Adenosine A2A receptors facilitate synaptic NMDA currents in CA1 pyramidal neurons

RESEARCH PAPER Adenosine A 2A receptors facilitate synaptic NMDA currents in CA1 pyramidal neurons Correspondence Ana Maria Sebastião, Instituto de Medicina Molecular, Faculdade de Medicina, Universidade de Lisboa, Avenida Professor Egas Moniz, Lisbon 1649-028, Portugal. E-mail: [email protected] Received 24 January 2018; Revised 18 July 2018; Accepted 21 August 2018 Francisco M Mouro 1,2, *, Diogo M Rombo 1,2, *, Raquel B Dias 1,2 , Joaquim A Ribeiro 1,2 and Ana M Sebastião 1,2 1 Instituto de Farmacologia e Neurociências, Faculdade de Medicina, Universidade de Lisboa, Lisbon, Portugal, and 2 Instituto de Medicina Molecular, Faculdade de Medicina, Universidade de Lisboa, Lisbon, Portugal *Co-first authors. BACKGROUND AND PURPOSE NMDA receptors play a key role in both synaptic plasticity and neurodegeneration. Adenosine is an endogenous neuromodulator and through membrane receptors of the A 2A subtype can influence both synaptic plasticity and neuronal death. The present work was designed to evaluate the influence of adenosine A 2A receptors upon NMDA receptor activity in CA1 hippocampal neurons. We discriminated between modulation of synaptic versus extrasynaptic receptors, since extrasynaptic NMDA receptors are mostly associated with neurodegeneration while synaptic NMDA receptors are linked to plasticity phenomena. EXPERIMENTAL APPROACH Whole-cell patch-clamp recordings were obtained to evaluate NMDA receptor actions on CA1 pyramidal neurons of young adult (5–10 weeks) male Wistar rat hippocampus. KEY RESULTS Activation of A 2A receptors with CGS 21680 (30 nM) consistently facilitated chemically-evoked NMDA receptor-currents (NMDAPSCs) and afferent-evoked NMDA-currents (NMDA-EPSCs), an action prevented by an A 2A receptor antagonist (SCH58261, 100 nM) and a PKA inhibitor, H-89 (1 μM). These actions did not reflect facilitation in glutamate release since there was no change in NMDA-EPSCs paired pulse ratio. A 2A receptor actions were lost in the presence of an open-channel NMDA receptor blocker, MK-801 (10 μM), but persisted in the presence of memantine, at a concentration (10 μM) known to preferentially block extrasynaptic NMDA receptors. CONCLUSION AND IMPLICATIONS These results show that A 2A receptors exert a positive postsynaptic modulatory effect over synaptic, but not extrasynaptic, NMDA receptors in CA1 neurons and, therefore, under non-pathological conditions may contribute to shift the dual role of NMDA receptors towards enhancement of synaptic plasticity. Abbreviations aCSF, artificial CSF; CaMKII, Ca 2+ /calmodulin-dependent protein kinase II; CGS 21680, 2-[4-(2-p-carboxyethyl) phenylamino]-50N-ethylcarbozamido adenosine; CNQX, 6-cyano-7-nitroquinoxaline-2,3-dione; DL-AP5, 2-amino-5phosphonovaleric acid; LTD, long-term depression; memantine, 3,5-dimethyl-1-adamantanamine hydrochloride; PPR, paired pulse ratio; SCH 58261, 7-(2-phenylethyl)-5-amino-2-(2-furyl)-pyrazolo-[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine; TTX, tetrodoxin citrate This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproductioninany medium, provided the original work is properly cited and is not used for commercial purposes. British Journal of Pharmacology British Journal of Pharmacology (2018) 175 4386–4397 4386 DOI:10.1111/bph.14497 © 2018 The Authors. British Journal of Pharmacology published by John Wiley & Sons Ltd on behalf of British Pharmacological Society. Introduction NMDA receptors are membrane-bound ionotropic glutamate receptors characterized by voltage-dependent activity, high permeability to calcium ions and slow activation/ deactivation kinetics (Doherty and Sladek, 2011). Calcium influx through NMDA receptors is crucial for synaptogenesis, experience-dependent synaptic remodelling and for longlasting changes in synaptic efficiency such as LTP and longterm depression (LTD). Indeed, NMDA receptors are key components in the induction of several forms of synaptic plasticity (Hunt and Castillo, 2012). Following depolarization of the postsynaptic membrane and release of magnesium blockade, a postsynaptic rise in calcium, mediated by the opening of NMDA receptors, activates important kinases such as calmodulin-dependent kinase II (CaMKII), PKA and PKC (Lau and Zukin, 2007), with an impact on synaptic reinforcement through fast signalling AMPA receptors at the postsynaptic membrane (Malenka and Nicoll, 1999; Morris et al., 2003; Collingridge et al., 2004). On the other hand, NMDA receptors are the main mediators of excitotoxicity, whenever there is an excessive and neurotoxic increase in intracellular calcium (Hardingham and Bading, 2010). The dual role of the NMDA receptor is related to receptor location relative to the synapse (Hardingham and Bading, 2010). Calcium influx through synaptic NMDA receptors is relatively well tolerated by neuronal cells and important for neuronal survival and synaptic plasticity, while an influx through extrasynaptic NMDA receptors is linked to neurotoxicity (Hardingham et al., 2002). Regulatory mechanisms that control NMDA receptor activity at specific membrane locations are, therefore, of particular importance and its knowledge relevant for strategies aiming to favour the balance towards neuronal survival and shape synaptic functioning. Adenosine is an ubiquitous neuromodulator influencing the action and activity of several neurotransmitter receptors (Sebastião and Ribeiro, 2015). Adenosine A 1 receptors are responsible for the adenosine-mediated inhibitory effects, including inhibition of neurotransmitter release and changes in postsynaptic membrane conductance (Dunwiddie and Masino, 2001), as well as inhibition of NMDA receptor-mediated currents at hippocampal neurons (De Mendonça et al.,1995).Theothersubtypeofhighaffinity adenosine receptors, the adenosine A 2A receptor, is coupled to Gs proteins, classically leading to stimulation of adenylate cyclase, a rise in intracellular cAMP levels and PKA activation. A 2A receptors are mostly expressed in the basal ganglia but are also expressed in lower levels at the hippocampus where they exert important neuromodulatory actions such as the control of other modulators (Sebastião and Ribeiro, 2015) and regulate synaptic plasticity (Dias et al., 2013a). Indeed, adenosine A 2A receptors at the CA1 area of the hippocampus enhance LTP (Costenla et al., 2011; Dias et al., 2012), facilitate the actions of the brain-derived neurotrophic factor (BDNF) at hippocampal synapses (Diógenes et al., 2004; Fernandes et al., 2008) with consequences for synaptic plasticity (Fontinha et al., 2008), and enhance AMPA-evoked currents recorded from CA1 pyramidal neurons (Dias et al., 2012) with an impact on ischaemiainduced plasticity phenomena (Dias et al., 2013b). Also, the degree of activation of A 2A receptors affects hippocampaldependent learning (Fontinha et al., 2009; Batalha et al., 2013), excitotoxicity and neuronal death (Jones et al., 1998; Cunha, 2016; Ribeiro et al., 2016). The activity of A 2A receptors regulates the function of NMDA receptors in the striatum (Wirkner et al., 2004) and plays a permissive role in the positive interaction between metabotropic glutamate receptors 5 (mGlu 5 receptors) and NMDA receptors in the hippocampus (Tebano et al., 2005; Sarantis et al., 2015; Kouvaros and Papatheodoropoulos, 2016; Krania et al., 2018), an action recently shown to be involved in synaptic plasticity deregulation under conditions of presynaptic A 2A receptor overexpression (Temido-Ferreira et al., 2018). A 2A receptors’activity is also required to gate a form of LTP mediated by postsynaptic NMDA receptors at mossy fibre hippocampal synapses (Rebola et al., 2008). However, despite the dual role of NMDA receptors, as well as of A 2A receptors upon LTP and excitotoxicity, no information so far exists on the direct influence of A 2A receptors on NMDA receptor-mediated currents. This issue is particularly important and needs to be investigated in the CA1 area of the hippocampus, given the major role of NMDA receptors in CA1 LTP and the vulnerability of this hippocampal area to excitotoxicity phenomena. We herein evaluated whether activation of adenosine A 2A receptors could affect NMDA receptor-mediated currents in CA1 pyramidal cells. Given the differences between synaptic and extrasynaptic NMDA receptor-mediated activity upon neuronal function (Hardingham and Bading, 2010), we focused on distinguishing if any action produced by A 2A receptors detected mostly resulted from modulation of synaptic or extrasynaptic NMDA receptor-mediated currents. Methods Hippocampal slice preparation and patch-clamp procedures Animal studies are reported in compliance with the ARRIVE guidelines (Kilkenny et al., 2010). All experiments were carried out on acute hippocampal slices obtained from 5to 10-week-old male Wistar rats (Harlan Iberica, Spain). All experiments followed the European Community Guidelines (Directive 2010/63/EU) and the Portuguese Law (DL 113/2013) for Animal Care for Research Purposes, and were approved by the “Instituto de Medicina Molecular” Internal Committee and the Portuguese Animal Ethics Committee - Direcção Geral de Veterinária. The animals were killed by decapitation under deep isoflurane anaesthesia. The brain was quickly removed, hemisected, and both hippocampi used to obtain transverse slices (300-μm-thick) cut on a Vibratome (VT 1000S; Leica, Nussloch, Germany) in ice-cold dissecting solution containing (in mM): sucrose 110; KCl 2.5; CaCl 2 0.5; MgCl 2 7; NaHCO 3 25; NaH 2 PO 4 1.25; glucose 7, bubbled with 95% O 2 and 5% CO 2 ,pH7.4.Sliceswerefirst incubated for 30 min at 35°C in artificial CSF (aCSF) containing (mM): NaCl 124; KCl 3; NaH 2 PO 4 1.25; NaHCO 3 26; MgSO 4 1; CaCl 2 2 and glucose 10, pH 7.4, gassed with 95% O 2 and 5% CO 2 ; A 2A receptors facilitate synaptic NMDA currents British Journal of Pharmacology (2018) 175 4386–4397 4387 after recovery for at least 1 h at room temperature, individual slices were set up for electrophysiological recordings. Throughout the recording period, the aCSF was supplemented with 5 μMglycineandnoMg 2+ was added. Individual slices were fixed with a grid in a recording chamber and continuously superfused by a gravitational superfusion system at 2–3mL·min 1 with aCSF (with glycine and no Mg 2+ ) at room temperature. Unless stated otherwise, drugs were added to the superfusion solution. Zero minutes in the figures correspond to the time when the first test solution was changed and indicate time at which the transfer was initiated. Around 2–3 min elapsed before the new solution reached the recording chamber; it took several minutes before the final concentration of the test drug was achieved in the recording chamber (dead volume of the recording chamber plus perfusion tubing: 10 mL). Patch pipettes (4–9MΩ) were pulled from borosilicate glass capillaries (1.5 mm outer diameter, 0.86 inner diameter, Harvard Apparatus), with a PC-10 Puller (Narishige Group) and filled with an intracellular solution containing (mM): K-gluconate 125; KCl 11; CaCl 2 0.1; MgCl 2 2; EGTA 1; HEPES 10; MgATP 2; NaGTP 0.3 and phosphocreatine 10, pH 7.3, adjusted with KOH (1 M), 280-290 mOsm. Electrode positioning and cell selection were performed under visual guidance using a Carl Zeiss