Amylin Modulates a Ventral Tegmental Area–to–Medial Prefrontal Cortex Circuit to Suppress Food Intake and Impulsive Food-Directed Behavior

Abstract

BACKGROUND:

A better understanding of the neural mechanisms regulating impaired satiety to palatable foods is essential to treat hyperphagia linked with obesity. The satiation hormone amylin signals centrally at multiple nuclei including the ventral tegmental area (VTA). VTA–to–medial prefrontal cortex (mPFC) projections encode food reward information to influence behaviors including impulsivity. We hypothesized that modulation of VTA-to-mPFC neurons underlies amylin-mediated decreases in palatable food-motivated behaviors.

METHODS:

We used a variety of pharmacological, behavioral, genetic, and viral approaches (n = 4–16/experiment) to investigate the anatomical and functional circuitry of amylin-controlled VTA-to-mPFC signaling in rats.

RESULTS:

To first establish that VTA amylin receptor (calcitonin receptor) activation can modulate mPFC activity, we showed that intra-VTA amylin decreased food-evoked mPFC cFos. VTA amylin delivery also attenuated food-directed impulsive behavior, implicating VTA amylin signaling as a regulator of mPFC functions. Palatable food activates VTA dopamine and mPFC neurons. Accordingly, dopamine receptor agonism in the mPFC blocked the hypophagic effect of intra-VTA amylin, and VTA amylin injection reduced food-evoked phasic dopamine levels in the mPFC, supporting the idea that VTA calcitonin receptor activation decreases dopamine release in the mPFC. Surprisingly, calcitonin receptor expression was not found on VTA-to-mPFC projecting neurons but was instead found on GABAergic (gamma-aminobutyric acidergic) interneurons in the VTA that provide monosynaptic inputs to this pathway. Blocking intra-VTA GABA signaling, through GABA receptor antagonists and DREADD (designer receptor exclusively activated by designer drugs)–mediated GABAergic neuronal silencing, attenuated intra-VTA amylin-induced hypophagia.

CONCLUSIONS:

These results indicate that VTA amylin signaling stimulates GABA-mediated inhibition of dopaminergic projections to the mPFC to mitigate impulsive consumption of palatable foods.

Excessive consumption of energy-dense, highly palatable foods contributes to weight gain and metabolic disorders (1–3). Moreover, consumption of palatable foods is associated with increased body mass index and predicts binge-eating behaviors (4,5). Baseline sensitivity to rewarding foods varies among individuals, and heightened reward sensitivity is positively associated with the amount of food consumed, the reinforcing value of palatable foods, body mass index, obesity-related health outcomes, and binge-eating behaviors (6,7). Thus, consuming highly rewarding food can promote a feed-forward cycle resulting in dysregulated eating and obesity. With the growing need for antiobesity therapeutics, it is critical to expand basic science knowledge about how energy balance systems influence motivational processes that contribute to excessive caloric intake.

Amylin is a peptide hormone co-secreted with insulin from pancreatic beta cells that improves glucoregulation and energy homeostasis and acts centrally at many distributed nuclei including the hindbrain and mesocorticolimbic areas to suppress food intake (8,9). Interestingly, central and peripheral administration of amylin analogues in rodents potently suppress intake of palatable foods while only modestly decreasing and even increasing intake of low-fat/low-sugar chow (10–12). High-fat/high-sugar foods activate central reward circuits and stimulate dopamine neurons of the ventral tegmental area (VTA) (13–15). The VTA expresses amylin receptors that are physiologically relevant for appetite control, and VTA selective knockdown of the core component of the amylin receptor complex, the calcitonin receptor (CTR), causes hyperphagia and excess weight gain in rats maintained on a high-fat diet (HFD) (16,17). In humans, the amylin analogue pramlintide promotes weight loss, reduces palatable food intake, and improves self-reported control of eating (18,19). Although no current amylin-based drug has Food and Drug Administration approval to treat obesity, recent evidence indicates that cagrilintide, a long-lasting amylin analogue, induced clinically significant body weight loss in overweight and obese individuals (20). Thus, there is considerable interest in amylin signaling as a next-generation antiobesity pharmaceutical target (21,22).

VTA dopaminergic neurons transmit information about a rewarding stimulus to, among other nuclei, the nucleus accumbens (NAc) and medial prefrontal cortex (mPFC) (23–25). Appetite-regulating hormones modulate VTA dopamine output to promote or discourage food seeking and consumption. Orexigenic factors stimulate VTA dopamine neuron firing while satiation signals diminish VTA dopaminergic activity (16,26–33). Most research has focused on the actions of leptin, insulin, GLP-1, and amylin to decrease VTA-to-NAc dopamine signaling (16,31,33,34). However, the regulation of satiation hormones on VTA-to-mPFC projections has largely been uninvestigated. This is surprising given that the mesocortical pathway is implicated in encoding the predicted value of a reward, reinforcement of reward cues, impulsive reward-directed behaviors, and the development of binge-eating disorder (15,24,25,35). Moreover, the mPFC is considered an important therapeutic target for treatment of binge-eating disorder (36), highlighting a potential key role of mesocortical signaling in maladaptive eating. Given the role of VTA amylin signaling in controlling motivated behavior and improving food choices (18,37–39), we investigated the mechanisms underlying amylin-mediated modulation of VTA-to-mPFC neurotransmission and motivated feeding. The results reported herein are the first to mechanistically show how amylin impacts the mesocortical circuit to control food reward–directed behaviors.

