| The protease of human immunodeficiency virus1 (HIV-PR) is an essential enzyme for antiviral treatments.
Carbon nanostructures of fullerene derivatives, have nanoscale dimension with a diameter comparable to the
diameter of the active site of HIV-PR which would in turn inhibit HIV. In this research, two dimensional
quantitative structure-activity relationships (2D-QSAR) of fullerene derivatives against HIV-PR activity were
employed as a powerful tool for elucidation the relationships between structure and experimental observations.
QSAR study of 49 fullerene derivatives was performed by employing stepwise-MLR, GAPLS-MLR, and PCAMLR models for variable (descriptor) selection and model construction. QSAR models were obtained with
higher ability to predict the activity of the fullerene derivatives against HIV-PR by a correlation coefficient
(R2 training) of 0.942, 0.89, and 0.87 as well as R2test values of 0.791, 0.67and 0.674 for stepwise-MLR, GAPLSMLR, and PCA -MLR models, respectively. Leave-one-out cross-validated correlation coefficient (R2CV) and Yrandomization methods confirmed the models robustness. The descriptors indicated that the HIV-PR inhibition
depends on the van der Waals volumes, polarizability, bond order between two atoms and electronegativities of
fullerenes derivatives. 2D-QSAR simulation without needing receptor's active site geometry, resulted in useful
descriptors mainly denoting “C60 backbone-functional groups” and “C60 functional groups” properties. Both
properties in fullerene refer to the ligand fitness and improvement van der Waals interactions with HIV-PR
active site. Therefore, the QSAR models can be used in the search for novel HIV-PR inhibitors based on
fullerene derivatives. |