Probe dependence of allosteric enhancers on the binding affinity of adenosine A1‐receptor agonists at rat and human A1‐receptors measured using NanoBRET

Background and Purpose Adenosine is a local mediator that regulates a number of physiological and pathological processes via activation of adenosine A1‐receptors. The activity of adenosine can be regulated at the level of its target receptor via drugs that bind to an allosteric site on the A1‐receptor. Here, we have investigated the species and probe dependence of two allosteric modulators on the binding characteristics of fluorescent and nonfluorescent A1‐receptor agonists. Experimental Approach A Nano‐luciferase (Nluc) BRET (NanoBRET) methodology was used. This used N‐terminal Nluc‐tagged A1‐receptors expressed in HEK293T cells in conjunction with both fluorescent A1‐receptor agonists (adenosine and NECA analogues) and a fluorescent antagonist CA200645. Key Results PD 81,723 and VCP171 elicited positive allosteric effects on the binding affinity of orthosteric agonists at both the rat and human A1‐receptors that showed clear probe dependence. Thus, the allosteric effect on the highly selective partial agonist capadenoson was much less marked than for the full agonists NECA, adenosine, and CCPA in both species. VCP171 and, to a lesser extent, PD 81,723, also increased the specific binding of three fluorescent A1‐receptor agonists in a species‐dependent manner that involved increases in B max and pK D. Conclusions and Implications These results demonstrate the power of the NanoBRET ligand‐binding approach to study the effect of allosteric ligands on the binding of fluorescent agonists to the adenosine A1‐receptor in intact living cells. Furthermore, our studies suggest that VCP171 and PD 81,723 may switch a proportion of A1‐receptors to an active agonist conformation (R*).

Adenosine is a local reactive metabolite that has a major role in regulating a number of physiological and pathological processes including inflammation, hypoxia, and cardiovascular regulation (Fredholm, Ijzerman, Jacobson, Linden, & Müller, 2011). Adenosine acts via four specific GPCRs, which have been denoted adenosine A 1 -, A 2A -, A 2B -, and A 3 -receptors (Fredholm et al., 2011). The A 1and A 3 -receptors preferentially couple to G i/o proteins and have an inhibitory action on adenylyl cyclase activity whilst the A 2Aand A 2B -receptors couple to G s proteins and stimulate cAMP formation (Fredholm et al., 2011;. The crystal structures of the A 2A -receptor in both antagonist (Jaakola et al., 2008) and agonist (Xu et al., 2011) bound conformations have been determined, and very recently, the structure of the adenosine A 1 -receptor has also been solved (Cheng et al., 2017;Glukhova et al., 2017), including an adenosinebound A 1 -receptor in complex with a G i -protein (Draper-Joyce et al., 2018).
Numerous selective agonists and antagonists for each adenosine receptor subtype are now available for the study of receptor function (see Fredholm et al., 2011;. In the case of the adenosine A 1 -receptor, a number of compounds have previously undergone evaluation for cardiovascular disease indications such as paroxysmal supraventricular tachycardia, atrial fibrillation, and angina pectoris . At the present time, the A 1 -receptor partial agonist neladenoson is undergoing clinical trial for heart failure (Meibom et al., 2017). However, the ubiquitous distribution of adenosine receptors in the body can often limit therapeutic application because of the effects of adenosine ligands on the same receptor in a different tissue or cell type .
Activation of cell surface adenosine receptors by endogenous adenosine requires it to be available at the extracellular surface of cells. Extracellular adenosine can rise as a consequence of several pathways (Fredholm et al., 2011). It can be formed intracellularly following various metabolic processes and be exported from cells via membrane transporters, or it can be formed in the extracellular space from adenine nucleotides released from cells. Once ATP or ADP is released, the nucleotide is broken down by nucleoside triphosphate diphosphohydrolases (e.g., CD39) and then ecto-5′-nucleotidase (CD73) to adenosine (Fredholm et al., 2011;Knapp et al., 2012). The intricacies of localized extracellular release of adenine nucleotides and subsequent production of adenosine following CD73 activity has recently provided insights into the role of adenosine A 1 -receptors in mediating localized analgesia in animals and humans (Goldman et al., 2010;Sowa, Voss, & Zylka, 2010;Street & Zylka, 2011). In addition, there is increasing evidence that adenosine A 1 -receptors may be involved in promoting angiogenesis and the release of VEGF in response to local hypoxia and neoplasia (Clark et al., 2007;Merighi et al., 2009).
From the foregoing argument, it is clear that localized regulation of adenosine production may have important therapeutic implications.
The potential for allosteric enhancers to provide highly localized augmentation of adenosine actions on target receptors is well established . However, the in vivo actions of allosteric regulators have not been extensively investigated, and there is a need to evaluate the potential for these small molecules to augment specific actions of adenosine in particular organs or cell types in a whole animal setting. Some limited success has been achieved in vivo with PD 81,723. Activation of adenosine A 1 -receptors has been shown to protect against renal ischaemia/reperfusion injury in experimental animals (Lee & Emala, 2000;Lee, Gallos, Nasr, & Emala, 2004;Park et al., 2012). However, despite the high homology between the species homologues of the A 1 -receptor, there is evidence for species What is already known • Adenosine is a local mediator that regulates physiological processes via activation of adenosine A 1receptors.
• Agonist activity can be regulated by drugs that bind to an allosteric site on the A 1 -receptor.

