Abstract
Striatal adenosine A1 receptor (A1R) activation can inhibit dopamine release. A1Rs on other striatal neurons are activated by an adenosine tone that is limited by equilibrative nucleoside transporter 1 (ENT1) that is enriched on astrocytes and is ethanol sensitive. We explored whether dopamine release in nucleus accumbens core is under tonic inhibition by A1Rs, and is regulated by astrocytic ENT1 and ethanol. In ex vivo striatal slices from male and female mice, A1R agonists inhibited dopamine release evoked electrically or optogenetically and detected using fast-scan cyclic voltammetry, most strongly for lower stimulation frequencies and pulse numbers, thereby enhancing the activity-dependent contrast of dopamine release. Conversely, A1R antagonists reduced activity-dependent contrast but enhanced evoked dopamine release levels, even for single optogenetic pulses indicating an underlying tonic inhibition. The ENT1 inhibitor nitrobenzylthioinosine reduced dopamine release and promoted A1R-mediated inhibition, and, conversely, virally mediated astrocytic overexpression of ENT1 enhanced dopamine release and relieved A1R-mediated inhibition. By imaging the genetically encoded fluorescent adenosine sensor [GPCR-activation based (GRAB)-Ado], we identified a striatal extracellular adenosine tone that was elevated by the ENT1 inhibitor and sensitive to gliotoxin fluorocitrate. Finally, we identified that ethanol (50 mm) promoted A1R-mediated inhibition of dopamine release, through diminishing adenosine uptake via ENT1. Together, these data reveal that dopamine output dynamics are gated by a striatal adenosine tone, limiting amplitude but promoting contrast, regulated by ENT1, and promoted by ethanol. These data add to the diverse mechanisms through which ethanol modulates striatal dopamine, and to emerging datasets supporting astrocytic transporters as important regulators of striatal function.
SIGNIFICANCE STATEMENT Dopamine axons in the mammalian striatum are emerging as strategic sites where neuromodulators can powerfully influence dopamine output in health and disease. We found that ambient levels of the neuromodulator adenosine tonically inhibit dopamine release in nucleus accumbens core via adenosine A1 receptors (A1Rs), to a variable level that promotes the contrast in dopamine signals released by different frequencies of activity. We reveal that the equilibrative nucleoside transporter 1 (ENT1) on astrocytes limits this tonic inhibition, and that ethanol promotes it by diminishing adenosine uptake via ENT1. These findings support the hypotheses that A1Rs on dopamine axons inhibit dopamine release and, furthermore, that astrocytes perform important roles in setting the level of striatal dopamine output, in health and disease.
Introduction
Striatal dopamine (DA) axons are major strategic sites for striatal neuromodulators to influence DA output (Rice et al., 2011; Sulzer et al., 2016; Nolan et al., 2020; Roberts et al., 2021). Adenosine acts at A1 receptors (A1Rs) and A2ARs on diverse neurons in striatum, and exogenous activation of striatal A1Rs, but not A2Rs, inhibit evoked DA release (Okada et al., 1996; O'Neill et al., 2007; O'Connor and O'Neill, 2008; Ross and Venton, 2015). Immunocytochemical studies of rat synaptosomes suggest that A1Rs can be localized directly to striatal DA axons (Borycz et al., 2007), but definitive evidence is lacking and other intermediary inputs to DA axons have not been excluded. In nucleus accumbens core (NAcC), adenosine can provide tonic A1R-mediated inhibition of glutamatergic and GABAergic transmission (Brundege and Williams, 2002; Adhikary and Birdsong, 2021). DA release in NAcC and caudate–putamen has been suggested to be under tonic A1R-mediated inhibition, as A1R antagonists increase extracellular DA levels in rats in vivo measured by microdialysis (Okada et al., 1996; Solinas et al., 2002; Quarta et al., 2004a, b; Borycz et al., 2007), but whether these effects were direct and local or involved intact long-loop circuits was not resolved. Furthermore, adenosine acts in a frequency-dependent manner on glutamate and GABA transmission across other nuclei, resulting in a stronger A1R-dependent inhibition of neurotransmission elicited by low-frequency versus high-frequency electrical stimulations (e.g., neocortex: Yang et al., 2007; Qi et al., 2017; Perrier et al., 2019; hippocampus: Moore et al., 2003; calyx of Held: Wong et al., 2006). It has not yet been established whether striatal A1Rs simply inhibit DA neurotransmission, or might also promote contrast in DA signals released by different firing rates, and whether in turn this is independent from striatal acetylcholine (ACh) or GABA circuits, which modify DA signal contrast (Rice and Cragg, 2004; Lopes et al., 2019).
Extracellular adenosine concentrations are limited by the activity of nucleoside transporters, most notably the equilibrative nucleoside transporter type 1 (ENT1; Young et al., 2008; Nguyen et al., 2015). ENT1 is expressed in striatum (Anderson et al., 1999; Jennings et al., 2001) and is especially abundant on astrocytes (Peng et al., 2005; Chai et al., 2017). ENT1 on astrocytes can regulate adenosine signaling in striatum and elsewhere (Nagai et al., 2005; Tanaka et al., 2011; Boddum et al., 2016; Cheffer et al., 2018; Hong et al., 2020), and in dorsal striatum, astrocytic ENT1 activity modulates reward-seeking behaviors (Hong et al., 2020; Kang et al., 2020). We recently identified that astrocytes regulate DA release, because of their expression of GABA transporters that regulate tonic GABAergic inhibition (Roberts et al., 2020), but whether ENT1 on astrocytes also modulates DA output by regulating adenosine tone at A1Rs has not been explored. Additionally, adenosine uptake by ENT1 is impaired by acute ethanol, which augments extracellular adenosine levels (Nagy et al., 1989, 1990; Choi et al., 2004) and contributes to the ataxic and hypnotic effects of ethanol through the activation of striatal A1Rs (Meng and Dar, 1995; Phan et al., 1997; Dar, 2001). Acute ethanol reduces DA release evoked in NAcC in brain slices (Yorgason et al., 2014, 2015), as does chronic intermittent ethanol exposure in mice (Karkhanis et al., 2015; Rose et al., 2016), raising the question of whether ethanol regulates DA by promoting A1R-mediated inhibition of DA release.
Here, we assessed A1R regulation of DA release in NAcC by detecting DA in real time using fast-scan cyclic voltammetry and detecting striatal adenosine using the genetically encoded fluorescent adenosine sensor [GPCR-activation based (GRAB)-Ado; Peng et al., 2020; Wu et al., 2020]. We reveal that A1Rs can tonically inhibit DA output though an ambient adenosine tone, and that A1Rs additionally regulate the activity sensitivity of DA release, independently from striatal acetylcholine or GABA inputs. Furthermore, we find that tonic A1R-mediated inhibition of DA release is regulated by adenosine uptake by ENT1 on astrocytes, and is dysregulated by ethanol.
Materials and Methods
Animals.
All procedures were performed in accordance with the Animals in Scientific Procedures Act 1986 (Amended 2012) with ethical approval from the University of Oxford, and under authority of a Project License granted by the UK Home Office. Experiments were conducted using male and female adult (6- to 12-week-old) C57BL/6J mice (Charles River) or heterozygote knock-in mice bearing an internal ribosome entry site (IRES)-linked Cre recombinase gene downstream of the gene Slc6a3, which encodes the plasma membrane DA transporter (Slc6a3IRES-Cre mice; B6.SJL-Slc6a3tm1.1(cre)Bkmn/J, stock #006660, The Jackson Laboratory) maintained on a C57BL/6J background. All mice were group housed and maintained on a 12 h light/dark cycle (light ON from 7:00 A.M. to 7:00 P.M.) with ad libitum access to food and water. Data from male and female mice were combined throughout as no differences in the effects of adenosine A1 receptor agonists or antagonists between sexes were observed (Table 1).
