Key target genes related to anti-breast cancer activity of ATRA: A network pharmacology, molecular docking and experimental investigation

Abstract

All-trans retinoic acid (ATRA) has promising activity against breast cancer. However, the exact mechanisms of ATRA's anticancer effects remain complex and not fully understood. In this study, a network pharmacology and molecular docking approach was applied to identify key target genes related to ATRA's anti-breast cancer activity. Gene/disease enrichment analysis for predicted ATRA targets was performed using the Database for Annotation, Visualization and Integrated Discovery (DAVID), the Comparative Toxicogenomics Database (CTD), and the Gene Set Cancer Analysis (GSCA) database. Protein-Protein Interaction Network (PPIN) generation and analysis was conducted via Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) and cytoscape, respectively. Cancer-associated genes were evaluated using MyGeneVenn from the CTD. Differential expression analysis was conducted using the Tumor, Normal, and Metastatic (TNM) Plot tool and the Human Protein Atlas (HPA). The Glide docking program was used to predict ligand-protein binding. Treatment response predication and clinical profile assessment were performed using Receiver Operating Characteristic (ROC) Plotter and OncoDB databases, respectively. Cytotoxicity and gene expression were measured using MTT/fluorescent assays and Real-Time PCR, respectively. Molecular functions of ATRA targets (n = 209) included eicosanoid receptor activity and transcription factor activity. Some enriched pathways included inclusion body myositis and nuclear receptors pathways. Network analysis revealed 35 hub genes contributing to 3 modules, with 16 of them were associated with breast cancer. These genes were involved in apoptosis, cell cycle, androgen receptor pathway, and ESR-mediated signaling, among others. CCND1, ESR1, MMP9, MDM2, NCOA3, and RARA were significantly overexpressed in tumor samples. ATRA showed a high affinity towards CCND1/CDK4 and MMP9. CCND1, ESR1, and MDM2 were associated with poor treatment response and were downregulated after treatment of the breast cancer cell line with ATRA. CCND1 and ESR1 exhibited differential expression across breast cancer stages. Therefore, some part of ATRA's anti-breast cancer activity may be exerted through the CCND1/CDK4 complex.

Keywords: All-trans-retinoic acid; Breast neoplasia; Drug-target prediction; Network pharmacology; Protein-protein interaction network.

Graphical abstract

Fig. 1

Protein-protein interaction network and gene/diseases enrichment analysis of ATRA targets. a) Venn diagram of databases used for ATRA target prediction. The total number of targets was 254: DrugBank = 30 (∼12 % of total targets), Binding Database = 22 (∼9 % of total targets), SWISS Target Prediction = 105 (∼41 % of total targets), CTD = 87 (∼34 % of total targets), and STITCH = 10 (∼4 % of total targets). b) Protein-protein interaction network of ATRA targets. Network nodes are orange and edges are grey. PPI enrichment p < 10⁻¹⁶, number of edges: 2074, clustering coefficient: 0.496. Yellow highlighted nodes are core targets of ATRA (Network hub genes), red edges are directly connected to core targets. c) Gene ontology (Molecular functions: ■, Biological processes:●, Cellular components:▲) and Wiki pathways(+). d) Enriched diseases retrieved from Set analyzer (red box), GAD diseases (yellow box), and DisGeNET (blue box). Counts represents the number of genes involved.

Fig. 2

Venn diagram of hub genes and final hub genes, and network modules. a) Venn diagram of the top 50 nodes in the PPIN of ATRA targets based on closeness centrality, betweenness centrality, degree and radiality. b-d) Modules of ATRA targets in the PPIN. Network nodes are orange and edges are grey. b) Module 1 (nodes:27, efges:173). c) Module 2 (nodes:41, edges:285). d) Module 3 (nodes:22, edges:60). e) Venn diagram of overlapped hub genes of ATRA targets in the PPIN with breast cancer-associated genes.

Fig. 3

PPIN, disease enrichment, pathway enrichment, and functional analysis of breast cancer-associated hub genes. a) PPIN of breast cancer-associated genes (final hub genes). Network nodes are yellow and edges are red. PPI enrichment p < 10⁻¹⁶, number of edges: 92, clustering coefficient: 0.873. b) Disease enrichment analysis using DisGeNET (blue box) and GAD diseases (red box). Counts represents the number of genes involved. c)Pathway enrichment analysis based on KEGG (■), Reactome (●), and Wiki pathways (▲). d) Role of breast cancer-associated genes in Activation (A) or Inhibition (I) of breast cancer-related pathways/mechanisms/processes. In each cell, 100 indicates a specific gene's association with a particular pathways/mechanisms/process, while 0 indicates non-association.

Fig. 4

Differential expression of key hub genes between breast cancer and normal tissues. a) Comparison of gene expression between paired tumor and adjacent normal tissues. b) Comparison of gene expression between tumor and normal samples from non-cancerous individuals. c) IHC sections showing examples of protein expression levels in breast carcinoma and normal tissues, randomly obtained from two different cancer patients and one non-cancerous individual. Brown color indicates higher expression. *There is no IHC section for the RARA gene in HPA [32].

Fig. 5

Molecular docking analysis of ATRA interaction with key hub genes CCND1, ESR1, MDM2 and MMP9. 2D representations of amino acids residues surrounding bound ATRA with: a) CCND1, b) ESR1, c) CCND1 in complex with CDK4, d) MMD2, and e) MMP9. Red arrows indicate hydrogen bonds. Depiction of ATRA interaction with: f) CCND1, g) CCND1 in complex with CDK4, and h) MMP9 in α-helix/β-sheet form.

Fig. 6

ROC curve analysis and differential genes expression for CCND1, ESR1, and MDM2 in breast cancer. a) Gene expression comparison between responders and non-responders to chemotherapy. b) ROC curve for treatment response prediction, with a higher Area Under the Curve (AUC) value indicating better discrimination between responders and non-responders to chemotherapy. TPR: True Positive Rate, TNR: True Negative Rate. c) Box plots for gene expression comparison in different breast cancer stages.

Fig. 7

Cell viability and genes expression changes after ATRA treatment. a) Viability of MCF7 and HFF cells after 24 h treatment with different concentrations (0–266 μg/ml) of ATRA. b) Fluorescent live/dead cell viability of MCF7 and HFF cells after 24 h treatment with the ATRA IC50 for HFF cells (200 μg/ml). c) Changes in expression levels of ESR1, MDM2, and CCND1 genes after 24 h treatment of MCF7 and HFF cells with the ATRA IC50 for MCF7 cells (85 μg/ml). p < 0.01 (**). Not significant (ns).