What this study adds
• This study demonstrates the power of fluorescent ligand NanoBRET approaches to study allosterism in cells.
• Positive allosteric modulators can switch a proportion of A 1 -receptors to an active agonist-binding conformation.

What is the clinical significance
• This study provides insights into allosteric mechanisms that may provide new opportunities for drug discovery. differences in the affinity of certain adenosine receptor ligands Szymańska et al., 2016).

| Constructs, cell lines, and cell culture
Human and rat Nluc-labelled adenosine A 1 -receptor (Nluc-A 1 R) constructs were generated as previously described by Stoddart, Johnstone, et al. (2015). In brief, the full-length sequence of Nluc luciferase from the pNL1.1 vector (Promega) was amplified and fused in-frame with the membrane signal sequence of the 5-HT 3A membrane localization signal sequence (pcDNA3.1 sig-Nluc; Soave, Stoddart, Brown, Woolard, & Hill, 2016). This was fused to the full-length human or rat sequence of the adenosine A 1 -receptor (with the methionine start signal removed) to the 3′ end of the sig-Nluc in pcDNA3.1. The resulting fusion protein contained a Gly-Ser linker between the Nluc open reading frame and the human or rat A 1 open reading frame. This resulted in the human and rat Nluc-A 1 R constructs.

| Cultured cells
HEK293T cells (ATCC Cat# CRL-3216, RRID:CVCL_0063) were maintained in DMEM supplemented with 2 mM L-glutamine and 10% fetal calf serum at 37°C 5% CO 2 . Once 70-80% confluent, cells were dislodged from the flask surface by gentle shaking after incubation in 0.25% trypsin and collected following centrifugation at 1000× g for 5 min. Cells were then seeded at 2-5 × 10,000 cells cm -2 . Mixed population human Nluc-A 1-AR and rat Nluc-A 1-AR cell lines were generated using Fugene HD (Promega) according to the manufacturer's instructions, and cells were then subjected to 1 mg/mL G418-selection pressure for 2 weeks.

| BRET human and rat Nluc-A 1 R ligand-binding assays
The fluorescent antagonist saturation, competition-binding, allosteric modulator binding cooperativity, and the fluorescent agonist saturations in the presence/absence of allosteric modulator assays were performed on the stably transfected HEK293T cells expressing human or rat Nluc-A 1 R. The cells were seeded 24 hr before experimentation in white walled, poly-D-lysine coated 96-well microplates (Thermo Scientific, Loughborough, UK) at a density of 25,000 cells per well.
The medium was replaced with HEPES-buffered saline solution (145 nM NaCl, 5 mM KCl, 1.7 mM CaCl 2 , 1 mM MgSO 4 , 10 mM HEPES, 2 mM sodium pyruvate, 1.5 mM NaHCO 3 , 10 mM D-glucose, pH 7.2-7.45), with the required concentration of fluorescent ligand, competing ligand, and/or allosteric modulator. For each experiment, ligands were added simultaneously, and the 96-well plate was incubated for 1 hr at 37°C (no CO 2 ). Following this, the Nluc substrate furimazine (Promega) was added to give a final concentration of 10 μM and then incubated for 5 min at 37°C. For all experiments, the luminescence and resulting BRET were measured using the PHERAstar FS plate reader (BMG Labtech) using filtered light emissions at 460 nm (80 nm bandpass) and >610 nm (longpass) at room temperature. The raw BRET ratio was calculated by dividing the >610 nm emission by the 460 nm emission.