Comparison of effects of adenosine A1 receptor agonists and antagonists on evoked mean peak [DA]o between sexes
Stereotaxic intracranial injections.
Mice were anesthetized with isoflurane and placed in a small animal stereotaxic frame (David Kopf Instruments). After exposing the skull under aseptic techniques, a small burr hole was drilled and adeno-associated viral solutions were injected at an infusion rate of 100 nl/min with a 32 gauge syringe (Hamilton) using a microsyringe pump (World Precision Instruments) and were withdrawn 5 min after the end of injection. To virally express channelrhodopsin-2 (ChR2) selectively in DA neurons, 600 nl/hemisphere AAV5-EF1α-DIO-hChR2(H134R)-eYFP (8 × 1012 genome copies/ml; UNC Vector Core Facility) encoding Cre-dependent ChR2 was injected bilaterally into the midbrain [anteroposterior (AP), −3.1 mm; mediolateral (ML), ±1.2 mm from bregma; dorsoventral (DV), −4.25 mm from exposed dura mater] of 6-week-old Slc6a3IRES-Cre mice, following previously described methods (Roberts et al., 2020). To overexpress ENT1 in striatal astrocytes, 600 nl/hemisphere AAV5-GfaABC1D-mENT1/mCherry-WPRE (1.6 × 1013 genome copies/ml; VECTOR BIOLABS) encoding a fluorescence protein-fused and functional ENT1 under the astrocyte-specific abbreviated glial fibrillary acidic protein promoter (GfaABC1D; Hong et al., 2020; Jia et al., 2020) or AAV5-GfaABC1D-mCherry-WPRE (6.1 × 1012 genome copies/ml; ETH Zurich Viral Vector Facility) for control fluorophore expression, was injected bilaterally into nucleus accumbens (AP, +1.3 mm; ML, ±1.2 mm from bregma; DV, −3.75 mm from exposed dura mater) of 6-week-old C57BL/6J wild-type mice. To virally express the fluorescent adenosine reporter GRAB-Ado (Peng et al., 2020; Wu et al., 2020), 800 nl/hemisphere AAV9-hSyn-GRAB-Ado1.0m (0.5 × 1013 genome copies/ml; WZ Biosciences) was injected bilaterally into nucleus accumbens or dorsal striatum (AP, +0.8 mm; ML, ±1.75 mm from bregma; DV, −2.40 mm from exposed dura mater) of 6-week-old C57BL/6J wild-type mice. Mice were used for experiments 2−4 weeks post-intracranial injection.
Brain slice preparation.
Acute brain slices were obtained from 8- to 12-week-old mice using standard techniques. Mice were culled by cervical dislocation within 1−2 h after the start of the light-ON period of the light cycle, and brains were dissected out and submerged in ice-cold cutting solution containing the following (in mm): 194 sucrose, 30 NaCl, 4.5 KCl, 1 MgCl2, 26 NaHCO3, 1.2 NaH2PO4, and 10 d-glucose. Coronal slices 300 μm thick containing striatum were prepared from dissected brain tissue using a vibratome (model VT1200S, Leica Microsystems) and transferred to a holding chamber containing artificial CSF (aCSF) containing the following (in mm): 130 NaCl, 2.5 KCl, 26 NaHCO3, 2.5 CaCl2, 2 MgCl2, 1.25 NaH2PO4, and 10 glucose. Sections were incubated at 34°C for 15 min before they were stored at room temperature (20–22°C) until recordings were performed. All recordings were obtained within 8 h of slicing. All solutions were saturated with 95% O2/5% CO2. Before recording, individual slices were hemisected, transferred to a recording chamber, and superfused at ∼2.5–3.0 ml/min with aCSF at 31–33°C.
Fast-scan cyclic voltammetry.
Evoked extracellular DA concentration ([DA]o) was measured in acute coronal brain slices using fast-scan cyclic voltammetry (FSCV) at carbon fiber microelectrodes (diameter, 7–10 μm) fabricated in-house (tip length, 70–120 μm) as used previously (Roberts et al., 2020). In brief, a triangular voltage waveform was scanned across the microelectrode (−700 to +1300 mV vs Ag/AgCl reference) at 800 V/s and at a scan frequency of 8 Hz using a Millar Voltammeter (Julian Millar, Barts and the London School of Medicine and Dentistry, London, UK). Microelectrodes were calibrated post hoc in 2 μm DA in each experimental solution. Microelectrode sensitivity to DA was between 10 and 40 nA/μm. Signals were attributed to DA because of the potentials of their characteristic oxidation (500–600 mV) and reduction (−200 mV) peaks. Currents at the oxidation peak potential were measured from the baseline of each voltammogram and plotted against time to provide profiles of [DA]o versus time. Recordings were conducted in the NAcC, within ∼100 μm of the anterior commissure, one site per slice. Electrical or light stimuli were delivered at 2.5 min intervals, which allow stable DA release to be sustained at ∼90–95% of original levels over the typical time course of experiments, in control conditions (Roberts et al., 2020). In experiments where [DA]o was evoked by electrical stimulation, a local bipolar concentric Pt/Ir electrode (inner diameter, 25 μm; outer diameter, 125 μm; FHC) was placed ∼100 μm from the recording microelectrode and stimulus pulses (duration, 200 μs) were given at 0.6 mA. We applied either a single pulse (1p) or trains of 2–20 pulses at 10–100 Hz. In experiments where [DA]o was evoked by light stimulation in slices prepared from Slc6a3IRES-Cre mice expressing ChR2, DA axons in NAcC were activated by TTL (transistor–transistor logic)-driven (Multi Channel Stimulus II, Multi Channel Systems) brief pulses (2 ms) of blue light (470 nm; 5 mW/mm2; OptoLED, Cairn Research), which illuminated the field of view (diameter, 2.2 mm; 10× water-immersion objective). Data were digitized at 50 kHz using a Digidata 1550A digitizer (Molecular Devices). Data were acquired and analyzed using Axoscope 11.0 (Molecular Devices) and locally written VBA scripts in Microsoft Excel (2013).
GRAB-Ado imaging.
