Bioactive compound screening. In total, 221 compounds were retrieved from literature, including 144 flavonoid compounds, 23 phenolic compounds and 37 steroid compounds and 17 other compounds(Supplementary Table S1). After identifying the relevant structure, 115 bioactive compounds were screened by using SwissADME with GI and DL, which included 87 flavonoid compounds, 22 phenolic compounds and 6 steroid compounds. A total of 378 potential targets of the bioactive compounds were obtain by using PharmMapper database(Supplementary Table S2).
Mining the RD target genes for the treatment of BC. A total of 984 BC-related genes were identified by merging these genes retrieved from DisGeNET, GeneCards, OMIM, PharmGKB, and TTD databases and removing duplicates(Supplementary Table S3). Further, 148 RD target genes for the treatment of BC were obtained from common targets of both BC and the chemical constituents of RD (Fig. 1, Supplementary Table S4).
GO and KEGG enrichment analysis. To reveal the biological function and the potential mechanism underlying of RD against BC, GO functional and KEGG pathway analyses were performed using DAVID database. A total of 227 biological processes (BP) were acquired (Supplementary Table S5). The top 20 BP terms results were shown in Fig. 2a, terms related to BP mainly focused on the negative regulation of apoptotic process, signal transduction, protein phosphorylation, and positive regulation of cell proliferation. In addition, 142 KEGG pathways were acquired (Supplementary Table S6). The top 20 KEGG signaling pathways results were shown in a bubble diagram (Fig. 2b), which mainly included the Pathways in cancer, PI3K-Akt signaling pathway, MAPK signaling pathway, Ras signaling pathway, Rap1 signaling pathway.
Construction of the compound-target-pathway network. Based on the KEGG pathway enrichment results, the top 20 pathways, 91 enriched genes corresponding to 115 compounds were selected to construct the compound-target-pathway network which included 226 nodes and 3280 edges (Fig. 3a, Supplementary Table S7). According to the network analysis, single compound affected multiple targets that participated in multiple pathways, in turn, multiple compounds were closely related unique targets and unique pathways. The results revealed that the processes of RD for the treatment of BC involve multicomponent, multitarget, and multipathway. In addition, among the BC-related signaling pathways, the PI3K-AKT signaling pathway, which had the second most count of enriched genes, could play an important role in the treatment of BC(Fig. 3b).
Construction of the PPI Network and Screening of hub targets. To identify biomarkers of RD treatment for BC, 148 RD target genes were imported into the STRING database to obtain PPI information. Next, a PPI network was visualized by Cytoscape v3.6.1, which included 146 nodes and 1668 edges (Fig. 4a). In addition, CytoHubba was used to construct a core PPI subnetwork which was composed of 10 genes with the highest degree value. And finally, the top 10 hub targets were precisely aligned as AKT1, ESR1, EGFR, SRC, CASP3, HSP90AA1, IGF1, MDM2, RHOA, and MAPK1 (Fig. 4b and Table 1).
Table 1
The core targets of RD for BC treatment and the topological parameters
Number | Gene name | Protein name | Degree |
1 | AKT1 | RAC-alpha serine/threonine-protein kinase | 94 |
2 | ESR1 | Estrogen Receptor 1 | 88 |
3 | EGFR | Epidermal growth factor receptor | 87 |
4 | SRC | Proto-oncogene tyrosine-protein kinase Src | 82 |
5 | CASP3 | Caspase-3 | 80 |
6 | HSP90AA1 | Heat Shock Protein 90 Alpha Family Class A Member 1 | 79 |
7 | IGF1 | Insulin Like Growth Factor 1 | 59 |
8 | MDM2 | MDM2 Proto-Oncogene | 56 |
9 | RHOA | Ras Homolog Family Member A | 55 |
10 | MAPK1 | Matrix metalloproteinase-1 | 55 |
Validation of hub targets in different databases. To elucidate possible therapeutic significance and prognostic value of 10 hub targets, mRNA expression and protein expression of the molecules were .analyzed. In the cases of invasive BC from the TCGA database in UALCAN (Fig. 5a), the mRNA levels of AKT1, ESR1, HSP90AA1, CASP3, SRC, MDM2 mRNA expression were significantly upregulated in tumor tissues. In turn, EGFR, IGF1 and MAPK1 mRNA expression were significantly downregulated, while RHOA mRNA expression was not significantly in tumor tissues.
