Anti-cancer targets and molecular mechanisms of formononetin in treating osteosarcoma based on network pharmacology

Study design

We used network pharmacology to study the anti-tumor activity of formononetin and determine its possible molecular mechanism in osteosarcoma treatment to provide a basis for further pharmacological action research and clinical application.

Screening of formononetin targets

Target proteins that interact with bioactive molecules of formononetin were obtained from TCMSP, DrugBanK (https://go.drugbank.com/), Pharm Mapper (https://pubchem.ncbi.nlm.nih.gov/), and Swiss databases (http://www.swisstargetprediction.ch/). We used the full name of the target protein to obtain the corresponding gene abbreviation via UniProt ID in the UniProt database. Intersection targets screened from the three databases were considered potential target protein targets.

Functional analysis of target proteins

Gene ontology (GO) is a common method for annotation of genes and their expression products. GO functional analysis by DAVID database and pathway enrichment analysis based on clusterProfiler were performed. P < 0.05 was considered the critical value of significant gene enrichment. R software “clusterProfiler” (version 3.8) was used to make bubble charts for visualization. Kyoto encyclopedia of genes and genomes (KEGG) was searched to analyze the signal pathway of drug targets. The STRING database was used to build protein-protein interaction (PPI) network (http://www.string-db.org/) and Cytoscape software (version 3.7.2) was used to draw the drug targets-proteins network diagram. Network Analyzer was utilized to analyze three parameters of the network including degree, betweenness, and closeness.

Screening of OS target gene

CTD (https://ctdbase.org/), CooLGeN, and DisGeNET (https://www.disgenet.org/) databases were used to establish a database of OS-related targets. The intersection targets of the three databases were considered promising OS target genes.

Screening and functional analysis of common targets of OS and formononetin

Commonly predicted targets were searched from the combination of formononetin’s targets from TCMSP, Pharm Mapper, and Swiss databases and the intersection of OS’s targets from CTD, CooLGeN, and DisGeNET databases. GO functional analysis, KEGG pathway analysis, and PPI network analysis were used to analyze common targets.

Construction of target-pathway pharmacology network of formononetin and OS

The OS protein interaction network was constructed by connecting OS-related proteins and other interacting proteins. Compound target OS target protein interaction network was constructed using the intersection compound of PPI target and OS PPI target. Cytoscape software (version 3.7.2) was used to draw a pharmacology network based on the relationship among ingredients, target proteins, and pathways.

Validation of target protein using gene expression omnibus dataset

The OS mRNA chip dataset numbered GSE126209 was searched in the Gene Expression Omnibus (GEO) database of the NCBI, including six samples of OS patients and five samples of the healthy control group. The original GSE126209 dataset was downloaded and read with “Affy” R package (version 3.8). Subsequently, standardization preprocessing was performed with the RMA method, including background correction, standardization, and normalization. Probes were annotated with the platform annotation file to remove probes not matching the gene. For different probes mapped to the same gene, their mean value was taken as the final expression value of this gene. Samples were divided into two groups: the OS group and the healthy control group. The p value for expression difference and expression fold change value were calculated using limma package. P value < 0.05 and | log₂FC | > 2 (four-fold relationship) were selected as the threshold for screening significantly differentially expressed mRNA. The volcano map and heat map were drawn by combining the differential gene expression matrix file. The protein target was the intersection of formononetin prediction target, multi-data osteosarcoma prediction target, and GEO differential gene.

Molecular docking verification of core compounds and core target genes

AutoDock Vina was used to docking core targets downloaded from the PBD database and the chemical structure of formononetin downloaded from the PubChem database.

Materials

Formononetin (purity ≥ 99.0%) and MTT and DAPI kits were purchased from Sigma (USA). Main antibodies Anti-p21 and anti-p53 were bought from Santa Cruz Biotechnology (USA). Anti-mouse and anti-rabbit-horseradish peroxide (HRP) IgG were purchased from Cell Signaling Technology (USA). Annexin V-FITC/PI Apoptosis Detection kit was acquired from Bestbio Company (China).

