Özet:

This study is devoted to the development of QSAR and pharmacophore models for screening of antiinflammatory activity among substituted (pyrrolo[1,2-a][1,2,4]triazino[2,3-c]quinazoline-5a-yl)carboxylic acids, obtained in earlier research of our scientist group. Genetic algorithm–multiple linear regression allowed to calculate two three-parameter QSAR equations for each model of experimental inflammation. It was found that the antiinflammatory activity in the "carrageenan" model depends on the descriptors GATS5m and B10[C-O] and the presence in the structure nROH functional group. Whereas, in the "formalin" model, pharmacological effect is determined by the descriptors E2p, R2e+, F09[N-F], that reveal positive contribution to the anti-inflammatory activity. The obtained equations are characterized by significant predictive power and are determined by both internal and external validation. The parameters of prognostic ability, namely the values of the cross-validation Q2LOO coefficient, are 0.8509 and 0.7653 respectively. Using pharmacophore modeling, it was found that substituted (pyrrolo[1,2-a][1,2,4]triazino[2,3- c]quinazoline-5a-yl)carboxylic acids bind with potential biotargets through the carboxyl moiety at the 5a position (potential participant in the formation of hydrogen bonds), regardless of the model of experimental inflammation. Additional factors influencing activity are electron withdrawing substituents, namely aromatic (fluorophenyl) cycle at the certain distance and oxygen at the second position in the molecule. The aforesaid shows the prospects for further structural modification of the molecule by replacing the triazole ring with azole one or by removing it while retaining the noted pharmacophore fragments. It is important that the developed models complement each other and can later be used for virtual screening of anti-inflammatory activity among quinazolines and its condensed analogues.

Anahtar Kelimeler: