Conditioned inhibition of amphetamine sensitization

: Repeated intermittent exposure to psychostimulants, such as amphetamine, leads to a progressive enhancement of the drug’s ability to increase both behavioral and brain neurochemical responses. The expression of these enhancements, known as sensitization, can be regulated by Pavlovian conditioned stimuli. Cues that are associated with drug experience can facilitate sensitization so that it only occurs in the presence of these stimuli (context-specific sensitization). In contrast, cues that are explicitly related to the absence of drugs (conditioned inhibitors) can prevent the expression of sensitization. We hypothesized that disrupting conditioned inhibition would enable amphetamine sensitization in new contexts. Using male Sprague Dawley rats and a two-context amphetamine conditioning procedure, we found that extinguishing amphetamine experience in one environment led to the loss of conditioned inhibition in a separate context. Thus, amphetamine-induced sensitized locomotion, as well as both enhanced dopamine and glutamate neurotransmission in the nucleus accumbens, were observed in a context where the drug was never experienced before. A similar loss of contextual control of sensitization was seen after using baclofen/muscimol microinjections to transiently inhibit the medial prefrontal cortex, basolateral amygdala, or ventral subiculum of the hippocampus. In other words, compared to control infusions, these intracranial injections of GABA-receptor agonists were able to block conditioned inhibitors from preventing the expression of sensitized locomotion. Together, these findings reveal the importance of conditioned inhibitors for regulating addiction-like behavior. The results suggest that dopaminergic and glutamatergic brain circuitry controls the context-specific expression of amphetamine sensitization


Introduction:
Addictive drugs cannot be administered in isolation as there are always cues present that become associated with the act of drug-taking and the drug's biopsychological effects. The process of learning about drug-related stimuli (associative conditioning) is controlled by various neurotransmitters and brain regions, including dopamine (DA) and glutamate (Glu) signaling in the nucleus accumbens (McEntee and Crook, 1993;Everitt et al., 1999;Robbins and Murphy, 2006;Shiflett and Balleine, 2011;Singer et al., 2016a;Cox and Witten, 2019). There are also non-associative processes that develop alongside cue-reward learning. For example, drug self-administration and repeated intermittent psychostimulant injections result in the non-associative sensitization of the drug's effects, wherein the ability of the drug to increase psychomotor function, as well as DA and Glu levels in the NAc, are enhanced (Pierce et al., 1996;Singer et al., 2009;Calipari et al., 2013;Carr et al., 2020;Allain et al., 2021). Thus, while distinct molecular pathways underlie drug conditioning and sensitization (Singer et al., 2014), both mechanisms also rely on neurotransmission in the NAc.
Conditioning and sensitization interact to mediate drug-related behaviors. It has been shown, for example, that following the induction of sensitization, animals primarily express sensitized movement in the presence of drug-associated cues (Stewart and Vezina, 1991;Anagnostaras and Robinson, 1996;Anagnostaras et al., 2002) and that motivation to selfadminister drug selectively occurs in the drug-paired context (Cortright et al., 2012). Similarly, increases or decreases of drug-induced striatal DA levels, in the presence or absence of drug cues, may influence motivation to pursue drug (Boileau et al., 2006;Vezina and Leyton, 2009;Leyton and Vezina, 2013). Understanding why this occurs may facilitate the development of psychological or pharmacological strategies for treating substance use disorders.
Memories may control the expression of sensitization in at least two ways: by (a) facilitating the expression of amphetamine sensitization in the presence of drug-cues (Anagnostaras and Robinson, 1996), and (b) preventing the expression of amphetamine sensitization in the presence of cues that predict the absence of drug (conditioned inhibition; Rescorla, 1969). It is challenging to determine the individual contributions of facilitation and inhibition to context-specific sensitization because conditioned inhibition appears to operate only to the extent that an excitatory representation of the drug is available somewhere . The excitatory representation can take various forms; for example, it may be defined as a drug experience in the presence or absence of specific cues.
Removing the excitatory representation of reward can result in sensitized responding being expressed in the presence of cues that previously functioned as conditioned inhibitors. In one early study, drug memories (presumably the 'excitatory representation' of the drug) were disrupted by shock, resulting in sensitized behavior in a context that had been previously unpaired with the drug (Anagnostaras et al., 2002). Another study used extinction procedures in every context to abolish the excitatory representation of drug; this again abolished contextual control over sensitization, which was observed in both drug-paired and drugunpaired environments (Stewart and Vezina, 1991; see Figure 1, Phase 2). The present experiments take a more nuanced approach, using an extinction procedure shown in fear conditioning experiments to selectively deactivate conditioned inhibition (Fowler et al., , 1991Lysle and Fowler, 1985). Accordingly, we extinguished the excitatory representation of drug in a specific environment for one group of animals (referred to as 'Unpaired-Disinhibited'; see methods and Figure 1). We hypothesized that, by disabling conditioned inhibition, these 'Unpaired-Disinhibited' animals would express sensitized responses in an environment that was previously 'unpaired' with drug. Therefore, we predicted that amphetamine-evoked behavior and neurotransmission in Unpaired-Disinhibited rats would resemble the sensitized responses of animals for which the context was always 'Paired' with the drug.
Both DA and Glu have been implicated in these excitatory and inhibitory processes. When animals are trained to self-administer psychostimulants, in the NAc these neurotransmitters are elevated in the presence of cues that signal reward availability, but are diminished when cues predict the absence of drug (e.g., research using discriminative stimuli; Weiss et al., 2000;Suto et al., 2013). It is possible, therefore, that conditioned inhibitors can decrease DA and Glu signaling in the NAc and that this might limit amphetamine's ability to evoke a sensitized behavioral response. The present experiments tested this possibility by assessing the effects of selectively weakening the conditioned inhibitory valence of a drugunpaired context. We anticipated that this would diminish the ability of such a drug-unpaired context to inhibit sensitized dopaminergic, glutamatergic, and behavioral responding to the drug.
While the NAc receives its DA inputs from the midbrain ventral tegmental area (Kalivas and Duffy, 1993), it also receives Glu afferents from several cortical sites, including the basolateral amygdala (BLA), the prefrontal cortex (PFC), and the ventral subiculum (vSUB) of the hippocampus (Britt et al., 2012). All three of these sites are essential for regulating the expression of conditioned behaviors and storing information related to conditioned cues (Gabriele and See, 2010;Stefanik and Kalivas, 2013;McGlinchey et al., 2016;Villaruel et al., 2018;Caballero et al., 2019;Otis et al., 2019). We used a transient pharmacological inhibition procedure to investigate the contribution of each of these cortical sites to conditioned inhibition of sensitized locomotor responding to amphetamine. Thus, the present experiments sought to unambiguously demonstrate conditioned inhibition of sensitized responding by using a behavioral extinction procedure to selectively deactivate it and a pharmacological inhibition approach to explore its neuronal circuitry.

Materials & Methods:
2.1 Subjects: Adult male Sprague-Dawley rats were used (n=126; Harlan, Madison, WI). Given the known sex differences in learning, motivation, and drug responses, and the ensuing likely divergent biology underlying qualitatively similar behaviors (see review by Daiwile et al., 2022), it is crucial that potential differences between male and female animals be investigated. Limitations in funding and shared facilities prohibited the inclusion of female rats in the present experiments. This work thus constitutes an initial step in exploring the contribution of conditioned inhibition to the expression of sensitization. While currently guiding these efforts, it awaits the complete neurobehavioral investigation of female rats. This work is ongoing.
In the present study, rats weighed 275-300g at the time of surgery or first injection. They were individually housed in a reverse cycle room, with food and water available ad libitum. All testing occurred during the dark phase of the cycle and was conducted according to an approved Institutional Animal Care and Use Committee (IACUC) protocol.

Behavioral Testing Equipment:
Two distinct test chambers were used: locomotor activity boxes in Experiments 1 and 3, and microdialysis chambers in Experiment 2. The locomotor activity boxes (22 x 43 x 33 cm) were constructed of opaque plastic with a tubular stainless-steel ceiling and floor. Two photocells, positioned 3.5cm above the floor and spaced 13cm apart along the longitudinal axis of the box, estimated horizontal locomotion. Interruptions of photo beams were recorded by a computer. Only movements longer than 0.5 seconds were recorded. The microdialysis chambers (38×32×34 cm) were constructed of the same materials described above for the locomotor activity boxes and were used to collect microdialysis samples on tests for sensitization.

Drugs:
D-amphetamine sulfate (Sigma-Aldrich) was dissolved in 0.9% NaCl and administered via intraperitoneal (IP) injections at a dose of 1.0mg/kg during conditioning and 0.75mg/kg during test sessions. Drugs used in the surgical preparation of animals, including ketamine, xylazine, and pentobarbital, were obtained from the University of Chicago Animal Resource Center.

