Relationships between conformational analysis and QSAR: the cases for the volatile anesthetic isoflurane and for a xylulokinase enzyme ligand

Abstract

The bioactive conformation of molecules used in 3D-QSAR (Tridimensional Quantitative Structure-Activity Relationships) is usually unknown. In this case, the use of molecular descriptors obtained from chemical structures optimized in a receptor-free environment can induce errors either in the prediction or interpretation of the QSAR models. In recent years, 3D-QSAR has been the most explored QSAR methodology, despite considering only the interactions governing the stability of a molecule in the isolated state, rather than in the biological environment. Therefore, this work aimed to evaluate how valuable 3D descriptors are in providing high performance QSAR models, by comparing the results with a bidimensional QSAR technique, which has been shown to be highly predictive and requires a low computation cost related to memory and disk space. The conformational analysis of the inhalational anesthetic isoflurane was theoretically carried out and the stable conformations were compared to the bioactive structures available in the Protein Data Bank (PDB). Also, a bidimensional MIA-QSAR (Multivariate Image Analysis applied to QSAR) modeling was performed for a series of fluorinated anesthetics similar to isoflurane. In another study, the role of an intramolecular hydrogen bond in 5-deoxy-5-fluoro-D-xylulose (DFX) on its bioconformation was theoretically analyzed. The outcomes indicate that a bidimensional QSAR methodology can be employed with reliability and similar prediction power than tridimensional techniques. In addition, this work showed that intramolecular interactions do not govern the stability of the bioconformation of the studied molecules, which is rather driven by an enzyme induced-fit.