Nucleus accumbens GABAergic inhibition generates intense eating and fear that resists environmental retuning and needs no dopamine

Abstract

Intense fearful behavior and/or intense appetitive eating behavior can be generated by localized amino acid inhibitions along a rostrocaudal anatomical gradient within medial shell of nucleus accumbens of the rat. This can be produced by microinjections in medial shell of either the GABA_A agonist muscimol (mimicking intrinsic GABAergic inputs) or the AMPA antagonist DNQX (disrupting corticolimbic glutamate inputs). At rostral sites in medial shell, each drug robustly stimulates appetitive eating and food intake, whereas at more caudal sites the same drugs instead produce increasingly fearful behaviors such as escape, distress vocalizations, and defensive treading (an antipredator behavior rodents emit to snakes and scorpions). Previously we showed that intense motivated behaviors generated by glutamate blockade require local endogenous dopamine and can be modulated in valence by environmental ambience. Here we investigated whether GABAergic generation of intense appetitive and fearful motivations similarly depends on local dopamine signals, and whether the valence of motivations generated by GABAergic inhibition can also be retuned by changes in environmental ambience. We report that the answer to both questions is ‘no’. Eating and fear generated by GABAergic inhibition of accumbens shell does not need endogenous dopamine. Also, the appetitive/fearful valence generated by GABAergic muscimol microinjections resists environmental retuning and is determined almost purely by rostrocaudal anatomical placement. These results suggest that NAc GABAergic release of fear and eating are relatively independent of modulatory dopamine signals, and more anatomically pre-determined in valence balance than release of the same intense behaviors by glutamate disruptions.

Introduction

Local inhibitions in medial shell of the nucleus accumbens (NAc) of the rat produce intense appetitive and/or fearful motivated behaviors organized along a rostrocaudal gradient resembling an “affective keyboard”. That is, microinjections of either the AMPA antagonist DNQX or GABA agonist muscimol induce intense appetitive behaviors such as eating (Maldonado-Irizarry et al., 1995; Stratford & Kelley, 1997), especially at rostral sites in medial shell (Reynolds & Berridge, 2001; Reynolds & Berridge, 2002; Reynolds & Berridge, 2003). Induction of an appetitve psychological state is supported also by demonstrations that rostral muscimol microinjections induce conditioned place preference, and potentiate ‘liking’ reactions to taste (Reynolds & Berridge, 2002; Faure et al., 2010). In contrast, progressively more caudal microinjections of these same drugs produce increasingly fearful or aversive behaviors, including distress vocalizations and escape attempts elicited by experimenter’s touch, conditioned place avoidance, and spontaneous emission of defensive treading/burying behavior, an anti-predator response which rodents use to throw debris at a localized threat (i.e. rattlesnake or shock prod) (Coss & Owings, 1978; Treit et al., 1981; Reynolds & Berridge, 2001; Reynolds & Berridge, 2002; Reynolds & Berridge, 2003; Faure et al., 2010). Microinjections at intermediate locations produced graded mixtures of appetitive and fearful behaviors.

While keyboard patterns of intense motivations are generated by both methods of inhibition, GABAergic versus glutamatergic effects differ in important respects. For example, while GABAergic muscimol microinjection amplifies positive hedonic “liking” reactions to tastes at rostral sites and conversely amplifies negative disgust reactions at caudal sites (Reynolds & Berridge, 2002; Faure et al., 2010), glutamate disruption fails to change affective ‘liking’ or disgust reactions, even when DNQX microinjection generates comparably intense levels of appetitive/fearful motivated behaviors (Faure et al., 2010).

Here we were interested in comparing motivations generated by GABAergic versus glutamatergic NAc inhibitions on two other features. One feature is that endogenous local dopamine is necessary for glutamate disruptions to generate either desire or dread. That is, mixing dopamine antagonists into a DNQX microinjection prevents generation of intense eating and fear (Faure et al., 2008; Richard & Berridge, 2011). However, given that the GABAergic inhibition generates changes in hedonic impact as well as motivation, and that dopamine does not mediate the hedonic impact of tastes (Peciña et al., 1997; Berridge & Robinson, 1998; Peciña et al., 2003), one might infer that the GABAergic inhibition should not require dopamine to generate intense motivations. Is GABAergic generation of eating and fear independent of local dopamine?

Another feature of the glutamate keyboard is that the ratio of appetitive to fearful behaviors is retuned by current environmental ambience, especially at intermediate rostrocaudal “keys”. In a familiar and comfortable home environment, DNQX microinjections at more than 80% of the generates positive appetitive behavior while the fear-generating zone shrinks to a far-caudal 20%; but in a stressfully loud and bright environment DNQX instead generates mostly fearful behavior at 80% of sites (Reynolds & Berridge, 2008; Richard & Berridge, 2011). Therefore, we also asked: Is GABAergic generation of appetitive and fearful motivations similarly retuned by environmental ambience?

Materials and Methods

Experimental Design

We assessed appetitive and fearful behavior generated by GABAergic inhibition via muscimol microinjections to assess the contribution of endogenous dopamine signaling, and to detect any influence of changes in emotional ambience of the environment. We assessed the role of endogenous dopamine by combining a D1 antagonist and D2 antagonist in the same microinjection with GABA_A agonist muscimol. We separately assessed the influence of emotional ambience by comparing motivations generated by muscimol microinjections or DNQX microinjections in three different environments: “Stressful” (noisy and bright), “Standard” lab, and “Home” (a rat’s own home-room: dark, quiet, familiar odor). In order to estimate the area of functional drug spread and to assess the neurobiological impact of dopamine blockade on GABAergic neuronal suppression, a Fos plume analysis was conducted on a separate group of rats.

Subjects

Male Sprague-Dawley rats [280 – 400 grams, total n = 95; dopamine dependence group, n = 18; environment shift groups, n = 59; Fos Plume group, n = 18] from an in-house breeding colony were housed on a reverse 12:12 light:dark cycle at ~21 °C with ad libitum access to both food and water. All experimental procedures were approved by the University Committee on the Use and Care of Animals at the University of Michigan and were carried out in accordance with the guidelines on animal care and use of the National Institutes of Health of the United States.

