granule lumen, ficolin-1-rich granule, lytic vacuole and pigment granule. A total of 67 enrichment processes were identified related to MF, including nuclear receptor activity, steroid binding, steroid hormone receptor activity, DNA-binding, transcription factor binding and beta-catenin binding. In order to further investigate the biological process of these targets, the KEGG pathways enrichment analysis was conducted, which resulted in a total of 202 entries (p<0.01), including FOXO signal and PI3K-AKT signaling pathway, endocrine resistance and estrogen signaling pathway.


Fig. 5: Top 20 most important entries of (A) BP; (B) MF; (C) CC in GO annotations and 21 important entries of (D) KEGG pathway enrichment analysis

The pathogenesis of NAFLD is closely related to insulin resistance[18]. A core regulatory mechanism in the FOXO signaling pathway is the phosphorylation of AKT which is the downstream of the PI3K for insulin response (fig. 6). A study showed that the FOXO signal pathway could lead to lipid accumulation and the denaturation of hepatocytes and promote the occurrence and development of NAFLD[19]. PI3K-AKT signal transduction pathway in hepatocytes is a common molecular mechanism, which is involved in metabolic dysfunction, including obesity, metabolic syndrome and NAFLD[20]. Cai et al. found that the insulin-PI3K/ AKT-p70 S6 Kinase (p70S6K) pathway played an important role in the early activation of hepatic stellate cells[21]. Interestingly, a study reported that paeoniflorin could have positive influence on improving NAFLD and obesity by regulating the PI3K/AKT pathway[22]. The results of our compound-target-pathway suggested that paeoniflorin could interact with the PI3K-AKT signaling pathway through AKT1, EGFR, Glycogen Synthase Kinase-3 Beta (GSK3B), HSP90AA1, Mouse Double Minute 2 homolog (MDM2) and Insulin-like Growth Factor 1 Receptor (IGF1R), while paeoniflorgenone could interact with this pathway through MAPK1 and Retinoid X Receptor Alpha (RXRA). In addition, there are also numerous endocrine-related pathways, such as endocrine resistance, estrogen signaling pathway and thyroid hormone signaling pathway. Zhu et al. found that the estrogen signaling pathway could inhibit fatty acid oxidation and TGs in the liver and increase peripheral fat mobilization[23].


Fig. 6: FOXO signaling pathway
Note: The green circle indicates that this gene is ZWD-NAFLD target gene and p stands for phosphorylation

Cytoscape v3.8.2 software was used to construct a compound-target-pathway network, which is shown in fig. 7. There were 328 edges and 91 nodes in the network, which showed that ZWD could treat NAFLD through its multiple compounds and multiple target genes. Among them, paeoniflorin, paeoniflorgenone, 14-acetyl-12-senecioyl-2E, 8Z, 10E-atractylentriol, dihydrocapsaicin, etc., might play important roles in the efficacy of ZWD for the treatment of NAFLD. The specific information of the compounds is listed in Table 1.


Fig. 7: Compound-target-pathway network
Note: Pink squares represent the active compounds in ZWD, blue squares represent the target genes and green squares represent the signal pathway of NAFLD

The top 10 active compounds (paeoniflorin, paeoniflorgenone, 14-acetyl-12-senecioyl-2E, 8Z, 10E-atractylentriol, dihydrocapsaicin, jesaconitine, 14-acetyl-12-senecioyl-2E, 8E, 10E-atractylentriol, poricoic acid B, poricoic acid A, 7, 9 (11)-dehydropachymic acid and 6-methylgingediacetate 2) obtained from the compound-target network analysis were docked with the five potential core target genes (ALB, AKT1, EGFR, CASP3 and PPARG) obtained from the PPI network analysis. The details of these compounds and target genes are listed in Table 2 and presented in fig. 8. The binding energies of all the compounds except jesaconitine were less than 0, showing a good binding affinity of these compounds with the top five target genes. Among them, the binding energy of paeoniflorgenone with ALB, AKT1, CASP3 and PPARG proteins was lower than that of the other compounds and the paeoniflorgenone-EGFR showed the lowest binding energy (-4.55); ASN-842 was the active site, showing hydrogen bond interactions (fig. 9). Docking prediction provided a preliminary basis for the further study of these drug targets.