Procedures:
Three separate experiments were conducted (n=126). We first used an extinction procedure to assess the effects of conditioned inhibition on sensitized locomotor (Experiment 1; n=23) and neurotransmitter (Experiment 2; n=35) responses to amphetamine. In Experiment 3 (n=68), we used pharmacology to determine how three cortical regions contributed to conditioned inhibition of the sensitized locomotor response.
The initial drug exposure phase was the same for all three experiments (Figure 1; Phase 1). It consisted of five three-day blocks. All rats received injections in the test chamber on the first day, injections in the home cage on the second, and were left undisturbed on the third. Rats in one group (Paired) received amphetamine in the test box and saline in the home cage. In contrast, rats in a second group (Unpaired) received saline in the test box and amphetamine in the home cage. Thus, the test box was unpaired with amphetamine in these rats, a procedure known to imbue this stimulus complex with conditioned inhibitory properties. Rats in a third group (Control) received saline in both environments. In all three experiments, subsequent testing (Phase 2) started the day after drug-exposure (Phase 1). There are three primary groups of rats (Paired, Unpaired, and Control) that are given IP injections of amphetamine ('A') or saline ('S') in either a test environment (squares) or their home cage (hexagons). All Experiments undergo similar 'Phase 1' testing: a three-day procedure that is repeated five times (rats are left alone ('X') in the home cage on each 'Day 3'). For Phase 2, rats are divided into different groups for Experiments 1-3. For comparison to previous research, the dashed arrow indicates how rats were studied in Stewart & Vezina (1991), where Unpaired rats were given saline in every environment. The right panel shows the Phase 3 tests for Experiment 1 (locomotion in the test environment), Experiment 2 (microdialysis for dopamine (DA) or glutamate (Glu)), and Experiment 3 (microinjections; saline (SAL) and baclofen/muscimol (BAC/MUS) into the basolateral amygdala (BLA), medial prefrontal cortex (mPFC), or ventral subiculum (vSUB)). For Phase 2, no differences were seen for Control rats regardless of whether they received saline in the Test or Home contexts (their data were combined).

Experiment 1:
Following drug-exposure (Section 2.4), rats were subjected to a second phase of testing ( Figure 1; Phase 2; Groups P1, U1, UD1, C1). Rats received six injections, one injection administered every other day. Paired rats (P1; n=5) were repeatedly administered saline in their home cage (their usual treatment in the drug exposure phase). Similarly, Control rats (C1; n=6) continued to receive saline in the locomotor activity box or home cage. Crucially, while some Unpaired rats (U1; n=7) continued to receive saline in the locomotor activity box, rats in a subset of this group (Unpaired-Disinhibited; UD1; n=5) were administered saline injections in the home cage (where they usually received amphetamine). The rationale behind this procedure was to remove the excitation afforded by the amphetamine (the unconditioned stimulus, US) experienced in the home cage, thereby diminishing the ability of the locomotor activity box stimulus complex to act as a conditioned inhibitor in these rats (Fowler et al., , 1991Lysle and Fowler, 1985). Thus, rats in four groups were tested. Notably, only rats in one group (Unpaired-Disinhibited) were subjected to extinction.
The day following this phase, all rats were tested for their locomotor response ( Figure  1; Phase 3), first to saline (conditioning test) and then amphetamine (test for sensitization). All rats were first administered saline (1.0ml/kg, IP) and immediately placed into the activity boxes. After one hour, all animals were administered amphetamine (0.75mg/kg, IP) and immediately returned to the activity boxes for two hours.

