The central amygdala gates exogenous glucagon-like peptide 1 signals

Abstract

Nuclei within the limbic system like the central amygdala (CeA) play a critical role in mediating fear, motivation, reward, and appetitive behavior. Although previous reports demonstrate the presence of the glucagon-like peptide-1 receptor (GLP-1R) in limbic nuclei, how limbic neurons mediate the actions of systemically administrated GLP-1R agonists is unclear. In this study, we investigated the CeA's response to peripherally administered GLP-1R agonist Exendin-4 (Ex-4) in vivo, and determined the functional requirement of select CeA neuron populations in acute Ex-4 induced hypophagia. Using fiber photometry, we observed that Ex-4 promoted a rapid and lasting activation of CeA neurons that was blocked by pretreatment with the GLP-1R antagonist Exendin-9. We then tested the functional requirement of CeA neuron activation in mediating Ex-4 induced hypophagia of standard grain chow using inhibitory chemogenetics. Chemogenetic inhibition of all CeA neurons significantly suppressed the hypophagic actions of Ex-4. Then using selective mouse Cre-drivers, we found that chemogenetic inhibition of protein kinase c delta (Prkcd^CeA) and GLP-1R (Glp1r^CeA), but not somatostatin (Sst^CeA), neurons also attenuates the full hypophagic effect of Ex-4. Having observed that inhibition of Glp1r^CeA modestly attenuated Ex-4 induced hypophagia of standard chow, we then tested whether these neurons might mediate Ex-4 suppression of energy-dense, palatable diet. We used intermittent high-fat diet (HFD) access and found that inhibition of Glp1r^CeA neurons significantly rescued the reduction of HFD consumption by Ex-4. Collectively, these data demonstrate that the CeA responds to peripherally administered GLP-1R agonists and that multiple CeA neuron populations are required for the complete effect of GLP-1R agonist mediated hypophagia.

Highlights

• •Peripheral administration of Ex-4 activates CeA neurons, in vivo.
• •Chemogenetic inhibition of CeA and genetically defined Prkcd^CeA and Glp1r^CeA neurons attenuates Ex-4 induced hypophagia.
• •Chemogenetic inhibition of Glp1r^CeA neurons attenuates Ex-4 hypophagia of palatable, energy-dense food.

1. Introduction

Type 2 diabetes and obesity are serious public health maladies that affect many developed nations. In the U.S., 25% of the population is obese, which brings increased risk for more serious health complications, e.g., cardiovascular disease, cancer, and the development of type 2 diabetes [1]. Over the past ten years, the FDA has approved glucagon-like peptide-1 receptor (GLP-1R) agonists like liraglutide or semaglutide for the treatment of type 2 diabetes and weight management [2]. These drugs promote insulin release, delay gastric emptying, and decrease appetite the latter of which is mediated through the central nervous system (CNS) [[3], [4], [5], [6], [7]]. However, the precise circuitry and mechanisms in the CNS recruited by GLP-1R agonists to reduce food intake have yet to be fully characterized. In particular, further research is needed to establish the direct vs indirect recruitment of neural circuits by GLP-1R agonists.

GLP-1R-expressing cells within the CNS play an important role on physiology and behavior [[3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57]]. GLP-1R-expressing cells are evolutionarily conserved and enriched in brain nuclei that respond to exogenous GLP-1R agonists [[8], [9], [10],30,37,[58], [59], [60], [61], [62]]. The hypothalamus and hindbrain contain many GLP-1R + cells and these nuclei are required to mediate the anorexigenic effects of GLP-1R agonists [3,5,20,21,26,58]. However, GLP-1R expression is also enriched in limbic sites such as the bed nucleus of the stria terminalis (BNST), lateral septum (LS), and central nucleus of the amygdala (CeA) which are key nuclei that regulate motivation, reward, and appetitive behavior [15,39,48,53,63]. While GLP-1R-expressing or other cells within the BNST, LS, and CeA regulate feeding, the role of neuronal populations in these limbic nuclei in mediating responses to exogenous GLP-1R agonists is less clear [39,48,53,57].

The CeA contains heterogeneous GABAergic cell types that express distinct signaling molecules and neuropeptides [[64], [65], [66], [67], [68], [69], [70], [71], [72]]. The CeA's composition and input/output architecture allows it to integrate local signaling while also sending and receiving projections throughout the CNS [73,74]. Our lab has shown that GLP-1R-expressing neurons are distinct from other known genetically-defined neurons in the CeA, are enriched within the medial division of the CeA, and have unique excitability and electrophysiological properties [54]. Of particular note, multiple Fos studies have shown that, regardless of the route of administration, the type or dose of GLP-1R agonist used, species, or subject sex, neurons in the CeA are consistently activated [75]. Despite these reports, the CeA's role in mediating hypophagia by exogenously administered GLP-1R agonists has not been reported.

Among the genetically defined populations with the CeA, protein kinase C-delta (Prkcd)-expressing neurons in the CeA have been identified as a convergent node for multiple anorexigenic signals. Activation of Prkcd^CeA neurons suppresses food intake and their inhibition attenuates the anorexigenic effects of cholecystokinin (CCK) and lithium chloride [76]. However, whether Prkcd^CeA neurons are required for the hypophagic effects of systemically administered GLP-1R agonists has not been investigated. Somatostatin (Sst)-expressing neurons marks a separate population within the CeA that has minimal overlap with Prckd^CeA neurons [54] and reciprocally inhibits Prkcd^CeA neurons. Additionally, Sst^CeA neurons mediate appetitive and aversive behaviors, but their contribution to GLP-1R agonist induced hypophagia is unknown [68,[77], [78], [79]]. In addition to these populations, we reported that Glp1r-expressing neurons in the CeA are distinct from Prkcd+ and Sst + neurons. Recent evidence has demonstrated a direct anatomical connection between the stomach and the CeA and site-specific infusion of GLP-1R agonists is sufficent to induce hypophagia [39]. Additionally, conditional Glp1r knockout in the CeA selectively attenuates palatable food consumption following systemic administration of a GLP-1R agonist and optogenetic activation of Glp1r^CeA neurons is sufficent to selective suppress palatable food intake [80]. However, whether Glp1r^CeA neurons or the aforementioned subpopulations are functionally required for the hypophagic effects of systemically administered GLP1-R agonist has not been tested. In this study, we tested whether distinct, genetically-defined neuronal populations within the CeA are functionally required for the hypophagic effect of peripheral GLP-1R agonist administration.