Axioskop 2FS upright microscope (Jena, Germany) equipped with a differential interference contrast-infrared (DIC-IR) CCD video camera (VX44, Till Photonics, Gräfelfing, Germany). Recordings were performed in voltage-clamp mode (Vh = 60 mv) with an EPC-7 amplifier (List Biologic, Campbell, CA) or Axopatch 200B amplifier (Axon Instruments, CA, USA). The junction potential was not compensated for, and offset potentials were nulled before giga-seal formation. Small voltage steps (5 mV, 50 ms) were delivered throughout the experiment to monitor the access resistance; the holding current was also constantly monitored, and when any of these parameters varied by more than 20%, the experiment was rejected. The current signal was low-pass filtered using a 3 and 10 kHz three-pole Bessel filter, digitized at 10 kHz using a Digidata 1322A board and registered by the Clampex software version 10.2 (Molecular Devices, Sunnyvale, CA, USA). Data analysis was performed with Clampfit10 (included in pCLAMP 10). NMDA receptor-mediated postsynaptic currents (NMDAPSCs) were evoked through an NMDA (150 μM)-filled micropipette coupled to a pressure application system (PicoPump PV820; World Precision Instruments) positioned close to the cell soma of the recording cell and injecting single pulses of 10 ms, 6–10 ψ, applied every 2 min. The concentration of NMDA in the micropipette, the frequency of stimulation, injection duration and pressure were chosen within a range used by others (Khosravani et al., 2008; Papouin et al., 2012; Laprell et al., 2015) to obtain stable NMDA-PSCs recordings under control conditions. The pressure of injection was adjusted from experiment to experiment, as needed to obtain a stable recording, and once defined, it was not changed up to the end of the experiment. Individual current recordings were used for analysis, and their amplitude is depicted in the figure time course panels. Current tracings in the figures are averages of five consecutive individual recordings, thus corresponding to averaged recordings for 10 min before test drug application and for 10 min during peak drug effect. In NMDA-PSC recording experiments, tetrodoxin citrate (TTX) was added to aCSF to prevent activity-dependent neurotransmitter release and minimize presynaptic interference of the recordings. Afferent-evoked EPSCs mediated by NMDA receptors (NMDA-EPSCs) were obtained by stimulating the Schaffer Collateral afferents with 0.2 ms rectangular pulses delivered once every 15 s through a concentric electrode (Harvard) placed in the stratum radiatum. Averages of eight consecutive individual recordings (2 min) were used for analysis. The aCSF was supplemented with CNQX (10 μM), to block AMPA and kainate glutamate receptors and gabazine (1 μM), to block GABA A receptor activity. In paired-pulse experiments, recordings were obtained with the same condition as single-pulse recordings but two consecutive NMDA-EPSCs were evoked 100 ms apart. In all experiments, whenever assessing the influence of a drug on NMDA receptor-mediated currents, the test drug was superfused in the aCSF only after at least 10 min of stable (non-continuous change in the same direction) recordings of current amplitude. The averaged amplitude values during the 10 min before test drug application were normalized to 100% (baseline amplitude), to allow comparison of the magnitude of drug effects in different experiments; the effect of the test drug was quantified as percentage change from that of baseline. Drugs CGS 21680 (a selective A 2A receptor agonist), SCH 58261 (a selective and competitive A 2A receptor antagonist), H-89 dihydrochloride hydrate (a selective inhibitor of cAMPdependent protein kinase) and (+)-MK-801 hydrogen maleate (dizocilpine; a non-competitive NMDA antagonist acting at the ion channel site) were obtained from SigmaAldrich (MO, USA). NMDA, DL-AP5 (a selective NMDA receptor antagonist) and gabazine (SR-95531; a selective and competitive GABA A receptor antagonist) were obtained from Ascent Scientific (Bristol, UK). TTX (sodium-channel blocker) and CNQX, an antagonist of AMPA/kainate receptors, were obtained from Tocris Bioscience (Bristol, UK). Memantine was obtained from Abcam Biochemicals (Cambridge,UK).Isoflurane was obtained from Abbot Laboratories (Barcelona, Spain) and 1,2-propylenglycol from Merck (NJ, USA). CGS 21680 (5 mM) and CNQX (100 mM) were prepared as a stock solution in DMSO. TTX (1 mM) was prepared in water, as well as NMDA (100 μM), DL-AP5 (50 mM), gabazine (5 mM), MK-801 (10 mM) and memantine (100 mM). Stock solutions were aliquoted and stored at 20°C until use. Dilutions of these stock solutions to the final concentration (in aCSF) were made freshly before each experiment. Statistical analysis The data and statistical analysis comply with the recommendations on experimental design and analysis in pharmacology (Curtis et al., 2015). Data presented were tested for normality (Shapiro–Wilk test), and when passed, parametric analyses were performed. Statistical significance was evaluated by Students two-tailed paired t-test (parametric) or Wilcoxon matched-pairs signed rank test (nonparametric), FMMouroetal. 4388 British Journal of Pharmacology (2018) 175 4386–4397 while comparing data from periods before and after drug perfusion. One-way ANOVA, with the Greenhouse–Geisser correction, followed by Tukey’s multiple comparison post test (parametric) or Friedman test followed by a Dunn’s multiple comparisons test (nonparametric) was used for multiple comparisons. All statistical analyses were performed in PRISM GraphPad 6.0 (La Jolla, CA, USA). Statistical significance was assumed when the Pvalue was below 0.05. All data are presented as mean ± SEM of nexperiments, where ncorresponds to the number of recorded cells from nhippocampal slices (only one cell for each slice was used in each experiment). The number of animals used for each set of experiments (N) is also indicated in the figure legends and Table 1. 