Cells culture

MG63 cells were cultured with Dulbecco’s Modified Eagle Medium (DMEM) including 10% FBS, 100 mg/L streptomycin, and 100 mg/L penicillin with 5% CO₂ at 37°C. Cells were digested with 0.25% pancreatic enzymes and passaged in a ratio of 1:3 every three days. The following experiments were set after the degrees of cell fusion attained 80%.

Observation of morphological variations and DAPI staining

MG63 cells were incubated at a concentration of 1 × 10⁵ cells per well in six-well culture plates. Experimental groups were treated with three different concentrations (2, 5, and 8 μg/mL) of formononetin for 48 h, whereas the control group (0 μg/mL) was treated with DMEM containing 10% FBS. After treatment, morphological appearances were observed with an inverted microscope.

DAPI staining was performed to establish cell apoptosis. Treated cells were fixed with 4% paraformaldehyde solution (37°C, 15 min) and their membrane penetrability increased using 0.1% TritionX-100 solution. Subsequently, treated cells were incubated with 1 g/ml (1:1000 dilution) DAPI for 10 min and washed with Phosphate Buffer Solution (PBS). A fluorescence microscope was used to observe the nuclear morphology of PBS-washed cells.

MTT assay

MG63 cells of logarithmic growth were used to make cell suspension and seeded at the density of 5 × 10³ cells/well in 96-well plates. Each well was added with 200 μL cell suspension. After incubation overnight, MG63 cells were cultured with various concentrations (0, 2, 5, 8 μg/mL) of formononetin for 48 h. Cells were then incubated with MTT at 37°C and 5% CO₂ for 4 h. Subsequently, 150 μL DMSO was added to each well to end the reaction. Finally, plates were vortexed for 10 min, and their absorbance at 490 nm was labeled with a microplate reader.

Cell scratch test

The effect of formononetin on adhesive and invasive abilities of MG63 cells was detected with a scratch test. First, six-well plates were marked with equal straight lines on the back with at least five lines crossing over each orifice. The cell suspension was added into plates at a concentration of about 1×10⁵ cells/well and cultured with 5% CO₂ at 37°C overnight. Then, transverse lines perpendicular to the straight lines were drawn on the back of these plates. Next, after washing with PBS, MG63 cells were incubated with diverse concentrations (0, 2, 5, 8 μg/mL) of formononetin for 24 h. Finally, cell migration was observed and photographed with an optical microscope.

Flow cytometric analysis

To investigate cell apoptosis, MG63 cells in the control group and experiment group were collected and rinsed severally with PBS. Cells were stained with PI and Annexin V-FITC solution for 20 min at room temperature and analyzed with flow cytometry.

Western blotting assay

Total proteins were divided into equal amounts and separated on SDS-PAGE gels. Then, the protein in gels was transferred to PVDF membranes. The membranes were blocked with 5% BCA for 1 h. Then electroblotted PVDF membranes were incubated overnight at 4°C with 1:500 primary antibodies including bcl-2, caspase-3, p21, and p53, followed by 1:2000 diluted secondary antibodies at room temperature for 1.5 h. Blots were then developed in the dark room and scanned with Bio-Rad (USA) GS800. Finally, quantitative analysis was performed with ImageJ 1.41 software.

Statistical analysis

All tests were repeated in three independent experiments. Data were analyzed with SPSS 16.0 and expressed as mean ± standard deviation (SD). One-way analysis of variance (ANOVA) followed by a post-hoc test was used to analyze the results. Values with p < 0.05 were considered statistically significant.