Cranial cannulation surgery

All rats were anesthetized using a mixture of ketamine (80 mg/kg) and xylazine (5 mg/kg) and pretreated with atropine (0.05 mg/kg) to prevent respiratory distress. After anesthesia induction, rats were placed in a stereotaxic apparatus (David Kopf Instruments), with the mouth bar set to 5.0 mm above intra-aural zero, so that cannulae could be angled to avoid puncturing the lateral ventricles. Bilateral stainless steel cannulae (14 mm, 23 gauge) were aimed 2 mm above points throughout the rostrocaudal extent of the medial shell, between coordinates anteroposterior (AP) +2.4 to +3.4 mm ahead of bregma, mediolateral (ML) +/− 1.0 mm from the midline, and dorsoventral (DV), +5.7 – 6.0 mm below skull. Cannulae were anchored to the skull using surgical screws and secured with dental cement; stainless steel obturators (28 gauge) were insert to prevent occlusion of the cannulae. Post-surgery, rats received cefazolin (75mg/kg) to prevent infection and carprofen (5 mg/kg) for analgesia. Rats were allowed to recover for 7 days before testing.

Drugs and intracerebral microinjections

Drug microinjections were always administered bilaterally in a 0.5 μl volume on test days spaced at least 48 hours apart. On test days, solutions were brought to room temperature (~21°C), inspected to confirm the absence of precipitation, and infused at a speed of 0.3 μl per minute using a syringe pump attached via PE-20 tubing to stainless steel injectors (16 mm, 29 gauge) which extended 2 mm beyond the end of the guide cannulae into medial shell. Injectors were left in place for 1 minute, and then obturators were replaced and rats were immediately placed in one of the testing chambers described below.

GABA/dopamine interaction experiment

For GABAergic muscimol microinjections in medial shell, rats (n=18) received a dose of muscimol (75 ng/0.5 μl per side) previously shown to generate intense eating and fear (Faure et al., 2010). Muscimol was microinjected either alone or, to test the role of endogenous dopamine, in combination with two dopamine D1 and D2 antagonists as a mixture in the same microinjection: the selective D1 antagonist SCH23390 R(+) – 7-chloro-8-hydroxy-3-methyl1-phenyl-2,3,4,5,-tetrahydro-1H-3-benzazepine) at a dose of 3 μg/0.5 μl per side, plus the selective D2 antagonist raclopride (3,5-dichloro-N-{[(2S)-1-ethylpyrrolidin-2-yl]methyl}-2-hydroxy-6-methoxybenzamide) at a dose of 5 μg/0.5 μl per side. These doses of dopamine antagonists have been shown previously to prevent DNQX microinjections in NAc shell from producing eating and fearful behaviors (Faure et al., 2008). Each rat received 4 microinjections in balanced order: 1) vehicle alone (ACSF), 2) muscimol alone, 3) dopamine antagonists alone (D1 & D2 dopamine antagonists without muscimol), and 4) muscimol plus D1 & D2 dopamine antagonists. Immediately after every microinjection the rat was placed in a chamber in a conventional laboratory testing room with normal daylight illumination conditions (white fluorescent light intensity 550–650 lux) and moderate intensity ambient noise (65 –70 decibels).

GABA versus glutamate environmental shift experiment

In order to compare environmental ambience modulation of intense motivations produced by glutamate disruption or GABA inhibition, each rat was tested in three different emotional environments either after DNQX or vehicle microinjections (6 tests), or after muscimol or vehicle microinjection (6 tests). DNQX was microinjected at a dose of 450 ng/0.5 μl per side and muscimol was injected at a dose of 75 ng/0.5 μl per side, doses which were selected to produce roughly equal levels of motivated behaviors from previous studies (Faure et al., 2010). Some rats (n=29) received microinjections of either DNQX or its vehicle (50% DMSO and 50% .15 M saline) on separate days, prior to being tested twice each in the Home, Stressful and Standard environments, for a total of 6 test days and 6 microinjections per rats. Other rats (n=30) received microinjections of either muscimol or its vehicle (ACSF) prior to being tested twice each of the 3 environments, also for a total of 6 test days. The order of drug and environment conditions was counterbalanced in each group.

The Standard lab environment was identical to test conditions described above, used in most previous studies and for the dopamine blockade experiment (standard lab room, florescent lighting, low to moderate levels of ambient noise). The Home environment test occurred in the rats’ own home-room, and was characterized by dim red lighting (5–10 lux), the familiar odors of the housing room, and low levels of ambient noise primarily consisting of sound from the ventilation systems (65–70 decibels) and from other rats. The Stressful environment was conducted in an unfamiliar lab room, featuring high-intensity sensory-stimulation of light and sound. Extra light was provided by additional incandescent lamps that were directed at the transparent test chamber (1000–1300 lux within the chamber). Loud, unpredictable sound was presented continuously through the test (raucous rock music from the continuous full-album soundtrack of “Raw Power” by Iggy & The Stooges [1973; Iggy Pop reissue 1997]; 80–86 decibels). In previous preference tests, rats have been shown to prefer the Home environment condition over the Standard environment condition and to prefer the Standard lab environment over the Stressful condition (Reynolds & Berridge, 2008).

Behavioral tests of unconditioned motivated behaviors

Prior to the first test day, rats were habituated to handling and procedures for 7 days: rats were handled for 10 minutes per day for 3 days, and then were habituated to the testing procedure and apparatus for 1 hour each on 4 additional days. On the 4th day of habituation rats received “mock” microinjections of vehicle before being placed in the chamber. On drug test days, rats received one of the microinjections described above and were placed immediately in a transparent testing chamber (23 × 20 × 45 cm) which contained pre-weighed food (~20g rats chow) and ad libitum water. The chamber always contained granular cob bedding (~ 3 cm deep) to allow expression of defensive treading behavior. Rats remained in the chamber for 60 minutes while behavior was videorecorded, to be coded later for analysis.

At the end of each session, rats were removed by the experimenter’s gloved hand using a standardized slow-approach hand motion in order to quantify any fearful distress calls, escape attempts or defensive bites elicited by human touch. Following a ~5 second approach towards the testing cage, the experimenter slowly reached towards the rat, taking ~2 seconds. Upon contact, the experimenter lightly brushed the side of the rat with gloved fingertips, taking ~1 sec, before lifting the rat from the chamber in a gentle movement that lasted ~2 sec. The observer recorded any a) attempts by the rat to escape when touched (e.g., frantic jumps or runs away from the hand), b) bites or attempts to bite the gloved hand, and c) audible distress vocalizations elicited by the approaching hand.

