Adenosine A1-A2A Receptor-Receptor Interaction: Contribution to Guanosine-Mediated Effects

Guanosine, a guanine-based purine nucleoside, has been described as a neuromodulator that exerts neuroprotective effects in animal and cellular ischemia models. However, guanosine’s exact mechanism of action and molecular targets have not yet been identified. Here, we aimed to elucidate a role of adenosine receptors (ARs) in mediating guanosine effects. We investigated the neuroprotective effects of guanosine in hippocampal slices from A2AR-deficient mice (A2AR−/−) subjected to oxygen/glucose deprivation (OGD). Next, we assessed guanosine binding at ARs taking advantage of a fluorescent-selective A2AR antagonist (MRS7396) which could engage in a bioluminescence resonance energy transfer (BRET) process with NanoLuc-tagged A2AR. Next, we evaluated functional AR activation by determining cAMP and calcium accumulation. Finally, we assessed the impact of A1R and A2AR co-expression in guanosine-mediated impedance responses in living cells. Guanosine prevented the reduction of cellular viability and increased reactive oxygen species generation induced by OGD in hippocampal slices from wild-type, but not from A2AR−/− mice. Notably, while guanosine was not able to modify MRS7396 binding to A2AR-expressing cells, a partial blockade was observed in cells co-expressing A1R and A2AR. The relevance of the A1R and A2AR interaction in guanosine effects was further substantiated by means of functional assays (i.e., cAMP and calcium determinations), since guanosine only blocked A2AR agonist-mediated effects in doubly expressing A1R and A2AR cells. Interestingly, while guanosine did not affect A1R/A2AR heteromer formation, it reduced A2AR agonist-mediated cell impedance responses. Our results indicate that guanosine-induced effects may require both A1R and A2AR co-expression, thus identifying a molecular substrate that may allow fine tuning of guanosine-mediated responses.


Introduction
Guanosine is a guanine-based purine nucleoside that has been shown to exert neuroprotective and neurotrophic effects in both in vitro and in vivo studies (for review, see [1]). Thus, it has been postulated as a good candidate for the management of several central nervous system (CNS) disorders, including neurodegenerative diseases (i.e., Parkinson's, Alzheimer's) or ischemia [1,2]. Brain ischemia is one of the major health disability conditions worldwide [3]. It occurs after a blood supply collapse that leads to a reduced level of oxygen and glucose within the affected brain area. Similarly, upon excitotoxicity and oxidative stress a failure of cellular bioenergetics occurs [4]. Importantly, a neuroprotective role of guanosine has been extensively investigated in animal and cellular models of ischemia, excitotoxicity and oxidative stress [5][6][7][8][9][10]. Indeed, we have demonstrated that guanosine prevents reactive oxygen species (ROS) generation and cell death in hippocampal slices subjected to the oxygen/glucose deprivation (OGD) [11].
The mechanism by which guanosine exerts its neuroprotective effects is still intriguing. Despite the identification of a putative guanosine binding site in rat brain membranes [12], a specific guanosine receptor has not yet been discovered. Importantly, it has been hypothesized that adenosine receptors (ARs) may play a role in mediating guanosine effects, although with some controversy. For instance, it has been reported that AR selective ligands do not compete for guanosine binding to rat brain membranes [13,14], whereas AR ligands were able to block some of the guanosine-dependent neuroprotective effects [15]. In line with this, a selective adenosine A 1 receptor (A 1 R) antagonist (DPCPX, 8-cyclopentyl-1,3-dipropylxanthine) and a selective A 2A receptor (A 2A R) agonist (CGS21680, 2-(4-(2-carboxyethyl)phenethylamino)-5 -N-ethylcarboxamidoadenosine) inhibited guanosine-mediated neuroprotection in hippocampal slices subjected to OGD [11]. Overall, these findings, including those using multimodal A 1 R and A 2A R ligand treatments, supported the notion that both A 1 R and A 2A R would participate in guanosine-mediated effects.
Interestingly, it has been hypothesized that adenosine A 1 and A 2A receptor-receptor interactions (i.e., heteromerization) might be behind some of the guanosine-mediated effects, thus pointing to the A 1 R/A 2A R heteromer as a putative molecular target for guanosine [16]. Indeed, the existence of A 1 R/A 2A R heteromers has been demonstrated in presynaptic terminals of striatal neurons controlling glutamate release [17], thus acting as an adenosine concentration-dependent switch [18]. Consequently, low to moderate concentrations of adenosine predominantly activate A 1 R within the A 1 R/A 2A R heteromer (i.e., inhibiting glutamate release), whereas moderate to high concentrations of adenosine also activate A 2A R, which, by means of the A 1 R-A 2A R intramembrane negative allosteric interaction, antagonizes A 1 R function, therefore facilitating glutamate release. Altogether, in view of the already known experimental indications, the A 1 R/A 2A R heteromer might be viewed as a potential target for guanosine, thus deserving further attention. Here, we aimed to assess the role of A 1 R and A 2A R interaction in guanosine-mediated effects. First, we studied the neuroprotective effects of guanosine in an ex vivo model of brain ischemia, both in wild-type and A 2A R deficient (A 2A R −/− ) mice; subsequently, we aimed to elucidate, in vitro, both the putative guanosine binding and activation of the A 1 R/A 2A R heteromer.

