Basolateral amygdala to nucleus accumbens projections differentially mediate flexibility of sign- and goal-tracking rats

Rats rely on communication between basolateral amygdala (BLA) and nucleus accumbens (NAc) to express lever directed approach in a Pavlovian lever autoshaping (PLA) task that distinguishes sign- and goal-tracking rats. While sign-tracking rats inflexibly respond to cues even after the associated outcome is devalued, goal-tracking rats flexibly suppress conditioned responding during outcome devaluation. Here, we sought to determine whether BLA-NAc communication in sign-trackers drives rigid appetitive approach that is insensitive to manipulations of outcome value. Using a contralateral chemogenetic inactivation design, we injected contralateral BLA and NAc core with inhibitory DREADD (hm4D-mcherry) or control (mcherry) constructs. To determine sign- and goal-tracking groups, we trained rats in five PLA sessions in which brief lever insertion predicts food pellet delivery. We sated rats on training pellets (devalued condition) or chow (valued condition) prior to systemic clozapine injections (0.1 mg/kg) to inactivate BLA and contralateral NAc during two outcome devaluation probe tests, in which we measured lever and foodcup approach. Contralateral BLA-NAc chemogenetic inactivation promoted flexible lever approach in sign-tracking rats, but disrupted flexible food-cup approach in goal-tracking rats. Consistent with a prior BLA-NAc disconnection lesion study, we find contralateral chemogenetic inactivation of BLA and NAc core reduces lever, but not foodcup approach in PLA. Together these findings suggest rigid appetitive associative encoding in BLA-NAc of sign-tracking rats hinders the expression of flexible behavior when outcome value changes.


1
Introduction 32 A body of evidence suggests that sign-and goal-tracking differences predict vulnerability to 33 Substance Use Disorder (SUD) ( basolateral amygdala (BLA) and nucleus accumbens (NAc) is necessary for the acquisition and 45 expression of lever approach that classifies ST rats (Chang et al., 2012). Here we aim to determine 46 the extent to which the incentive salience process supported by BLA-NAc core communication 47 interferes with the expression of flexibility in ST rats during outcome devaluation. 48 BLA and NAc are critically involved in Pavlovian incentive learning processes including 49 second order conditioning (SOC) and outcome devaluation. SOC is a learning process that relies 50 upon the positive incentive value of the conditioned stimulus (CS), while outcome devaluation relies 51 upon the current value of the unconditioned stimulus (US) (Holland & Rescorla, 1975 impair both SOC  and lever approach (the approach response 57 characterizing ST rats), while leaving intact food cup-directed behavior (the approach response 58 characterizing GT rats) (Chang et al., 2012). Taken  relying on CS or US value. Given tracking-related behavioral differences in incentive salience 66 processing and flexibility, we hypothesize that the BLA to NAc communication drives rigid CS 67 approach in ST rats and outcome value sensitive behavior in GT rats. 68 The primary prediction of our hypothesis is that contralateral chemogenetic inactivation of 69 BLA and NAc core will make ST rats more flexible in outcome devaluation. contralateral NAc core. We excluded some rats from subsequent analyses due to poor health or 94 misplaced viral expression based on histological analysis (Figure 4), resulting in 72 rats being 95 included in our analyses. The PCA characterization completed after surgery for viral injections 96 resulted in the following number of rats in each group: ST n = 20 (mCherry n = 9 (n = 5 female, n = 4 97 male), hM4Di n = 11 ( n = 7 female, n = 4 male), GT = 22 (mCherry n = 10 ( n = 4 female, n = 6 98 male), hM4Di n = 12 ( n = 7 female, n = 5 male), and INT n = 28 (mCherry n = 18 ( n = 10 female, n 99 = 8 male), hM4Di n = 10 ( n = 3 female, n = 7 male). 100 We conducted behavioral experiments in individual standard experimental chambers (25 x 27 x 101 30 cm; Med Associates) located outside of the colony room. Each chamber was housed in an 102 individual sound-attenuating cubicle with a ventilation fan. During PLA and devaluation probe tests, 103 each chamber had one red house light (6 W) located at the top of the wall that was illuminated for the 104 duration of each session. The opposite wall of the chamber had a recessed foodcup (with photo beam 105 detectors) located 2 cm above the grid floor. The foodcup had an attached programmed pellet 106 dispenser to deliver 45 mg food pellets (catalog#1811155; Test Diet Purified Rodent Tablet (5TUL);  107 protein 20.6%, fat 12.7%, carbohydrate 66.7%). One retractable lever was positioned on either side 108 of the foodcup, counterbalanced between subjects, 6 cm above the floor. Sessions began with the 109 illumination of the red house light and lasted ~26 minutes. 110