An Olympus BX51WI microscope equipped with a 470 nm OptoLED light system (Cairn Research), Iris 9 Scientific CMOS camera (Teledyne Photometrics), 525/50 nm emission filter (Cairn Research), and 10×/0.3 numerical aperture water-immersion objective (Olympus) was used for wide-field fluorescence imaging of GRAB-Ado in striatal slices. Image acquisition was controlled using Micro-Manager 1.4. Electrical stimulation, LED light, and image acquisition were synchronized using TTL-driven stimuli via Multi Channel Stimulus II (Multi Channel Systems). Image files were analyzed with MATLAB R2017b and Fiji 1.5. For experiments measuring changes to basal, nonstimulated extracellular adenosine levels, images (exposure duration, 100 ms) were acquired every 30 s during brief exposure (1 s) to blue light (470 nm; 1 mW/mm2; OptoLED, Cairn Research) for a 40 min window. We extracted fluorescence intensity from the region of interest (ROI; 150 × 150 μm) and derived a background-subtracted fluorescence intensity (Ft) by subtracting background fluorescence intensity from an equal-sized ROI where there was no GRAB-Ado expression (i.e., cortex). Data are expressed as a change in fluorescence (ΔF/F0) and were derived by calculating [(Ft – F0)/F0], where F0 is the average Ft of the first 20 acquired images (initial 10 min). For experiments measuring extracellular adenosine levels in response to trains of electrical stimulation or experiments calibrating GRAB-Ado signals to applications of known concentrations of exogenous adenosine, images were acquired at 10 Hz (exposure duration, 100 ms) during continuous blue light (470 nm; 1 mW/mm2; OptoLED, Cairn Research) for a 4 min recording window. For evoked adenosine release, electrical stimulus pulses (100 pulses at 50 Hz; pulse duration, 200 μs; 0.6 mA) were given by a local bipolar concentric Pt/Ir electrode (inner diameter, 25 μm; outer diameter, 125 μm; FHC) at the 30 s time point. Bath applications of exogenous adenosine were also applied at the 30 s time point. Ft was extracted from an ROI (150 × 150 μm) ∼50 μm from the stimulating electrode, and ΔF/F0 was derived by calculating [(Ft – F0)/F0], where F0 is the average fluorescence intensity over the 10 s window (100 images) before onset of electrical stimulation or bath application of exogenous adenosine.
Drugs.
Adenosine (25–100 μm), (+)-bicuculline (10 μm), dihydro-β-erythroidine hydrobromide (DHβE; 1 μm), nitrobenzylthioinosine (NBTI; 10 μm), and tetrodotoxin (TTX; 1 μm) were obtained from Tocris Bioscience. CGP 55845 hydrochloride (CGP; 4 μm), 8-cyclopentyl-1,3-dimethylxanthine (CPT; 10 μm), and dipropylcyclopentylxanthine (DPCPX; 2 μm) were obtained from Abcam. Ethanol (EtOH; 50 mm), and N6-cyclopentyladenosine (CPA; 15 μm) were obtained from Sigma-Aldrich. Fluorocitrate (FC) was prepared as previously described (Paulsen et al., 1987; Roberts et al., 2020). In brief, d,l-fluorocitric acid Ba3 salt (Sigma-Aldrich) was dissolved in 0.1 m HCl, and the Ba2+ was precipitated with 0.1 m Na2S04 and then centrifuged at 1000 × g for 5 min. Supernatant containing fluorocitrate was used at a final concentration of 100 μm for experimentation.
Immunocytochemistry.
To confirm viral ENT1-mCherry expression in non-neuronal cells, direct mCherry fluorophore expression was compared with indirect immunofluorescence for neuronal marker NeuN. ENT1-mCherry-expressing mice were anesthetized with an overdose of pentobarbital and was transcardially perfused with PBS, followed by 4% paraformaldehyde (PFA) in 0.1 m phosphate buffer, pH 7.4. Brains were removed and postfixed overnight in 4% PFA. Coronal sections were cut using a vibrating microtome (model VT1000S, Leica) at a thickness of 50 µm and collected in a 1 in 4 series. Sections were stored in PBS with 0.05% sodium azide until processing. Upon processing, sections were washed in PBS and then blocked for 1 h in a solution of PBS Triton X-100 (0.3%; PBS-Tx) containing 10% normal donkey serum (NDS). Sections were then incubated in primary antibodies overnight in PBS-Tx with 2% NDS at 4°C. The primary antibody used was rabbit anti-NeuN (1:500; catalog #R-3770–100, BIOSENSIS). Sections were then incubated in species-appropriate fluorescent secondary antibodies with minimal cross-reactivity for 2 h in PBS-Tx with 2% NDS at room temperature. The secondary antibody used was donkey anti-rabbit Alexa Fluor 405 (1:500; catalog #ab175651, Abcam). Sections were washed in PBS and then mounted on glass slides and coverslipped using Vectashield (Vector Laboratories). Coverslips were sealed using nail varnish and stored at 4°C. Confocal images were acquired with a confocal laser-scanning microscope (model FV3000, Olympus) using a 20× and 40× objective and filters for appropriate excitation and emission wave lengths (Olympus Medical).
Results
Adenosine A1Rs inhibit striatal DA release
Previous studies have shown that agonists for striatal A1Rs inhibit DA release evoked by discrete electrical stimulations in dorsal and ventral striatum (O'Neill et al., 2007; O'Connor and O'Neill, 2008; Ross and Venton, 2015). We first corroborated these observations in NAcC for DA release evoked by single stimulus pulses and detected using FSCV in ex vivo coronal slices (Fig. 1A). Bath application of A1R agonist CPA (15 μm) reduced evoked [DA]o by ∼50% (Fig. 1B; t(5) = 15.2, p < 0.0001, paired t test). This effect of CPA was prevented by prior application of A1R antagonist DPCPX [2 μm; Fig. 1B; F(1,11) = 88.6, p < 0.0001, two-way repeated-measures (RM) ANOVA for the effect of CPA in the absence vs presence of DPCPX; Fig. 1E; t(11) = 9.99, p < 0.0001, unpaired t test], confirming an action via A1Rs.
Adenosine A1R agonists inhibit DA release in NAcC. A, Experimental setup for FSCV for DA detection in wild-type mouse NAcC in acute slices. Inset, Typical evoked DA voltammogram. B–D, Left, Summary of mean peak [DA]o evoked by a 1p before and after application of A1R agonist CPA (15 μm) normalized to predrug baseline level (dotted line). Shaded areas represent time points used to obtain data on right, mean [DA]o transients in either drug-free ACSF (blue n = 6 experiments/4 mice), or in the presence of A1R antagonist DPCPX (2 μm; green, n = 8 experiments/5 mice; B) or nAChR antagonist DHβE (1 μm; dark blue, n = 5 experiments/4 mice; C), or GABAA and GABAB antagonists (+)-bicuculline (10 μm) and CGP 55 845 (4 μm; purple, n = 5 experiments/4 mice; D). Control data in C and D are from B. E, Mean peak [DA]o following application of CPA (15 μm; as a percentage of predrug baseline, data are summarized from B–D). F, Viral injection into midbrain of Slc6a3IRES-Cre mice for the expression of ChR2-eYFP in DA axons for optogenetic-evoked DA release. G, As in B but [DA]o evoked optogenetically by 1p or 5p of light at 25 Hz (n = 8 experiments/5 mice). H, Ratio of peak [DA]o evoked by 5p:1p ratio at 25 Hz before (gray) and after (blue) application of CPA (15 μm). ***p < 0.001, paired t tests (B–D, G, H), unpaired t tests (E), and two-way repeated-measures ANOVA (B–D). Error bars indicate SEM.
