Systematic investigation of the mechanism of herbal medicines for the treatment of prostate cancer

Abstract

Due to various unpleasant side effects and general ineffectiveness of current treatments for prostate cancer (PCa), more and more people with PCa try to look for complementary and alternative medicine such as herbal medicine. However, since herbal medicine has multi-components, multi-targets and multi-pathways features, its underlying molecular mechanism of action is not yet known and still needs to be systematically explored. Presently, a comprehensive approach consisting of bibliometric analysis, pharmacokinetic assessment, target prediction and network construction is firstly performed to obtain PCa-related herbal medicines and their corresponding candidate compounds and potential targets. Subsequently, a total of 20 overlapping genes between DEGs in PCa patients and the target genes of the PCa-related herbs, as well as five hub genes, i.e., CCNA2, CDK2, CTH, DPP4 and SRC were determined employing bioinformatics analysis. Further, the roles of these hub genes in PCa were also investigated through survival analysis and tumour immunity analysis. Moreover, to validate the reliability of the C-T interactions and to further explore the binding modes between ingredients and their targets, the molecular dynamics (MD) simulations were carried out. Finally, based on the modularization of the biological network, four signaling pathways, i.e., PI3K-Akt, MAPK, p53 and cell cycle were integrated to further analyze the therapeutic mechanism of PCa-related herbal medicine. All the results show the mechanism of action of herbal medicines on treating PCa from the molecular to systematic levels, providing a reference for the treatment of complex diseases using TCM.

Keywords: MD simulations; bioinformatics analysis; herbal medicines; mechanism of action; prostate cancer.

Figure 1

The detailed workflow of this study.

Figure 2

(A) Network map showing keywords in PCa-related literature. (B) Pie chart of the proportion of TCM targets in the treatment of PCa. (C, D) Violin diagram of OB and DL content. (E) Key target gene in herbs. (F) Correlation map of herbs target genes. (G) Degree value of target genes in the treatment of prostate cancer with TCM.

Figure 3

(A) The C-T network. Circles and octagons represent the chemicals from ten herbals and all their potential targets, respectively. (B) T-F network. 28 targets related to direct anticancer therapy 39 targets related to immunization therapy and 28 targets related to inflammation therapy. The overlapped targets in the middle are the common targets of all three therapies. (C) The heatmap of C-T interaction analysis. (D) The heatmap of T-F network analysis.

Figure 4

(A) The volcano plots for the 1880 DEGs in the GSE134073 dataset. (B) The overlapped genes between DEGs in PCa and the target genes of the PCa-related herbs. (C) The correlation between overlapped genes and drugs. (D) Heatmaps of significantly differentially expressed genes based on high Gleason score and low Gleason score groups. (E) A heatmap of the correlation of overlapped genes. (F) Gene Ontology (GO) enrichment of overlapped genes. (G) Visualized functional enrichment and gene interaction analysis results.

Figure 5

(A–F) The expression of five hub genes in TCGA database between PCa and peritumoral tissues. (G–K) Kaplan-Meier survival analysis of five prognostic genes in TCGA cohort. (L) Immunohistochemical analysis of five genes in HPA database.

Figure 6

Using the TIMER dataset to analyze the relationship between the expression of five hub genes and tumor purity and immune cells. (A) CCNA2, (B) CDK2, (C) CTH, (D) DPP4, (E) SRC.

Figure 7

The heat map of the docking results between the active compounds of ten herbs and five hub targets. (A) Lobeliae Chinensis Herba. (B) Hedyotis Diffusae Herba. (C) Scutellariae Barbatae Herba. (D) Atractylodes Macrocephala Koidz. (E) Curcumae Rhizoma. (F) Poria Cocos (Schw.)Wolf. (G) Hedysarum Multijugum Maxim. (H) Cornus Officinalis Sieb. Et Zucc. (I) Coicis Semen (J) Licorice.

Figure 8

(A) The RMSD for CDK2 with Eriodictyol over 10 ns. (B–E) The binding mode of the CDK2-Eriodictyol complex. (F) The RMSF for CDK2-Eriodictyol complex. (G, H) The energy of the key amino acids from CDK2-Eriodictyol complex.

Figure 9

(A) The RMSD for DPP4 with Calycosin over 10 ns. (B–E) The binding mode of the DPP4-Calycosin complex. (F) The RMSF for DPP4-Calycosin complex. (G, H) The energy of the key amino acids from DPP4-Calycosin complex.

Figure 10

(A) The Compounds pathway network. The target node and the pathway node are linked. The yellow rectangle, red circle and octagon represent the target, pathway and compound respectively. (B) Distribution of ten herbs targets in compressed PCa pathway.