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 PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander et al., 2017a,b,c). Results Activation of adenosine A 2A receptors facilitates NMDA-PSCs in CA1 pyramidal neurons To study the actions of A 2A receptor activation on NMDA receptor function, we recorded CA1 pyramidal cells in whole-cell voltage-clamp mode (Vh = 60 mV) in hippocampal slices. NMDA receptor currents were evoked by pressure applying NMDA (150 μM,10ms)closetothesomaofthe recorded cell every 2 min (NMDA-PSCs). After a stable baseline, the A 2A agonist, CGS 21680, was applied at a concentration (30 nM) selective for A 2A receptors (Jarvis et al., 1989), resulting in an increase in NMDA-PSCs amplitude that was already visible 2–3 min after starting the changeover of the solution and reached a maximum after 40 min of perfusion (to 133 ± 13% of baseline level; Figure 1A,D, Table 1). This is in accordance with the slow kinetics of CGS 21680 binding to its receptor (Hothersall et al., 2017) and the time for activation of GPCR and kinase-dependent signalling pathways (Lohse et al., 2008). Following the maximum effect, the amplitude of NMDA-PSCs started to decrease while continuously perfusing CGS 21680 and returned to near pre-control levels (to 115 ± 8.4% of baseline) within 20–30 min after CGS 21680 washout. The effects of the A 2A agonist upon NMDA-PSCs recorded from slices from 5to 7or 8to 10-week-old rats did not differ appreciably. Both rise time (measured from 10 to 90% of peak value) and decay time(decayto50%ofpeakvalue)ofNMDA-PSCswerenot changed during perfusion of CGS 21680 (Figure 1B, Table 2) indicating no change in NMDA-PSCs kinetics (rise time change from baseline: 7.3 ± 9.4%; decay time change from baseline: 1.2 ± 9.4%). To assess if the facilitatory action of CGS 21680 could be attributed to A 2A receptor activation, we tested if this action could be prevented by the selective A 2A antagonist SCH58261 (100 nM; Zocchi et al., 1996; Rombo et al., 2015). For this purpose, SCH58261 was added to the perfusion for at least 30 min before adding the agonist to allow equilibrium with the antagonist before assessing the effect of the agonist. The presence of SCH58261 fully prevented the facilitatory action of CGS 21680 (% change by the agonist: to 105 ± 4.7% of baseline; Figure 1C,D), strongly suggesting a selective A 2A receptor-mediated effect on NMDA-PSCs. By itself, SCH58261 (100 nM) did not affect NMDA-PSCs amplitude (% change by the antagonist: to 104 ± 13% of baseline; Figure 1C,D) suggesting the lack of endogenous action of adenosine on NMDA receptor-mediated currents, under the experimental conditions used. Given that A 2A receptor activation with CGS 21680 can also facilitate AMPA receptor-mediated currents, enhancing surface expression of Glu 1 receptor subunits (Dias et al., 2012), we excluded the contribution of AMPA receptors for the recorded PSCs by perfusing the AMPA/kainate receptor antagonist, CNQX (10 μM),withnoeffectonNMDA-PSC (to 99 ± 4.5% of baseline; Figure 1E). Consistent with a selective NMDA receptor-mediated current, the perfusion of the NMDA receptor antagonist, DL-AP5 (50 μM), caused a fast decrease in NMDA-PSCs amplitude to 19 ± 2.3% of baseline after 10 min of perfusion (Figure 1E). Knowing that, in the young-adult brain, three quarters of NMDA receptors are extrasynaptically located (Hardingham and Bading, 2010), it is likely that extrasynaptic NMDA receptors greatly contribute to NMDA-PSCs evoked by pressure application of the drug over the neuronal cell body. Thus, we performed a set of experiments testing the actions of memantine, a drug that, at selected concentrations, preferentially blocks extrasynaptic NMDA receptors (Xia et al., 2010). Compatible with an important contribution of extrasynaptic NMDARs NMDA receptors to NMDA-PSCs, perfusion of memantine (1 μM) significantly decreases NMDA-PSCs to 49 ± 3.2% of baseline (Figure 1F,G). No statistically significant change in NMDA-PSC kinetics was observed(Table2).Inthepresenceofmemantine,CGS 21680 (30 nM) was still able to increase NMDA-PSCs (to 152 ± 15% of baseline values recorded before applying CGS 21680; Figure 1F inset, Figure 1G, Table 1), suggesting that the facilitatory action of A 2A receptors upon NMDA receptor activity in CA1 pyramidal cells does not occur exclusively through extrasynaptic NMDA receptors. Table 1 Comparison between the effect of the A 2A receptor agonist, CGS 21680 (30 nM), on NMDA currents at CA1 pyramidal neurons Condition % of change NMDA-PSCs 133 ± 13 (n=6;N=6) NMDA-PSCs (with MEM) 152 ± 15 (n=7;N=4) NMDA-EPSCs(noMEM) 135±5.2(n=11;N=6) NMDA-EPSCs (with MEM) 139 ± 9.7 (n=8;N=4) The values represent the action of CGS 21680 (30 nM) on NMDA currents when compared to baseline level (10 min before drug perfusion, P<0.05, Students paired t-test). The number of experiments (n)andthenumberofanimalsused(N) for each condition are shown in brackets. The magnitude of the effect did not differ among different conditions (multiple comparison one-way ANOVA was not statistically significant, P= 0.82). MEM, memantine. A 2A receptors facilitate synaptic NMDA currents British Journal of Pharmacology (2018) 175 4386–4397 4389 Figure 1 A 2A receptor activation enhances chemical-evoked NMDA-PSCs. (A) Average normalized time course changes of the effect of 30 nM CGS 21680 (horizontal bar), a selective A 2A agonist, on chemically-evoked NMDA-PSCs peak amplitude (n=6,N= 6). Zero time in the abscissae corresponds to the time where the first test solution started to be perfused. Data are presented as mean ± SEM. Top insets: tracings (average of five consecutive currents, 10 min) obtained from representative experiments before (1) and 40 min after (2) CGS 21680 superfusion. *P<0.05, baseline versus CGS 21680 effect (Wilcoxon matched-pairs test). (B) Rise time (measured from 10 to 90% of peak value, ms) and decay time (decay to 50% of peak value, ms) values (left and right panels, respectively) for each individual experiment, with a line connecting values at baseline and after 40 min of CGS 21680 perfusion. Histogram depicts mean ± SEM for each set of values. ns, not significant (rise time, n=6,P=0.84anddecaytime, n=6,P= 0.63, Wilcoxon matched-pairs test). (C) Time course of averaged normalized changes in NMDA-PSC peak amplitude induced by superfusion of SCH58261 (100 nM) (a selective A 2A antagonist) (n=3,N= 3, left panel) and CGS 21680 (30 nM) while in the presence of SCH58261 (100 nM) (n=4,N= 4, right panel). ns, not significant; left panel: baseline versus CGS 21680 (n=3,P= 0.25); right panel: SCH 58261 versus SCH 58261 + CGS 21680 (n=4,P= 0.13) (Wilcoxon matched-pairs test). (D) Summary plot showing NMDA-PSC peak amplitude (pA) of all individual cells shown in (A) and (C). (E) Averaged normalized time course changes of the effect of CNQX (10 μM; upper horizontal bar) and DL-AP5 (50 μM; bottom horizontal bar) on NMDA-PSC peak amplitude. (F) Average normalized time course changes of the effect of memantine (MEM; 1 μM, preferentially extrasynaptic NMDA receptor antagonist, upper horizontal bar) and CGS 21680 (30 nM, a selective A 2A agonist, bottom horizontal bar), on chemically-evoked NMDA-PSCs peak amplitude (n=7,N= 4); data presented as mean ± SEM; inset (obtained from the period limited by dashed line rectangle) highlights the effect of CGS 21680 normalized to baseline already in the presence of memantine (baseline recalculated from 40 to 50 min of time course). Right upper inset: sample traces (average of five consecutive currents, 10 min) taken from a representative experiment during baseline (1), memantine superfusion (2) and memantine + CGS 21680 superfusion (3). Right bottom panel: summary plot showing NMDA-PSC peak amplitude (pA) of all individual cells shown in left panel. (G) Baseline normalized NMDA-PSC amplitude values for each individual experiment, with a line connecting values from memantine perfusion and memantine + CGS 21680 perfusion. Histogram depicts mean ± SEM for each set of values. *P<0.05; n=7,N= 4, one-way ANOVA followed by Tukey’smultiple comparisons test. FMMouroetal. 4390 British Journal of Pharmacology (2018) 175 4386–4397 Activation of adenosine A 2A receptors predominantly facilitates synaptic NMDA receptor activity The results presented above suggest that synaptic-located NMDA receptors are targeted by adenosine A 2A receptors. To confirm this, we evaluated whether CGS 21680 could enhance NMDA currents when the receptors predominantly contributing to the currents were indeed synaptic NMDA receptors. We thus recorded afferent-evoked EPSCs by electrically stimulating in stratum radiatum in the continuous presence of CNQX (10 μM) and gabazine (5 μM), to block AMPA/kainate and GABA A receptors respectively. Under such conditions, the facilitatory effect of the A 2A agonist, CGS 21680 (30 nM), was maintained, enhancing NMDA-EPSCs amplitude to 135 ± 5.2% of baseline (Figure 2A,B, Table 1). Again, the effect of the A 2A agonist upon NMDA-EPSCs recorded from slices from 5to 7or 8to 10-week-old rats did not differ significantly. CGS 21680 may act at presynaptic A 2A receptors to enhance glutamate release from Schaffer collaterals with impact upon AMPA receptor-mediated synaptic transmission (Lopes et al., 2002; Dias et al., 2012). We thus assessed paired pulse ratios (PPR) to understand whether a presynaptic influence of A 2A receptor could impact upon the NMDA receptormediated synaptic currents. The paired-pulse ratio (PPR = 2nd EPSC amplitude /1st EPSC amplitude )oftwoconsecutiveNMDAEPSCs 100 ms apart was analysed, and no significant alterations were detected in the presence of CGS 21680 (99 ± 6.8% of baseline; Figure 2A,B), indicating that A 2A receptor’sactiononNMDA-EPSCsdoesnotresultfrom enhanced release of glutamate. The CGS 21680 effect was not accompanied by changes in NMDA-EPSCs kinetics as evidenced by unaltered rise and decay times of both first and second pulses (rise time change from baseline: 2.2 ± 3.2% for first pulse and 2.3 ± 6.6% for second pulse; decay time change from baseline: 0.1 ± 4.8% for first pulse and 0.7 ± 3.9% for second pulse; Figure 2C, Table 2). As expected, the NMDA-EPSCs were fully blocked by the perfusion of the NMDA antagonist DL-AP5 (50 μM) (currents decreased to 7.9 ± 1.2% of baseline; Figure 2D), confirming the NMDA-mediated nature of the evoked responses. Altogether, the above experiments suggest that synaptic NMDA receptors, rather than extrasynaptic ones, are targeted by A 2A receptors. To further assess this possibility, the next series of experiments were designed to better discriminate between A 2A receptor modulation over synaptic or extrasynaptic NMDA receptors. We first evaluated the possibility that peri and extrasynaptic NMDA receptors could be contributing to the recorded EPSCs, since they can be recruited either by glutamate spillover into extrasynaptic compartments or during intense synaptic activity. Thus, we recorded NMDA-EPSCs (0.067 Hz, paired-pulse) and perfused an open-channel NMDA