Chemogenetic activation of mesoaccumbal Gamma-Aminobutyric Acid projections selectively tunes responses to predictive cues when reward value is abruptly decreased

Abstract

Background:

Mesolimbic circuits regulate the attribution of motivational significance to incentive cues that predict reward, yet this network also plays a key role in adapting reward-seeking behavior when the contingencies linked to a cue unexpectedly change. Here we asked whether mesoaccumbal gamma-aminobutyric acid (GABA) projections enhance adaptive responding to incentive cues of abruptly altered reward value, and whether these effects were distinct from global activation of all ventral tegmental area (VTA) GABA circuits.

Methods:

We used a viral targeting system to chemogenetically activate mesoaccumbal GABA projections in male rats during a novel cue-dependent operant Value Shifting (VS) task, in which the volume of a sucrose reward associated with a predictive cue is suddenly altered, from the beginning and throughout the session. We compared the results with global activation of VTA GABA neurons, which will activate local inhibitory circuits and long loop projections.

Results:

We found that activation of mesoaccumbal GABA projections decreases responding to incentive cues associated with smaller-than-expected rewards. This tuning of behavioral responses was specific to cues associated with smaller-than-expected rewards, but did not impact measures related to consuming the reward. In marked contrast, activating all VTA_(GABA) neurons resulted in a uniform decrease in responding to incentive cues irrespective of changes in the size of the reward.

Conclusions:

Targeted activation of mesoaccumbal GABA neurons facilitate adaptation in reward-seeking behaviors. This suggests that these projections may play a very specific role in associative learning processes.

INTRODUCTION

Altering the motivational properties of a predictive cue when the value of a reward unexpectedly changes is critical for survival. The mesolimbic system is heavily implicated in incentive motivational processes, as well as in adapting behavior when reward contingencies change (1-6). While dopamine neurotransmission from the ventral tegmental area (VTA) to the nucleus accumbens (NAc) is fundamental to both processes, γ-aminobutyric acid (GABA) neurons are also present in the VTA (7-9). The role of these neurons in regulating reward-seeking behavior remains unclear. While the VTA contains primarily GABA interneurons, a subpopulation of VTA_(GABA) neurons project to the NAc (VTA_(GABA)→NAc, (10-12)). Approximately 1/3 of the total VTA neurons that project to the NAc are GABAergic (13). Optogenetic stimulation of all VTA_(GABA) neurons, including interneurons, decreases anticipatory licking responses to reward predicting odors during a classical conditioning task (14). Combined with data demonstrating that activation of VTA_(GABA) neurons inhibit dopamine neuronal firing during reward expectancy, others suggest that VTA_(GABA) regulation of dopamine contributes to encoding reward prediction error during associative learning processes (14, 15). Specific to VTA_(GABA) projections to the NAc, activation of VTA_(GABA) terminals in the NAc enhances discrimination between a cue predicting an aversive foot shock from a non-paired cue via GABAergic inhibition of cholinergic interneurons (CINs) (16). Yet there is scant evidence that VTA_(GABA)→NAc projections play any role in responding to reward predictive cues. We and others have reported that directly activating VTA_(GABA) terminals in the NAc does not impact sucrose consumption, cue-induced anticipatory licking in mice, or operant responding to reward predictive cues in rats (17, 18).

The primary target of VTA_(GABA)→NAc projections are tonically active CINs (16). CINs “pause” when learning new cue-reward associations (16) and are active when reward-seeking is disadvantageous (e.g. during satiety, (19, 20)). It is possible that VTA_(GABA)→NAc projections are engaged when establishing or adapting to new reward contingencies, and may have less influence when associations are well-established. Here we test the hypothesis that increased VTA_(GABA)→NAc neurotransmission specifically facilitates adapting behavioral responses when the reward value associated with predictive cues abruptly changes. To test this, we chemogenetically activated mesoaccumbal GABA projections selectively during a novel cue-dependent operant task in rats, in which the magnitude of a natural reward (i.e. sucrose) associated with a predictive cue was suddenly altered (Value Shifting, VS). We compared these effects to global activation of all VTA_(GABA), which will activate GABA interneurons and projections throughout limbic and cortical brain regions.