METHODS AND MATERIALS

Study Design

This study was designed to explore the role of VTA amylin signaling on VTA-to-mPFC activity and control of motivated feeding behavior in rats. Immunohistochemical quantification of cFos responsivity in the mPFC first assessed how intra-VTA amylin signaling modulated mPFC activation following a food reward. Given that impulsivity is a behavior largely controlled by the mPFC, we tested the impact of intra-VTA amylin on a food reward–directed behavioral impulsivity assay (differential reinforcement of low rates of responding [DRL] task) to determine whether VTA amylin signaling also affected mPFC-influenced behavior. We hypothesized that intra-VTA amylin may modulate mPFC activity by decreasing dopamine release from VTA-to-mPFC projections, so we pretreated the mPFC with dopamine receptor agonists to see how this affected the food intake and body weight–reducing effects of intra-VTA amylin and used fiber photometry to measure how intra-VTA amylin signaling altered mPFC dopamine release in response to a food reward. To determine whether the amylin receptor is expressed on VTA-to-mPFC projecting neurons, we retrogradely labeled this pathway with fluorogold and co-stained the VTA using immunohistochemistry and fluorescent in situ hybridization (FISH) for CTR. Because no CTR was detected on VTA-to-mPFC projections, we hypothesized that amylin signaling on VTA GABAergic (gamma-aminobutyric acidergic) interneurons may inhibit VTA-to-mPFC dopaminergic neurons. We pretreated the VTA with GABA receptor antagonists and silenced VTA GABA neurons using an inhibitory DREADD (designer receptor exclusively activated by designer drugs) to determine the role of intra-VTA GABA signaling in mediating the anorectic effects of intra-VTA amylin. Lastly, we used a monosynaptic retrograde rabies virus to label neurons that innervated VTA-to-mPFC projections and co-stained for Ctr and Gad1 (glutamate decarboxylase 1) using FISH to identify GABAergic CTR-expressing neurons in the VTA that synapse onto VTA-to-mPFC projections. See Supplemental Methods for more details.

Animals

Adult male Sprague Dawley rats (Charles River) or Long-Evans rats expressing cre-recombinase under the Gad1 promoter (GAD1Cre+) and GAD1Cre− littermate controls were used for all feeding studies. Rats were individually housed under a 12-hour light/dark cycle in a temperature- and humidity-controlled satellite vivarium and had ad libitum access to water and chow (5001; LabDiet) or a 60% HFD (D12492; Research Diets). Experiments were conducted under the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and all procedures were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania, the University of Southern California, and the University of Illinois at Chicago. See Supplemental Methods for more details.

Statistics

Sample sizes were determined by power analysis. All food intake and body weight studies (mPFC dopamine receptor agonists, VTA GABA receptor antagonists, VTA inhibitory DREADD) were analyzed using two-way repeated measures analysis of variance (ANOVA) followed by Tukey post hoc tests, analyzing multiple comparisons within a single time point. To compare artificial cerebrospinal fluid (aCSF) and amylin infusion on mPFC cFos expression, a two-way ANOVA with Bonferroni post hoc comparison within a buret type was performed. To analyze aCSF, amylin, and salmon calcitonin (sCT) injection on choice food intake, a two-way ANOVA with Tukey post hoc comparison within a diet type was used. To quantify results from in vivo fiber photometry experiments, the mean signal during the 10 seconds after a behavioral event (e.g., pellet retrieval) was measured on each trial and averaged across trials for each session. To compare aCSF and amylin infusions on pellet retrieval, a one-tailed paired t test was used. To compare aCSF and amylin infusions on impulsivity assayed with a DRL task, a two-tailed paired t test was used. All data are expressed as mean ± SEM. For all statistical tests, p < .05 was considered significant. All data were analyzed using GraphPad Prism 9.3.1 software (GraphPad Software; https://www.graphpad.com).

RESULTS

VTA Amylin Signaling Attenuates Food Reward–Induced mPFC Neuronal Activation and Impulsive-like Behaviors

First, we examined whether the native hormone amylin impacts food choice as has been shown for amylin peptide analogues and the CTR agonist sCT (10–12). Intraperitoneal injection of both amylin (20 μg/kg) and sCT (5 μg/kg) selectively decreased acute intake of HFD without altering chow intake in rats with access to both diets (Figure 1A), suggesting that CTR and amylin receptor agonists more powerfully attenuate intake of high-palatability foods. Consumption of rewarding foods, particularly lipid-rich foods, is associated with VTA dopamine neuron activation and concurrently induces expression of the immediate early gene cFos in downstream nuclei including the mPFC (13). We hypothesized that amylin signaling in the VTA reduces the hedonic value and intake of palatable foods by limiting food reward–induced activation of the VTA-to-mPFC circuit.

Figure 1.

Intra-VTA amylin signaling attenuates reward-induced mPFC neuronal activation and impulsive food-directed behavior. One-hour food intake following IP injection of amylin (20 μg/kg), sCT (5 μg/kg), or vehicle (aCSF) (A) (diet effect F_1,18 = 11.83, p = .0029; treatment effect F_1.981,35.65 = 8.963, p = .0007; interaction F_2,36 = 13.70, p < .0001). The consumption of water or intralipid (10%) for 1 hour following bilateral intra-VTA administration of amylin (0.4 μg/100 nL) or vehicle (aCSF) (B) (buret effect F_1,21 = 3.772, p = .013) and the mPFC cFos counts in these rats 90 minutes after drug injection (C) (treatment effect F_1,17 = 16.30, p = .009). The efficiency (rewards/lever presses) (D) (t₆ = 2.703), number of correct (E) (t₆ = 2.441) and incorrect (F) (t₇ = 0.6478) lever presses, and number of rewards earned (G) (t₇ = 0.3136) in 24 hour-fasted rats trained to perform a differential reinforcement of low rate of responding task after 1 hour following bilateral intra-VTA injection of amylin (0.4 μg/100 nL) or vehicle (aCSF). Direct comparisons are indicated. Data are presented as mean ± SEM. For (A), data are within subjects, n = 10/treatment, and bars that do not share a common letter (a, b) are significantly different. For (B, C), data are between subjects, n = 6–7/treatment. For (D–G), data are within subjects, n = 7–8/treatment. aCSF, artificial cerebrospinal fluid; Amy, amylin; HFD, high-fat diet; IP, intraperitoneal; mPFC, medial prefrontal cortex; NS, not significant; sCT, salmon calcitonin; veh, vehicle; VTA, ventral tegmental area.

