Generation of conditional deletion of Drd2 in WFS1 neurons to parse the role of D2R in a subcircuit spanning across the EA and TS
To determine whether coordinated EA and TS D2R signaling is required for the optimization of approach and/or avoidance behaviors, we first searched for available markers/tools to delete Drd2 expression in a subcircuit spanning across the EA and TS. After close inspection, we selected a neuronal subpopulation defined by the expression of the Wolfram syndrome 1 (WFS1) protein based on its robust expression in the core nuclei defining the EA, namely the Acb core (AcbC) and shell (AcbSh), the oval nucleus of the BNST (BNSTov), the lateral part of the CeA (CEl),(Fig. 1a) as well as in the amygdalostriatal transition (AST) [33], and in the tail of the striatum (TS) (Fig. 1c). Similar patterns were observed when the distribution of tdTomato-positive neurons was analyzed in tamoxifen-inducible Wfs1-CreERT2 mice crossed with the mouse reporter line Ai14 (tdTomatoWFS1) indicating that Cre-dependent recombination triggered in this mouse line accurately reflects the endogenous WFS1 promoter activity (Figure S1).
We then analyzed the degree of overlap between the expression of WFS1 and D2R using Drd2-eGFP reporter mice (Fig. 1a, c). Consistent with our previous observation, ~ 40% of neurons co-expressed WFS1/GFP in AcbC and AcbSh [34] (Fig. 1a-b and Table S1). Double immunofluorescence analysis also revealed that a majority of D2R-containing neurons of the BNSTov (~ 69%), the CEl (~ 75%), the AST (~ 79%) and the lateroventral TS (~ 60%) were WFS1-positive (Fig. 1b, d and Table S1), a result confirmed by the presence of Drd2 mRNA in Wfs1+ neurons of the EA and TS (Fig. 1e-f). We therefore crossed Drd2loxP/loxP mice [41] with the Wfs1-CreERT2 mouse line (named thereafter Drd2WFS1-i-cKO) [34] to temporally delete Drd2 in a restricted subcircuit spanning across the EA and TS. Three weeks after tamoxifen treatment, the analysis of D2R expression by immunofluorescence confirmed the decrease of D2R levels in the Acb and TS of Drd2WFS1-i-cKO (Fig. 1g, h), indicating that conditional deletion of Drd2 in WFS1-positive neurons may represent a suitable tool to parse the role of EA and TS D2R in behaviors elicited by appetitive or aversive stimuli.
Deletion Of D2r In Wfs1 Neurons Does Not Impair Food-seeking Behavior In Fed Mice
Recent studies indicate that the EA and TS regulate appetitive behaviors [35–40]. Because D2R play an important role in motivation, we investigated whether D2R signaling in the EA-TS circuit modulates reward-seeking behaviors. Fed (non-food deprived) control and Drd2WFS1-i-cKO mice were first trained during the dark phase on a fixed-ratio 1 (FR1) reinforcement schedule for isocaloric palatable food delivery for 7 consecutive days (Fig. 2a-c). After 3 days of training, control and Drd2WFS1-i-cKO mice equally discriminated between the active (delivery of a single pellet) and inactive (no delivery) lever (Fig. 2b). Both groups similarly increased their total number of active lever presses and consumed pellets under FR1 (Fig. 2b, c and Figure S2a-b) as well as under a FR5 reinforcement schedule (Fig. 2b, c and Figure S2c-d). No differences were found between males and females (Figure S2 and S3). Next, we analyzed lever presses under a progressive ratio (PR) schedule of reinforcement during which the number of lever presses required to earn a pellet incrementally increased. On average, the highest ratio completed (breaking point) to obtain palatable pellets was similar between control and Drd2WFS1-i-cKO mice suggesting that the motivation and vigor elicited by palatable food-seeking and consumption were not impaired in Drd2WFS1-i-cKO mice (Fig. 2d-e). Altogether, these results indicate that EA-TS D2R signaling is not critical for food-seeking behavior in mice fed under an ad libitum regime.
