The active components of Erzhi wan and their anti-Alzheimer's disease mechanisms determined by an integrative approach of network pharmacology, bioinformatics, molecular docking, and molecular dynamics simulation

Abstract

Erzhi Wan (EZW), a classic Traditional Chinese Medicine formula, has shown promise as a potential therapeutic option for Alzheimer's disease (AD), yet its mechanism remains elusive. Herein, we employed an integrative in-silico approach to investigate the active components and their mechanisms against AD. We screened four active components with blood-brain barrier permeabilities from TCMSP, along with 307 corresponding targets predicted by SwissTargetPrediction, PharmMapper, and TCMbank websites. Then, we retrieved 2260 AD-related targets from Genecards, OMIM, and NCBI databases. Furthermore, we constructed the protein-protein interaction (PPI) network of the intersected targets via the STRING database and performed the GO and KEGG enrichment analyses using the "clusterProfiler" R package. The results showed that the intersected targets were intimately related to the p53/PI3K/Akt signaling pathway, serotonergic synapse, and response to oxygen level. Subsequently, 25 core targets were found differentially expressed in brain regions by bioinformatics analyses of GEO datasets of clinical samples from the Alzdata database. The binding sites and stabilities between the active components and the core targets were investigated by the molecular docking approach using Autodock 4.2.6 software, followed by pocket detection and druggability assessment via the DoGSiteScorer server. The results showed that acacetin, β-sitosterol, and 3-O-acetyldammarenediol-II strongly interacted with the druggable pockets of AR, CASP8, POLB, and PREP. Eventually, the docking results were further cross-referenced with the literature research and validated by 100 ns of molecular dynamics simulations using GROMACS software. Binding free energies were calculated via MM/PBSA strategy combined with interaction entropy. The simulation results indicated stable bindings between four docking pairs including acacetin-AR, acacetin-CASP8, β-sitosterol-POLB, and 3-O-acetyldammarenediol-II-PREP. Overall, our study demonstrated a theoretical basis for how three active components of EZW confer efficacy against AD. It provides a promising reference for subsequent research regarding drug discoveries and clinical applications.

Keywords: AD pathology; Computational biology; Erzhi wan; Network pharmacology; Oxidative stress; p53/PI3K/Akt signaling pathway.

Graphical abstract

Fig. 1

The workflow of the integrated computational approach to study the mechanism of Erzhi Wan (EZW) against Alzheimer's disease (AD). The workflow contains four sections including target screening from database retrieval, network construction and analysis, bioinformatics analysis upon Gene Expression Omnibus (GEO) datasets, and molecular docking and molecular dynamics simulation verifications.

Fig. 2

Initial exploration of the potential efficacy that Erzhi Wan (EZW) exhibited against Alzheimer's disease (AD). (A) Venn diagram displaying the common targets of EZW and AD. (B) Bubble diagram of the KEGG enrichment pathways in the ascending order of adjusted P-value per category. (C) UpsetR plot displaying the intersections of KEGG enrichment sets.

Fig. 3

The anti-AD Targets-Component-Herb network. The octagons represent herbs. The squares and the circles represent the active components and their corresponding intersected genes respectively, while the node sizes are proportional to the degrees independently between groups.

Fig. 4

The protein-protein interaction (PPI) analysis of the 140 potential targets. (A) PPI network. The nodes represent the target genes, while the node sizes are proportional to the node degrees. The top 10 targets ranked by degrees are highlighted in purple. (B–D) Clusters 1–3 identified by clustery analysis of MCODE. The seed nodes of individual cluster are displayed in red. Seed node was not assigned for cluster 2 by the MCODE algorithm. The KEGG enriched terms of each cluster are listed below the graphic. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 5

The results of GO and KEGG enrichment analysis. (A) Bubble diagram of the top four KEGG enrichment pathways in the ascending order of logP values per category. (B) GO enrichment terms. Top 10 GO terms of biological process (BP), cellular components (CC), and molecular functions (MF) are displayed in the ascending order of logP values individually. The discussed BP terms are highlighted in red square. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 6

Identification of the core targets correlating with Aβ and tau pathology based on bioinformatics analysis. (A) Circular bar plot displaying the targets that significantly correlated with Aβ and tau pathology. The bar sizes represent the convergent functional genomic (CFG) points. (B) The volcano plot showing the differentially expressed genes (DEGs) between control and Alzheimer's disease (AD) patients. (C–G) Results manifesting DEGs expression profiles from the GEO datasets. For the hippocampus, sample sizes are n = 66 and n = 74 in the control and AD patient group respectively. Results are shown in mean ± SD. (H) Venn diagram showing the targets distribution in 4 brain regions. (I) Interaction network displaying connections among DEGs, active components, and herbs. The circle, square, and the octagons are DEGs, active components, and herbs, respectively.

Fig. 7

Results of molecular docking. (A) Heatmap of the binding affinities (kcal/mol) between the active components and the core targets. (B–F) The graphic illustrations of the binding sites in two- and three dimensions (2D and 3D). (B) AR-acacetin. (C) AR-pratensein. (D) CASP8-acacetin. (E) POLB-β-sitosterol. (F) PREP-3-O-acetyldammarenediol-II. The binding pocket, zoomed-in views of hydrogen bonds in binding sites, ligand-interacted residues, and the corresponding interaction types are shown from left to right.

Fig. 8

Results of molecular dynamics (MD) simulation. (A) Root mean square deviation (RMSD). (B) Root mean square fluctuation (RMSF). Intensely fluctuated residues are highlighted. (C) Radius of gyration (ROG). (D) Solvent-accessible surface area (SASA). (E) Number of hydrogen bonds across the simulation period. (F) Evolution of molecular mechanics/Poisson-Boltzmann surface area (MM/PBSA) energies in last 50 ns simulations. (G) Per-residue decompositions of MM/PBSA energies. (H–I) The two-dimensional free energy landscape (FEL). The maximum and minimum energy basins are depicted in red and blue, respectively. The ligand-interacted residues and the corresponding interaction types of minimum energy conformations are illustrated in two- and three dimensions. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)