cells Article Adenosine A2A Receptor Antagonists Affects NMDA Glutamate Receptor Function. Potential to Address Neurodegeneration in Alzheimer’s Disease Rafael Franco 1,2, * , Rafael Rivas-Santisteban 1,2 , Mireia Casanovas 1,2 , Alejandro Lillo 3 , Carlos A. Saura 2,4 and Gemma Navarro 2,3, * 1 Departament de Bioquímica i Biomedicina Molecular, Universitat de Barcelona, 08028 Barcelona, Spain; rrivasbioq@gmail.com (R.R.-S.); mcasanfe8@gmail.com (M.C.) 2 Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas, Instituto de Salud Carlos III, Valderrebollo, 5, 28031 Madrid, Spain; carlos.saura@uab.cat 3 Departament of Biochemistry and Physiology, Faculty of Pharmacy and Food Science, Universitat de Barcelona, 08028 Barcelona, Spain; alilloma55@gmail.com 4 Institut de Neurociències, Department de Bioquímica i Biologia Molecular, Universitat Autònoma de Barcelona, Bellaterra, 08193 Barcelona, Spain * Correspondence: rfranco123@gmail.com (R.F.); g.navarro@ub.edu (G.N.)  Received: 27 March 2020; Accepted: 19 April 2020; Published: 26 April 2020  Abstract: (1) Background. N-methyl d-aspartate (NMDA) ionotropic glutamate receptor (NMDAR), which is one of the main targets to combat Alzheimer’s disease (AD), is expressed in both neurons and glial cells. The aim of this paper was to assess whether the adenosine A2A receptor (A2A R), which is a target in neurodegeneration, may affect NMDAR functionality. (2) Methods. Immuno-histo/cytochemical, biophysical, biochemical and signaling assays were performed in a heterologous cell expression system and in primary cultures of neurons and microglia (resting and activated) from control and the APPSw,Ind transgenic mice. (3) Results. On the one hand, NMDA and A2A receptors were able to physically interact forming complexes, mainly in microglia. Furthermore, the amount of complexes was markedly enhanced in activated microglia. On the other hand, the interaction resulted in a novel functional entity that displayed a cross-antagonism, that could be useful to prevent the exacerbation of NMDAR function by using A2A R antagonists. Interestingly, the amount of complexes was markedly higher in the hippocampal cells from the APPSw,Ind than from the control mice. In neurons, the number of complexes was lesser, probably due to NMDAR not interacting with the A2A R. However, the activation of the A2A R receptors resulted in higher NMDAR functionality in neurons, probably by indirect mechanisms. (4) Conclusions. A2A R antagonists such as istradefylline, which is already approved for Parkinson’s disease (Nouriast® in Japan and Nourianz® in the US), have potential to afford neuroprotection in AD in a synergistic-like fashion. i.e., via both neurons and microglia. Keywords: G-protein-coupled receptors; functional selectivity; microglia; neuroprotection; cognition; signaling 1. Introduction Alzheimer’s disease (AD) is characterized by two pathological hallmarks, amyloid plaques composed of β-amyloid peptides and neurofibrillary tangles composed of hyperphosphorylated tau protein. Patients are currently treated with two drug types: N-methyl d-aspartate ionotropic glutamate receptor (NMDAR) modulators and acetylcholinesterase inhibitors. An anti-AD approved drug acting on NMDAR, memantine (marketed in many Countries as Namenda® ), is a negative Cells 2020, 9, 1075; doi:10.3390/cells9051075 www.mdpi.com/journal/cells Cells 2020, 9, 1075 2 of 16 allosteric modulator acting as a low-affinity open channel blocker. The drug was developed to find a weak negative modulator as both strong or weak NMDAR activities are noxious [1–3]. Unfortunately, the efficacy of current drugs is low and none prevent the disease progression [4]. It is well established that AD correlates with NMDAR functional alterations that cannot be addressed by drugs acting in the orthosteric center; in fact, NMDAR-related drugs used in AD patients are allosteric modulators of the receptor. Adenosine is an autacoid, i.e., a hormone-like locally acting molecule, that exerts metabolic and regulatory functions in almost any tissue and cell type of the mammalian body. In the brain, it is one of the main neuromodulators acting via four G-protein-coupled receptors (GPCRs): A1 , A2A , A2B and A3 . The adenosine A2A receptor (A2A R), a Gs -coupled GPCR, is heavily expressed in the motor control brain areas [5] but is also expressed in other CNS regions [5]. Remarkably, an antagonist of the receptor, istradefylline, has been recently approved as a first-in-class drug for the treatment of Parkinson’s disease [6–8] (Nouriast® in Japan and Nourianz® in the US). Apart from combating motor symptoms and/or minimizing the side effect of anti-parkinsonian medication, A2A R antagonists may be neuroprotective. Mechanisms of neuroprotection are based in in vitro and in vivo pharmacological studies. Pioneering work using adenosine itself led to a review entitled “Cerebral Protection by Adenosine” [9]. Based on data using brain ischemia-reperfusion models, it was already evident that adenosine could achieve the same neuroprotective effects as NMDAR blockers. The data came from studies focused on neurons, and the molecular mechanisms, including the type of adenosine receptor, were not known. As pointed out below, NMDAR are expressed in glia where adenosine receptors are also expressed. It is noteworthy that the A2A R is upregulated in activated microglia [10]. More recent studies conducted to explore the therapeutic possibilities to combat AD have used different assays with genetic ablation or a pharmacological blockade of the adenosine A2A receptor, resulting in neuroprotection in different models. Briefly, taking into account the most recent reports, the expression of the A2A receptor is altered even in peripheral blood cells from patients with AD or vascular dementia [11]. In addition, deletion of the receptor was beneficial in a tauopathy mouse model [12] and A2A R antagonists were protective in both the APPswe/PS1dE9 [13] and the triple 3×Tg-AD transgenic AD models [14]. Finally it should be noted that the early synaptic events seemed mediated by the A2A R in the 3×Tg-AD transgenic AD mouse model [15]. Based on the extensive background, the aim of this paper was to investigate whether the activation of the A2A R may regulate NMDAR function in both neurons and microglia cells. Although NMDAR function is instrumental for neurotransmission, senescent neurons have little resources to prevent death and rely on the support provided by glial cells. Among them, microglia are of interest since A2A R expression is enhanced in activated microglia where the NMDAR is also expressed [16–19]. Our results suggested that A2A R antagonists may provide neuroprotection via the modulation of NMDAR functionality. 