Abstract
It is well established that glutamate plays an important role in drug-induced and cue-induced reinstatement of drug seeking. However, the role of glutamate in drug reward is unclear. In this study, we systemically evaluated the effects of multiple glutamate transporter (GLT) inhibitors on extracellular glutamate and dopamine (DA) in the nucleus accumbens (NAc), intravenous cocaine self-administration, intracranial brain-stimulation reward (BSR), and reinstatement of cocaine seeking in male and female rats. Among the five GLT inhibitors we tested, TFB-TBOA was the most potent. Microinjections of TFB-TBOA into the NAc, but not the ventral tegmental area (VTA), or dorsal striatum (DS), dose-dependently inhibited cocaine self-administration under fixed-ratio and progressive-ratio (PR) reinforcement schedules, shifted the cocaine dose–response curve downward, and inhibited intracranial BSR. Selective downregulation of astrocytic GLT-1 expression in the NAc by GLT-1 antisense oligonucleotides also inhibited cocaine self-administration. The reduction in cocaine self-administration following TFB-TBOA administration was NMDA GluN2B receptor dependent, and rats self-administering cocaine showed upregulation of GluN2B expression in NAc DA- and cAMP-regulated phosphoprotein 32 (DARPP-32)-positive medium-spiny neurons (MSNs). In contrast, TFB-TBOA, when locally administered into the NAc, VTA, or ventral pallidum (VP), dose-dependently reinstated cocaine-seeking behavior. Intra-NAc TFB-TBOA-evoked drug-seeking was long-lasting and NMDA/AMPA receptor dependent. These findings, for the first time, indicate that glutamate in the NAc negatively regulates cocaine's rewarding effects, while an excess of glutamate in multiple brain regions can trigger reinstatement of drug-seeking behavior.
SIGNIFICANCE STATEMENT It is well known that glutamate plays an important role in relapse to drug seeking. However, the role of glutamate in drug reward is less clear. Here, we report that TFB-TBOA, a highly potent glutamate transporter (GLT) inhibitor, dose-dependently elevates extracellular glutamate and inhibits cocaine self-administration and brain-stimulation reward (BSR), when administered locally into the nucleus accumbens (NAc), but not other brain regions. Mechanistic assays indicate that cocaine self-administration upregulates NMDA-GluN2B receptor subtype expression in striatal dopaminoceptive neurons and activation of GluN2B by TFB-TBOA-enhanced glutamate inhibits cocaine self-administration. TFB-TBOA also reinstates cocaine-seeking behavior when administered into the NAc, ventral tegmental area (VTA), and ventral pallidum (VP). These findings demonstrate that glutamate differentially regulates cocaine reward versus relapse, reducing cocaine reward, while potentiating relapse to cocaine seeking.
Introduction
Glutamate is the primary excitatory neurotransmitter in the central nervous system and has been implicated in a number of neuropsychiatric disorders including substance use disorders and depression (Uys and LaLumiere, 2008; Knackstedt and Kalivas, 2009; Niswender and Conn, 2010; Lewerenz and Maher, 2015; Takahashi et al., 2015; Henter et al., 2018). In preclinical studies, rats display a decrease or deficit in basal glutamate transmission in the nucleus accumbens (NAc) after extinction from chronic administration of cocaine, heroin, or nicotine (Park et al., 2002; McFarland et al., 2003; Moussawi et al., 2011; Gipson et al., 2013; Shen et al., 2014a). Re-exposure to these drugs of abuse produces an increase in extracellular NAc glutamate and reinstatement of drug-seeking behavior, suggesting that glutamate plays a pivotal role in relapse to drug seeking (Knackstedt et al., 2010a, b; Shen et al., 2014b). Cocaine has been particularly well-studied in this regard and considerable research has linked glutamate dysfunction with cocaine relapse behavior (Kalivas, 2004; Wise, 2009; Schmidt and Pierce, 2010; Scofield et al., 2015). However, it is unknown whether increased extracellular glutamate in the NAc or other brain regions also alters the initial rewarding effects of cocaine.
Previous studies have shown that rats self-administer NMDA receptor (NMDAR) antagonists (MK-801 and CPP) directly into the NAc (Carlezon and Wise, 1996a). Moreover, pretreatment with MK-801 significantly enhanced cocaine reinforcement and increased break-points for cocaine self-administration (Ranaldi et al., 1996; Pierce et al., 1997), suggesting that NMDARs in the NAc are critically involved in cocaine reward. This is supported by a recent finding that ketamine, a potent NMDAR antagonist, is highly effective in the treatment of severe depression and substance use disorders (Jones et al., 2018; Henter et al., 2021). Indeed, the United States Food and Drug Administration (FDA) has recently approved S-ketamine for the treatment of depression in humans. However, the neural mechanisms underlying the antidepressant effects of ketamine are also poorly understood. In addition, we have previously reported that selective mGluR5 antagonists (MPEP, MTEP) and mGluR7 agonist (AMN-082) inhibited cocaine self-administration and cocaine-enhanced brain-stimulation reward (BSR; Li et al., 2010, 2018). Importantly, these compounds also dose-dependently increased NAc extracellular glutamate, but not DA, suggesting that a glutamate receptor mechanism may also underlie the pharmacological action of these mGluR compounds. However, direct evidence linking NAc glutamate and cocaine self-administration is lacking.
In the present report, we employed pharmacological and gene-targeting approaches to manipulate multiple excitatory amino acid transporters (EAATs). Five EAAT subtypes have been identified – each with unique properties. EAAT1 (the glutamate-aspartate transporter, GLAST) and EAAT2 (glutamate transporter, GLT-1) are expressed primarily on astrocytes (Schmidt and Pierce, 2010). EAAT3, EAAT4, and EAAT5 are almost exclusively expressed in neurons with varied regional distributions. EAAT3 has a wide expression pattern, whereas EAAT4 and EAAT5 are restricted to the cerebellum and retina, respectively. GLT-1 is responsible for the majority (>90%) of glutamate reuptake and therefore controls extracellular glutamate homeostasis (Spencer and Kalivas, 2017; Todd and Hardingham, 2020).
We first tested five EAAT inhibitors with varying subtype specificity (Fig. 1A) to determine which one could selectively elevate extracellular glutamate levels in the NAc and alter cocaine self-administration. UCPH-101 (IC50: 0.66 μm) and L-β-BA (IC50: 0.8 μm) are selective EAAT1 and EAAT3 inhibitors, respectively (Bunch et al., 2009; Abrahamsen et al., 2013). Dihydrokainic acid (DHK) is a GLT-1 (EAAT2) inhibitor with an IC50 of 31 μm (Bunch et al., 2009). DL-TBOA is a nonselective EAAT inhibitor (IC50 for EAAT1: 70 μm, EAAT2: 6 μm, EAAT3: 6 μm, EAAT4: 4.4 μm, EAAT5: 3.0 μm; Shimamoto et al., 2000; Shigeri et al., 2001) and TFB-TBOA is a potent astrocytic EAAT 1/2 inhibitor (IC50 for EAAT1: 0.022 μm, EAAT2: 0.017 μm, EAAT3: 0.3 μm; Shimamoto et al., 2004). We found that TFB-TBOA produced the most robust elevation in extracellular glutamate without impacting DA levels and consequently selected this EAAT inhibitor to further assess the role of NAc glutamate in cocaine self-administration and electrical BSR. We then used a genetic manipulation approach to down-regulate GLT-1 expression in the NAc and then observed the effects GLT-1 knock-down on cocaine self-administration in rats. To determine the neural mechanisms underlying glutamate modulation of cocaine self-administration, we used glutamate receptor subtype-specific antagonists and immunohistochemical assessments in conjunction with self-administration. Finally, we examined whether TFB-TBOA-enhanced extracellular glutamate in different brain regions can reinstate extinguished cocaine-seeking behavior in a way similar as cocaine-enhanced glutamate does. Unexpectedly, elevation of extracellular NAc glutamate by TFB-TBOA inhibited the acute rewarding effects of cocaine via a NMDA-GluN2B receptor mechanism, but triggered reinstatement of cocaine seeking via both AMPA and NMDA receptor mechanisms.
Materials and Methods
Animals
Male and female Long–Evans rats (Charles River Laboratories) were used in the experiments. All animals were housed individually in a climate-controlled animal room on a reversed light–dark cycle with free access to food and water. All experimental procedures were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of the United States National Research Council and were approved by the National Institute on Drug Abuse Animal Care and Use Committee.
