The impact of inosine on hippocampal synaptic transmission and plasticity involves the release of adenosine through equilibrative nucleoside transporters rather than the direct activation of adenosine receptors

Abstract

Inosine has robust neuroprotective effects, but it is unclear if inosine acts as direct ligand of adenosine receptors or if it triggers metabolic effects indirectly modifying the activity of adenosine receptors. We now combined radioligand binding studies with electrophysiological recordings in hippocampal slices to test how inosine controls synaptic transmission and plasticity. Inosine was without effect at 30 μM and decreased field excitatory post-synaptic potentials by 14% and 33% at 100 and 300 μM, respectively. These effects were prevented by the adenosine A₁ receptor antagonist DPCPX. Inosine at 300 (but not 100) μM also decreased the magnitude of long-term potentiation (LTP), an effect prevented by DPCPX and by the adenosine A_2A receptor antagonist SCH58261. Inosine showed low affinity towards human and rat adenosine receptor subtypes with K_i values of > 300 µM; only at the human and rat A₁ receptor slightly higher affinities with K_i values of around 100 µM were observed. Affinity of inosine at the rat A₃ receptor was higher (K_i of 1.37 µM), while it showed no interaction with the human orthologue. Notably, the effects of inosine on synaptic transmission and plasticity were abrogated by adenosine deaminase and by inhibiting equilibrative nucleoside transporters (ENT) with dipyridamole and NBTI. This shows that the impact of inosine on hippocampal synaptic transmission and plasticity is not due to a direct activation of adenosine receptors but is instead due to an indirect modification of the tonic activation of these adenosine receptors through an ENT-mediated modification of the extracellular levels of adenosine.

Introduction

Adenosine is a neuromodulator in the brain, acting mainly through inhibitory A₁ receptors (A₁R) and facilitatory A_2AR [29]. The combined action of these two receptors is paramount to assist encoding salience in neuronal networks, whereby A₁R decrease basal transmission and A_2AR bolster synaptic plasticity [1]. The importance of this neuromodulation system is best heralded by the impact of its imbalance in different neuropsychiatric diseases and the neuroprotection afforded by bolstering A₁R function or dampening A_2AR overfunction [18]. In fact, noxious brain stimuli trigger the outflow of adenosine, which has a short half-life due to efficient re-uptake through equilibrative nucleoside transporters (ENTs) driven by an effective intracellular metabolism by astrocytic adenosine kinase and ubiquitous adenosine deaminase [64]. Noxious brain conditions also lead to inosine outflow, involving ENTs, which reach extracellular concentrations approximately 5–10 times higher than adenosine [8, 39, 55].

Inosine was reported to have very low affinity for brain A₁R and A_2AR [6, 47, 69] and is considered to be inactive on adenosine receptors based on the founding studies of the purinergic system [71]. Radioligand binding studies are typically performed in the presence of adenosine deaminase (ADA), an enzyme that converts adenosine present in the preparations to inosine, and the formed inosine does not disturb the studies [3]. However, several recent studies have illustrated a neuroprotective role of inosine [20, 25, 30, 38, 52, 53, 60, 62, 63], but their use of whole animals, which are not particularly amenable to dissect detailed molecular mechanisms, did not allow clarifying if inosine was directly acting through activation of adenosine receptors or rather through an indirect mechanism.

Since adenosine neuromodulation of neuronal activity in hippocampal slices is well-established to involve a combined action of A₁R-mediated inhibition of synaptic transmission and an A_2AR-mediated facilitation of synaptic plasticity, namely, long-term potentiation [46], with a discrete effect of A_2BR [34] or A₃R [24, 44], we now resorted to a hippocampal slice model in the present study to clarify if and how inosine affects neuronal activity.

Materials and methods

Animals

Male C57BL/6 J mice (8–12 weeks old; total of 30) and male Wistar rats (8–12 weeks old; total of 32) were from Charles River (Barcelona, Spain). Animals were maintained in groups of two to five per cage in a temperature-controlled room (22 ± 1 °C), with free access to food and water, and with a 12-h light/dark cycle (lights on at 7:00 am). Animals were handled according to ARRIVE guidelines, as approved by CNC Ethical Committee for Animal Research (ORBEA-128/2015).