2. Materials and Methods 2.1. Reagents Lipopolysaccharide (LPS) and interferon-γ (IFN-γ) were purchased from Sigma Aldrich (St Louis, MO, USA); receptor ligands were purchased from Tocris Bioscience (Bristol, UK). Agonists were: N-methyl d-aspartate (NMDA) 4-[2-[[6-Amino-9-(N-ethyl-β-d-ribofuranuronamidosyl)-9 H-purin-2-yl]amino]ethyl]benzenepropanoic acid hydrochloride (CGS-21680). Antagonists were: (5S,10R)-(+)-5-Methyl-10,11-dihydro-5H-dibenzo [a,d]cyclohepten-5,10-imine maleate (MK-801) and 2-(2-Furanyl)-7-(2-phenylethyl)-7H-pyrazolo[4,3-e] [1,2,4]triazolo[1,5-c]pyrimidin-5-amine (SCH-58261). Cells 2020, 9, 1075 3 of 16 2.2. Fusion Proteins To create A2A R-Renilla luciferase (RLuc and A2A R-yellow fluorescent protein (YFP) fusion molecules, the human version of adenosine A2A R cDNA lacking the stop codon, was obtained by PCR and subcloned to a Renilla luciferase (RLuc)-containing vector (pRLuc; PerkinElmer, Wellesley, MA, USA) and a yellow fluorescent protein (YFP)-containing vector (pEYEP-N1; Clontech, Heidelberg, Germany) using sense and antisense primers harboring unique restriction sites for HindIII and BamHI. A similar approach was used to generate the cDNAs for GluN1-RLuc and caldendrin-YFP fusion proteins. 2.3. APP Transgenic Mouse Model of Alzheimer’s Disease (AD) APPSw,Ind transgenic mice (line J9; C57BL/6 background), expressing human APP695 harboring the familial AD-linked Swedish (K670N/M671L) and Indiana (V717F) mutations under the PDGFβ promoter, were obtained by crossing APPSw,Ind to non-transgenic (WT) mice [20]. Animals come from a colony established by co-author Carlos A. Saura in the Autonomous Barcelona University; animals came directly from the laboratory who developed the transgenic animal after signing the ad hoc Transfer Agreement. 2.4. Cell Culture and Transfection HEK-293T human embryonic kidney cells from the American Type Culture Collection (ATCC) were grown in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Paisley, Scotland, UK) supplemented with 2 mM l-glutamine, 100 U/mL penicillin/streptomycin, MEM Non-Essential Amino Acids Solution (1/100) and 5% (v/v) heat inactivated fetal bovine serum (FBS) (Gibco, Paisley, Scotland, UK). The cells were maintained in a humid atmosphere of 5% CO2 at 37 ◦ C. The cells were transiently transfected using Polyethylenimine (PEI, Sigma Aldrich, St. Louis, MO, USA) as previously described [21]. To prepare mice primary microglial cultures (C57BL/6 wild type or transgenic mice), the brain was removed at postnatal days 2 to 4. The microglial cells were isolated as described in [22] and grown in a DMEM medium supplemented with 2 mM l-glutamine, 100 U/mL penicillin/streptomycin, and 5% (v/v) heat inactivated FBS. For neuronal primary cultures, the hippocampus from mouse embryos (E19) was removed and the neurons were isolated as described by Hradsky et al., 2013 [23]. Cells were grown in a neurobasal medium supplemented with 2 mM l-glutamine, 100 U/mL penicillin/streptomycin, supplement (2% v/v) with B27 (Gibco). For cAMP assays, cells were grown on 6-well plates at a density of 500,000 cells/well, for ERK 1/2 phosphorylation assays, cells were placed in 96-well plates at a density of 50,000 cells/well; for proximity ligation assay cells were placed in 12-well plates with coverslips. Cell counting was assessed using trypan blue and a countless II FL automated cell counter (Thermo Fisher Scientific, Waltham, MA, USA). Experiments were carried out 15 days later and the medium was replaced every 5–7 days. The animal handling and protocols were conducted in accordance with the European Council Directive 2010/63/UE as well as in keeping with the current Spanish legislation (RD53/2013). The ethics committee of the two institutions (University of Barcelona and Autonomous University of Barcelona) were in charge of law implementation. 2.5. Immunocytochemistry The transfected HEK-293T cells or primary microglial culture cells seeded in coverslips were fixed in 4% paraformaldehyde for 15 min and washed twice with phosphate-buffered saline (PBS) containing 20 mM glycine, before permeabilization with PBS-glycine containing 0.2% Triton X-100 (5 min incubation for the HEK-293T cells and 15 min for the microglial culture cells). The HEK-293T cells were treated for 1 h with PBS containing 1% bovine serum albumin (BSA), labeled with mouse monoclonal anti-RLuc antibody (1/100; mAB4400, EMD Millipore, Darmstadt, Germany) and subsequently treated with Cy3 anti-mouse (1/200; 715-166-150, Jackson ImmunoResearch (red)) immunoglobulin G (IgG) (1 h each). Microglial cells were treated for 1 h with PBS containing 1% BSA and labelled with a mouse Cells 2020, 9, 1075 4 of 16 anti-iNOS (1/100; NOS2 (C-11): sc-7271; SCB) antibody, a mouse monoclonal anti-arginase I (1/100; 610708; BD Biosciences, San Jose, CA, USA) antibody or a rabbit polyclonal anti-Ki-67 (1/100; ab15580; Abcam, Cambridge, UK) antibody, and subsequently treated with a Cy3-conjugated anti-rabbit (1/200; 711-165-152; Jackson ImmunoResearch (red), West Grove, PA, USA) or anti-mouse (1/200; 715-166-150; Jackson ImmunoResearch (red), West Grove, PA, USA) IgG secondary antibodies (1 h each). The nuclei were stained with Hoechst (1/100; Sigma Aldrich, St. Louis, MO, USA). The samples were washed several times and mounted with 30% Mowiol (Calbiochem, San Diego, CA, USA). The images were obtained in a Leica SP2 confocal microscope (Leica Microsystems). The instrument was equipped with an apochromatic 63X oil-immersion objective (N.A. 1.4), and 488 nm and 561 nm laser lines. 