Surgical Procedures 111
We rapidly anesthetized rats with 5% isoflurane and maintained them at 2-3% isoflurane 112 (Vetone, Boisie, ID) throughout the procedure. We maintained body temperature with a heating pad 113 during the procedure. Prior to the first incision, we administered a subcutaneous injection of the 114 analgesic carprofen (5mg/kg) and subdermal injection of the local anesthetic lidocaine (10mg/ml at 115 incision site). We secured rats in the stereotaxic apparatus (model 900, David Kopf Instruments, 116 Tujunga, CA) and leveled the skull by equating lambda and bregma in the dorsal ventral plane. We 117 lowered 10 μl Hamilton syringes (Hamilton, Reno, NV) into the brain targeting the BLA and 118 contralateral NAc core (counterbalanced) using the following coordinates: BLA: (AP -3.0 mm, ML ± 119 5.0 mm, DV -8.6 mm 0° from midline) NAc core: (AP +1.8 mm, ML ± 2.5 mm, DV -7.0 mm -6° 120 from midline) relative to bregma skull surface (Paxinos & Watson, 2007

Measurements 149
During PLA acquisition and probe tests, we collected three behavioral measurements during 150 the 10 s CS (lever) period. All behavioral measurements were automatically collected and scored via 151 MED-PC computer software (Med Associates, Georgia, VT). For foodcup and lever contacts, we 152 recorded the total number of contacts and latency to first contact for all sessions. On trials in which 153 no contact occurred, we recorded a latency value of 10s. We calculated the lever or foodcup 154 probabilities by dividing the number of trials that a lever or foodcup contact was made by total 155 number of trials in the session. 156 The criterion used for behavioral characterization of sign-and goal-tracking phenotype was 157 based on a Pavlovian Conditioned Approach (PCA) analysis (Meyer et al., 2012) determined by 158 averaging PCA scores during training sessions four and five. The PCA score quantifies the variation 159 between lever directed (sign-tracking) and foodcup directed (goal-tracking) behaviors. Each rat's 160 PCA score is the average of three difference score measures (each ranging from -1.0 to +1.0): (1)  161 preference score, (2) latency score, and (3) probability score. The preference score is the number of 162 lever presses during the CS, minus the foodcup pokes during the CS, divided by the sum of these two 163 measures. The latency score is the average latency to make a foodcup poke during the CS, minus the 164 latency to lever press during the CS, divided by the duration of the CS (10 s). The probability score is 165 the probability to lever press, minus the probability to foodcup poke observed throughout the session. 166 Sign-tracking PCA scores range from +0.33 to +1.0, goal-tracking PCA scores range from -0.33 to -167 1.0, and intermediate group PCA scores range from -0.32 to +0.32. 168

Histology 169
After completion of behavioral testing, we deeply anesthetized rats with isoflurane and 170 transcardially perfused them with 100 ml of 0.1 M PBS followed by 400 ml 4% paraformaldehyde in 171 0.1 M sodium phosphate, pH 7.4. We removed brains and post-fixed them in 4% paraformaldehyde 172 for two hours before transfer to a 30% sucrose 4% paraformaldehyde solution in 0.1 M sodium 173 phosphate for 48 hours at 4°C. We then rapidly froze them via dry ice and stored them at -20°C until 174 sectioning. We collected 50 μm coronal sections through the entire extent of the nucleus accumbens 175 and amygdala via a cryostat (Lecia Microsystems). We mounted sections on slides and verified viral 176 expression in BLA and NAc core using anatomical boundaries defined by Paxinos and Watson 177 (Paxinos & Watson, 2007) using a confocal microscope. The observer was blind to the condition and 178 behavior of each animal. 179 mean of group 1, M2 is mean of group 2, and SDpooled = √ (s12 + s22 ) / 2, which is the pooled 191 standard deviation of the two groups (Cohen, 1988). This approach allows us to interpret potential 192 sex effects that aren't appropriately powered for typical statistical analysis. We follow general 193 guidance for interpreting effect sizes where small effect d = 0.2, medium effect d = 0.5, and large 194 effect d = 0.8 or larger (Cohen, 1988), and note instances that future studies should be powered to 195 explore sex as a biological variable. 196 3