Fig. 8: Molecular docking energy diagram


Fig. 9: Molecular docking of paeoniflorgenone with EGFR, (A) Overall diagram; (B) Magnified diagram, showing detailed docking, the dashed yellow lines represent hydrogen bonds and (C) Plane drawing

The docking of the compounds with target proteins and their binding free energy, which were less than 1, suggested that the 10 major core compounds had a good binding affinity for the five core target proteins. The binding energy of paeoniflorgenone was the lowest among the 10 compounds, indicating its good binding affinity with all the five target proteins. The paeoniflorgenone-EGFR showed the lowest binding energy (-4.55) and ASN-842 was the active site for hydrogen bonding.

In order to verify whether the 10 components predicted by molecular docking exist in ZWD, UPLC-Q-TOFMS was performed to analyze the chemical constituents of ZWD. Full scanning was carried out in positive and negative ion mode and the analysis showed that the chromatogram of representative Base Peak Intensity (BPI) in positive and negative ion mode was as shown in fig. 10. A total of 67 compounds were detected by UniFi software (Table 3). Among them, paeoniflorin, paeoniflorgenone and poricoic acid B were proved to exist in ZWD. Therefore, it can be speculated that paeoniflorin, paeoniflorin and poricoic acid B may play a major role in ZWD in the treatment of NAFLD.


Fig. 10: Representative base peak chromatogram of ZWD in positive and negative ionization mode, respectively, (A) Negative ionization mode and (B) positive ionization mode

