New N,5'-di- and N,2,5'-trisubstituted adenosine derivatives were synthesized in good overall yields. Appropriate 5-O-alkyl-substituted ribose moieties were coupled to 6-chloropurine or 2,6-dichloropurine via Vorbrüggen's glycosylation method. Subsequent amination and deprotection of the intermediates yielded compounds 18-35. Binding affinities were determined for rat adenosine A1 and A2A receptors and the human A3 receptor. The ability of compounds 18-35 to inhibit forskolin-induced (10 microM) cyclic AMP (cAMP) production and their ability to stimulate guanosine 5'-O-(3-[35S]thio)triphosphate ([35S]GTPgammaS) binding, via either the adenosine A1 receptor or the adenosine A3 receptor, were assessed. N-Cyclopentyl-substituted adenosine derivatives displayed affinities in the low nanomolar range for the adenosine A1 receptor, whereas N-(3-iodobenzyl)-substituted derivatives had high affinity for the adenosine A3 receptor. Compound 22 had the highest affinity for the adenosine A1 receptor (K(i) value of 16 nM), and compounds 20 and 26 had the highest affinities for the adenosine A3 receptor (K(i) values of 4 and 3 nM, respectively). A chlorine substituent at the 2-position either did not affect or slightly increased the adenosine A1 receptor affinity, whereas the A3 receptor affinity was affected differently, depending on the N-substituent. Furthermore, the introduction of chlorine slightly increased the A3/A1 selectivity ratio. At the 5'-position, an O-methyl substituent induced the highest adenosine A1 receptor affinity, whereas an O-ethyl substituent did so for the A3 receptor. All compounds showed partial agonistic effects in both the cAMP and [35S]GTPgammaS assays, although more marked in the latter assay. In general, the 2-chloro derivatives seemed to have lower intrinsic activities compared to the 2-H-substituted compounds on both the adenosine A1 and the adenosine A3 receptors. The compounds with an N-(3-iodobenzyl) substituent displayed the lowest intrinsic activities. Finally, all compounds also showed partially antagonistic behavior in the [35S]GTPgammaS assay.