Behavioral coding of videorecorded behaviors

Videorecorded behaviors were scored offline by observers blind to drug conditions. The incidence of elicited fearful distress vocalizations, escape dashes, and bite attempts directed at the experimenter’s hand were scored when the rat was gently picked up at end of the test session (Reynolds & Berridge, 2003), at which time total grams of chow pellets consumed were also recorded. Behaviors emitted spontaneously and videotaped during the 1-hr test were subsequently scored by experimenters blind to treatment for the total cumulative duration (seconds) for each of the following: eating behavior (involving both appetitive approach and voluntary initiation of ingestion plus consummatory chewing and swallowing of food), drinking behaviors (licking from water spout), fearful defensive treading/burying behavior (defined as active spraying or pushing of bedding with rapid alternating thrusts of the forepaws, spatially directed generally towards the brightly lit front or corners of the cage) and grooming behavior (a stereotyped sequence described in (Aldridge et al., 1993)). Observers scored the total number for behaviors which tended to occur as discrete events, including appetitive behaviors such as food carrying (transportation of food pellets in the mouth) and food sniffs (sniffing near the food for at least 1 second), and two general motor activities: rearing (forepaws at least one inch off the floor) and cage crosses (forepaws and head cross the halfway point of the cage).

Histology

Following behavioral testing rats were euthanized with an overdose of sodium pentobarbital, and their brains were removed and fixed in 10% paraformaldehyde for 1–2 days and in 25% sucrose solution for 3 days. In order to assess microinjection site locations, brains were sliced at 60 microns on a freezing microtome, and stained with Cresyl violet. Microinjection sites were mapped onto coronal slices from a rat brain atlas (Paxinos & Watson, 2007), which were then used to extrapolate the position of each site on a sagittal slice. This allowed for the presentation on the same maps of most of the rostrocaudal and dorsoventral extent of medial shell. Functional effects on appetitive and fearful behaviors were mapped using color-coding to express intensity of changes in motivated behaviors for individual behaviorally-tested rats. Symbols were sized to match the maximal diameter of Fos plumes as found here and previously (Faure et al., 2010; Richard & Berridge, 2011). For rostral versus caudal statistical comparisons, sites were classified as rostral shell if they were located more than 1.4 mm ahead of bregma, and as caudal if they fell behind this benchmark.

Fos-like protein immunohistochemistry

To map the size of the area where microinjections were likely to have impacted local tissue, as well as to assess the neurobiological effects of dopamine antagonists on GABA inhibition with muscimol, we conducted a Fos plume analysis on a separate group of rats after a single microinjection in order to detect maximal spread of neural impact. This was important because we have previously shown that the intensity and diameter of changes in Fos expression is suppressed following several DNQX microinjections (Richard and Berridge, 2011). That finding indicates maximal neural impact spread may best be assessed after a single microinjection (Richard and Berridge, 2011), just as behavioral efficacy of microinjections may be strongest initially (Kelley et al., 2000). Therefore we analyzed Fos after only a single microinjection in order to avoid potential mapping problems due to underestimation of impact spread.

Fos plumes are used specifically as a measure of diameter for the spread of neural impact locally surrounding a drug microinjection. No assumptions are required that Fos expressing neurons are identical to receptor-bearing neurons (local neurons that are indirectly modulated via synaptic recruitment cannot be excluded from contributing to behavioral effects of a microinjection, so are also included in measures of impact spread), or that Fos expressing neurons are the ones functionally responsible for generating an observed behavior. Nor do we assume that Fos elevation necessarily signifies neuronal depolarization or that Fos inhibition signifies hyperpolarization (Dragunow & Faull, 1989). Information regarding the diameter of Fos plumes is useful for identifying how far the immediate neural impact of a drug microinjection spreads from its center. The clear spatial cut-off of local continguous Fos plumes allows local impact spread to be easily distinguished from circuit-level Fos changes at more distant sites, which are typically separated by unaffected spaces with no change in Fos expression extending up to several millimeters from the site. Our study used Fos plumes to estimate maximal spread of local impact. To construct maps for localization of function, we combined that information about spread diameter with functional information about the behavioral consequences of drug microinjections at each mapped site.

Rats used for Fos analysis (n=18) were anesthetized with an overdose of sodium pentobarbital and transcardially perfused 90 minutes after bilateral microinjection of 1) vehicle in NAc (n = 5), 2) muscimol (n = 6), 3) dopamine antagonists (n = 3), or 4) mixture (n = 3), or 5) no injection (n = 1) for Fos plumes analysis. Brains were removed and placed in 4% paraformaldehyde for 4 – 24 hours, and then transferred to 25% sucrose (in 0.1 M NaPB) for 3 days. Brains were sliced at 40 microns on a freezing microtome, and processed for Fos-like immunoreactivity using NDS, goat anti-cfos (Santa Cruz Biotechnology, Santa Cruz, CA) and donkey anti-goat Alexa Fluor 488 (Invitrogen, Carlsbad, CA) as described previously (Faure et al., 2008; Reynolds & Berridge, 2008; Faure et al., 2010). Sections were mounted, air-dried and coverslipped with ProLong Gold antifade reagent (Invitrogen).

Fos plume analysis

Immunoreactivity for Fos-like protein was visualized using a Leica microscope equipped for fluorescent microscopy, using a filter with an excitation band at 480–505 nm for Fos-positive cells and images were taken using MCID Core software. For analysis of drug spread, Fos plume images were taken in the areas surrounding the microinjection with the most intense areas of Fos expression, just caudal to the end of the injector tip, surrounding a small focal point of necrosis. Fos labeled cells were individually counted within successive blocks (50 μm × 50 μm), along 8 radial arms emanating from the center of the necrosis, with 10x magnification (Figure 1). Zones of Fos elevation (or “plumes”) were assessed as described previously for muscimol and dopamine antagonist microinjections (Reynolds & Berridge, 2008). Fos plume estimates of the diameter of DNQX microinjection impact were adopted from a previous study that used the same dose and volume of DNQX microinjections in NAc medial shell (Richard & Berridge, 2011).

Figure 1. Summary graphs of the effect of local dopamine blockade on muscimol-generated eating and fearful behaviors.