Surgery
The surgical procedures are the same as we reported previously (25). Briefly, a microrenathane catheter was implanted into a jugular vein, and its free end fed subcutaneously along the scalp – connecting to a 22-gauge stainless connector mounted to the skull by stainless-steel screws and dental acrylic. To ensure patency, catheters were flushed daily with gentamicin (0.1 mg/ml) and heparin (30 IU/ml) following surgery. For intracranial microinjections, different groups of rats were placed in a stereotaxic frame and implanted bilaterally with guide cannulae into the NAc-shell (AP from bregma: +1.7, ML: ±2.0; DV: −5.5 mm, 6° from vertical), ventral tegmental area (VTA; AP −5.2, ML: ±1.8; DV: −7.5 mm, 8°), ventral pallidum (VP; AP −0.3, ±ML 3.2, DV −5.5 mm, 6°) or dorsal striatum (DS; AP: 2.0, ML: ±2.5; DV: −3.5 mm, 6°). Obturators were inserted into the guide cannulae. The self-administration head mount and microinjection guide cannulae were anchored to the skull using four stainless steel jewelers screws and dental acrylic. Bilateral drug microinjections were conducted via a 30-gauge injector through implanted guide cannula with the injector tip ∼0.25 mm below the tips of the guide cannula. Cannula placements were verified after the completion of experiments by standard histologic localization techniques. The microinjection area was identified by the injection volume (0.5 µl, which may diffuse to an area of ∼1.2 mm2 as assessed by blue dye microinjection) and the injection site (∼0.25 mm below the tip of a guide cannula).
Apparatus
Standard MED associates operant test chambers were used for self-administration experiments. Each box was equipped with two levers (active and inactive) and a house light set to on at the beginning of each session. Presses on the active lever triggered a compound light and tone cue. Data were analyzed using MED PC software. Locomotor activity was measured in open field chambers (Accusan Instruments) and data were collected via the VersaMax data analysis system.
Experiment 1: in vivo brain microdialysis
In vivo microdialysis was run to determine which EAAT inhibitors produced a consistent and dose-dependent increase in glutamate. Thirty-eight rats were implanted with two guide cannulae (CMA Microdialysis AB) in the NAc. After 5–7 d of recovery from surgery, microdialysis probes (CMA Microdialysis AB) were inserted into the guide cannulae (∼ 2 mm beyond the tips). Artificial CSF (aCSF; 5 mm KCl, 140 mm NaCl, 1.4 mm CaCl2, 1.2 mm MgCl2, 5.0 mm glucose, and 0.2 mm PBS) was perfused through the probe at a rate of 0.1 µl/min via a syringe pump (Bioanalytical SystemsIN). The following day, the pump speed was adjusted to 2 µl/min. Baseline dialysate levels were collected for 30 min, after which GLT inhibiters were locally perfused into the NAc over an ascending dose range such that a higher dose was administered every 30 min. A dose range of 1, 10, 100, and 300 μm was used, except for DL-TBOA (1, 10, 100, 1000 μm) and TFB-TBOA (1, 10, 100 μm). Microdialysis samples were collected every 20 min in tubes containing 10-µl 0.1 m perchloric acid to prevent DA degradation. Dialysate DA and glutamate samples were frozen at −80°C until later processing using HPLC with electrochemical and fluorometric detection.
Experiment 2: cocaine self-administration under FR reinforcement
Next, each of the EAAT inhibitors mentioned above was microinjected into the NAc, VTA, or DS before cocaine self-administration to determine whether an increase in extracellular glutamate alters cocaine's rewarding effects. A total of 56 animals underwent intravenous catheterization surgery combined with intracranial cannula placement. Following successful jugular catheterization, rats were fixed in a stereotaxic frame. Two guide cannulae (Plastics One) were implanted bilaterally into the NAc, VTA, or DS. Following recovery, subjects were trained to lever press for cocaine (1 mg/kg/infusion) under an FR1 reinforcement schedule for five sessions. Then cocaine self-administration continued under FR2 reinforcement at 0.5 mg/kg per infusion until stable responding was observed (<20% variability in responding across three consecutive sessions and a ratio of at least 2:1 active to inactive lever presses). Each session was 3 h long with a cap of 50 infusions to prevent overdose. Rats randomly received either vehicle or one of the five EAAT inhibitors intracranially, at a rate of 1 µl/min via syringe pump, 30 min before cocaine self-administration testing. UCPH-101, DHK, and L-β-BA were administered at 3 or 10 µg/side, DL-TBOA at 1 or 3 µg/side, and TFB-TBOA at 0.05, 0.1, or 0.3 µg/side. TFB-TBOA was also infused into the VTA (0.01, 0.05 µg/side) and DS (1, 3 µg/side). These drug doses were selected based on pilot studies (in which higher doses produced seizures or death). Between tests, subjects underwent an additional three to five sessions of cocaine self-administration until the baseline response was re-established. A maximum of three microinjections were administered per brain region in each animal.
Experiment 3: cocaine dose–response self-administration
To further assess the ability of TFB-TBOA to alter cocaine self-administration, cocaine was presented across a wide range of doses following TFB-TBOA microinjections. Subjects (n = 10) underwent sessions in which multiple doses of cocaine (0.031, 0.0625, 0.125, 0.25, 0.5, or 1 mg/kg/infusion) were available under FR2 reinforcement. The changing cocaine doses were administered by modifying the infusion volume and time the syringe pump remained on every 20 min. When a pattern of consistent responding was observed, testing began with TFB-TBOA. During each test session, subjects received intra-NAc TFB-TBOA (0, 0.1, or 0.3 µg/side) 30 min before testing in a counterbalanced manner with three to five sessions between the tests.
Experiment 4: cocaine self-administration under progressive-ratio (PR) reinforcement
To determine whether the reinforcing efficacy of cocaine was altered by TFB-TBOA infusions, PR testing was instituted. Specifically, subjects (n = 14) were given access to 0.5 mg/kg/infusion cocaine under a PR reinforcement schedule. Within each session the number of lever presses required for each infusion incrementally increased according to the following pattern: 1, 2, 4, 6, 9, 12, 15, 20, 25, 32, 40, 50, 62, 77, 95, 118, 145, 178, 219, 268, 328, 402, 492, 603, and 737. The break-point was defined as the number of presses made for the last cocaine infusion before subjects stopped responding for 1 h. Daily PR testing continued until stable break-point levels (<3 ratio increment difference) were observed for three consecutive sessions. Using a counterbalanced design, subjects received microinjections of TFB-TBOA (0, 0.1, 0.3 µg/side) 30 min before PR sessions. Break-points were reestablished between TFB-TBOA tests.
Experiment 5: sucrose self-administration
To assess whether the effects of TFB-TBOA were specific to cocaine reward, a nondrug reinforcer (sucrose) was presented for self-administration. Subjects undergoing oral sucrose self-administration received intracranial cannulae in the NAc (n = 6) or DS (n = 7) but were not catheterized. The procedures mirrored those used for cocaine self-administration. Presses on the active lever led to a 5% liquid sucrose delivery (0.1 ml) in a food tray and presentation of a compound tone and light cue. Animals responded on an FR2 reinforcement schedule and sessions lasted 1 h with a maximum of 100 sucrose deliveries permitted.
Experiment 6: open-field locomotion
To determine whether TFB-TBOA-induced reduction in cocaine self-administration is because of sedation or locomotor impairment, we observed the effects of TFB-TBOA on basal and cocaine-enhanced open-field locomotion. Rats were allowed to acclimate to the locomotor chambers for 2 d before testing and each session lasted 3 h. Separate locomotor experiments were run in two groups of rats (n = 8). In the first group, subjects were placed in the locomotor chambers for a 1-h habituation period and subsequently removed and given intra-NAc infusions of vehicle or 0.3 µg TFB-TBOA. After 30 min they were again placed in the locomotor chambers and locomotor activity was measured for 3 h. In the second experiment, subjects underwent another 1 h habituation period before intra-NAc infusion of vehicle, 0.1, or 0.3 µg TFB-TBOA; 30 min after the microinjections, cocaine (10 mg/kg, i.p.) was administered and subjects were placed in the open-field chambers for an additional 3 h of locomotor recording.