Experiment 2:
Intracranial (IC) implantation surgery was performed before Phase 1 drug exposure, 4-7 days after rats arrived. Briefly, rats were placed in a stereotaxic instrument with the incisor bar 5.0mm above the interaural line and implanted with a unilateral guide cannula (CMA11; Harvard Apparatus) aimed at the NAc (A/P: +3.6 mm; M/L: ±1.5 mm; D/V: −6.5 to −8.5 mm from the surface of the skull). The D/V coordinates refer to the placement of the active portion of the microdialysis probe (for dialysate measurements, NAc core and shell were not differentiated due to the size of the probe). Cannulae were angled laterally at 10° from the vertical and secured with dental acrylic anchored to skull screws. After surgery, CMA11 obturators were inserted flush to the cannula tips. After surgical recovery (5-7 days), rats were acclimatized to later microdialysis procedures by connecting their headcaps to the spring tether and swivel in the microdialysis chamber, but without lowering the microdialysis probe into the NAc. Rats next underwent amphetamine conditioning (five 3-day blocks; Section 2.4; Figure  1, Phase 1). As in Experiment 1, rats were then subjected to a second phase designed to extinguish conditioned inhibition in Unpaired-Disinhibited rats (UD2; n=6) as outlined above ( Figure 1, Phase 2; Groups P2, U2, UD2, C2). The remaining n/group were 10, 10, and 9 for the Paired (P2), Unpaired (U2), and Control (C2) groups, respectively.
The evening before testing in the microdialysis chambers, rats were briefly immobilized with isoflurane, and a concentric microdialysis probe lowered into the NAc via the guide cannula (CMA11; 2mm active membrane length protruding from the tip of the guide cannula). Rats were connected via a steel spring tether to a liquid swivel and collection vial positioned outside each chamber. Probes were perfused with a modified Ringer's dialysate (145mM Na+, 1.2mM Ca2+, 2.7mM K+, 1.0mM Mg2+, and 150mM Cl−; pH = 7.4; 0.3μl/min overnight and 1.5μl/min during testing the following day).
Paired, Unpaired, and Control rats remained in the microdialysis test chamber overnight (with food and water freely available). Thus, little evidence of a conditioned response was expected on the test day in Paired rats (this was not a focus of the present experiments). Unpaired-Disinhibited rats were placed in the home cage overnight (with food and water freely available) and were moved to the microdialysis chamber before testing the next day. Housing Unpaired-Disinhibited rats in the home cage was done to avoid having these rats sit in the microdialysis test chamber overnight with no amphetamine, thereby possibly reinstating the inhibitory valence of this environment (i.e., this is the place where the unpaired rats did not get amphetamine). This is in fact what happened in a control experiment in which Unpaired-Extinguished rats were placed overnight in the test chamber: these rats subsequently showed conditioned inhibition on the test ( Figure S2). Neurotransmission was measured in the microdialysis chambers for all rats on the test day (Figure 1; Phase 3). Microdialysis samples were collected in 20-min intervals in awake, freely-moving animals. Three baseline samples were collected (B1, B2, B3). All rats were then injected with saline (1.0ml/kg, IP), and three additional samples were collected (S1, S2, S3). Finally, all rats were administered amphetamine (0.75mg/kg, IP), and nine more samples were collected (A1-A9).
Dialysate samples (30μl) were split into two 15μl aliquots that were immediately frozen at -80 o C for later HPLC assessment of DA and Glu levels. For DA, samples were added to a 0.04M sodium acetate mobile phase and injected onto a 10-cm column packed with 3μm C-18 ODS particles maintained at 35°C. DA was detected with a Coulochem detector (conditioning cell at +300 mV; analytical cell electrodes at -50 and -350mV) in line with a dualpiston Shimadzu pump set at 1.0ml/min and a diaphragm type pulse dampener. Glu content was assayed by Dr. Zhi-Bing You at the Behavioral Neuroscience Research Branch of the National Institute of Drug Abuse (You et al., 2001).
Tests for amphetamine sensitization began the next day (Figure 1; Phase 3). All rats were subjected to two amphetamine sensitization tests three days apart and administered in counterbalanced order. On one test, saline was microinjected IC, whereas on the other test, a cocktail of baclofen (GABA-B agonist; 0.3nmol/side) and muscimol (GABA-A agonist; 0.03nmol/side) was infused. Both control and baclofen/muscimol microinjections (Sigma-Aldrich) were performed manually (0.3μl/side over one minute with one additional minute for diffusion). This combination of GABA receptor agonists was chosen based on previous work showing that it reliably inhibits projection cells in the BLA, mPFC, and vSUB (Peters et al., 2008;Bossert et al., 2016). Rats received a challenge injection of amphetamine (0.75 mg/kg, IP) one hour after the microinjections, and their horizontal locomotor activity was measured for two hours in the activity boxes.

Histology:
After the experiment, rats implanted with chronic indwelling cannulae were administered a lethal dose of pentobarbital and perfused with saline and 10% formalin. Brains were removed and postfixed in 10% formalin. Coronal sections (40μm) were mounted onto gelatin-coated slides and subsequently stained with cresyl violet for verification of microdialysis probe and injection cannula tip placements. Only rats with the active portion of the microdialysis probe or microinjection cannula tips in the targeted site were retained for statistical analyses (Figures S3-S4).

Analysis:
All experiments were analyzed using repeated-measures between-within ANOVA with exposure group as the between factor and time as the within factor. Differences between groups were analyzed using Scheffé post hoc tests. Because multiple replications were performed in the microdialysis studies, inter-experiment variation was introduced. For this reason, ANOVAs in these experiments were conducted on data that were normalized to each rat's baseline DA and Glu levels.