2.6. Bioluminescence Resonance Energy Transfer (BRET) Assays For the BRET assays, the HEK-293T cells were transiently co-transfected with a constant amount of cDNAs encoding for GluN1-RLuc and GluN2 and with increasing amounts of cDNAs corresponding to A2A R-YFP or caldendrin-YFP. Forty-eight hours post-transfection, the cell suspension was adjusted to 20 µg of protein using a Bradford assay kit (Bio-Rad, Munich, Germany) and BSA for standardization. To quantify the protein-YFP expression, fluorescence was read in a Mithras LB 940 (Berthold Technologies, Bad Wildbad, Germany) equipped with a high-energy xenon flash lamp, using a 10 nm bandwidth excitation filter at 485 nm reading. For BRET and BRET with bimolecular complementation (BiFLC) measurements, the readings were collected 1 min after the addition of 5 µM coelenterazine H (Molecular Probes, Eugene, OR, USA) using a Mithras LB 940, which allowed the integration of the signals detected in the short-wavelength filter at 485 nm and the long-wavelength filter at 530 nm. To quantify the protein-RLuc expression, luminescence readings were performed 10 min after the 5 µM coelenterazine H addition using a Mithras LB 940. The net BRET was defined as ((long-wavelength emission)/(short-wavelength emission)) − Cf , where Cf corresponds to ((long-wavelength emission)/(short-wavelength emission)) for the donor construct expressed alone in the same experiment. The GraphPad Prism software (San Diego, CA, USA) was used to fit the data. BRET is expressed as milli BRET units, mBU (net BRET × 1000). 2.7. Cyclic Adenylic Acid (cAMP) Determination Two hours before initiating the experiment, culture medium for HEK-293T-transfected or primary neuronal or glial cells was exchanged by serum-starved DMEM medium. Then, the cells were detached, resuspended in a growing medium containing 50 µM zardaverine (Tocris Bioscience, Bristol, UK) and plated in 384-well microplates (2500 cells/well), pretreated (15 min) with the corresponding antagonists (SCH-58261 for A2A R and MK-801 for NMDAR) or vehicle and stimulated with agonists (CGS-21680 for A2A R and NMDA for NMDAR) (15 min) before adding 0.5 µM forskolin or vehicle (15 min). The readings were performed after a 1 h incubation (room temperature). Homogeneous time-resolved fluorescence energy transfer (HTRF) measures were performed using the Lance Ultra cAMP kit (PerkinElmer, Waltham, MA, USA). Fluorescence at 665 nm was analyzed on a PHERAstar Flagship microplate reader equipped with an HTRF optical module (BMG Lab technologies, Offenburg, Germany). 2.8. Extracellular Signal-Regulated Kinase (ERK) Phosphorylation Determination To determine the ERK1/2 phosphorylation, 40,000 HEK-293T cells/well, 50,000 microglia cells/well or 50,000 neurons/well were plated in transparent 96-well microplates and kept in the incubator for 48 h (HEK-293T cells) or 12 days (microglia and neuronal culture cells). Two to four h before initiating the experiment, the medium was substituted for a serum-starved DMEM medium. Then, the cells were pre-treated at room temperature for 10 min with the specific antagonists (SCH-58261 for A2A R and MK-801 for NMDAR) or vehicle in a serum-starved DMEM medium and stimulated for an additional 10 min with the specific agonists (CGS-21680 for A2A R and N-methyl d-aspartate (NMDA) for NMDAR) or vehicle. The cells were then washed twice with cold PBS before the addition of a lysis buffer (20 min treatment in constant agitation). Subsequently, 10 µL of each supernatant was Cells 2020, 9, 1075 5 of 16 placed in white ProxiPlate 384-well microplates and the ERK 1/2 phosphorylation was determined using the AlphaScreen® SureFire® kit (Perkin Elmer, Waltham, MA, USA) following the instructions of the supplier and using an EnSpire® Multimode Plate Reader (PerkinElmer, Waltham, MA, USA). 2.9. Assessment of Dynamic Mass Redistribution (DMR) The cell mass redistribution induced upon receptor activation was detected by illuminating the underside of a biosensor with a polychromatic light and measuring the changes in the wavelength of the reflected monochromatic light that was a sensitive function of the index of refraction. The magnitude of this wavelength shift (in picometers) was directly proportional to the amount of DMR. HEK-293T cells and neuronal and microglial primary cultures were seeded in 384-well sensor microplates to obtain 70–80% confluent monolayers constituted of approximately 10,000 cells per well. Prior to the assay, the cells were washed twice and incubated for 2 h with assay buffer (Hank’s balanced salt solution (HBSS) with 20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer, pH 7.15) containing 0.1% DMSO (24 ◦ C, 30 µL/well). Hereafter, the sensor plate was scanned and a baseline optical signature was recorded for 10 min before adding the 10 µL of the specific antagonists (SCH-58261 for A2A R and MK-801 for NMDAR), that were recoded for 30 min followed by the addition of 10 µL of the specific agonists (CGS-21680 for A2A R and NMDA for NMDAR); all the test compounds were dissolved in the assay buffer. The cell signaling signature was determined using an EnSpire® Multimode Plate Reader (PerkinElmer, Waltham, MA, USA) by a label-free technology. The results were analyzed using the EnSpire Workstation Software v 4.10. 