Four-Dimensional Structure–Activity Relationship Model to Predict HIV-1 Integrase Strand Transfer Inhibition using LQTA-QSAR Methodology
journal contributionposted on 23.07.2012 by Eduardo B. de Melo, Márcia M. C. Ferreira
Any type of content formally published in an academic journal, usually following a peer-review process.
Despite highly active antiretroviral therapy (HAART) implementation, there is a continuous need to search for new anti-HIV agents. HIV-1 integrase (HIV-1 IN) is a recently validated biological target for AIDS therapy. In this work, a four-dimensional quantitative structure–activity relationship (4D-QSAR) study using the new methodology named LQTA-QSAR approach with a training set of 85 HIV-1 IN strand transfer inhibitors (INSTI), containing the β-diketo acid (DKA) substructure, was carried out. The GROMACS molecular dynamic package was used to obtain a conformational ensemble profile (CEP) and LQTA-QSAR was employed to calculate Coulomb and Lennard–Jones potentials and to generate the field descriptors. The partial least-squares (PLS) regression model using 14 field descriptors and 8 latent variables (LV) yielded satisfactory statistics (R2 = 0.897, SEC = 0.270, and F = 72.827), good performance in internal (QLOO2 = 0.842 and SEV = 0.314) and external prediction (Rpred2 = 0.839, SEP = 0.384, AREpred = 4.942%, k = 0.981, k′ = 1.016, and |R02 – R0′2 = 0.0257). The QSAR model was shown to be robust (leave-N-out cross validation; average QLNO2 = 0.834) and was not built by chance (y-randomization test; R2 intercept = 0.109; Q2 intercept = −0.398). Fair chemical interpretation of the model could be traced, including descriptors related to interaction with the metallic cofactors and the hydrophobic loop. The model obtained has a good potential for aid in the design of new INSTI, and it is a successful example of application of LQTA-QSAR as an useful tool to be used in computer-aided drug design (CADD).