Results:
The present work investigated how conditioned inhibition regulates the contextspecificity of amphetamine sensitization. We found that extinction procedures can selectively deactivate conditioned inhibition, and this disinhibited sensitized responding in a previously drug-unpaired environment.

Experiment 1:
After amphetamine conditioning and saline exposure sessions (Figure 1, Phases 1 &  2), locomotion was assessed during a test session following saline and amphetamine injections. Paired rats, who were assessed in the drug-associated test environment, displayed conditioned locomotion after saline ( Figure S1) and sensitized locomotion after amphetamine (Figure 2). In contrast, Unpaired-Disinhibited rats did not show a conditioned locomotor response in the test environment because they were never administered amphetamine in the context ( Figure S1). Instead, removing the excitatory representation of the drug via extinction procedures resulted in sensitized locomotion for amphetamine in the test environment for Unpaired-Disinhibited animals (Figure 2).
The ANOVA conducted on the test for sensitization results found significant effects of group [F(3,19)=5.71, p<0.01] and time [F(3,57)=68.11, p<0.001], and a significant group x time interaction [F(9,57)=2.53, p<0.01]. Post-hoc Scheffé comparisons found that both Paired and Unpaired-Disinhibited rats displayed significantly greater locomotion than Unpaired and Control rats in the first hour following amphetamine (p<0.001-0.05). Thus, when the excitatory representation of the amphetamine US was removed for Unpaired-Disinhibited rats, sensitization was no longer inhibited; it was rendered visible in these animals.

Figure 2: Removal of conditioned inhibition reveals sensitized locomotor responding to amphetamine in Unpaired-Disinhibited rats.
During Phase 2 of the study, Unpaired-Disinhibited rats were given saline injections in the environment where they previously experienced amphetamine (i.e., in their home cage). As a result, these animals then showed a sensitized locomotor response in the test environment, similar to Paired-group animals. Therefore, the test environment no longer provided conditioned inhibition of amphetamine-induced locomotor responding. The top bar graph shows total horizontal locomotor counts in the two hours following amphetamine, while the bottom displays the time course of this effect. Both Paired (n=5) and Unpaired-Disinhibited (n=5) rats displayed an enhanced locomotor response to amphetamine when compared to Unpaired (n=7) or Control animals (n=6). ***-*, p<0.001-0.05. Error bars represent ± SEM.

Experiment 2:
Given the importance of NAc DA and Glu in the expression of drug sensitization, we next investigated how neurotransmission was impacted by deactivating conditioned inhibition selectively in Unpaired-Disinhibited rats. Like the locomotor sensitization results, the procedure spared enhanced DA ( Figure 3) and Glu (Figure 4) responding in Paired relative to Unpaired and Control animals, while enabling these enhanced responses in Unpaired-Disinhibited rats.
No significant group differences in DA neurotransmission following saline were observed (Figure 3). In contrast, the ANOVA conducted on the results obtained after amphetamine revealed significant effects of group [F(3,31)=7.82, p<0.001] and time [F(8,248)=72.72, p<0.001], as well as a significant group x time interaction [F(24,248)=5.0, p<0.001]. Post-hoc Scheffé comparisons again found that both Paired and Unpaired-Disinhibited rats displayed significantly greater peak DA responses than either Unpaired or Control animals (20-80 minutes post-amphetamine; p<0.01-0.05). DA responses in the NAc were observed in Paired (n=10), Unpaired (n=10), Unpaired-Disinhibited (n=6), and Control (n=9) animals following saline (SAL) and amphetamine (AMPH). As described in Figure 1 (Phase 3, Exp. 2), DA was recorded while rats were in the microdialysis chamber (previously in Phase 1, only Paired rats received amphetamine in this chamber). The top bar graphs display mean DA responses, and the bottom chart shows the time course of the effects (B1-B3 are baseline measurements; S1-S3 are post-saline measurements; A1-A9 are post-amphetamine measurements). While saline had no impact on DA levels, IP amphetamine administration significantly increased DA levels in the NAc for both Paired and Unpaired-Disinhibited animals. These results provide further support for the findings of Figure 2, demonstrating that the loss of conditioned inhibition can result in sensitized behavioral and DA responses to amphetamine in contexts that were previously unpaired with the drug. Data are shown as % baseline. *-***, p<0.001-0.05, vs Unpaired and Control rats. Error bars represent ± SEM.
The saline and amphetamine challenges evoked slightly different patterns of Glu transmission in the NAc (Figure 4). Following saline, the ANOVA detected a significant effect of group [F(3,19) The saline-induced increase in Glu transmission for Paired and Unpaired-Disinhibited animals was maintained following the amphetamine injection. Only a significant effect of group was found [F(3,19)=4.96, p<0.01] and observed in five of the nine sampling times post amphetamine (p<0.001-0.05 as revealed by post-hoc Scheffé comparisons). Thus, while both Paired and Unpaired-Disinhibited rats displayed only a sensitized NAc DA response, these animals showed an increase in Glu transmission after the saline injection stimulus, suggesting a distinct contribution to conditioned responding by this neurotransmitter.