2.10. Determination of Cytoplasmic Calcium Ion Level Increase HEK-293T cells were transfected with the cDNAs for human A2A R, for GCaMP6 calcium sensor, and/or for both the GluN1 and GluN2 subunits of the NMDAR [24]. Forty-eight hours after transfection, 150,000 HEK-293T cells/well were plated in 96-well black, clear bottom microtiter plates and were incubated with Mg2+ -free Locke’s buffer (154 mM NaCl, 5.6 mM KCl, 3.6 mM NaHCO3 , 2.3 mM CaCl2 , 5.6 mM glucose and 5 mM HEPES, pH 7.4) supplemented with 10 µM glycine. The cells were treated with the specific antagonists (SCH-58261 for A2A R and MK-801 for NMDAR) for 10 min, followed by the addition of the receptor agonists, CGS-21680 for A2A R and NMDA for NMDAR, just a few seconds before the readings. The fluorescence emission intensity of the GCaMP6 was recorded at 515 nm upon excitation at 488 nm on the EnSpire® Multimode Plate Reader (PerkinElmer, Waltham, MA, USA) for 225 s every 5 s. 2.11. In Situ Proximity Ligation Assay (PLA) PLA was performed with reagents from Sigma Aldrich and following the protocols of the supplier. In brief, microglial and neuronal primary cells grown on glass coverslips were fixed in 4% paraformaldehyde for 15 min, washed with PBS containing 20 mM glycine to quench the aldehyde groups, and permeabilized with the same buffer containing 0.05% Triton X-100 (15 min). After 1 h incubation at 37 ◦ C with blocking solution, the cells were treated with specific antibodies against A2A or NMDA receptors: mouse monoclonal anti-A2A R (1/100, Millipore, Darmstadt, Germany) or rabbit polyclonal anti-GluN1 antibody (1/200, Millipore). The cells were processed using the PLA probes detecting mouse and rabbit antibodies (Duolink II PLA probe anti-rabbit plus and Duolink PLA probe anti-mouse minus; Sigma Aldrich) and were prepared with Hoechst (1/200; Sigma Aldrich, St. Louis, MO, USA) using a mounting medium. The images were obtained in a Leica SP2 confocal microscope (Leica Microsystems, Mannheim, Germany). For each field of view a stack of two channels (one per staining) and 3 to 4 Z stacks with a step size of 1 µm were acquired. The quantification of the cells containing one or more red spots versus the total cells (blue nucleus), and in cells containing spots, the ratio r (number of red spots/cell), were determined by the Duolink Image tool software (Sigma Aldrich, St. Louis, MO, USA). Cells 2020, 9, 1075 6 of 16 2.12. Statistical Analysis The data in the graphs are the mean ± SEM (n = 5, at least). The GraphPad Prism software version 7 (San Diego, CA, USA) was used for the data fitting and statistical analysis. The Kolmogorov-Smirnov test with the correction of Lilliefors was used to evaluate the normal distribution and the Levene test was used to evaluate the homogeneity of variance. A one-way ANOVA followed by the post-hoc Bonferroni’s test were used when comparing multiple values. When a pair of values were compared, the Student’s t test was used. Significant differences were considered when the p value was < 0.05. 3. Results 3.1. NMDA Receptors May Directly Interact with Adenosine A2A Receptors N-methyl-d-aspartate receptor (NMDAR) activation regulates synaptic plasticity and neuronal survival. However, an increase in the NMDAR activity is associated to excitotoxicity and cell death and this is the main reason it is targeted by one of the existing anti-AD drugs. On the other hand, adenosine is a neuromodulator and one of the main mediators is the A2A receptor (A2A R), which is expressed in both CNS neurons and glial cells. To address a potential interaction between the two receptors, we first performed immunocytochemical assays in a heterologous expression system. HEK-293T cells were transfected with cDNAs for A2A R-RLuc or for GluN1 fused to RLuc and the GluN2B subunit of NMDAR (transfection of the two subunits was necessary for reconstituting a functional NMDA receptor). The membrane and cytoplasmic expression of both: A2A and NMDA receptors was observed (Figure 1A). Then, colocalization was addressed in cells transfected with the cDNAs for A2A R-YFP and for GluN1-RLuc and GluN2B subunits. The degree of colocalization was significant (Figure 1B) but while it demonstrated expression in the same compartment(s), it did not allow the discovery of direct protein–protein interaction. Accordingly, the A2A -NMDA receptor interaction was assayed by bioluminescence resonance energy transfer (BRET) using HEK-293T cells expressing a reconstituted NMDAR fused to RLuc and increasing amounts of A2A R-YFP. The saturable BRET curve indicated a specific interaction between the A2A and the NMDA receptors (BRETmax 141 ± 14 and BRET50 155 ± 22). As a negative control, the A2A R was substituted by the calcium sensor protein, caldendrin, and the result was a linear relationship that was indicative of unspecificity (Figure 1C). In summary A2A R may interact with NMDAR but not with caldendrin in living transiently transfected HEK-293T cells. 3.2. Functional Properties of A2A -NMDA Receptor Heteromer Complex Adenosine A2A R couple to Gs proteins, activating adenylate cyclase and increasing cAMP intracellular levels. In preliminary assays, we demonstrated that cytosolic cAMP levels increased when HEK-293T cells expressing A2A R were treated with the selective ligand, CGS-21680 (Supplementary Figure S1A). Furthermore, this effect was specific, because it was blocked by pre-treatment with the selective antagonist SCH-58261. Neither NMDA, nor a NMDAR antagonist, MK-801, induced any effect (Supplementary Figure S1A). Similar results were obtained in extracellular signal-regulated kinase (ERK) phosphorylation and label-free dynamic mass redistribution (DMR) assays (Supplementary Figure S1B,C). As expected, cytosolic calcium did not increase upon A2A R stimulation (Supplementary Figure S1A–D) but it did increase when HEK-293T cells were transfected with reconstituted NMDAR and treated with NMDA (Supplementary Figure S1A–H). Whereas the stimulation with NMDA did not alter the cytosolic cAMP levels (Supplementary Figure S1A–E), it induced MAPK phosphorylation and modified the DMR outputs. Although the effects were specific and blocked by a selective NMDAR antagonist, MK-801 (Supplementary Figure S1F,G), they were not altered by either a selective A2A R agonist, CGS-21680, or a selective A2A R antagonist, SCH-58261. Once receptor-mediated signaling was characterized in the cells expressing A2A R or NMDAR receptors, cross-modulation was assayed in the co-transfected HEK-293T cells in which the cAMP levels, MAPK phosphorylation, label-free DMR and the calcium release signals were analyzed. Remarkably, the cAMP data revealed that the signal obtained after the A2A R stimulation was blocked by both A2A R compared, the Student’s t test was used. Significant differences were considered when the p value was < 0.05. 