METHODS AND MATERIALS

Subjects

Long Evans rats (Envigo, Indianapolis, IN) weighing between 290-320 grams were individually housed in 12h/12h light/dark, with lights on at 3AM. Food and water were available ad libitum. All procedures complied with the Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Care and Use Committee at the University at Buffalo.

Adeno-associated virus (AAV)

We used a combinatorial viral method to target specific neuronal subtypes including dopamine (21-23) and GABA neurons (18). Here, we co-infused EF1α-DIO-hM3D-mCherry-AAV2/10 with GAD1-Cre-AAV2/10 into the VTA (n=13, Figure 1A). Cre recombinase is needed to reorient the excitatory DREADD transgene into a translatable position relative to the EF1α promoter. The GAD1-Cre-AAV2/10 expresses Cre only in neurons that express glutamate decarboxylase 1. Co-infusion results in hM3D expression in GAD1+ GABAergic neurons (18). Notably, the population of dopamine neurons that also make GABA use a biochemical pathway independent of GAD1, and appear to be GAD1 negative (24). Control rats (n=9) were co-infused with GAD1-CRE-AAV2/10 and CMV-DIO-tdTomato-AAV2/10. Targeted activation of mesoaccumbal GABA was achieved by microinfusing clozapine-N-oxide (CNO) into the NAc, while global activation of VTA_(GABA) occured by systemic administration of CNO (Figure 1A, E). Details of surgeries and CNO treatment are in the Supplement.

Figure 1. Viral targeting of activating DREADDs to VTA GABA neurons in the incentive cue and value shifting tasks.

(A) Diagram illustrating viral targeting of VTA GABA neurons and localized activation of projections in the nucleus accumbens, (B) VTA neuron bodies (TH in green, hM3D in red) and (C) Expression of hM3D mCherry DREADD in NAc terminals. (D) Extent of virus expression within the VTA. (E) Location of the tip of the injection cannula in rats used for the VS experiments, for clarity both hM3D and tdTomato rats are shown together. (F) Schematics depicting the events in the incentive cue task the rats were trained and maintained on (left panel), the Value Shifting probe test (right panel), and the potential order of reward blocks in any given probe session (bottom panel). ml, medial mammillary nucleus, SN, substantia nigra, AAV10, AAV2/10.

Behavioral Apparatus

Operant chambers

Med-Associates operant chambers (Georgia, VT) in sound-attenuating cubicles were equipped with two illuminated nosepoke ports on either side of a liquid receptacle (i.e. reward cup) equipped with an infrared entry detector. The left nosepoke port was designated the active nosepoke, while the right port was designated as the inactive nosepoke. A white houselight and tone generator speaker were located on the opposite wall. Each chamber was also equipped with a pump to deliver 10% sucrose solution to the receptacle. During operant training, a 10 ml syringe was used to deliver ~15 μl/s 10% sucrose solution, while a 30 ml syringe was used during free drinking sessions to deliver ~45 μl/s 10% sucrose solution.

Behavioral Training

Incentive cue (IC) task

Rats were trained on the Incentive Cue (IC) task, which was modified from others and described elsewhere (18, 25-27), and in detail in the Supplement. During initial IC task training, the rats responded during a distinct 8-second audiovisual cue to obtain a ~64 μl sucrose reward, and importantly this volume was kept constant throughout the 1 hr session (Figure 1F, left).

Value Shifting (VS) Task

Once rats reached criterion performance in the IC task, we probed rats with the Value Shifting (VS) task to assess the effect of altering the reward. In this task, diagrammed in Figure 1F (right), the IC predicting reward availability remained unchanged throughout all sessions. However, we divided the 1 hr session into three 20 min blocks where a successful response to the IC resulted in the delivery of one of three fixed volumes of sucrose (16 μl, 48 μl or 128 μl), which was achieved by changing the time the pump and associated CS+ was on (1s, 3s, 8s). There were ~35 ICs in each block, and every IC in that 20 min block received the same designated volume of sucrose per successful response. The order of the three 20 min blocks, and the corresponding reward sizes, were randomized by the computer during the 1 hr session. All possible order permutations were equally represented in the VS probe tests (Figure 1F, bottom panel). Rats were maintained on the IC task (~64 μl for 1 hr, Figure 1F, left) between VS task probes, with at least 2 days between probe sessions.