Next, we investigated the correlation between 10 hub targets mRNA levels and different tumor stages of BC in the UALCAN databases. As shown in Fig. 5b, When compared with normal, the mRNA levels of EGFR, IGF1 and MAPK1 were decreased significantly in stages one to four, while the expression of AKT1, ESR1, HSP90AA1, CASP3, SRC, MDM2 increased significantly, but HSP90AA1 and MDM2 were not significant in stage four. In addition, the expression of RHOA in different tumor stages has different degrees of expression than that in normal but not significant.
We further study the protein level of 10 hub targets in BC tissues. As shown in Fig. 6, the protein expression of EGFR, IGF1 and RHOA was not different between in normal breast tissues and BC tissues. In turn, the other 7 hub targets were expressed in different degrees between in normal breast tissues and BC tissues, indicating that there were significant differences. Compared with normal breast tissues, the expression levels of AKT1, ESR1, HSP90AA1, CASP3, SRC, and MDM2 were increased in BC tissues, while the expression levels of MAPK1 was decreased in BC tissues. These results indicated that aberrant expression of several hub targets also occurs at the protein level in patients with BC.
We next explored prognostic significance of 10 hub targets in BC patients using the Kaplan-Meier Mapper database. As shown in Fig. 7, of the 10 genes, we found that ESR1 (hazard ratio [HR] = 0.62, logrank P = 5.3e-07), HSP90AA1 (HR = 1.37, logrank P = 0.001), IGF1 (HR = 0.65, logrank P = 9.5e-06) were significantly correlated with OS, respectively. In contrast, AKT1(HR = 096, logrank P = 0.87), EGFR(HR = 1.1, logrank P = 0.66), CASP3(HR = 1.04, logrank P = 0.68), SRC(HR = 0.97, logrank P = 0.31), MDM2(HR = 1.05, logrank P = 0.15), MAPK1 (HR = 1.31, logrank P = 0.05) and RHOA(HR = 1.37, logrank P = 0.76) was no statistical difference with OS in BC patients. The results indicated that the low expressions of ESR1 and IGF1, the high expressions of HSP90AA1 were associated with significantly poor survival.
We next analyzed the frequency and types of gene changes in 10 hub targets in 818 BC patients using the cBioPortal tool. As shown in Fig. 8a, the overall alteration rate of 10 hub targets was 19% (159/818), and genetic alteration rate of individual genes was varied from 0.2 to 4%, of which the variation rate of AKT1 and MDM2 is the highest (4%) and the variation rate of IGF1 is the lowest (0.2%). In addition, among the 4 different breast cancer types, the genetic alteration rate ranged from 11.36–20.34% (Fig. 8b). These results indicated that the types of genetic variation in 10 hub targets include gene mutation, gene amplification, deep deletion and multiple alterations, of which the mutation type of AKT1, EGFR, ESR1, SRC, and MDM2 was mainly gene amplification.
Molecular docking validation. To further validate and explore the mode of interaction between targets gene and effective compounds, 10 hub targets and their corresponding 108 compounds were selected to perform molecular docking. In total of 103 compounds were ultimately screened with docking scores < − 7.0 to 10, which indicated that they have strong binding activity (Supplementary Table S8). As shown in Fig. 9a, the largest proportion (84.5%) of flavonoid compounds suggested that it is the main ingredient in anti-breast cancer of RD, while phenolic(11.7%) and other compounds(3.8%) were relatively lower. In addition, due to the plethora of molecular docking results, according to reports in the literature, 12 compounds with anti-breast cancer efficacy were selected to plot as a heatmap(Fig. 9b, Table 2). It can be seen from the Fig. 9b that 12 compounds with AKT1, EGFR and ESR1 all had good binding activities. We selected the molecular docking results of AKT1 with the main active compounds, and visual analysis was carried out by PyMol software(Fig. 9c). Finally, based on the analysis of molecular docking, sanguis draconis flavones (SDF) were selected in the follow-up vitro experiment.