Striatal cholinergic interneurons (ChIs) have A1Rs (Alexander and Reddington, 1989; Ferré et al., 1996; Preston et al., 2000; Song et al., 2000) through which A1R agonists can hyperpolarize ChIs and inhibit ACh release (Richardson and Brown, 1987; Brown et al., 1990; Preston et al., 2000). Because ChIs operate strong control over DA release via nAChRs (Jones et al., 2001; Rice and Cragg, 2004; Zhang and Sulzer, 2004) and can mediate the effects of other striatal neuromodulators on DA (Britt and McGehee, 2008; Hartung et al., 2011; Stouffer et al., 2015; Kosillo et al., 2016; Lemos et al., 2019), they might indirectly mediate the control of DA by A1Rs. We explored whether A1Rs can modulate striatal DA release in the absence of these potential actions on ChIs and nAChRs. We used nAChR antagonist DHβE (1 μm) to inhibit nAChRs, as described previously (Rice and Cragg, 2004; Threlfell et al., 2012), after which subsequent application of CPA nonetheless reduced [DA]o evoked by single stimulus pulses by ∼50% (Fig. 1C; t(4) = 15.9, p < 0.0001, paired t test), an effect not different from that seen in the absence of DHβE (Fig. 1C; F(1,9) = 1.22, p = 0.29, two-way RM ANOVA: main effect of drug; Fig. 1E; t(9) = 1.21, p = 0.259, unpaired t test), indicating that A1Rs can suppress DA release independently from any indirect effects via ChI inputs to nAChRs.
DA release throughout striatum is also under tonic inhibition by GABA through action at GABAA and GABAB receptors (Lopes et al., 2019; Roberts et al., 2020), and we tested whether A1Rs require this GABA input to inhibit DA release. However, in the presence of GABAA and GABAB receptor antagonists, bicuculline (10 μm) and CGP 55845 (5 μm), respectively, the A1R agonist CPA (15 μm) significantly reduced [DA]o evoked in NAcC by single electrical pulses by ∼40% (Fig. 1D; t(4) = 6.10, p = 0.0036, paired t test), an effect that was not significantly different from that seen in the absence of GABA receptor antagonists (Fig. 1D; F(1,9) = 0.019, p = 0.892, two-way RM ANOVA: main effect of drug; Fig. 1E; t(9) = 0.259, p = 0.802, unpaired t test), indicating that A1 receptor agonists can inhibit DA release independently from indirect effects acting via striatal GABA modulation of DA.
To further test whether the inhibition of DA release by A1Rs requires the coactivation of other striatal neuron types or inputs, we tested whether A1Rs modulate DA release evoked optogenetically by targeted light activation of DA axons. Optogenetic activation of DA axons minimizes the activation of other striatal neurons that occurs with nonselective electrical stimulation; optogenetically evoked DA release is independent of at least nAChR input (Threlfell et al., 2012; Melchior et al., 2015). We expressed ChR2-enhanced yellow fluorescent protein (eYFP) in DA neurons and axons in Slc6a3IRES-Cre mice using an established viral approach (Fig. 1F), as we previously described (Roberts et al., 2020). CPA suppressed [DA]o evoked in NAcC by single brief (2 ms) blue light pulses (Fig. 1G; t(7) = 16.2, p < 0.0001, paired t test), indicating that A1Rs can inhibit DA release independently from not just ChI input to nAChRs, but also without requiring the coincident activation of inputs from other neuron types, suggesting that A1R inhibition of DA release could be via direct action on DA axons, although actions via an unidentified tonically active input that is not GABA are possible.
Striatal A1Rs enhance the frequency and activity sensitivity of DA release
Previous studies exploring whether striatal A1Rs regulate DA release have predominantly focused on the effects on release evoked by single, discrete stimuli, and impact on release by the full range of physiologically relevant frequencies has not been explored. Adenosine acts in a frequency-dependent manner on glutamate and GABA transmission in other brain nuclei, resulting in a stronger A1R-dependent inhibition of neurotransmission elicited by low-frequency versus high-frequency electrical stimulations, and therefore operating like a high-pass input filter on neurotransmitter release (e.g., neocortex: Yang et al., 2007; Qi et al., 2017; Perrier et al., 2019; hippocampus: Moore et al., 2003; calyx of Held: Wong et al., 2006). Other inputs and mechanisms that inhibit DA release by single stimuli can also promote the frequency dependence of DA release, for instance GABA (Lopes et al., 2019; Roberts et al., 2020), consistent with a reduction in the initial release probability and a consequent relief of some short-term depression (Condon et al., 2019). Using optogenetic activation, we observed a sizeable enhancement in the 5p:1p ratio (for 25 Hz) of light-evoked [DA]o in the NAcC following application of CPA (Fig. 1H; t(7) = 10.4, p < 0.0001, paired t test), suggesting that striatal A1Rs might promote the sensitivity of DA release to firing rate.
We used electrical stimulation to investigate the effects of A1R activation on DA release across a full range of physiological DA neuron firing frequencies (1, 10, 50, and 100 Hz), with varying pulse numbers (5, 10, 15, and 20 p). These experiments were conducted in the presence of nAChR antagonist DHβE (1 μm) to eliminate the effects of ChIs at nAChRs, which profoundly restrict the apparent frequency and activity sensitivity of evoked DA release (Rice and Cragg, 2004; Zhang and Sulzer, 2004). A1R activation by CPA (15 μm) inhibited DA release inversely and significantly with stimulation frequency (Fig. 2A,B; F(3,18) = 53.9, p < 0.0001, one-way RM ANOVA), such that CPA significantly increased the frequency dependence of evoked DA release (Fig. 2C; F(1,7) = 38.2, p = 0.0005, two-way RM ANOVA: main effect of drug; F(3,21) = 23.0, p < 0.0001, frequency × drug interaction). In addition, A1R activation by CPA inhibited DA release in a manner that varied inversely with the number of pulses in a train (50 Hz; Fig. 2D,E; F(6,24) = 31.1, p < 0.0001, one-way RM ANOVA), which significantly increased the pulse-dependence of evoked DA release (Fig. 2F; F(1,7) = 46.7, p = 0.0002, two-way RM ANOVA: main effect of drug; F(4,28) = 28.2, p < 0.0001, pulse number × drug interaction). Therefore, striatal A1Rs do not simply inhibit DA release, but have a dynamic outcome depending on activity, that promotes the contrast in DA signals released by different activity.
Adenosine A1R agonists modify the frequency and activity sensitivity of DA release in NAcC. A, Mean [DA]o transients evoked by one or five electrical pulses in control conditions (gray) or in A1R agonist CPA (15 μm; blue) in NAcC (n = 7 experiments/5 mice). B, Mean peak [DA]o from A normalized to control conditions versus stimulation frequency. C, Mean peak [DA]o from A normalized to 1p in each condition versus stimulation frequency. D, Mean [DA]o transient evoked by 1–20 pulses (50 Hz) of electrical trains in control conditions (gray) or in A1R agonist CPA (15 μm; blue) in NAcC (n = 7 experiments/5 mice). E, Mean peak [DA]o from D normalized to control conditions versus pulse number. F, Mean peak [DA]o from D normalized to 1p in each condition versus number of pulses. *p < 0.001, one-way repeated-measures ANOVA with Sidak's multiple comparisons to 1p (B, E), two-way repeated-measures ANOVA with Sidak's multiple comparisons (C, F). Sigmoidal nonlinear curve fits. Error bars indicate SEM of DHβE (1 μm) present throughout.
Adenosine operates a tonic inhibition on DA release
We next addressed whether A1Rs provide an endogenous inhibition of DA release levels and support the contrast in DA signals released by different activity. Previous in vivo microdialysis studies indicate that endogenous adenosine might even exert a tonic A1R-mediated inhibition of DA release, as systemic dosing or intrastriatal infusions of A1R antagonists (as well as competitive A1 and A2A receptor antagonist caffeine; Solinas et al., 2002; Quarta et al., 2004a) increase extracellular striatal DA levels (Okada et al., 1996; Quarta et al., 2004b; Borycz et al., 2007). These effects might involve indirect actions via long loop circuits that are intact in vivo that could modulate DA neuron firing in midbrain. We tested therefore whether endogenous inhibition of DA release by A1Rs could be localized to the NAcC and whether it occurs without coactivation of other neurons, indicative of an tonic inhibition of DA release, by exploring the effects of A1R antagonists on DA release in NAcC evoked either electrically or optogenetically in coronal striatal slices.