2. Results

2.1. Peripheral GLP-1R signaling activates CeA neurons, in vivo

Post hoc Fos immunolabeling studies demonstrate that peripheral and intracerebroventricular administration of GLP-1R agonists activate the CeA [75]. To investigate this activation in vivo, we used the genetically-encoded calcium indicator GCaMP7 paired with fiber photometry recordings to capture the CeA's response to systemic administration of a GLP-1R agonist, Exendin-4 (Ex-4), in freely behaving mice. The CeA is primarily a GABAergic nucleus, therefore we bilaterally injected Slc32a1(vGat)-IRES-Cre mice with AAV-hSyn-FLEX-GCaMP7f into the CeA followed by bilateral implantation of fiber optics dorsal to the injection site (Figure 1A), resulting in the expression of GCaMP7f throughout the CeA and correctly targeted fiber optics (Figure 1B). Following recovery from surgery, habituation to behavior chambers, and handling, mice were subjected to randomized treatment (i.p.) receiving either Saline and Saline (Sal/Sal), Saline and Ex-4 (5 μg/kg) (Sal/Ex-4), Exendin-9 (50 μg/kg) and Saline (Ex-9/Sal), and Ex-9 and Ex-4 (Ex-9/Ex-4) (Figure 1C) during a recording session. Each mouse received two injections of their respective treatment regimen at 1 h timepoints over a 2.5 h recording (Figure 1D). In this approach, Ex-4 is always administered as the 2nd injection, with either Sal or Ex-9 administered as pretreatment at the 1sthour timepoint, allowing the assessment of GLP-1R-dependent activation. As expected due to the CeA's role in encoding pain and aversive stimuli, all mice showed strong and rapid responses to the injection itself (Figure 1E, S1A). For Sal/Ex-4 treatment, however, we observed a rapid and sustained activation of CeA neurons following 2nd treatment administration of Ex-4 when compared to Sal/Sal, Ex-9/Ex-4, and Ex-9/Sal (Figure 1E). No sustained activation of CeA neurons were observed following administration of the 1st treatment (S1A). Examination of Ex-4 responses in individual mice revealed that several mice show sustained and prominent increases in GCaMP signal when compared to saline alone (Figure 1F). We quantified the net area under the curve (AUC) from 15 to 30′ post-injection of the 2nd treatment, where the animal's response to injection has subsided (S1A). This analysis uncovered a significant increase in the Sal/Ex-4 treatment condition when compared to Sal/Sal, Ex-9/Ex-4, and Ex-9/Sal (Figure 1G). In comparison, there was no significant difference of net AUC 15–30′ post-injection of the 1st treatment between any of the conditions (S1B). We observed no significant difference in the peak response between any of the treatments following any drug administration (Figure 1H, S1C). In post hoc examination of the CeA in these mice, we confirmed GCaMP7 expression and documented fiber optic termination sites for each mouse (Fig. S1D–G). Collectively, these data demonstrate that systemic administration of the GLP-1R agonist Exendin-4 results in sustained activation of the CeA and is mediated through GLP-1R signaling.

Figure 1.

Systemic administration of the long acting GLP-1R agonist Exendin-4 results in lasting activation of CeA neurons in vivo. A. Surgical schematic of bilateral injection of AAV-hSyn-FLEX-GCaMP7 and fiber optic implants in CeA of Sc32a1(vGat)-Cre mice (n = 15–21). B. Representative image of GCaMP7 expression in the CeA and fiber optic termination site from a vGat-Cre injected mouse taken at 20× magnification. Scale bar = 200 μM. C. Table of treatment (Tx) regimen, randomized for each animal, given (i.p.) during each trial. D. Timeline for each photometry recording trial (0–2.5 h) with Tx given at 1- and 2-h timepoints. Shaded regions indicate 1 h time windows surrounding each injection period (1st injection, Blue; 2nd injection, Yellow) used for plotted traces. E. Z-scored ΔF/F traces of GCaMP7 fluorescence during a 1 h window (Panel D; yellow) surrounding each 2nd respective Tx for each trial (15–21 mice). F. Heatmap of calcium transients during 1Hr window for each animal during Saline only or Ex-4 only treatment (n = 21 mice). Data were sorted by descending area-under-the-curve calculations. G. Quantification of the net area under the curve (AUC) 15–30 min post 2nd Tx for each trial (n = 15–21 mice). p = 0.0014; Post-test: Sal/Sal vs Sal/Ex-4: p = 0.0081; Sal/Sal vs Sal/Ex-9: p > 0.9620; Sal/Sal vs Ex-9/Ex-4: p = 0.9932; Sal/Ex-4 vs Ex-9/Ex-4: p = 0.0429; Sal/Ex-4 vs Ex-9/Sal: p = 0.0277; Ex-9/Sal vs Ex-9/Ex-4: p = 0.9886. H. Quantification of the maximum Z-Score peak post-injection for each treatment (n = 15–21 mice). p = 0.6878. For G-H, data were analyzed using a mixed-effects model with Turkey's multiple comparisons. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 Error Bars for G-H represent standard error of the mean (SEM).

2.2. Inhibition of vGat^CeA neurons attenuates GLP-1-induced hypophagia

Having observed a rapid and sustained activation of CeA neurons in response to a systemic GLP-1R agonist, we tested the requirement of CeA neuron activation in Ex-4-mediated hypophagia. We used a chemogenetic strategy to reversibly silence all CeA neurons prior to Ex-4 administration. As the CeA is a GABAergic nucleus, we bilaterally injected vGat-Cre mice with either AAV-hSyn-DIO-mCherry or AAV-hSyn-DIO-hM4d-mCherry into the CeA (Figure 2A), resulting in mCherry expression throughout the CeA (Figure 2B). To validate the function of hM4d, we used whole-cell patch clamp electrophysiology in ex vivo CeA brain slices and recorded from hM4d-mCherry + neurons (Figure 2C). In current clamp mode, bath application of the designer ligand deschloroclozapine (DCZ, 1 μM) resulted in significant hyperpolarization after a 5-minute baseline recording (Figure 2D&E). Similarly, bath application of DCZ to CeA hM4d-expressing neurons resulted in a significant increase in rheobase (Figure 2F&H) and a significant decrease in the number of action potentials generated by current steps (Figure 2G&I). These data demonstrate that activation of hM4D by DCZ results in inhibition of neural activity and silencing of CeA neurons.

Figure 2.