receptor blocker, MK-801 (10 μM), to use-dependently inhibit synaptically evoked NMDA-EPSCs and isolate extrasynaptically recruited NMDA receptors. In fact, MK-801 perfusion decreased NMDA-EPSCs to 31 ± 1.9% of baseline (Figure 3A,B), without altering NMDA receptor current kinetics (Table 2), revealing a strong component of synaptic NMDA receptor activity on electrical-evoked stimulation conditions. We then perfused CGS 21680 (30 nM) in the presence of MK-801, aiming to activate A 2A receptors in conditions in which extrasynaptic NMDA receptors are the ones mostly recruited. Under these conditions CGS 2160 caused no statistically significant changes on NMDA-EPSCs amplitude (to 111 ± 12.6% of baseline values recorded before applying CGS 21680; Figure 3A,B) or current kinetics Table 2 Comparison between the rise and decay times of NMDA receptor-mediated currents in different recordings and conditions Recordings Condition Rise time (ms) Statistics (vs. baseline) Decay time (ms) Statistics (vs. baseline) NMDA-PSC (Figure 1A) Baseline 146 ± 36 n=6,P=0.84 591±112 n=6,P=0.63 CGS 21680 (30 nM) 147 ± 31 571 ± 104 NMDA-PSC (MEM) (Figure 1F) Baseline 188 ± 27 n=7,P=0.94 650±124 n=7,P=0.94 MEM (1 μM) 190 ± 39 653 ± 149 NMDA-EPSC (1stP) (Figure 2A) Baseline 14 ± 2.8 n=11,P=0.95 32±4.2 n=7,P=0.81 CGS 21680 14 ± 2.7 32 ± 4.3 NMDA-EPSC (2ndP) (Figure 2A) Baseline 12 ± 1.7 n=11,P=0.98 33±4.0 n=7,P=0.84 CGS 21680 12 ± 1.6 33 ± 3.8 NMDA-EPSC (MK-801) (Figure 3A) Baseline 22 ± 1.9 n=10,P=0.05 37±3.1 n=10,P=0.71 MK-801 (10 μM) 23 ± 2.9 36 ± 3.0 MK-801 + CGS 21680 19 ± 1.7 35 ± 3.9 NMDA-EPSC (MEM) (Figure 3D) Baseline 40 ± 4.2 n=8,P=0.11 76±5.3 n=8,P=0.13 MEM (1 μM) 36 ± 4.7 71 ± 3.8 MEM + CGS 21680 31 ± 5.6 72 ± 5.0 No statistically significant changes were observed in any of the conditions evaluated. 1stP, first pulse; 2ndP, second pulse; MEM: memantine. A 2A receptors facilitate synaptic NMDA currents British Journal of Pharmacology (2018) 175 4386–4397 4391 (Table 2), indicating the lack of A 2A receptor action upon extrasynaptic NMDA receptors. We next performed a set of experiments testing the actions of memantine on afferent-evoked NMDA-EPSCs. Perfusion of memantine (1 μM) caused no significant change in NMDA-EPSCs amplitude (to 96 ± 4.8% of baseline; Figure 3C,D) or kinetics (Table 2). Consistent with A 2A receptor action being preferentially upon synaptic NMDA receptors, perfusion of CGS 21680 in the presence of memantine lead to a facilitation of NMDA-EPSCs to 139 ± 9.7% of baseline values recorded before perfusion of CGS21680 (Figure 3C,D), with no change in current kinetics (Table 2), similar to what was observed in the absence of memantine (Figure 2A, Table 1). Finally, we tested for the dependency of A 2A receptors actions on classical signalling pathway involving cAMP/PKA activation (Fredholm et al., 2011). To test this, we first perfused a selective inhibitor of cAMP-dependent protein kinase, H-89, at a concentration (1 μM) that causes marked inhibition of PKA (Bain et al., 2007) without inhibiting other kinases that could possibly be involved in the synaptic actions of A 2A receptors, such as ERK or MAPK,PKCαor CAMK (Bain et al., 2007). H-89, per se,hadnoeffecton NMDA-EPSCs (to 97 ± 2.8% of baseline before H-89, Figure 3E,F) but completely prevented the action of CGS 21680 on NMDA-EPSCs (to 108 ± 3.8% of baseline before adding CGS 21680; Figure 3E,F), a finding compatible with the involvement of canonical pathway for A 2A receptors, the cAMP/PKA signalling cascade, in the modulating effect of A 2A receptors on NMDA receptors. Altogether, our results strongly indicate a preferential action of A 2A receptors on synaptic and not extrasynaptic NMDA receptors via cAMP/PKA signalling. Figure 2 Adenosine A 2A receptor activation potentiates afferent-evoked NMDA receptor-mediated currents (NMDA-EPSCs). (A) Average normalized time course changes in afferent-evoked NMDA-EPSCs amplitude (first pulse amplitude, open symbols) and PPR (second vs. first pulse amplitude) induced by superfusion of selective A 2A receptor agonist, CGS 21680 (30 nM) (horizontal bar). Zero time in the abscissae corresponds to the time where the first test solution started to be perfused. Data are presented as mean ± SEM (n= 11, N=6).Right upper inset:sample traces (average of eight consecutive currents, 2 min) taken from a representative experiment; before (1) and 30 min after CGS 21680 (2) superfusion; trace showing 30 min after CGS 21680 superfusion was normalized for baseline trace (1 + 2norm) to point out the PPR; the stimulus artefact for each trace was removed for better visualization. Right bottom panel: summary plot showing baseline-normalized NMDA-EPSCs peak amplitude (first pulse amplitude, Amp) and PPR (second vs. first pulse amplitude) following addition of CGS 21680. Circles correspond to individual experiments; triangles represent mean ± SEM. *P<0.05; ns, not significant (n=11,N=6,pairedt-test). (B) Summary plot showing NMDA-EPSC peak amplitude (first pulse amplitude, pA) (left panel) and PPR (second vs. first pulse amplitude) (right panel) of all individual cells shown in (A). (C) Rise time (measured from 10 to 90% of peak value, ms) and decay time (decay to 50% of peak value, ms) values (left and right panels, respectively) for first and second NMDA-EPSC pulses of each individual experiment, with a line connecting values at baseline and after 30 min of CGS 21680 perfusion. Histograms depict mean ± SEM for each set of values. ns, not significant [rise time first pulse, n=11,P= 0.95; rise time second pulse, n= 11, P= 0.98; decay time second pulse, n= 11, P=0.84 (Student’spairedt-test) and decay time first pulse, n=7,P= 0.81, Wilcoxon matched-pairs test]. (D) Averaged normalized time course changes of the effect of DL-AP5 (50 μM) (horizontal bar) on NMDA-EPSC peak amplitude. Right upper inset: sample traces (average of eight consecutive currents, 2 min) taken from a representative experiment before (1) and 20 min after DL-AP5 (2) superfusion; the stimulus artefact for each trace was removed for better visualization. Right bottom panel: summary plot showing NMDA-EPSC peak amplitude (first pulse amplitude, pA) of all individual cells shown in right panel (3: baseline; 4, DL-AP5). FMMouroetal. 