| Data analysis
Data were presented and analysed using Prism 7 software (GraphPad software, San Diego, CA, USA).
Saturation-binding curves were simultaneously fitted to obtain the total and non-specific components using the following equation: where B max is the maximal level of specific binding, [B] is the concentration of fluorescent ligand in nM, K D is the equilibrium dissociation constant in nM, M is the slope of the linear non-specific binding component, and C is the y-axis intercept.
Competition NanoBRET data was fitted using a one-site sigmoidal competition curve given by the following equation: where [A] is the concentration of competing drug, NS is the nonspecific binding, n is the Hill coefficient, and IC 50 is the concentration of ligand required to inhibit 50% of the specific binding of the fluorescent ligand.
The IC 50 values from competition-binding curves were used to calculate the K i of the unlabelled ligands using the Cheng-Prusoff equation:

BJP
where [L] is the concentration of fluorescent ligand in nM, and K D is the dissociation constant of the fluorescent ligand in nM. The K D values used were obtained from the saturation-binding experiments.
Pooled fluorescent agonist saturation assays obtained in the presence and absence of a fixed concentration of allosteric modulator were simultaneously fitted to the following equation: The slope of the non-specific binding component M was kept constant (equivalent to the slope of the binding curve obtained in the presence of 1 μM DPCPX in the same experiments), and a partial F test was used to determine whether a significantly better fit was obtained with individual parameters for B max and K D for each curve (control vs. that obtained in the presence of VCP171 or PD 81,723) when compared with sharing the parameters between curves.

| Statistical analysis
The statistical analyses in this study comply with the recommendations on experimental design and analysis in pharmacology (Curtis et al., 2018). Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test, partial F test, or unpaired Student's t test. In all cases, differences were considered significant at P < 0.05. All statistical analysis was performed using GraphPad Prism 7.03 (RRID:SCR_002798). In all cases, individual experiments were performed in triplicate, and statistical analysis was performed on the data obtained from five or six repeat experiments. The fluorescent A 1-receptor agonist, ABEA-X-BY630, was synthesized as previously described by Middleton et al. (2007). The fluorescent A 1receptor agonist, BY630-X-(D)-A-(D)-A-G-ABEA, was synthesized as described by Stoddart, Vernall, et al. (2015). Fugene HD transfection reagent and furimazine were from Promega (Southampton, UK). All other reagents were from Sigma-Aldrich (Gillingham, UK).

| Nomenclature of targets and ligands
Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org/, the common portal for data from the IUPHAR/BPS Guide to PHARMA-COLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander et al., 2017).
3.1 | Measurement of the specific binding of CA200645 to rat and human adenosine A 1 -receptors using NanoBRET We have recently described a bioluminescence energy transfer approach (NanoBRET) to monitor ligand-receptor interactions in living HEK293T cells expressing the human A 1 -receptor tagged on its Nterminus with the luminescence protein Nluc (Stoddart, Johnstone, et al., 2015). Here, we have compared the ligand-binding characteristics of Nluc-tagged human and rat adenosine A 1 -receptors using the fluorescent antagonist ligand CA200645 (Stoddart, Johnstone, et al., 2015). Binding experiments were performed over a large range of concentrations of CA200645 (1-500 nM) and yielded clear saturable components of specific binding for both species receptor homologues with negligible non-specific binding detected in the presence of a high concentration of the A 1 -receptor selective antagonist DPCPX (1 μM).
The K D values obtained for CA200645 for the specific binding component were 33.84 ± 10.15 nM (n = 6) and 35.44 ± 4.66 nM (n = 6) for the human and rat Nluc-A 1 -receptors, respectively.

