Pharmacoinformatics Protocol to Modulate the Tumor Suppressor Gene RUNX3 in Breast Canc

Abstract

The development and progression of cancer is a multi-step process involving multiple cellular signaling cascades causing its prevention and treatment challenging in the world. The role of epigenetics in the initiation and progression of breast cancer has led investigators to target DNA Methyltransferase1 (DNMT1) to rescue the normal functioning of tumor suppressor gene, RUNX3 to prevent breast carcinogenesis. It has been recently established that overexpression of DNMT1 induces local hypermethylation at the proximal promoter region of RUNX3 repressing its expression in breast cancer. In present project, we established an integrated network of DNMT1 and RUNX3 regulators to understand dynamics of the biological system that lead to tumor invasion. Qualitative modelling of BRN by Rene Thomas illustrated that onset of oncogene c-myc along with the persistent suppression of TSG RUNX3 contributes to the overexpression of DNMT1. Therefore, we adopted a combined structure and ligand based pharmacoinformatics protocol to explore the binding hypothesis of DNMT1 and constructed 2D and 3D models against it. Till date there are two FDA approved drugs available against DNMT1, 5-Azacitidine and Decitabine, which have certain limitations like, toxicity and inactivation by cytidine deaminases. Thus, in this research study we proposed to rescue the normal working of RUNX3 by targeting DNMT1 and designing less toxic hits against DNMT1. Machine learning models including, Neural network model with classification accuracy of 95% and decision tree model with 97% classification accuracy have been developed by using a training set of already known inhibitors against DNMT1 followed by cross validation with test sets. Models were efficient to make prediction on test sets with accuracy greater than 60%. Molecular docking studies revealed the diverse inhibitor binding potential of DNA methyl transferase1. Further, we identified crucial amino acid residues Thr 1528, Arg 1310, Lys 1535 Ser 1230 and Gly 1231 that play important role towards binding of highly active antagonists within DNMTI. Binding poses from docking studies were implied to generate class specific pharmacophore models that displayed accuracy greater than 70% for each model. It has been inferred that pharmacophore features including hydrogen bond donor, hydrophobic, and hydrogen bond acceptor as important features for the inhibition of DNMT1. Projection of DNMT1 structure onto the identified features complements the presence of complementary amino acids which reflect the robustness of the applied protocol. Overall, we have developed an integrated pipeline that can be used to screen new modulators against DNMT1 to minimize the local methylation at promoter region to reactivate RUNX3 in breast cells. The proposed protocol can be used in future for drug repurposing.