Recent Advances on A3 Adenosine Receptor Antagonists by QSAR Tools

Feng      Luan; Fernanda      Borges; M.      Natalia D. S. Cordeiro

Abstract

Adenosine receptors (ARs) are widespread on virtually every human organ/tissue, and have long been considered promising therapeutic targets in a wide range of conditions, ranging from cerebral diseases to cancer, including inflammatory disorders. The knowledge acquired up to date in relation to ARs, in particular regarding the molecular biology of the A3 AR has provided a solid basis that led to the proposal of this receptor as a novel therapeutic target enabling the rational design and development of potent and selective A3 AR ligands. This review attempts to summarize the most recent developments in the A3 research field, focusing in particular on Quantitative Structure-Activity Relationships (QSAR) based studies that supported so far the design of new, potent and selective human A3 AR antagonists. In addition, a classical QSAR modeling study carried out on two series of pyrazolo-triazolopyrimidine derivatives is presented as a case study. Specifically, a systematic evaluation of linear and non-linear models along with a variety of structure representations and feature selection tools is reported. The combination of these techniques (neural networks to capture non-linear relationships in the data and feature selection to prevent over-fitting) was found to produce QSAR models with good overall accuracy and robustness, as well as predictivity on external data. Moreover, the study indicated that the antagonist activity of these derivatives is largely explained by electrostatic, steric and hydrogen-bonding factors, highlighting the role of the size, shape and type of inhibitor in forming effective blocking of the A3 AR subtype. The developed QSAR models could then be usefully employed to design new compounds selectively active towards the A3 adenosine receptor.

Keywords: Human A3 adenosine receptor subtype, Quantitative structure–activity relationships, Multiple linear regression, Radial basis function neural networks, Antagonists

« Previous Next »