Kiralj R., Ferreira M. M. C., "Comparative chemometric and QSAR/SAR study of structurally unrelated substrates of a MATE efflux pump VmrA from V. parahaemolyticus: prediction of multidrug resistance", QSAR Comb. Sci., 27(3), 314-329 (Mar 2008).
[Article]
 

Abstract.
Multidrug resistance activities pEC₅₀ of sensitive, resistant, and moderately resistant strains of the pathogenic fungus Penicillium digitatum against triflumizole, fenarimol, bitertanol, pyrifenox, cycloheximide, acriflavine, and 4-nitroquinoline-N-oxide were studied by principal component and hierarchical cluster analyses. Genome descriptors for fungal cytochrome P450 sterol 14a-demethylase and multidrug efflux pump PMR1 were generated and correlated with pEC₅₀ by partial least squares. Toxicants were modeled at the PM3 level. Novel Activity-Structure Relationships (ASRs) were established to predict toxicant structural features from biological activities and to identify and classify the strains. New types of relationships to model and predict biological activities are Quantitative Genome-Activity Relationship (QGAR) and Quantitative Genome/Structure-Activity Relationship (QGSAR). QGAR for demethylation inhibitors had a reasonable regression [Q²=0.79, R²=0.81, Standard Error of Validation (SEV)=0.34] and was extended into QGSAR with improved statistics (Q²=0.85, R²=0.87, SEV=0.29). Conformers of toxicants with common hydrogen bonding and aromatic ring geometry indicate possible interactions with receptors such as cytochromes, efflux pumps, and regulatory proteins which activate fungal multidrug resistance.

Keywords.
Agent-Receptor Interactions; Molecular Descriptors; Multidrug Resistance; Regression Models; Semiempirical Methods.

Keywords Plus.