Study on the mechanism of Danshen-Guizhi drug pair in the treatment of ovarian cancer based on network pharmacology and in vitro experiment

Screening of effective chemical component and collecting targets of Danshen-Guizhi drug pair

The main effective components of the Danshen-Guizhi drug pair were collected from the TCMSP (https://tcmspw.com/tcmsp.php). All components of the drug pair were screened by TCMSP based on absorption, distribution, metabolism, and excretion (ADME), including oral bioavailability (OB), drug similarity (DL), and lipids-water partition coefficient (AlogP) (Lee et al., 2019). OB is the percentage of oral drugs absorbed into systemic circulation, DL represents the similarity between the active ingredients of Traditional Chinese medicine and drugs, AlogP is one of the binding forces between drugs and receptors, and the occurrence of drug pharmacokinetic process is related to the hydrophobicity of drugs (Liu et al., 2013). All components of the Danshen-Guizhi drug pair were screened under the following conditions: oral bioavailability (OB) ≥30%, drug-like property (DL) ≥0.18, and lipid-water partition coefficient (AlogP) ≤5 (Liu et al., 2016). The main active compounds of the Danshen-Guizhi drug pair and their respective protein targets were obtained, the collected targets were transformed into gene names in the Uniprot protein database (https://www.uniprot.org/uploadlists/), and invalid genes were deleted.

Acquisition of disease-related targets

Enter “Ovarian Cancer” in Genecards database (https://www.genecards.org/) and OMIM database (https://omim.org/) to search for potential therapeutic targets of ovarian cancer.

Construction of PPI network and screening of core targets

The ovarian cancer-related gene targets obtained in Genecards were crossed with the drug targets collected in TCMSP to obtain the potential targets for drug treatment of ovarian cancer, and the VENNY diagram was plotted. Then the intersection targets were imported into the STRING (https://www.string-db.org/) database to restrict the species to human (“Homo sapiens”), and the protein-protein interaction network function enrichment analysis was carried out to obtain the PPI network. The results were topologically analyzed by Cytoscape3.8.0 software plug-in for potential therapeutic targets, and PPI network diagram of potential therapeutic targets of the Danshen-Guizhi drug pair for ovarian cancer was constructed.

In order to explore the mechanism of the Danshen-Guizhi drug pair in the treatment of ovarian cancer, the obtained PPI of disease and drug were combined with the intersection network in Cytoscape3.8.0 software, and then the topology network was analyzed by Cytoscape3.8.0 software plug-in to screen the core targets.

Enrichment analysis of Danshen-Guizhi drug pair in the treatment of ovarian cancer

The Matescap database (http://metascape.org/gp/index.html#/main/step1) is a powerful online tool for gene enrichment analysis, which can be used for gene annotation and analysis resources. The potential targets of the Danshen-Guizhi drug pair in the treatment of ovarian cancer were input into the Matescape database, and the species was limited to “H. sapiens”. P < 0.01 was set for enrichment analysis of GO molecular function, cell composition and KEGG pathway.

Danshen-Guizhi drug pair “component-target-pathway” network construction

The network diagrams of active components, targets and pathways of drugs were drawn by Cytoscape 3.8.0 software. The network topology parameters of active components and targets of drugs were analyzed by CytoScape 3.8.0 plug-in CytoNCA. The core targets and main active components of drugs were judged according to the network topology parameters of Degree, Betweenness and Closeness.

Molecular docking verification

The main components corresponding to the target protein were searched from the TCMSP database (https://tcmspw.com/tcmsp.php), downloaded and saved as Mol2 format. The “component-target-pathway” network of the Danshen-Guizhi drug pair was analyzed, and the core targets with degree greater than 20 were selected for docking. The corresponding PDB ID of the core targets was found in the PBD (http://www1.rcsb.org/) database and saved as pdb format. Then the target was dehydrated and hydrogenated to optimize ligand in Pymol (Ren et al., 2020). Finally, the molecular docking was performed by AutoDock software to calculate the binding ability of the main active components of the Danshen-Guizhi drug pair to the core targets.

MTS assay

SKOV3 cells and A2780 cells were seeded in 96-well plates at 100 μL/well (about 1 × 10⁴), cultured at 37 °C in 5% CO₂ incubator for 24 h. Luteolin, salviolone, tanshinone IIA and β-sitosterol were added at concentrations 0, 10 or 20 μm. Luteolin, β-sitosterol and tanshinone IIA were obtained from Sigma (USA) and salviolone was obtained from Tianjin Wanxiang Hengyuan Technology Co., LTD. After incubation for 2 h, the cell vitality was detected by MTS kit (G3582; Promega, Madison, WI, USA), and the optical density was detected at 450 nm.