Drugs

Inosine, DPCPX (1,3-dipropyl-8-cyclopentylxanthine; Cat# 1111), SCH58261 (2-(2-furanyl)-7-(2-phenylethyl)-7H-pyrazolo[4,3-e][1,2,4]triazolo[1,5-c]pyrimidin-5-amine; Cat# 2270), and dipyridamole (Cat# 0691) were from Tocris (Bristol, UK). Adenosine deaminase (ADA; Cat# 10102105001) and NBTI (S-(p-nitrobenzyl)-6-thioinosine; Cat# N2255) were from Sigma. [³H]SCH58261 (specific activity of 77 Ci/mmol; prepared by Amersham) was generously offered by Dr. Ennio Ongini (Shering-Plough, Milan, Italy) and [³H]DPCPX (specific activity of 102.1 Ci/mmol) was from DuPont NEN (Boston, MA, USA). The precursor of [³H]2-chloro-N⁶-cyclopentyladenosine (CCPA) [41], [³H]PSB-11 [51], and [³H]PSB-603 [10] were synthesized as previously described and tritium-labeled by Pharmaron, Cardiff, UK. [³H]CGS21680 and [³H]NECA (5′-N-ethylcarboxamido-adenosine) were purchased from PerkinElmer Life and Analytical Science (Rodgau-Jügesheim, Germany). 2-Chloroadenosine and NECA were purchased from Sigma-Aldrich (Taufkirchen, Germany) and Tocris Biosciences (Bristol, UK).

SCH58261, dipyridamole, and NBTI were made up as a 5 mM stock solution in dimethylsulfoxide (DMSO). DPCPX was made up into a 5 mM stock solution in 99% DMSO and 1% NaOH (1 M). All drug stock solutions were diluted directly into the superfusion solution to the appropriate final concentration. DMSO, in the maximal concentration applied to the preparations, was devoid of effects on synaptic plasticity experiments. Except for inosine, all drugs used in functional studied were used at supramaximal but selective concentrations based on our previous studies: 100 nM DPCPX [61], 50 nM SCH58261 [58], 2 U/mL ADA [17], 10 µM dipyridamole, and 5 µM NBTI [2].

Extracellular electrophysiological recordings

Electrophysiological recordings of synaptic transmission and plasticity in hippocampal slices were carried out as previously described [46]. Briefly, after deep halothane anesthesia, mice were sacrificed by decapitation and the hippocampus was dissected to prepare 400-µm-thick transverse slices from the dorsal hippocampi using a McIlwain tissue chopper (Brinkmann Instruments, USA, RRID:SCR_015798). Slices were allowed to recover functional and energetically for at least 1 h in a preincubation chamber (BSC-PC prechamber, Harvard Apparatus, Holliston, USA, No. PY2 65–0076) with gassed ACSF at 34 °C. Individual slices were then transferred to a submerged recording chamber with 1-mL capacity (Harvard Apparatus, MA, USA) and continuously superfused at a flow rate of 3 mL/min with oxygenated ACSF (control) kept at 30.5 °C (TC-202A Bipolar Temperature Controller, Harvard Apparatus, MA, USA). A bipolar concentric electrode was placed in the proximal CA1 stratum radiatum for stimulation of the Schaffer fibers every 20 s with regular pulses of 0.1 ms delivered with a Grass S44 square pulse stimulator (Grass Technologies, RI, USA). The orthodromically evoked field excitatory post-synaptic potentials (fEPSP) were recorded through an extracellular microelectrode pipette filled with NaCl 4 M yielding a 1–2-MΩ resistance and placed in the CA1 stratum radiatum targeting the distal dendrites of pyramidal neurons. Recordings were obtained with an ISO-80 amplifier (World Precision Instruments, Hertfordshire, UK), digitized using an analog-to-digital converter (BNC-2110, National Instruments, Newbury, UK), and analyzed using the WinLTP version 2.20.1 software (WinLTP Ltd., UK, RRID: SCR_008590) [5]. An input/output curve (I/O) (percentage of maximum fEPSP slope versus stimulus intensity) was obtained to select a stimulation intensity yielding ∼40% of maximal fEPSP response. After a minimum period of 30 min to fully stabilize baseline fEPSP responses in the absence or presence of tested drugs and resetting stimulation to yield ∼40% of maximal fEPSP response, LTP was induced by a high-frequency stimulation (HFS) train (100 Hz for 1 s) [46]. LTP magnitude was calculated as the percentage change between the average slope of the ten averaged potentials taken between 50 and 60 min after terminating LTP induction in relation to the average slope of the fEPSP measured during the 10 min that preceded LTP induction. The effect of drugs on LTP was assessed by comparing LTP magnitude in the absence and presence of the drug in experiments carried out in different slices from the same animal [46].