Animals
Wild-type and A 2A R −/− CD-1 male and female mice [20] weighing 25-50 g were used at 2-3 months of age. The University of Barcelona Committee on Animal Use and Care (CEEA-UB) approved the protocol (Code 10033, 04/02/2018). Animals were housed and tested in compliance with the guidelines described in the Guide for the Care and Use of Laboratory Animals [21] and following the European Union directives (2010/63/EU), FELASA and ARRIVE guidelines. Mice were housed in groups of five in standard cages with ad libitum access to food and water and maintained under a 12-h dark/light cycle (starting at 7:30 AM), 22 • C temperature, and 66% humidity (standard conditions).

OGD Protocol
Mice were euthanized by cervical dislocation and hippocampi rapidly removed and placed in an ice-cold Krebs-Ringer bicarbonate buffer (KRB) ( PO 4 , and 5 HEPES), where 10 mM d-glucose was replaced by 10 mM 2-deoxy-glucose and equilibrated with a 95% N 2 /5% CO 2 gas mixture, as described previously [5] After 15 min of OGD the media of the slices was replaced by oxygenated KRB and maintained for 2 h for evaluation of cellular viability and ROS generation. Guanosine (100 µM), when present, was added 15 min before (in KRB) and during OGD (in OGD buffer), and maintained in the re-oxygenation period (2 h), when the OGD buffer was replaced by physiological KRB.

Measurement of ROS Production
For evaluating ROS generation, slices were incubated with 80 µM 2 ,7 -dichlorofluorescein diacetate (DCFH-DA; Sigma-Aldrich) for 30 min [23]. Then, subsequent to the OGD/reoxygenation protocol, slices were washed twice with KRB and maintained for 15 min before adding DCFH-DA. H 2 DCFDA diffuses through the cell membrane, and it is hydrolyzed by intracellular esterases to the non-fluorescent form dichlorofluorescin (DCFH). Afterwards, DCFH can react with intracellular H 2 O 2 to form dichlorofluorescein (DCF), a green fluorescent dye. Slices were then transferred to a 96-well black plate containing 200 µL of KRB, and fluorescence was read (excitation 480 nm, emission 525 nm) using a POLARStar plate reader (BMG Labtech).

Plasmid Constructs
The cDNA encoding the human A 1 R tagged at its N-terminal tail with the O6-alkylguanine-DNA alkyltransferase (i.e., A 1 R SNAP ) cloned in pRK5 vector (BD PharMingen, San Jose, CA, USA) was a gift from Prof. Jean-Philippe Pin (CNRS, Montpellier, France). Thus, to perform functional assays A 2A R SNAP [24] and A 1 R SNAP were used. Also, A 2A R RLuc and A 1 R YFP constructs [17] were used to perform classical BRET (Bioluminescence Resonance Energy Transfer) assays. Finally, to perform NanoBRET experiments with the MRS7396 fluorescent antagonist, we created an A 2A R NanoLuc sensor (A 2A R NL ). To this end, the cDNA encoding the human A 2A R was amplified by polymerase chain reaction from the pECFP-A 2A R vector using the primers: FA2AEco (5 -GCCGGAATTCCCCATCATGGGCTCC TCGGTGTAC-3 ) and RA2ANot (5 -CGCGGCGGCCGCtcaggacactcctgctccatcctggg-3 ). The amplified A 2A R insert was then cloned into the EcoRI/NotI sites of pNLF1-secN vector (Promega, Stockholm, Sweden) containing a hemagglutinin (HA) epitope tag. All the constructs were verified by DNA sequencing.