Application of the A1R antagonist CPT (10 μm) significantly enhanced [DA]o evoked by single or five pulses (50 Hz) of electrical stimulation by ∼30–40% (Fig. 3A; 1p: t(6) = 5.10, p = 0.0022; 5p: t(6) = 5.13, p = 0.0022; paired t tests) and in turn, decreased the ratio of [DA]o evoked by the 5p:1p ratio (Fig. 3B; t(6) = 3.72, p = 0.0099, paired t test), indicative of an underlying regulation of DA release by endogenous adenosine. Caffeine (20 μm) also significantly enhanced [DA]o evoked by single or five pulses (50 Hz) of electrical stimulation by ∼20–10% (Fig. 3C; 1p: t(5) = 3.69, p = 0.014; 5p: t(5) = 3.49, p = 0.018; paired t tests) and decreased the ratio of [DA]o evoked by the 5p:1p ratio (Fig. 3D; t(5) = 3.59, p = 0.016, paired t test). We then investigated whether A1R-mediated inhibition of DA release by endogenous adenosine might arise through an ambient adenosine tone, by exploring whether A1R antagonist CPT could promote DA release when evoked using optogenetic stimulation to activate DA axons selectively without coactivation of other striatal cell types that might provide a source of endogenous adenosine. CPT significantly enhanced [DA]o evoked by single and five pulses (25 Hz) of light in Slc6a3IRES-Cre ChR2-expressing mice by ∼20–30% (Fig. 3E; 1p: t(7) = 5.58, p = 0.0008; 5p: t(7) = 5.51, p = 0.0009; paired t tests) and decreased the ratio of [DA]o evoked by the 5p:1p ratio (25 Hz; Fig. 3F; t(7) = 13.50, p < 0.0001, paired t test). These data therefore suggest that striatal DA release is under tonic inhibition by an ambient adenosine tone at A1Rs, which promotes contrast in DA signals released by different activity.
Adenosine A1R antagonists enhance DA release and correspondingly reduce the activity sensitivity of DA release in NAcC. A, Left, Summary of mean peak [DA]o before (gray) and after the application of A1R antagonist CPT (10 μm; green) normalized to predrug baseline (dotted line) and right, corresponding mean [DA]o transients evoked by 1p and 5p at 50 Hz in NAcC (n = 7 experiments/4 mice). Shaded areas are used to obtain illustrated transients of [DA]o and for statistical comparisons. B, Ratio of peak [DA]o evoked by 5p:1p ratio at 50 Hz before (gray) and after (blue) application of CPT (10 μm). Data summarized from A. C, D, As in A and B, but before (gray) and after (green) application caffeine (20 μm; n = 6 experiments/4 mice). E, F, As in A and B, but [DA]o evoked optogenetically by 1p or 5p of light at 25 Hz in Slc6a3IRES-Cre mice (n = 8 experiments/5 mice). *p < 0.05, **p < 0.01, ***p < 0.001, paired t tests. Error bars indicate SEM.
ENT1 on astrocytes is a regulator of tonic A1R-mediated inhibition of DA release
We next tested the hypothesis that striatal ENT1 and, by association, astrocytes, by governing ambient adenosine levels (Nagai et al., 2005; Young et al., 2008; Tanaka et al., 2011; Nguyen et al., 2015; Cheffer et al., 2018; Hong et al., 2020), might determine the level of tonic A1R-mediated inhibition of evoked DA release in NAcC. We first inhibited ENT1 activity by pretreatment of slices with the selective ENT1 inhibitor NBTI (10 μm, for 45–60 min), which has previously been shown to increase adenosine levels in striatum (Pajski and Venton, 2010). ENT1 inhibition itself attenuated [DA]o evoked by single and five pulses (50 Hz) of electrical stimulation (Fig. 4A; 1p: t(12) = 4.38, p = 0.0009; 5p: t(12) = 3.02, p = 0.0107; unpaired t tests). Furthermore, after pretreatment with NBTI, the A1R antagonist CPT (10 μm) enhanced [DA]o evoked by single electrical pulses to a significantly greater degree than in control slices (Fig. 4B; F(1,12) = 27.17, p = 0.0002, two-way RM ANOVA: main effect of drug). CPT decreased the ratio of [DA]o evoked by 5p:1p ratio (50 Hz) in both conditions (Fig. 4C; F(1,12) = 35.48, p < 0.0001, two-way RM ANOVA: main effect of drug), but there was a significant statistical interaction between NBTI and CPT (Fig. 4C; F(1,12) = 9.48, p = 0.0096, two-way RM ANOVA: NBTI × CPT interaction) borne out by a greater decrease in this ratio in slices pretreated with NBTI compared with control slices (Fig. 4D; t(12) = 3.08, p = 0.0096, unpaired t test). These data suggest that ENT1 regulates, and in particular limits, how A1 receptors tonically inhibit DA release and support its associated activity sensitivity.
Adenosine uptake by ENT1 on striatal astrocytes regulates tonic A1R-mediated inhibition of DA release. A, Mean [DA]o transients (left) and corresponding mean peak [DA]o (right) evoked by a 1p or 5p at 50 Hz in NAcC in the absence (gray; n = 6 experiments/5 mice) or the presence of ENT1 inhibitor NBTI (10 μm; orange; n = 8 experiments/5 mice). B, Summary of mean peak [DA]o before (gray) and after application of A1R antagonist CPT (10 μm) normalized to predrug baseline level (dotted line; left) and right, corresponding mean [DA]o transients evoked by 1p and 5p at 50 Hz in NAcC in the absence (green; n = 6 experiments/5 mice) or the presence of ENT1 inhibitor NBTI (10 μm; orange; n = 8 experiments/5 mice). C, D, Ratio of peak [DA]o evoked by 5p:1p ratio at 50 Hz (C) and Δ5p:1p ratio (D) before (gray) and after application of CPT (10 μm) in the absence (left, green) or presence of ENT1 inhibitor NBTI (10 μm; right, orange). Data are summarized from B. E, Delivery to NAcC of viral fluorescence-tagged ENT1 driven by an astrocytic promoter to test the impact of increased astrocyte-specific expression of ENT1 on DA release detected locally in NAcC. F, Immunofluorescence for mCherry (top, red), mCherry-tagged ENT1 (bottom, red), and neuronal marker NeuN (cyan) in NAcC. Scale bars, 40 µm. ac, Anterior commissure. G, Mean [DA]o transients (left) and corresponding mean peak [DA]o (right) evoked by a 1p or 5p at 50 Hz in NAcC in slices overexpressing ENT1 (ENT1-OX; red; n = 9 experiments/4 mice) or expressing control fluorophore mCherry (dark gray; n = 8 experiments/4 mice) under an astrocyte-specific promoter. H, Summary of mean peak [DA]o before (gray) and after the application of A1R antagonist CPT (10 μm) normalized to predrug baseline (dotted line; left) and right, corresponding mean [DA]o transients evoked by 1p and 5p at 50 Hz in NAcC in slices overexpressing ENT1 (ENT1-OX; red; n = 9 experiments/4 mice) or expressing control fluorophore mCherry (dark gray; n = 8 experiments/4 mice). I, J, Ratio of peak [DA]o evoked by 5p:1p ratio at 50 Hz (I) and Δ5p:1p ratio (J) before (light gray) and after application of CPT (10 μm) in slices overexpressing ENT1 (right, red) or expressing control fluorophore mCherry (left, dark gray). Data are summarized from H. p < 0.05, **p < 0.01, ***p < 0.001, unpaired t test (A, D, G, J), two-way repeated-measures ANOVA (B, C, H, I) with Sidak's multiple-comparisons test (C, I). Error bars indicate SEM.