Behavioral controls and histological verification

Free drinking, locomotor activity, the Decreasing Reward IC task controls and histological verification are detailed in the Supplement.

DATA ANALYSIS AND STATISTICAL ANALYSIS

For operant behavior, our primary metrics were the response ratio of active nosepoke responses to an IC (#rewarded nosepokes/#ICs), the nosepoke latency (time; T) in response to the IC (T_nosepoke - T_IC), and the latency to enter the reward cup (T_{cup entry} - T_nosepoke) during the VS probe trials. We also examined a number of secondary metrics to assess whether the treatment caused behavioral changes in unconstrained responses. These include unrewarded active nosepokes, accuracy (#rewarded nosepokes/#total active nosepokes), active nosepokes per IC (#total active nosepokes/#IC), reward cup entries per reward (#total cup entries/#rewards), and total inactive nosepokes. We adopted a sequential strategy of analysis for these experiments, which is described in detail in the Supplement.

RESULTS

VIRAL TARGETING VTA_(GABA) NEURONS

Specificity of our combinatorial viral system has been previously verified (18). Briefly, when CNO was applied, nearby spontaneously firing dopamine neurons were inhibited. Moreover, optical stimulation of GABA targeted channelrhodopsin-2 produced a GABA specific optical inhibitory post synaptic current. As previously established, and demonstrated here (Figure 1B), immunohistochemistry for tyrosine hydroxylase (TH) established that hM3D expression occurred primarily in TH- neurons. Additionally, there is robust terminal expression of hM3D in the NAc (Figure 1C), and the spread of the hM3D-mCherry DREADD through the VTA (Figure 1D).

ACTIVATION OF MESOACCUMBAL GABA PROJECTIONS DECREASES REWARD-SEEKING WHEN REWARD SIZE IS ABRUPTLY DECREASED

We assessed how activation of mesoaccumbal GABA projections regulates reward-seeking behaviors during unpredicted changes in reward contingencies during the VS probe task (Figure 2). When comparing the overall response ratio during each reward size across all treatments and groups, the response ratio was significantly decreased when the reward size decreased to 16 μl (Figure 2A). Three-way Mixed Effects analysis showed fixed effects of reward (volume) F_2,40=23.56 p<0.0001, and virus (hM3D/tdTomato) F_1,40=5.362 p=0.0258, and a virus x treatment (CNO/vehicle) interaction F_1,40=9.470 p=0.0038. Importantly, these effects were specific to rats expressing hM3D and were not present with either vehicle or in DREADD-free controls. Comparing response ratios in the first and last 5 minutes of each reward bin, we found a significant decrease only in the last 5 minutes of the 16 μl reward block compared to vehicle (Figure 2B). Three-way Mixed Effects analysis of hM3D rats showed fixed effects of reward F_2,24=7.016 p=0.004, treatment F_1,12=15.79 p=0.0018, and reward x bin (first 5 min/last 5 min) F_2,24=13.49 p=0.0001, treatment x bin F_1,12=9.335 p=0.01, and reward x treatment x bin F_2,24=6.081 p=0.0073 interactions. However, in tdTomato controls responding was not sensitive to CNO between the first and last 5 minutes of any reward volume. Yet, they remained sensitive to abrupt changes in reward volume overall (Three-way Mixed Effects analysis, fixed effects of reward F_2,18=4.020 p=0.036). Moreover, tdTomato rats reduced their responding to the cue between the beginning and end of the 16 μl reward block (reward x bin interaction F_2,18=5.222 p=0.0163).

Figure 2. Activation of VTA(GABA) terminals in the NAc reduces responses to reward predictive cues reinforced by smaller-than-expected rewards.

All data are shown as mean±SEM. (A) The overall response ratio significantly decreased after intra-NAc CNO in rats expressing hM3D compared to vehicle, but not in rats expressing tdTomato, when the reward size was 16 μl. (B) Within the first and last 5 minutes of each reward size bin, intra-NAc injections of CNO only reduced IC performance in hM3D rats at the end of the 16 μl bin. (C) Responses to individual successive pairs of ICs during the 16 μl bin highlights an adaptive response only in CNO treated hM3D rats after experiencing a change in reward contingency. Asterisks represent significant differences between CNO and VEH determined by a Holms-Sidak post-hoc test (*p<0.05, **p<0.01, ***p<0.005, ****p<0.001). The sample size for groups in (A) is found at the base of each column.