Component name Neutral mass  (Da) Observed neutral mass  (Da) Observed m/z Mass error  (mDa) Mass error (ppm) Observed RT  (min) Detector counts Response Adducts Total  fragments found
Furfuryl alcohol 98.03678 98.0374 143.0356 0.6 4.4 0.43 430 430 HCOO+ 1
Epinephrine 416.15299 416.1552 455.1184 2.2 4.9 0.43 1149 1149 K+ 40
16-Oxoacetylpyrochloric acid 570.39204 570.3945 609.3576 2.4 4 0.45 388 388 K+ 42
Paradisamide 421.28282 421.2832 444.2724 0.3 0.8 0.46 1283 400 Na+ 50
Coryneine 195.12593 195.1268 196.1341 0.9 4.4 0.46 619 619 H+ 12
1-Galloyl-β-D-glucose 332.07435 332.0751 331.0678 0.7 2.2 0.62 8711 7759 H- 19
Gallic acid 170.02152 170.022 169.0148 0.5 3 0.74 42699 40055 H- 2
5-Hydroxymethyl-2-furaldehyde 126.03169 126.0322 125.025 0.5 4.4 0.74 6362 6017 H- 3
1-O-β-D-Glucopyranosylpaeonisuffrone 360.14203 360.1432 405.1414 1.1 2.8 0.81 17152 8794 HCOO+, H- 12
Carmicheline 377.25661 377.2566 378.2639 0 0.1 1.24 395 395 H+ 0
Senbusine  B 423.26209 423.2619 424.2692 -0.2 -0.4 1.91 772 772 H+ 1
Atractylenolide II 232.14633 232.147 233.1542 0.6 2.7 2.01 2012 1710 H+ 5
d-Catechin 290.07904 290.08 289.0727 0.9 3.2 2.24 26247 22781 H- 8
Nonyn 357.23039 357.231 358.2383 0.7 1.8 2.25 20376 16960 H+ 1
Benzoic acid 122.03678 122.0372 123.0445 0.4 3.6 2.26 496 496 H+ 1
Dictysine 347.24604 347.247 348.2542 0.9 2.6 2.27 433 433 H+ 2
Oxypaeoniflorin 496.15808 496.1591 495.1518 1 2.1 2.28 44480 35623 H-, HCOO+ 21
Paeoniflorin 480.16316 480.1649 481.1721 1.7 3.5 3.04 2749 2186 H+ 10
Fuziline 453.27265 453.2739 454.2812 1.3 2.8 3.08 1526 1526 H+ 3
Nyonin 437.27774 437.2789 438.2862 1.2 2.7 3.18 33673 27072 H+ 6
Kaempferol-3,7-di-O-β-D-glucoside 610.15338 610.1551 609.1478 1.7 2.8 3.21 326 326 H- 1
β-Eudesmol 222.19837 222.1978 245.187 -0.6 -2.4 3.27 697 697 Na+ 0
6-O-β-D-Glucopyranosyllactinolide 362.15768 362.1584 361.1511 0.7 2 3.31 1142 1142 H- 14
(Z)-(1S,5R)-β-Pinen-10-yl-β-vicanoside 446.2152 446.2168 491.215 1.6 3.3 3.62 563 563 HCOO+ 1
Denudatine 343.25113 343.2524 344.2597 1.3 3.7 3.72 4188 3440 H+ 4
Ethyl salicylate 166.06299 166.0637 165.0565 0.7 4.4 3.93 2153 2153 H- 14
Poricoic acid B 196.07356 196.0744 197.0817 0.8 4.2 3.94 9904 9086 H+ 18
Benzyl alcohol 108.05751 108.058 109.0652 0.5 4.2 3.94 2463 2463 H+ 4
Delbruline 479.2883 479.2903 480.2976 2 4.2 4.42 373 373 H+ 3
Paeoniflorigenone 318.11034 318.1115 319.1188 1.2 3.7 5.3 1695 1695 H+ 2
Lactiflorin 462.1526 462.1546 507.1528 2 3.9 6.27 1006 785 HCOO+ 0
Paeonilactone C 318.11034 318.1119 319.1192 1.6 4.9 7.86 2726 2355 H+ 7
Hokbusine A 603.30435 603.3056 604.3128 1.2 2 10.17 48550 36036 H+ 5
Benzoylaconitine 603.30435 603.3057 648.3039 1.3 2.1 10.18 5395 3794 HCOO+ 2
Benzoylhypaconine 573.29378 573.2947 574.302 0.9 1.6 11.42 72877 54681 H+ 5
Atractylenolide IV 306.14672 306.148 305.1407 1.2 4 12.27 277 277 H- 7
Beiwutine 647.29418 647.2967 648.304 2.5 3.9 12.65 1232 838 H+ 1
Benzoyloxypaeoniflorin 600.18429 600.1854 599.1781 1.1 1.8 13.53 602 602 H- 25
3-Deoxynivalenol 629.32 629.3219 628.3146 1.9 3 13.84 417 417 H- 5
Aconitine 645.31491 645.3173 646.3246 2.4 3.7 14.02 538 538 H+ 4
Benzoylpaeoniflorin 584.18938 584.1922 585.1995 2.8 4.8 14.48 464 464 H+ 2
Palbinone 358.21441 358.215 357.2077 0.6 1.7 14.49 319 319 H- 1
Poricoic acid A 512.35017 512.3483 551.3114 -1.9 -3.5 14.77 856 856 K+ 3
Delbruine 465.27265 465.2744 466.2816 1.7 3.7 14.8 1321 901 H+ 3
Paeonilactone 168.11503 168.1156 169.1229 0.6 3.5 14.87 1947 1947 H+ 5
6-Gingerol 294.18311 294.184 317.1732 0.9 2.8 14.99 5834 5014 Na+ 38
Hypaconitine 615.30435 615.3066 638.2958 2.3 3.6 15 393 393 Na+ 14
Poricoic acid D 514.32944 514.3301 513.3229 0.7 1.3 15.06 905 642 H- 2
25-Hydroxy-3-epidehydrotumulosic acid 500.35017 500.3498 499.3425 -0.4 -0.8 15.23 1282 955 H- 3
Poricoic acid B 484.31887 484.3196 483.3123 0.7 1.4 15.25 217 217 H- 0
Hydroxy-atractylenolide 248.14124 248.1425 247.1352 1.2 5 15.26 222 222 H- 0
3β-hydroxy-atractylon 232.14633 232.1473 233.1545 0.9 4 16.03 7081 6265 H+ 4
3β,16α-Dihydroxylauroster-8,24-dien-21-oic acid 472.35526 472.356 471.3487 0.8 1.6 16.2 217 217 H- 3
Poricoic acid F 498.33452 498.3365 497.3293 2 4 16.24 309 309 H- 4
3-Epidehydrotumulosic acid 484.35526 484.3564 483.3491 1.1 2.3 16.39 452 452 H- 3
Dehydrotumulosic acid 486.37091 486.3718 485.3645 0.9 1.8 16.48 386 386 H- 3
Atractylenolide A 230.13068 230.1315 231.1387 0.8 3.4 16.53 438 438 H+ 2
3β-Hydroxy-11α,12α-epoxyolean-28-13β-olide 486.33452 486.3369 485.3296 2.4 4.8 16.61 338 338 H- 2
Ginger brain 276.17254 276.1732 277.1805 0.7 2.5 16.7 4270 3648 H+ 4
Poricoic acid C 482.33961 482.3401 481.3328 0.4 0.9 16.77 353 353 H- 1
β-Red myrcene 204.1878 204.1885 205.1958 0.7 3.6 17.08 431 431 H+ 0
Pachymic acid 528.38147 528.3828 527.3756 1.4 2.6 17.67 485 485 H- 2
β-aqua-anisidine 136.1252 136.1258 137.1331 0.6 4.4 18.18 1095 1095 H+ 14
Salsolinol 193.11028 193.1109 194.1182 0.7 3.5 18.55 2682 2682 H+ 2
Eburicoic acid 470.376 470.3781 493.3674 2.1 4.3 18.6 646 646 Na+ 4
Hexadecane 256.24023 256.2408 255.2336 0.6 2.4 18.82 15674 13670 H- 7
γ-Sitosterol 414.38617 414.3875 453.3506 1.3 2.9 18.88 1913 1913 K+ 1