To understand the effects of amylin on this circuit, we investigated whether intra-VTA amylin could diminish a lipid reward–induced increase in mPFC cFos. In rats trained to drink intralipid from a buret, intra-VTA amylin (0.4 μg) decreased intralipid intake compared with vehicle treatment (Figure 1B). On the test day, one half of the rats received a water buret, and as expected they reduced their intake from the previous intralipid day regardless of VTA treatment (Figure 1B). Surprisingly, amylin decreased mPFC cFos counts in both water and intralipid groups (Figure 1C), which may be attributed to amylin modulating dopamine release induced by the buret presentation, which represents a predictive cue that the rats associate with an intralipid reward (40). Because VTA amylin receptor activation controls dopaminergic projections to the NAc (16), we also quantified cFos in the NAc. Intra-VTA amylin did not significantly alter cFos counts in the NAc shell or core (Figure S1A, B).

The mPFC is a critical neural substrate regulating reward-driven behavior and is implicated in modulating impulsive behavior (41,42). We assessed the impact of VTA amylin treatment on a DRL task. DRL is an established operant chamber paradigm where rats must withhold from lever pressing for 20 seconds to receive a reward, which illustrates mPFC-mediated alterations in impulsive food-seeking behaviors (43–45). Under 24-hour food-restricted conditions, intra-VTA amylin (0.4 μg)-treated rats received more rewards per correct lever presses than control rats (Figure 1D; efficiency measure). Accordingly, vehicle-treated rats tended to press the paired lever more frequently (Figure 1E) (p = .0504), yet amylin- and vehicle-treated rats ultimately received the same number of reward pellets (Figure 1F). Thus, amylin increased the like-lihood of waiting the required period of time before pressing the reward lever, demonstrating less impulsive-like behavior. In fact, the efficiency measure in control rats fasted for 24 hours fell from the typical efficiency measure under fed conditions (~0.5) (46), while amylin-treated rats were able to retain their training. Both treatment groups pressed the incorrect lever similarly (Figure 1G). Importantly, amylin treatment did not decrease the number of rewards obtained, suggesting that the rat’s hunger or motivation to work for a reward was not diminished by VTA amylin signaling, but rather impulsive food-directed behavior was decreased.

Intra-VTA Amylin Decreases Palatable Food Intake by Limiting Phasic Dopamine Release and Dopamine Receptor Activation in the mPFC

Dopamine signaling at both D₁ and D₂ receptors are involved in stimulating impulsive behaviors (47–49), and feeding increases PFC dopamine levels (50). Indeed, an estimated 50% to 55% of total VTA neurons are dopaminergic (51,52), and intra-VTA amylin signaling has previously been shown to limit dopamine release at the NAc during sucrose pellet retrieval (16). We set out to determine whether VTA amylin action mitigates impulsive behavior by reducing dopamine signaling in the mPFC. Chow-maintained rats with bilateral cannula targeting both the mPFC and VTA (Figure 2A, B) were pretreated with subthreshold doses of a dopamine D₁ and D₂ receptor agonist cocktail (DR-A) (D2R/D3R agonist quinpirole [6 μg] and D1R/D5R agonist SKF-81297 [1.2 μg]) or vehicle in the mPFC prior to administration of intra-VTA amylin or vehicle. In vehicle-pretreated rats, intra-VTA amylin decreased food intake during the first hour following injections compared with vehicle controls, and this was not affected by DR-A pretreatment (Figure 2C). Cumulative food intake at 3, 6, and 24 hours after injections and 24-hour body weight change were not different between any treatment groups (Figure 2C, D). Given the role of the VTA in hedonic food intake and amylin’s preferential suppression of HFD (Figure 1A), we hypothesized that intra-VTA amylin would induce a more robust anorectic response in rats fed an HFD. Indeed, amylin suppressed HFD intake, and this hypophagic response was blocked by mPFC DR-A pretreatment at 3- and 24-hour food intake (Figure 2E). Control amylin-injected rats lost more body weight in 24 hours than both mPFC DR-A pretreated groups (Figure 2F). To understand whether signaling at mPFC dopamine D₁ or D₂ receptors was responsible for attenuating VTA amylin-induced hypophagia, we tested SKF-81297 or quinpirole separately against intra-VTA amylin. Neither DR-A alone was able to account for the effects seen with the combined cocktail (Figure 2G, H), suggesting that dopamine signaling at both D₁ and D₂ receptors in the mPFC is necessary for amylin-mediated suppression of palatable food intake.

Figure 2.

mPFC dopamine 1 and 2 receptor agonism mitigates the hypophagic effect of intra-VTA amylin on HFD. Cannula placement in the mPFC (A) and VTA (B). Cumulative chow food intake (C) (1-h VTA treatment effect F_1,13 = 14.20, p = .0023) and corresponding 24-hour body weight change (D) (VTA treatment effect F_1,13 = 4.870, p = .046), cumulative HFD intake (E) (1-hour mPFC treatment effect F_1,16 = 5.862, p = .028; VTA treatment effect F_1,16 = 15.16, p = .0013; 24-hour mPFC treatment effect F_1,16 = 5.475, p = .033; VTA treatment effect F_1,16 = 5.925, p = .027; interaction F_1,10 = 11.20, p = .0074) and corresponding 24-hour body weight change (F) (mPFC treatment effect F_1,16 = 7.524, p = .014) in rats injected in the mPFC with pretreatment of subthreshold doses of a DR-A cocktail (DR2/DR3 agonist quinpirole 6 μg/100 nL and DR1/DR5 agonist SKF-81297 1.2 μg/100 nL) or vehicle before intra-VTA injection of amylin (0.4 μg/100 nL) or vehicle (aCSF). Cumulative HFD intake (G) (1-h VTA treatment effect F_1,17 = 29.15, p < .0001; 24-hour VTA treatment effect F_1,17 = 11.87, p = .0031) and corresponding 24-hour body weight change (H) in rats injected in the mPFC with pretreatment of subthreshold doses of either quinpirole (6 μg/100 nL) or SKF-81297 (1.2 μg/100 nL) or vehicle before intra-VTA injection of amylin (0.4 μg/100 nL) or vehicle (aCSF). For (C–F), bars that do not share a common letter (a, b, c) within a time point are significantly different. For (G, H), direct comparisons are indicated. Data are presented as mean ± SEM. Each experiment is within subjects, and for (C, D), n = 10–13/treatment; for (E, F), n = 15–16/treatment; for (G, H), n = 11–15/treatment. *p < .05, **p < .01, ***p < .001. ACd, dorsal anterior cingulate cortex; aCSF, artificial cerebrospinal fluid; Amy, amylin; DR-A, dopamine receptor agonist; HFD, high-fat diet; IL, infralimbic cortex; mPFC, medial prefrontal cortex; NS, not significant; OFC, orbitofrontal cortex; PL, prelimbic cortex; Q, quinpirole; SK, SKF-81297; veh, vehicle; VTA, ventral tegmental area.