Dynamic changes in response ratio occurred on a cue by cue basis during the 16 μl reward block only after intra-accumbal CNO treatment in hM3D-expressing rats (Figure 2C, Two-way Mixed Effects analysis, fixed effects of cue pair F_11,132=4.743 p<0.0001, treatment F_1,12=14.88 p=0.0023, and a cue pair x treatment F_11,132=3.325 p=0.0006 interaction). This effect was not observed in tdTomato controls.

We next examined whether activation of VTA_(GABA)→NAc projections influences the vigor of reward-seeking in the VS probe task (Figure 3). Rats changed their latency to nosepoke to the IC relative to the reward size, and there was an overall effect of CNO treatment compared to vehicle (Figure 3A). Three-way Mixed Effects analysis showed fixed effects of reward F_2,40=26.64 p<0.0001, and treatment F_1,20=5.347 p=0.0315. However, there were no overall effects on virus type, and no significant interactions between reward, virus type, and treatment. Within the first and last 5 minutes of each reward bin, we found that nosepoke latency was significantly increased after CNO compared to vehicle only in the last 5 minutes of the 16 μl reward block (Figure 3B). In hM3D rats, there were fixed effects of reward F_2,24=9.766 p=0.0008, treatment F_1,12=8.911 p=0.0114, and a reward x bin F_2,24=8.093 p=0.0021 interaction. In tdTomato controls, a similar analysis found an overall difference in nosepoke latency depending on the reward size (Three-way Mixed Effects analysis, fixed effects of reward F_2,16=6.973 p=0.0066) and the nosepoke latency at the beginning and end of each reward block differed according to the reward (reward x bin interaction, F_2,16=4.670 p=0.0253); the CNO treatment had no effect. We found no significant differences in the latency to enter the reward cup to consume reward after a rewarded nosepoke, either as a function of reward, virus type, or treatment (Figure 3C). Further, there were no differences in reward cup latencies between the first and last 5 minutes of each reward bin across all of the variables tested (Figure 3D).

Figure 3. Activation of VTA(GABA) terminals in the NAc increases latency to respond to cues that predict lower than-expected-reward, but not latency to collect the reward.

All data are shown as mean±SEM. (A) Overall latency to nosepoke increased when the reward size was reduced to 16 μl, but there were no significant differences within each reward size. (B) Within the first and last 5 minutes of each reward size, intra-NAc injections of CNO only increased the latency to nosepoke to the IC in hM3D rats at the end of the 16 μl bin. (C) No effect of intra-NAc treatment was seen in latencies to enter the reward cup after responding to the IC. (D) No within-reward size effects on latency to enter the reward cup were seen after intra-NAc treatment. Asterisks represent significant differences between CNO and VEH determined by a Holms-Sidak post-hoc test (*p<0.05).

Finally, we examined how mesoaccumbal GABA activation during the VS probe task influenced unreinforced responses (unrewarded active nosepokes, accuracy, total active nosepokes per IC, reward cup entries per reward, and responses in the inactive nosepoke). While four of the metrics of unconstrained behavior during the VS task showed a significant effect of reward size (Figure 4, Three-way Mixed Effects analysis, fixed effects of reward, unrewarded active nosepokes: F_2,40=7.201 p=0.0021, accuracy: F_2,40=4.663 p=0.0151, active nosepokes per IC F_2,40=35.94 p< 0.0001, and reward cup entries per reward F_2,40=10.39 p=0.0002), demonstrating that rats were adapting their responses as reward sizes changed, there were no effects of virus type, treatment, or significant interactions in any group. There were no significant effects observed in inactive nosepoke responses.

Figure 4. Activation of VTA(GABA) terminals in the NAc has no effect on four possible response outcomes in the VS task.

All data are shown as mean±SEM. There were no significant effects of intra-NAc treatment on unrewarded active nosepokes (A), accuracy (the number of rewarded nosepokes per active nosepokes) (B), active nosepokes per IC presented (C), reward consummatory behavior as measured by reward cup entries per reward earned (D), or changes in discriminating between the active and inactive nosepoke (E).