Ea-ts D2r Signaling Regulates Homeostatic-dependent Food-seeking Behavior
The metabolic state of the animals influences the motivational drive associated with food-seeking behavior [42–45]. To determine whether changes in recent food availability may unveil a role of EA-TS D2R signaling in food-seeking behavior, we trained single-housed food-restricted male and female control and Drd2WFS1-i-cKO mice on a FR1 lever press schedule for isocaloric palatable food delivery for 5 consecutive days (Fig. 3a). As expected, the total number of active lever presses and pellets consumed increased in food-restricted control mice compared to fed control mice (Fig. 3b-c and Fig. 2b-c), an effect observed in both male and female mice (Figure S4a-b and Figure S5a-b). In contrast, both male and female Drd2WFS1-i-cKO mice showed a reduced number of active lever presses and consumed pellets on FR1 schedule compared to control mice, suggesting that food restriction has a limited impact on food-seeking behavior in mice lacking functional EA-TS D2R signaling (Fig. 3b-c, Figure S4a-b and Figure S5a-b). These differences between genotypes were maintained in both male and female mice as the cost of reward increased under the FR5 schedule (Fig. 3b-c, Figure S4c-d and Figure S5c-d). However, both control and Drd2WFS1-i-cKO mice were able to adjust their behavior in face of increasing effort requirement since the number of pellets consumed during the last session of FR1 and FR5 remained the same (Fig. 3c, Figure S5c-d). These differences were not the result of impaired ability of Drd2WFS1-i-cKO to discriminate between the active and inactive levers as the discrimination indexes (ratio active/inactive lever presses calculated during the first and last day of both FR1 and FR5) were equivalent between control and Drd2WFS1-i-cKO mice (Fig. 3d and Figure S5e-f). However, the higher ratio of active lever presses/consumed pellets observed during FR5 in control mice compared to Drd2WFS1-i-cKO mice suggest that the latter optimized more efficiently their behavior (Fig. 3e and Figure S5g-h). Indeed, only control mice gradually increased their premature impulsive responses on the active lever during the time-out period, indicating that the impulsivity traits associated to reward-seeking behaviors in food-restricted conditions rely, at least in part, on intact EA-TS D2R signaling (Figure S6).
Because D2R have been implicated in cognitive flexibility [46], we also tested male and female control and Drd2WFS1-i-cKO mice in a reversal learning task by switching the active and inactive levers (Figure S7). Despite an increased number of inactive lever presses on the first day of reversal learning (session 10, Figure S7a), control and Drd2WFS1-i-cKO mice rapidly re-learnt the new rule and efficiently readjusted their behavioral sequences (active vs inactive lever) to obtain a similar number of pellets as the last FR5 session (Figure S7a-b). Moreover, in both male and female, lever presses during the time-out period remained low in Drd2WFS1-i-cKO compared to control mice (Figure S7c). Altogether, these findings indicate that Drd2WFS1-i-cKO mice have normal cognitive flexibility and quickly adapt their food-seeking strategies to sudden changes in environmental cues.
Ea-ts D2r Signaling Modulates The Willingness And/or Vigor Of Reward-seeking Behavior
We then analyzed lever presses in both male and female control and Drd2WFS1-i-cKO mice under PR. On average, the number of lever presses as well as the highest ratio completed to obtain palatable pellets were lower in Drd2WFS1-i-cKO mice compared to control mice, an effect only observed in males (Fig. 3f and Figure S5i-j). To determine whether this phenotype could be related to decreased hedonic value of palatable food, we exposed 12-h fasted control and Drd2WFS1-i-cKO mice to high fat diet (HFD; 60% of kcal from fat) or normal chow (NC; 5% of kcal from fat) during a 2-hour choice test. We found that both male and female control and Drd2WFS1-i-cKO mice, which did not differ in body weight loss following food-deprivation (ct: 23.69 ± 0.9 g; Drd2WFS1-i-cKO: 23.093 ± 0.9 g), showed a similar tropism toward the HFD (Fig. 3g and Figure S8), and consumed the same amount of HFD (Fig. 3g and Figure S8). Taken together, our results suggest that reduced food-seeking behaviors observed in food-deprived Drd2WFS1-i-cKO mice during the operant conditioning mainly result from the use of an optimized behavioral strategy which consists in lowering the energy costs involved in obtaining food (effort vs no effort paradigms) rather than altered hunger sensitivity or food preference.
To confirm that EA-TS D2R signaling regulates homeostasis-dependent food-seeking behavior, immediately after PR, control and Drd2WFS1-i-cKO mice were maintained on an FR5 reinforcement schedule for 4 additional consecutive days while being fed ad libitum in their home cages. As expected, total lever presses and consumed pellets strongly decreased in fed control male and female mice (Fig. 3h and Figure S9a-b). However, while control male mice now performed the same number of lever presses and consumed the same number of pellets than the Drd2WFS1-i-cKO mice, control female mice showed a slight but significant reduced number of active lever presses and pellets consumed compared to female Drd2WFS1-i-cKO mice (Figure S9a-b). On the other hand, discrimination and performance indexes were equivalent between the two groups (Fig. 3i-j and Figure S9c-d). Finally, the number of active lever presses during the time-out period drastically decreased in control mice suggesting that premature responses were a direct consequence of the internal state triggered by food-restriction (Figure S10). Altogether, our findings suggest that EA-TS D2R signaling regulates homeostasis-dependent adaptation of food-seeking strategies.