Having established that pharmacologically activating mPFC dopamine receptors blocks intra-VTA amylin-induced anorexia and weight loss, we next wanted to measure the effect of VTA amylin signaling on real-time mPFC dopamine levels in response to a food reward. We performed in vivo fiber photometry in rats trained to retrieve noncontingent high-fat/high-sugar pellets in an operant chamber. Rats were transfected with an AAV (adeno-associated virus) expressing the GRAB_DA2h dopamine sensor (53) and implanted with an optic fiber targeting the mPFC (Figure 3A, B) and a VTA cannula for drug administration (Figure 3C, D). Recordings show mPFC dopamine released following pellet retrieval increased in both amylin-(0.4 μg) and vehicle-treated rats, but intra-VTA amylin significantly decreased the magnitude of dopamine signaling (Figure 3E). In fact, the average dopamine signal during the first 10 seconds following pellet retrieval was ~75% lower in amylin-treated rats (Figure 3F). These recordings support the idea that VTA amylin signaling decreases VTA-directed dopamine release in the mPFC.

Figure 3.

Intra-VTA amylin decreases fiber photometry–detected mPFC dopamine release in response to a food reward. Optic fiber placements for mPFC dopamine recording in experimental rats shown on Paxinos and Watson coordinates (A) and representative image (B). Cannula placement for intra-VTA injections shown on Paxinos and Watson coordinates (C) and representative image (D). ΔF/F-score recording traces (E) from optic fibers in rats presented with a sucrose pellet 1 hour following intra-VTA injection of amylin (0.4 μg/100 nL) or vehicle (aCSF) and the mean ΔF/F-score during the 10 seconds after pellet retrieval (F). Direct comparisons are indicated. Data are presented as mean ± SEM. Data are within subjects and n = 4. aCSF, artificial cerebrospinal fluid; GRABDA, G protein–coupled receptor activation-based dopamine; mPFC, medial prefrontal cortex; veh, vehicle; VTA, ventral tegmental area.

VTA-to-mPFC Projecting Neurons Do Not Express the Calcitonin Receptor

Having shown that intra-VTA amylin decreased mPFC markers of neuronal activation (cFos) and phasic dopamine release in response to food reward and mitigated impulsive food-directed behavior, we next investigated how amylin modulates the VTA-to-mPFC circuit. To determine whether VTA-to-mPFC dopaminergic neurons express amylin receptors to directly affect mesocortical dopamine release, we injected the retrograde tracer fluorogold into the mPFC (Figure 4A) and visualized fluorogold-labeled neurons in the VTA (Figure 4B). We first performed immunohistochemistry for the dopamine neuron marker tyrosine hydroxylase (TH) and the primary subunit of the amylin receptor, CTR. Surprisingly, no CTR staining was observed on fluorogold-labeled neurons in the VTA (Figure S2A–F). We did observe fluorogold and TH colabeling as expected, and previous reports have estimated that 30% to 40% of VTA-to-mPFC projecting neurons are dopaminergic (52,54,55).

Figure 4.

Retrograde FG-labeled VTA→mPFC-projecting neurons do not express the calcitonin receptor. Representative image of mPFC fluorogold injection (A) and area of analysis for fluorogold tracing in the VTA (B) (yellow outline represents VTA). Panels (B–H) depict VTA sections stained for messenger RNA of the amylin receptor subunit (Ctr; pink), dopamine neurons (Th; yellow), GABA neurons (Gad1; green), and glutamate neurons (Vglut2VGLUT2; red) with fluorescence in situ hybridization followed by immunohistochemistry to rescue FG staining (blue). (C, D) Representative rostral VTA section stained for Ctr, Th, and Gad1 (C) and enhancement of the white outline in panel C (D). (E, F) Representative mid-VTA section stained for Ctr, Th, and Vglut2 (E) and enhancement of the white outline in panel E (F). (G, H) Representative caudal VTA section stained for Ctr, Gad1, and Vglut2 (G) and enhancement of the white outline in panel G (H). Quantification of neuronal population of all FG-labeled neurons in the rostral, mid, and caudal VTA as number/section (I) and as a percentage of FG neurons/section (J). Quantification of neuron population of all CTR-expressing neurons in the rostral, mid, and caudal VTA as number/section (K) and as a percentage of CTR neurons/section (L). Data are presented as mean ± SEM. Each staining combination was performed on slides from 6 different FG-injected animals, so n = 18 slides were quantified for each VTA level. cp, cerebral peduncle; CTR, calcitonin receptor; dtg, dorsal tegmental decussation; FG, fluorogold; GABA, gamma-aminobutyric acid; GAD1, glutamate decarboxylase 1; IP, interpeduncular nucleus; IPDM, interpeduncular nucleus dorsomedial; ml, medial lemniscus; mlf, medial logitudinal fasciculus; mPFC, medial prefrontal cortex; mtg, mammillotegmental tract; scp, superior cerebellar peduncle; TH, tyrosine hydroxylase; VGLUT2, vesicular glutamate transporter 2; VTA, ventral tegmental area; vtg, ventral tegmental decussation.