Time spent eating (A), amount of food intake (B), time spent defensive treading/burying (C), as well as incidence of distress vocalizations in response to human touch (D), elicited by vehicle (grey), dopamine antagonist combination of raclopride and SCH23390 (black), muscimol (red) and mixture of muscimol and dopamine antagonist (yellow) in rostral (n=9) and caudal (n=7) regions of medial NAc shell. Errors bars indicate SEM, * p < 0.05, ** p < 0.01 change from vehicle, # p < 0.05, ## p < 0.01 change from dopamine antagonists alone, pairwise comparisons using Sidak corrections for multiple comparisons (eating, food intake and defensive treading) or McNemar’s test (distress vocalizations). Muscimol alone and mixture were never significantly different. Summary maps below show the localization of microinjections that produced primarily appetitive (green symbols), defensive (red symbols) or mixed appetitive and defensive (yellow symbols) in rostral (white) versus caudal (grey) shell.

Statistical analysis

The effects of dopamine blockade on muscimol-induced motivated behaviors were assessed using a three-factor mixed within- and between-subject ANOVA (muscimol X dopamine blockade X anatomical level [rostral versus caudal]). When significant effects were found, rats were split into rostral and caudal groups and additional two-way (muscimol X dopamine blockade) and one-way ANOVAs (main effect of drug) were conducted, including pairwise comparisons with Sidak corrections. For analysis of binomial data (vocalizations, escape attempts and bite attempts) Cochran’s Q and McNemar’s repeated measures tests were used. The effects of DNQX or muscimol in the three different emotional environments were compared using a three-factor mixed within- and between-subjects ANOVA (drug X anatomical level [rostral versus caudal] X environment [Stressful versus Standard versus Home]) for all parametric behaviors. When significant effects were found, rats were split into rostral and caudal groups and additional one-way ANOVAs were conducted on the difference between vehicle and drug between the three environments, including pairwise comparisons with Sidak corrections. We also assessed the percent of rats which met criteria for appetitive, defensive or mixed valence behaviors, and tested whether that changed between the three environments for DNQX versus muscimol, using Cochran’s Q and McNemar’s repeated measures test, and whether the proportion of rats that switched differed between DNQX and muscimol using Pearson’s chi-squared test.

Results

Dopamine blockade experiment: dopamine is not needed for GABAergic motivation

Local blockade of dopamine D1 and D2 receptors failed to impair the ability of GABAergic muscimol microinjections to generate intense levels of motivated behaviors (5-times to 75-times above vehicle control levels). Neither the generation of appetitive eating nor of fearful defense/escape behaviors was prevented by the addition of dopamine antagonists to the muscimol microinjection, and muscimol-generated levels always remained high above vehicle levels regardless of whether dopamine antagonists were present or absent (Figures 1 and 2). In rostral shell muscimol microinjections generated robust tripling of food intake and of time spent eating, with or without dopamine antagonists (Figures 1a and 2a and b; eating, interaction of muscimol by placement, F(1, 14) = 14.51, p = 0.002). The addition of dopamine antagonists did not reduce these intense appetitive behaviors generated at rostral sites, and eating remained intense, at more than double vehicle levels (Figures 1a and 2a; rostral shell, eating, interaction of muscimol and D1/D2 antagonists, F (1,8) = <0.01, p = 0.996). By comparison, caudal shell sites of muscimol microinjection with or without dopamine antagonists generated a 60-fold increase in defensive treading behavior (interaction of muscimol by placement, F(1,14) = 6.15, p = 0.026), as well as distress vocalizations, escape attempts and sometimes bite attempts in response to the experimenter’s approach and touch after the treading test (Figures 1c and 2d; vocalizations and escape attempts, p < 0.001). Dopamine antagonists did not alter the level of fearful defensive treading generated by muscimol at caudal sites, which remained at more than 50-fold vehicle levels (Figures 1c and 2b; interaction of muscimol and D1/D2 antagonists, F(1, 6) = 0.86, p = 0.389). Dopamine blockade also had no effect on caudal distress vocalizations, escape attempts, or bite attempts observed after the mixture of muscimol and dopamine antagonists when compared to muscimol alone (Figures 1d and 2c; McNemar’s Test, vocalizations, p = 0.375; escape attempts, p = 0.180; bite attempts, p = 1.000).

Figure 2. Effects of dopamine antagonism on muscimol-induced eating and defensive fearful behaviors.

Fos plume maps (n=16) of the generation in medial shell of eating (A), defensive treading (B), and fearful calls, escape attempts and bite attempts (C) by muscimol (left) and a mixture of muscimol plus dopamine antagonists (right). Local dopamine blockade failed to prevent muscimol generated eating or fearful behaviors, despite its previously reported ability to prevent DNQX-induced eating and fear, and its generally suppressive effects. Histograms bars below the maps show behavior as percent of vehicle (eating, A; treading, B) or percent of subjects (calls, escape attempts and bite attempts, C) for each behavior at rostrocaudal level as marked along the medial shell (error bars = SEM).

The persistence of intense GABAergic eating and fear after local dopamine blockade contrasts with the dramatic reduction of motivated behaviors generated by corticolimbic glutamate disruptions we previously reported to be caused by similar dopamine blockade (Faure et al., 2008; Richard & Berridge, 2011). However, by themselves, microinjection of dopamine antagonists here (without muscimol) did suppress lower spontaneous baseline levels of appetitive eating and food intake below vehicle to nearly zero (Figure 1a and b; time spent eating, general effect of D1/D2 antagonists, F (1,15) = 7.853, p = 0.014; food intake grams consumed, average of 0.25 ± 0.68 g under D1/D2 antagonists vs 1.65 ± 1.28 g under vehicle; food intake, general effect of D1/D2 antagonists, F (1,15) = 8.674, p = 0.012).

Appetitive versus fearful motivational valence generated by GABAergic inhibition is purely anatomically determined and resists change by environmental ambience

Muscimol microinjections in NAc shell generated the same rostrocaudal keyboard-like pattern of intense eating, fearful and mixed valence behaviors across all three environments, resisting environmental retuning (Figures 3 and 4). At rostral sites in medial shell muscimol generated intense appetitive behavior equally in Home, Standard, and Stressful environments (main effect on eating, F(1,27) = 6.309, p = .018; interaction with placement, F(2,27) = 5.672, p = .009). At more caudal locations in medial shell, muscimol generated intense defensive behaviors in all three environments (main effect on treading, F(1,27) = 4.511, p = .043; interaction with placement, F(2,27) = 6.388, p < .005), such that at least 46% of rats emitted distress vocalizations to the experimenter, and at least 21% of rats attempted to escape in each environment (McNemar’s Tests, vocalizations: Standard, p = .000, Stressful, p = .004, Home p = .000; escape attempts: Standard, p = .031, Stressful, p = .008, Home, p = .031). Switching between environments had no detectable influence on the intensity of muscimol-elicited appetitive eating at any site (muscimol by environment: eating, F(4,54) = 1.874, p = .128) and did not alter the size of the appetitive zone in which eating was elicited (Cochran’s Q(2) = .400, p = .819). Environmental changes also did not alter fearful escape attempts or distress vocalizations elicited by the approach of the experimenter’s gloved hand (vocalizations, Cochran’s Q(2) = .500, p = .779; escape attempts, Cochran’s Q(2) = .727, p = .695), and altered only one category of fearful reaction: spontaneous defensive treading.

