Growth inhibition of AGS on CRC
MTT assay was used to evaluate the influence of AGS on the viability of the CRC cells. As shown in Fig. 2, treatments with PEAGS, EAAGS, NBAGS (0 ~ 1000 μg/mL) for 48 h resulted in the significant inhibition of proliferation on HCT116 and SW620 cells. The IC50 (50% inhibitory concentration) values of NBAGS, EAAGS, PEAGS were 197.24, 264.85, 15.45 μg/mL on HCT116 cells (Fig. 2a), and 523.6, 323.59, 150.31 μg/mL on SW620 cells (Fig. 2b), respectively, indicating that all the three fractions of AGS could significantly inhibit the growth of CRC cells. These data suggested that AGS would be a good candidate for CRC prevention or therapy.
Collection forchemical structure and targetsinformation of AGS
Network pharmacology analysis was performed to explore the interactions between AGS constituents and potential targets. 11 active ingredients including 4-O-galloylbergenin, 11-O-galloylbergenin, 11-O-protocatechuoylbergenin, 11-O-syringylbergenin, ardisiacrispin B, bergenin, epicatechin-3-gallate, gallic acid, quercetin, stigmasterol, stigmasterol-3-O-β-D-glucopyranoside were identified by database searching (Fig. 3), and the numbers of the targets for 4-O-galloylbergenin, 11-O-galloylbergenin, 11-O-protocatechuoylbergenin, 11-O-syringylbergenin, ardisiacrispin B, bergenin, epicatechin-3-gallate, gallic acid, quercetin, stigmasterol, stigmasterol-3-o-β-D-glucopyranoside were 69, 86, 79, 79, 51, 41, 111, 17, 54, 56, 88, respectively (Supplementary Table S1). A total of 173 targets related to the bioactive components were picked out after deleting the reappeared targets.
Collection for disease targets
A total of 21572 targets related to CRC disease were obtained by searching the Genecards database (Supplementary Table S2). The top 10 “high response” genes were screened out according to the score of relevance, including BRCA2, BRCA1, TP53, MSH2, APC, MSH6, MLH1, CDH1, PTEN, PMS2 .
Prediction for candidate targets of AGS against CRC
As the Venn diagram showed in Fig. 4a, a total of 170 overlapped genes (Supplementary Table S3) were identified by matching the therapeutic target genes of CRC and target genes of AGS. Cytoscape 3.7.2 software was applied to construct the “drug-component-target-disease” network. The “SJT-component-target-RA” network was built up by importing the crossover genes of AGS & CRC and the potential active components into the system (Fig. 4b).
PPI network construction and drug-disease key targets prediction
Based on the crossover targets of AGS & CRC, the PPI network was built up by the String database. As shown in Fig. 5a, there were 169 nodes and 239 edges in the network diagram and the average node degree was 2.83. The top 30 core genes were screened out, of which the node degrees of SRC, MAPK1, ESR1, HSP90AA1, MAPK8 were greater than 12 (Fig. 5b). According to the results, the core targets were predicted to be SRC, MAPK1, ESR1, HSP90AA1, MAPK8, which had more connections than other genes.
Enrichment analysis for key targets
As a result, GO analysis showed that the numbers of BP, CC, and MF of AGS against CRC were 1079, 44, and 132, respectively (Supplementary Table S4), and the top 20 GO analysis of BP, CC, MF had been shown as graphical bubbles. According to the results, AGS would mainly participate in the biological process of steroid metabolic process (Fig. 6a) and the cytoplasm is the major reaction site (Fig. 6b) in the treatment of CRC, during which the central molecular function would probably include the steroid hormone receptor activity, nuclear receptor activity, transcription factor activity, steroid binding and endopeptidase activity (Fig. 6c). Furthermore, a total of 96 signaling pathways (Supplementary Table S5) were screened out through KEGG pathway enrichment analysis, and the top 20 signaling pathways were shown as a bar graph (Fig. 6d), among which the MAPK signaling pathway, Lipid, and atherosclerosis, Proteoglycans in cancer, Prostate cancer, Adherens junction, Endocrine resistance, Progesterone-mediated oocyte maturation, Relaxin signaling pathway, FoxO signaling pathway, Apoptosis had been proved to be the major pathways related to CRC treatment.