Chemogenetic inhibition of CeA neurons attenuates systemic Exendin-4-mediated hypophagia. A. Surgical schematic of bilateral injection of either AAV-hSyn-DIO-mCherry (n = 8/sex) or AAV-hSyn-hM4d-mCherry (n = 8/sex) in the CeA of Sc32a1(vGat)-Cre mice. B. Representative image of hM4d-mCherry expressing CeA tissue from vGat-Cre injected mouse taken at 20× magnification. Scale bar = 200 μM. C. Schematic of vGat^CeA neurons in current clamp using whole-cell patch-clamp. D. Average electrophysiological trace of vGat^CeA neurons in current clamp using whole-cell patch-clamp following DCZ (1 μM) wash-on in the presence of tetrodotoxin and synaptic blockers. E. Quantification of change in membrane potential (D) pre- and post- DCZ wash-on. n = 8 cells. Wilcoxon test p = 0.0156. F. Rheobase of hM4d-expressing CeA neurons at baseline and after 5′ of 1 μM DCZ washon, paired t test p = 0.0156 G. Current step-evoked action potentials of hM4d-expressing CeA neurons at baseline and after 5′ of 1 μM DCZ washon, two-way repeated measures ANOVA Time X Drug: F(1.978, 17.80) = 9.954, p = 0.0013; Time: F (1.809, 16.28) = 12.24, p = 0.0007; Drug: F (1.000, 9.000) = 11.69, p = 0.0076 H. Representative traces of current ramp evoked action potentials in one cell at baseline and after DCZ. I. Representative traces of current step (+140) evoked action potentials in one cell at baseline and after DCZ. J. Representative illustration of mouse interacting with Feeding Experimental Device (FED3) under fixed-ratio 1 paradigm in order to receive a food pellet. K. Pellets retrieved by all vGat-Cre mice during each treatment (Tx) session. Group (Gr, mCherry/hM4d) X Tx: F(2.173, 65.20) = 5.194, p = 0.0067; Gr effect: F(1, 30) = 7.226, p = 0.0116; Tx effect: F(2.173, 65.20) = 102.2, p < 0.0001. Post tests - Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.3755), hM4 (p = 0.0064); DCZ/Veh vs Veh/Veh: mCh (p = 0.8494), hM4 (p = 0.0294). L. Active port pokes by all vGat-Cre mice during each Tx session. Gr X Tx: F(2.721, 81.62) = 4.528, p = 0.007; Gr effect: F(1, 30) = 5.558, p = 0.0251; Tx effect: F(2.721, 81.62) = 45.38, p < 0.0001. Post tests - Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9862), hM4 (p = 0.1180); DCZ/Veh vs Veh/Veh: mCh (p = 0.0679), hM4 (p = 0.5978). M. Table of treatment (Tx) regimen that is randomized for each animal. N. Pellets retrieved by male vGat-Cre mice during each Tx session. Gr X Tx: F(1.726, 24.17) = 2.909, p = 0.0804; Gr effect: F(1, 14) = 4.386, p = 0.0549; Tx effect: F(1.726, 24.17) = 96.91, p < 0.0001. Post tests - Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9974), hM4 (p = 0.0064); DCZ/Veh vs Veh/Veh: mCh (p = 0.3424), hM4 (p = 0.2469). O. Active port pokes by male vGat-Cre mice during each Tx session. Gr X Tx: F(1.838, 25.73) = 4.791, p = 0.0192; Gr effect: F(1, 14) = 4.210, p = 0.0594; Tx effect: F(1.838, 25.73) = 43.70, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9594), hM4 (p = 0.0382); DCZ/Veh vs Veh/Veh: mCh (p = 0.9480), hM4 (p = 0.6168). P. Timeline for each fasted-refeed/treatment session (i.p.) and endpoint (60 min) for data acquisition for analysis. Each animal is exposed to each Tx regime (M) at least once over the course of 4 sessions. Q. Pellets retrieved by female vGat-Cre mice during each Tx session. Gr X Tx: F(2.283, 31.96) = 5.795, p = 0.0054; Gr effect: F(1, 14) = 2.559, p = 0.1320; Tx effect: F(2.283, 31.96) = 37.68, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.1649), hM4 (p = 0.2600); DCZ/Veh vs Veh/Veh: mCh (p = 0.2246), hM4 (p = 0.1749). R. Active port pokes by female vGat-Cre mice during each Tx session. Gr X Tx: F(2.017, 28.24) = 2.858, p = 0.0736; Gr effect: F(1, 14) = 1.540, p = 0.2350; Tx effect: F(2.017, 28.24) = 15.51, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p > 0.9999), hM4 (p = 0.9399); DCZ/Veh vs Veh/Veh: mCh (p = 0.0366), hM4 (p = 0.9643). For K-L, N-O, and Q-R, data were analyzed using two-way repeated measures ANOVA with Tukey's multiple comparisons. Datapoints for each mouse during each Tx session were normalized to their respective Veh/Veh session. # Indicates a post test to Veh/Veh condition. Error bars for K-L, N-O, and Q-R represent SEM. Raw data shown in Fig. S2.

We then investigated the effect of chemogenetic inhibition of the CeA on GLP-1R-agonist-mediated hypophagia. To do this, we used a noninvasive method that allowed quantal assessment of both food intake and food seeking using the open-source FED3 device (Figure 2J) [82]. The use of FED3 enabled us to precisely quantify the number of nose pokes and pellet retrievals after delivery of GLP-1R agonists or any drug treatment in individual mice. After surgery, mice were singly-housed and trained to self-feed under a fixed-ratio 1 (FR1) paradigm where one poke in the active port resulted in the delivery of one 20 mg precision grain pellet. This operant paradigm allows the assessment of appetitive food-seeking (having to nose-poke the active port to dispense a pellet) and consummatory (retrieving and consuming the pellet) behaviors in discrete, time-stamped nose poke and pellet retrieval events within defined event windows, while closely approximating free feeding given its minimal operant demand [82]. While active pokes and pellets retrieved are expected to match closely under a FR1 paradigm, mice can demonstrate an increase in food-seeking behaviors by repeatedly poking the active port during active pellet dispensal [79]. Mice may select the active port as many times as they want but the individual pellet dispensed must be first retrieved before another one can be dispensed. We allowed all mice at least one week of self-feeding before beginning drug experiments. To test our hypothesis, we used four randomized drug treatment combinations (Figure 2M) in both mCherry (control) and hM4d-mCherry-expressing animals. The dosage of Ex-4 (5 μg/kg) was titrated to produce a ∼50% suppression in food intake, which allows for modulation of food intake in response to DCZ in either direction. To promote overall food intake, we food-deprived animals for 24 h prior to an experimental day. On the experimental day, animals received a randomized two-drug treatment regimen prior to FED3 access and refeeding (Figure 2P). As expected, we observed that mice consume 30–40 precision pellets (0.6–0.8 g) and ∼50 active port pokes in the first hour after food deprivation (Figure 2K-L, S2A-B), and Ex-4 pretreatment resulted in a 50–60% reduction in both pellets retrieved and active pokes performed in both sexes of mice. We focused on the first hour due to the rapid pharmacokinetics of DCZ [83]. Here we present data in normalized and raw pellet formats in the main figures and supplements, respectively. In control animals, DCZ treatment prior to Ex-4 resulted in no change in pellet retrieval and active pokes relative to Veh/Ex-4 conditions. In hM4d animals, however, we observed that DCZ treatment mitigated the hypophagic effect of Ex-4, but not completely. This resulted in a significant increase in pellet retrieval and active pokes in DCZ/Ex-4 conditions relative to Veh/Ex-4 in the hM4d group. We observed this phenomenon in male hM4d-expressing mice (Figure 2N-O, S2C-D), and although a similar trend was apparent in female mice, we failed to detect a significant attenuation of Ex-4 hypophagia in response to DCZ (Figure 2Q-R, S2E-F). In parallel, we also tested the effect of DCZ alone (DCZ/Veh). Inhibition of CeA neurons produced a subtle and significant increase in pellet retrieval, but not active pokes (Figure 2K-L, S2A-B) relative to Veh/Veh, with a nonsignificant increase observed in both male and female hM4d mice (Figure 2N-O & Q-R, S2C-F). Collectively, these data demonstrate that activation of CeA neurons is required for the complete hypophagic effect of the GLP-1R agonist Ex-4.