3. Results Cells 2020, 9, 1075 7 of 16 3.1. NMDA Receptors May Directly Interact with Adenosine A2A Receptors N-methyl-D-aspartate receptor (NMDAR) activation regulates synaptic plasticity and neuronal and NMDAR antagonists (Figure 2A). Such a property is known as cross-antagonism and would be survival. However, an increase in the NMDAR activity is associated to excitotoxicity and cell death useful to identify A2A -NMDA receptor complexes in natural sources. In fact, cross-antagonism is and this is the main reason it is targeted by one of the existing anti-AD drugs. On the other hand, considered a heteromer print [25,26]. Coactivation of the A2A and NMDA receptors led to a decrease adenosine is a neuromodulator and one of the main mediators is the A2A receptor (A2AR), which is in the CGS-21680-induced effect, indicating that the NMDAR activation impacted on adenosine A2A R expressed in both CNS neurons and glial cells. To address a potential interaction between the two signaling. In the MAPK phosphorylation assays, the activation of both A2A and NMDA receptors receptors, we first performed immunocytochemical assays in a heterologous expression system. were able to activate the MAPK pathway. In addition, whereas the A2A R-mediated signaling was HEK-293T cells were transfected with cDNAs for A2AR‐RLuc or for GluN1 fused to RLuc and the smaller when the two receptors were co-expressed, the NMDAR-mediated signaling was stronger. GluN2B subunit of NMDAR (transfection of the two subunits was necessary for reconstituting a This result indicated a possible potentiation of A2A R over the NMDAR signaling when forming the functional NMDA receptor). The membrane and cytoplasmic expression of both: A2A and NMDA A2A -NMDA receptor complexes. However, the coactivation of both receptors did not result in an receptors was observed (Figure 1A). Then, colocalization was addressed in cells transfected with the additive effect. It is noteworthy that any signal was counteracted by either the NMDAR or A2A R cDNAs for A2AR‐YFP and for GluN1‐RLuc and GluN2B subunits. The degree of colocalization was selective antagonists, i.e., cross-antagonism was also detected. (Figure 2B). DMR data were similar significant (Figure 1B) but while it demonstrated expression in the same compartment(s), it did not to those found in cAMP and MAPK assays. The A2A R- and NMDAR-agonist-induced signals were allow the discovery of direct protein–protein interaction. Accordingly, the A2A‐NMDA receptor blocked by both the NMDAR and A2A R selective antagonists, while the coactivation of both receptors interaction was assayed by bioluminescence resonance energy transfer (BRET) using HEK-293T cells produced a stronger signal than that of the NMDA but smaller than that of the A2A R agonist. Finally, expressing a reconstituted NMDAR fused to RLuc and increasing amounts of A2AR‐YFP. The in terms of calcium mobilization, the coactivation of both receptors produced a similar effect to that saturable BRET curve indicated a specific interaction between the A2A and the NMDA receptors induced by the NMDAR stimulation but interestingly, the NMDA effect was blocked by both NMDA (BRETmax 141 ± 14 and BRET50 155 ± 22). As a negative control, the A2AR was substituted by the calcium and A2A receptor antagonists (A2A receptor agonists did not yet induce cytosolic calcium increases) sensor protein, caldendrin, and the result was a linear relationship that was indicative of unspecificity (Figure 2D). The results may be explained by i) the A2A R expression increasing the NMDAR function (Figure 1C). In summary A2AR may interact with NMDAR but not with caldendrin in living and ii) the NMDAR activation blocking adenosine A2A R signaling and iii) a cross-antagonism due to transiently transfected HEK-293T cells. inter-protomer allosteric communication within the A2A R and NMDAR heteromer. Figure 1. The N-methyl d-aspartate (NMDA) and the adenosine A2A receptors interact to form heteromeric complexes. In (A,B), HEK-293T cells were transfected with 0.5 µg cDNA corresponding to the A2A receptor (A2A R)-Renilla luciferase (RLuc) (A left) or 0.5 µg cDNA corresponding to the GluN1-RLuc in the presence of 0.3 µg cDNA corresponding to GluN2 (A right) or co-transfected with 0.41.µgThe Figure cDNA for A2AD N-methyl R-yellow -aspartatefluorescent (NMDA)protein (YFP) and the and 0.4 A adenosine µg2AcDNA corresponding receptors interact to toform the GluN1-RLuc in the presence heteromeric complexes. of 0.25 In (A,B), µg cDNAcells HEK-293T corresponding to thewith were transfected GluN2 0.5 (B). Confocal μg cDNA microscopy corresponding images are shown. The receptors fused to RLuc were identified by immunocytochemistry (red) and the proteins fused to YFP were identified by its own fluorescence (green). Colocalization is shown in yellow in the merge image. Scale bar: 20 µm. In (C), the bioluminescence resonance energy transfer (BRET) saturation experiments were performed in HEK-293T cells transfected with 0.3 µg of cDNA corresponding to the GluN1-RLuc, 0.2 µg of cDNA corresponding to the GluN2 and increasing amounts of cDNA corresponding to A2A R-YFP (0.1 µg to 0.5 µg) or caldendrin-YFP (0.1 µg to 0.7 µg) as a negative control. The relative amount of BRET is given as a function of 1000× the ratio between the fluorescence of the acceptor (YFP) and the luciferase activity of the donor (RLuc). BRET is expressed as milli BRET units (mBU) and is given as the mean ± SEM of 6 different experiments grouped as a function of the amount of BRET acceptor. Cells 2020, 9, x 8 of 16 antagonism due to inter-protomer allosteric communication within the A2AR and NMDAR heteromer. Cells 2020, 9, 1075 8 of 16 Figure 2. Functional characterization of the A2A -NMDA receptor heteromer in HEK-293T cells. In (A) to (D), HEK-293T cells expressing A2A R (0.5 