To better characterize the VTA-to-mPFC neuronal population and confirm the absence of Ctr messenger RNA on these projections, we performed FISH to stain for messenger RNA identifying dopaminergic (Th), GABAergic (Gad1), glutamatergic (Vglut2 [vesicular glutamate transporter 2]), and amylin receptor–expressing neurons (Ctr). Representative images are shown from rostral (Figure 4C, D), mid (Figure 4E, F), and caudal (Figure 4G, H) FISH-stained fluorogold-labeled VTA. We again observed no expression of Ctr in VTA fluorogold-labeled neurons, confirming that amylin receptor signaling does not directly modulate mesocortical neurons. Individual channels for the inserts displayed in Figure 4D, F, H are shown in Figure S3A–O.

We observed that VTA-to-mPFC projecting fluorogold-labeled neurons were more concentrated in the rostral VTA and decreased in more caudal sections (Figure 4I). The phenotype of fluorogold neurons in the rostral, mid, and caudal VTA (Figure 4I, J) and for the whole VTA (Figure S3P, Q) are consistent with other reports of VTA-to-mPFC projections (52,54).

Interestingly, CTR-positive neurons were more abundant in the mid and caudal VTA (Figure 4K). Classification of VTA CTR-expressing neurons revealed a prominent GABAergic population in more caudal VTA sections (Figure 4L). Previously, 62.6% of VTA CTR was quantified on dopaminergic neurons (16). Here, we report that in the whole VTA, 44.6% of CTR-positive neurons coexpressed TH, half of which were double-labeled with VGLUT2, while 20.9% expressed VGLUT2 alone, 28.1% expressed GAD1, and a remaining 6.4% did not coexpress one of these markers (Figure S3R, S).

Blocking VTA GABA Signaling Eliminates VTA Amylin-Induced Hypophagia

Despite having multiple lines of evidence supporting that VTA amylin signaling decreases mPFC activation and function, we failed to detect CTR expression on VTA-to-mPFC monosynaptic projections, which argues against amylin regulating this circuit through a direct mechanism. Importantly, the amylin receptor is made of CTR heterodimerized to a receptor activity–modifying protein (RAMP 1, 2, or 3) (56,57), and the VTA expresses comparable levels of the 3 RAMPs (17). Signaling at amylin receptors stimulates intracellular cAMP formation, ERK phosphorylation, and calcium mobilization (58), all of which would be expected to increase neurotransmission. Thus, if the amylin receptor was expressed on dopaminergic VTA-to-mPFC projections, its activation would likely increase dopamine release in the mPFC and not decrease it as was observed here. The VTA has GABA interneurons whose activation inhibits neighboring dopamine neuron transmission (59,60). Although this analysis did not distinguish between interneurons versus projecting neurons, roughly 30% of all VTA CTR is expressed on GABA neurons (Figure 3S). Therefore, we hypothesized that VTA amylin receptor signaling, through the stimulation of local GABA release, decreases the activation of mPFC-projecting dopaminergic neurons.

To investigate this potential mechanism, we tested whether blockade of VTA GABA receptors would prevent amylin-induced hypophagia. Pretreatment with subthreshold doses of a GABA_A and GABA_B receptor antagonist cocktail (GABA_A antagonist bicuculline methiodide [10 ng] and GABA_B antagonist 2-hydroxysaclofen [0.25 μg]) attenuated the anorectic response of intra-VTA amylin (0.4 μg), reaching significance at 6, 12, and 24 hours after injections and tended to reverse amylin-induced 24-hour body weight loss (p = .07) (Figure 5A, B). Antagonizing GABA receptors immediately weakened the effect of intra-VTA amylin by decreasing the amylin-induced delay to consume the first meal (Figure 5C). Intra-VTA amylin diminished the first meal size independent of pretreatment (Figure 5D).

Figure 5.

Pharmacological antagonism of VTA GABA_A and GABA_B receptors or DREADD-induced silencing of VTA GABA neurons attenuates intra-VTA amylin-induced hypophagia. Cumulative HFD intake (A) (1-hour second injection effect F_1,15 = 23.20, p = .0002; 24-hour second injection effect F_1,15 = 11.38, p = .0042; treatments interaction F_1,5 = 14.94, p = .012), 24-hour body weight change (B) (second injection effect F_1,13 = 8.074, p = .014; treatments interaction F_1,6 = 8.720, p = .026), latency to the first meal (C) (pretreatment effect F_1,48 = 7.764, p = .0076; second injection effect F_1,48 = 15.75, p = .0002), first meal size (D) (second injection effect F_1,14 = 12.31, p = .0035), and meal length (seconds/meal) (E) (1-hour second injection effect F_1,14 = 14.94, p = .0017) and meal size (grams/meal) (F) (1-hour second injection effect F_1,14 = 23.15, p = .0003; 24-hour treatments interaction F_1,7 = 7.167, p = .032) in rats injected with intra-VTA pretreatment of subthreshold doses of a GABA receptor antagonist cocktail (GABA_A antagonist bicuculline methiodide 10 ng/100 nL and GABA_B antagonist 2-hydroxysaclofen 0.25 μg/100 nL) or vehicle before intra-VTA injection of amylin (0.4 μg/100 nL) or vehicle (aCSF). Cumulative HFD intake (G) (1-hour IP treatment effect F_1,11 = 4.874, p = .049; VTA treatment effect F_1,11 = 17.93, p = .0014; 24-hour treatments interaction F_1,8 = 5.602, p = .046) and 24-hour body weight change (H) in GadCre1 and cumulative HFD intake (I) (1-hour VTA treatment effect F_1,14 = 14.78, p = .0018; 24-hour VTA treatment effect F_1,14 = 10.81, p = .0054) and 24-hour body weight change (J) (VTA treatment effect F_1,15 = 4.444, p = .052) in control GadCre- littermates injected IP with pretreatment of CNO (1 mg/kg) before intra-VTA injection of amylin (0.4 μg/100 nL) or vehicle (aCSF). Both GadCre+ and GadCre− rats received intra-VTA injections of an inhibitory cre-dependent DREADD AAV during surgery. For all panels, bars that do not share a common letter (a, b, c) within a time point are significantly different. Data are presented as mean ± SEM. For (A–F), data are within-subjects n = 12–14/treatment. For (G–H), data are within-subjects n = 10–12/treatment. For (I, J), data are within-subjects n = 15/treatment. AAV, adeno-associated virus; Amy, amylin; ant, antagonist; CNO, clozapine N-oxide; DREADD, designer receptor activated by designer drugs; GABA, gamma-aminobutyric acid; HFD, high-fat diet; IP, intraperitoneal; NS, not significant; sal, saline; veh, vehicle; VTA, ventral tegmental area.