NanoBRET Experiments
The NanoBRET assay was performed on stably expressing (A 2A R NL ) HEK-293T cells, transiently transfected (or not) with A 1 R SNAP , according to [25]. In brief, cells were re-suspended in HBSS, and seeded into poly ornithine coated white 96-well plates. After 24 h, cells were challenged with/without the non-labelled A 2A R antagonist (SCH442416) or guanosine and incubated for 1 h at 37 • C. Subsequently, the fluorescent ligand (MRS7396) was added and the plate and returned to 37 • C for 1 h. Finally, coelenterazine-h (Life Technologies Corp.) was added at a final concentration of 5 µM, and readings were performed after 5 min using a CLARIOStar plate reader (BMG Labtech). The donor and acceptor emissions were measured at 490-510 nm and 650-680 nm, respectively. The raw NanoBRET ratio was calculated by dividing the 650 nm emission by the 490 nm emission. In competition studies, results were expressed as a percentage of the maximum signal obtained (mBU; milliBRET Units).

cAMP Assay
cAMP accumulation was measured using the LANCE ® Ultra cAMP Kit (PerkinElmer, Waltham, MA, USA) as previously described [26]. In brief, transfected (A 2A R SNAP or A 2A R SNAP + A 1 R SNAP ) HEK-293T cells were firstly incubated for 1 h at 37 • C with stimulation buffer (BSA 0.1%, ADA 0.5 units/mL, zardaverine 2 µM; in serum-free DMEM) and later on with CGS21680 for 30 min at 37 • C. Thereafter, cells were transferred to a 384-well plate in which reagents were added following manufacturer's instructions. After 1 h at room temperature, Time-Resolved-Fluorescence Resonance Energy Transfer (TR-FRET) was determined by measuring light emission at 620 nm and 665 nm by means of a CLARIOstar plate reader (BMG Labtech).

Intracellular Calcium Determinations
The A 1 R-mediated intracellular Ca 2+ accumulation was assessed by means of a luciferase reporter assay based on the expression of the nuclear factor of activated T-cells (NFAT), as previously described [27]. In brief, cells were transfected with the cDNA encoding the A 1 R, the NFAT-luciferase reporter (pGL4-NFAT-RE/luc2p; Promega) and the yellow fluorescent protein (pEYFP-N1; Promega). After 36 h post-transfection, cells were incubated with the indicated drugs for 6 h. Subsequently, cells were harvested with passive lysis buffer (Promega), and the luciferase activity of cell extracts was determined using a luciferase Bright-Glo TM assay (Promega) in a POLARStar plate-reader (BMG Labtech) using a 30-nm bandwidth excitation filter at 535 nm.

Label-Free Cellular Impedance Assay
The xCELLigence Real-Time Cell Analyzer (RTCA) system (ACEA Biosciences, San Diego, CA, USA) was employed to measure changes in cellular impedance correlating with cell spreading and tightness, thus being widely accepted as a morphological and functional biosensor of cell status [28][29][30]. Thus, 16-well E-plates (ACEA Biosciences) were coated with 50 µL fibronectin (10 µg/mL) at 37 • C for 1 h before being washed three times with 100 µL MilliQ-water before use. The background index for each well was determined with 90 µL of stimulation buffer (supplemented DMEM with ADA 0.5 U/mL and zardaverine 10 µM) in the absence of cells. Data from each well were normalized to the time point just before compound addition using the RTCA software providing the normalized cell index (NCI). Subsequently, HEK-293T cells permanently expressing the A 2A R SNAP construct [31] in the absence or presence of A 1 R SNAP (90 µL resuspended in stimulation buffer) were then plated at a cell density of 40,000 cells/well and grown for 18 h in the RTCA SP device station (ACEA Biosciences) at 37 • C and in an atmosphere of 5% CO 2 before ligand (i.e., CGS21680 and/or guanosine) addition. Cell index values were obtained immediately following ligand stimulation every 15 s for a total time of at least 50 min. For data analysis, the area under the curve (AUC) for each NCI trace response was quantified and normalized to the basal.