Radioligand binding assays

Radioligand binding assays at human and rat adenosine receptor subtypes were performed as previously described using membranes from CHO cells heterologously expressing the different adenosine receptors or rat brain membranes, prepared as previously described [3, 10]. All experiments were performed in the presence of 2 U/mL ADA using the suitable radioligands for each receptor subtype, as indicated below in each subsection.

After the indicated incubation time, the assay mixtures were filtered through GF/B glass fiber filters using a Brandel harvester (Brandel, Gaithersburg, MD, USA). Filters were washed three times (3–4 mL each) with ice-cold 50 mM Tris–HCl buffer, pH 7.4. Then, the filters were transferred to scintillation vials, incubated for 9 h with 2.5 mL of scintillation cocktail (Luma Safe, PerkinElmer), and counted in a liquid scintillation counter (Tri-Carb 2810 TR) with a counting efficiency of $\sim$ 53%. All data were analyzed with GraphPad Prism (GraphPad Inc., La Jolla, CA).

Radioligand binding assays at human and rat A₁ adenosine receptors

Competition binding experiments using the agonist radioligand [³H]2-chloro-N⁶-cyclopentyladenosine (CCPA) were performed in a final volume of 400 µL containing 10 µL of test compound dissolved in DMSO, 190 µL buffer (50 mM Tris–HCl, pH 7.4), 100 µL of radioligand solution in the same buffer (final concentration 1 nM), and 100 µL of a suspension of a membrane preparation (human: 6 µg of protein per vial in buffer, recombinantly expressed human A₁ adenosine receptor in CHO cells; lot #1604613, PerkinElmer; rat: rat brain cortical membrane preparation, 150 µg of protein per vial). Non-specific binding was determined in the presence of 2-chloroadenosine (CADO) (final concentration 10 µM) and the incubation time was 90 min at room temperature.

Radioligand binding assays at human and rat A_2A adenosine receptors

Competition binding experiments using the agonist radioligand [³H]CGS21680 were performed in a final volume of 400 µL containing 10 µL of test compound dissolved in DMSO, 190 µL buffer (50 mM Tris–HCl, 10 mM MgCl₂, pH 7.4), 100 µL of radioligand solution in the same buffer (final concentration 5 nM), and 100 µL of membrane preparation (human: 4 µg of protein per vial in buffer, recombinantly expressed in CHO cells; rat: rat brain striatal membrane preparation, 200 µg protein per vial). Non-specific binding was determined in the presence of 5′-N-ethylcarboxamidoadenosine (NECA; final concentration 10 µM) and the incubation period was 120 min at room temperature.

Radioligand binding assays at human and rat A_2B adenosine receptors

Competition binding experiments with [³H]NECA were performed in a final volume of 1 mL containing 10 µL of test compound dissolved in 100% DMSO, 790 µL buffer (50 mM Tris–HCl, 10 mM MgCl₂, pH 7.4), 100 μL of radioligand solution in the same buffer (final concentration 30 nM), and 100 μL of membrane preparation (human: 20 μg protein per vial in the same buffer of Lot #1599597 or 9.5 µg protein per vial in the same buffer of Lot # 2302323, PerkinElmer; rat: 70 µg protein per vial in buffer). Non-specific binding was determined in the presence of 250 μM (final concentration) NECA and the incubation time was 4 h at 4 °C.

Competition binding experiments with [³H]PSB-603 were performed in a final volume of 1 mL containing 10 µL of test compound dissolved in DMSO, 790 µL buffer (50 mM Tris–HCl, pH 7,4), 100 μL of radioligand solution in the same buffer (final concentration 0.3 nM), and 100 μL of membrane preparation (10 μg protein per vial in the same buffer, Lot 1599597, PerkinElmer). Non-specific binding was determined in the presence of 10 µM (final concentration) DPCPX and the incubation time was 75 min at room temperature.