Experiment 7: cocaine self-administration with intra-NAc GLT-1 antisense oligonucleotide infusion
To determine whether EAAT2 (GLT-1) is the major target that TFB-TBOA acts on in the above experiments, accumbal GLT-1 was downregulated via a specific antisense oligo before cocaine self-administration. Rats (n = 33) received IV catheterization surgery and bilateral guide cannula implantation into the NAc for GLT-1 oligo infusion. After recovery, subjects underwent cocaine self-administration following the procedures described in experiment 2. Rats that acquired stable cocaine self-administration were implanted with mini-osmotic pumps (Alzet #2002) containing non-sense (control) or antisense GLT-1 oligo and connected to the NAc cannulae with polyethylene tubing placed subcutaneously. Lypophilized phosphodiester oligo was diluted with aCSF and loaded into the pumps (see Rothstein et al., 1996). Subjects received vehicle (non-sense oligo), 240, or 720 ng/d antisense for a total of 7 d. The pumps continuously infused non-sense or antisense oligos into the NAc at a rate of 1 µl/h. The sequence for the GLT-1 antisense oligo was: 5′-ATATTGTTCACCCTCGGTTGAT-3′ and for the GLT-1 random (non-sense oligo): 5-ATCAACCGAGGGTGCCAACAATAT-3′. Self-administration behavior was tested from day 1 to day 7 after oligo infusion began. Following the last self-administration session, rats were deeply anaesthetized and perfused through the ascending aorta with 0.01 m PBS, followed by 4% paraformaldehyde in PBS. Brains were then removed and fixed overnight in 4% paraformaldehyde, followed by cryoprotection in 30% sucrose at 4°C for 24–48 h.
Serial coronal sections (25 μm) containing the striatum were cut using a cryostat (Leica 3050S). Free floating sections were pretreated with 0.3% hydrogen peroxide in methanol for 20 min, then washed with PBS, and incubated for 1 h with a blocking solution composed of 0.2% Triton X-100 and 5% normal goat serum in PBS. Sections were then incubated with a primary antibody to GLT-1 (1:1000; Calbiochem brand of MilliporeSigma) in the blocking solution for 24 h at 4°C. After rinsing in PBS, sections were incubated in biotinylated secondary antisera (1:200) of the Vectastain (Elite) ABC kits for 2 h. Following PBS rinses, sections were incubated in an avidin–biotin–horseradish peroxidase complex (ABC) for 1 h at 22°C. Finally, the antigen–antibody complexes were visualized using 0.03% 3,3′-diaminobenzidine (DAB; Sigma-Aldrich) as the chromogen.
Experiment 8: cocaine self-administration in the presence of glutamate receptor antagonism
To study the receptor mechanisms underlying TFB-TBOA action in cocaine self-administration, additional groups of rats were used to evaluate the effects of pretreatment with various glutamate receptor antagonists on TFB-TBOA action. 50 rats were catheterized and received intra-NAc guide cannulae before the start of self-administration training. When stable responding was observed, DL-AP5 (AP5) (a NMDA antagonist, 0, 2, 10 µg/side) or DNQX (an AMPA/kainate receptor antagonist, 0, 2, 10 µg/side) was microinjected into the NAc and after 30 min TFB-TBOA (0, 0.3 µg/side) was infused into the NAc. The cocaine self-administration session began 30 min after the second microinjection with two to three baseline sessions between the tests. Since AP5, but not DNQX, blocked TFB-TBOA-induced reduction in cocaine self-administration, we further examined the effects of NMDAR subtype-specific antagonists on TFB-TBOA's action. NVP-AAM077 (a selective GluN2A inhibitor, 0, 2 µg/side), ifenprodil (a nonselective GluN2B inhibitor, 0, 2, 10 µg/side) or Ro-25-6981 (a selective GluN2B blocker, 0, 1 µg/side) were locally infused into the NAc 30 min before TFB-TBOA (0 or 0.3 µg/side) microinjections and cocaine self-administration.
Experiment 9: double-label immunofluorescence
In the above experiment, we found that NMDARs containing the GluN2B subunit underlie the attenuating effect of TFB-TBOA on cocaine reward. To further examine whether cocaine self-administration alters expression of the NMDA GluN2B subtype in DA- and cAMP-regulated phosphoprotein 32 (DARPP-32)-labeled dopaminoceptive neurons in the NAc, double label immunohistochemistry (IHC) was performed. Subjects exposed to cocaine self-administration (n = 3) or yoked saline controls (n = 3) were perfused and brains removed and stored as described above. Coronal 12-μm sections containing the NAc were cut on a cryostat. Sections were blocked with 4% BSA with 0.5% Triton X-100 phosphate buffer for 2 h on the bench. Following a series of washes, sections were incubated with the primary antibodies against GluN2B and DARPP-32 (1:200) overnight (GluN2B antibody from Millipore; DARPP-32 antibody from Santa Cruz) to detect GluN2B levels in DARPP-32+ medium-spiny neurons (MSNs) in the NAc core and shell. The day after washing, sections were incubated with secondary antibodies, donkey anti-goat Alexa Fluor 594 and donkey antirabbit Alexa Fluor 488 for 2 h at room temperature. Following rinses, sections were coverslipped, mounted, and images were processed with a fluorescence microscope (Nikon Eclipse 80i). All images were taken and presented under identical optical conditions. The DARPP-32-immunostaining and GluN2B-immunostaining density in each brain section was quantified using ImageJ software. The DARPP-32+ and GluN2B+ neurons in the NAc were counted manually in a double-blinded manner. Densitometric analysis and DARPP-32-GluN2B colocalization cells were counted in three to four stereologically selected sections per brain with 100-μm spacing between sections from the rostral to the caudal level under 40× magnification (De Biase et al., 2017; Shen et al., 2018). The mean density and cell count from three to four selected sections in each animal subjected to statistical analysis.
Experiment 10: electrical BSR
To assess whether TFB-TBOA produces a change in BSR thresholds, rats were anesthetized (60 mg/kg sodium pentobarbital, i.p.) and a unilateral monopolar stainless steel stimulating electrode (Plastics One) was surgically implanted and targeted at the medial forebrain bundle (MFB) in the lateral hypothalamus (AP +2.5 mm, ML +1.7 mm, and DV −8.4 mm). Subjects were allowed a minimum of 7 d to recover, before the start of experiments. Electrical ICSS was conducted in standard operant chambers (Med-Associates), each of which contained a wall-mounted retractable lever with a cue light situated above. Lever presses resulted in a 500-ms train of 0.1-ms rectangular cathodal pulses through the electrode, followed by retraction of the lever and a 500-ms “timeout” in which further lever presses did not produce brain stimulation. After acquiring brain stimulation-reinforced lever pressing, animals were presented with a series of 16 different pulse frequencies, ranging from 141 to 25 Hz in descending order. At each pulse frequency, animals could lever press for two 30-s trials, after which the pulse frequency was decreased by 0.05 log units. The response rate for each frequency was defined as the average number of lever presses during the two 30-s trials at that frequency. The BSR threshold (θ0) was defined as the minimum frequency at which an animal responded for rewarding stimulation and Ymax was defined as the maximal rate of responding (number of lever presses for BSR per unit of time). After stable ICSS responding was achieved (<10% variation in θ0 over five consecutive days; Spiller et al., 2019) animals were injected with cocaine (0, 2, 10 mg/kg, i.p.) or intra-NAc TFB-TBOA (0, 0.1, or 0.3 µg/side) 30 min before the test session. After each test, animals received additional self-stimulation sessions until a new θ0 baseline was established and then re-tested with a different dose of cocaine or TFB-TBOA.
Experiment 11: TFB-TBOA-induced reinstatement of cocaine-seeking behavior
We next sought to determine whether local increases in extracellular glutamate by TFB-TBOA could reinstate drug seeking following extinction. Subjects with intracranial guide cannulae implanted in the NAc (n = 22), DS (n = 11), VP (n = 11), or VTA (n = 10) were trained to self-administer cocaine on an FR2 reinforcement schedule followed by extinction training. During extinction, saline replaced cocaine and the cues previously accompanying drug delivery were absent. Extinction training continued until the extinction criteria were met (≤10 active lever presses for at least 2–3 d). Subjects were then divided into different dose groups and TFB-TBOA was microinjected into the NAc or DS (0, 1, and 3 µg/side), VP (0, 0.3, 1, 3 µg/side), or VTA (0, 0.05, 0.1 µg/side) 30 min before reinstatement tests. TFB-TBOA doses were counterbalanced with no more than three tests in each group of rats. After each test three to five sessions of extinction training were run until lever responding was extinguished. To delineate which glutamate receptor subtype underlies TFB-TBOA-induced drug seeking, two additional groups of rats with intra-NAc cannulae (n = 6) were pretreated with AP5 (2 µg/side) or DNQX (2 µg/side) 30 min before assessments of TFB-TBOA-induced reinstatement.
