Characterization of Novel A3 Adenosine Receptor Allosteric Modulators

INTRODUCTION: A3 adenosine receptor (A3 AR) agonists are currently being evaluated in clinical trials for treatment of inflammation, cancer, and neuropathic pain. To circumvent complications associated with the use of direct agonists of GPCRs (selectivity, dose-limiting side-effects), we have pursued development of A3 adenosine receptor positive allosteric modulators (PAMs). These compounds potentiate signaling while allowing for spatiotemporal specificity and decreased off target-effects. METHODS: Prior structure-activity-relationship (SAR) studies identified the 1H-imidazo[4,5]quinoline-4-amine, LUF6000, and the 2,4-disubstituted quinoline, LUF6096, as exhibiting human A3 AR PAM activity. We observed a ~2-3-fold increased maximal effect in G protein-dependent signaling assays, although they had the tendency to reduce agonist potency. In addition, these lead compounds were substantially less potent at rodent A3 ARs. The purpose of the present investigation was to re-evaluate two expanded compound series based on LUF6000 and LUF6096 in greater detail, with emphasis on examining effects on orthosteric ligand potency, G protein-dependent versus G protein-independent signaling, and efficacy versus mouse A3 adenosine receptors. Based on known species differences, human/mouse chimeric receptors were prepared to determine the region of the receptor comprising the allosteric ligand binding pocket. RESULTS: Our evaluation centered on a series of 2-substituted imidazoquinolinamine and comparable 4-substituted quinoline derivatives consisting of 3-12-membered cycloalkyl rings and their carbon-bridged analogues. This was based on our observation during an initial screen that the 2-cycloheptyl and 2-cyclooctyl imidazoquinolinamine analogues increased affinity of the orthosteric radioligand ([125 I]I-AB-MECA) in equilibrium binding assays in a manner that did not slow its (apparent) association binding rate. Among this series, compounds with larger 2-substituents (c4-9) were found to increase the maximal efficacy of the orthosteric agonist Cl-IB-MECA in [35 S]GTPγS binding assays ~2-fold. Most prominently, the 2-cyclononanyl derivative increased the efficacy of Cl-IB-MECA (~2.5-fold) while also increasing its potency. 4-substituted quinoline derivatives displayed a similar SAR with larger 4-substituents (c4-9) increasing maximal efficacy - however, none of the tested derivatives increase orthosteric radioligand binding affinity at equilibrium. While human A3 G-protein coupling profiles were similar for all agonists examined, with robust coupling being detected to all Gαi proteins, probe-dependent effects were observed for β-arrestin-2 recruitment. Moreover, while LUF6000 and LUF6096 potentiated agonist-induced Gαi protein activation, enhancement of β-arrestin-2 recruitment was agonist-dependent. Studies with human/mouse chimeric receptors indicated that modulators based on the imidazoquinolinamine template do not function as 'vestibule' modulators but instead bind intracellularly. CONCLUSIONS: We have identified improved PAMs for the human A3 receptor that: 1) not only increase agonist efficacy, but also increase agonist potency (and binding affinity), and 2) may bias receptor signaling towards G protein-dependent signaling. Furthermore, we have evidence supporting a novel intracellular allosteric binding pocket on the A3 AR. © FASEB.