2.3. Inhibition of Pkcd^CeA neurons rescues GLP-1R agonist-induced hypophagia

Prkcd is an enriched gene that marks neurons in the lateral CeA (CeL). Prkcd^CeA neurons mediate the influence of other anorexigenic signals like CCK and inhibit feeding when activated [76]. With that, we investigated the role of Pkcd^CeA neurons in mediating Ex-4-induced hypophagia. Similar to our previous strategy to silence all GABAergic neurons in the CeA, we bilaterally injected either mCherry or hM4d into the CeA of Prkcd-Cre mice (Figure 3A), resulting in mCherry expression in the CeL (Figure 3B). We followed the same experimental approach as in Figure 2J, M, and P. In control animals, DCZ treatment prior to Ex-4 resulted in no change in pellet retrieval and active pokes relative to Veh/Ex-4 conditions (Figure 3C-D, S3A-B). In hM4d animals, however, there was a significant increase in pellet retrieval and active pokes relative to Veh/Ex-4. Similar to the inhibition of all CeA neurons, attenuation of Ex-4-mediated hypophagia was incomplete. When separated by sex, we observed a significant increase of pellets retrieval in both male and female hM4d mice (Figure 3E&G, S3C&E). For active pokes, we only observed a significant increase in female, but not male, hM4d mice (Figure 3F&H, S3D&F). When examining the effects of DCZ/Veh alone, inhibition of Pkcd^CeA neurons significantly increased pellet retrieval and active pokes in hM4d mice (Figure 3C-D, S3A-B) relative to Veh/Veh. In male mice, a nonsignificant increase was observed (Figure 3E-F S3C-D), but female mice increased their pellet retrieval and active pokes (Figure 3G-H, S3E-F). Collectively, these data demonstrate that activation of Pkcd^CeA neurons is required for the complete hypophagic effect of Ex-4.

Figure 3.

Chemogenetic inhibition of Prkcd^CeA neurons potently attenuates systemic Exendin-4-mediated hypophagia. A. Surgical schematic of bilateral injection of either AAV-hSyn-DIO-mCherry (n = 8/sex) or AAV-hSyn-hM4d-mCherry (n = 8/sex) in the CeA of Prkcd-Cre mice. B. Representative image of hM4d-mCherry-expressing CeA tissue from Prkcd-Cre mouse taken at 20× magnification. Scale Bar = 200 μM. C. Pellets retrieved by all Prkcd-Cre mice during each treatment (Tx) session. Group (Gr, mCherry/hM4d) X Tx: F(1.702, 51.06) = 18.03, p < 0.0001; Gr effect: F(1, 30) = 31.68, p < 0.0001; Tx effect: F(1.702, 51.06) = 145.8, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9740), hM4 (p < 0.0001); DCZ/Veh vs Veh/Veh: mCh (p = 0.6740), hM4 (p = 0.0013) D. Active port pokes by all Prkcd-Cre mice during each Tx session. Gr X Tx: F(2.003, 60.09) = 7.274, p = 0.0015; Gr effect: F(1, 30) = 14.91, p = 0.0006; Tx effect: F(2.003, 60.09) = 99.40, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9050), hM4 (p = 0.0047); DCZ/Veh vs Veh/Veh: mCh (p = 0.8286), hM4 (p = 0.0118) E. Pellets retrieved by male Prkcd-Cre mice during each Tx session. Gr X Tx: F(1.426, 19.96) = 4.670, p = 0.0316; Gr effect: F(1, 14) = 9.769, p = 0.0074; Tx effect: F(1.426, 19.96) = 49.28, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.3768), hM4 (p = 0.0200); DCZ/Veh vs Veh/Veh: mCh (p = 0.7379), hM4 (p = 0.1827) F. Active port pokes by male Prkcd-Cre mice during each Tx session. Gr X Tx: F(1.474, 20.64) = 2.889, p = 0.0911; Gr effect: F(1, 14) = 10.99, p = 0.0051; Tx effect: F(1.474, 20.64) = 39.49, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9868), hM4 (p = 0.3144); DCZ/Veh vs Veh/Veh: mCh (p = 0.9710), hM4 (p = 0.2576) G. Pellets retrieved by female Prkcd-Cre mice during each Tx session. Gr X Tx: F(2.088, 29.24) = 20.65, p < 0.0001; Gr effect: F(1, 14) = 29.14, p < 0.0001; Tx effect: F(2.088, 29.24) = 125.3, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.6484), hM4 (p = 0.0004); DCZ/Veh vs Veh/Veh: mCh (p = 0.9369), hM4 (p = 0.0010). H. Active port pokes by female Prkcd-Cre mice during each Tx session. Gr X Tx: F(2.399, 33.58) = 6.529, p = 0.0025; Gr effect: F(1, 14) = 4.738, p = 0.0471; Tx effect: F (2.399, 33.58) = 66.73, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.7470), hM4 (p = 0.0151); DCZ/Veh vs Veh/Veh: mCh (p = 0.3738), hM4 (p = 0.0372). For C-H, data were analyzed using two-way repeated measures ANOVA with Tukey's multiple comparisons. Datapoints for each mouse during each Tx session were normalized to their respective Veh/Veh session. # Indicates a post test to Veh/Veh condition. Error bars for C-H represent SEM. Raw data in Fig. S3.