GLOBAL ACTIVATION OF VTA_(GABA) NEURONS DECREASES REWARD-SEEKING REGARDLESS OF CHANGES IN REWARD SIZES

In contrast to targeted activation of the mesoaccumbal GABA projections, global activation of VTA_(GABA) neurons, including local interneurons and projections to limbic and cortical regions, uniformly reduced responding to reward predictive ICs during the VS task (Figure 5). The response ratio was significantly decreased across all reward sizes after systemic CNO treatment (Figure 5A, Three-way Mixed Effects analysis, fixed effects of reward F_2,40=16.40 p<0.0001, and virus F_1,40=11.14 p=0.0018, treatment F_1,20=28.47 p<0.0001, and a virus x treatment interaction F_1,40=18.36 p=0.0001). The effects of systemic CNO were limited to hM3D rats. Global activation decreased the response ratio within both the first and last 5 minutes across most reward bins compared to vehicle treatment (Figure 5B, Three-way Mixed Effects analysis, fixed effects of reward F_2,24=6.148 p=0.007, bin F_1,12=88.41 p<0.0001, and a reward x bin F_2,24=5.826 p=0.0087 interaction). There was a modest proportional change in response ratio with reward volume in tdTomato controls (Three-way Mixed Effects analysis, fixed effects of reward F_2,16=7.827 p=0.0043), and also observed a reward x treatment x bin interaction (F_2,16=6.173 p=0.0103). However, no clear trend was apparent and comparing individual responses between CNO and vehicle revealed no significant differences.

Figure 5. Global activation of VTA(GABA) neurons reduces responding to reward predictive cues regardless of reward size.

All data are shown as mean±SEM. (A) Overall response ratio significantly decreased after systemic CNO in rats expressing hM3D compared to vehicle and tdTomato rats across all reward sizes. (B) Within the first and last 5 minutes of each reward size, systemic CNO treatment reduced IC performance in hM3D rats at both the beginning and end of the 16 μl and 128 μl reward bin. While CNO also reduced response ratios at the beginning and end of the 48 μl bin, it was only significant at the end. No differences between treatments were seen in tdTomato controls. (C) Responses to individual successive pairs of ICs during the 16 μl bin highlights a uniform decrease in responding to cues in CNO treated hM3D rats. Asterisks represent significant differences determined by a Holms-Sidak post-hoc test (*p<0.05, **p<0.01, ***p<0.005, ****p<0.001).

A cue by cue analysis of the response ratio performance in the 16 μl reward block demonstrated that global VTA_(GABA) activation in hM3D rats produced a uniform reduction in response ratio from the beginning of the block (Figure 5C, Two-way Mixed Effects analysis, fixed effects of cue pair F_11,132=3.115 p=0.0009 and treatment F_1,12=43.11 p<0.0001). In tdTomato controls, there was no effect of CNO treatment, although there was an overall modest decrease in responding on a cue by cue basis in the 16 μl reward block (Two-way Mixed Effects analysis, fixed effects of cue pair F_11,88=1.932 p=0.0457).

Global VTA_(GABA) activation also altered other VS probe task metrics, including an overall increase in nosepoke latencies in all reward blocks in hM3D rats treated with systemic CNO (Figure 6A, Three-way Mixed Effects analysis, fixed effects of reward F_2,40=26.78 p<0.0001, virus F_1,40=8.026 p=0.0072, treatment F_1,20=6.994 p=0.0155, and a virus x treatment interaction F_1,20=7.884 p=0.0077). Subsequent analysis showed nosepoke latency was increased in hM3D, CNO treated rats compared to vehicle, between the first and last 5 minutes of each reward block (Figure 6B, fixed effects of reward F_2,24=9.226 p=0.001, treatment F_1,12=11.65 p=0.0051), although post-hoc tests revealed that this effect was not consistently statistically significant across all reward sizes. No effects were seen in tdTomato controls. Likewise, global VTA_(GABA) activation increased the latency to enter the reward cup after a successful response regardless of reward size (Figure 6C, fixed effects of virus F_1,40=6.808 p=0.0127, treatment F_1,20=13.39 p=0.0016, and a virus x treatment interaction F_1,40=21.89 p<0.0001). Reward cup latency between the first and last 5 minutes of each reward bin also increased regardless of whether it was at the beginning or end of the reward bin, and occurred during all reward sizes compared to vehicle (Figure 6D, fixed effects of treatment F_1,12=31.30 p=0.0001).