Molecular docking analysis
The docking analysis was performed to assess the binding effect and pattern between the active ingredients of AGS and the identified core targets. The results of SP molecular docking between SRC (PBD ID: 3G5D) and the 6 characteristic active components were shown in Table 1. The active sites of SRC were x: 9.27, y: -37.78, z: -4.32, determined by the protein-ligand. Four active compounds, including 11-O-galloylbergenin, 11-O-protocatechuoylbergenin, 11-O-syringylbergenin, epicatechin-3-gallate, can interact with SRC well, with the docking scores ranging from -7.965 to -6.595. Compared with the SRC ligand (1N1, -6.5), the docking scores of 11-O-protocatechuoylbergenin, epicatechin-3-gallate interacting with SRC were higher, with the docking score of -7.082, -7.965. As shown in Fig. 7a, 11-O-protocatechuoylbergenin could form H bonds with ALA-390 (2.9 Å), ASN-391 (3.1 Å), LYS-295 (3.0 Å), THR-338 (3.3 Å), ILE-336 (3.4 Å), and epicatechin-3-gallate (Fig. 7b) could develop H bonds with GLU-310 (3.1Å), GLU-280 (3.1 Å), ASP-404 (3.1 Å), GLN-275 (3.0 Å), MET-341 (2.7 Å and 2.9 Å). The results of SP molecular docking between MAPK1 (PBD ID: 1PME) and the 6 characteristic active components were shown in Table 1. The active sites of the MAPK1 were x: -12.73, y: 13.48, z: 40.75, determined by the MAPK1 protein-ligand (SB2). All the docking scores of the active compounds were lower than that of the MAPK1 ligand (SB2, -7.509). 11-O-galloylbergenin, epicatechin-3-gallate had relative high-level of interactions with MAPK1, with docking scores of -6.571, -6.43. As shown in Fig. 7c, 11-O-galloylbergenin could develop H bonds with ASN-154(2.8 Å), LEU-103(3.1 Å), LYS-54(3.3 Å), ASP-111(2.8 Å), LYS-114(3.0 Å), and epicatechin-3-gallate(Fig. 7d) developed H bonds with ASN-154(2.6 Å), ALA-52(3.4 Å), ASP-167(3.4 Å), GLY-37(2.8 Å). The SP molecular docking results of ESR1 (PBD ID: 1A52) with the 6 characteristic active components were shown in Table 1. The active sites of the ESR1 were x: 95.12, y: 92.31, z: 109.75, determined by the ESR1 protein-ligand. The 4-O-galloylbergenin failed to dock with ESR1. According to the docking scores results, 11-O-galloylbergenin, 11-O-protocatechuoylbergenin, 11-O-syringylbergenin could not interact with ESR1 well, with the scores ranging from -5.91 to -1.7, while bergenin, epicatechin-3-gallate had greater levels of interactions with ESR1, with the docking scores of -8.133, -8.797, higher than the docking score of the ligand (EST, -6.7). As shown in Fig. 7e, bergenin could develop H bonds with HIS-524 (2.7 Å and 2.8 Å) and THR-347 (3.0 Å), and epicatechin-3-gallate (Fig. 7f) could form H bonds with GLU-353 (4.9 Å and 2.5 Å), LEU-387 (3.3 Å), ASP-351 (2.6 Å and 3.2 Å), THR-347 (2.7 Å), LYS-529 (2.8 Å). The SP molecular docking results of HSP90AA1 (PBD ID: 7lt0) with the 6 characteristic active components were shown in Table 1. The active sites of the HSP90AA1 were x: -31.94, y: -10.74, z: -25.24, determined by the HSP90AA1 protein-ligand. According to the docking scores, all 6 active compounds didn’t interact with HSP90AA1 well, with the scores ranging from -8.847 to -6.955, lower than the docking score of the ligand (ONJ, -9.144). 11-O-protocatechuoylbergenin, 11-O-syringylbergenin had relatively high levels of interactions with HSP90AA1, with docking scores of -8.847, -8.288. As shown in Fig. 7g, 11-O-protocatechuoylbergenin could develop H bonds with GLY-135 (2.8 Å and 2.8 Å), SER-52 (3.1 Å), ASP-93 (2.9 Å), LEU-103 (3.2 Å), and 11-O-syringylbergenin (Fig. 7h) could form H bonds with GLY-135 (2.7 Å), THR-184 (3.1Å). The SP molecular docking results of MAPK8 (PBD ID: 1UKI) with the 6 characteristic active components were shown in Table 1. The active sites of MAPK8 were x: 2.22, y: 39.06, z: 29.48, determined by the protein-ligand. All the active compounds could not interact with MAPK8 well, with the scores ranging from -7.374 to -5.835, lower than the docking score of the MAPK8 ligand (537, -9.397). 11-O-galloylbergenin, 11-O-syringylbergenin had high levels of interactions with MAPK8, with docking scores of -7.106, -7.374. As shown in Fig. 7i, 11-O-galloylbergenin could develop H bonds with GLU-109 (3.0 Å and 3.5 Å), ASP-169 (2.7 Å and 2.8 Å), MET-111 (3.4 Å and 3.0 Å), ASN-114 (3.5 Å), and 11-O-syringylbergenin(Fig. 7j) could develop H bonds with LYS-55(3.3 Å), GLN-37 (3.5 Å), GLU-109 (3.0 Å), MET-111 (3.3 Å), ASN-114 (3.4 Å and 3.0 Å), SER-155 (3.0 Å and 2.9 Å), ASP-169 (2.8 Å).
Altogether, the results showed that the targets proteins SRC and ESR1 had stronger docking capability with AGS than the other targets. Four compounds (11-O-galloylbergenin, 11-O-protocatechuoylbergenin, 11-O-syringylbergenin, epicatechin-3-gallate) had a higher docking score than the ligand-protein of SRC and two compounds (bergenin, epicatechin-3-gallate) docked better than the ligand-protein of ESR1 (Table 1) which indicates that AGS had multiple ingredients and multiple targets against CRC and potential to become an anti-CRC drug.