Integrating network pharmacology and Mendelian randomization to explore potential targets of matrine against ovarian cancer

2.1Cells culture and treatment

OC cell lines A2780 (#CL-0013, Procell, Wuhan, China) and SKOV3 cells (#CL-0215, Procell) were cultured in RPMI-1640 medium (Gibco, CA, USA) at 37^∘C with 5% CO₂. Matrine purchased from Wuhan Zhongbiao Science & Technology CO. (Wuhan, China) was diluted into 2 mg/mL and then used to treat cells for 24 h.

2.2Cells counting kit 8 (CCK-8) analysis for cell viability

SKOV3 and A2780 cells were inoculated in a 96-well plate (100 μL/cell). After being treated with matrine for 24 h, 10 μL CCK-8 solution (Beyotime, Shanghai, China) was added into each well and incubated with cells for 2 h. Absorbance at 450 nm was determined with a microplate reader (Wuxi Hiwell Diatek, Jiangsu, China).

2.3Flow cytometry for cell apoptosis

Flow cytometry was performed to assess the cell apoptosis in SKOV3 and A2780 cells. After culture for 24 h, cells were gathered and suspended in 195 μL FITC binding buffer (Beyotime). Then cells were stained with Annexin V-FITC (5 μL) and propidium iodide (10 μL) at 25^∘C for 15 min and 10 min, respectively.

2.4Cell migration and invasion

Would healing was used to assess of cell migration. SKOV3 and A2780 cells were lated in 6-well plates (5 × 10⁵ cells/well). Then a sterilized pipette was used for scratches and cells were incubated in serum-free DMEM for 24 h.

Transwell were employed to assess cell invasion. SKOV3 and A2780 cells were seeded into upper chamber, and DMEM containing 10% FBS was added to the lower chamber. After 24 h incubation, cells were washed with PBS, fixed with methanol, and stained with 0.1% crystal violet.

2.5Data acquisition and preprocessing

The GeneCards (www.genecards.org) [14], Herbal Ingredients’ Targets Platform (HIT; http://hit2.badd-cao.net) [15], SwissTargetPrediction (www.swisstargetprediction.ch) [16], and SuperPred (http://prediction. charite.de/index.php) [17] databases were adopted to explore and predict the possible matrine-related targets. OC-related genes were downloaded from GeneCards and DisGeNET (http://www.disgenet.org) [18] online platforms. The GSE66957 containing mRNA profile information of 57 OC cases and 12 controls and GSE26712 including mRNA profile information of 185 OC cases and 10 controls were searched from the Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) database.

2.6Identification of potential matrine-related targets against OC

The limma package was implemented to obtain differentially expressed genes (DEGs) in OC with detecting conditions of P< 0.05 and |logFoldChange(FC)|> 1 in the GSE66957 cohort. Subsequently, the VennDiagram package was performed to select intersecting genes by overlapping OC-related genes, matrine-related targets, and DEGs.

2.7PPI network development and core targets selection

Search Tool for the Retrieval of Interacting Genes (STRING; https://string-db.org) [19] platform was adopted to explore the molecular interactions of intersecting genes with score_cutoff > 0.4 and size_cutoff < 10 as the filtering conditions. Subsequently, the PPI network was visualized using Cytoscape software. Furthermore, the cytoHubba plugin was applied to get the core targets in the PPI network, and the candidate core targets were finally detected by intersecting the results of Maximal Clique Centrality (MCC), Edge Percolated Component (EPC), Maximum Neighborhood Component (MNC), and Degree algorithms using VennDiagram. The core targets were input into the online website GEPIA (http://gepia.cancer-pku.cn/detail.php) to validate their expression levels with |log₂FC|⩾ 1 and P⩾ 0.01 as the cutoff values. The differential expression of these core targets was further validated in the dataset GSE26712.

2.8Mendelian randomization (MR)

TwoSampleMR package in R software (Version 4.2.2) was employed to conduct MR analysis. The exposure-related drug-targeting instrumental variables were harmonized with the outcome datasets. Four statistical methods, including inverse variance weighting (IVW), weighted mode, weighted median, and simple mode, were applied to explore potential causal relationships between core targets and OC. P< 0.05 was considered statistical significance. The extract_instruments function was applied to obtain SNPs independently affecting the core targets and analyzed the causal relationship between the obtained SNPs and OC based on genome-wide association studies (GWAS, https://gwas.mrcieu.ac.uk/datasets/). The heterogeneity test was performed by MR Egger and IVW methods.