Figure 6. Global activation of VTA(GABA) neurons increases the latencies to respond to cues and collect the reward irrespective of reward size.

All data are shown as mean±SEM. (A) Overall latency to nosepoke increased with all reward sizes after systemic CNO pretreatment compared to controls. (B) Within the first and last 5 minutes of each reward size, VTA(GABA) activation in hM3D rats increased nosepoke latencies overall in both the first and last 5 minutes of each reward bin, although most of this could be attributed to the first 5 min during the 128 μl reward block. No changes were seen in tdTomato controls. (C) Latencies to enter the reward cup significantly increased in hM3D rats given systemic CNO across all reward sizes compared to controls. (D) Latencies to enter the reward cup were significantly greater in hM3D rats during the first 5 min of the 16 and 128 μl reward blocks, and the last 5 minutes of the 48 μl bin. No within-reward block effects on latency to reward were seen in tdTomato controls. Asterisks represent significant differences determined by a Holms-Sidak post-hoc test (*p<0.05, **p<0.01, ***p<0.005).

Finally, global VTA_(GABA) activation during the VS probe task influenced unrewarded active nosepokes, accuracy, total active nosepokes per IC, reward cup entries per reward, and responses in the inactive nosepoke. Similar to VTA_(GABA)→NAc projection activation, rats adapted their responses as reward sizes changed, except for responses in the inactive nosepoke (Figure 7, Three-way Mixed Effects analysis, fixed effects of reward, unrewarded active nosepokes, F_2,40=6.274 p=0.0043, accuracy: F_2,40=3.594 p=0.0367, active nosepokes per IC F_{2, 40}=12.48 p<0.0001, and reward cup entries per reward F_{2, 40}=4.049 p=0.025). However, in contrast to targeted VTA_(GABA)→NAc projection activation, global VTA_(GABA) activation decreased active nosepokes per IC across all reward blocks (fixed effects of virus F_1,40=8.002 p=0.0073, treatment F_1,20=27.61 p<0.0001, and there was a significant virus x treatment interaction, F_1,40 p=12.26 p=0.0012). No significant effects were observed in inactive nosepoke responses.

Figure 7. Systemic activation of VTA(GABA) neurons selectively impacts four measures of possible response outcomes to the IC in the VS task.

All data are shown as mean±SEM. There were no significant effects of CNO treatment on unrewarded active nosepokes (A) and accuracy (the number of rewarded nosepokes per active nosepokes) (B), while there was a decrease in active nosepokes per IC as a result of treatment across all reward sizes compared to controls (C). There were no effects of VTA-GABA activation on reward cup entries per reward earned (D), or changes in the number of inactive nosepokes (E). Asterisks represent significant differences determined by a Holms-Sidak post-hoc test (**p<0.01, ***p<0.005).

DISCUSSION

This study tested the hypothesis that VTA_(GABA) projections to the NAc facilitate adapting behavioral responses in the face of changing reward contingencies. We selectively activated mesoaccumbal GABA terminals using an activating DREADD, while changing the volume of a sucrose reward associated with a reward predictive cue. We discovered three main novel findings, first, that activation of mesoaccumbal GABA projections selectively tunes behavioral responses to predictive cues when the reward is decreased. Secondly, this attenuation in responding occurred after experience with the altered reward, and became more pronounced over the 20 min bin. Third, the tuning was specific to responding to the IC, and did not impact reward consumption. These results contrast our previous work in which activation of VTA_(GABA)→NAc projections did not affect responding when the reward volume was held constant. Additionally, we show that global activation of all VTA_(GABA) neurons (GABA projections to different regions and GABA interneurons), attenuated responding irrespective of experience with the new reward contingency and also attenuated reward consumption during the task.