Radioligand binding assays at human and rat A₃ adenosine receptors

Competition binding experiments with [³H]5′-N-ethylcarboxamidoadenosine (NECA) were performed in a final volume of 400 µL containing 4 µL of test compound dissolved in DMSO, 200 µL of buffer (50 mM Tris–HCL, 10 mM MgCl₂, 1 mM EDTA, pH 8.25), 100 µL of radioligand solution in the same buffer (final concentration 10 nM), and 100 µL of membrane preparation (human: 9 µg protein per vial in buffer, Lot #1619113, PerkinElmer or 30 µg of protein per vial in buffer, transiently transfected CHO-S cell membrane preparations; rat: 35 µg of protein per vial in buffer, Lot # 1648417, PerkinElmer or 30 µg of protein per vial in buffer, transiently transfected CHO-S cell membrane preparations). Non-specific binding was determined in the presence of N⁶-((R)-2-phenylisopropyl)adenosine (final concentration 100 µM) and the incubation time was 180 min at room temperature.

Competition binding experiments with [³H]PSB-11 were performed in a final volume of 400 µL containing 4 µL of test compound dissolved in DMSO, 196 µL buffer (50 mM Tris–HCL, pH 7.4), 100 µL of radioligand solution in the same buffer (final concentration 1 nM), and 100 µL of membrane preparation (150 µg protein per vial in buffer). Non-specific binding was determined in the presence of N⁶-((R)-2-phenylisopropyl)adenosine (final concentration 100 µM) and the incubation time was 45 min at room temperature.

Radioligand binding assays at synaptic A₁R and A_2AR

The competition by inosine (0.1–1000 μM) of the binding of the selective A₁R antagonist [³H]DPCPX (2 nM) or of the selective A_2AR antagonist [³H]SCH58261 (2 nM) was performed as previously described [45] using synaptosomal membranes from rat cerebrocortical synaptosomes, prepared as previously described [57]. Briefly, the binding of [³H]DPCPX (2 nM) or of [³H]SCH58261 (2 nM) was carried out for 2 h at 37 °C and for 4 h at room temperature (23–25 °C), with 62–110 μg of membrane protein in a final volume of 200 μL in an incubation solution containing 50 mM Tris–HCl and MgCl₂ (2 mM for A₁R, 10 mM for A_2AR) at pH 7.4, with 4 U/mL adenosine deaminase. Specific binding was determined by subtraction of the non-specific binding which was measured in the presence of 2 μM XAC. The binding reactions were stopped by vacuum filtration through Whatman GF/C glass fiber filters, followed by washing of the filters and reaction tubes with 5 mL of the incubation solution, kept at 4 °C. The filters were then placed in scintillation vials, and 2 mL of scintillation liquid (Scintran Cocktail T; Wallac, Turku, Finland) was added. Radioactivity bound to the filters was determined after 12 h with an efficiency of 55–60% for 2 min. Binding data was expressed as the percentage of binding of [³H]DPCPX (2 nM) or [³H]SCH58261 (2 nM) in the absence of competitor and adjusted by non-linear regression assuming one single binding site and complete displacement to first calculate the IC₅₀ values and then the corresponding K_i values according to the Cheng and Prusoff equation using the previously determined K_D values of [³H]DPCPX (1.2 nM) and of [³H]SCH58261 (0.6 nM) binding to synaptosomal membranes [45].

Statistics

In all experimental procedures, four or more animals were used for each parameter analyzed and the individual sample size (n = number of animals) is specified for each experiment. All data are presented as mean ± SEM and significance was considered at p < 0.05 using a Student’s t test to determine the effects of a drug versus baseline, whereas comparison between two groups was carried out using a paired Student’s t test.

Results

To investigate the impact of inosine on synaptic transmission and plasticity in representative neuronal circuits such as in the hippocampus, it is first important to define the concentration levels of inosine expected to occur in the brain parenchyma. Typically, the measured concentration of extracellular inosine in brain preparations is approximately 10 times greater than that of adenosine [8, 39, 55]. Since extracellular adenosine levels under non-noxious conditions in hippocampal excitatory synapses are in the low micromolar range [17, 21, 23], we tested inosine in concentrations ranging from 30 to 300 μM.

Involvement of adenosine receptors in the ability of inosine to control synaptic transmission and plasticity

In Schaffer fibers-CA1 pyramid synapses of hippocampal slices from adult mice, inosine did not significantly affect the evoked field excitatory post-synaptic potential (fEPSPs) at a concentration of 30 μM (Fig. 1A), but depressed fEPSPs in an apparent concentration-dependent manner at higher concentrations with an inhibition of 13.8 ± 3.5% (n = 6, p = 0.011) at 100 μM and of 32.6 ± 2.9% (n = 6, p < 0.001) at 300 μM (Fig. 1A), essentially as previously reported by others [4]. This likely resulted from a presynaptic effect, since 100–300 μM inosine increased paired-pulse facilitation using inter-pulse intervals of either 25 ms or 50 ms (Fig. 1B). When studying long-term potentiation (LTP) as a representative plastic process in hippocampal synapses, we observed that 100 μM inosine did not significantly alter LTP magnitude (9.65 ± 1.73% inhibition; n = 6, p = 0.736), but 300 μM inosine decreased LTP magnitude by 71.0 ± 18.8% (n = 6, p < 0.001) (Fig. 1C, D).