Statistics
Data are represented as mean ± standard error of mean (SEM). The number of samples/animals (n) in each experimental condition is indicated in the corresponding figure legend. Comparisons among experimental groups were performed by Student's t-test and ANOVA, using GraphPad Prism 6.01 (San Diego, CA, USA), as indicated. Statistical difference was accepted when p < 0.05.

Guanosine-Mediated Neuroprotection in Hippocampal Slices Depends on A 2A R Expression
It has been postulated that ARs might be involved in guanosine-mediated responses in vivo [16]. Within this line of inquiry, we first interrogated whether A 2A R expression is necessary for guanosine-mediated neuroprotection, a well-known guanosine effect in vivo [1]. To this end, we subjected hippocampal slices from wild-type (i.e., A 2A R +/+ ) and A 2A R −/− mice to an OGD protocol in the presence or absence of guanosine. Indeed, significant cell death (p < 0.001) and ROS production (p = 0.0359) were observed in A 2A R +/+ hippocampal slices subjected to the OGD protocol ( Figure 1A,B). Interestingly, guanosine (100 µM) was able to prevent these effects, thus cellular viability significantly increased (p = 0.0012) and ROS production decreased (p = 0.0389) ( Figure 1A,B), as previously reported [5,11]. Importantly, under the same experimental conditions, in hippocampal slices obtained from A 2A R −/− mice, guanosine failed to prevent OGD-mediated cell death (p = 0.005) and ROS production (p = 0.0279) ( Figure 1A,B), thus losing its neuroprotective effect. Overall, these results suggested that A 2A R expression was necessary for guanosine-mediated neuroprotection.

A 2A R Ligand Binding is Affected by Guanosine upon A 1 R Coexpression
Once we demonstrated that the neuroprotective effect of guanosine was A 2A R-dependent, we aimed to assess the putative direct interaction of guanosine with A 2A R through ligand binding studies. To this end, we engineered a fluorescent ligand BRET-based assay to assess A 2A R ligand binding in living cells (Figure 2A). We used a fluorescent A 2A R antagonist (MRS7396) that is able to engage in a BRET process upon interacting with a cell surface A 2A R tagged with the NanoLuciferase (NL) at its N-terminus (i.e., A 2A R NL ) (Figure 2A). MRS7396 is a BODIPY630/650 derivative of SCH442416 [19], which upon A 2A R binding can act as an acceptor chromophore for NanoLuciferase emission (490 nm) in a BRET process. Thus, we challenged stable A 2A R NL -expressing cells with increasing concentrations of MRS7396, in the presence/absence of non-labelled SCH442416. Interestingly, a bell-shaped binding saturation hyperbola, with a K D = 4.8 ± 2.7 nM, was obtained for MRS7396, while in the presence of a saturating concentration of SCH442416 (1 µM) the binding was displaced ( Figure 2B). Our results showed that the NanoBRET binding assay was a robust and reliable way to assess A 2A R ligand binding. Accordingly, we next assessed possible guanosine effects on A 2A R orthosteric binding by performing a competition assay with a fixed concentration of MRS7396 (10 nM) (occupying~80% of receptors at equilibrium) and increasing concentrations of guanosine. Interestingly, under these experimental conditions, guanosine was unable to alter MRS7396 binding to A 2A R NL ( Figure 2C), thus indicating that guanosine does not orthosterically bind to A 2A R, as previously reported [12,13].
Since A 2A R heteromerizes with A 1 R [17], and some of the physiological effects of guanosine were modulated by A 1 R ligands [32,33], we investigated whether A 1 R/A 2A R heteromer formation affected AR-related guanosine-dependent effects. To this end, we first recreated the formation of A 1 R/A 2A R heteromers in HEK-293T cells by transfecting A 2A R RLuc and A 1 R YFP constructs and monitoring A 2A R/A 1 R heteromerization by a classical BRET approach ( Figure A1). Interestingly, neither adenosine nor guanosine incubation altered A 1 R/A 2A R heteromer formation ( Figure A1). Subsequently, we assessed the impact of A 1 R co-expression in A 2A R binding of MRS7396 using our NanoBRET binding assay. Notably, in A 1 R-A 2A R doubly expressing cells, guanosine (100 µM) was able to significantly reduce by 19 ± 4% (p = 0.0138) the binding of MRS7396 to the A 2A R NL , thus indicating that the A 1 R/A 2A R heteromer might play a potential role in AR-related guanosine-dependent effects ( Figure 2C). nM) (occupying ~80% of receptors at equilibrium) and increasing concentrations of guanosine. Interestingly, under these experimental conditions, guanosine was unable to alter MRS7396 binding to A2AR NL ( Figure 2C), thus indicating that guanosine does not orthosterically bind to A2AR, as previously reported [12,13].