Figure 4: Removal of conditioned inhibition reveals elevated NAc Glu neurotransmission in Unpaired-Disinhibited rats.
Glu responses in the NAc were observed in Paired (n=5), Unpaired (n=6), Unpaired-Disinhibited (n=6), and Control (n=6) animals following saline (SAL) and amphetamine (AMPH). As described in Figure 1 (Phase 3, Exp. 2), Glu was recorded while rats were in the microdialysis chamber (previously in Phase 1, only Paired rats received amphetamine in this chamber). The top bar graphs display mean Glu responses, and the bottom chart shows the time course of the effects (B1-B3 are baseline measurements; S1-S3 are post-saline measurements; A1-A9 are post-amphetamine measurements). Unlike the DA results (Figure 3), following both saline and amphetamine Glu levels in the NAc were higher in both Paired and Unpaired-Disinhibited animals compared to Unpaired and Control rats. Data are shown as % baseline. *-***, p<0.001-0.05, vs Unpaired and Control rats. Error bars represent ± SEM.

Experiment 3:
Given the robust Glu response in Unpaired-Disinhibited rats in Experiment 2, we investigated how Glu projections to the NAc might modulate the conditioned inhibition of amphetamine sensitization. Accordingly, in separate groups of rats, we microinjected a cocktail of GABA receptor agonists (baclofen/muscimol) into the BLA, mPFC, or vSUB. We observed that the inactivation of any of these brain regions disinhibited sensitized amphetamine-induced locomotion in Unpaired rats ( Figure 5). This loss of context-specific sensitization mirrors the results obtained in Experiments 1 and 2, where Unpaired-Disinhibited animals displayed a previously-concealed sensitized response to amphetamine.
Separate ANOVAs were conducted for each brain region and test (following intracranial (IC) saline and IC baclofen/muscimol). As expected, and in agreement with previous results, amphetamine sensitization is context-specific under normal conditions (IC saline). Significant effects of group and time were detected following the microinjection of saline in the BLA [group : F(2,19) was also found in the latter case. Post hoc Scheffé comparisons confirmed that Paired rats displayed significantly greater amphetamine-induced locomotion compared to Unpaired and Control rats (p<0.001-0.01) following IC saline in all brain regions.
In contrast, inactivating any of the brain regions tested (BLA, mPFC, or vSUB) had a significant impact on amphetamine-induced locomotion in Unpaired rats. ANOVAs showed significant effects of group, time, and interactions following IC baclofen/muscimol in the BLA Post hoc Scheffé comparisons showed that the significant differences between Paired and Unpaired rats relative to controls were confined to the first hour of the test (p<0.001-0.05). Together, these results indicate that the BLA, mPFC, and vSUB are each individually required to regulate the context-specific nature of amphetamine sensitization. Following control microinjections of saline into either the BLA (left), mPFC (middle), or vSUB (right), only Paired-group rats showed a sensitized locomotor response to amphetamine, demonstrating the context-specific nature of this behavior. As described in Figure 1 (Phase 3, Exp. 3), locomotion following microinjections was recorded while rats were in the test chambers (not their home cages; previously in Phase 1, only Paired rats received amphetamine in the test chamber). The top bar graphs show total two-hour horizontal locomotor counts, while line graphs on the bottom display the time course of amphetamine-induced locomotion (***-*, p<0.01-0.05, Paired vs Unpaired and Control). Infusion of BAC/MUS into the specified brain regions did not affect the response to amphetamine of Paired and Control rats but disinhibited the sensitized locomotor response to amphetamine in Unpaired rats on the test for sensitization ( † † †- †, p<0.01-0.05, Paired and Unpaired vs Control). Error bars represent ± SEM.

