Quantification of adenosine A₁ receptor biased agonism: Implications for drug discovery

Abstract

Adenosine A₁ receptor (A₁AR) stimulation is a powerful protective mechanism in cerebral and cardiac ischemia⿿reperfusion injury. Despite this, therapeutic targeting of the A₁AR for the treatment of ischemia⿿reperfusion injury has been largely unsuccessful, as high concentrations of prototypical A₁AR agonists impart significant hemodynamic effects, particularly pronounced bradycardia, atrioventricular block and hypotension. Exploiting the phenomenon of biased agonism to develop ligands that promote A₁AR cytoprotection in the absence of adverse hemodynamic effects remains a relatively unexplored, but exciting, approach to overcome current limitations. In native systems, the atypical A₁AR agonists VCP746 and capadenoson retain cytoprotective signaling in the absence of bradycardia, a phenomenon suggestive of biased agonism. The current study used pharmacological inhibitors to investigate A₁AR mediated cytoprotective signal transduction in a CHO FlpIn cell background, thus identifying candidate pathways for quantitative bias profiling, including cAMP, extracellular signal-regulated kinases 1 and 2 and Akt1/2/3. Subsequently, effects on cell survival and the bias profile of VCP746 and capadenoson were determined and compared to that of the prototypical A₁AR agonists, NECA, R-PIA, MeCCPA and CPA. We found that prototypical agonists do not display significant bias for any of the pathways assessed. In contrast, VCP746 and capadenoson show significant bias away from calcium mobilization relative to all pathways tested. These studies demonstrate that quantitative ⿿fingerprinting⿿ of biased agonism within a model system can enable ligands to be clustered by their bias profile, which in turn may be predictive of preferential physiologically relevant in vivo pharmacology.

Introduction

Adenosine is an ubiquitous purine nucleoside that is involved in a range of physiological functions [1]. The nucleoside achieves this functionality by binding to one of the four adenosine receptors (ARs), namely, the A₁AR, A_2AAR, A_2BAR, and A₃AR, all of which are G protein-coupled receptors (GPCRs) [2]. The A₁AR and A₃AR preferentially couple to G_i/o proteins, whilst the A_2AAR and A_2BAR preferentially couple to G_s proteins [2], [3].

Under conditions of cellular stress, such as inflammation and hypoxia, adenosine can reach micromolar concentrations in the pericellular environment, acting via adenosine receptors to promote tissue protection and repair [4], [5]. This is particularly pertinent in ischemia⿿reperfusion injury (IRI), where lack of blood supply coupled with pro-inflammatory reperfusion, leads to serious organ and tissue damage [6], [7]. Activation of A₁ARs can elicit powerful cytoprotection in myocardial or cerebral IRI [8], [9], [10], likely involving signaling downstream of the mitogen activated protein kinases (MAPK), the protein kinase B (Akt/PKB) family of serine/threonine protein kinases and a number of protein kinase C (PKC) isoforms [11], [12], [13], [14], [15], [16], [17], [18], [19].

Whilst A₁AR agonism promotes profound cardioprotection, clinical trials of A₁AR agonists have shown only limited success, as the trialed dosing was suboptimal due to undesirable hemodynamic side effects, including bradycardia [20]. Under these conditions, A₁AR-mediated bradycardia largely occurs via an increase in the current of inward rectifying potassium channels in the pacemaker cells of the heart [21], [22], [23]. Additionally, in the presence of increased sympathetic tone, A₁AR activation can reduce heart rate via inhibition of adenylate cyclase activation and subsequent effects on hyperpolarization-activated cyclic nucleotide-gated channels [22]. Consequently, the difficulty of A₁AR-targeted medicines is that both the therapeutic effect, cytoprotection, and the clinically limiting effect, bradycardia, are mediated by the same target.

One approach that may allow for the separation of on-target beneficial effects from adverse effects is exploitation of biased agonism. This is a phenomenon whereby different ligands, acting at the same GPCR and in the same cellular background, can promote distinct receptor conformations that bias the signal either towards or away from a subset of the cellular pathways normally mediated by activation of the receptor [24]; in the case of the A₁AR, this could represent signaling promoting cytoprotection in the absence of signaling producing on-target bradycardia. Importantly, we recently designed a novel bitopic (hybrid orthosteric/allosteric) A₁AR ligand, 4-(5-amino-4-benzoyl-3-(3-(trifluoromethyl)phenyl)thiophen-2-yl)-N-(6-(9-((2R,3R,4S,5R)-3,4-dihydroxy-5 (hydroxymethyl)tetrahydrofuran-2-yl)-9H-purin-6-ylamino)hexyl)benzamide (VCP746), that displayed biased agonism in a heterologous expression system, as well as reducing isolated neonatal rat cardiomyocyte death due to simulated ischemia while having no effect on isolated rat atrial rate [25]. This finding provided the first proof of concept that potentially therapeutically useful biased agonism can be observed at the A₁AR, but also raises a number of important questions. For instance, is the bias promoted by VCP746 due to its atypical extended structure, given that it is a hybrid molecule composed of orthosteric and allosteric pharmacophores? Can other atypical A₁AR ligands promote similar or distinct patterns of bias? Is it possible to develop a ⿿signaling fingerprint⿿ in a recombinant model system that can be used to group A₁AR ligands with regards to their propensity to engender physiologically relevant bias in a native cellular system?

The aim of the current study was to begin addressing these questions. We first extended our preliminary determination of VCP746 bias by identifying a broad range of signaling pathways linked to A₁AR activation in a heterologous system, that are potentially relevant to cytoprotection, and can thus be used to generate a ⿿bias fingerprint⿿ as a surrogate predictor of physiologically relevant bias. We subsequently compared the bias profile of VCP746 to that of the prototypical A₁AR agonists 5⿲-N-ethylcarboxamidoadenosine (NECA), (R-)-N6-phenylisopropyladenosine (R-PIA), 2-chloro-N-cyclopentyl-2⿲-methyl adenosine (MeCCPA) and N6-cyclopentyladenosine (CPA), as well as another atypical agonist, capadenoson (Fig. 1). The latter was chosen because it is also an extended non-nucleoside molecule that mediates cardioprotection in the absence of adverse hemodynamic effects [26], [27]. We find that both VCP746 and capadenoson display a distinct pattern of bias, whereas canonical A₁AR ligands do not, providing further support for the role of quantitative bias profiling in model systems to better inform potential drug candidate selection.