A 2A R Signalling, but Not A 1 R, is Modulated by Guanosine in an A 1 R Coexpression-Dependent Manner
Given that guanosine reduced A 2A R binding in an A 1 R-expression-dependent manner, we next aimed to determine whether guanosine also impinged into A 2A R signaling. Accordingly, we determined the effects of guanosine in A 2A R-mediated cAMP accumulation upon agonist incubation. In A 2A R-expressing cells, the selective A 2A R full agonist CGS21680 induced a concentration-dependent cAMP accumulation (pEC 50 = 7.98 ± 0.08), indicating that the receptor was expressed and functional at the plasma membrane ( Figure 3A). Subsequently, we challenged cells with a fixed concentration of CGS21680 (200 nM) and evaluated the effects of increasing concentrations of guanosine in A 2A R-dependent cAMP accumulation. As shown in Figure 3B, guanosine did not preclude A 2A R-mediated cAMP accumulation. Conversely, in cells doubly expressing A 1 R and A 2A R, guanosine (100 µM) was able to significantly reduce, by 19 ± 3% (p = 0.0460), the A 2A R-mediated cAMP accumulation ( Figure 3B). These results supported the hypothesis that the effects of guanosine might be dependent on an A 1 R-A 2A R interaction.
Interestingly, our NanoBRET-based binding results and cAMP determinations in the absence and presence of A 1 R suggested a direct involvement of this receptor in guanosine-mediated blockade of A 2A R ligand binding and signaling. Thus, to ascertain whether guanosine would directly interact with A 1 R we assessed its impact on A 1 R-dependent signaling. To this end, A 1 R-mediated calcium responses in HEK-293T cells were determined through a homogenous bioluminescence reporter assay system using a NFAT response element controlling luciferase gene expression. While the activation of A 1 R, via application of the agonist N 6 -R-phenylisopropyladenosine (R-PIA, 50 nM), increased intracellular Ca 2+ , the incubation with guanosine (100 µM) did not promote intracellular Ca 2+ mobilization ( Figure 4A). Similarly, when A 1 R-expressing cells were treated with R-PIA in the presence of increasing concentrations of guanosine, A 1 R-dependent intracellular Ca 2+ mobilization was not affected, as observed in doubly A 1 R and A 2A R transfected cells ( Figure 4B). Overall, these results indicated that guanosine did not interact with A 1 R, thus ruling out any orthosteric A 1