Discussion:
Typically, following repeated intermittent amphetamine exposure, animals only express sensitized responses to amphetamine in the presence of cues that are predictive of the drug. It is widely believed that such findings occur because drug-paired contexts can facilitate sensitization (Anagnostaras and Robinson, 1996). However, sensitization may also be environment-specific because contextual cues that predict the absence of drugs can block sensitization via conditioned inhibition (Rescorla, 1969;Lysle and Fowler, 1985;Stewart and Vezina, 1991). Crucially, conditioned inhibition can only occur to the extent that the excitatory representation of the drug remains intact (Fowler et al., , 1991Lysle and Fowler, 1985). In the present studies, we manipulate this excitatory representation of the drug and provide evidence for conditioned inhibition of sensitization to amphetamine. Thus, conditioned inhibition provides yet another mechanism whereby environmental contexts can regulate the expression of sensitized behavioral (present findings; Stewart and Vezina, 1991;Vezina and Leyton, 2009;Cortright et al., 2012), biochemical (present findings;Boileau et al., 2006;Vezina and Leyton, 2009), and cellular (Singer et al., 2009(Singer et al., , 2014(Singer et al., , 2016b(Singer et al., , 2016c responding to drugs like amphetamine. In Experiments 1 and 2, the experience of receiving amphetamine in a particular environment was extinguished in a subset of rats (Unpaired-Disinhibited). This resulted in the emergence of sensitized locomotor and dopaminergic responses, and heightened glutamatergic responses to amphetamine in a context where rats had learned not to expect the drug. Thus, by removing the excitatory representation of the drug via extinction procedures, we also took away the ability of conditioned inhibitors to block sensitization. Next, because we observed elevated Glu neurotransmission following the removal of conditioned inhibition, we investigated the role of glutamatergic inputs to the NAc (Britt et al., 2012). In Experiment 3, we found that inhibiting the BLA, mPFC, or vSUB enabled sensitized responding in a context where amphetamine was never experienced before, suggesting that these regions normally contribute to conditioned inhibitory processes.
Together, the results implicate DA and Glu in the NAc in enhanced context-specific drug responses. Future work will investigate brain pathways underlying these effects, helping to determine what processes are necessary and sufficient for encoding conditioned inhibition. Such studies will be performed using both male and female rats, and will determine whether there are sex differences in the ability of conditioned inhibitory cues to regulate drug-pursuit (including self-administration).

Neurobiological Pathways
The present report suggests that connections between multiple brain regions may be crucial for mediating conditioned inhibition and, therefore, should be studied in more detail in future studies. For example, in Unpaired rats tested in the microdialysis chamber or activity box, why are amphetamine-induced DA and Glu levels usually suppressed, and how would inhibiting the mPFC, BLA, or vSUB reverse this effect to promote a sensitized locomotor response? One possibility is that DA functions as a neuromodulator of mPFC pyramidal neurons (Seamans and Yang, 2004); during conditioned inhibition, the elevation of DA levels in the mPFC due to amphetamine could inhibit these glutamatergic projections to the NAc, perhaps through DA D2 receptor signaling on pyramidal neurons. Alternatively, it could also be that activation of mPFC neurons indirectly suppresses DA neurotransmission in the NAc ; this may occur because mPFC neurons excite GABAergic cells within the ventral tegmental area (VTA), which in turn inhibit dopaminergic cells projecting to the NAc (Carr and Sesack, 2000). As observed during conditioned inhibition, the resulting reduction of DA transmission in the NAc may be reversed by inactivating the mPFC (Rouillon et al., 2008).
Bi-directional interactions between the mPFC and BLA likely control the ability of conditioned inhibitors to regulate the expression of amphetamine sensitization. First, inhibiting the BLA could decrease the excitability of mPFC pyramidal neurons; as described above, this may disinhibit dopaminergic input from the VTA to the NAc (Sotres-Bayon et al., 2012). Second, stimulation of the mPFC inhibits BLA excitatory activity indirectly through activation of GABAergic interneurons within the BLA (Rosenkranz and Grace, 2002). Consequently, mPFC afferents to the BLA may inhibit BLA Glu outputs to the NAc. Such inhibition could underlie the ability of conditioned inhibitors to decrease extracellular levels of Glu in the NAc and prevent the expression of locomotor sensitization. In our study, it is possible that enhanced Glu transmission in the NAc of Unpaired-Disinhibited rats could result from a weakened PFC-BLA pathway.
Conditioned inhibition may also rely on poorly understood GABAergic interneurons within the PFC (Wearne and Cornish, 2019). Some evidence suggests that projections from the BLA and vSUB could regulate these inhibitory PFC circuits. For example, inactivation of the ventral hippocampus results in decreased firing of interneurons within the PFC (Sotres-Bayon et al., 2012). In our experiments, reducing vSUB activity via either microinjections or potentially through extinction procedures in Unpaired rats may, therefore, allow for the mPFC to release Glu in the NAc and disinhibit the expression of sensitization. However, as discussed earlier, increasing activation of the mPFC pyramidal neurons may also decrease DA release in the NAc (Carr and Sesack, 2000; and potentially dampen the ability of the BLA to release Glu in the NAc (Rosenkranz and Grace, 2002). These points aside, it is possible that neurons that project from the BLA or vSUB to the mPFC either directly or indirectly impact the function of the PFC-NAc pathway, but not the PFC-VTA-NAc or PFC-BLA-NAc routes. It is also likely that other brain regions, such as the paraventricular thalamus, modulate the PFC's ability to influence neurotransmission in the NAc and behavioral output (Otis et al., 2019). Together, these findings suggest that the mPFC is crucial to controlling conditioned inhibition and that disruption of the mPFC can result in the loss of context-specific sensitization.