Experimental design and statistical analyses
Both the between-subjects (experiments 1, 3, 5, 6, 8, 10) and within-subjects designs (experiments 2, 4, 7, 9, 11) were used to evaluate the effects of EAAT inhibitors on cocaine self-administration and reinstatement of drug seeking as well as other measurements (such as BSR, locomotion, and GluN2 expression) in this study. All data were expressed as mean ± SEM and were analyzed using one-way or two-way ANOVA where appropriate. Significant main effects and interactions were followed by post hoc Student–Newman–Keuls tests for multiple group comparisons. Statistical analyses were performed using Systat.11-Statistics II software (Systat Software) with a threshold for statistical significance of p < 0.05.
Results
EAAT inhibitors increase extracellular levels of glutamate, but not DA, in the NAc
To determine which EAAT inhibitor is more potent and selective in elevating extracellular glutamate without producing significant side-effects (such as seizure), we locally infused EAAT inhibitors into the NAc to observe DA and glutamate, responsivity. The five inhibitors selected target astrocyte EAAT1 and EAAT2 and/or neuronal EAAT3 (Fig. 1A). They produced a significant increase in glutamate, but not DA, relative to baseline (Fig. 1B–F). Surprisingly, selective EAAT1 (UCPH-101), EAAT2 (DHK), and EAAT3 (L-β-BA) inhibitors did not produce a steady dose-dependent increase in glutamate. Instead, they generated a “pulsed increase,” a rapid rise and fall in extracellular glutamate after each drug dose (Fig. 1B–D). A two-way repeated measures ANOVA on the UCPH-101 data over the dose (time) revealed a significant DA/glutamate main effect (F(1,8) = 9.2, p < 0.05; Fig. 1B) and a dose main effect (F(14,112) = 1.94, p < 0.05). Since UCPH-101 has no effect on extracellular levels of DA, the glutamate/DA main effect reveals a change in extracellular glutamate after UCPH-101 administration, indicating that this EAAT inhibitor selectively altered extracellular glutamate, not DA, levels in the NAc. This explanation also applies to other datasets below as shown in Figure 1C–F. An identical analysis on the DHK data (Fig. 1C) did not detect a DA/glutamate main effect (F(1,14) = 2.83, p > 0.05), but there was a significant dose main effect (F(14,196) = 2.52, p < 0.05) and dose × DA/glutamate interaction (F(14,196) = 2.70, p < 0.01). A two-way repeated measures ANOVA for the L-β-BA data (Fig. 1D) revealed a significant DA/glutamate main effect (F(1,11) = 5.94, p < 0.05), dose main effect (F(14,154) = 4.64, p < 0.001) and dose × DA/glutamate interaction (F(14,154) = 5.270, p < 0.001). Post hoc individual group comparisons indicated a significant increase in glutamate after local infusion of each EAAT inhibitor into the NAc compared with baseline (Fig. 1B–D). These data demonstrate that selective inhibition of EAAT1, EAAT2, or EAAT3 produces a short-term increase in extracellular glutamate. This spike may reflect a compensatory increase in glutamate reuptake mediated by the other EAAT subtypes when one transporter is blocked.
Effects of five EAAT inhibitors on extracellular dopamine (DA) and glutamate in the NAc. A, Diagram showing the cellular distributions of EAAT1, EAAT2, and EAAT3 in the brain and the targets each EAAT inhibitor acts on; B–D, Effects of intra-NAc local infusion of the EAAT1 (UCPH-101), EAAT2 (DHK), or EAAT3 (L-β-BA) inhibitors on extracellular DA and glutamate, respectively. E, F, Effects of the nonselective EAAT inhibitor (DL-TBOA) or the astrocytic EAAT1 and EAAT2 inhibitor (TFB-TBOA) on extracellular DA and glutamate. All EAAT inhibitors increased extracellular glutamate, but not DA, in the NAc; *p < 0.05, **p < 0.01, ***p < 0.001, compared with the baseline before the drug infusion.
We next examined the effects of the subtype nonselective EAAT inhibitor DL-TBOA and the astrocytic EAAT inhibitor TFB-TBOA on extracellular DA and glutamate levels. We found that both TBOA analogs produced a dose-dependent increase in glutamate, with no effects on DA levels (Fig. 1E,F). TFB-TBOA is more potent and effective in elevating extracellular glutamate than DL-TBOA at the same doses (1, 10, 100 μm; Fig. 1E,F). When higher TFB-TBOA doses (>100 μm) were tested, a number of rats had seizures likely because of an excess of extracellular glutamate. To avoid this problem, we exclusively used lower doses (1, 10, 100 μm). A two-way repeated measures ANOVA for the DL-TBOA data (Fig. 1E) revealed a significant DA/glutamate main effect (F(1,10) = 6.51, p < 0.05), dose main effect (F(14,140) = 7.18, p < 0.001), and dose × DA/glutamate interaction (F(14,140) = 9.16, p < 0.01). A similar two-way repeated measures ANOVA for the TFB-TBOA data (Fig. 1F) also revealed a significant DA/glutamate main effect (F(1,10) = 13.55, p < 0.01), dose main effect (F(11,110) = 5.84, p < 0.001), and dose × DA/glutamate interaction (F(11,110) = 2.75, p < 0.01). Post hoc individual group comparisons indicated a significant increase in extracellular glutamate at 100–1000 μm DL-TBOA (Fig. 1E) and 1–100 μm TFB-TBOA (Fig. 1F) relative to baseline sampling.
DL-TBOA and TFB-TBOA inhibit cocaine self-administration under FR reinforcement
We then sought to determine whether microinjections of the five EAAT inhibitors into the NAc, DS, or VTA would alter cocaine self-administration under an FR2 reinforcement schedule in rats. We found that intra-NAc microinjection of UCPH-101 (one-way repeated measures ANOVA, F(2,20) = 0.97, p > 0.05), DHK (F(2,22) = 1.72, p > 0.05), or L-β-BA (F(2,23) = 2.37, p > 0.05) failed to alter intravenous cocaine self-administration at 3 or 10 µg/side (Fig. 2A,B). In contrast, intra-NAc microinjections of either DL-TBOA or TFB-TBOA produced a dose-dependent reduction in cocaine self-administration. Microinjections of TFB-TBOA were more effective at lower doses relative to DL-TBOA (Fig. 2C). One-way repeated measures ANOVAs for each TBOA compound revealed a significant main effect of treatment (DL-TBOA: F(2,24) = 4.36, p < 0.05; TFB-TBOA: F(3,26) = 11.49, p < 0.001; Fig. 2C). Post hoc comparisons confirmed that pretreatment with DL-TBOA significantly reduced cocaine infusions at 3 µg/side relative to vehicle. Similarly, 0.1 and 0.3 µg TFB-TBOA robustly inhibited responding for cocaine.
Effects of microinjections of EAAT inhibitors on cocaine self-administration under an FR2 reinforcement schedule. A, Representative histologic images showing the tips of guide cannulae and the predicted microinjection areas in the NAc, DS, and VTA. B, Intra-NAc microinjections of UCPH-101, DHK, or L-β-BA did not alter cocaine self-administration. C, Intra-NAc DL-TBOA or TFB-TBOA dose-dependently attenuated lever presses for cocaine. D, TFB-TBOA microinjections in the VTA or DS had no effect on cocaine self-administration. E, F, G, Representative cocaine self-administration (infusions) records showing the extinction-like pattern of cocaine self-administration (infusions) observed 30 min after TFB-TBOA microinjection into the NAc (D), but not into the VTA (E) or DS (F); *p < 0.05, ***p < 0.001, compared with the vehicle control group.
On the other hand, microinjections of TFB-TBOA into the VTA (0.01, 0.05 µg/side) or DS (1, 3 µg/side; Fig. 2D) had no effect on cocaine self-administration (VTA: F(2,20) = 1.36, p > 0.05; DS: F(2,18) = 1.25, p > 0.05; Fig. 2D). Lower doses of TFB-TBOA were selected by default for VTA microinjection because of the high risk of seizures with larger doses in this brain region. Figure 2E–G shows representative records of cocaine self-administration sessions illustrating a cessation in cocaine self-administration after intra-NAc 0.3 µg TFB-TBOA administration, but not after intra-VTA or intra-DS administration. This effect lasted ∼90 min and gradually waned.