4392 British Journal of Pharmacology (2018) 175 4386–4397 Figure 3 Synaptic, but not extrasynaptic, NMDA-EPSCs are affected by adenosine A 2A receptor activation through cAMP/PKA signalling. (A) Average normalized time course changes of the effect of MK-801 (10 μM, an open-channel NMDA receptor blocker, upper horizontal bar) and CGS 21680 (30 nM, a selective A 2A receptor agonist, bottom horizontal bar), on electrically-evoked NMDA-EPSCs peak amplitude (n=10,N= 5). Zero time in the abscissae corresponds to the time when the first test solution started to be perfused. Data are presented as mean ± SEM; inset (obtained from the period limited by dashed line rectangle) highlights the effect of CGS 21680 normalized to baseline already in the presence of MK-801 (baseline recalculated from 35 to 45 min of time course). Right upper inset: sample traces (average of 8 consecutive currents, 2 min) taken from a representative experiment during baseline (1), MK-801 superfusion (2) and MK-801 + CGS 21680 superfusion (3); the stimulus artefact for each trace was removed for better visualization. Right bottom panel: summary plot showing NMDA-EPSC peak amplitude (pA) of all individual cells shown in left panel. (B) Baseline-normalized NMDA-EPSC amplitude values for each individual experiment, with a line connecting values during MK-801 perfusion and MK-801 + CGS 21680 perfusion. Histogram depicts mean ± SEM for each set of values. *P<0.05; ns, not significant, n=10, N= 5, one-way ANOVA followed by Tukey’s multiple comparisons test. (C) Average normalized time course changes for the effect of memantine (1 μM, preferentially extrasynaptic NMDA receptor antagonist, upper horizontal bar) and CGS 21680 (30 nM, a selective A 2A receptor agonist, bottom horizontal bar), on electrically-evoked NMDA-EPSCs peak amplitude (n=8,N= 4); data presented as mean ± SEM; inset (obtained from the period limited by dashed line rectangle) highlights the effect of CGS 21680 normalized to baseline already in the presence of Memantine (baseline recalculated from 35 to 45 min of time course). Right upper inset: sample traces (average of eight consecutive currents, 2 min) taken from a representative experiment during baseline (1), memantine superfusion (2) and memantine + CGS 21680 superfusion (3); the stimulus artefact for each trace was removed for better visualization. Right bottom panel: summary plot showing NMDA-EPSC peak amplitude (pA) of all individual cells shown in left panel. (D) Baseline-normalized NMDA-EPSC amplitude values for each individual experiment, with a line connecting values during memantine perfusion and memantine + CGS 21680 perfusion. Histogram depicts mean ± SEM for each set of values. ns, not significant; *P<0.05; n=8,N= 4, one-way ANOVA followed by Tukey’s multiple comparisons test. (E) Average normalized time course changes of the effect of H-89 (1 μM, a selective inhibitor of cAMP-dependent protein kinase, upper horizontal bar) and CGS 21680 (30 nM, a selective A 2A agonist, bottom horizontal bar), on electrically-evoked NMDA-EPSCs peak amplitude (n=5,N= 3); data presented as mean ± SEM. Right upper inset:sample traces (average of eight consecutive currents, 2 min) taken from a representative experiment during baseline (1), H-89 superfusion (2) and H89 + CG S21680 superfusion (3); the stimulus artefact for each trace was removed for better visualization. Right bottom panel: summary plot showing NMDA-EPSC peak amplitude (pA) of all individual cells shown in left panel. (F) Baseline-normalized NMDA-EPSC amplitude values for each individual experiment, with a line connecting values during H-89 perfusion and H-89 + CGS 21680 perfusion. Histogram depicts mean ± SEM for each set of values. ns, not significant; n=5,N=5,Kruskal–Wallis test followed by Dunn’s multiple comparisons test. A 2A receptors facilitate synaptic NMDA currents British Journal of Pharmacology (2018) 175 4386–4397 4393 Discussion The main finding in the present work is that adenosine A 2A receptors facilitate postsynaptic NMDA currents in CA1 pyramidal neurons and that this facilitation preferentially occurs at the synaptic rather than extrasynaptic level. A 2A receptors at the hippocampus are scarcely distributed but are known to affect hippocampal synaptic transmission at different levels. A predominant role of hippocampal A 2A receptors is to facilitate the action of other neuromodulators, such as BDNF (Diógenes et al., 2004), or receptors for neurotransmitters, such as glutamate AMPA receptors (Dias et al., 2012) and mGlu 5 receptors (Tebano et al., 2005). Presynaptically, A 2A receptors are also known to increase glutamate release, although in young rats this action is small and mostly results from a blockade of A 1 receptor-mediated inhibitory effect (Lopes et al., 2002). In addition, A 2A receptors at the hippocampus also enhance GABA release (Cunha and Ribeiro, 2000), an action that is synapse-specific resulting in a disinhibition of glutamatergic neurons (Rombo et al., 2015). All these A 2A receptor-mediated actions may impact synaptic plasticity (Dias et al., 2013a; Sebastião and Ribeiro, 2015), as demonstrated by A 2A receptor facilitation of LTP (Costenla et al., 1999; Dias et al., 2012) and post-ischaemia plasticity phenomena (Dias et al., 2013b). The influence of A 2A receptors on hippocampal NMDA receptors has been much less explored. There is, however, evidence on dentate gyrysCA3 synapses that A 2A receptor activation unmasks a new form of NMDA receptor-dependent LTP that most likely occurs at the postsynaptic level (Rebola et al., 2008). However, evidence for a direct postsynaptic influence of A 2A receptors upon NMDA receptors was until now lacking. The data obtained in the present study have revealed this missing link, clearly showing that NMDA receptors are targets for A 2A receptors likely through the canonical cAMP/PKA signalling pathway, and importantly, we showed that this action is evident in young animals and occurs predominantly at synaptic rather than extrasynaptic receptors. Furthermore, this action was clearly detected in the CA1 hippocampal area where synaptic plasticity mostly depends upon NMDA receptors. A 2A receptors are known to affect memory tasks dependent on intact hippocampal activity, an action that seems to depend on the age and physiological state of the animal. Thus, while in young animals, A 2A receptor blockade disturbs hippocampal-dependent conditional learning as well as prevents the concomitant increase in synaptic potentials at the hippocampus (Fontinha et al., 2009), supporting a facilitatory action of A 2A receptor synaptic plasticity and memory acquisition, in aged animals, where spurious LTP occurs at Schaffer CA1 synapses (Diógenes et al., 2007), A 2A receptor blockade restores plasticity towards normal values (Costenla et al., 2011).Thesamealsooccursin chronically stressed animals where A 2A receptor blockade has been shown to counteract exacerbated plasticity and to facilitate learning (Batalha et al., 2013), thus suggesting anegativeinfluence of A 2A receptors upon hippocampaldependent tasks. This apparent contradiction on the influence of A 2A receptors in learning in young versus aged and/or diseased animals may result from the different roles of A 2A receptor in heathy young animals, aged animals or under disease conditions (Sebastião and Ribeiro, 2009). ThesameisalsoknowntooccurwithNMDAreceptors. Indeed, it is widely accepted that synaptic NMDA receptor activation is necessary for several forms of memory encoding and retrieval. Thus, our results showing that A 2A receptors can facilitate synaptic NMDA receptors in young animals help to explain the negative effects of A 2A receptor antagonists on certain learning paradigms in healthy young animals. In contrast, neurodegenerative conditions are known to increase activation of NMDA receptors and calciumpermeable AMPA receptors, leading to hippocampal synaptic plasticity impairment (Diógenes et al., 2012; Ferreira et al., 2017; Tanqueiro et al., 2018). There is plenty of evidence that under pathological conditions, A 2A receptor signalling is exacerbated, which contributes to pathogenesis and leads to early cognitive and synaptic dysfunction as well as neurodegeneration (Canas et al., 2009; Cognato et al., 2010; Batalha et al., 2013; Kadowaki-Horita et al., 2013; Kaster et al., 2015; Li et al., 2015a,b; Hu et al., 2016; Laurent et al., 2016; Viana da Silva et al., 2016). Under such pathological conditions, A 2A receptor blockade restores learning and/or memory (Canas et al., 2009; Cognato et al., 2010; Batalha et al., 2013; Kadowaki-Horita et al., 2013; Kaster et al., 2015; Li et al., 2015b; Viana da Silva et al., 2016). A 2A receptor-mediated dysfunctions may even precede NMDA receptor dysfunction (Viana da Silva et al., 2016). In most of the above-mentioned studies, the dysfunctional A 2A receptor activity predominantly occurs at the presynaptic level, enhancing glutamate release, therefore, being prone to indirectly facilitate NMDA receptor activation, which under pathological conditions may provide a positive feedback loop towards neurodegeneration. Interestingly, in healthy aged animals, enhanced synaptic plasticity and NMDA receptor activation seem to compensate cognitive decline (Pinho et al., 2017). One may thus think that while in disease conditions, A 2A receptors exacerbate excitotoxicity and, under non-diseased conditions, a facilitatory action of A 2A receptors on several steps of excitatory synaptic transmission may compensate some loss of function in the aged. Importantly, since synaptic and extrasynaptic NMDA receptors have different roles in plasticity and in neurodegeneration, with the extrasynaptic NMDA receptors being more prone to gate neurodegeneration and synaptic NMDA receptors mostly involved in synaptic plasticity (Hardingham and Bading, 2010), a drug that could selectively facilitate synaptic NMDA receptors without affecting extrasynaptic ones could be a candidate to selectively promote plasticity with minimal consequences on degeneration. Selective A 2A receptor agonists may prove to usefully fulfilthistask. In conclusion, our results showing that A 2A receptors predominantly affect synaptic NMDA receptors thus suggest that, at least in young animals, A 2A receptor activation may contribute to a shift in the dual role of NMDA receptors towards enhancement of synaptic plasticity. Further studies in healthy aged animals are desirable to clarify weather this selective influence of A 2A receptors on synaptic NMDA receptors, rather than extrasynaptic receptors, may prove beneficial to enhance cognition without triggering neurodegeneration. FMMouroetal. 4394 British Journal of Pharmacology (2018) 175 4386–4397