R-dependent trans-inhibition of A 2A R function in A 1 R-A 2A R expressing cells.
In A2AR-expressing cells, the selective A2AR full agonist CGS21680 induced a concentrationdependent cAMP accumulation (pEC50 = 7.98 ± 0.08), indicating that the receptor was expressed and functional at the plasma membrane ( Figure 3A). Subsequently, we challenged cells with a fixed concentration of CGS21680 (200 nM) and evaluated the effects of increasing concentrations of guanosine in A2AR-dependent cAMP accumulation. As shown in Figure 3B, guanosine did not preclude A2AR-mediated cAMP accumulation. Conversely, in cells doubly expressing A1R and A2AR, guanosine (100 µM) was able to significantly reduce, by 19 ± 3% (p = 0.0460), the A2AR-mediated cAMP accumulation ( Figure 3B). These results supported the hypothesis that the effects of guanosine might be dependent on an A1R-A2AR interaction.  Finally, we assessed the functional activity of guanosine using the label-free technology. To this end, the whole-cell guanosine-mediated impedance responses were monitored in living cells expressing A 2A R in the absence or presence of A 1 R using a biosensor method, as previously reported [34]. First, we tested CGS21680-mediated changes in morphology (i.e., impedance) of A 2A R SNAP expressing HEK-293T cells, which were recorded in real-time. Interestingly, addition of CGS21680 resulted in a significant (p = 0.015) increase of impedance, which was blocked by incubation with the selective A 2A R antagonist ZM241385 ( Figure 5A,B). In addition, guanosine did not affect the cell basal morphology (p = 0.6105) nor its CGS218680-mediated changes (p = 0.1217) ( Figure 5B). However, in doubly expressing A 1 R/A 2A R cells guanosine significantly reduced (p < 0.0106) cell basal morphology and precluded (p < 0.0001) the CGS218680-induced increase in cellular impedance ( Figure 5B). Again, these results indicated that the A 1 R-A 2A R co-expression may play a potential role in AR-related guanosine-dependent cellular effects. activation of A1R, via application of the agonist N 6 -R-phenylisopropyladenosine (R-PIA, 50 nM), increased intracellular Ca 2+ , the incubation with guanosine (100 µM) did not promote intracellular Ca 2+ mobilization ( Figure 4A). Similarly, when A1R-expressing cells were treated with R-PIA in the presence of increasing concentrations of guanosine, A1R-dependent intracellular Ca 2+ mobilization was not affected, as observed in doubly A1R and A2AR transfected cells ( Figure 4B). Overall, these results indicated that guanosine did not interact with A1R, thus ruling out any orthosteric A1Rdependent trans-inhibition of A2AR function in A1R-A2AR expressing cells. Finally, we assessed the functional activity of guanosine using the label-free technology. To this end, the whole-cell guanosine-mediated impedance responses were monitored in living cells expressing A2AR in the absence or presence of A1R using a biosensor method, as previously reported [34]. First, we tested CGS21680-mediated changes in morphology (i.e., impedance) of A2AR SNAP expressing HEK-293T cells, which were recorded in real-time. Interestingly, addition of CGS21680 resulted in a significant (p = 0.015) increase of impedance, which was blocked by incubation with the selective A2AR antagonist ZM241385 ( Figure 5A and B). In addition, guanosine did not affect the cell basal morphology (p = 0.6105) nor its CGS218680-mediated changes (p = 0.1217) ( Figure 5B). However, in doubly expressing A1R/A2AR cells guanosine significantly reduced (p < 0.0106) cell basal morphology and precluded (p < 0.0001) the CGS218680-induced increase in cellular impedance ( Figure 5B). Again, these results indicated that the A1R-A2AR co-expression may play a potential role in AR-related guanosine-dependent cellular effects.

Discussion
Guanosine is a purine nucleoside with widely demonstrated extracellular neuromodulatory effects in the CNS, but so far without an identified receptor. Based on the use of selective ligands, ARs have been proposed as possible targets to explain guanosine-mediated effects in animal and cellular models of ischemia. However, at present, the mechanism of action of guanosine is not clear.