Fig. 1.

The inhibitory effect of inosine on synaptic transmission and plasticity in mouse hippocampal slices is mediated by adenosine A₁ and A_2A receptors. A Although inactive at a concentration of 30 μM, inosine concentration dependently inhibited synaptic transmission at concentrations of 100–300 μM. The inserts correspond to representative fEPSPs under each condition. B Inosine (100–300 μM) increase paired pulse facilitation with inter-pulse intervals of 25 and 50 ms, suggesting that inosine-induced inhibition of synaptic transmission is exerted presynaptically. C Although devoid of significant effects at 100 μM, at 300 μM inosine depressed the magnitude of long-term potentiation induced by a high-frequency (HF) train (100 Hz for 1 s) applied at time 0. The inserts correspond to representative fEPSPs under each condition, dashed lines before delivering the HF, and filled lines 60 min after the HF train. D Average values of LTP magnitude calculated as the variation of fEPSP slope before and 60 min after HF stimulation. E The selective A₁ receptor antagonist DPCPX (100 nM) prevented the ability of 300 μM inosine to decrease synaptic transmission. F DPCPX also curtailed the ability of inosine to inhibit LTP (insert with average values). G The selective A_2A receptor antagonist SCH58261 (50 nM) also fully eliminated the ability of inosine to inhibit LTP (insert with average values). Data are mean ± SEM of 5–6 determination (independent mice). *p < 0.05 with paired Student’s t test, ns: non-significant

Since it is well established that hippocampal synaptic transmission in adult rodents is mostly under the control of adenosine A₁ receptors (A₁R), without a relevant participation of A_2AR [16, 19] with only discrete modulatory roles of A₃R and A_2BR that seem strictly dependent on A₁R function [24, 34, 44], we tested the impact of a supramaximal concentration of the selective A₁R antagonist DPCPX (100 nM; see [61] on the inhibition by inosine of synaptic transmission and plasticity. DPCPX (100 nM) caused the expected increase of fEPSP slopes (25.5 ± 7.8%; n = 6, p = 0.022; Fig. 1E), due to the disinhibition of synaptic transmission by eliminating the tonic activation of inhibitory A₁R by endogenous adenosine (e.g., [61]. As shown in Fig. 1E, F, DPCPX abrogated the ability of 300 μM inosine to depress synaptic transmission (Fig. 1E,n = 6, p = 0.458 vs. DPCPX alone) and LTP magnitude (Fig. 1F; n = 5, p = 0.104 vs. DPCPX alone). As an important control, we confirmed that two successive application of 300 μM inosine separated by 30 min caused a similar depression of synaptic transmission (29.9 ± 2.8% inhibition in the first application and 39.1 ± 3.4% inhibition in the second application; n = 4, p = 0.066).

Given that the A_2AR is the predominant adenosine receptor controlling hippocampal LTP (reviewed in [18], with inconsistent effects of A₃R [16, 48] and A_2AR control A₁R function in the hippocampus [42, 43], we further tested the impact of a supramaximal concentration of the selective A_2AR receptor SCH58261 (50 nM; see [58] on the inhibitory effects of inosine on synaptic plasticity. As shown in Fig. 1G, SCH58261 (50 nM) abrogated the ability of 300 μM inosine to significantly reduce LTP magnitude (n = 6, p = 0.311 vs. SCH58261 alone).

Overall, these findings indicate that the ability of inosine to affect synaptic transmission and plasticity is strictly dependent on the functioning of the two adenosine receptors — A₁R and A_2AR — with more evident effects on the control of synaptic transmission and plasticity (reviewed in [29].