TFB-TBOA inhibits cocaine self-administration under multiple-dose and PR reinforcement
To further evaluate the role of extracellular glutamate in cocaine reward, TFB-TBOA was administered into the NAc before an assessment of cocaine self-administration using a dose–response curve. This behavioral measure is thought to assess the effects of a given drug on cocaine's rewarding efficacy (Hiranita et al., 2009; Keck et al., 2013). Figure 3A demonstrates that TFB-TBOA induced a significant downward shift in the cocaine dose–response curve, indicating a reduction in cocaine reward. A two-way repeated measures ANOVA revealed a significant main effect of TFB-TBOA treatment (F(2,20) = 7.49, p < 0.05) and an interaction between TFB-TBOA treatment and cocaine dose (F(10,20) = 3.84, p < 0.01) but no main effect of cocaine dose (F(5,20) = 1.58, p > 0.05). Post hoc individual group comparisons indicated that both 0.1 and 0.3 µg TFB-TBOA produced a significant reduction in cocaine infusions compared with the vehicle control group (Fig. 3A).
Effects of intra-NAc TFB-TBOA on cocaine or sucrose reward and locomotor activity. A, Intra-NAc TFB-TBOA microinjections shifted the cocaine self-administration dose–response curve downward. B, Intra-NAc TFB-TBOA lowered break-points for cocaine self-administration under a PR reinforcement schedule. C, Intra-NAc TFB-TBOA inhibited oral sucrose self-administration under a FR2 reinforcement schedule. D, Microinjections of TFB-TBOA into the DS had no effect on sucrose self-administration. E, Microinjections of TFB-TBOA into the NAc had no effect on cocaine-induced hyperactivity. F, Intra-NAc TFB-TBOA failed to alter basal locomotor activity; *p < 0.05, **p < 0.01, ***p < 0.001, compared with vehicle control group.
We also looked at how intra-NAc TFB-TBOA affected responding for cocaine under PR reinforcement, to determine how motivated subjects were to pursue cocaine (Richardson and Roberts, 1996; Xi et al., 2005). The results showed that TFB-TBOA decreased responding on a PR schedule and lowered break-points for cocaine (Fig. 3B). A one-way ANOVA revealed a main effect of TFB-TBOA treatment (F(2,21) = 9.18, p < 0.01). Post hoc comparisons indicated that both 0.1 and 0.3 µg TFB-TBOA significantly decreased break-points for cocaine (Fig. 3B).
TFB-TBOA inhibits oral sucrose self-administration
Next, we examined whether TFB-TBOA would have similar effects on a nondrug reinforcer, sucrose. We found that microinjection of TFB-TBOA into the NAc (Fig. 3C), but not the DS (Fig. 3D), significantly inhibited sucrose self-administration. A one-way ANOVA revealed a main effect of TFB-TBOA treatment (F(2,12) = 8.95, p < 0.01; Fig. 3C). Post hoc individual group comparisons indicated that TFB-TBOA, at 1 µg, significantly reduced sucrose intake relative to the vehicle group.
TFB-TBOA fails to alter basal or cocaine-enhanced locomotor activity
To eliminate the possibility that the TFB-TBOA-induced reduction in cocaine self-administration was due to nonspecific locomotor impairment, the effects of intra-NAc TFB-TBOA on basal or cocaine-enhanced locomotor activity was measured. TFB-TBOA, at the same doses that inhibit cocaine self-administration, had no effect on cocaine-induced hyperlocomotion (Fig. 3E) or basal levels of locomotion (Fig. 3F). A two-way repeated measures ANOVA did not reveal a TFB-TBOA treatment main effect or a TFB-TBOA × time interaction, although a time (cocaine) main effect was observed (F(23,322) = 19.68, p < 0.001; Fig. 3E). Similarly, a two-way repeated measures ANOVA on locomotor activity following TFB-TBOA administration failed to reveal a main effect of treatment (F(1,129) = 3.16, p > 0.05; Fig. 3F).
TFB-TBOA inhibits electrical BSR in rats
To evaluate whether TFB-TBOA alters brain reward function, we used the highly sensitive BSR paradigm (Spiller et al., 2019). Figure 4A illustrates the experimental method with the location of a stimulation electrode in the MFB at the lateral hypothalamus. Figure 4B shows representative rate-frequency functions for BSR, indicating the BSR threshold θ0 value and Ymax value under baseline conditions and following cocaine or TFB-TBOA administration. Cocaine (2, 10 mg/kg, i.p.) dose-dependently decreased the BSR threshold θ0 value and shifted the response curve to the left (F(2,30) = 4.25; p < 0.001; Fig. 4C), demonstrating that cocaine potentiates the rewarding effects of brain stimulation. Intra-NAc TFB-TBOA (0.1, 0.3 µg/side) produced a dose-dependent increase in θ0 and shifted the response curve to the right (F(2,15) = 4.41; p < 0.05; Fig. 4D), signifying a reduction in BSR after TFB-TBOA administration. TFB-TBOA by itself had no effect on Ymax (Fig. 4E), which indicates that drug administration does not impair locomotion.
Effects of intra-NAc administration of TFB-TBOA on electrical BSR in rats. A, A diagram illustrating the location of a stimulation electrode in the MFB at the anterior–posterior level of the lateral hypothalamus. B, A schematic diagram, showing that cocaine shifted the curve to the left and decreased the BSR stimulation threshold (θ0), while TFB-TBOA shifted the curve to the right and increased θ0. C, D, Cocaine dose-dependently decreased, while TFB-TBOA dose-dependently increased BSR thresholds (θ0). E, TFB-TBOA failed to alter Ymax; *p < 0.05, ***p < 0.001, compared with the vehicle control group.
GLT-1 antisense oligonucleotide infusion attenuates cocaine self-administration
Given that GLT-1 is responsible for >90% glutamate reuptake (Spencer and Kalivas, 2017; Todd and Hardingham, 2020) and TFB-TBOA is an EAAT1 and EAAT2 (GLT-1) inhibitor, we further evaluated whether down-regulation of GLT-1 expression by chronic infusion of GLT-1 antisense oligonucleotide (oligo) inhibits cocaine self-administration. Figure 5A shows the experimental timeline of events. Figure 5B,C depicts representative GLT-1-immunostaining in the NAc in rats microinjected with GLT-1 non-sense (control) or antisense oligo. Based on the anatomic conformation of the cells, it appears that GLT-1 is mainly expressed in astrocytes. There was a significant reduction in GLT-1 optical density in rats given the antisense infusion compared with the non-sense oligo control group (p < 0.001; Fig. 5D). Intra-NAc local perfusion of the GLT-1 antisense oligo significantly and dose-dependently inhibited cocaine self-administration under an FR2 reinforcement schedule throughout the 7-d oligo infusion period. A two-way repeated measures ANOVA revealed a main effect of antisense treatment (F(2,27) = 18.69, p < 0.001; Fig. 5E). Post hoc individual group comparisons discovered a significant reduction in cocaine self-administration on days 2, 3, 5, 6, and 7 after starting the continuous oligo infusion at 720 ng/d (Fig. 5E).
Downregulation of GLT-1 expression via antisense oligo infusions into the NAc inhibits cocaine self-administration. A, Experimental timeline of events for GLT-1 antisense infusions and cocaine self-administration. B, C, Representative images of GLT-1-immunostaining in NAc sections taken at three magnifications (4×, 20×, 40×) from rats exposed to non-sense (control) or antisense oligos. D, GLT-1 density was significantly lower in the NAc of rats receiving GLT-1 antisense oligos than those receiving non-sense oligos. E, Intra-NAc antisense oligo administration inhibited cocaine self-administration in a dose-dependent manner; *p < 0.05, **p < 0.01, ***p< 0.001, compared with vehicle control group.
TFB-TBOA inhibits cocaine self-administration by a NMDAR-GluN2B receptor mechanism
We next investigated which glutamate receptors mediate TFB-TBOA's inhibitory effects on cocaine self-administration. DNQX, a selective AMPA-kainate antagonist, failed to alter the TFB-TBOA-induced reduction in cocaine self-administration (Fig. 6A, left panel). A one-way repeated measures ANOVA revealed a significant TFB-TBOA treatment main effect (F(3,19) = 16.55, p < 0.001), while post hoc comparisons demonstrated no difference in the numbers of cocaine infusions between different DNQX dose groups. DNQX alone produced a significant reduction in cocaine self-administration at 10 µg (F(2,18) = 20.68, p < 0.001; Fig. 6A, left panel). These findings suggest that TFB-TBOA-attenuated cocaine self-administration is not mediated by an AMPA/kainate receptor mechanism.