2.9Molecular docking

The three-dimensional (3D) protein structure of the core targets was obtained from the RCSB Protein Data Bank (RCSB) using the website http://www.pdb.org/. To isolate the modified and ligand components and eliminate water molecules, the Pymol software was utilized. Additionally, the 3D structure of matrine was acquired from PubChem, hydrogen atoms were added, and charges were calculated. Subsequently, Autodock (Version 4.2.6) software was employed to perform docking simulations with the core targets. A stable connection site was determined based on a threshold of limited binding energy ⩽-5.0 kcal⋅mol ^{- 1} [20].

2.10Quantitative real-time PCR (qRT-PCR)

Total RNA was isolated using Trizol (Invitrogen, CA, USA). Reverse transcription utilized the cDNA Reverse Transcription Kit (TIANGEN, Beijing, China) to synthesize cDNA. Quantitative PCR was conducted using SYBR Green PCR Master Mix (Lifeint, Xiamen, China) on the StepOnePlus Real-Time PCR System (Bio-rad, CA, USA). Specific primers designed for CCND1, IL1B, p53, and GAPDH (internal control) were used in a 20 μL reaction volume. The cycling conditions included an initial denaturation of 95^∘C for 3 min, followed by 40 cycles of 95^∘C, 12 s and 62^∘C, 40 s, with subsequent melt curve analysis. Expression levels were calculated using the 2 ^{- ΔΔCT} method and normalized to GAPDH. Primer sequences were as follows: GAPDH: forward, 5’-3’-GAA GGT GAA GGT CGG AGT CA, reverse, 5’-3’-GAC AAG CTT CCC GTT CTC AG; CCND1: forward, 5’-3’-TCC TAC TAC CGC CTC ACA, reverse, 5’-3’-ACC TCC TCC TCC TCC TCT; IL1B: forward, 5’-3’-GCT CGC CAG TGA AAT GAT GG, reverse, 5’-3’-AAC ACG CAG GAC AGG TAC AG; P53: forward, 5’-3’-CCT GAG GTT GGC TCT GAC TG, reverse, 5’-3’-CTT CTT TGG CTG GGG AGA GG.

2.11Statistical analysis

Statistical analysis of the RT-qPCR data was performed using GraphPad Prism software 8.0. After confirming that the data met the assumptions of normality and equal variance, we employed Student’s t-test to compare the means between the two groups. Each set of experiments was conducted in triplicate, and results were expressed as mean ± standard deviation. P< 0.05 was considered statistically significant.

3.1Effects of martrine in cell viability, apoptosis, migration, and invasion in vitro

To investigate the effects of matrine on OC, A2780 and AKOV3 cells were cultured. Significantly, matrine reduced the cell viability, migration, invasion and increased the cell apoptosis of A2780 and AKOV3 cells (P< 0.01, Fig. 1A–D), indicating the therapy effectiveness in OC.

3.2Potential targets of matrine against OC

To further investigate the targets of matrine on OC, network pharmacology and mendelian randomization were conducted. Totally 635 possible matrine targets and 5,513 OC-related genes were obtained from the online websites after removing duplicates (Fig. 2A–B, Supplementary Table 1). A total of 8,001 DEGs were detected from the GSE66957 cohort, which comprised 4,990 up-regulated genes and 3011 down-regulated genes (Fig. 2C–D, Supplementary Table 2). Finally, 105 potential targets of matrine against OC were confirmed by overlapping 635 possible targets of matrine, 5,513 OC-related genes, and 8,001 DEGs (Fig. 2E, Supplementary Table 3).

Figure 2.

Identification of potential targets of matrine against OC. (A) The possible targets of matrine are from HIT, SwissTargetPrediction, SuperPred, and GeneCards databases, respectively. (B) OC-related genes were obtained from the GeneCards and DisGeNET online websites, respectively. (C) Volcanic map of DEGs in GSE66957 cohort. (D) Heat map of the expression levels of DEGs between OC cases and controls from GSE66957 cohort. Blue represents down-regulation, and red represents up-regulation. (E) Venn diagram shows 105 interacting genes among OC-related genes, targets of matrine, and DEGs. OC: ovarian cancer; DEGs: differentially expressed genes.

Figure 3.

Screening of hub targets. (A) PPI network of 105 intersecting genes. (B) Four core modules (Degree, EPC, MCC, and MNC) in the PPI network. (C) Venn diagram shows the six hub targets among four core modules. PPI: protein-protein interaction.