µg of cDNA), GluN1 (0.5 µg of cDNA) and GluN2 (0.5 µg of 2. Figure cDNA) (A–C)characterization Functional or expressing Aof 2A R the(0.5 A2A of cDNA), ‐NMDA µg GluN1 receptor (0.5 µg ofincDNA), heteromer HEK-293TGluN2 (0.5 cells. (A)of In µg tocDNA) and 6GCaMP (D), HEK-293T cellscalcium sensor expressing A2A(0.75 R (0.5µgμgof of cDNA) (D)GluN1 cDNA), were pre-incubated or notand (0.5 μg of cDNA) withGluN2 1 µM of the (0.5 μgA2A of R antagonist cDNA) (A to SCH-58261 (SCH)Aor C) or expressing 2ARwith (0.51μg µM ofof the NMDA cDNA), GluN1antagonist, (0.5 μg of MK-801 (MK), followed cDNA), GluN2 (0.5 μg ofby treatment cDNA) with 100 nM and 6GCaMP of the calcium A2A R(0.75 sensor agonist μg ofCGS-21680 cDNA) (D) (CGS), were15 µM of NMDA pre-incubated or or notboth, withand1 μMthe ofcAMP the Alevels (A), extracellular 2AR antagonist SCH-58261signal-regulated (SCH) or with (ERK) 1 µ1/2 M phosphorylation (B), representative of the NMDA antagonist, MK-801 traces (MK),of dynamicbymass followed redistribution treatment with 100(DMR) nM of (C) the and A2ARthe representative agonist CGS-21680 traces of intracellular (CGS), 15 µ M of NMDACa2+ responses or both, over and thetime cAMP (D)levels were(A), determined. Values extracellular signal-regulated ± SEM are the mean(ERK) ofphosphorylation 1/2 10 to 12 different(B), experiments. representativeERK 1/2 phosphorylation levels and cAMP increases are expressed as percentage over traces of dynamic mass redistribution (DMR) (C) and the representative traces of intracellular Ca2+ basal. A one-way ANOVA followed responses over time by(D)a Bonferroni multiple comparison were determined. Values arepost-hoc the mean test±showed SEM of a significant 10 to 12 effect over different 100% (* p < 0.05, experiments. ERK **1/2p < 0.01, *** p < 0.001). phosphorylation levels and cAMP increases are expressed as percentage over basal. A one-way ANOVA followed by a Bonferroni multiple comparison post-hoc test showed a 3.3. A2A -NMDA Receptor Heteromer Complex Expression in Resting and Activated Microglia significant effect over 100% (* p < 0.05, ** p < 0.01, *** p < 0.001). Due to the renowned interest in glial cells as targets to combat neurodegenerative diseases and 3.3. dueA2A to‐NMDA Receptor of the expression Heteromer Complexin both receptors Expression in Resting these cells, we next and Activatedthe analyzed Microglia A2A -NMDA receptor complex expression in the microglial primary cultures from wild type mice. Interestingly, the proximity Due to the renowned interest in glial cells as targets to combat neurodegenerative diseases and ligation assay (PLA) showed 23% of cells with clusters of receptor complexes depicted as red dots due to the expression of both receptors in these cells, we next analyzed the A2A‐NMDA receptor (two redexpression complex dots per cell) (Figure in the 3A,B). Moreover, microglial when the primary cultures microglia from was mice. wild type activated using LPS the Interestingly, and IFN-γ, theligation proximity A2A R-NMDAR assay (PLA)heteromer showedexpression 23% of cells was markedly with clustersenhanced, of receptor with 92% of cells complexes showing depicted as red dots and an eight-fold increase in dots/cell (16 dots in activated cells versus two in red dots (two red dots per cell) (Figure 3A,B). Moreover, when the microglia was activated using LPS resting cells) (Figure and IFN-γ,3A,B). the Similar A2AR-NMDARassays were performed heteromer in hippocampal expression was markedly neuronal primary enhanced, withcultures; 92% of28%cellsof neurons red showing showed dots red anddots. Overall, these an eight-fold results increase in showed dots/cellthat (16the A2Ain-NMDA dots receptor activated complexes cells versus twomay in play a relevant role in activated microglia cells. resting cells) (Figure 3A,B). Similar assays were performed in hippocampal neuronal primary cultures; 28% of neurons showed red dots. Overall, these results showed that the A2A‐NMDA receptor complexes may play a relevant role in activated microglia cells. Cells 2020, 9, x 9 of 16 Cells 2020, 9, 1075 9 of 16 Figure Figure Expression and 3.3.Expression and functionality functionality ofofthe A2A the A2Aand andNMDA NMDA receptors heteromers receptors heteromers in thein microglial the microglial and neuronal primary cultures. Panels (A,B). The proximity ligation assay and neuronal primary cultures. Panels (A)–(B). The proximity ligation assay (PLA) was performed(PLA) was performed in in mice microglial primary cultures treated or not with 1 µM lipopolysaccharide (LPS) and 200 U/mL mice microglial primary cultures treated or not with 1 µ M lipopolysaccharide (LPS) and 200 U/mL interferon-γ (IFN-γ) to activate the microglia and in the mice hippocampal primary cultures of neurons interferon-γ (IFN-γ) to activate the microglia and in the mice hippocampal primary cultures of using the primary antibodies specific for the A2A and NMDA receptors. In all cases, the cell nuclei were neurons using the primary antibodies specific for the A2A and NMDA receptors. In all cases, the cell stained with Hoechst (blue). The confocal microscopy images are shown (superimposed sections) in nuclei which were stainedappear heteromers with Hoechst (blue). as red clusters (inThe confocal neurons microscopy or microglia). Scaleimages bars = 30areµm.shown (superimposed The bar graph sections) in which heteromers appear as red clusters (in neurons or microglia). (B) shows the number of the red dots/cell (r) and the numbers above the bars indicate the percentage Scale bars = 30 μm. Theof bar cellsgraph (B) shows presenting the number red dots. Values are ofthe themean red dots/cell ± SEM (n(r) = and 8). Athe numbers one-way ANOVAabovefollowed the barsbyindicate theBonferroni’s percentagemultiple of cellscomparison presentingpost-hoc red dots. testValues are the were used mean ± SEM for statistical (n =(*** analysis < 0.001, 8).p A one-way ANOVA versus control -resting- followed microglia). by Bonferroni’s Panels (C–H) multiple display post-hoc comparison the microglial test