Computational study of the interactions among structural analogues of acyl homoserine lactones (AHLs) and the Agrobacterium tumefaciens TraR binding site

Abstract

Background: In the present investigation, relationships between a set of 34 analogues of N-acyl-L-homoserine lactones (AHL) and the TraR receptor were studied. The aim was to use molecular modeling as a strategy for elucidating important aspects of the mechanism of chemical signaling in the Gram-negative bacteria Agrobacterium tumefaciens, with the idea of ​​identifying some of analogues’ structural characteristics and molecular interactions with the active site of the TraR receptor. Methods: For this purpose, we combine two molecular modeling strategies: molecular docking and three-dimensional quantitative structure-activity relationship (3D-QSAR). First, the molecular docking methodology was applied to a series of 34 analogues of AHL on the TraR transcriptional receptor to simulate the binding of analogues at the active TraR site. Secondly, 3D-QSAR models were generated to describe the correlation with the experimental biological activity using partial least squares (PLS) calculations and steric and electrostatic properties, which theoretically predict the activity of the 34 AHL analogues through statistical parameters and evaluate the prediction of the models obtained. Two alignment models were constructed; one using the optimized structures of the 34 analogues (ligand-based model) and another using the conformations of the best poses generated in the docking with TraR (receptor-based model). Results: The outcomes obtained for each protein-ligand complex showed that the Aspartic acid 70 and Threonine 129 residues are residues that participate in the formation of hydrogen bonds, while residues Alanine 38, Leucine, 40, Tyrosine 53, Glutamine 58, Tyrosine 61, Phenylalanine 62 and Valine 72 form hydrophobic interactions. These interactions are important in determining the antagonistic activity of the analogues under study against TraR. Conclusions: The ligand-based model produces better statistical results expressed in terms of several rigorous evaluation criteria, such as Q2 and R2 for the data sets than those of the receptor-based model.