Synthesis, CoMFA analysis, and receptor docking of 3,5-diacyl-2,4-dialkylpyridine derivatives as selective A(3) adenosine receptor antagonists

3,5-Diacyl-2,4-dialkyl-6-phenylpyridine derivatives have been found to be s elective antagonists at both human and rat A(3) adenosine receptors (Li et al. J. Med. Chem. 1998, 41, 3186-3201), In the present study, ring-constrai ned, fluoro, hydroxy, and other derivatives in this series have been synthe sized and tested for affinity at adenosine receptors in radioligand binding assays. K-i values at recombinant human and rat A(3) adenosine receptors w ere determined using [I-125]AB-MECA (N-6-(4-amino-3-iodobenzyl)-5 '-N-methy lcarbamoyladenosine). Selectivity for A(3) adenosine receptors was determin ed vs radioligand binding at rat brain A(1) and A(2A) receptors, and struct ure-activity relationships at various positions of the pyridine ring (the 3 - and 5-acyl substituents and the 2- and 4-alkyl substituents) were probed. At the 5-position inclusion of a beta-fluoroethyl (7) or a gamma-fluoropro pyl ester (26) was favorable for human A(3) receptor affinity, resulting in K-i values of 4.2 and 9.7 nM, respectively, while the pentafluoropropyl an alogue was clearly less potent at human A(3) receptors. At the 2-, 3-, and 4-positions, fluoro or hydroxy substitution failed to enhance potency and s electivity at human A(3) receptors. Several analogues were nearly equipoten t at rat and human A(3) receptors. To further define the pharmacophore conf ormationally, a lactam, a lactone, and thiolactones were tested in adenosin e receptor binding. The most potent analogue in this group was compound 34, in which a thiolactone was formed between 3- and 4-positions and which had a K-i value of 248 nM at human A(3) receptors. Using affinity data and a g eneral pharmacophore model for A(3) adenosine receptor antagonists recently proposed, we applied comparative molecular field analysis (CoMFA) to obtai n a three-dimensional quantitative structure-activity relationship for pyri dine derivatives, having good predictability (r(pred)(2) = 0.873) for compo unds in the test set. A rhodopsin-based model of the human A(3) receptor wa s built, and the pyridine reference ligand 2,3,4,5-tetraethyl-6-phenylpyrid ine-3-thiocarboxylate-5-carboxylate (MRS 1476) was docked in the putative l igand binding site. Interactions between receptor transmembrane domains and the steric and the electrostatic contour plots obtained from the CoMFA ana lysis were analyzed.