Differences in DA and Glu Neurotransmission
Deserving of further investigation is our finding that Glu, but not DA, levels in the NAc were increased following administration of saline to Paired and Unpaired-Disinhibited rats. We might not expect a dopaminergic response because (a) Paired rats were already in the dialysis chamber overnight, and thus any conditioned DA response to saline would be blunted, and (b) the saline injection was not a DA-evoking drug-associated stimulus when administered in the dialysis chamber for the Unpaired-Disinhibited rats.
There may be various reasons for increased Glu levels after saline in Paired and Unpaired-Disinhibited animals. For Paired animals, the saline injection could still evoke a conditioned Glu response. Such findings could indicate that drug-associated stimuli (i.e., saline injections) exert a stronger influence over Glu than DA in the NAc. The observed Glu response in Paired rats could indicate that this drug-associated stimulus was resistant to extinction. In contrast, this was clearly not the case for NAc DA, as there was no elevation in DA following the saline injection.
For both Paired and Unpaired-Disinhibited rats, the robust Glu response evoked by the saline injection may have masked the ability of amphetamine to also increase Glu levels. It is also possible that stress associated with the IP saline injection could enhance Glu signaling in the NAc (Campioni et al., 2009). While there was no injection-related increase in Glu for Control and Unpaired rats, amphetamine can cross-sensitize stress responses and this effect is dependent on NAc Glu (Pacchioni et al., 2007). Therefore, if Unpaired-Disinhibited animals had cross-sensitized injection-stress, and if this response lost its context-specificity (like amphetamine sensitization), enhanced stress-related Glu transmission may have been observed after the saline injection in these animals. Such Glu neurotransmission could underlie the enhanced locomotion observed in Unpaired-Disinhibited animals soon after the saline injection ( Figure S1).

Conclusions & Implications:
There is evidence for the contextual control of the expression of stimulant sensitization and enhanced drug pursuit in both people and animals. Conditioned inhibition could explain why detoxified volunteers with a cocaine use disorder have reduced striatal DA responses following a psychostimulant challenge (Volkow et al., 1997); the laboratory setting is not associated with drugs and thus functions as a conditioned inhibitor (Vezina and Leyton, 2009). Supporting this idea, if cocaine users are allowed to prepare and take the drug in the lab as they would typically, then striatal DA responses show evidence of sensitization as a function of past drug use (Cox et al., 2009), but not when drug-taking paraphernalia and other drug-taking cues are absent (Casey et al., 2014). Interestingly, drug-related cues provided in a cocaine cue video did not enable striatal DA responding in cocaine misusers, suggesting that these may not be as effective as those involved in the actual drug self-administration (Volkow et al., 2014). Regarding these findings, it is important to remember that participants in these studies have varying histories of drug use; this may impact the results. For example, individuals tested in Cox et al. (2009) likely used less drug than subjects in the Volkow et al. (1997) study. Such research nonetheless highlights the possibility of incorporating conditioned inhibitors as 'extinction reminder cues' to help patients reduce their drug use (Rosenthal and Kutlu, 2014).
Others have demonstrated that, in rats trained to self-administer psychostimulants, DA and Glu transmission in the NAc is reduced when cues predict drug unavailability (Weiss et al., 2000;Suto et al., 2013). Our results help explain why this might occur: conditioned inhibitors prevent the expression of behavioral sensitization. The findings suggest that the mPFC, BLA, and vSUB contribute to blunting drug-induced DA and Glu transmission in the NAc when an individual is exposed to cues that predict the absence of drug (i.e., conditioned inhibitors). Future work will need to investigate how specific cell types and projections contribute to conditioned inhibition, with the objective of developing therapies to stimulate these systems and reduce problematic drug pursuit for treatment-seeking individuals.