2.4. Inhibition of Sst^CeA neurons have no effect on GLP-1R agonist-induced hypophagia

Somatostatin marks a neuronal population within the CeL that has minimal overlap with Pkcd^CeA [76]. In mice, Sst^CeA neurons mediate both appetitive and aversive behaviors and are interconnected with Pkcd^CeA neurons [68,[77], [78], [79]]. Similar to our previous experiments, we targeted Sst^CeA neurons to determine if they mediate GLP-1R agonist-induced hypophagia. We bilaterally injected either mCherry or hM4d into the CeA of Sst-Cre mice (Figure 4A), resulting in mCherry expression throughout the CeA, predominantly within the CeL (Figure 4B). In both control and hM4d animals, DCZ treatment prior to Ex-4 resulted in no significant differences in either pellet retrieval or active pokes relative to Veh/Ex-4 conditions (Figure 4C-D, S4A-B). When separated by sex, we observed no significant differences in male (Figure 4E-F, S4C-D) and female (Figure 4G-H, S4E-F) mice. When examining the effects of inhibition alone, inhibition of Sst^CeA neurons significantly decreased pellets retrieved in hM4d male mice of their normalized data (Figure 4E, DCZ/Veh) but not in their raw data (S4C) relative to Veh/Veh. Ultimately, these data demonstrate that activation of Sst^CeA neurons is not required for the complete hypophagic effect of Ex-4.

Figure 4.

Chemogenetic inhibition of Sst^CeA neurons has no effect on systemic Exendin-4-mediated hypophagia. A. Surgical schematic of bilateral injection of either AAV-hSyn-DIO-mCherry (n = 8/sex) or AAV-hSyn-hM4d-mCherry (n = 8/sex) in CeA for Sst-Cre mice. B. Representative image a hM4d-mCherry-expressing tissue from Sst-Cre mouse taken at 20× magnification. Scale bar = 200 μM. C. Pellets retrieved by all Sst-Cre mice during each treatment (Tx) session. Group (Gr, mCherry/hM4d) X Tx: F(1.893, 56.79) = 1.333, p = 0.2712; Gr effect: F(1, 30) = 1.295, p = 0.2641; Tx effect: F(1.893, 56.79) = 124.6, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.2854), hM4 (p = 0.3097); DCZ/Veh vs Veh/Veh: mCh (p = 0.9548), hM4 (p = 0.5872). D. Active port pokes by all Sst-Cre mice during each Tx session. Gr X Tx: F(2.322, 69.66) = 0.6302, p = 0.5585; Gr effect: F(1, 30) = 0.8424, p = 0.3660; Tx effect: F(2.322, 69.66) = 114.0, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.8651), hM4 (p = 0.8351); DCZ/Veh vs Veh/Veh: mCh (p = 0.9998), hM4 (p = 0.6391). E. Pellets retrieved by male Sst-Cre mice during each Tx session. Gr X Tx: F(2.214, 31.00) = 4.811, p = 0.0128; Gr effect: F(1, 14) = 3.963, p = 0.0664; Tx effect: F(2.214, 31.00) = 100.3, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.3220), hM4 (p = 0.6511); DCZ/Veh vs Veh/Veh: mCh (p = 0.9972), hM4 (p = 0.0199). F. Active port pokes by male Sst-Cre mice during each Tx session. Gr X Tx: F(3, 42) = 2.583, p = 0.0793; Gr effect: F(1, 14) = 4.338, p = 0.0561; Tx effect: F(2.469, 34.56) = 73.58, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9993), hM4 (p = 0.8733); DCZ/Veh vs Veh/Veh: mCh (p = 0.9531), hM4 (p = 0.0686). G. Pellets retrieved by female Sst-Cre mice during each Tx session. Gr X Tx: F(1.899, 26.59) = 0.07238, p = 0.9277; Gr effect: F(1, 14) = 0.002488, p = 0.9609; Tx effect: F(1.899, 26.59) = 50.50, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.8671), hM4 (p = 0.5803); DCZ/Veh vs Veh/Veh: mCh (p = 0.9259), hM4 (p = 0.9829). H. Active port pokes by female Sst-Cre mice during each Tx session. Gr X Tx: F(2.112, 29.56) = 0.1134, p = 0.9026; Gr effect: F(1, 14) = 0.2487, p = 0.6257; Tx effect: F(2.112, 29.56) = 46.86, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.5993), hM4 (p = 0.9575); DCZ/Veh vs Veh/Veh: mCh (p = 0.9669), hM4 (p = 0.9926). For C-H, data were analyzed using two-way repeated measures ANOVA with Tukey's multiple comparisons. Datapoints for each mouse during each Tx session were normalized to their respective Veh/Veh session. # Indicates a post test to Veh/Veh condition. Error bars for C-H represent SEM. Raw data in Fig. S4.

2.5. Inhibition of Glp1r^CeA neurons attenuates GLP-1R agonist-induced hypophagia

GLP-1Rs are expressed in the CeA, enriched within the medial division of the CeA (CeM), and do not overlap with other genetic markers like Somatostatin and PKCδ [54]. Exogenous administration of GLP-1R agonists consistently activates the CeA and decreases food intake, but whether Glp1r^CeA neurons are necessary to mediate GLP-1R agonist-induced hypophagia has not been tested [75]. We bilaterally injected mCherry or hM4d into the CeA of Glp1r-Cre mice (Figure 5A), resulting in robust mCherry expression within the CeM as we previously demonstrated (Figure 5B). In control animals, DCZ treatment prior to Ex-4 resulted in no change in pellets retrieval and active pokes relative to Veh/Ex-4 conditions (Figure 5C-D, S5A-B). In hM4d animals; however, we observed a significant increase of pellet retrieval and a non-significant increase of active pokes relative to Veh/Ex-4 conditions in their normalized data (Figure 5C-D), although treatment-induced changes were not detected in raw pellet retrieval and active pokes (S5A-B). When grouped data were parsed by sex, we observed a nonsignificant increase in male (Figure 5E-F, S5C-D) and female (Figure 5G-H, S5E-F) hM4d mice. When examining the effects of DCZ/Veh alone, Glp1r^CeA neuron inhibition did not significantly alter pellet retrieval and active pokes of hM4d mice (Figure 5C-H, S5A-F) relative to Veh/Veh. Collectively, these data demonstrate that activation of Glp1r^CeA neurons is required for the complete hypophagic effect of Ex-4.