We then tested conversely whether an upregulation of ENT1 specifically on astrocytes could diminish the level of endogenous A1R-mediated inhibition of DA release. We targeted fluorescence-tagged ENT1 to astrocytes in the NAcC (Fig. 4E,F) using a viral approach already validated for striatal astrocytes (Hong et al., 2020; Jia et al., 2020). Astrocyte-targeted ENT1 overexpression (ENT1-OX) increased [DA]o evoked by single and five pulses (50 Hz) of electrical stimulation compared with targeted mCherry controls (Fig. 4G; 1p: t(15) = 3.29, p = 0.0049; 5p: t(15) = 3.49, p = 0.0033; unpaired t tests). Furthermore, with ENT1-OX, the A1R antagonist CPT (10 μm) enhanced [DA]o evoked by single electrical pulses to a significantly lesser degree than in mCherry controls (Fig. 4H; F(1,15) = 20.12, p = 0.0004, two-way RM ANOVA: main effect of drug). CPT decreased the ratio of [DA]o evoked by 5p:1p (50 Hz) in ENT1-OX and mCherry controls (Fig. 4I; F(1,15) = 46.28, p < 0.0001, two-way RM ANOVA: main effect of drug), but there was a significant statistical interaction between ENT1-OX and CPT (Fig. 4I; F(1,15) = 9.54, p = 0.0075, two-way RM ANOVA: ENT1-OX × CPT interaction), which was borne out by a smaller decrease in this ratio following CPT application in ENT1-OX than in controls (Fig. 4J; t(15) = 3.09, p = 0.0075, unpaired t test). Together, these data suggest that ENT1 on astrocytes in particular can support adenosine uptake and set the level of inhibition and regulation of activity dependence of DA release by ambient adenosine acting at A1Rs.
To resolve directly whether ENT1, and astrocytes, govern striatal adenosine levels, we detected adenosine levels by imaging the recently developed GRAB-Ado sensor (Peng et al., 2020; Wu et al., 2020), a virally expressed genetic reporter, injected into striatum (Fig. 5A). We confirmed that striatal GRAB-Ado sensor fluorescence responded to adenosine concentrations in a concentration-dependent manner on application of exogenous adenosine to slices (Fig. 5B; F(3,35) = 28.73, p < 0.0001, one-way ANOVA), and with a large range of dF/F. To assess the impact of ENT1 on ambient adenosine tone in NAcC, we imaged GRAB-Ado fluorescence before and after ENT1 inhibition, in the absence of any stimulation. Bath application of ENT1 inhibitor NBTI (10 μm) increased fluorescence compared with vehicle controls (Fig. 5C; F(1,18) = 57.8, p < 0.0001, two-way RM ANOVA: main effect of NBTI), indicating that ENT1 limits extracellular adenosine levels. To test whether ENT1 on astrocytes participate in this mechanism, we tested whether metabolic inhibition of astrocytes limited ENT1 function. We pretreated slices with the gliotoxin fluorocitrate (100 μm, for 45–60 min), which has been established to induce metabolic arrest in astrocytes, render them inactive, and prevent the effects of astrocytic transporters (Paulsen et al., 1987; Henneberger et al., 2010; Bonansco et al., 2011; Boddum et al., 2016; Roberts et al., 2020). The effects of ENT1 inhibitor NBTI on GRAB-Ado fluorescence were occluded after pretreatment with fluorocitrate (Fig. 5D; F(1,14) = 14.32, p = 0.002, two-way RM ANOVA: main effect of FC).
Striatal adenosine tone is regulated by ENT1 activity on striatal astrocytes. A, Viral delivery of GRAB-Ado to striatum for imaging extracellular adenosine levels ex vivo in acute striatal slices. B, Peak ΔF/F0 in response to increasing concentrations of exogenously applied adenosine (0–100 μm; n = 9–11 slices/4–6 mice). C, Nonstimulated ΔF/F0 GRABAdo signal in NAcC normalized to before drug application (dotted line) in control conditions (dark gray, n = 10 experiments/5 mice) or before and after application of NBTI (10 μm; orange, n = 10 experiments/5 mice). D, As in C, but in the presence of gliotoxin fluorocitrate (100 μm; pink, n = 6 experiments/4 mice). E, Mean transients (left) and mean peak (right) ΔF/F0 GRABAdo signal evoked by 100 electrical pulses (50 Hz) in drug-free control conditions (n = 9 slices/5 mice) or the presence of NBTI (10 μm; orange; n = 7 experiments/5 mice), TTX (1 μm; blue; n = 9 experiments/5 mice), or fluorocitrate (100 μm; pink; n = 8 experiments/5 mice). F, Mean ΔF/F0 decay phase normalized to peak from E. *p < 0.05, **p < 0.01, ***p < 0.001, two-way repeated-measures ANOVA (C, D), and one-way ANOVA with Sidak's multiple-comparison test (E). Nonlinear regression with extra-sum-of-squares F test; τ = decay time constant (F). Error bars indicate SEM.
To further characterize and validate the role of ENT1 and astrocytes in striatal adenosine signaling, we explored their impact on the dynamics of extracellular adenosine transients evoked electrically by trains of electrical stimulation (100 pulses, 50 Hz; Fig. 5E). Evoked increases in extracellular adenosine concentrations exhibited relatively extended rise times (∼30 s) and clearance times, as reported previously in cultured hippocampal neurons and acute hippocampal and medial prefrontal cortex mouse brain slices (Wu et al., 2020), and surprisingly release was activated via a mechanism that was not prevented by Nav blocker TTX (1 μm). (Fig. 5E; F(3,29) = 13.60, p < 0.0001, one-way ANOVA; TTX vs control: p > 0.99, Sidak's multiple comparisons). In slices pretreated with the ENT1 inhibitor NBTI (10 μm, for 45–60 min), the peak of evoked adenosine levels was attenuated (Fig. 5E; F(3,29) = 13.60, p < 0.0001, one-way ANOVA; NBTI vs control: p = 0.001, Sidak's multiple comparisons), but the clearance time constant was extended (Fig. 5F; F = 8543, p < 0.0001, extra-sum-of-squares F test; control, τ = 108.4; NBTI, τ = 564.5), indicating reduced release and uptake, and seen previously in cultured hippocampal neurons (Wu et al., 2020). In slices pretreated with the gliotoxin fluorocitrate (100 μm, for 45–60 min), both the peak level of adenosine and the time constant for clearance were elevated (Fig. 5E; F(3,29) = 13.60, p < 0.0001, one-way ANOVA; FC vs control: p = 0.049, Sidak's multiple comparisons; Fig. 5F; F = 8543, p < 0.0001, extra-sum-of-squares F test; control, τ = 108.4; FC, τ = 217.0), indicating that ENT1 on astrocytes supports adenosine uptake.