Ea D2r Signaling Regulates Defensive Behaviors
Beside approaching/appetitive behaviors, the EA and the TS also regulate defensive/avoidance behaviors. We therefore assessed whether EA D2R signaling participates in the optimization of defensive behavioral strategies employed when facing a threat. To determine the role of EA-TS D2R signaling in the selection of defensive behaviors, we first evaluated whether Drd2WFS1-i-cKO mice were able to discriminate between auditory cues associated (CS+) or not (CS-) with a threat and learn how to avoid a threat by shuttling from one compartment to another during CS + presentation (Fig. 4a). During the first session, both male and female control and Drd2WFS1-i-cKO mice shuttled exclusively after the delivery of the shocks indicating that during this phase escape responses predominate (Fig. 4b and Figure S11b). Upon subsequent trials, control male mice gradually switched from escape to avoidance responses following CS + presentation. In contrast, Drd2WFS1-i-cKO male mice and female mice from both genotypes failed to do so although discriminative learning was preserved in all groups as supported by the lower number of avoidance responses following CS- presentation (Fig. 4b-c and Figure S11b-c). These differences were not the result from altered sensory systems (hearing and somatosensation) since male and female from both genotypes displayed similar auditory thresholds (Figure S12a), mechanical activity of the auditory outer hair cells (Figure S12b), auditory brainstem responses (ABR) (Figure S12c-d) as well as tactile and thermal sensitivity (Figure S13 and data not shown).
To test whether Drd2WFS1-i-cKO mice bias the selection of defensive behaviors toward passive strategies, we then analyzed freezing responses, used as a proxy of passive defensive strategy adopted to face threatening situations, in Drd2WFS1-i-cKO male mice (Fig. 4d). During the conditioning (day 1), control and Drd2WFS1-i-cKO male mice displayed equivalent freezing responses to the auditory cues associated (CS+) or not (CS-) with a threat (Fig. 4e-f). On the test day (day 2), high levels of freezing were observed in both groups following CS + presentation (one Drd2WFS1-i-cKO mouse displaying less than 20% of freezing during the CS + presentation was excluded from the analysis). In contrast, CS- presentation evoked low freezing responses in both control and Drd2WFS1-i-cKO mice (Fig. 4g). Of note, contextual discrimination learning was similar in both groups (Figure S14), further suggesting that discriminative learning does not require EA-TS D2R signaling. We then evaluated the evolution of defensive behavioral responses by repeatedly presenting the CS+ (12 times) during 2 consecutive days (day 3–4) (Fig. 4d). While the freezing responses gradually diminished over the course of CS + re-exposure in control mice, Drd2WFS1-i-cKO mice maintained high levels of freezing which were similar to those observed immediately after conditioning (Fig. 4g-h). Together, our results indicate that Drd2WFS1-i-cKO mice adapted their behavioral defensive strategy less efficiently, maintaining a high level of freezing responses when exposed to sensory stimuli associated with threats.
Ea D2r Biases Innate Defensive Behaviors Induced By A Visual Threat
Finally, we investigated whether EA-TS D2R signaling also contributes to the selection of defensive strategies in response to innate threats. To do so, freezing and/or escape responses to looming visual stimuli mimicking an aerial predator were assessed in control and Drd2WFS1-i-cKO male mice (Fig. 5a). At day 1, mice of both genotypes displayed a high level of defensive responses (escape and/or freezing), which gradually decreased across sessions (Fig. 5b). Moreover, both groups began their escape or freezing responses with similar latencies after stimulus onset (Fig. 5c). However, analysis of the independent freezing and flight responses revealed that at day 2, as well as in the subsequent test session (day 7 and 8), Drd2WFS1-i-cKO mice strongly biased their innate defensive behaviors towards freezing (Fig. 5d-e). Consequently, the number of entries into the shelter was strongly reduced in Drd2WFS1-i-cKO mice compared to control animals (Fig. 5f). Altogether, these findings indicate that EA D2R signaling favors the selection of active defensive behaviors in response to visual innate threat.