primary were used cultures treated (D,G) for statistical or not(*** p < analysis (C,F)versus 0.001, with 1 µM LPS and control 200 U/mL -resting- IFN-γ and microglia). the mice Panels (C)hippocampal to (H) display primary cultures of the the microglial neurons primary cultures (E,H) that were pre-incubated or not with 1 µM of the A2A R antagonist SCH-58261 (SCH) or with 1 µM treated (D,G) or not (C,F) with 1 µ M LPS and 200 U/mL IFN-γ and the mice hippocampal primary of the NMDAR antagonist MK-801 (MK) followed by treatment with 100 nM of the A R agonist cultures of the neurons (E,H) that were pre-incubated or not with 1 μM of the A2AR2Aantagonist SCH- CGS-21680 (CGS), 15 µM NMDA or both, and the cAMP levels (C–E) and ERK 1/2 phosphorylation 58261 (SCH) or with 1 µ M of the NMDAR antagonist MK-801 (MK) followed by treatment with 100 signal (F–H) were determined. The values are the mean ± SEM of 10 to 12 different experiments. nMTheof ERK the A 2AR agonist CGS-21680 (CGS), 15 µ M NMDA or both, and the cAMP levels (C to E) and 1/2 phosphorylation levels and cAMP increases were expressed as a percentage over basal. ERK 1/2 phosphorylation A one-way ANOVA followed signal by a(FBonferroni to H) were determined. multiple comparisonThe values post-hocaretestthe meana±significant showed SEM of 10 to 12 different effect over 100% (* p < The experiments. ** p <1/2 0.05, ERK *** p < 0.001). phosphorylation 0.01, levels and cAMP increases were expressed as a percentage over basal. A one-way ANOVA followed by a Bonferroni multiple comparison post-hoc The functional cross-talk was assayed by the analysis of the cAMP intracellular levels in the test showed a significant effect over 100% (* p < 0.05, ** p < 0.01, *** p < 0.001). microglia cells and in neurons. In both, the resting and the activated microglia, only the activation of the A2A R results in cAMP responses, that were reverted by the A2A R antagonist and by the NMDAR The functional cross-talk was assayed by the analysis of the cAMP intracellular levels in the antagonist. We observed similar results in neurons but without cross-antagonism, probably reflecting microglia cells and in neurons. In both, the resting and the activated microglia, only the activation of the A2AR results in cAMP responses, that were reverted by the A2AR antagonist and by the NMDAR antagonist. We observed similar results in neurons but without cross-antagonism, probably reflecting Cells 2020, 9, 1075 10 of 16 that not all NMDA receptors were interacting with the A2A R (Figure 3C–E). The activation of either the NMDAR or the A2A R resulted in the MAPK pathway activation. Interestingly, coactivation led to a small non additive effect only in microglia cultures, once more supporting the idea that not all NMDA receptors in neurons were interacting with the A2A R. However, cross-antagonism was detected in both the microglial and neuronal cells (Figure 3F–H). 3.4. A2A R-NMDAR Heteromer Expression is Elevated in Primary Microglia from APPSw/Ind Mice Alzheimer’s disease (AD) is one of the most prevalent neurodegenerative diseases worldwide; the causes are unknown and current drugs have little efficacy. NMDAR function is enhanced in the initial stages of AD, thus leading to altered intracellular calcium handling and the gradual loss of synaptic function [27,28]. In addition, A2A R expression markedly increases in neurodegenerative diseases with an inflammatory component. Accordingly, we moved to assess the expression of A2A R-NMDAR heteromers in a transgenic AD mouse model. We first analyzed the expression of A2A -NMDA receptor complexes in the primary cultures of microglia and neurons from the hippocampus of APPSw,Ind mice. The hippocampal microglia and neuronal primary cultures were obtained from two-day-old pups or 19-day-old fetuses, respectively, and were independently cultured. PLA data analysis was blindly done before knowing the genotyping results. Remarkably, whereas 29% of the cultured microglia from the controls animals showed red dots, the percentage of cells from transgenic animals was 75% (three dots/cell in control versus six dots/cell in transgenic) (Figure 4A,B). The same assay type in neurons led to a low percentage of cells expressing receptor clusters (11% in control and 16% in transgenic) (Figure 4A,B). These results suggested that microglia cells were resting in control animals and activated in AD-mice. After determining the occurrence of A2A -NMDA receptor complexes in microglia primary cultures, we moved to characterize the activated microglia phenotype in the APPSw,Ind mice model. When microglia cells are activated, they can evolve showing different characteristics, with two opposite phenotypes: M1 microglia inducing a pro-inflammatory state and cytotoxic effects and M2 microglia inducing an anti-inflammatory state and neuroprotective effects. Then, the microglia primary cultures of the control and APPSw,Ind mice models were prepared and analyzed by immunocytochemistry. Through the analysis, an important increase in inducible nitric oxide synthase (iNOS) a marker of the M1 phenotype, was observed in microglia from APPSw,Ind mice compared to controls (Figure 4C,D). Moreover, fluorescence signal quantitation indicated an important increase in arginase-1 (Arg-1), a marker of the M2 phenotype, in APPSw,Ind –derived microglia (Figure 4C,D). These results indicated an important increase in both M1 and M2 microglia in the AD-mice model. Interestingly, when the same assay was repeated by pretreating the APPSw,Ind -derived cultures for one week with the A2A R antagonist SCH-58261, a significant decrease in iNOS and a small but significant increase in Arg-1 fluorescence were observed, indicating that A2A R antagonists skew activated microglia towards the neuroprotective M2 phenotype. Then, no alteration in microglia proliferation marker was observed upon analysis of the Ki-67 fluorescence signal in the APPSw,Ind mice models treated or not with SCH-58261 and control (WT) animals (Figure 4C,D). and were independently cultured. PLA data analysis was blindly done before knowing the genotyping results. Remarkably, whereas 29% of the cultured microglia from the controls animals showed red dots, the percentage of cells from transgenic animals was 75% (three dots/cell in control versus six dots/cell in transgenic) (Figure 4A,B). The same assay type in neurons led to a low percentage Cells 2020, 9, 1075 of cells expressing receptor clusters (11% in control and 16% in transgenic) (Figure 4A,B). 