EtOH increases tonic A1R-mediated inhibition of DA release
Our data indicate that astrocytic ENT1 function regulates tonic A1R-mediated inhibition of DA output in NAcC. Acute exposure to ethanol is documented to increase extracellular adenosine levels in many brain nuclei, including striatum, via inhibition of adenosine uptake by ENT1 (Nagy et al., 1989, 1990; Choi et al., 2004). Given that ethanol also attenuates evoked DA release in NAc (Yorgason et al., 2014, 2015; Karkhanis et al., 2015; Rose et al., 2016), we tested whether acute ethanol exposure might increase tonic A1R-mediated inhibition of DA release via impaired ENT1 function. We pretreated slices with ethanol (2–3 h), at a concentration (50 mm) that correlates to a blood alcohol concentration of 230 mg/dl in humans and is consistent with what late-stage alcoholics achieve (Brick and Erickson, 2009). We first confirmed that pretreating slices with ethanol using this paradigm reduced [DA]o evoked by single and five pulses (50 Hz) of electrical stimulation in NAcC (Fig. 6A; 1p: t(12) = 2.74, p = 0.018; 5p: t(12) = 2.67, p = 0.020; unpaired t tests). Next, we found that the effect of A1R antagonism with CPT (10 μm) on [DA]o evoked by single electrical pulses was elevated compared with control slices (Fig. 6B; F(1,12) = 18.04, p = 0.0011, two-way RM ANOVA: main effect of drug). A1 receptor antagonism with CPT decreased the ratio of [DA]o evoked by 5p:1p ratio (50 Hz; Fig. 6C; F(1,12) = 68.51, p < 0.0001, two-way RM ANOVA: main effect of drug), and there was a significant interaction between ethanol and CPT (Fig. 6C; F(1,12) = 9.36, p = 0.0099, two-way RM ANOVA: ethanol × CPT interaction), which was borne out by a more pronounced decrease after ethanol than in control slices (Fig. 6D; t(12) = 3.16, p = 0.0082, unpaired t test). To test directly whether ethanol in this paradigm impaired adenosine uptake by ENT1, we imaged tonic extracellular adenosine levels with the GRAB-Ado sensor during application of ENT1 inhibitor NBTI (10 μm). NBTI increased tonic extracellular adenosine levels in NAcC to a significantly lesser degree in slices preincubated with ethanol than in controls (Fig. 6E; F(1,17) = 11.30, p = 0.0037, two-way RM ANOVA: main effect of drug; t(17) = 2.60, p = 0.019, unpaired t test). These data indicate that tonic A1R-mediated inhibition of DA axons by adenosine in the NAcC is elevated by acute ethanol exposure, paralleled by an underlying attenuated uptake of adenosine uptake by ENT1.
EtOH enhances tonic adenosine A1R-mediated inhibition of DA release. A, Mean [DA]o transients (left) and corresponding mean peak [DA]o (right) evoked by 1p or 5p at 50 Hz in NAcC in the absence (gray; n = 7 experiments/4 mice) or presence of 50 mm EtOH (pink; n = 7 experiments/4 mice). B, Summary of mean peak [DA]o before (gray) and after application of A1R antagonist CPT (10 μm) normalized to predrug baseline level (dotted line; left) and right, corresponding mean [DA]o transients evoked by 1p and 5p at 50 Hz in NAcC in the absence (green, n = 7 experiments/4 mice) or presence of 50 mm EtOH (pink; n = 7 experiments/4 mice). C, D, Ratio of peak [DA]o evoked by 5p:1p ratio at 50 Hz (C) and Δ5p:1p ratio (D) before (gray) and after application of CPT (10 μm) in the absence (left, green) or presence of 50 mm EtOH (right, pink). Data are summarized from B. E, Summary (left) and mean peak (right) of unstimulated ΔF/F0 GRAB-Ado signal in NAcC normalized to predrug baseline (dotted line) before and after the application of ENT1 inhibitor NBTI (10 μm) in the absence (gray; n = 10 experiments/5 mice) or presence of 50 mm EtOH (pink; n = 9 experiments/5 mice). *p < 0.05, **p < 0.01, ***p < 0.001, unpaired t test (A, D), two-way repeated-measures ANOVA (B, C, E) with Sidak's multiple comparisons (C). Error bars indicate SEM.
Discussion
Here, we reveal that DA release in NAcC is under a tonic inhibition by ambient adenosine levels acting at A1Rs, and that ENT1, located at least in part on striatal astrocytes, governs the level of this tonic inhibition. Moreover, we reveal that ethanol promotes A1R inhibition of DA release, through elevating adenosine levels by diminishing adenosine uptake via ENT1. These data support the emerging concept that astrocytes play important roles in setting the level of striatal DA output, in health and disease.
Direct versus indirect actions of A1Rs
Our data provide functional evidence for direct regulation of DA release by A1Rs. Immunocytochemical studies in rat striatal synaptosomes indicate that dopaminergic axons can contain A1Rs (Borycz et al., 2007), but direct immunocytochemical evidence in situ is currently lacking. We excluded indirect actions of A1Rs via key candidate pathways, namely cholinergic inputs that act via nAChRs, and GABAergic networks that act via GABAA/GABAB receptors on DA axons. Previous reports have excluded effects of glutamatergic modulation (Borycz et al., 2007; O'Connor and O'Neill, 2008). Moreover, we found that A1R agonists inhibited DA release evoked by even single short optogenetic stimulation of ChR2-expressing DA axons, which should not coactivate other striatal neurons, suggesting that the locations for A1Rs that regulate DA are either a currently undisclosed tonically active cell type with currently unknown actions on DA or, more parsimoniously, DA axons themselves.
Tonic inhibition
A1R antagonists enhanced DA release evoked by single short optogenetic stimulation, suggesting that A1R activation does not require stimulation of any other network, and is tonically activated by an ambient striatal adenosine tone. An adenosine tone at striatal A1Rs has previously been described on glutamate inputs in NAcC (Brundege and Williams, 2002; Choi et al., 2004). Extracellular adenosine concentrations in the brain have been suggested to be in the range of 25–250 nm under basal conditions, which are sufficient to activate high-affinity A1Rs (Dunwiddie and Masino, 2001). In further support for a resting adenosine tone in NAcC, we could detect extracellular adenosine with the GRAB-Ado sensor, following application of an ENT1 inhibitor, in the absence of any striatal stimulation. The source of adenosine is still undetermined, but can arise from the catabolism of ATP from neuronal and non-neuronal sources (Latini and Pedata, 2001), including astrocytes (Corkrum et al., 2020).
