Isoprenylcysteine carboxyl methyltransferase inhibitors: QSAR, docking and molecular dynamics studies

Abstract

A quantitative structure-activity relationship (QSAR) investigation was conducted to build models that correlate the structures of 273 compounds to their actions against Isoprenylcysteine Carboxyl Methyltransferase (ICMT) protein. Genetic Algorithm (GA) was used to select the final set of descriptors and create the correlation models that associate the structural characteristics of the chemicals to their biological activity. Three QSAR models (four descriptor based) were identified having coefficient of determination (R2) for the training set as 0.7183, 0.7358 and 0.7475 and for external validation set 0.813, 0.746 and 0.728. Artificial Neural Network(ANN) was used to establish the strength of the QSAR models. ANN model assessed the descriptor's contribution to the structure-activity association. The descriptors that formed the best models were used to form the ANN model. The correlation coefficient(R) of the ANN model was = 0.919. The MSE was 0.24 for the training set, 0.30 for the validation set and 0.15 for the test set. In accordance with the statistical findings, the QSAR models were significant and exhibited excellent stability with respect to data fluctuation in the validation technique. The docking study revealed the best molecule, which formed a hydrophobic contact and hydrogen bond with ICMT amino acid residues and had a docking affinity of -9.5 kcal/mol. Molecular dynamics studies were also performed to understand the interactions between the protein and the ligands. Ligand 264 complex trajectory showed stable behavior with lesser fluctuations and supported the molecular docking result. The results of the QSAR, molecular docking and dynamics study can be further analysed to develop specific inhibitors of ICMT.