Detection of PTGS2 mRNA expression by qRT-PCR

Molecular docking results showed that the docking results of salvianone with seven main targets were the best. In order to further explore the role of salviolone in ovarian cancer, we detected the effect of salviolone on PTGS2 mRNA expression in ovarian cells. SKOV3 cells and A2780 cells were seeded in a 6-well plate and added with 0 and 20 μm of salviolone for 24 h. Cells were collected, total RNA was extracted, and PTGS2 mRNA was quantitatively detected. Data were represented as the Mean ± SD of three independent experiments.

Detection of PGE2 expression by ELISA

SKOV3 cells were seeded in 6-well plates at 1 × 10⁵cells per well. Cells were incubated in serum-free medium. The cells were incubated with 0, 20 μm salviolone for 2 h. The cell supernatant was collected and PGE2 was measured by ELISA kit (ab133021; Abcam, Cambridge, UK). The absorbance (OD) of the sample was measured at 415 nm. The average value was calculated by three repeated experiments.

Statistical analysis

SPSS Statistics 20.0 and GraphPad Prism v8.0.2 were used for statistical analysis. The data are expressed as Mean ± SD. Student’s t-test was used to analyze the difference in effect of luteolin, salviolone, β-sitosterol as well as tanshinone IIA on ovarian cancer cells, compared with the control group, and P < 0.05 was considered to be statistically significant.

Obtaining active components and potential targets of Danshen-Guizhi drug pair

Through searching the TCMSP database, 202 effective active ingredients of Danshen and 220 effective active ingredients of Guizhi were obtained. According to ADME, 54 effective active ingredients of Danshen and six effective active ingredients of Guizhi were obtained. A total of 916 targets of effective components of Danshen and 74 targets of Guizhi were obtained. After deleting duplicate targets, 60 effective active ingredients and 136 targets were obtained, as shown in Table 1. The collected targets were transformed into genes in the Uniprot database.

Acquisition of ovarian cancer-related targets

A total of 8,366 potential targets related to ovarian cancer were obtained by Genecards human gene database. According to the target targets with Score greater than the median, the potential targets for the treatment of ovarian cancer were determined. Finally, 1,119 therapeutic targets for ovarian cancer were obtained (Table S1).

Construction and analysis of PPI network

A total of 119 ovarian cancer-related gene targets obtained in Genecards were crossed with 136 drug targets, and 66 potential targets for drug treatment of ovarian cancer were obtained. VENNY map was drawn, as shown in Fig. 1. The PPI network was constructed by STRING software for 66 targets. The network nodes represent proteins, and the edges represent the combination of proteins and proteins. The number of nodes is 66,1109. Cytoscape 3.8.0 software was used to construct the PPI network diagram of potential targets for the treatment of ovarian cancer by the Danshen-Guizhi drug pair, and the depth of node color as well as the size of node were related to the degree value. Gene targets can be screened from the color of node size: PTGS1, TP53, PTGS2, AKT1, MAPK1, JUN, AR, EGFR, ESR1, etc. These can be considered to be the core targets in the target interaction network of the Danshen-Guizhi drug pair for the treatment of ovarian cancer (Fig. 2).

In order to reveal the pharmacological mechanism of the Danshen-Guizhi drug pair in the treatment of ovarian cancer, the PPI of disease (Fig. 3A) and drug (Fig. 3B) were combined to obtain the intersection network (Fig. 3C). The degree value of all nodes was calculated according to the median of two times. All nodes with the degree value greater than two times the median were selected to extract the network, and the Hithubs1 network (Fig. 3D) was obtained. The median and each attribute value of nodes in the Hithubs network were calculated. All nodes with the degree value greater than the median were selected to extract the network again, and the Hithubs2 network (Fig. 3E) was obtained. Finally, 70 key targets (Fig. 3F) were obtained.

Enrichment analysis of potential targets for the treatment of ovarian cancer by Danshen-Guizhi drug pair

The potential targets of Danshen-Guizhi drug pair in the treatment of ovarian cancer were input into Matescape database for analysis, and the limited type was “H. sapiens”. The personalized analysis was carried out, and the P < 0.01 was set. The enrichment analysis of potential targets of the Danshen-Guizhi drug pair in the treatment of ovarian cancer was obtained, as shown in Fig. 4. The biological processes of GO enrichment include: apoptotic signaling pathway, response to oxidative stress, response to reactive oxygen species, regulation of mitotic cell cycle, cellular response to lipid, response to organic cyclic compound and response to steroid hormone.