Reward-seeking is multifaceted and engages multiple reward-related neural circuits (10, 29-31). Several different processes have been ascribed to mesolimbic circuitry, including evaluating the outcome of a goal-directed action against an established, learned value (14, 15). For example, negative reward prediction is a type of behavioral learning that occurs when an earned reward is worse than predicted (32), resulting in a subsequent decrease in reward-seeking. In our study, rats were well trained to respond to receive a fixed volume of reward (64 μl), but during VS probe tests the reward was abruptly decreased or increased from that obtained during training sessions. Under these conditions, mesoaccumbal VTA_(GABA) activation selectively decreased the response ratio and increased the nosepoke latency to the cue newly predicting a smaller volume of sucrose reward (16 μl). A central component of a learned process is that the changed reward outcome must be experienced before an alteration in reward-seeking behavior can occur. Indeed, in the VS probe rats initially responded to the cue predicting the smallest reward at levels identical to controls. Yet as they gained experience with the new reward contingency upon subsequent cue presentations, activation of VTA_(GABA)→NAc projections facilitated a decrease in responding to the new, less-than-expected reward. In addition, the latency to respond to the cue predicting the smallest reward increased at the end of the bin, compared to vehicle controls. Together, this suggests that activation of VTA_(GABA)→NAc projections selectively facilitated a progressive decrease in the choice to respond to the cue over time, along with a decrease in the vigor of the response (33).

Future studies will determine if the decreases facilitated by mesoaccumbal GABA activation during the VS probe task persist when the task is repeated the next day. Also, additional work will be needed to distinguish if these decreases in responding are due to violating established expectations of reward, or because of an alternative mechanism, such as potentiated reactivity to a smaller reward. It should be noted that when other rats are challenged on a Decreasing Reward IC task, where the reward always decreases systematically over time, their performance during the first session is very similar to vehicle pretreated rats challenged with the VS task (Figure S2). This suggests that in both groups, the rats are initially unaware of the changed IC-reward contingencies and adapt their behavior after experience. Indeed, given 25 repeated daily sessions, rats trained on the Decreasing Reward task showed similar reward-seeking adaptations as the mesoaccumbal GABA activated rats during a VS challenge. This suggests that the effects we report here are due to engaging experience-dependent processes rather an increased sensitivity to a smaller reward volume. Alternately, the changes in responding to the cue when rewards are abruptly decreased may be secondary to altered general arousal. However, the effects of mesoaccumbal GABA activation were limited to behaviors specific to responding to the cue itself, without changes in the latency to enter the reward cup to obtain a reward, unrewarded nosepokes, accuracy, reward cup entries, inactive nosepokes, or locomotion in an open field (Figure S1), suggesting that arousal is unlikely to account for the behavior observed here. Interestingly, neither the choice nor response vigor was increased when there was an unpredicted larger-than-expected reward. It is possible that VTA_(GABA)→NAc projections specifically contribute to forming new stimulus-outcome associations only when rewards are less than expected. However, during the development of the VS task, we discovered that the task is more sensitive in detecting behavioral flexibility in the face of a smaller-than-expected reward. Therefore, we cannot preclude the possibility the rats reached a performance ceiling during this instrumental task with sucrose reinforcement, thereby masking increases in responding with larger-than-expected reward.

To our knowledge, this is the first demonstration that VTA_(GABA)→NAc projections alter cue processing for rewards. We and others have shown that selectively activating these projections does not alter cue-induced reward-seeking (17, 18). One study demonstrated that this pathway enhances discrimination of cues predicting aversive stimuli (16). In that study, mice generalized freezing to two distinct tones, but optogenetic activation of VTA_(GABA) terminals in the NAc during conditioning resulted in the mice freezing less to the non-predictive tone, while freezing was maintained for the tone predicting the foot shock (16). In this way, the non-predictive tone was marked as less motivationally relevant. The authors determined this effect occurred through selective inhibition or “pausing” of local CINs in the NAc. Future studies examining whether our results also occur through accumbal CINs will be useful in determining if this is a common pathway for facilitating learning new cue-reinforcement contingencies regardless of valence. Additionally, future studies should delineate the functional role of VTA_(GABA) terminals in the NAc core and shell subregions. Notably, studies of VTA_(GABA) projections thus far indicate heterogenous projection patterns to these subregions (11, 34, 35) that appear distinct from dopamine.

