Modelos computacionais para previsão de afinidade entre ligantes e proteínas alvos para o desenvolvimento de fármacos

Abstract

Antibiotics are the most successful drugs of the 20th century and, probably, of the entire history of medicine. However, as the years went by, discoveries of new antimicrobial compounds became increasingly scarce and bacterial resistance is in evidence. From this, we selected the enzyme Trans-2-Enoyl (ACP) Reductase (InhA) (E.C. 1.3.1.9) as one of the focuses of this work, because it plays a crucial role in the anti-tuberculosis treatment. In the same way that new strains of resistant bacteria can bring complications for the coming years of public health, neoplasms are diseases that have been known for many years, but which still need to be resolved quickly and without serious side effects. Cancer can be defined as a set of cells with uncontrolled growth and the ability to invade new tissues. A directed look to this need for new chemotherapy drugs against neoplasms, we chose the enzyme Cyclin-dependent Kinase type 2 (CDK2) (E.C. 2.7.11.22) as another target of the work, mainly due to its controlling activity of the eukaryotic cell cycle. In line with the current needs exposed before, the general objective of the present work is to determine the structural bases for the inhibition of the enzymes InhA and CDK2 with a focus on the interactions that occur in the protein-ligand system. The work was carried out using the methods of Bioinspired Computing, a field of Natural Computing, which bases its approach on processes observed in nature. The methodological basis of the study followed the steps of performing molecular docking to find energetic terms that best described the interactions of each of the enzymes with possible non-covalent ligands. With the aid of the SAnDReS software, Machine Learning Methods were used, based on the classic energy terms present in the MVD, AD4 and Vina programs, polynomial score functions were constructed in an attempt to predict the degree of affinity between the two biological systems before cited and possible candidates for inhibitors. For InhA, the two polynomial functions, Polscore231 (Vina) ( = 0.709; p-value1 <0.03) and Polscore345 (AD4) ( = 0.717; p-value1 <0.03) obtained satisfactory statistical values, placing themselves as good options in inhibitor selection studies. For CDK2, the Polscore60 (MVD) polynomial function ( = 0.328; p-value1 <0.02) was the best option both in predicting the affinity of a set of structures with a resolution less than 1.5Å (HRIC50), and for the set of structures containing only CDK's2. From the correlation values obtained for each of the functions, is suggested that in later studies the polynomial functions are used in the selection of candidates for possible new drugs with inhibitory action on the catalytic site of these two enzymes.