Figure 5.

Chemogenetic inhibition of Glp1r^CeA neurons subtly attenuates systemic Exendin-4-mediated hypophagia. A. Surgical schematic of bilateral injection of either AAV-hSyn-DIO-mCherry (n = 8/sex) or AAV-hSyn-hM4d-mCherry (n = 8/sex) the in CeA of Glp1r-Cre mice. B. Representative image of hM4d-mCherry expressing CeA tissue from Glp1r-Cre mouse taken at 20× magnification. Scale Bar = 200 μM. C. Pellets retrieved by all Glp1r-Cre mice during each treatment (Tx) session. Group (Gr, mCherry/hM4d) X Tx: F(2.194, 65.83) = 1.148, p = 0.3270; Gr effect: F(1, 30) = 0.2528, p = 0.6188; Tx effect: F(2.194, 65.83) = 118.2, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.6219), hM4 (p = 0.0290); DCZ/Veh vs Veh/Veh: mCh (p = 0.0644), hM4 (p = 0.2987). D. Active port pokes by all Glp1r-Cre mice selected during each Tx session. Gr X Tx: F (2.155, 64.66) = 1.107, p = 0.3400; Gr effect: F(1, 30) = 0.6597, p = 0.4231; Tx effect: F (2.155, 64.66) = 67.79, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9937), hM4 (p = 0.1459); DCZ/Veh vs Veh/Veh: mCh (p = 0.1627), hM4 (p = 0.4084). E. Pellets retrieved by male Glp1r-Cre mice during each Tx session. Gr X Tx: F(2.311, 32.35) = 0.2455, p = 0.8140; Gr effect: F(1, 14) = 0.01258, p = 0.9123; Tx effect: F(2.311, 32.35) = 80.73, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.8308), hM4 (p = 0.4447); DCZ/Veh vs Veh/Veh: mCh (p = 0.2264), hM4 (p = 0.6472). F. Active port pokes by male Glp1r-Cre mice during each Tx session. Gr X Tx: F(2.689, 37.65) = 2.263, p = 0.1030; Gr effect: F(1, 14) = 2.070, p = 0.1722; Tx effect: F(2.689, 37.65) = 79.00, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.1063), hM4 (p = 0.2277); DCZ/Veh vs Veh/Veh: mCh (p = 0.5166), hM4 (p = 0.4588). G. Pellets retrieved by female Glp1r-Cre mice during each Tx session. Gr X Tx: F(1.972, 27.61) = 1.095, p = 0.3479; Gr effect: F(1, 14) = 0.3242, p = 0.5781; Tx effect: F(1.972, 27.61) = 42.94, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.4631), hM4 (p = 0.1093); DCZ/Veh vs Veh/Veh: mCh (p = 0.3607), hM4 (p = 0.5846). H. Active port pokes by female Glp1r-Cre mice during each Tx session. Gr X Tx: F(1.747, 24.45) = 0.6367, p = 0.5171; Gr effect: F(1, 14) = 0.009802, p = 0.9225; Tx effect: F (1.747, 24.45) = 21.74, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.6059), hM4 (p = 0.6198); DCZ/Veh vs Veh/Veh: mCh (p = 0.3114), hM4 (p = 0.5861). For C-H, data were analyzed using two-way repeated measures ANOVA with Tukey's multiple comparisons. Datapoints for each mouse during each Tx session were normalized to their respective Veh/Veh session. # Indicates a post test to Veh/Veh condition. Error bars for C-H represent SEM. Raw data in Fig. S5.

2.6. Inhibition of Glp1r^CeA neurons attenuates the effect of Ex-4 on palatable and energy-dense food

In a previous study, we demonstrated a population of prepronociceptin-expressing neurons in the CeA was important for the consumption of energy-dense and highly palatable food [66]. Furthermore, one of the hallmark effects of GLP-1R agonists is their ability to reduce preference for and consumption of energy-dense, palatable foods [12,84,85]. Thus, we reasoned that Glp1r^CeA neurons might be tuned to mediate suppression of palatable high-fat diet (HFD) by Ex-4.

To test this hypothesis, we generated new mice using the same approach used in Figure 5. As FED3 is incompatible with the use of HFD, we used standard HFD pellets weighed manually. To avoid entrainment, ad libitum chow fed mice were allowed access to HFD once every four days in combination with chow but receiving randomized two drug treatments as described in Figure 2M. In this way, we used a model of hedonic hyperphagia to test the contribution of Glp1r^CeA neurons to the effects of exogenous Ex-4 (Figure 6A-B). In mCherry control animals, 5 μg/kg Ex-4 produced a ∼60% reduction in HFD intake while also increasing the relative percent/amount of chow during this access period (Figure 6C-D, S6A-B). We observed no effect of DCZ pretreatment in this group with Ex-4 or on its own. In contrast, DCZ pretreatment of the hM4d group produced a highly significant attenuation of the hypophagic effects of Ex-4 on HFD intake (Figure 6C,S6A). This effect was further reflected in total kilocalorie (kCal) consumption of both HFD and chow, with hM4d animals consuming significantly more during DCZ/Ex-4 relative to Veh/Ex-4 (S6C). Although a decreasing trend was observed, DCZ did not significantly rescue the increase in percent/amount of chow intake or decreased HFD preference seen with Ex-4 (Figure 6D, S6B). When separated by sex, we observed similar phenomena. Chemogenetic inhibition of Glp1r^CeA neurons produced a significant attenuation of Ex-4's suppression of HFD intake in both male and female mice (Figure 6E&G, S6D&G). This attenuation was also seen in total kCal consumed in hM4d male mice (S6F), with a similar but non-significant trend observed in females (S6I) without producing a significant rescue of HFD preference in both male and female mice (Figure 6F&H, S6E&H). Similar to the combined intake observations, DCZ pretreatment did not significantly rescue HFD preference or percent/amount chow intake in either sex (Figure 6F&H, S6E&H).

Figure 6.