Table 2
The anti-tumor active substances in RD
Different types | Com ID | Com Name | Antitumor Effect | References |
Flavonoid | M2 | 7,4′-Dihydroxyflavone | Inhibiing proliferation | 74 |
| M10 | Quercetin | Inhibiing proliferation, metastasis and angiogenesis, cell cycle arrest, inducing apoptosis | 70 |
| M11 | Isorhamnetin | Inhibiing proliferation, inducing apoptosis | 71 |
| M13 | Liquiritigenin | Inhibiting proliferation, and migratory, DNA methyltransferase | 25 |
| M14 | Pinocembrin | Inhibiting proliferation and metastasis | 75 |
| M15 | 7-Hydroxyflavanone | Inhibiting proliferation | 19 |
| M22 | Isoliquiritigenin | Inhibiting metastasis, neoangiogenesis | 76 |
| M27 | Loureirin A | Inducing apoptosis | 24 |
| M83 | Resveratrol | Inhibiting migration and metastasis, inducing apoptosis, cell cycle arrest | 77 |
| M84 | Pterostilbene | Inhibiting metastasis and proliferation | • 78 |
Phenolic | M147 | Syringaresinol | Inhibiting proliferation | • 79 |
| M151 | Secoisolariciresinol | Inhibiting proliferation | 80 |
Experimental Validation In Vitro
SDF inhibited the proliferation of MCF-7 cells. To verify SDF’s anti-tumor effect on BC, MCF-7 cells were treated with different concentrations of SDF for 24 h, and cell viability was determined using the CCK-8 assay. Figures 10a and b showed that with increasing concentrations of SDF, the MCF-7 cells viability gradually decreased, and the inhibition rate of MCF-7 cells in a concentration-dependent manner. The IC50 was 22.68 ± 1.68ug/mL. This result suggested that SDF can inhibit the proliferation of MCF-7 cells MCF-7, and thus 10 and 20 µg/mL for 24h were chosen in the follow-up experiment.
SDF induced cell apoptosis and cell cycle arrest in MCF-7 Cells. To examine if SDF induced apoptosis in MCF-7 cells, apoptosis cells was assessed by morphological changes and quantitative analysis. Figure 11a indicated the morphology changes of control and SDF treated MCF-7 cells were observed by light microscope. Subsequently, Annexin V/PI staining assay was used to confirm these observations. As shown in Fig. 11b, the percentage of apoptotic cells increased from 30.7 ± 0.25% and 43.3 ± 0.1.75% in MCF-7 cells in a dose-dependent manner. These findings suggested that SDF is able to effectively trigger apoptosis in MCF-7 cells.
Cell cycle arrest is related to both inhibition of proliferation and induction of apoptosis. As shown in Fig. 11c, the increase in S-phase cell population was accompanied by a corresponding decrease in G0/G1 and G2/M cell populations. The results showed that SDF arrested MCF-7 cells at the S phase in a dose-dependent manner.
Effect of SDF on 10 core target mRNA Expression in MCF-7 Cells. To verify the reliability of the network pharmacology prediction results, mRNA levels of 10 hub targets genes were detected by RT-PCR in MCF-7 cells. As shown in Fig. 12, the mRNA levels of AKT1, EGFR, ESR1, HSP90AA1, SRC, IGF1, MDM2, MAPK1 and RHOA were significantly down-regulated in SDF treated groups compared with controls(P < 0.05), while the mRNA levels of CASP3 and were significantly up-regulated (P < 0.05). These results showed that AKT1, EGFR, ESR1, HSP90AA1, SRC, IGF1, MDM2, CASP3, MAPK1 and RHOA can closely related to the BC process as the SDF treatment targets.
Effect of SDF on the PI3K/Akt Pathway in MCF-7 Cells. We further explored the molecular mechanism of SDF involved in inducing apoptosis in MCF-7 cells. The PI3K/AKT pathway is likely to be a key pathway associated with the effect of SDF against BC. The protein expression levels of PI3K, Akt, other critical genes enriched in PI3K/Akt pathway, and BAX were detected by western blot. As shown in Fig. 13, the protein levels of BCL-2, CDK2, PI3K and AKT were significantly decreased(P < 0.05). In contrast, the protein levels of BAX and Caspase 3 were significantly increased(P < 0.05). The results revealed that SDF can induce apoptosis in MCF-7 cells via the PI3K/AKT pathway.