A1Rs modify DA signal contrast for firing frequency
We found that A1R activation not only limits the overall amplitude of DA output, but also promotes the contrast in DA signals released by different firing rates and pulse numbers, and, vice versa, that A1R antagonists reduce this contrast. The strength of A1R-mediated inhibition of DA release varied with adenosine concentration, resulting from ENT1 inhibition or overexpression, and, furthermore, was greater for lower frequencies of activation of DA axons. These data indicate that an adenosine tone preferentially limits DA output during the lower frequencies of DA neuron activity that represent tonic activity, while leaving relatively intact the DA release evoked by the higher frequencies associated with phasic activity in DA neurons. The preferential inhibition of DA release by low frequencies of activity parallels the effects of GABA input to DA axons (Lopes et al., 2019; Roberts et al., 2020, 2021) and could indicate a preferential influence of striatal adenosine on DA functions that are proposed to be mediated by low frequencies of activity (e.g., the ongoing monitoring of reward value and its changes; Wang et al., 2021). We identified an underlying change in the dynamic short-term plasticity (STP) of DA output that corresponds to this change to frequency filtering, which again parallels the effect of GABA on STP in DA release (Lopes et al., 2019; Roberts et al., 2020). Elsewhere in the brain, A1R activation, including through ambient adenosine, can modulate STP of glutamate transmission through a presynaptic mechanism (Moore et al., 2003; Wong et al., 2006; Yang et al., 2007; Qi et al., 2017; Perrier et al., 2019). A1Rs are Gi/o coupled, and so their activation causes the inhibition of adenylyl cyclase, the activation of potassium channels, and the inactivation of voltage-gated calcium channels (VGCCs; Haas and Selbach, 2000). We have recently revealed that mechanisms that determine axonal excitability, particularly potassium-dependent processes, strongly gate STP of DA release (Condon et al., 2019), while calcium largely gates amplitude. We therefore hypothesize that A1Rs on striatal DA axons likely inhibit the overall amplitude of DA release via reduced VGCC activity while simultaneously gating the STP of DA release by modifying axonal excitability via potassium-dependent conductances. Future studies are needed to establish these potential mechanisms.
ENT1 and astrocytes are key regulators of tonic A1R inhibition of DA release
We established that ENT1 in NAcC limits ambient adenosine levels and tonic A1R inhibition of DA axons, and thereby indirectly facilitates DA release. ENT1 usually facilitates adenosine uptake from the extracellular milieu, but can reverse to release adenosine (Parkinson et al., 2011), and ENT1 has been shown to facilitate adenosine release evoked by electrical stimulation in cultured hippocampal neurons (Wu et al., 2020). Intriguingly, we found that adenosine release driven by electrical stimulation was TTX insensitive and sensitive to ENT1 inhibition. Together, our data show that ambient adenosine in NAcC is limited by uptake via ENT1 but can be released via ENT1 reversal in some conditions. There is an additional ENT in striatum, ENT2 (Jennings et al., 2001; Anderson et al., 1999), whose roles we did not explore. ENT1 is thought to be responsible for the majority of adenosine transport and, subsequently, the key regulator of extracellular adenosine levels across the brain (Young et al., 2008).
We found a major role for ENT1 located in striatal astrocytes, although we did not test or exclude roles for ENT1 located on neurons. The glial metabolic poison fluorocitrate limited the effects of ENT1 inhibitors on adenosine tone, while overexpressing ENT1 in astrocytes boosted evoked DA release by limiting tonic A1R inhibition. This strategy for viral overexpression of ENT1 in striatal astrocytes has previously been shown to increase ENT1 expression by ∼30%, but also increases striatal GFAP expression and modifies astrocyte morphology (Hong et al., 2020). ENT1 expression has also been shown to regulate astrocyte-specific excitatory amino acid transporter 2 (EAAT2) and aquaporin-4 expression in NAcC (Wu et al., 2010; Lee et al., 2013). These additional astrocyte-specific modifications might also contribute to boosted DA signaling in NAcC, beyond mechanisms involving tonic inhibition by adenosine tone. Regardless, astrocytic ENT1 in striatum has previously been reported to play key roles in reward-seeking behaviors (Hong et al., 2020; Kang et al., 2020), which our data suggest could be mediated by underlying changes to DA signaling.
The role for ENT1 on astrocytes in gating DA output parallels our recent finding that GABA transporters (GAT1 and GAT3) on astrocytes in dorsal striatum set the level of tonic GABAergic inhibition of DA release (Roberts et al., 2020). Furthermore, EAAT2, enriched on astrocytes, limits glutamate-mediated inhibition of DA release (Zhang and Sulzer, 2003), and, thus, our collective findings point to astrocytic transporters across neurotransmitter categories as regulators of DA release.
Dysregulation of A1R inhibition of DA release by ethanol
To probe the wider potential significance of the regulation of DA in NAcC by adenosine tone and ENT1, we explored whether A1R inhibition of DA release was modified by ethanol. Ethanol has been shown to increase extracellular adenosine levels by impairing adenosine uptake via ENT1 after acute exposure (Nagy et al., 1989, 1990; Choi et al., 2004) and, separately, to reduce DA release following acute application (Yorgason et al., 2014, 2015) or following chronic intermittent exposure in mice (Karkhanis et al., 2015; Rose et al., 2016). Here we bridge these different findings by revealing that ethanol can reduce DA output via a boosted tonic A1R inhibition. While pretreating slices with 50 mm ethanol to understand the disruption to cellular and circuit function has limitations in informing effects on behavior, we speculate that this mechanism could help to explain how striatal A1Rs contribute to the ataxic and hypnotic effects of ethanol (Meng and Dar, 1995; Phan et al., 1997; Dar, 2001), as well as how acute ethanol exposure attenuates concurrent GABA corelease from DA axons in dorsal striatum (Kim et al., 2015).
Chronic ethanol exposure is thought to evoke an adaptive response, resulting in decreased ENT1 expression, and therefore a reduced ability for ethanol to increase extracellular adenosine (Nagy et al., 1989). Indeed, rats permitted daily access to ethanol for 8 weeks exhibit downregulated ENT1 gene expression in NAc (Bell et al., 2009). ENT1-null mice exhibit reduced hypnotic and ataxic responses to ethanol, increased ethanol consumption, and decreased adenosine tone in NAc, while viral-mediated rescue of ENT1 expression in NAc reduces ethanol consumption (Choi et al., 2004; Jia et al., 2020). Furthermore, growing evidence implicates striatal adenosine signaling in the neurobiological adaptations of other drugs of abuse (Bachtell, 2017; Ballesteros-Yáñez et al., 2018). Repeated cocaine administration enhances adenosine uptake, reduces adenosine tone, and reduces plasma membrane A1R expression in NAcC (Manzoni et al., 1998; Toda et al., 2003), while striatal μ-opioid receptor activation by opioids results in reduced striatal adenosine tone in dorsomedial striatum (Adhikary and Birdsong, 2021). Dysregulated tonic A1R inhibition of glutamate afferents in NAc is thought to be a key circuit adaptation underlying ethanol abuse and other drugs of abuse (Choi et al., 2004; Chen et al., 2010; Wu et al., 2010; Nam et al., 2011); however, given the role that adenosine tone plays in setting the level and activity dependence of DA output we describe here, we speculate that dysregulated tonic A1R inhibition of DA release in NAcC might also be an important circuit adaptation underlying drug abuse.
In conclusion, we show here that A1Rs can tonically inhibit DA output though an ambient adenosine tone, and that A1Rs additionally regulate the activity sensitivity of DA release, preferentially impacting on release by low frequencies. Furthermore, we find that tonic A1R inhibition of DA release is regulated by ENT1 and astrocytes, and is dysregulated by ethanol. These data provide a further mechanism through which ethanol modulates striatal DA function and corroborates emerging data highlighting astrocytic transporters as important regulators of striatal function.
Footnotes
This work was supported by UK Medical Research Council Grant MR/V013599/1 (to S.J.C. and B.M.R.); University of Oxford John Fell Fund Grant 0008310 (to B.M.R.); a Junior Research Fellowship to B.M.R. from St John's College, Oxford University; and a Biotechnology and Biological Sciences Research Council Doctoral Training grant to J.A.L.
The authors declare no competing financial interests.
- Correspondence should be addressed to Stephanie J. Cragg at Stephanie.Cragg{at}dpag.ox.ac.uk or Bradley M. Roberts at Bradley.Roberts{at}dpag.ox.ac.uk