Chemogenetic inhibition of Glp1r^CeA neurons attenuates systemic Exendin-4-mediated reduction in palatable food intake. A. Surgical schematic of bilateral injection of either AAV-hSyn-DIO-mCherry (n = 8/sex) or AAV-hSyn-hM4d-mCherry (n = 8/sex) into the CeA of Glp1r-Cre mice. B. Representative illustration of timeline for each intermittent high-fat diet (HFD) access/treatment (Tx) session (i.p.) and endpoint (60 min) for data acquisition for analysis. Each animal is exposed to each Tx regime (Fig. 2M) at least once over the course of 4 sessions. C. Relative percentage of HFD consumed by all Glp1r-Cre mice during each Tx session. Group (Gr, mCherry/hM4d) X Tx: F(1.969, 59.07) = 8.847, p = 0.0005; Gr effect: F(1, 30) = 14.86, p = 0.0006; Tx effect: F(1.969, 59.07) = 120.9, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0. 0.9742), hM4 (p = 0.0001); DCZ/Veh vs Veh/Veh: mCh (p = 0.6910), hM4 (p = 0.1105). D. Percent of chow consumed by all Glp1r-Cre mice selected during each Tx session. Gr X Tx: F(1.972, 59.15) = 1.221, p = 0.3019; Gr effect: F(1, 30) = 0.1699, p = 0.6831; Tx effect: F(1.972, 59.15) = 31.78, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9977), hM4 (p = 0.4654); DCZ/Veh vs Veh/Veh: mCh (p = 0.8345), hM4 (p = 0.9997). E. Relative percentage of HFD consumed by male Glp1r-Cre mice during each Tx session. Gr X Tx: F(1.922, 26.91) = 4.252, p = 0.0261; Gr effect: F(1, 14) = 11.15, p = 0.0049; Tx effect: F(1.922, 26.91) = 51.52, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9999), hM4 (p = 0.0288); DCZ/Veh vs Veh/Veh: mCh (p = 0.9490), hM4 (p = 0.6828). F. Percent of chow consumed by male Glp1r-Cre mice during each Tx session. Gr X Tx: F(2.072, 29.00) = 4.412, p = 0.0202; Gr effect: F(1, 14) = 6.603, p = 0.0223; Tx effect: F(2.072, 29.00) = 25.02, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9657), hM4 (p = 0.8740); DCZ/Veh vs Veh/Veh: mCh (p = 0.6030), hM4 (p = 0.8750). G. Relative percentage of HFD consumed by female Glp1r-Cre mice during each Tx session. Gr X Tx: F(1.739, 24.34) = 6.232, p = 0.0085; Gr effect: F(1, 14) = 4.157, p = 0.0608; Tx effect: F(1.739, 24.34) = 71.24, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.7839), hM4 (p = 0.0235); DCZ/Veh vs Veh/Veh: mCh (p = 0.8976), hM4 (p = 0.3790). H. Percent of chow consumed by female Glp1r -Cre mice during each Tx session. Gr X Tx: F(1.994, 27.92) = 2.987, p = 0.0669; Gr effect: F(1, 14) = 2.269, p = 0.1542; Tx effect: F(1.994, 27.92) = 14.70, p < 0.0001. Post tests: Veh/Ex4 vs DCZ/Ex4: mCh (p = 0.9557), hM4 (p = 0.0638); DCZ/Veh vs Veh/Veh: mCh (p = 0.9774), hM4 (p = 0.9803). For C-H, data were analyzed using two-way repeated measures ANOVA with Tukey's multiple comparisons. For C, E, G, datapoints for each mouse during each Tx session were normalized to their respective Veh/Veh session. For D, F, H, percentage of chow consumed was calculated by dividing the consumed gram amounts of standard chow pellet by the summation of consumed gram amount of both standard chow pellet and HFD pellet in respect to each animal for each Tx session. # Indicates a post test to Veh/Veh condition. Error bars for C-H represent SEM. Raw data in Fig. S6.

Consistent with FR1 refeed in Figure 5, these findings confirm that Glp1r^CeA inhibition attenuates Ex-4-induced hypophagia during operant and free-access feeding paradigms. These data demonstrate that activation of Glp1r^CeA neurons is required for the complete suppression of palatable food intake produced by Ex-4.

3. Discussion

In this study, we observed that peripheral administration of Ex-4 increases CeA neuronal activity in freely behaving mice and that this was blocked by a systemic GLP-1R antagonist. Importantly, we observed heterogenous responses across animals, and this may be due to fiber placement and/or recording of specific subpopulations. Further efforts are needed to reveal the specific cell types activated by Ex-4 or other GLP-1R compounds and the mechanisms, direct or indirect, by which this activation occurs through GLP-1R.

At a behavioral level, chemogenetic inhibition of total and select CeA neuronal populations increased feeding in the presence of systemic Ex-4; however, Ex-4 induced hypophagia was incompletely rescued. It is likely that the engagement of parallel anorexigenic circuitry throughout the CNS continues to mediate GLP-1R signaling, as well as the potential contribution of uncharacterized CeA cell types. Indeed, the partial attenuation observed during complete CeA inhibition underscores that the CeA operates as a critical node within a broader, brain-wide anorexigenic system. Gabery and others have shown that systemic GLP-1R agonists directly activate the hindbrain as shown by the induction of Fos and accumulation of fluorescently-conjugated semaglutide. They also observed activation of the CeA and other brain regions where fluorescently-conjugated semaglutide was not detected. These data suggest that there exist primary and secondary sites of activation in response to peripheral GLP-1R agonists. This is consistent with the anatomical wiring of the CeA positioning it to gate and translate upstream GLP-1R signals into coordinated behavioral outputs and supported by our prior data [12,22,75]. In contrast, Qiao and others reported that an electrical lesion of the CeA in male rats minimally attenuated the hypophagic effects of Ex-4 on homeostatic chow intake but did attenuate high sucrose consumption following a 24 h fast [57]. We speculate that a CeA lesion disrupts fibers of passage and eliminates multiple opposing populations that promote and constrain the actions of peripheral GLP-1R agonists. In support of this notion, Prkcd^CeA and Sst^CeA neurons inhibit one another to regulate fear learning [78].

Beyond parallel circuitry, the population specificity of our findings suggests that the CeA gates the hypophagic effect of exogenous GLP-1R agonists through heterogeneous and functionally distinct neuronal populations. The robust attenuation of Ex-4 hypophagia by Prkcd^CeA inhibition indicates that upstream projections likely converge on the CeA along with parallel GLP-1R circuitry in other CNS nuclei. This is consistent with Prkcd^CeA neurons being known to mediate anorexigenic signals [76]. Additionally, that Sst^CeA inhibition produced no significant attenuation of Ex-4 induced hypophagia highlights the specificity of these effects. The opposing relationship between Prkcd^CeA and Sst^CeA neurons as demonstrated for fear learning and pain encoding raises the possibility that intact reciprocal inhibition within the CeA is important for the processing and translation of GLP-1R signals into behavioral outputs. Silencing one population may disinhibit the other, potentially explaining why full rescue was never achieved. Taken together, these findings suggest that exogenous GLP-1R agonists engage the CeA through select neuronal populations, particularly Prkcd^CeA neurons, and likely coordinate with downstream nuclei to gate the complete hypophagic response to peripheral GLP-1R agonist administration.