Receptor mechanisms underlying TFB-TBOA-induced reduction in cocaine self-administration. A, Intra-NAc pretreatment with DNQX had no effect on TFB-TBOA-induced reduction in cocaine self-administration. DNQX alone (10 µg) inhibited cocaine self-administration. B, Pretreatment with accumbal AP5 dose-dependently reversed TFB-TBOA-mediated decrease in cocaine self-administration. C, NVP-AAM077 pretreatment (selective GluN2A blocker) failed to alter TFB-TBOA-induced suppression of cocaine self-administration. D, Intra-NAc ifenprodil (a GluN2B blocker) blocked the decrease in cocaine self-administration caused by TFB-TBOA in a dose-dependent manner. E, Microinjection of Ro-25-6981 (a selective GluN2B blocker) into the NAc prevented the TFB-TBOA-induced reduction in cocaine self-administration; *p < 0.05, **p < 0.01, ***p < 0.001, compared with the vehicle group without TFB-TBOA; #p < 0.05, ##p < 0.01, ###p < 0.01, compared with each glutamate receptor antagonist vehicle group in the presence of TFB-TBOA.
We then examined the effects of AP5, a selective NMDAR blocker, on TFB-TBOA's action. AP5 blocked the TFB-TBOA-induced reduction in cocaine self-administration (Fig. 6B). A one-way repeated measures ANOVA showed a main effect of TFB-TBOA treatment (F(3,20) = 9.52, p < 0.001). Post hoc comparisons demonstrated a significant difference in responding for cocaine in subjects administered TFB-TBOA and pretreated with 10-µg AP5 relative to those given vehicle. Unlike DNQX, AP5 by itself failed to alter cocaine self-administration.
To further determine which NMDAR subtypes underlie TFB-TBOA's effects, additional antagonists with specificity for the NR2A (GluN2A) subtype (NVP-AAM077) or NR2B (GluN2B) subtype (ifenprodil, Ro-25-6981) were microinjected into the NAc-shell before TFB-TBOA administration. NVP-AAM077 pretreatment failed to alter TFB-TBOA's action on cocaine self-administration (F(2,13) = 13.09, p < 0.001; Fig. 6C). However, pretreatment with ifenprodil (F(2,19) = 9.81, p < 0.001; Fig. 6D) or Ro-25-6981 (F(2,14) = 38.83, p < 0.001; Fig. 6E) significantly blocked the TFB-TBOA-induced reduction in cocaine self-administration. When these NMDAR subtype antagonists were administered alone, ifenprodil, but not NVP-AAM077 or Ro-25-6981, produced a significant reduction in cocaine self-administration (F(2,12) = 7.74, p < 0.01; Fig. 6D).
Cocaine self-administration upregulates GluN2B expression in the NAc
It is currently unknown how an excess of extracellular glutamate can modulate cocaine reward via the NMDA-GluN2B receptor. However, cocaine elevates extracellular DA by blocking the DA transporter (DAT). Subsequently, DA acts on downstream dopaminoceptive neurons that express DARPP-32 in the NAc and other DA projection regions (Kebabian and Greengard, 1971; Walaas et al., 1983). We hypothesized that cocaine self-administration may up-regulate GluN2B expression in DARPP-32+ dopaminoceptive neurons in the NAc, providing a mechanism by which TFB-TBOA or glutamate inhibits cocaine self-administration. To test this hypothesis, we used double-staining IHC assays to examine GluN2B and DARPP-32 expression in the NAc of rats trained to self-administer cocaine or a yoked saline group. Figure 7A–D depicts representative images of GluN2B-immunostaining and DARPP-32-immunostaining in the NAc shell and core. Cocaine self-administration did not alter the total density of DARPP-32 (NAc-shell, T = −1.12, p > 0.05; NAc-core, T= 0.34, p > 0.05; Fig. 7E) or GluN2B (NAc-shell, T = −0.22, p > 0.05; NAc-core, T = −2.77, p > 0.05; Fig. 7F). However, co-localization of GluN2B and DARPP-32 was significantly increased in rats with a history of cocaine self-administration compared with the yoked saline control group (NAc shell, T = 5.56, p < 0.001; NAc-core, T = 3.38, p < 0.05; Fig. 7G), suggesting that cocaine self-administration up-regulates GluN2B expression in DARPP-32-positive neurons, while decreases GluN2B expression in other non-DARPP-32-expressing cells.
Cocaine self-administration increases NMDA-GluN2B receptor expression in NAc DARPP32 neurons. A–D, Representative fluorescent images illustrating GluN2B cells (green), DARPP32 cells (red), and their co-expression (yellow) in the NAc shell (A, B) or core (C, D) in rats with a history of cocaine self-administration or yoked saline treatment. E, No difference in the density of DARPP32 was observed between cocaine-treated and saline-treated rats in either the NAc core or shell. F, No difference in the density of GluN2B was observed between cocaine-treated and saline-treated rats. G, Cocaine self-administration increased GluN2B expression in DARPP32 neurons in the NAc core and shell relative to controls; *p < 0.05, ***p< 0.001, compared with saline control rats.
TFB-TBOA reinstates cocaine seeking after injections into the NAc, VTA, or VP
As TFB-TBOA administration enhances extracellular glutamate, we were interested in determining whether microinjections of this compound into the NAc and other reward-related brain regions (i.e., VTA, VP, DS; Fig. 8A–D) would reinstate drug-seeking behavior, similar to a cocaine priming injection. Rats were trained for cocaine self-administration, followed by extinction. Reinstatement testing began once the extinction criteria were met. We found that bilateral microinjections of TFB-TBOA into the NAc (F(2,24) = 4.81, p < 0.05; Fig. 8E), VP (F(3,19) = 10.25, p < 0.001; Fig. 8F), or VTA (F(2,18) = 4.68, p < 0.05; Fig. 8G), but not DS (F(2,16) = 0.35, p > 0.05; Fig. 8H), produced robust, dose-dependent reinstatement. Interestingly, intra-NAc TFB-TBOA reinstatement was long-lasting. High levels of lever responding continued for at least 5 d after TFB-TBOA administration on the first reinstatement test day (T1; Fig. 8I). A two-way repeated measures ANOVA on lever presses following intra-NAc TFB-TBOA showed a significant TFB-TBOA treatment main effect (F(2,28) = 14.23, p < 0.01; Fig. 8I). In contrast, the reinstatement response occurred only on the first test day when TFB-TBOA was microinjected into the VTA (Fig. 8J).
Reinstatement responses caused by intracranial microinjections of TFB-TBOA into the NAc, VP, VTA, or DS in rats after extinction from cocaine self-administration. A–D, Representative histologic images showing the traces of guide cannulae in the brain and the predicted microinjection areas in the NAc (A), VP (B), VTA (C), or DS (D). E–H, Microinjections of TFB-TBOA into the NAc (E), VP (F), and VTA (G), but not the DS (H), induced reinstatement of cocaine seeking in a dose-dependent manner following extinction from cocaine self-administration. I, J, The time course of active lever responding (drug seeking) before and after TFB-TBOA microinjections into the NAc (I) or VTA (J), illustrating that intra-NAc TFB-TBOA caused a long-term increase in cocaine-seeing behavior. K, Pretreatment with AP5 (2 µg/side, NAc) or DNQX (2 µg/side, NAc) blocked TFB-TBOA-induced reinstatement of cocaine-seeking behavior; *p < 0.05, **p < 0.01, ***p< 0.001, compared with vehicle (E–G, K) or baseline (I, J); ##p < 0.01 compared with the group administered (Veh + 3 µg/side TFB-TBOA) (K).
We also examined the receptor mechanisms underlying TFB-TBOA-induced cocaine seeking. Pretreatment with either AP5 (2 µg) or DNQX (2 µg) blocked the reinstatement response induced by microinjections of TFB-TBOA into the NAc (Fig. 8K). A one-way ANOVA revealed a significant treatment main effect (F(3,24) = 9.35, p < 0.001). Post hoc comparisons indicated that subjects administered TFB-TBOA produced a significant reinstatement response relative to vehicle controls, which was blocked by AP5 and DNQX (Fig. 8F).