11 of 16 These results suggested that microglia cells were resting in control animals and activated in AD-mice. Figure Figure 4. Microglial 4. Microglial markers markers in primary in primary cultures cultures fromfrom APP APP mice. In Sw,Indmice. Sw,Ind In (A,B), (A,B), proximity proximity ligation ligation assays (PLAs) were performed in microglial and in the hippocampus neuronal primary cultures of assays (PLAs) were performed in microglial and in the hippocampus neuronal primary cultures of control control (WT)(WT) and and APPAPP mice mice Sw,Ind Sw,Ind using using primary primary antibodiesspecific antibodies specificfor forAA2A and NMDA 2A and NMDA receptors. receptors. In In all cases, all cases, the cell the cell nuclei nuclei werewere stained stained with with Hoechst Hoechst (blue).The (blue). Theconfocal confocalmicroscopy microscopy images images are are shown shown (superimposed sections) in which the heteromers appear as red clusters (in neurons or microglia). Scale bars = 30 µm (neurons and microglia). The bar graph (B) shows the number of red dots/cell (r) and the numbers above the bars indicate the percentage of cells presenting red dots. Values are the mean ± SEM (n = 6). A one-way ANOVA followed by Bonferroni’s multiple comparison post-hoc test were used for statistical analysis (*** p < 0.001, versus WT -control-). In (C,D), immunocytochemical assays were performed in the primary cultures of microglia APPSw,Ind mice or control (WT) animals pretreated or not for one week with the A2A R antagonist SCH-58261. The staining was performed using the antibodies that detected either arginase-1 marker, the antiproliferation cell protein Ki-67 or nitric oxide synthase and a Cy3-conjugated secondary antibody (red). The fluorescence was quantified in all the panels using the Fiji program and normalization by cell number. Representative images in all conditions are shown in (C). A one-way ANOVA followed by Bonferroni’s multiple comparison post-hoc test were used for statistical analysis (* p < 0.05, ** p < 0.01; versus WT). 3.5. A2A R Activation Negatively Modulates NMDA Receptor Signaling in Microglia but it Increases NMDAR Signaling in Neurons from APPSw,Ind Mice To analyze the adenosine regulatory effects over NMDAR functionality in the AD model, primary cultures of microglia and neurons were prepared from the hippocampus of APPSw,Ind mice and control animals. Intracellular cAMP levels, ERK1/2 phosphorylation and DMR were determined upon receptor activation and coactivation. On the one hand, A2A R activation led to an increase in cAMP levels in the neurons and microglia from control and transgenic animals. As expected, NMDA did not modify cAMP levels but counteracted the action of A2A R agonists. In addition, the selective NMDAR antagonist reverted the effect of CGS-21680 (cross-antagonism) (Figure 5A–D). The MAPK pathway activation was similar Cells 2020, 9, 1075 12 of 16 in microglia from control and transgenic animals, with a similar effect in the individual activation or the coactivation of receptors. In these cells, bidirectional cross-antagonism was found (Figure 5E,F). In neurons coactivation led to a more robust effect than individual treatments. We found a partial cross-antagonism that fits with the hypothesis that, in neurons, not all NMDAR are directly interacting with the A2A R (Figure 5G,H). DMR recordings in glial cells were only obtained upon A2A R activation with NMDA being ineffective and with a lack or a small cross-antagonism. A relevant finding was underscored using neurons, because those from the control animals were much less responsive to CGS-21680 than those from the transgenic animals. Moreover, NMDA also induced a robust response in the cells from transgenic animals; however, coactivation did not lead to synergism or additive effects (Figure 5I–L). Cells 2020, 9, x 12 of 16 Figure Figure 5. 5.Functionality FunctionalityofofthetheAA -NMDA 2A2A ‐NMDA receptor receptor heteromer heteromer in in the the APP micemodel Sw,Indmice APPSw,Ind modelofof Alzheimer’s disease(AD). In (A–L), the microglial and neuronal primary cultures Alzheimer’s disease(AD). In (A) to (L), the microglial and neuronal primary cultures from the from the hippocampus of the WT and APP hippocampus of the WTmice Sw,Ind (E,H) and APP were pre-incubated with vehicle, 1 µM of the A2A R Sw,Ind mice (E,H) were pre-incubated with vehicle, 1 μM ofantagonist the A2AR SCH-58261 antagonist SCH-58261 (SCH) or 1 µ M of the NMDAR antagonist, MK-801 (MK), followed with (SCH) or 1 µM of the NMDAR antagonist, MK-801 (MK), followed by treatment by 100treatment nM of thewith A2A100 R agonist nM of theCGS-21680 (CGS), A2AR agonist 15 µM NMDA CGS-21680 (CGS), 15 orµboth, M NMDA and the cAMP or both, andlevels (A–D), the cAMP ERK 1/2 phosphorylation levels (A to D), ERK 1/2signal (E–H) and the phosphorylation representative signal (E to H) andtraces of dynamic mass the representative tracesredistribution of dynamic (DMR) massover time (I–L)(DMR) redistribution were determined. over time (I to Values L) were the mean ± Values aredetermined. SEM ofare10the to 12 different mean ± SEMexperiments. of 10 to 12 ERK 1/2 phosphorylation different experiments. levels ERK and 1/2 cAMP increases are phosphorylation expressed levels as percentage and cAMP increasesover arebasal. A one-way expressed as ANOVA followed percentage overby a Bonferroni basal. A one-way multiple ANOVA comparison followed bypost-hoc test showed a Bonferroni multipleacomparison significant effect over post-hoc basal p < 0.05,a ** test(*showed p < 0.01,effect significant *** p