Experimental Design and Statistical Analysis
Results 197

Acquisition of Pavlovian Lever Autoshaping 198
We trained rats for five days in Pavlovian Lever Autoshaping to determine tracking groups 199 prior to outcome devaluation testing. We used a Pavlovian Conditioned Approach Index (Fig. 1A,  200 see methods for calculation) that takes into account the number of lever and foodcup contacts ( Fig.  201 1B-C), latency to contact, and probability of contact for both lever and foodcup. We analyzed the 202 lever autoshaping training data using six separate mixed-design, repeated measures ANOVAs with 203 the between-subjects factor of Tracking (ST, INT, GT) with the within-subjects factors of Session (1-204 5). In Table 1  We hypothesized that ST rats rely on BLA-NAc core to drive rigid appetitive approach. To test 215 this a priori hypothesis, we examined the extent to which BLA-NAc core contralateral chemogenetic 216 inactivation altered the preferred response ST rats during satiety devaluation tests. For ST rats the 217 preferred response is lever contacts ( Fig. 2A), while for GT rats the preferred response is foodcup 218 contacts (Fig. 2B). Notably, mCherry ST control rats showed no difference in lever contact between 219 valued and devalued tests, confirming their insensitivity to devaluation, consistent with prior reports 220 (Keefer et al., 2020;Nasser et al., 2015). ST rats expressing hm4di showed greater lever contact 221 during valued compared to devalued tests (t(10)=2.582, p=0.027), indicating devaluation sensitivity 222 in ST rats with contralateral chemogenetic inactivation of BLA-NAc core ( Fig. 2A)  NAc also produces deficits in both initial acquisition and terminal levels of lever directed behavior, 306 the preferred response of sign-tracking rats (Chang et al., 2012 impedes flexibly in ST rats, but facilitates flexibility in GT rats, suggest individual or methodological 344 differences that bias CS or US processing may account for the diverse role for BLA-NAc in inventive 345 learning processes. food cup approach is driven by male rats. While the primary objective of this study was to include 357 both sexes, not to probe sex differences, our exploratory analyses suggest that some sex effects may 358 warrant further investigation. In particular, one testable working hypotheses includes the possibility 359 that the devaluation sensitivity of lever approach that is unmasked by BLA-NAc core inactivation 360 may be sex-specific. The present approach to include and report effects for both sexes ensures we do 361 not rely solely on male rats to determine the causal role of brain circuit contributions to behavior. 362 The present work does not include the ipsilateral control group that is typical of traditional 363 disconnection designs. In brief, our work employs contralateral chemogenetic inactivation of the 364 BLA and NAc core. To demonstrate that effects are attributable to disrupted BLA-NAc core 365 communication, rather than inactivation of these two regions alone, an ipsilateral control (in which 366 communication between the structures is still possible unilaterally) is often employed. For practical 367 reasons, we were unable to include an ipsilateral control group. However, we are not the first to 368 contralaterally inactivate these regions, and a body of evidence demonstrates no effect of ipsilateral 369 disconnection of the BLA and NAc in similar tasks. Contralateral disconnection of the BLA and NAc 370 disrupts lever-directed approach in Pavlovian lever autoshaping both early and late in training. 371 Critically, ipsilateral controls performed similarly to sham lesioned rats, suggesting unilateral 372 functional communication between BLA and NAc is sufficient to support lever directed behavior 373 (Chang et al., 2012