Screening active ingredients of RP
After combining the RP ingredients found in the following databases, there were 40 components, of which 15 components were retrieved from the TCPSP database, 12 components were retrieved from the ETCM database, 10 components were retrieved from the herb database and 3 components were retrieved from the literature (Supplementary Table S1). The retrieved ingredients were submitted to the admetSAR website (http://lmmd.ecust.edu.cn/admetsar2) for further screening based on the results of human intestinal absorption (HIA), Caco-2, human oral bioavailability (HOB), and plasma protein binding (PPB) (Supplementary Table S2). The results showed that there were 9 components with good absorption and distribution properties, and their chemical constituents are mainly isoflavone, coumarins, and alkaloids (Table 1). Although the Caco-2 and HOB of puerarin predicted by admetSAR were lower, it has an important biological role in previous studies, so it was temporarily retained for further study.
Intersection of Related Targets
SwissTargetPrediction is a network tool designed to predict the most likely protein targets in small molecules by reverse screening based on similarity principles. Submit the SMILES descriptions of puerarin and the screened active ingredients to the SwissTargetPrediction database and output the results with the similarity probability >0 to the target (Supplementary Table S3). By combining the targets of active ingredient and deleting duplicates, we obtained 226 targets related to RP. A total of 21,894 colon cancer-related targets were obtained from GeneCards and OMIM. Then we inputted RP and CC targets to Venny 2.1, and the results showed that there were 219 common gene targets between RP and CC (Supplementary Table S4) (Fig. 2).
Analysis of Target PPI Network
The STRING database is used to show the link between proteins participating in specific biological functions. We entered common targets into the "Multiple proteins" of STRING database, selected the highest confidence level and hided the disconnected nodes in the network to obtain the interaction between them. Then the results of TSV format were imported into Cytoscape, and 104 nodes and 218 edges were displayed (Fig. 3). By calculating and visualizing the targets degree values in PPI network, the genes with higher degree are SRC, LYN, JAK2, MAPK14, MAPK8, PTK2, PTK2B, EGFR, NFKB1, JAK1, PTPN6, SYK, FGR and ESR1.
GO and KEGG Analysis
We used Metascape for GO function and KEGG pathway analysis to further understand the mechanism of RP on CC, and set P Value <0.01 to be significant. GO function results includes three parts: biological process (BP), cell component (CC) and molecular function (MF). In BP GO terms, peptidyl-tyrosine phosphorylation, regulation of cell adhesion, response to drug et al. may be associated with tumor regulation. CC terms are mainly enriched in perinuclear region of cytoplasm, membrane raft, focal adhesion, etc. In MF GO terms, protein kinase activity, kinase binding, and transmembrane receptor protein tyrosine kinase activity may be associated with tumors(Fig. 4). In KEGG enrichment pathways, EGFR tyrosine kinase inhibitor resistance, PI3K-Akt signaling pathway, apoptosis, and NF-kappa B signaling pathway is involved in apoptosis and cancer regulation (Fig. 5). These pathways may be critical in the treatment of CC. The results of KEGG analysis were compared with PPI network, and 7 targets with higher degree were randomly selected for receptor proteins in molecular docking, including SRC, JAK2, MAPK14, EGFR, NFKB1, ESR1, and IL2.
Compounds Target Network Analysis
The first 20 pathways with the largest number of genes were chosen to construct a RP-components-targets-pathways network with 183 nodes and 626 edges (Supplementary Table S5)(Fig. 6). The red node refers to drug; The green node represents pathways; The blue nodes represent targets; The yellow nodes indicate components. The edges indicate their interactions. Each compound interacts with multiple targets in the graph, suggesting that the effect of RP on CC may be a synergistic effect of multiple targets. According to the degree of topological parameters of network, five high-level components were selected as ligand molecules in molecular docking. Of the five ingredients, three isoflavones component, namely formononetin, daidzein, and 3'-Methoxydaidzein; one alkaloid component, namely sitosterol; and one coumarin component, namely scoparone.
Molecular Docking Analysis
The PDB Entry number of the targets structure selected from the PDB database is EGFR (5UG9), JAK2(3UGC), MAPK14(2FST), NFKB1(1SVC), ESR1(3CBP), IL2(4NEJ), and SRC (1O43). We conducted molecular docking between receptor proteins and ligand molecules through AutoDockTools 1.5.6. The result of AutoDockTools was output in the form of affinity score, which is the core parameter of AutoDockTools (Table 2). The lower the affinity score, the better the binding effect. PYMOL software visualized the docking complexes and binding residues of 3'- Methoxydaidzein and formononetin ligand molecule with MAPK14 receptor protein (Fig. 7).
RP inhibits proliferation of colon cancer cells
The anti-cancer effect of RP at different concentrations (0, 5, 10, 15, and 20 μg/ml) on SW480 cells for 24,48 and 72 hours was verified by CCK-8 experiment. The proliferation of SW480 cells treated with different concentrations of RP decreased in a dose- and time-dependent manner(Fig. 8). The proliferation ability of SW480 cells decreased linearly under low concentration of RP, but did not change much under high concentration of RP. The IC50 values of 24,48 and 72 h after RP treatment were 14.9, 9.8, and 8.0 μg/ml in SW480 cells, respectively.
RP inhibits migration of colon cancer cells
The effect of RP on SW480 cells migration was studied by scratch test (Fig. 9). As shown in the figure, at 12h, the cell scratches of the three groups were reduced by different treatment methods, and the reduction range of the control group was the largest; at 48h, the control group was further reduced by a large margin, but there was no significant change in the two groups after treatment with RP medium. The results showed that the effect of RP on SW480 cell migration was more significant over time than that of the control group. It suggested that RP could inhibit the invasion and migration activity of SW480 cells.