To better understand how GABA receptor antagonism in-terrupts amylin’s suppression of feeding, we examined meal pattern structures. The number of meals and time spent engaged in eating were comparable between intra-VTA amylin-treated rats regardless of pretreatment (Figure S4A, B). Intra-VTA amylin robustly suppressed the meal length (seconds/meal) and size (grams/meal) 1 hour after injections compared with controls (Figure 5E, F). The amylin-induced decrease in average meal size 6 hours after treatment was completely reversed with GABA receptor blockade (Figure 5F). These data suggest that intra-VTA amylin signaling through stimulating local GABA release inhibits food intake primarily by decreasing meal size without affecting meal number or length. Earlier consumption of the first meal by blocking GABA receptors prior to amylin treatment also suggests that local GABA signaling is necessary for amylin-induced reductions in the impulsive drive to consume palatable foods.

While antagonizing VTA GABA_A and GABA_B receptors attenuated the anorectic response to amylin, the VTA receives numerous GABAergic inputs (61). Thus, pharmacological manipulation of GABA signaling could interfere with many off-target circuits. To target GABA release more specifically without affecting incoming GABA signaling to the VTA, we utilized a cre-dependent inhibitory DREADD AAV-construct to selectively silence VTA GABAergic neurons in transgenic rats expressing cre-recombinase under the Gad1 promoter (Figure S4C, D). In GAD1Cre+ rats, intraperitoneal injection of the DREADD ligand clozapine N-oxide (1 mg/kg) prior to intra-VTA amylin (0.4 μg) significantly increased food intake compared with vehicle amylin-treated rats at 1, 3, and 6 hours after injections (Figure 5G). In GAD1Cre− control rats, amylin treatment independent of pretreatment decreased food intake compared with controls, and rats did not exhibit food intake differences between clozapine N-oxide or vehicle pre-treatments at any time point (Figure 5I). Twenty-four-hour body weight change was not different between any treatment groups for either cre1 or cre- rats (Figure 5H, J).

VTA GABAergic Monosynaptic Inputs to mPFC-Projecting Neurons Express the Calcitonin Receptor

These data demonstrate that chemogenetic silencing of VTA GABAergic neurons eliminates intra-VTA amylin-induced hypophagia. However, DREADD expression was not genetically restricted to GABA interneurons. Therefore, off-target VTA GABAergic projections (62,63) were also silenced in this manipulation. To provide histological evidence that amylin signaling on VTA GABA interneurons is upstream of VTA-to-mPFC projecting neurons, we used a retrograde rabies virus strategy to label neurons that are presynaptic to VTA-to-mPFC projecting neurons (64,65).

To selectively target the VTA-to-mPFC pathway, we injected a retrograde cre-expressing AAV into the mPFC (Figure 6A) and 2 GFP (green fluorescent protein)–tagged cre-dependent helper viruses necessary for rabies virus expression into the VTA (66). One week later, the mCherry-tagged EnvA G-deleted rabies virus was injected into the same VTA location (Figure 6B). The different fluorescent tags on the helper and rabies viruses distinguish VTA-to-mPFC projecting starter cells (expressing GFP and mCherry) from monosynaptic retrograde inputs to these starter cells (expressing mCherry only) (Figure S5A–C). Injection of the helper viruses and rabies virus without first injecting the retrograde cre-AAV in the mPFC does not result in viral transfection or monosynaptic retrograde labeling (Figure S5D–F). Injection of only the helper viruses or the rabies virus alone without retrograde cre-expression also did not result in any VTA transfection (data not shown).

Figure 6.

RV retrograde labeling of monosynaptic inputs to VTA→mPFC–projecting neurons show calcitonin receptor expression on GABAergic inputs. Representative image of medial prefrontal cortex injected with a retrograde cre-AAV–expressing GFP (A) and VTA injected first with 2 cre-dependent helper viruses expressing GFP and then 1 week later with a EnvA G-deleted RV expressing mCherry whose expression is dependent on the helper viruses (B). The RV is taken into helper virus–expressing starter neurons (colabeled with GFP and mCherry) and then travels retrogradely one synapse to label monosynaptic inputs to the starter neurons (labeled only with mCherry). Representative image of VTA stained for messenger RNA of the amylin receptor (Ctr; pink) and GABA neurons (Gad1; blue) with fluorescence in situ hybridization followed by immunohistochemistry to rescue GFP (green) and mCherry (red) staining (C). Enhancement of the white boxes in panel C (D, F, H) and individual channels showing RV, Gad1, Ctr, and GFP labeling in the enhancement area D (E), area F (G), and area H. (I) Slides were stained from n = 7 different RV-injected animals. Cg1, cingulate cortex; CLi, caudal linear nucleus of the raphe; CTR, calcitonin receptor; GABA, gamma-aminobutyric acid; GAD1, glutamate decarboxylase 1; GFP, green fluorescent protein; IP, interpeduncular nucleus; IPR, interpeduncular nucleus caudal; M, motor cortex; mPFC, medial prefrontal cortex; PBP, parabrachial pigmented nucleus; PrL, prelimbic cortex; ml, medial lemniscus; RV, rabies virus; VTA, ventral tegmental area.

To identify amylin receptor expression on local GABAergic inputs to VTA-to-mPFC projections, we performed FISH staining for Gad1 and Ctr messenger RNA on rabies virus–injected VTA sections (representative image Figure 6C). Typically, 2 or 3 examples of Gad1/Ctr/mCherry-positive but GFP-negative cells, indicating a CTR-expressing GABAergic neuron that is presynaptic to VTA-to-mPFC projections, were observed per 20-mm section (Figure 6D, F, H). Individual channels for rabies virus, Gad1, Ctr, and GFP staining with a yellow arrow denoting the triple-labeled neuron are shown in Figure 6E, G, I. These results support that amylin receptors are expressed on GABAergic VTA neurons that provide input to mPFC projections and offer mechanistic evidence that VTA amylin signaling reduces palatable food intake by stimulating local GABA release to inhibit dopaminergic output in the mPFC.

DISCUSSION

The VTA and mPFC are implicated in behaviors related to reward seeking, impulsivity, and attention around feeding and other hedonic stimuli. Many postprandial hormones including leptin, insulin, GLP-1, and amylin are known to act in the VTA to control motivated behavior around appetite and substance use disorders (31,32,67,68). This is the first study to investigate how a satiation hormone affects mesocortical pathway activity and food reward–directed behaviors. Using multiple behavioral, pharmacological, and genetic approaches, we demonstrated that amylin signaling in the VTA reduced palatable food intake, impulsive behavior for a food reward, and mPFC dopamine release, likely through activation of local VTA GABA neurotransmission and inhibition of VTA-to-mPFC projecting dopamine neurons.

Although dopamine in the VTA is the most extensively studied neurotransmitter in the control of motivated behaviors, inhibitory signaling within the VTA is essential in regulating dopamine output. An estimated 50% to 80% of synaptic inputs to VTA dopaminergic neurons are GABAergic (69,70). GABA signaling in the VTA arises from both local GABA interneurons and long-range GABA projections (59,71–74). In fact, rewarding stimuli signaling outside the VTA often stimulate long-range GABA projections that inhibit VTA GABA interneurons and ultimately disinhibit dopamine neurons (67,73,75,76).

Increasing VTA GABA signaling is an established mechanism for satiation hormones like GLP-1 and leptin to decrease intake of palatable foods (33,67). Recently, leptin was found to also inhibit food intake and dopamine release to the NAc by activating VTA GABA interneurons (33). Leptin and amylin signaling in the VTA, which cooperatively promotes negative energy balance (68), appear to have many parallels including a sizable percentage (31% for leptin, 55% for amylin) of receptors expressed on TH-negative VTA neurons and no receptor expression on VTA-to-mPFC projecting neurons (33).

The results presented herein add amylin to the list of satiation hormones that increase VTA GABA signaling to inhibit dopamine transmission. We demonstrated that antagonism of VTA GABA receptors or genetic silencing of VTA GABA neurons blocks the anorectic response to intra-VTA amylin. While we did not specifically identify VTA interneurons, because VTA GABA interneurons lack consistent expression of canonical interneuronal markers that are common to other brain regions (71), we identified CTR expression on GABAergic neurons in the VTA that synapse onto long-range projections to the mPFC. Additionally, there may be CTR expressed on presynaptic terminals that modulate GABAergic input to the VTA, although we did not investigate this possible contribution.

There is a longstanding debate about the source of endogenous amylin that activates midbrain structures, including the VTA. Peripherally released amylin from pancreatic beta cells (77) could act in the VTA as amylin crosses the blood-brain barrier through a saturable mechanism or simple diffusion (78,79). In fact, radiolabeled amylin and fluorescent sCT are detected in the midbrain and VTA 10 and 30 minutes after peripheral administration, respectively (38,78). Alterna-tively, central sources of amylin have been identified in the preoptic area in lactating dams (80,81) and widespread centrally in neonatal pups (82). Higher levels of amylin gene expression in females in hypothalamic structures have also been reported (83), but this potential source and its projections have not been confirmed. A third possibility is that other neuropeptides that are known to or that are likely activate the amylin receptor complex, such as calcitonin-gene related peptide, sCT-like peptide, and amylin-like peptide (84–86) are the central ligands for VTA amylin receptors.

mPFC dopamine signaling from VTA terminals is complex, having both excitatory and inhibitory properties depending on the target receptor, the behavioral outcomes of which depend on mPFC activity state and synaptic dopamine concentrations (87–89). VTA dopamine released in the mPFC, particularly the infralimbic cortex, reinforces reward-seeking behavior and enhances attention to a stimulus cue (24,25). In fact, PFC neural activity increases immediately upon cue presentation, and this encoding information to the PFC is dependent on VTA dopamine (35). The changes we observed in mPFC phasic dopamine release during pellet retrieval occurred over seconds and may reflect amylin-dependent regulation of the reward signal evoked by a stimulus cue. Altering the response to a food cue and in turn the decision to engage with the food reward may explain how intra-VTA amylin-induced changes in dopamine neurotransmission that last seconds suppress food intake for hours.

There is not a clear delineation between VTA-to-mPFC and VTA-to-NAc projections in impulsive versus attentive behaviors (24,90–93). Our findings that intra-VTA amylin decreased impulsive food-directed behavior on a DRL task did not specifically target the VTA-to-mPFC pathway; thus, we cannot separate the involvement of projections to the mPFC versus the NAc. Nevertheless, these results show that intra-VTA amylin signaling curbs impulsive action toward obtaining a food reward.

Amylin receptor signaling is an attractive pharmaceutical target for the treatment of obesity because it improves body weight and appetite control in humans, has been shown to restore leptin responsiveness in obesity, and is cooperative with GLP-1 receptor ligands in food and body weight control (11,19,94–96). Our results that amylin lowers intake of high-fat/high-sugar food over chow, mitigates impulsive palatable food-directed behaviors, and modulates mesocortical signaling raises the possibility that amylin-based therapeutics may be especially effective in treating segments of the obese patient population who meet clinical criteria for binge-eating disorder. In this context, our data may offer a mechanism that explains the improved control of eating and reduced binge-eating scores in obese patients following pramlintide treatment (18). When viewed collectively, our findings highlight a novel mechanism for amylin’s regulation of the mesocortical pathway and broaden our understanding for developing pharmacotherapies to improve dysregulated feeding.

Supplementary Material

Supplementary material cited in this article is available online at https://doi.org/10.1016/j.biopsych.2023.07.011.

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.
Antibody	Rabbit Anti-cFos	Research Signaling	2250S
Antibody	Donkey Anti-Rabbit Secondary AlexaFluor 594	Jackson Immunoresearch
Antibody	Chicken Anti-GFP	Abcam	ab13970
Antibody	Donkey Anti-Chicken Secondary AlexaFluor 488	Jackson Immunoresearch
Antibody	Chicken Anti-Tyrosine Hydroxylase	Abcam	ab76442
Antibody	Rabbit Anti-CTR	Abcam	ab11042
Antibody	Rabbit Anti-Fluorogold	Fluorochrome, LLC
Antibody	Donkey Anti-Rabbit Secondary AlexaFluor 405	Jackson Immunoresearch
Antibody	Rabbit Anti-mCherry	Abcam	ab167453
Chemical Compound or Drug	sCT	Bachem	47931-85-1
Chemical Compound or Drug	Amylin	Bachem	124447-81-0
Chemical Compound or Drug	quinpirole hydrochloride	Sigma	85798-08-9
Chemical Compound or Drug	SKF-81297 hydrobromide	Sigma	253446-15-0
Chemical Compound or Drug	bicuculline methiodide	Tocris	40709-69-1
Chemical Compound or Drug	2-hydroxysaclofen	Tocris	117354-64-0
Chemical Compound or Drug	Clozapine N-oxide	Tocris	34233-69-7
Chemical Compound or Drug	Ketamine	Butler Animal Health Supply
Chemical Compound or Drug	Xylazine	Anased
Chemical Compound or Drug	Acepromazin	Butler Animal Health Supply
Commercial Assay Or Kit	RNAscope Multiplex Fluorescent V2	Advanced Cell Diagnostics, Inc.	323110
Commercial Assay Or Kit	rnCALCR FISH probe	Advanced Cell Diagnostics, Inc.	477791-C1
Commercial Assay Or Kit	rnTH FISH probe	Advanced Cell Diagnostics, Inc.	314651-C3
Commercial Assay Or Kit	rnGAD1 FISH probe	Advanced Cell Diagnostics, Inc.	316401-C2
Commercial Assay Or Kit	rnVGLUT2 FISH probe	Advanced Cell Diagnostics, Inc.	317011-C3
Genetic Reagent	DREADD AAV (pAAV-hSyn-DIO-hM4D(Gi)-mCherry	Addgene	44362
Genetic Reagent	AAV9-hsyn-GRAB_DA2h	Addgene	140554
Genetic Reagent	AAV2(retro)-eSYN-EGFP-T2A-iCre-WPRE	Vector Biolabs	VB4855
Genetic Reagent	EnvA G-Deleted Rabies-mCherry	Salk Institute for Biological Studies	32636
Genetic Reagent	pAAV-syn-FLEX-splitTVA-EGFP-tTA	Addgene	100798
Genetic Reagent	pAAV-TREtight-mTagBFP2-B19G	Addgene	100799
Organism/Strain	Rat: LE-Tg(Gad1-iCre)3Ottc	Rat Resource & Research Center
Organism/Strain	Rat: Sprague-Dawley	Charles River
Other	Chow	LabDiet	5001
Other	High fat diet	Research Diets	D12492
Other	artificial cerebrospinal fluid	Harvard Apparatus
Other	bilateral guide cannula: 26-gauge	Plastics One
Other	mircopump depressed: PHD 2000	Harvard Apparatus
Other	C&B Metabond	Parkell
Other	25 mLburet	Ace Glass Inc.	5758–04
Other	Microscope slides	Superfrost Plus, Fisher Scientific	12-550-15
Other	5% normal donkey serum	Jackson Immunoresearch
Other	Antifade Mounting Medium with DAPI	Vector Laboratories, Inc	Cat No.: H-1200
Other	Fluorescence Microscope	Keyence	Model: BZ-X810
Other	45 mg high-fat high-sugar food pellets	Bio-Serv	Product No: F05989
Other	Light-emitting diodes	Doric Lenses
Other	Filter cube	Doric Lenses	FMC4
Other	Visible Femtowatt Photoreceiver	Newport	Model: 2151
Other	Operant Chambers	Med Associates Inc	Product No. ENV-009A-CT
Other	Modular Pellet Dispenser	Med Associates Inc	Product No. ENV-203M-45
Other	Head Entry Detector	Med Associates Inc	Product No. ENV-254-CB
Other	Syringe Infusion Pump	New Era Pump Systems	Model NE-4000
Other	10% buffered formalin solution	Sigma Aldrich	Product No. HT501320
Other	Stag Microtome	Leica Biosystems	SM2010R
Other	Fluoromount G	SouthernBiotech	0100–01
Other	optic fiber: flat 400-μm core, 0.48 numerical aperture	Doric Lensees Inc
Other	Flurogold: 2% in aCSF	Fluorochrome, LLC
Other	Data Acquistion System	Tucker Davis Technologies	Model: RZ5P
Software; Algorithm	Synapse Suite	Tucker Davis Technologies
Software; Algorithm	GraphPad Prism 9.3.1	GraphPad Software