| Inhibition of binding by A 1 -receptor ligands
The binding affinities of non-fluorescent A 1 -receptor ligands at the two species homologues were then determined from competitionbinding studies in the presence of 25 nM CA200645 (Table 1).
In the case of VCP171, 3, 10, or 30 μM concentrations of this allosteric regulator not only produced a clearer decrease in the specific binding of CA200645 alone to the human A 1 -receptor but also produced significant decreases in the IC 50 for adenosine, CCPA, and NECA (Figure 2e,g,h; Table 2) without producing a significant change in the IC 50 of capadenoson (Figure 2f; Table 2).
VCP171 produced a significant increase in the specific binding of the adenosine analogue ABA-X-BY630 to both the human ( Figure 5a) and rat A 1 -receptors (Figure 5c). Partial F test analysis of the non-linear regression fits to the combined data shown in Figure 3 confirmed that the two binding parameters (pK D and B max ) differed significantly between the control and VCP171 (10 or 30 μM) curves (i.e., they could not be shared; Figure 5a,c). In the case of the human receptor, this could also be ascribed to a significant change in the B max value (partial F test). Analysis of the mean parameters from the individual repeat experiments (Tables 3 and 4) confirmed a significant increase in B max and pK D values for the human (Table 3) but not the rat A 1 -receptor (Table 4) (Table 4).
In the case of the NECA derivative ABEA-X-BY630, significant increases in specific binding were detected with VCP171 at both A 1receptor species homologues (Figure 6a,c). At the human A 1 -receptor, this was due to significant changes in both pK D and B max (Table 3), whereas for the rat homologue, it was more dependent upon an increase in B max (Table 4). PD 81,723 did not significantly change any of the binding parameters for the human (Table 3) or rat (Table 4) A 1 -receptors, but there was a very small elevation in overall specific binding at the human A 1 -receptor but not the rat (Figure 6b,d).
For the tripeptide linker variant of ABEA-X-BY630 (AAG-ABEA-X-BY630) a similar profile was observed to that obtained with ABEA-X-BY630. Significant increases in specific binding were detected with VCP171 at both A 1 -receptor species homologues (Figure 7a,c). At the human A 1 -receptor, this appeared to be due to significant changes in both pK D and B max (Table 3); although for the rat homologue, this was more dependent on an increase in pK D (Table 4). F test analysis of the combined data for the human A 1 -receptor also indicated an effect of VCP171 on pK D (Figure 7a). In the case of PD 81,723, no consistent of effect of this allosteric enhancer was observed on AAG-ABEA-X-BY630 binding in the rat although a small significant increase in overall specific binding was detectable at the human A 1 -receptor.
This enabled accurate determination of the binding affinities of competing A 1 -receptor ligands to be made. This work confirmed the species differences in the affinity of DPCPX reported previously in brain membrane homogenates between the human and rat A 1 -receptor homologues (Maemoto et al., 1997). Interestingly, the allosteric regulators PD 81,723 and VCP171 produced a small direct inhibition of CA200645 binding at high concentrations (>100 μM for PD 81,723 and 10-100 μM VCP171) that was more evident at the human A 1 -receptor.
*P < 0.05 compared to 0 allosteric modulator; one-way ANOVA, post hoc Tukey's test. . Data are expressed as mean ± SEM. *P < 0.05 fitted parameters (both K D and B max ) curves significantly different from control (without allosteric modulator; partial F test) antagonist DPCPX (Glukhova et al., 2017). It was noticeable, however, that these authors did also identify a putative secondary binding pocket in this inactive A 1 -receptor structure that could represent a site that is involved in allosteric regulation (Glukhova et al., 2017).
A striking feature of the inactive human A 1 -receptor crystal structure obtained in complex with the covalently bound antagonist DU-172 is that the ECL2 residues form an α-helix that extends away from the transmembrane regions of the receptor in a manner that is almost perpendicular to the plane of the membrane (Glukhova et al., 2017;Figure 8a). This is a region of the receptor that mutagenesis studies have suggested is crucial to both the functional efficacy of the agonist NECA (Nguyen, Baltos, et al., 2016) and to the ability of PD 81,723 and VCP171 to elicit allosteric effects on orthosteric agonist binding (Nguyen, Vecchio, et al., 2016). It is therefore possible that ECL2 undergoes a conformational change following agonist binding to bring these residues in closer juxtaposition to the large binding pocket of the A 1 -receptor that contains the orthosteric binding site (Glukhova et al., 2017). However, the recent structure of the adenosine-occupied The lack of a significant effect of allosteric enhancers on the binding of the partial A 1 -receptor agonist capadenoson also suggests a reduced ability of this agonist to switch the receptor from R to R* or indeed to produce a different "partially active" R* conformation.