These results substantially expand our earlier work, in which we observed no change in cue-induced responding after activating VTA_(GABA) projections to the NAc in the standard IC task, where reward size was kept constant (18). The VS task has different reward contingencies and likely engages reward-seeking processes that are distinct from the standard IC task. The VS probe differs notably from the IC task in that the reward sizes are randomly assigned to the same predictive cue within the session. To prevent rats from becoming over-trained on this VS task and to preserve the novelty of the changed reward contingencies, these sessions were conducted sparingly and always after several days of IC task sessions. Importantly, activation of VTA_(GABA)→NAc projections altered responding in the VS challenge even when our subjects had a strong pre-established cue-reward expectancy due to extensive prior training in the IC task. In contrast, global activation of VTA_(GABA) neurons decreases responding, increases nosepoke and reward latencies, and increases the active nosepokes per IC during the VS challenge, irrespective of the reward size. This supports our earlier findings that global VTA_(GABA) activation preferentially inhibits the incentive salience of the reward predictive cues (18) possibly through VTA_(GABA) interneuron inhibition of dopamine neurons, activity of VTA_(GABA) projections to other brain regions, or some combination. During the VS probe, changes in performance after global VTA_(GABA) activation are present at the beginning of each reward bin, which is a defining characteristic of incentive motivational processes, namely that changes in reward-seeking behavior occurs before experience with the reward itself (2).

Similar to our previous work, we found that targeted VTA_(GABA)→NAc activation does not change sucrose consumption under free-drinking conditions or generalized locomotion (Figure S1). This suggests that the decreases in responding to cues associated with a lower-than-expected reward did not result from a suppression of appetite or locomotor activity. Indeed, when the rats do respond in the other reward bins, they do so with similar speed, and are thus capable of emitting faster responses (Figure 3). Finally, since high doses of CNO may be converted to CZP in vivo (36), all of our behavioral metrics for both intra-NAc and systemic administration of CNO included comparisons with DREADD-free controls. CZP did not alter the behavioral metrics of the VS probe, demonstrating there is no confound from the potential metabolism of CNO to CZP in this study (Figure S1).

In summary, our data demonstrate that targeted activation of VTA_(GABA) projections to the NAc preferentially regulates associative learning related to new stimulus-outcomes, possibly underlying negative reward prediction processes. This specific behavioral consequence in a robust operant model is distinct from global VTA_(GABA) activation, which appears to preferentially regulate the attribution of incentive salience to reward predictive cues, regardless of the value of the subsequent reward. Further studies will be needed to determine the precise local circuitry within the NAc that contribute to these reward-seeking behaviors.

KEY RESOURCES TABLE

Resource Type	Specific Reagent or Resource	Source or Reference	Identifiers	Additional Information
Add additional rows as needed for each resource type	Include species and sex when applicable.	Include name of manufacturer, company, repository, individual, or research lab. Include PMID or DOI for references; use “this paper” if new.	Include catalog numbers, stock numbers, database IDs or accession numbers, and/or RRIDs. RRIDs are highly encouraged; search for RRIDs at https://scicrunch.org/resources.	Include any additional information or notes if necessary.
virus	GAD1-Cre-AAV2/10	Caroline Bass Laboratory		custom packaged in laboratory
virus	DIO-hM3D-mCherry-AAV2/10	Caroline Bass Laboratory		custom packaged in laboratory
virus	DIO-tdTomato-AAV2/10	Caroline Bass Laboratory		custom packaged in laboratory
plasmid	GAD1-Cre-pACP	Caroline Bass Laboratory		custom constructed in laboratory
plasmid	DIO-tdTomato-pACP	Caroline Bass Laboratory		custom constructed in laboratory
plasmid	pAAV-hSyn-hM3D(Gq)-mCherry	Addgene	Plasmid #50474
Antibody	Tyrosine Hydroxylase Antibody	Immunostar	RRID:AB_572268
Antibody	Goat Anti-Mouse IgG1 Antibody, Alexa Fluor 488 Conjugated	Invitrogen	RRID:AB_2535764
Antibody	Goat Anti-Rabbit IgG (H+L) Highly Cross-adsorbed Antibody, Alexa Fluor 555 Conjugated	Invitrogen	RRID:AB_2535850
Antibody	Anti-RFP (RABBIT) Antibody Biotin Conjugated Min X Hu Ms and Rt Serum Proteins - 600-406-379	Rockland	RRID:AB_828390
Chemical Compound or Drug	clozapine-n-oxide (CNO)	NIDA Drug Supply Program	NOCD-135
Commercial Assay Or Kit	Vectashield Vibrance	Vector Laboratories	H-1700
Organism/Strain	Long Evans (blue spruce) outbred rats	Envigo	HsdBlu:LE