Discussion
In this study, we systemically evaluated the effects of extracellular glutamate on cocaine self-administration, BSR, and reinstatement of drug-seeking behavior via EAAT inhibitors. We found the following. (1) All five EAAT inhibitors, particularly the astrocytic EAAT1/2 inhibitor TFB-TBOA, selectively increased extracellular glutamate, but not DA, in the NAc in a dose-dependent manner. (2) Accumbal microinjections of TFB-TBOA dose-dependently inhibited cocaine self-administration under both FR and PR reinforcement schedules, and shifted the cocaine dose–response curve downward, indicating a reduction in cocaine reward. (3) In an electrical BSR paradigm, cocaine and TFB-TBOA produced opposite effects, e.g., cocaine enhanced, while TFB-TBOA attenuated BSR. (4) Selective downregulation of GLT-1 expression in the NAc also inhibited cocaine self-administration, suggesting GLT-1 as the major target of TFB-TBOA. (5) Cocaine self-administration upregulated GluN2B expression in NAc DARPP-32+ neurons and pharmacological blockade of GluN2B, not AMPA or GluN2A, receptor subtype blocked TFB-TBOA-induced reduction in cocaine self-administration, suggesting a NMDA-GluN2B receptor mechanism underlying glutamate modulation of drug reward. And finally, (6) microinjections of TFB-TBOA into the NAc, VTA and VP, but not the DS, dose-dependently reinstated cocaine seeking, an effect that was blocked by either NMDA or AMPA/kainate receptor antagonists. Notably, intra-NAc TFB-TBOA-induced reinstatement responding was long-lasting compared with that by cocaine priming, suggesting that the NAc is a highly sensitive or vulnerable region in reinstatement response to glutamate. Overall, these findings suggest that glutamate differentially regulates drug reward versus relapse, inhibiting cocaine reward, while potentiating relapse to drug-seeking behavior.
Previous work has concentrated on changes in GLT-1 (EAAT2) expression following chronic administration of drugs of abuse (Roberts-Wolfe and Kalivas, 2015; Smaga et al., 2020). It has been demonstrated that during cocaine self-administration and withdrawal GLT-1 is downregulated in the NAc core and shell (Knackstedt et al., 2010a; Fischer-Smith et al., 2012; Reissner et al., 2014, 2015; LaCrosse et al., 2017; Bechard et al., 2018; Sepulveda-Orengo et al., 2018). Interestingly, GLT-1 agonism has been shown to have beneficial effects in relapse models. Indeed, elevation of GLT-1 expression by chronic administration of ceftriaxone, a commonly used β-lactam antibiotic, attenuates cocaine or cue-induced reinstatement of drug-seeking behavior (Sari et al., 2009; Knackstedt et al., 2010a; Smaga et al., 2020; Niedzielska-Andres et al., 2021). Similarly, systemic administration or intracranial microinjections of a cystine prodrug, N-acetylcysteine, increases nonvesicular glutamate release via cystine-glutamate exchangers, which subsequently rescued cocaine-induced decreases in basal glutamate transmission and attenuated cocaine-primed reinstatement of drug-seeking behavior (Baker et al., 2003; Knackstedt et al., 2010a; Kupchik et al., 2012), suggesting an important role of glutamate in relapse to drug-seeking behavior. However, it is important to note that ceftriaxone and N-acetylcysteine are not GLT-1 or glutamate specific modulators. They have multiple targets complicating interpretation of these findings. For example, ceftriaxone inhibits penicillin-binding proteins on the inner membrane of the bacterial cell wall and N-acetylcysteine increases glutathione production (Atkuri et al., 2007; Zelenitsky et al., 2018). Furthermore, ceftriaxone-induced upregulation of GLT-1 expression does not necessarily translate to accelerated glutamate uptake, as one report found that ceftriaxone application had no effect on basal and activity-dependent glutamate clearance (Wilkie et al., 2021). Altogether, these findings highlight the importance of studying novel glutamate-specific agents to understand the role of glutamate in relapse prevention and CNS function.
In the present study, we, for the first time, used selective EAAT inhibitors to manipulate extracellular glutamate levels and then observed the effects of glutamate on cocaine self-administration and reinstatement of drug-seeking behavior. We found that selective inhibition of EAAT1 (UCPH-101), EAAT2 (DHK), or EAAT3 (L-β-BA) caused a significant increase in extracellular glutamate but failed to alter cocaine self-administration. This is likely because of the pulsatile pattern of glutamate efflux observed following each dose of the inhibitor, during which accumbal glutamate rapidly rose and fell. These transient fluctuations may reflect enhanced glutamate uptake by other EAATs after one subtype is blocked. However, when multiple EAAT subtypes were inhibited by DL-TBOA (a subtype nonselective EAAT inhibitor) and TFB-TBOA (an astrocyte EAAT1/2 inhibitor), a steady increase in extracellular glutamate was observed. Among the five EAAT inhibitors we tested, TFB-TBOA displayed the highest potency and efficacy. At very low doses, intra-accumbens TFB-TBOA produced a dose-dependent increase in glutamate efflux and a reduction in cocaine self-administration. Based on these findings, TFB-TBOA was selected as our major tool to address the role of glutamate in drug reward and relapse in all subsequent experiments.
The key finding in this study is the reduction in acute cocaine reward following elevation of extracellular glutamate in the NAc by TFB-TBOA. Specifically, TFB-TBOA administration attenuated cocaine self-administration under FR reinforcement conditions with a classic extinction-like pattern: cessation in cocaine self-administration immediately after drug infusion, a reduction in PR break point for cocaine reward, and a downward shift in the cocaine dose–response curve in the absence of locomotor suppression. In addition, an excess of NAc glutamate inhibited sucrose self-administration, suggesting a common mechanism underlying drug and food reward. We have previously reported that mGluR5 antagonists (MPEP, fenobam) and mGluR7 agonists (AMN-082) can dose-dependently inhibit cocaine or food self-administration (Li et al., 2008, 2009, 2018; Keck et al., 2013, 2014). Given that these mGluR compounds also produce an increase in NAc extracellular glutamate, but not DA (Li et al., 2008, 2009, 2010, 2018), a striatal glutamate mechanism may also explain how these mGluR compounds inhibit cocaine self-administration.
As described above, our data also support a brain region-specific modulation of cocaine reward. TFB-TBOA microinjections attenuated cocaine reward only when administered into the NAc, without effect after microinjections into the VTA or DS. However, considerably lower doses (0.01, 0.05 µg/side) of TFB-TBOA were infused into the VTA relative to the NAc (0.1, 0.3 µg/side), because of the high risk of seizure. Therefore, it is unclear whether the VTA is relatively insensitive to TFB-TBOA administration or higher doses are required to inhibit cocaine self-administration. Larger doses (1, 3 µg/side) of TFB-TBOA were microinjected in the DS but failed to alter cocaine self-administration.
Another critical finding is the identification of a NMDAR subtype-specific (GluN2B) mechanism that may underlie glutamate-mediated reduction in cocaine reward. This is supported by the finding that a NMDA antagonist (AP5), but not an AMPA/kainite antagonist (DNQX), blocked TFB-TBOA-induced inhibition of cocaine self-administration. NMDARs are heteromers containing two N1 and two or three N2 subunits (Massey et al., 2004). Four N2 subunits have been identified. Multiple NMDAR subtypes can be formed from two N1 and two N2 subunits including NR2A (GluN2A), NR2B (GluN2B), NR2C (GluN2C), NR2D (GluN2D; van Zundert et al., 2004). The NMDA-GluN1 receptor is mainly expressed within glutamatergic synapses and mediates synaptic NMDAR signaling, while the NMDA-GluN2B receptor is localized at extrasynaptic sites (Rumbaugh and Vicini, 1999; Tovar and Westbrook, 1999; Köhr, 2006; Flores-Soto et al., 2013). Accordingly, it was hypothesized that glutamate derived from astrocyte EAAT blockade may target extrasynaptic GluN2B receptors (Fellin et al., 2004; Fig. 9). To test this hypothesis, animals were administered a selective GluN2A (NVP-AAM077) or GluN2B (ifenprodil or Ro-25-6981) receptor antagonist. We found that GluN2B antagonists blocked TFB-TBOA's action on cocaine self-administration, while the GluN2A antagonist did not, implicating NMDA-GluN2B receptor involvement in glutamate-mediated inhibition of cocaine self-administration. This finding may well explain why systemic or intra-NAc local administration of NMDAR antagonists is rewarding and potentiates cocaine's rewarding effects (Carlezon and Wise, 1996a,b; Ranaldi et al., 1996; Pierce et al., 1997) since blockade of NMDA-GluN2B receptor may disinhibit glutamate-mediated suppression of cocaine reward. It is important to note that ifenprodil alone, at a higher dose (10 µg/side), significantly inhibited cocaine self-administration in the absence of TFB-TBOA, while Ro-25-6981, a highly selective GluN2B antagonist, did not. This may be related to the somewhat poor selectivity of ifenprodil for GluN2B over GIRK channels, alpha1 adrenergic, serotonin and σ receptors (Chenard et al., 1991; Williams, 1993; McCool and Lovinger, 1995; Kobayashi et al., 2006; Karakas et al., 2011; Ishima and Hashimoto, 2012). As such, ifenprodil's off target actions likely underlie the reduction in cocaine self-administration.