Figure 3. Summary maps of behavior and percent flipped by changes in ambience.

Summary maps (A) show behavior produced by muscimol (top) or DNQX (bottom) in either the Home (left), Standard (middle) or Stressful (right) environments. Each subject (n=59) was designated as producing primarily appetitive (green symbols), defensive (red symbols) or mixed appetitive and defensive (yellow symbols) motivated behavior following DNQX microinjection. In a Standard environment, purely appetitive eating behavior and food intake (criteria for including a site was a >200% increase in eating) was primarily stimulated in rostral shell by DNQX and muscimol. Fearful distress calls, escape attempts and spontaneous emission of defensive treading-burying (criteria for including a site was a >500% increase in treading over vehicle levels, or emission of a defensive reaction to the experimenter) were primarily stimulated in caudal shell by DNQX. Mixed sites met criteria for both motivations. Testing in the Stressful environment shifted fearful behavior produced by DNQX into more rostral regions, whereas testing in the Home environment virtually eliminated all defensive behavior produced by DNQX. Muscimol produced a similar rostrocaudal gradient of eating and fear regardless of environment.

Figure 4. Maps of changes in motivational valence between the Home and Stressful environment.

Summary maps show sites where DNQX (A) or muscimol (B) generated either intense eating behavior (top; green; > 200% of vehicle) in the Home environment but not in the Stress environment, or intense defensive behavior (bottom; red; > 500% of vehicle level treading or defensive reaction to the experimenter) in the Stressful environment but not in the Home environment. Sites mapped in white produced the same valence of behavior in both the Home and Stressful environment. Criteria for designating a site as appetitive was a >200% increase in eating, criteria for designating a site as defensive was a >500% increase in treading over vehicle levels, or emission of a defensive reaction to the experimenter. The percentage of rats that “flipped” between mainly defensive/mixed in the Stressful environment and purely appetitive in the Home environment was significantly greater in rats that received DNQX (C) than rats that received muscimol (D), and the number of rats who “flipped” when given muscimol was not significant.

The only environmental modulation effect observed was that the dark, quiet and familiar Home environment virtually eliminated spontaneous emission of defensive treading (F(1,26) = 8.551, p < .007; McNemar’s Tests, versus Standard, p = .005; versus Stressful, p = .002; Figure 5). However, the noisy and bright Stressful environment did not increase defensive treading behavior over the already intense levels produced by caudal muscimol microinjections in the Standard environment (muscimol by environment: F(1,26) = .163, p = .690), and failed to switch any rostral sites from generating purely appetitive eating to defensive treading (Stressful versus Standard, McNemar’s Test, p = 1.000). Also, the reactive and even more extreme threat-evoked defensive reactions of distress vocalizations and escape attempts that were elicited by the experimenter’s touch remained equally intense and unchanged across all three environments after caudal muscimol microinjections (distress vocalizations, Cochran’s Q(2) = .500, p = .779; escape attempts, Cochran’s Q(2) = .727, p = .695; Figure 6).

Figure 5. Effect of changing environmental ambience on defensive treading produced by DNQX or muscimol.

Fos plume maps (n=59) of the generation of defensive treading by muscimol (top) or DNQX (bottom) in the Home (left), Standard (middle), or Stressful (right) environments. Testing in the Stressful environment had no effect on muscimol-generated treading, but the Home environment nearly eliminated muscimol-generated treading. This may be due to the removal of all visual cues that the animals usually tread toward. DNQX treading was produced at more rostral locations in a Stressful environment, but was nearly eliminated by testing in a Home environment. Histogram bars show treading as percent of vehicle at each rostrocaudal level as marked along the medial shell (error bars = SEM).

Figure 6. Effect of changing environmental ambience on defensive reactions to the experimenter produced by DNQX or muscimol.

Fos plume maps (n=59) of the generation of defensive reactions to the experimenter by muscimol (top) or DNQX (bottom) in the Home (left), Standard (middle), or Stressful (right) environments. Changing environment between the familiar Home or the aversive Stressful environments had no effect on defensive reactions produced by muscimol, which were robustly generated regardless of environment ambience. In the Stressful environment, defensive reactions to the experimenter produced by DNQX treading were generated at more rostral locations in shell, but were nearly eliminated by testing in a Home environment. Histogram bars show percent of rats emitting distress calls (yellow), distress calls and escape attempts (red) and distress calls, escape attempts and bite attempts (red) at each rostrocaudal level as marked along the medial shell.

Thus the overall pattern of rostral shell generation of intense eating, middle generation of mixed eating/fear, and caudal generation of intense fear appeared stable, and determined purely by anatomical location of each GABAergic microinjection that induced local hyperpolarization. The valence tuning of individual microinjection sites never shifted across environments. Only the behavioral intensity of spontaneous defensive treading for middle and caudal sites was at all influenced by environment, and only by being dampened in the presence of the familiar and comfortable Home conditions. All other aspects of intense motivated behaviors generated by muscimol microinjections were invulnerable to environmental changes.

By contrast to GABAergic inhibition, the keyboard pattern generated by DNQX microinjections was powerfully shifted both anatomically and in behavioral intensity generated at particular sites by changes in environmental ambience, as expected (Figures 3 and 4a; interaction with environment: eating, F(2,44) = 7.359, p = .002; treading, F(2,44) = 4.985, p = .011; distress vocalizations, Cochran’s Q(2) = 11.556, p = .003; escape attempts, Cochran’s Q(2) = 6.500, p = .039). The Stressful environment nearly doubled the size of the medial shell zone where defensive treading was generated, in comparison to the Standard environment, to more than 55% of sites (versus 31% in Standard), more than quadrupling the defensive zone in comparison to the Home environment (overall effect of environment on % defensive sites, Cochran’s Q(2) = 10.391, p = .006; McNemar’s Test, p = .002, Figure 5). Likewise, the Stressful environment doubled the zone in which DNQX produced attempts to escape from the experimenter’s hand, compared to the Standard environment (to 7% of sites; McNemar’s Tests, Lab versus Stressful, p = .25; Home versus Stressful, p = .125; Figure 6) and quadrupled the zone where DNQX produced reactive distress vocalizations (to 28%; McNemar’s Tests, Lab versus Stressful, p = .070; Home versus Stressful, p = .008).

Conversely, the familiar Home environment nearly eliminated the fearful caudal zone where DNQX generated spontaneous defensive treading behavior (Cochran’s Q(2) = 10.391, p = .006; difference between Stressful and Home environments, McNemar’s Test, p = .002, Figure 5), and completely eliminated generation of reactive responses such as distress vocalizations and escape attempts to the experimenter’s hand (McNemar’s Test, Lab versus Home, p = .500, Stressful versus Home, p = .008; Figure 6). At the same time, the Home environment expanded the “appetitive zone” where DNQX produced purely appetitive behavior to more than 75% of medial shell (versus 31% in the Stressful environment; Figure 3), consistent with previous findings (Reynolds & Berridge, 2008). Testing the Stressful environment slightly, but non-significantly, reduced the intensity of DNQX-induced eating (DNQX by environment, F(1,23) = 3.404, p = .078).

Fos plume analysis of muscimol and dopamine antagonist microinjections

Mapping of effects was aided by measurements of drug impact spread for muscimol microinjections, as reflected in Fos plumes of about 0.5 mm in diameter. Microinjections of muscimol alone in NAc medial shell altered Fos expression in an area with a total radius of 0.24 mm: containing a small 0.15 mm radius excitatory plume center (volume = .014 mm³), of elevated Fos expression (150% of vehicle levels), encompassed by a larger of 0.24 mm radius (volume = 0.06 mm³) inhibitory or “anti-plume” surround (Fos reduced to <75% of vehicle levels; Figure 8), similar to our previous reports (Faure et al., 2010). Dopamine D1 & D2 antagonists, combined together but without muscimol, produced a tiny excitatory center (volume = .005 mm³) of 150% of vehicle level Fos expression up to .10 mm from the microinjection center, surrounded by a very large 0.41 mm radius robust inhibitory “anti-plume”. The anti-plume contained a dense inhibitory middle layer of 0.16 radius (volume = .018 mm³) of intense Fos suppression (< 50% of vehicle levels), contained within a larger 0.41 mm radius (volume = .29 mm³) but less intense anti-plume of moderate Fos suppression (50%–75% of vehicle levels), consistent with previous reports (Faure et al., 2008). Combining muscimol and dopamine antagonists in the same cocktail microinjection produced a similarly sized 0.4 mm radius anti-plume, containing even larger 0.31 mm radius middle layer of intense Fos suppression (volume = .135 mm³), surrounded by a thinner shell (0.1 mm shell width; 0.4 mm total outer radius) of moderate Fos suppression (50%–75% of vehicle levels; total volume = .24 mm³). The diameter of estimated spread for DNQX microinjection was based on a previously reported total radius of 0.38 mm (+/− .05 mm, SEM) (Richard and Berridge, 2011).

Figure 8. Fos plume analysis.

Fos expression was assessed following microinjections of vehicle, muscimol, dopamine antagonists alone, or mixture of muscimol and dopamine antagonists. Fos labeled cells were individually counted within successive blocks (50 μm × 50 μm), along 8 radial arms emanating from the center of the site, with 10x magnification. For vehicle microinjections colors indicate levels of Fos expression of 3x (red), and 2x (orange) levels of Fos expression found in normal (uninjected) tissue. For drug microinjections, colors indicate levels of Fos expression of 1.5x (yellow), .75x (light blue), and .50x (darker blue) vehicle level Fos expression. Line graphs show that muscimol (red) reduced Fos expression approximately ~.35 mm away from the microinjection center, and that dopamine antagonists (black) and mixture (yellow) reduced Fos expression from .15 mm to .40 mm away from the microinjection center.

Discussion

GABAergic release of intense motivations by muscimol microinjection in NAc appears to be more anatomically predetermined by rostrocaudal location in medial shell than equivalent motivation generation by glutamate disruptions (Reynolds & Berridge, 2008). GABAergic release of eating and fear is also more autonomous of endogenous dopamine inputs (Faure et al., 2008; Richard & Berridge, 2011).

Here the addition of D1 (or D2) dopamine antagonists to GABAergic muscimol microinjections did not impair the generation of intense levels of eating and food intake at rostral sites nor of fearful spontaneous defensive treading behavior and of reactive frantic escape attempts and distress calls upon being touched at caudal sites. By contrast, addition of D1 antagonist to a DNQX microinjection does block generation of either eating or fear by NAc glutamate disruption, and addition of a D2 antagonist additionally blocks generation of fear, as we previously showed (Richard & Berridge, 2011).

Regarding environmental modulation, the valence of appetitive versus fearful motivation generated by the GABAergic inhibition of NAc resisted re-tuning by ambience changes. Muscimol microinjections always released anatomically-determined mixtures of intense motivations, based solely on keyboard position along the rostrocaudal axis of medial shell, regardless of whether rats were exposed to the comfortable Home environment, the neutral Standard lab, or the loud and bright Stressful environment. By contrast, we confirmed that those environmental changes powerfully retuned the relative size of appetitive versus fearful valence-generating zones produced by glutamate disruption.

The sole environmental retuning effect observed for GABAergic muscimol microinjections was that the dark Home environment nearly eliminated spontaneous defensive treading-burying, which in Standard and Stressful environments was directed specifically toward the front of the transparent cage. However, reactive distress calls and frantic escape jumps away from the experimenter’s hand at the end of the session persisted in the dark Home environment after caudal muscimol microinjections, remaining unsuppressed, and those defensive escape reactions elicited by human touch arguably reflect even more intense fear than defensive treading. That pattern suggests that at best the Home environment suppresses moderate fear generation displayed in defensive treading-burying but does not suppress more intense fear generation displayed in reactive distress calls and frantic escape attempts (elicited by the potentially-threatening stimulus of an approaching human hand). Additionally, an alternative explanation exists for why spontaneous treading was reduced: the dark Home environment (illuminated only by dim red light, which scotopic rat vision does not detect well) made it more difficult for rats to see visual stimuli toward which defensive treading is usually directed (sight of objects in the room beyond the cage, reflected light from curved corners of the cage). Thus it may have been the removal of eliciting visual stimuli that was more responsible for reducing defensive treading, rather than suppression of fear. In the lighted Standard and Stressful environments, a rat typically threw bedding toward the reflecting corners or toward the transparent front wall of the chamber facing the room, often building a defensive mound of bedding between those stimuli and the rat (Reynolds & Berridge, 2001). In the absence of those eliciting visual stimuli under darkness, conceivably even an actively fearful rat might not emit much spontaneous treading behavior, but could still react defensively to a to the multi-modal stimulus of an approaching hand. Especially considering that alternative explanation, and that all other aspects of the rostrocaudal pattern of eating and fear generated by GABA inhibition remained intact across the three environments, we conclude that GABAergic muscimol generation of intense motivations is more robust, resistant to environmental retuning of valence, and neuroanatomically pre-determined than similar motivations produced by glutamate disruption (Faure et al., 2008; Reynolds & Berridge, 2008).

Bases for subcortical autonomy

Why should the GABAergic NAc keyboard for generating desire versus dread be more anatomically pre-determined and autonomous of endogenous dopamine than the corresponding glutamatergic keyboard? One possible answer derives from the subcortical and intrinsic nature of endogenous GABA signals in NAc, compared to the corticolimbic nature of glutamate signals. That is, GABA signals would normally be delivered to spiny neurons in medial shell site by neighboring intrinsic NAc medium spiny neurons or interneurons, or by GABAergic afferent projections that arose from other subcortical structures such as ventral pallidum, extended amygdala (e.g., BNST), lateral hypothalamus, or brainstem (Brog et al., 1993; Sun & Cassell, 1993; Churchill & Kalivas, 1994; Vanbockstaele & Pickel, 1995; Meredith et al., 2008).

Environmental ambience signals by comparison might ordinarily be delivered to NAc largely by ‘top-down’ corticolimbic glutamatergic projections from prefrontal cortex, or other glutamatergic inputs from cortex-related structures such as basolateral amygdala, hippocampus or thalamus (Beckstead, 1979; Christie et al., 1987; Fuller et al., 1987; Goldin et al., 2008). Intrinsic subcortical GABA signals therefore might be somewhat “bottom-up” and relatively autonomous from “top-down” environmental modulation. By this view of GABAergic autonomy, the resistance of GABAergic release of appetitive and fearful motivation tox changes in emotional environment reflects the resistance of emotional processes generated by intrinsically subcortical circuits to top-down modulation by factors that can influence corticolimbic regulation of emotion (Beauregard et al., 2001; Russell, 2003; Davidson, 2004; Barrett et al., 2007; Posner et al., 2007).

Likewise, while glutamate signals to NAc neurons interact closely with mesolimbic dopamine signals, often on the same spine (Cepeda et al., 1993; Calabresi et al., 1997; Brady & O’Donnell, 2004; Tecuapetla, 2010), GABA signals to NAc neurons may be less modulated by co-occurring dopamine inputs. Although dopamine may modify pre-synaptic GABA release from NAc interneurons (Bracci et al., 2002; Centonze et al., 2002; Tecuapetla et al., 2007; Towers & Hestrin, 2008), we are aware of no studies reporting that the post-synaptic hyperpolarizing effects of GABA_A receptor activation on medium spiny neurons are modulated by dopamine. For example, dopamine transporter knock-down mice, which have increased levels of dopamine in the synapse, have normal GABA-receptor mediated synaptic currents onto striatal medium spiny neurons (Wu et al., 2007). GABAergic generation of eating and fear by muscimol microinjections may mimic natural post-synaptic GABA signals that do not depend on post-synaptic dopamine, and so it is understandable that intense muscimol-generated motivations are not impaired by blockade of D1 or D2 inputs. The dopamine-independent nature of GABAergic eating and fear is also consistent with subcortical GABAergic generation of ‘liking’ or disgust reactions that reflect intense hedonic impact, psychological processes which are relatively independent of mesolimbic dopamine (Peciña et al., 1997; Berridge & Robinson, 1998; Peciña et al., 2003; Faure et al., 2010). Here, muscimol microinjection would mimic post-synaptic GABA release effects by directly binding to GABA_A receptors in medial shell, and hence may have acted downstream of dopamine modulation. This account of muscimol microinjection effects does not exclude the possibility that, under natural conditions, changes in dopamine may modulate motivated behaviors by presynaptic modulation of endogenous GABA release in NAc (Bracci et al., 2002; Centonze et al., 2002; Tecuapetla et al., 2007; Towers & Hestrin, 2008).

It is also possible that hyperpolarizations produced by GABAergic muscimol are more robust or intense at the neuronal level than hyperpolarizations by DNQX glutamatergic blockade (Koos et al., 2004), which merely diminishes “up states” to suppress EPSPs (Meredith et al., 1993; Pennartz et al., 1994; Kiyatkin & Rebec, 1999; Meredith, 1999; O’Donnell, 1999; Suwabe et al., 2008; Jeun et al., 2009). GABAergic signaling may more potently inhibit NAc medium spiny neurons by acting directly to open Cl- channels on soma or proximal dendrites (Sun & Cassell, 1993; Johnson et al., 1994; Behrends et al., 2002; Goetz et al., 2007), whereas glutamatergic AMPA receptors are localized more distally at the ends of dendrite spines (Meredith et al., 1990; Sesack & Pickel, 1992; Johnson et al., 1994; Chen et al., 1998), allowing dopamine released on spines or dendrites to play a greater role.

NAc keyboard pattern

Just as a musical keyboard generates many different notes, depending on which key is tapped, the NAc ‘affective keyboard’ for generating desire versus dread released many different ratios of appetitive to fearful motivated behaviors depending on the precise rostrocaudal position of a GABAergic or glutamatergic microinjection. Our data suggest that the anterior-posterior length of medial shell, which extends nearly 2.5 mm, contains multiple ‘key’ levels or a ratio continuum of substrates for opposite motivations. That is, each microinjection here released its own individualized ratio of valence-generating substrates (i.e., either at least three detectable rostrocaudal levels or a gradual continuum capable of producing many more potential ratios).

The different ratios appear to be generated by mechanisms contained close to microinjection sites, rather than due to distant drug diffusion to another part of medial shell. That is suggested especially by the relatively small diameter of muscimol Fos plumes, which were just under 0.5 mm (0.24 mm radius), as well as the just slightly larger diameter of DNQX Fos plumes, just under 0.8 mm (0.38 mm radius). As caveat, Fos plumes are not infallible indicators for neurobiological impact spread. However, plumes do at least provide some objective information on how far a microinjected drug’s impact may extend. It therefore seems worthwhile to seriously consider the implications of measured plume spread, at least given the absence of evidence that muscimol microinjection impact extends any further. One implication of relatively small 0.5 to 0.8 mm diameters of impact spread is not only that substrates for both motivations must have been reachable within that spread from most shell positions, but also that the different ratio mechanisms must have been locally contained.

Why can’t a simpler mechanism fully account for our effects, say, in which the front half of shell simply produced eating and the caudal half produced fearful behavior? The reason is that two discrete non-overlapping substrates cannot easily account for the multiple quantitative ratios of appetitive and defensive behavior seen as sites moved rostral to caudal, or for the capacity for most sites in medial shell to release both appetitive and fearful motivations. That is, the two halves of medial shell did not respond as simple qualitative on/off mechanisms. Instead moves of position along the AP axis produced proportional shifts in the ratio of intense motivated behaviors.

Perhaps the most compelling evidence against a simple 2-key mechanism for differently valenced motivations contained in the two halves of NAc shell are the glutamatergic data which confirmed that each half of NAc medial shell must also contain at least some substrates for generating its opposite-valenced motivation. That is, the rostral half of medial shell must also contain at least sparse fear-releasing substrates, and the caudal shell must also contain at least sparse eating-releasing substrates. This is demonstrated by the ability of glutamate-disrupting DNQX microinjections fully contained in the rostral half of shell to release fear (in a stressful environment), and for microinjections contained in the caudal half to release eating (in a comfortable home environment). A qualitative two-mechanism hypothesis cannot easily explain such reversal patterns.

One way of reconciling all these observations would be to conceive of NAc eating/fear mechanisms as two cloud continuums for desire/dread substrates extending through medial shell. One substrate is densest near the rostral pole for generating appetitive eating, but stretches in declining density nearly to the caudal pole. A second is densest near the caudal pole for generating fearful responses, but stretches in declining density nearly to the rostral pole. The two mechanisms would overlap in varying densities or efficacies throughout nearly the entire 2.5-mm rostrocaudal extent of medial shell. As consequence, the ‘affective keyboard’ pattern of behavioral ratios we observed here would be produced by the relative densities of substrates within the diameter of each single microinjection along the AP continuum. Of course, the challenge of identifying what this actually means in terms of neuronal mechanisms for determining the valence of intense motivations released by a microinjection remains a major task for the future.

Disinhibition of rostral versus caudal NAc shell outputs

The neuronal mechanism by which NAc inhibitions generated motivations is likely release or disinhibition of motivation generators in target structures, as NAc projection neurons are primarily GABAergic, and would be expected to tonically inhibit downstream neurons (Taber & Fibiger, 1997; Kelley, 1999; Roitman et al., 2008; Krause et al., 2010). Temporary inhibition of NAc neurons by GABAergic muscimol microinjection would remove downstream GABAergic inhibition, releasing a subpopulation of recipient neurons in a target structure to become relatively excited and generate intense motivations. Downstream target structures for potential release include ventral pallidum, ventral tegmentum and lateral hypothalamus, which all receive GABAergic inputs that originate throughout the rostrocaudal extent of medial shell (Mogenson et al., 1983; Zahm & Heimer, 1990; Heimer et al., 1991). Caudal NAc shell also likely projects to the extended amygdala, including the central and medial amygdala, the ventrolateral bed nucleus of the stria terminalis and the sublenticular gray, as well as potentially to the retrorubral area, the substantia nigra, and the parabrachial nucleus (Heimer et al., 1991; Zahm & Heimer, 1993; Usuda et al., 1998). Rostral shell projects to specific targets including particular subregions of the lateral preoptic area, lateral hypothalamus and ventral pallidum (Usuda et al., 1998; Thompson & Swanson, 2010; Zahm et al., 2013). Similarly, relative inhibitions of localized NAc neurons by DNQX microinjections, which block excitatory glutamate inputs to NAc such as from corticolimbic projections, might similarly disinhibit distinct subpopulations in target structures to release the generation of intense motivations. Rostrocaudal differences in potential targets of release may therefore contribute to different ratios observed along the NAc rostrocaudal axis in generation of fear versus eating.

Conclusions

The NAc generation of intense motivations by localized GABA inhibition obeys different rules than generation by glutamatergic disruption regarding dopamine and environmental retuning, even though GABA inhibition and glutamate disruption both produce similar appetitive and fearful behaviors organized along the same NAc rostrocaudal gradient. The GABAergic generation of intense eating and fearful motivations is solely determined by the anatomical position of the muscimol microinjections in NAc medial shell. The valence and intensity of GABAergic appetitive and fearful motivations are relatively immune to modulation by environmental ambience and free of any need for local dopamine inputs. By contrast, the keyboard pattern of desire and dread generated by glutamate disruption is easily retuned by changes in emotional environment, as well as highly dependent on endogenous dopamine inputs.

This distinction between glutamatergic motivations that are modifiable via top-down psychological and mesolimbic dopamine control, compared to robust GABAergic motivational states that are not, may have implications for understanding mechanisms of some psychiatric disorders and psychopathologies of emotional well-being. Flips in the valence of pathologically intense motivational salience may occur more easily when they primarily involve NAc corticolimbic or other glutamatergic mechanisms, than when motivations are generated by subcortical GABAergic mechanisms that are intrinsic to NAc. Glutamate/dopamine modifiability may relate to why the incentive salience of amphetamine addiction can flip valence from appetitive to the fearful salience of amphetamine psychosis, or why schizophrenic patients can have not only heightened fearful salience of paranoia but may also show exaggerated incentive salience for appetitive stimuli (Elman et al., 2006; Featherstone et al., 2007; Jensen et al., 2008; Howes & Kapur, 2009). In contrast, motivations released by GABAergic inhibitions in NAc-related circuits may be more anatomically coded, consistently valenced, hedonically laden, and less amenable to regulation by top-down control (Faure et al., 2010; Watson & Naragon-Gainey, 2010).

Figure 7. Effect of changing environmental ambience on eating produced by DNQX or muscimol.

Fos plume maps (n=59) of the generation of eating by muscimol (top) or DNQX (bottom) in the Home (left), Standard (middle), or Stressful (right) environments. Changing environmental ambience had inconsistent effects on eating produced by DNQX or muscimol. Histogram bars show treading as percent of vehicle at each rostrocaudal level as marked along the medial shell (error bars = SEM).