Impact of inosine on the binding of radioligands to adenosine receptors

To test the ability of inosine to directly bind to adenosine receptors, we first carried out radioligand binding studies using cell membrane preparations of cells heterologously expressing the different human adenosine receptor subtypes or membrane preparations obtained from the rat cerebral cortex, where A₁R are predominant, and from rat striatum, where A_2AR is most abundant (e.g., [57]. Agonist rather than antagonist radioligands were employed in order to label the high-affinity conformation for agonists. As can be seen in Table 1, inosine was only able to displace binding to either human or rat A₁R, albeit with very low potency (K_i approximately 100 µM) and was essentially ineffective in displacing binding to human or rat A_2AR and A_2BR within the estimated concentrations of inosine occurring in the brain (n = 3–4). Likewise, it showed no affinity for the human A₃R, but significant affinity for the rat A₃R (K_i value of 1.37 ± 1.08 µM; n = 3) (Fig. 2A).

Table 1.

Affinity of inosine at human and rat adenosine receptor subtypes determined in radioligand binding assays

Ki (µM)a (or percent inhibition at indicated concentration)
A1 vs. [3H]CCPA	A2A vs. [3H]CGS21680	A2B vs. [3H]NECA	A3 vs. [3H]NECA
Human 94 ± 18 µM Rat 101 ± 22 µM	Human  > 300 µM (6%) Rat  > 300 µM (15%)	Human  > 300 µM (40%)b Rat  > 300 µM (0%)	Human  > 300 µM (20%)c Rat 1.37 ± 1.08 µM

^aDetermined in 3 to 4 separate experiments. K_i values are shown in bold

^bVersus the antagonist radioligand [³H]PSB-603, the following data were determined: 0–4% inhibition at 300 µM (in the presence of ADA)

^cVersus [³H]PSB-11

Bold indicates values of potentially physiological significance

Fig. 2.

Ability of inosine to displace radioligand binding to rat A₃ adenosine receptors, but not to rat A₁ or A_2A adenosine receptors. A Concentration-dependent inhibition by inosine of the binding of the non-selective adenosine receptor ligand [³H]NECA to membranes of Chinese hamster ovary suspension (CHO-S) cells recombinantly expressing rat adenosine A₃ receptors. The data are mean ± SEM of 3 experiments, yielding a K_i value 1.37 µM. B Concentration-dependent inhibition by inosine of the binding of the adenosine A₁ receptor selective antagonist [³H]DPCPX to membranes of synaptosomes from the rat cerebral cortex. The data are mean ± SEM of 4 experiments, yielding a K_i value 108 μM. C Concentration-dependent inhibition by inosine of the binding of the adenosine A_2A receptor selective antagonist [³H]SCH58261 to membranes of synaptosomes from the rat cerebral cortex. The data are mean ± SEM of 4 experiments, yielding a K_i value 200 μM. Note that the displayed data only allow a direct estimation of the IC₅₀ values of inosine but not of the K_i values, which require computing the concentration of the radioligand and is K_D for the receptor

Given that synaptic transmission and plasticity are controlled by synaptic A₁R and A_2AR, we next assessed the ability of inosine to compete for A₁R and A_2AR binding in synaptic membranes. In the absence of competitors, the specific binding of the selective A₁R antagonist, [³H]DPCPX (2 nM), to cerebrocortical synaptosomal membranes was 1230 ± 30 fmol/mg protein (n = 4) and that of the selective A_2AR antagonist, [³H]SCH58261 (2 nM), was 42.2 ± 4.2 fmol/mg protein (n = 4). Inosine displayed low affinity to displace binding to synaptic A₁R and A_2AR, with estimated K_i values of 108 μM (95% confidence interval of 84–138 μM, n = 4) and of 200 μM (95% confidence interval of 130–307 μM, n = 4), respectively (Fig. 2B, C).

Involvement of endogenous extracellular adenosine in the ability of inosine to control synaptic transmission and plasticity

In order to unravel if the effects of inosine were due to a control of the levels of extracellular adenosine tonically activating adenosine receptors, we tested the effects of inosine in the presence of a supramaximal concentration of adenosine deaminase (ADA; 2 U/mL) to eliminate extracellular adenosine levels [17]. As shown in Fig. 3A–C, ADA (2 U/mL) caused the expected increase of fEPSP slopes (31.8 ± 5.0%; n = 6, p = 0.001; Fig. 3A) and abrogated the ability of 300 μM inosine to depress synaptic transmission (Fig. 3A; n = 6, p = 0.753 vs. ADA alone) and LTP magnitude (Fig. 3B, C; n = 4–5, p = 0.156 vs. ADA alone).

Fig. 3.

