Frontier Molecular Orbital Energies as Descriptors for Prediction of Antiglycating Activity of N-Hydroxybenzoyl-Substituted Thymine and Uracil

• Published in: Pharmaceutical chemistry journal 2021-10, Vol.55 (7), p.648-654
• Main Authors: Litvinov, R. A., Vasil’ev, P. M., Brel’, A. K., Lisina, S. V.
• Format: Article
• Language: English
• Subjects: Analysis; Medicine; Neural networks; Organic Chemistry; Pharmacology/Toxicology; Pharmacy; Pyrimidines; Rankings
• ISSN: 0091-150X; 1573-9031
• DOI: 10.1007/s11094-021-02474-1

Dependences of the antiglycation activity of N -hydroxybenzoyl-substituted thymine and uracil derivatives with regular changes of compound structure (nitrogenous base type, number of substitutions, hydroxy group position in benzoyl radical) on their frontier molecular orbital energies (E LUMO , E HOMO ) and the HOMO-LUMO gap (∆E OMO-LUMO ) were studied. A continuous symbatic dependence between the experimentally established and calculated parameters was not found using a ranked correlation analysis. Clustered values of antiglycating activity and calculated quantum-chemical parameters could form homogeneous data groups by using multiple range correlation analysis if the analysis was supplemented with information about the variable molecular structural fragments as information on the binary/ternary properties. This indicated that the dependence of the antiglycating activity of the compounds on E HOMO , E LUMO , and ∆E HOMO-LUMO was complicated. However, it could be approximated by a discrete model if the analyzed data were supplemented with information about the variable structural fragments. Neural network modeling was performed to formulate a model of the dependence of the antiglycating activity exclusively on the calculated E HOMO , E LUMO , and ∆E HOMO-LUMO values. It was shown that the dependence of the antiglycating activity on E HOMO , E LUMO , and ∆E HOMO-LUMO could be described highly accurately by a correlation neural network model. The resulting dependence was used to formulate a quantum-chemical QSAR model for predicting the antiglycating activity of N -hydroxybenzoyl derivatives of thymine and uracil based on the molecular orbital energies.