Data availability statement

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

Formononetin target protein database establishment and target protein analysis

In total, 121 target proteins were collected from Traditional Chinese Medicine Systems Pharmacology (TCMSP), Pharm Mapper, and Swiss database. The intersecting target protein was ESR1, which revealed a relatively more significant role of ESR1 in the play (Figure 2A, 2B). Using the David website and KEGG, these targets were screened with GO enrichment analysis (Supplementary Figure 1A, 1B) and KEGG enrichment analysis (Supplementary Figure 1C, 1D). Serotonin-containing synaptic pathway (GO ID is hsa04726), one of the signal pathways of formononetin, contains 11 formononetin target genes, including MAOA, HTR2A, HTR2C, ALOX12, ALOX15, PTGS1, MAOB, RAF1, PTGS2, PRKACA and SLC6A4 (Supplementary Figure 2).

Figure 2. (A) The 2D and 3D chemical structure of formononetin. (B) The predicted targets of formononetin.

Construction of PPI network diagram and formononetin target-pathway network

STRING database was used to draw a PPI network diagram of the targets (Figure 3A), which presented both direct and indirect regulatory relationships between targets. PPI network diagram comprised 121 nodes and 406 edges (PPI enrichment p-value < 1.0e⁻¹⁶). The average node degree is 6.71. The average local clustering coefficient is 0.493. The expected number of edges is 193. The nodes represent proteins, and the connections represent interactions between proteins. The more connections there are, the greater the degree of connection. The abscissa depicted 20 gene relationship pairs in the network diagram, including EGFR, ESR1, JUN, PTGS2, SIRT1, NOS3, AR, PPARG, ESR2, MAPK14, ACHE, F2, HSP90AB1, IL4, CYP19A1, IL2, PRKACA, ADRA1A, ADRB2, and GSK3B, in descending order of the number of related pairs of genes in the network diagram (Figure 3B). Considering ESR1 is the key target of formononetin, we also did the PPI network diagram of ESR1, and Figure 3C is the ESR1 protein interaction diagram. The PPI network diagram comprised 11 nodes and 39 edges (PPI enrichment p-value is 4e^−0.5). Cytoscape software was used to draw a network diagram demonstrating the pathway regulation mechanism of formononetin on related target proteins (Figure 3D). The yellow circles indicate formononetin targets, and the blue arrows indicate the KEGG pathways.

Figure 3. (A) Construction of formononetin network. (B) The abscissa represents the number of related pairs of genes in the network diagram. (C) The ESR1 protein interaction diagram. (D) The formononetin targets gene network.

Molecular docking verification of ESR1 protein and formononetin

In order to study the key target ESR1 deeper, the molecular docking verification of SATB2 and formononetin was performed. In a schematic diagram of docking of ESR1 protein and formononetin (Figure 4A–4C), the affinity between formononetin and the ESR1 target protein is −7.7 kcal/mol, indicating a strong interaction between ligand and receptor. The formononetin ligand forms a hydrogen bond with three residues in ESR1 protein, namely, ARG-394, LEU-346, and LEU-387.

Figure 4. (A) The electrostatic potential energy diagram of ESR1 and formononetin ligand. (B) The 2D structural diagram showing the interaction between ESR1 and formononetin ligand. (C) The structural diagram illustrating the interaction between ESR1 and formononetin ligand.

Screening and validation of OS target genes

We explored a total of 823 target proteins from CTD, CooLGeN, and DisGeNET databases (Figure 5A). In total, 21 common prediction targets were searched from combined formononetin targets collected from TCMSP, Pharm Mapper, and Swiss database and intersection targets of OS from CTD, CooLGeN, and DisGeNET database (Figure 5B). GEO dataset was used to validate the target protein. In total, 635 differential genes were searched from Chip GSE126209. Of these, 367 were up-regulated whereas 268 were down-regulated (Figure 5C, 5D). SATB2 protein target was identified from predicted formononetin intersection targets, OS targets, and GEO differential genes (Figure 5E). A boxplot extracted the expression levels in the GEO database (Figure 5F). Comprehensively, we considered that the STAB2 might serve as the key target in treating osteosarcoma.

Figure 5. (A) A Venn diagram of the predicted targets of OS. (B) A Venn diagram of OS and formononetin targets. (C, D) A calorimetric map and differential gene volcanic map for the OS group and normal group based on GSE126209 chip in the GEO database. (E) A Venn diagram showing the prediction targets for the OS in the GEO database. (F) A boxplot extracting the expression levels in the GEO database.

GO and KEGG analysis of formononetin anti-OS targets

We performed GO enrichment analysis (Figure 6A, 6B) and KEGG enrichment analysis (Figure 6C, 6D) on 21 prediction targets common to formononetin anti-OS through the David website and KEGG. Only the top 20 GO IDs are shown. These targets were involved in Endocrine resistance pathways, MicroRNAs in the cancer pathway, C-type lectin receptor signaling pathway, Estrogen signaling pathway, and Cellular senescence pathway.

Figure 6. (A, B) The bar chart and bubble chart showing top 20 GO ID. (C, D) The bar chart and bubble chart showing the top 20 KEGG pathways of potential target genes of Form in OS.

Construction of PPI network diagram and formononetin target-pathway network

STRING database was used to draw a PPI network diagram of the targets (Figure 7A) demonstrating direct and indirect regulatory relationships between targets. The Number of nodes is 21. The Number of edges is 59. The average node degree is 5.62. Average local clustering coefficient is 0.648. The expected number of edges is 19. PPI enrichment p-value is 3.86e⁻¹³. The nodes represent proteins, and the connections represent interactions between proteins. The more connections there are, the greater the degree of connection. In core gene analysis results (Figure 7B), the abscissa depicted 15 gene relationship pairs in the network diagram, namely EGFR, JUN, ESR1, PTGS2, MAPK14, SIRT1, IL2, CHEK1, ESR2, ABCB1, CCNA2, PPARG, CA9, EZR, MIF in descending order of the number of related pairs of genes in the network diagram. Cytoscape software was used to draw a pharmacology network diagram demonstrating the pathway regulation mechanism of formononetin-OS-related protein targets based on KEGG pathway analysis (Figure 7C). The yellow circles indicate formononetin-OS prediction targets, and the blue arrows indicate the KEGG pathways.

Figure 7. Construction of PPI network diagram (A) The protein interaction diagram. (B) The abscissa represents the number of related pairs of genes in the network diagram. (C) The formononetin-OS target gene network.

Molecular docking verification of SATB2 and formononetin

In order to study the key target SATB2 further, the molecular docking verification of SATB2 and formononetin was performed. In a schematic diagram of docking of SATB2 protein and formononetin (Figure 8), the affinity between formononetin and SATB2 target protein was −7 kcal/mol, indicating a strong interaction between SATB2 and formononetin. The formononetin ligand formed five hydrogen bonds with four residues in SATB2 protein, including ARG-87, ASN-81, ARG-77, and ILE-9.

Figure 8. (A) The electrostatic potential energy diagram of SATB2 and formononetin ligand. (B) The structural diagram showing the interaction between SATB2 and formononetin ligand. (C) The 2D structural diagram illustrating the interaction between ESR1 and formononetin ligand.

Effects of formononetin on morphology and DAPI staining of MG63 cells

When observed under an optical microscope, MG63 cells in the control group were intact, spindle-shaped, adhered to the wall, and had unclear cell boundaries (Figure 9A (a)). Treatment with 2 μg/mL, 5 μg/mL or 8 μg/mL formononetin for 48 h induced cell apoptosis. The cells appeared shrunk and spherical (Figure 9A (b–d)). Compared with the control group, apoptosis in experimental groups increased in a concentration- dependent manner. Cells were treated with DAPI for 48 h to assess cell apoptosis. Likewise, an intact nucleus and no evidence of apoptosis was observed in no-formononetin-treated control group. However, in the experimental groups treated with 2 μg/mL, 5 μg/mL, and 8 μg/mL formononetin for 48 h, apoptosis was observed (Figure 9B (a′–d′)).