The inhibitory effect of inosine on synaptic transmission and plasticity in mouse hippocampal slices is strictly dependent on extracellular adenosine, likely released through equilibrative nucleoside transporters. A The removal of extracellular adenosine by superfusing adenosine deaminase (ADA; 2 U/mL) prevented the ability of 300 μM inosine to decrease field excitatory post-synaptic potentials (fEPSPs) in Schaffer fiber-CA1 pyramid synapses of mouse hippocampal slices. B Likewise, inosine did not significantly modify the magnitude of long-term potentiation (LTP) induced by a high-frequency train (100 Hz for 1 s) applied at time 0, as quantified in (C). D The blockade of equilibrative nucleoside transporters with the simultaneous presence of NBTI (5 µM) and dipyridamole (10 µM) eliminated the ability of 300 μM inosine to inhibit synaptic transmission. E Likewise, inosine did not significantly modify LTP magnitude, as quantified in (C). Data are mean ± SEM of 4–6 determination (independent mice). ns: non-significant with paired Student’s t test

It should be kept in mind that the basal extracellular levels of adenosine within hippocampal excitatory synapses have been estimated to be in the range of 0.14–4 μM [17, 21, 23], making it unlikely that the full conversion of extracellular adenosine by exogenously added ADA would generate an amount of extracellular inosine able to occlude any eventual effect of exogenously added inosine, especially since we observed that a concentration of inosine (30 μM, which is over 10 times greater than that estimated for extracellular adenosine) was devoid of effects on synaptic transmission (see Fig. 3B). Therefore, the ability of ADA to eliminate the effects of inosine indicates that the ability of inosine to affect synaptic transmission and plasticity is mostly dependent on its ability to control of the extracellular levels of adenosine, thereby indirectly affecting adenosine receptor activation.

Role of nucleoside transporter-derived extracellular adenosine in the ability of inosine to control synaptic transmission and plasticity

Since equilibrative nucleoside transporters (ENTs) use both inosine and adenosine as substrates [54], the most parsimonious explanation to understand the strict dependency on extracellular adenosine for inosine to control synaptic transmission and plasticity would be that the exposure to increase extracellular levels of inosine would drive an exchange with adenosine through ENTs. To test this hypothesis, we studied the effects of inosine in the simultaneous presence of effective concentrations (see [2] of an ENT-1 inhibitor, NBTI (5 μM), and a mixed ENT1/2 inhibitor dipyridamole (10 μM). As shown in Fig. 3D, E, the simultaneous presence of NBTI and dipyridamole abrogated the ability of 300 μM inosine to depress synaptic transmission (Fig. 3D,n = 4, p = 0.121 vs. NBTI + dipyridamole alone) and LTP magnitude (Fig. 3E, F,n = 5, p = 0.229 vs. NBTI + dipyridamole alone).

Discussion

The present study shows that inosine does not appear to directly engage neuronal A₁R or A_2AR, but instead it indirectly modifies the tonic activation of these adenosine receptors through an ENT-mediated alteration of the extracellular levels of adenosine. In fact, we observed that the selective blockade of A₁R and A_2AR blunted the ability of inosine to affect synaptic transmission and plasticity, indicating that these receptors are critically involved in the ability of inosine to affect neuronal function in hippocampal slices. Moreover, the removal of extracellular adenosine with ADA abrogated the effects of inosine on synaptic transmission and plasticity, showing that inosine does not directly engage A₁R and/or A_2AR, but strictly requires extracellular adenosine to modulate synaptic transmission and plasticity. Finally, the observation that the effects of inosine are blunted by the blockade of ENTs through the combined use of NBTI and dipyridamole shows that inosine affects synaptic transmission and plasticity via an ENT-mediated modification of the extracellular levels of adenosine controlling adenosine receptor-mediated neuromodulation. This adenosine-mediated control of synaptic transmission in hippocampal synapses is mostly operated by presynaptic adenosine receptors [43], although the engagement of post-synaptic adenosine receptors may also play a role in fine-tuning adaptive changes of the adenosine modulation system upon repeated recruitment [13]. Notably, we observed an increased variability in synaptic transmission upon washout of inosine, which is also abrogated by ADA and by inhibitors of ENTs, hinting at an eventual alteration of the bioavailability of adenosine receptors that remains to be investigated. This further indicates that different effects of inosine on synaptic function seem all dependent on extracellular adenosine released through ENTs.