The results of KEGG enrichment analysis pathway were shown in Table 2. Pathways in cancer, hepatitis B, endocrine resistance, proteoglycans in cancer, AGE-RAGE signaling pathway in diabetes complications, PI3K-Akt signaling pathway, apoptosis, platinum drug resistance, TNF signaling pathway, IL-17 signaling pathway, progesterone-mediated oocyte maturation, and the prolactin signaling pathway were included in Fig. 5.

Construction of “component-target-pathway” network of Danshen-Guizhi drug pair

The results of effective active components, potential targets and enrichment pathways of the Danshen-Guizhi drug pair were analyzed by Cytoscape3.8.0 built-in plug-in CytoNCA, the core components and core targets of Danshen-Guizhi drug pair in the treatment of ovarian cancer were obtained, as shown in Fig. 6.

Cytoscape analysis results showed that the active ingredients DS7 luteolin, DS52 tanshinone IIA, DS48 salviolone, GZ2 beta-sitosterol, DS32 dihydrotanshinone lactone, DS28 cryptotanshinone and so on, among which the highest degree of luteolin was 58, the betweenness centrality was 0.204, and the closeness centrality was 0.476. It can be judged that luteolin was the core component of the Danshen-Guizhi drug pair in the treatment of ovarian cancer, followed by tanshinone (connectivity was 46, betweenness centrality was 0.048, and closeness centrality was 0.422), salviolone (connectivity was 39, betweenness centrality was 0.099, and closeness centrality was 0.414), and β-sitosterol (connectivity is 38, betweenness centrality was 0.083, closeness centrality was 0.431), which can be considered to be effective core components in the treatment of ovarian cancer.

The core target nodes were sorted by degree value (connectivity): PTGS2 (connectivity 61), CA2 (connectivity 46), NCOA1 (connectivity 33), HSP90AA1 (connectivity 31), ADRB2 (connectivity 29), SCN5A (connectivity 29), PTGS1 (connectivity 29), CHRNA7 (connectivity 28), OPRM1 (connectivity 28), CHRM1 (connectivity 27), RXRA (connectivity 26), ESR1 (connectivity 25), F2 (connectivity 25), ADRA1A (connectivity 25), ACHE (connectivity 24), CHRM3 (connectivity 23), RELA (connectivity 21), DRD1 (connectivity 21), AR (connectivity 20), and NCOA2 (connectivity 20). According to the degree value, the above targets can be judged as effective core targets for the treatment of ovarian cancer, and PTGS2 is the main core target.

Molecular docking analysis

In AutoDock docking, the score of the binding ability between active ingredients and core targets of the Danshen-Guizhi drug pair was greater than 4.25, indicating that there was a binding ability between active ingredients and targets, greater than 5.25, indicating that the molecule has a good relationship with the target, and greater than 7.0, indicating that the molecule has an excellent binding ability to the target (Hsin, Ghosh & Kitano, 2013). The results of molecular docking analysis showed that the docking scores of salviolone and tanshinone IIA with the core targets were higher, and the docking scores of PTGS2, AR and PTGS1 with the active components were higher. In the total docking scores, 89% of the total docking scores were higher than 4.25, and 68% of the total docking scores were higher than 5.25, indicating that the predicted targets had good binding ability with the active components. The results of molecular docking were shown in Fig. 7. The higher the score was, the deeper the color was. The highest value was 8.16.

Inhibition of core components on ovarian cancer cells

The effects of luteolin, salviolone, β-sitosterol as well as tanshinone IIA on SKOV3 and A2780 cells were determined by MTS assay. Compared with the control group, the activity of SKOV3 cells (Fig. 8A) and A2780 cells (Fig. 8B) were significantly inhibited by 10 μm, 20 μm luteolin, salviolone, β-sitosterol as well as tanshinone IIA, as shown in Fig. 8.

Expression of PTGS2 mRNA and PGE2

The expression of PTGS2 mRNA was evaluated by qRT-PCR. The expression of PTGS2 in SKOV3 and A2780 cells were significantly inhibited by salviolone (Fig. 9A). The expression of PGE2 was determined by ELISA. The expression of PGE2 in SKOV3 cells was significantly down-regulated by salviolone (Fig. 9B).