Diagram showing the astrocyte EAAT-glutamate-extrasynaptic NMDA-GluN2B (NR2B) receptor mechanism underlying TFB-TBOA modulation of cocaine reward and relapse. TFB-TBOA is a highly potent and selective astrocytic EAAT1/2 inhibitor. Chronic cocaine administration causes a reduction in basal glutamate transmission and NMDA-GluN2B up-regulation in NAc DARPP-32+ dopaminoceptive neurons. Intra-NAc local administration of TFB-TBOA blocks glutamate reuptake causing an increase in extracellular glutamate levels. Glutamate then activates extrasynaptic NMDA-GluN2B receptors, which subsequently antagonizes cocaine-induced or DA-induced reduction in DARPP-32+ dopaminoceptive neurons (D2-MSNs). This causes a reduction in cocaine reward in the presence of cocaine and an increase in relapse to cocaine seeking in the absence of cocaine. NR2A (GluN2A); NR2B (GluN2B).
Next, we sought to determine which type of neurons express GluN2B, and therefore, may underlie the pharmacological actions of TFB-TBOA in cocaine self-administration. The rewarding effects of cocaine are mediated primarily by inhibition of the DAT in the mesolimbic DA system, which subsequently elevates extracellular DA and activates DA receptors in downstream postsynaptic neurons (Kreek et al., 2012). We hypothesized that NMDA-GluN2B receptors expressed in dopaminoceptive neurons in the mesocorticolimbic DA system may underlie glutamate-mediated inhibition of cocaine reward. Dopaminoceptive neurons in the NAc, DS, olfactory tubercle, and amygdala receive dopaminergic and glutamatergic inputs and highly express DARPP-32 (Svenningsson et al., 2004; Fernandez et al., 2006). As such, we focused on GluN2B expression of DARPP-32+ neurons within the NAc. Cocaine self-administration did not alter GluN2B total density in the NAc core and shell in rats, consistent with our previous work (Yang et al., 2017). However, by examining the cellular distribution, we found that GluN2B expression was upregulated in DARPP-32+ neurons in subjects exposed to cocaine self-administration, but not in yoked saline controls (Fig. 7), suggesting that activation of NMDA-GluN2B receptor on DARPP-32+ neurons in the NAc (and also possibly in other DA projecting regions) may underlie glutamate modulation of cocaine self-administration and reward.
It is unknown precisely how GluN2B receptors in DARPP32-expressing neurons modulate cocaine reward. In the striatum, DARPP-32 is expressed in both D1-expressing and D2-expressing MSNs. Recent optogenetic studies indicate that activation of NAc D1-MSNs is critically associated with positive reinforcement, while activation of NAc D2-MSNs are mostly associated with aversion or reward-attenuation (Lobo et al., 2010; Kravitz et al., 2012), although not all evidence supports this dichotomy (Soares-Cunha et al., 2016; Natsubori et al., 2017). Accordingly, it is hypothesized that the acute rewarding effects of cocaine are most likely mediated by activation of D1-MSNs via Gs-coupled D1 receptors and inhibition of D2-MSNs via Gi-coupled D2 receptors (Lobo et al., 2010; Kravitz et al., 2012; Yawata et al., 2012; Smith et al., 2013). Based on this D1- versus D2-MSN hypothesis, we speculate that GluN2B upregulation may occur mainly in D2-MSNs following cocaine self-administration. Thus, activation of GluN2B by glutamate would counteract DA-mediated inhibition of D2-MSNs, and therefore, attenuate cocaine or DA reward. More studies are required to further address this cell type-specific neural circuitry underlying glutamate modulation of drug reward.
Very little work has focused on glutamate-induced inhibition of cocaine reward as discussed above. However, glutamatergic activity during cocaine reinstatement has been well described. Indeed, elevation in NAc extracellular glutamate is a key neurobiological feature of cocaine-primed reinstatement (for review, see Uys and LaLumiere, 2008; Knackstedt and Kalivas, 2009; Schmidt and Pierce, 2010). To determine whether an increase in extracellular glutamate caused by TFB-TBOA produces similar relapse-like effects as cocaine-priming, TFB-TBOA was microinjected into several brain regions in rats after extinction from cocaine self-administration. We found that microinjections of TFB-TBOA into the NAc, VTA and VP, but not the DS, dose-dependently reinstated robust cocaine-seeking behavior. Notably, intra-NAc TFB-TBOA-induced reinstatement was remarkably long-lasting, suggesting that glutamate in the NAc is a sensitive trigger or risk factor in precipitating relapse to drug-seeking behavior. These findings are overall consistent with prior work demonstrating that cocaine-enhanced glutamate release in the NAc, but not other brain regions, is vital for cocaine primed reinstatement (McFarland et al., 2003; Xi et al., 2006, 2010; Knackstedt and Kalivas, 2009; Li et al., 2010, 2018; Gipson et al., 2013; Shen et al., 2014a). We also looked at the receptor mechanisms underlying TFB-TBOA-induced reinstatement of cocaine seeking. We found that both NMDA and AMPA receptor antagonists are able to block the effects of TFB-TBOA on reinstatement. This parallels prior work in which antagonism of AMPA or NMDA (including GluN2A or GluN2B) receptors attenuated cocaine-induced and cue-induced reinstatement (Cornish and Kalivas, 2000; Bäckström and Hyytiä, 2006, 2007; Ping et al., 2008; Gipson et al., 2013). These findings suggest that different glutamate receptor mechanisms underlie glutamate modulation of drug reward versus relapse. Unexpectedly, TFB-TBOA, when locally administered into the VTA or VP, but not the DS, also reinstated cocaine-seeking behavior, indicating that glutamate in multiple brain regions can trigger relapse to drug-seeking.
It is important to note that the above finding in which NMDA antagonism blocks TFB-TBOA-mediated reinstatement conflicts with previous reports demonstrating that NMDA antagonists (MK-801, AP5, LY235959) can reinstate drug-seeking behavior in rats after extinction from cocaine self-administration (De Vries et al., 1998; Park et al., 2002; Famous et al., 2007; D'Souza and Markou, 2014). The precise reasons underlying this discrepancy are unclear. Given the presence of multiple functional NMDAR subtypes as described above, it is likely that different NMDAR subtype mechanisms may underlie actions produced by NMDA antagonists in reinstatement of drug seeking in the presence or absence of cocaine priming. In addition, previous studies have shown that the NAc core versus shell subregions are differentially involved in drug reward versus reinstatement of drug-seeking (Peters et al., 2008; Knackstedt and Kalivas, 2009; Knackstedt et al., 2010b). In the present study, the major focus was on the functional role of extracellular glutamate in drug reward and our microinjections primarily targeted to the NAc-shell. It is possible that blockade of NMDARs in the shell versus core may also differentially alter reinstatement responses to cocaine or NMDA antagonist alone. Further work is required to address this issue.
In conclusion, the present study indicates that elevation of extracellular glutamate differentially modulates drug reward versus relapse to drug seeking, inhibiting cocaine reward during cocaine self-administration, but potentiating reinstatement of cocaine seeking. This reward-attenuating effect is mediated by glutamate activity at extrasynaptic NMDA-GluN2B receptor subtypes on NAc dopaminoceptive neurons, while the reinstatement response is mediated by activation of AMPA and NMDARs. These findings provide novel insight into the role of glutamate in drug reward versus relapse to drug-seeking behavior.
Footnotes
This work was supported by the Medication Development Program, Intramural Research Program, National Institute on Drug Abuse Grant DA000633-01.
The authors declare no competing financial interests.
- Correspondence should be addressed to Zheng-Xiong Xi at zxi{at}mail.nih.gov