Table 3: Identification of The Compounds of ZWD by UPLC-Q-TOF/MS Analysis

Referring to the relevant literature[24] and the experimental results in this study, 1 mM FFA was selected as model concentration. As shown in fig. 11, the results of CCK-8 assay showed that ZWD at concentrations ranging from 9.75-156 mg/ml+1 mM FFA displayed no cytotoxicity on cells.


Fig. 11: Cell viability after treatment with the different concentrations of ZWD and 1 mM FFA, **p<0.01 and ****p<0.0001 vs. control group

Treatment with FFA can induce significant lipid accumulation in cells. After the cells were treated with the different doses of ZWD for 24 h, the ORO staining showed a significant difference between the experimental group and model group, there were obvious differences when the cells were treated with 19.5 mg/ml and 39 mg/ml of ZWD. The TG contents in HepG2 cells in the experimental group were lower than those in the control group, as shown in fig. 12. These results suggested that 19.5 mg/ml (p≤0.05) and 39 mg/ ml (p≤0.01) of ZWD could reduce lipid accumulation in a concentration-dependent manner.


Fig. 12: Lipid concentration of HepG2 cells treated with the different concentrations of ZWD and 1 mM FFA, *p<0.05 and **p<0.01 vs. model group; ##p<0.01 vs. control group. All experiments were repeated at least two times

The results of cell-based assays showed that 1 mM FFA could significantly induce lipid accumulation in the HepG2 cells. After the intervention of ZWD, cell viability enhanced, which might be due to the effect of Fuzi because it significantly promote cell viability even in low concentration[25]. The pathogenesis of NAFLD is a ‘three-hit’ process, including steatosis, lipotoxicity and inflammation, which impair the hepatocyte viability, eventually leading to death and exacerbation of NAFLD[26]. Therefore, it was speculated that the improvement of cell vitality by ZWD might be the mechanism of its therapeutic effect. The intracellular lipid contents were decreased by ZWD in a concentration-dependent manner, where 19.5 mg/ml (p≤0.05) and 39 mg/ml (p≤0.01) of ZWD could reduce lipid accumulation. For validation purposes, RTqPCR was performed to test the expression of EGFR, PPARG, HSP90AA1, MAPK1 and PI3K. RT-qPCR experiments proved ZWD induced down-regulation of EGFR mRNA levels and up-regulation of HSP90AA1, MAPK1 and PI3K mRNA levels.