The likeliness of indirect effects of inosine mediated by ENT-derived control of the extracellular levels of adenosine to control synaptic transmission and plasticity is further supported by our binding studies evaluating the influence of inosine on the binding of radioligands to adenosine receptor subtypes. In fact, we observed that inosine could only consistently affect the binding to A₁R, but with a low potency of around 100 μM for either heterologously expressed human A₁R or rat brain A₁R, in agreement with previous reports by others obtained in different preparations [28, 53, 56, 69]. In contrast, inosine poorly displaced the binding of selective A_2AR ligands to human A_2AR or to rat brain A_2AR, although it could displace binding to synaptic A_2AR with a low potency. This is in agreement with some reports [6], but in contrast to theoretical predictions [68], and these differences may be due to an eventual impact on the binding properties of A_2AR upon its homomerization and heteromerization (reviewed in [27] and interaction with different intracellular partners (reviewed in [18]. Finally, whereas it seems that inosine does not significantly displace A_2BR binding in spite of previous suggestions [22], the present observation indicates that inosine fails to affect binding to human A₃R but displays affinity for rodent A₃R. This is consonant with findings that inosine may displace A₃R binding and directly activate A₃R in different peripheral preparations [15, 28, 31, 33, 35, 37, 65]. However, it should be considered that A₃R displays the largest inter-species pharmacological differences among adenosine receptors [50] and its distribution, subcellular localization, and function in the brain are still poorly studied, with discrete effects on synaptic transmission and plasticity in healthy rodents [12, 24, 44].

Although the present findings exclude the hypothesis that inosine may directly activate adenosine receptors to control hippocampal synaptic transmission and plasticity, they do not discard the possibility that inosine may eventually affect non-synaptic adenosine receptors to afford brain neuroprotection. Brain dysfunction mostly involves a deregulation of synaptic function at its onset [67], but there is abundant evidence that non-synaptic mechanisms can precipitate synaptic dysfunction to trigger or amplify brain dysfunction [49]. Accordingly, adenosine-mediated control of neurodegeneration can be mostly accounted by the alteration of A₁R and A_2AR controlling synaptic function [18], but there is evidence that non-neuronal adenosine receptors might also affect neurodegeneration [1]. In particular, although the role of A₃R in the control of neurodegeneration is still unclear [11, 14, 26, 40, 59, 66], it is possible that inosine may directly engage non-neuronal A₃R, namely, glial A₃R (see [9, 36], A₃R in myeloid/lymphoid cells (e.g., [7, 32], or A₃R in mast cells [70], to affect neurodegeneration in rodents. However, the assumption of this A₃R-mediated mechanism of inosine neuroprotection requires excluding the presently reported ability of inosine to increase the extracellular levels of adenosine likely through competitive inhibition of ENTs’ activity, which emerges as the most likely mechanism of inosine-mediated neuroprotection.

In conclusion, the present study shows that inosine does not directly affect neuronal A₁R and A_2AR but rather indirectly controls adenosine-mediated neuromodulation by altering ENT-mediated control of the extracellular levels of adenosine. Although the present study does not rule out the possibility that inosine may directly affect non-synaptic adenosine receptors to afford neuroprotection or other non-synaptic responses, it clearly indicates that the proposal for the involvement of a direct modulation by inosine of adenosine receptor function requires excluding the involvement of an indirect effect on adenosine receptors resulting from inosine-induced adenosine outflow through ENTs. It is hoped that future studies with ENT1 and/or ENT2 knockout mice may be helpful to further confirm this hypothesis.

Author contribution

P. V., J. P. L., and C. R. L. carried out the electrophysiological experiments and analyzed the data; S. H. and C. V. carried out the receptor binding studies and analyzed the data; and J. P. L., R. A. C., and C. E. M. supervised the project and wrote the manuscript.

Funding

This study was supported by La Caixa Foundation (HP17/00523), Centro 2020 (CENTRO-01–0145-FEDER-000008: BrainHealth2020 and CENTRO-01–0246-FEDER-000010), FCT (POCI-01–0145-FEDER-03127, UIDB/04539/2020, and IF/01492/2015), and Deutsche Forschungsgemeinschaft (FOR2685, SFB1328).

Data availability

Data can be made available upon reasonable request.

Declarations

Ethics approval

This study was approved by the ORBEA_128_2015/04122015 and certified by Direção Geral de Alimentação e Veterinária (DGAV; 0421/000/000/2016 Ref. 014420).

Conflict of interest

RAC is a scientific consultant for the Institute for Scientific Information on Coffee (ISIC). All other authors declare no conflict of interests.