Thus, for example, recent structural information published for the   . Data are expressed as mean ± SEM. *P < 0.05 fitted parameters (both K D and B max ) curves significantly different from control (without allosteric modulator; partial F test). # P < 0.05 fitted parameter for K D significantly different from control (partial F test) highlighted is E172 that appears to be important for the direct binding of PD 81,723 and VCP171 to the human A 1 -receptor (Nguyen, Vecchio, et al., 2016). It is notable that the mutations identified by Nguyen, Baltos, et al. (2016) do not include residues that are different between species. The amino acid sequence that represents the α-helix of EL2 in the human sequence is highlighted in grey in Figure 9a. Helix prediction analysis using PredictProtein (Rost, Yachdav, & Liu, 2004) confirmed that there was no change in helix propensity in the two species and that the α-helix is in the same position for the rat protein sequence. It is notable that four of the five residues that are different between rat and human A 1 -sequence are in the α-helical region. This suggests that the M162V change (which is adjacent to the helix domain and also to two residues mutated by Nguyen, Vecchio, et al., 2016) may underlie some of the subtle changes in allosteric action between the two species.
A direct effect of both VCP171 and PD 81,723 on agonist binding was also evident from NanoBRET studies using fluorescent A 1 -receptor-agonists. This was particularly clear for VCP171 where the allosteric modulator significantly increased the level of specific binding of all three fluorescent agonists tested at the human A 1 -receptor. This was a consequence of both an increase in maximal binding capacity (B max ) and affinity (pK D ). Consistent with this positive allosteric effect, we have previously shown that PD 81,723 (10 μM) can slow the dissociation of BY630-ABA from human A 1 -receptors expressed in CHO cells (May et al., 2010). The effect of VCP171 was, however, also species dependent with an increase in both binding parameters evident with all three fluorescent agonists at the human A 1 -receptor, but the effects were limited to B max (ABEA-X-BY630) and pK D (AAG-ABEA-X-BY650) for particular fluorescent agonists at the rat species homologue. It is also worth pointing out that our results suggest that the functionalization and addition of fluorophore to the agonist chemical scaffold has been achieved at a point that does not clash with the allosteric modulator binding site.
The data obtained with VCP171 at the human A 1 -receptor are consistent with an increase in the proportion of a higher affinity agonist conformation (R*; Figure 9c) that was only detectable with values might also suggest that this property underlies the ability of allosteric enhancers to produce direct agonist actions in the absence of orthosteric agonists (Nguyen, Vecchio, et al., 2016). Thus, the increased formation of active receptor conformations (R*) in the presence of the allosteric regulator may lead to increased stimulation of intracellular signalling pathways. An increased conversion of inactive receptor (R) to active receptor conformations (R*) by VCP171 and PD 81,723 may also explain the decrease in specific binding of the fluorescent antagonist CA200645 observed above, although the structure of the inactive receptor also indicates that the allosteric ligands can bind to the orthosteric binding site (Glukhova et al., 2017).
In summary, the present study has shown that PD 81,723 and VCP171 can elicit positive allosteric effects on the binding affinity of orthosteric agonists at both the rat and human adenosine A 1 -receptors. This work also confirms that these two allosteric regulators exhibit both probe and species homologue dependence. Thus, the allosteric effect on the highly selective partial agonist capadenoson is much less marked than for the full agonists NECA and adenosine in both species. In addition, at higher concentrations, both allosteric regulators have a direct inhibitory effect on the binding of the orthosteric fluorescent antagonist CA200645 that is consistent with the suggestion from crystallographic studies that indicates that they can also bind directly to the orthosteric binding site of the A 1 -receptor. Finally, VCP171 and, to a lesser extent, PD 81,723, were also able to increase the specific binding of three fluorescent A 1 -receptor agonists in a species-dependent manner that involved increases in B max and pK D . This latter effect may provide new insights into the mechanisms by which allosteric enhancers can elicit functional responses in the absence of orthosteric A 1 -receptor agonists.