DESIGN, SCREENING AND EXPLORING NOVEL METAL-THIOSEMICARBAZONE COMPLEXES TARGETING HT-29 CELL LINE USING QSPR-BASED IN SILICO MODELING

Tóm tắt

In silico models were used to calculate the stability constants (logb₁₁) of metal ions and thiosemicarbazone in complex solutions. The 0-3D molecular descriptors, physicochemical, and quantum parameters were produced using the semi-empirical quantum computation PM7 and PM7/sparkle as well as the molecular geometric structure. Multivariate linear regression (MLR) and artificial neural networks (ANN) were used to build the quantitative structure and property relationship (QSPR). The constructed QSPR_MLR model involves descriptors such as: k0, core-core repulsion, xvp10, ⁵C, and xch5. The statistical values of the model were pointed out: R²_train = 0.847, R²_adj = 0.834, Q²_LOO = 0.764, SE = 1.371, and F-stat = 66.20. The neural network QSPR_ANN I(5)-HL(8)-O(1) model was also developed with the statistical values: R²_train = 0.976, Q²_test = 0.956, and Q²_validation = 0.926. A series of new thiosemicarbazone derivatives and complexes of this ligand and metal ions were designed using these models. These complexes were screened using the Applicability Domains (AD) and Outliers technique. The ability prediction of models applied complexes on the test data set agrees with those from the experimental literature. Besides, the results of fifteen new complexes were selected because they were within the predictive application domain. Next, nine of fifteen new complexes overcame drug-likeness analysis of Lipinski and Veber rules. Finally, nine typical complexes simulated the docking on protein for anti-colorectal cancer (code: 6GUE-PDB). The results screened out six novel complexes considered potential inhibitors to the HT-29 cell line in supporting colorectal cancer treatment.