In this study, ent-kaurenoic acid derivatives were obtained by microbial transformation methodologies and tested against breast cancer cell lines (MCF-7). A multivariate quantitative-structure activity relationship (QSAR) analysis was performed taking into account both microbial transformation derivatives and other analogues previously reported in literature to give some insight into the main features behind the cytotoxic activity displayed by kaurane-type diterpenes against MCF-7 cells. The partial least square regression (PLS) method was employed in the training set and the best PLS model was built with a factor describing 69.92% of variance and three descriptors (logP, εHOMO and εHOMO-1) selected by the Ordered Predictors Selection (OPS) algorithm. The QSAR model provided reasonable regression (Q2 = 0.64, R2 = 0.72, SEC = 0.29 and SEV = 0.33). The model was validated by leave-N-out cross-validation, y-randomization and external validation (R2pred = 0.89 and SEP = 0.27). The selected descriptors indicated that the activity was mainly related to electronic parameters (HOMO and HOMO-1 molecular orbital energies), as well as to logP. These findings suggest that higher activity values are directly related with both higher logP and frontier orbital energy values. The positive relationship between these orbitals and the activity suggests that the ent-kaurenoic acid analogues interaction with the target involves charge displacement, which is entirely consistent with the literature. Based on these findings, three compounds were proposed and one of them was synthesized and tested. The experimental result confirmed the activity predicted by the model.
Keywords: Copaifera ssp (Fabaceae-caesalpinoiodidaeae); Kaurenoic acid derivatives; MCF-7 breast cancer cells; Multivariate QSAR analysis; PLS regression.
Copyright © 2018 Elsevier Ltd. All rights reserved.