Discussion
Guanosine is a purine nucleoside with widely demonstrated extracellular neuromodulatory effects in the CNS, but so far without an identified receptor. Based on the use of selective ligands, ARs have been proposed as possible targets to explain guanosine-mediated effects in animal and cellular models of ischemia. However, at present, the mechanism of action of guanosine is not clear. Here, we show that A 2A R expression was crucial for guanosine-mediated protective effects in an ex vivo model of brain ischemia. In addition, when examining guanosine effects in a controlled heterologous system, we were able to reveal the importance of a proposed A 1 R-A 2A R interaction mediating guanosine effects, both in A 2A R-ligand binding and in receptor function.
In the OGD ischemia model in hippocampal slices, we previously showed that guanosine induced a neuroprotective effect (increase of glutamate uptake) that was inhibited by activation of A 2A R by CGS2180 [11]. This effect of CGS21680 in abolishing a guanosine-evoked increase in glutamate uptake in an OGD protocol was also observed in cultured astrocytes expressing the astrocytic glutamate transporter Glt-1 [15]. Therefore, here we evaluated guanosine's neuroprotective effects in A 2A R −/− mice and revealed an important role for this receptor. Thus, in A 2A R −/− hippocampal slices, we observed a loss of the neuroprotective effects of guanosine (increasing viability and controlling ROS production in OGD conditions) that were observed in slices from wild-type mice ( Figure 6A). This result, consistent with previous data, pointed to ARs as possible targets for guanosine [35,36], prompting us to further explore the mechanism by which guanosine might act. ( Figure 6A). This result, consistent with previous data, pointed to ARs as possible targets for guanosine [35,36], prompting us to further explore the mechanism by which guanosine might act. While guanosine does not interfere with A1R-dependent signaling, it modulates A2AR binding and intracellular signaling (i.e., cAMP accumulation and cellular morphology) only in A1R-A2AR co-expressing cells. Therefore, A1R and A2AR may constitute a molecular substrate involved in guanosine-mediated effects, but the precise mechanism of action of guanosine involving ARs is still lacking.
Our NanoBRET-based sensor data suggested that, as previously reported [13], guanosine apparently does not bind directly to the A2AR. However, in A1R/A2AR cells, it was possible to observe a guanosine-mediated partial displacement of A2AR-ligand binding ( Figure 6B). Together with the ex vivo data, this result would indicate that the mechanism of action of guanosine would be mediated by this receptor-receptor entity. Indeed, previous data showing both DPCPX-and pertussis toxindependent blockade of protective effects of guanosine in hippocampal slices subjected to OGD [11], supported the dependence on functional A1Rs coupled to a G-protein to mediate guanosine effects.
We found that guanosine reduced A2AR orthosteric binding only in A1R-A2AR expressing cells. Thus, we evaluated whether guanosine could modulate A2AR-dependent signaling under the same experimental conditions. Interestingly, while guanosine did not preclude CGS21680-induced cAMP accumulation in A2AR-expressing cells, it reduced A2AR-mediated cAMP accumulation in doubly in an A 1 R-dependent manner. While guanosine does not interfere with A 1 R-dependent signaling, it modulates A 2A R binding and intracellular signaling (i.e., cAMP accumulation and cellular morphology) only in A 1 R-A 2A R co-expressing cells. Therefore, A 1 R and A 2A R may constitute a molecular substrate involved in guanosine-mediated effects, but the precise mechanism of action of guanosine involving ARs is still lacking.
Our NanoBRET-based sensor data suggested that, as previously reported [13], guanosine apparently does not bind directly to the A 2A R. However, in A 1 R/A 2A R cells, it was possible to observe a guanosine-mediated partial displacement of A 2A R-ligand binding ( Figure 6B). Together with the ex vivo data, this result would indicate that the mechanism of action of guanosine would be mediated by this receptor-receptor entity. Indeed, previous data showing both DPCPX-and pertussis toxin-dependent blockade of protective effects of guanosine in hippocampal slices subjected to OGD [11], supported the dependence on functional A 1 Rs coupled to a G-protein to mediate guanosine effects.
We found that guanosine reduced A 2A R orthosteric binding only in A 1 R-A 2A R expressing cells. Thus, we evaluated whether guanosine could modulate A 2A R-dependent signaling under the same experimental conditions. Interestingly, while guanosine did not preclude CGS21680-induced cAMP accumulation in A 2A R-expressing cells, it reduced A 2A R-mediated cAMP accumulation in doubly A 1 R-A 2A R transfected cells, as observed in the ligand-binding assay ( Figure 6B). Additionally, the evaluation of guanosine effects on the functional activity of ARs using the label-free technology confirmed that guanosine-mediated cell impedance responses were dependent on A 1 R-A 2A R co-expression. Hence, our results indicate that guanosine could attenuate A 2A R signaling (i.e., agonist-mediated cAMP accumulation and cell impedance responses) in an A 1 R-dependent manner ( Figure 6B). On the other hand, when the A 1 R-dependent signaling (i.e., intracellular Ca 2+ mobilization) was assessed, guanosine was unable to modulate receptor's function both in singly and doubly A 1 R-A 2A R transfected cells. Taken together, our results suggest that while guanosine did not signal through A 1 R, it requires this receptor to exert its A 2A R modulatory effect, which could indicate that the A 1 R/A 2A R heteromer might be a molecular substrate for guanosine.
The A 1 R/A 2A R heteromer displays some functional characteristics similar to that reported for other AR-containing oligomers, for instance A 2A R combined with the dopamine D 2 receptor (D 2 R) or the cannabinoid CB 1 receptor (CB 1 R) [37]. Interestingly, these receptor heteromers have been shown to exert reciprocal receptor-receptor allosteric antagonistic interactions [38]. Precisely, an A 1 R/A 2A R heteromer-mediated transmembrane-dependent negative allosteric interaction at the ligand-receptor binding level has been described [39]. In addition, co-activation of both receptors led to a canonical protein Gs-Gi antagonistic interaction at the level of the adenylyl cyclase [40]. This situation makes it difficult to conclude whether an effect in a given signaling pathway is caused by either the allosteric or the canonical interaction. Thus, our data showing that guanosine was able to modulate AR functioning (i.e., cAMP assay) only in cells expressing A 1 R and A 2A R do not permit a clear determination of the interaction at the intracellular level (i.e., canonical protein Gs-Gi antagonistic interaction). However, considering the whole picture, it seems likely that guanosine effects in the physiological context may depend on the co-expression of both receptors and their and interaction. Indeed, guanosine did not disrupt the A 1 R/A 2A R heteromer, as observed by a saturable BRET signal, similar to that obtained following adenosine treatment, and by membrane co-localization of A 1 R and A 2A R in guanosine-treated cells ( Figure A1).
Overall, our data suggest an important role for the A 1 -A 2A receptor-receptor interaction in guanosine-mediated effects. Thus, while our results seem to rule out an eventual guanosine-mediated A 1 R-A 2A R canonical antagonistic interaction, further investigation is needed to ascertain whether guanosine may either modulate the well-known A 1 R-A 2A R allosteric interaction or an indirect mechanism of action yet to be discovered.