3.3Identification of core targets

The PPI network including 578 interactions among 105 potential targets of matrine was developed and shown in Fig. 3A. In addition, four core modules in the PPI network consisted of genes with the top 10 scores calculated by MCC, MNC, EPC, and Degree algorithms exhibited in Fig. 3B. By intersecting the four core modules, six core targets, consisting of TP53, STAT3, CCND1, VEGFA, IL1B, and CCL2, were further obtained (Fig. 3C). Moreover, the expression levels of six core targets were verified in the GEPIA database, and the results suggested that the expression of TP53, CCND1, VEGFA, IL1B, and CCL2 was remarkably higher in the OC samples than in the control group, whereas there was no significant difference in STAT3 (Fig. 4). Moreover, validation of the core targets differential expression in the GSE26712 dataset further revealed that TP53, CCND1, VEGFA, IL1B, and CCL2 levels were elevated in OC samples (Fig. 5).

Figure 4.

Box plots show the expression levels of CCL2, CCND1, IL1B, STAT3, TP53, and VEGFA between OC cases and controls based on the GEPIA database. Red indicates OC samples, while blue indicates normal samples. OC: ovarian cancer. ^{* P <} 0.05.

Figure 5.

Box plots show the expression levels of CCL2, CCND1, IL1B, STAT3, TP53, and VEGFA between OC cases and controls in validation dataset GSE26712. Red indicates OC samples, while blue indicates normal samples. OC: ovarian cancer. ^*P< 0.05, ^**P < 0.01, ^*** P< 0.001, ^***** P< 0.0001.

3.4Evaluating the potential association between core targets and OC

The SNP characteristics of CCND1, IL1B, TP53, VEGFA, and OC were meticulously presented in Supplementary Table 4, providing comprehensive insights into their genetic attributes. The causal correlations between the levels of core targets (CCND1, IL1B, TP53, and VEGFA) and the risk of OC were evaluated and shown in Fig. 6A–D, the results demonstrated that VEGFA and IL1B were protective factors for OC, while CCND1 and TP53 were risk factors for OC. The funnel plot exhibited a roughly symmetrical causal effect (Fig. 6E–H). Supplementary Table 5 demonstrated that CCND1 (Weighted median, β= 0.084, P= 0.021), IL1B (Weighted median, β=-0.146, P= 0.002; IVW, β=-0.088, P= 0.002; Weighted mode, β=-0.198, P= 0.001), TP53 (Weighted median, β= 0.129, P= 0.036), and VEGFA (Weight median, β=-0.146, P= 0.02; IVW, β=-0.146, P= 0.002; Weighted mode, β=-0.198, P= 0.001) were associated with the risk of OC. MR heterogeneity and multiple validity tests of exposure factors and OC were shown in Supplementary Table 6.

3.5Matrine-targets docking analysis

To gain deeper insights into the molecular interactions, four genes (CCND1, IL1B, TP53, and VEGFA) detected by MR analysis were subjected to molecular docking as protein receptors interacting with matrine (CID: 91466). Based on the selection criteria of binding energy less than -5.0 kcal/mol and the ability to form hydrogen bonds between the receptor and ligand [20], the final docking results were obtained (Table 1). The results demonstrated that the binding energy scores of matrine target CCND1, IL1B, and TP53 were -5.918, -5.953, and -5.968 kcal⋅mol ^{- 1}, respectively (Fig. 7A–C). These findings indicated that the three identified genes do indeed bind stably to matrine.

Figure 6.

Mendelian randomization study results. (A-D) Scatter plot showing the causal effect of CCND1, IL1B, TP53, and VEGFA on the risk of OC. (E-H) Funnel plots show overall heterogeneity of MR estimates of the effect of CCND1, IL1B, TP53, and VEGFA on OC. OC: ovarian cancer.

Figure 7.

Molecular docking site. (A) CCND1-matrine. (B) IL1B-matrine. (C) TP53-matrine. (D) RT-qPCR detected the mRNA expression of CCND1, IL1B, and P53. OC cell line A2780 and SKOV3 cells were treated with matrine. ^{* P <} 0.05, ^**P < 0.01.

3.6Validation of molecular docking in vitro

We then validated the effect of matrine on levels of target genes CCND1, IL1B, and TP53 in OC cell lines A2780 and SKOV3.The expression levels of CCND1 and IL1B were decreased while P53 expression was increased in the marine-treated cells compared to control cells (P< 0.05, Fig. 7D). These results further indicated that matrine may treat OC by regulating the CCND1, IL1B, and TP53.