The network pharmacology results identified potential key targets and pathways of ZWD active against NAFLD. In order to verify the reliability of network pharmacological prediction results, we detected the mRNA level of these key genes by qRT-PCR. As shown in fig. 13, RT-qPCR results showed that mRNA expression of EGFR and PPARG was increased in the model group compared with that in the control group, and after three concentration gradient of ZWD treated for 24 h, the expression of EGFR mRNA was decreased, the expression of PPARG mRNA showed that there was an inconsistent change trend between high concentration and low concentration. The mRNA expression of HSP90AA1, MAPK1 and PI3K was decreased in the model group compared with that in the control group, and after treatment, the mRNA expression of HSP90AA1, MAPK1 and PI3K was increased.


Fig. 13:The expression of NAFLD-related genes after treatment with ZWD, *p<0.05, **p<0.01 and ***p<0.001 vs. model group and #p<0.05, ##p<0.01 vs. control group

The results of our compound-target-pathway suggested that the active ingredients in ZWD could interact with the PI3K-AKT signaling pathway through EGFR, HSP90AA1 and MAPK1. The activated PI3K/AKT signaling pathway can exert anti-inflammatory, anti-oxidative stress, anti-apoptotic and autophagic regulatory effects through downstream pathways and related proteins[27], these processes are closely related to the pathogenesis of NAFLD. MAPK1 has been shown to be involved in cellular processes of autophagy, lipid metabolism, proliferation, migration[28] and the activation of MAPK 1/3 ameliorates liver steatosis in leptin receptor-deficient mice[29]. The function of HSP90AA1 is to promote the maturation, structural maintenance and proper regulation of specific target proteins involved in cell cycle control and signal transduction[30]. Thus, our findings show that ZWD can positively regulate NAFLD in multiple ways through active ingredients and targets.

In summary, this study demonstrated a strategy to optimize conventional network pharmacology and explained the molecular mechanism of ZWD in NAFLD treatment was closely associated with 5 core genes (including AKT1, EGFR, CASP3, PPARG and IGF1R), which were involved in important signal pathways (including FOXO signaling pathway and PI3K-AKT signaling pathway). RT-qPCR analyses revealed that the therapeutic effect of ZWD might be achieved via down-regulation of EGFR mRNA levels and up-regulation of HSP90AA1, MAPK1 and PI3K mRNA levels.

Author’s contributions:

Chun-Jiang Zhang and Zhi-Yan Fang contributed equally to this work. Tian-e Zhang, Yan-Qiu Wang and Chun-Jiang Zhang: Conceptualization; Chun-Jiang Zhang, Zhi-Yan Fang, Zheng Luo, Hai-Yan Zhu and Lili Huang: Methodology and visualization; Wen-Ying Huai and Yan-Qiu Wang: Software; Lili Huang and Yan-Qiu Wang: in vitro cells-based experiment; Chun- Jiang Zhang and Tian-e Zhang: Writing-original draft; Tian-e Zhang and Yu-Qin Tang: Writing-review and editing; Tian-e Zhang and Qiao-Zhi Yin: Supervision and project administration. All authors have read and agreed to the published version of the manuscript.


This study was supported by the Key Research and Development Program of Science and Technology Department of Sichuan Province (grant number 2020JDZH0018), The fourth National TCM Doctor (Clinical And Basic) Excellent Talents Research and Study Program (grant number CMM No. 24 [2017]), the Sichuan Provincial Famous TCM Doctor Studio Construction Project (Sichuan TCM 003113005003), the Special Research Project of Sichuan Provincial Administration of Traditional Chinese Medicine (grant number 2021MS099), Technological innovation research and development projects of Chengdu Science and Technology Bureau (Grant number 2021-YF05- 02379-SN) and the “Double First-class” construction project of Chengdu University of Traditional Chinese Medicine (030041043).

Conflict of interests:

The authors declared that there are no conflicts of interest regarding the publication of this paper.