Figure 9. The ((A) a–d) showed the results observed under the inverted microscope after being treated by different concentrations of formononetin for 48 h. Meanwhile, ((B) a′–d′) showed the results treated with DAPI for 48 h to assess cell apoptosis. (a, a′) The MG63 cells were treated with 0μg/mL Form for 48 h (the control group). (b, b′) The MG63 cells were treated with 5 μg/mL Form for 48 h. (c, c′) The MG63 cells were treated with 8 μg/mL Form for 48 h. (d, d′) The MG63 cells were treated with 10 μg/mL Form for 48 h.

Effects of formononetin on proliferation of MG63 cells

To verify whether formononetin could inhibit cell proliferation, MG63 cells were cultured with diverse concentrations of formononetin (0, 2, 5, and 8 μg/mL) for 48 h. Cell proliferation was assessed using an MTT assay. The results showed that cell survival rate declined with increasing formononetin concentration (Figure 10A), demonstrating that formononetin has an inhibitory effect on MG63 proliferation.

Figure 10. (A) The inhibitory effect of different concentrations of formononetin (0, 2, 5, 8 μg/mL) on the proliferation of MG63 cells. Data are presented as the means ± SD of three independent experiments. ^*p < 0.05, ^**p < 0.01 vs. control group. (B) Images of the wounds of cell layer at 0 h (a–d) and after 48 h (a′–d′), showing the width of scratches with the effect of different concentrations of formononetin (0, 2, 5, 8 μg/mL) on the migratory ability of MG63 cells. (C) Representative results of Annexin-V/PI staining for formononetin. (a–d) Cells were treated with 0, 2, 5, 8 μg/mL formononetin for 48 h. (D) Wound healing assay of formononetin treated MG63 cells. (E) The apoptosis rate of MG63 cells after formononetin treatment.

Effect of formononetin on migration of MG63 cells

The effect of different concentrations of formononetin (0, 2, 5, and 8 μg/mL) on the migration ability of MG63 cells was analyzed using cell scratch test. As shown in Figure 10B (a, a′–d, d′) and 10D, the cell migration was significantly accelerated in the control group compared with the experimental groups after 48 h.

Effect of formononetin induced apoptosis in MG63 cells

Flow cytometry analysis was performed to confirm apoptotic cell damage. In the control group, few apoptotic cells were detected (Figure 10C (a)). However, cells treated with 2 μg/mL formononetin exhibited remarkable apoptosis of up to 17.07% (Figure 10C (b)). In addition, treatment with 5 μg/mL and 8 μg/mL formononetin for 48 h increased apoptotic rate to 31.06% and 55.77%, respectively (Figure 10C (c, d) and 10E).

Formononetin stimulated the mitochondrial pathway and inhibited the expression of SATB2

Cellular levels of bcl-2, caspase-3, p53, p21, and SATB2 protein in different concentrations of formononetin (0, 2, 5, and 8 μg/mL) were demonstrated in MG63 cells using a western blotting assay (Figure 11). Analysis of cell cultures incubated with various concentrations of formononetin for 48 h showed a decrease in gray ratios of bcl-2 with increasing concentrations of formononetin. Further, experimental groups had significantly lower gray ratios of bcl-2 than the control group. Expression levels of caspase-3, p53, and p21 in the experiment groups were higher than in the control group and increased concomitantly with formononetin concentrations. The cellular level of SATB2 protein in different concentrations of formononetin (0, 2, 5, 8 μg/mL) after 48 h treatment was suppressed in MG63 cells as shown through western blotting assay. The gray ratio of SATB2/β-actin of the experimental groups was significantly lower than that of the control group and decreased gradually as the drug dose was increased.

Figure 11. (A) Western blotting results showing expression of cleaved caspase-3, p21, and p53. Formononetin inhibited bcl-2 expression (B) and induced the expression of cleaved caspase-3 (C) through the p21/p53 (D, E) pathway and inhibited the expression of SATB2 (F). Data are presented are the means ± SD of three independent experiments. ^*p < 0.01 vs. control group.