Conclusions
In summary, our results revealed that certain AR-related guanosine-mediated effects rely on A 1 R and A 2A R co-expression. Indeed, in ex vivo experiments, the well-known guanosine-mediated neuroprotective effect depends on A 2A R expression. Thus, guanosine failed to protect A 2A R −/− mouse hippocampal slices from ischemia-induced damage. In addition, while guanosine did not interfere with A 1 R-mediated signaling, it modulated A 2A R binding and intracellular signaling only in A 1 R-A 2A R co-expressing cells. Overall, our results suggest that A 1 R and A 2A R may constitute a molecular substrate involved in guanosine effects, but the precise mechanism of action of guanosine involving ARs still is intriguing. bell-shaped BRET saturation curve (BRET 50 = 0.38 ± 0.07 and BRET max = 90 ± 6), thus indicating the formation of constitutive A 1 R-A 2A R complexes in living cells ( Figure A1B). Importantly, under the same experimental conditions, the treatment with either adenosine (100 µM) or guanosine (100 µM) for 2 h did not alter the physical proximity of A 1 R and A 2A R. Thus, neither the BRET 50 [F (2,30) = 1.524, p-value = 0.2343] nor the BRET max [F (2,30) = 0.3135, p-value = 0.7333] was significantly affected by adenosine or guanosine incubation ( Figure A1B). Overall, these results corroborated the formation of A 1 R/A 2A R heterocomplexes in living cells, as previously described [17], and that these complexes were not affected by adenosine or guanosine, consistent with the general notion that GPCR homo-and heteromerization is often constitutive. . Cells transiently transfected with A2AR SNAP and A1R SNAP and incubated with vehicle or guanosine (100 µM) for 2h. Cells were processed for immunocytochemical (ICC) detection of A2AR (red) and A1R (green) using specific antibodies (see Appendix A1). Merged images reveal codistribution of A2AR SNAP and A1R SNAP (yellow). Scale bar: 100 µm. (B) BRET saturation curve between A2AR and A1R. BRET was measured in HEK-293T cells co-expressing A2AR Rluc and A1R YFP constructs and incubated with vehicle, adenosine (100 µM) or guanosine (100 µM) for 2 h. Cells were cotransfected with a fixed amount of A2AR Rluc and increasing amounts A1R YFP . Plotted on the X-axis is the fluorescence value obtained from the YFP, normalized with the luminescence value of the Rluc constructs 10 min after coelenterazine h incubation and in the Y-axis the corresponding BRET ratio (× 1000). mBU: mBRET units. Results are expressed as mean ± SEM of four independent experiments grouped as a function of the amount of acceptor fluorescence.