Throughout the majority our chemogenetic experiments, we assessed both pellet retrieval and active pokes to distinguish consummatory and appetitive components of feeding behavior [19,20,51]. All animals were well trained on the FR1 paradigm prior to treatment exposure and each animal's Veh/Veh session served as a within-subject baseline for their learned operant strategy [82]. Therefore, treatment-induced changes in active pokes reflect shifts in appetitive behavior, the motivation to seek food, beyond changes in consumption itself (pellet retrieval) [19,20]. Generally, active poke effects paralleled pellet retrieval effect, suggesting that CeA chemogenetic inhibition in the presence of Ex-4, and without it, modulates the motivation to seek food, in addition to the amount consumed. The magnitude and significance of active pokes effects, however, did not always match pellet effects across all populations and sexes. Chemogenetic inhibition of male vGat^CeA significantly rescued both pellet retrieval and active pokes during Ex-4 treatment, while inhibition of male Prckd^CeA male mice only rescued pellet retrieval and not active pokes. Chemogenetic inhibition of female Prckd^CeA mice rescued both pellet retrieval and active pokes but neither were significant for female vGat^CeA. These dissociations may reflect differences in specific CeA populations modulating appetitive versus consummatory components of feeding [79], even more so potentially depending on sex, or could arise from higher variability in active pokes due to individual differences in mice's operant strategy. Importantly, where active pokes effects diverge from pellet retrieval effects, the direction of these changes is consistently aligned.

While this study focused on the role of the CeA in mediating the hypophagic actions of exogenous GLP-1R agonists, our data also suggest that our identified populations likely play an endogenous role in constraining food intake under our assay conditions. Specifically, chemogenetic inhibition of all CeA neurons and Prkcd^CeA neurons alone increased food intake following a 24 h fast. That exogenous GLP-1R agonists coopt existing anorexigenic circuitry to mediate its effects is not surprising and replicates previous reports [76]. For GLP-1R populations, chemogenetic inhibition of GLP-1R-expressing cells in the paraventricular nucleus of the hypothalamus, lateral septum, or dossal vagal complex all increase food intake in the absence of any exogenous drug perturbation [21,50,86]. These baseline effects, however, also raises important interpretive consideration. Our strategy using chemogenetic inhibition alone cannot fully distinguish whether CeA populations directly gate Ex-4's hypophagic effects or whether CeA inhibition produces a parallel hyperphagic effect that sums with intact GLP-1R signaling elsewhere. However, the effect of DCZ alone varies across populations; Prckd^CeA inhibition produced a robust increase in food consumption, vGat^CeA inhibition produced a modest increase, Sst^CeA inhibition produced a modest decrease in males only, and Glp1r^CeA inhibition did not significantly alter feeding. Additionally, Glp1r^CeA inhibition still significantly attenuated Ex-4's suppression of palatable food, while Sst^CeA inhibition failed to attenuate Ex-4. This heterogeneity is not readily explained by a purely additive model. Nonetheless, we cannot exclude the contribution of parallel pathway contributions. Future studies using complementary gain-of-function and circuit-level approaches would assist in resolving this distinction.

In these studies, we examined sex differences in food consumption during chemogenetic inhibition of the CeA neuronal populations. Depending on the population of interest, we either did or did not detect sex differences. Across species, endogenous hormones such as estrogens and androgens modulate sex differences in feeding. [87]. Furthermore, exogenous GLP-1R agonists produce stronger hypophagic effects in females in a estrus cycle-dependent manner [38,88,89]. In our studies, we did not account for the estrus cycle in female subjects and this may have impacted responses to Ex-4 during chemogenetic inhibition and the observed variability, particularly when inhibiting whole CeA. However, there is a notable contrast where activation of Prkcd^CeA neurons is necessary in female mice for Ex-4 induced hypophagia relative to both male and female mice vGat^CeA and even male Prkcd^CeA mice. This was also observed when assessing total kCal consumed during intermittent HFD assay, where activation Glp1r^CeA neurons are necessary in male mice for Ex-4 induced hypophagia relative to female Glp1r^CeA mice. We speculate that select neuronal populations may be sensitive to cycle-dependent hormonal and affect the response to Ex-4 during chemogenetic inhibition, potentially indicating that select neuronal populations may be modulated by sex-dependent factors. Further, studies of this design would benefit from carefully tracking estrus cycle or including ovariectomized females in combination with estrogen analog treatments.

Chemogenetic inhibition of Glp1r^CeA neurons produced a notably stronger attenuation of Ex-4 hypophagia on HFD consumption (30%) compared to standard grain-based diet (10%). The attenuation of Ex-4 hypophagia on grain-based diet by Glp1r^CeA inhibition, while significant in normalized data, was not detected in raw intake values. This underscores the modest magnitude of this effect on homeostatic feeding. In contrast, the rescue of HFD consumption was robust across both normalized and raw values, both combined and separated sexes. This rescue was further reinforced by changes in total kCal intake, which was significantly increased in hM4d mice, particularly in males, during chemogenetic inhibition in the presence of Ex-4. This dissociation posits that Glp1r^CeA neurons are preferentially engaged by Ex-4 to suppress palatable, energy-dense food intake rather than homeostatic feeding of calorically neutral food. Importantly, when animals were presented with both HFD and standard chow during intermittent HFD assays, Ex-4 decreases HFD preference resulting in more chow intake. Glp1r^CeA inhibition resulted in a nonsignificant decrease in this change in preference in both sexes combined and in females specifically. This indicates that Glp1r^CeA inhibition did not just rescue HFD consumption but also returned preference towards palatable, energy-dense diet consumption. Consistent with a recent publication, this behavioral evidence suggests that Glp1r^CeA neurons may suppress hedonic-driven rather than homeostatic feeding [80]. Overall, this distinction may be due to differences in diet palatability. In a previous study, we demonstrated that Pnoc^CeA neurons are activated following intermittent HFD access and specifically promote palatable food consumption [66]. Given that inhibiting Glp1r^CeA neurons attenuates Ex-4 induced hypophagia while Pnoc^CeA neurons can promote palatable, energy-dense food consumption, we speculate that these two populations may function in a reciprocal manner within the CeA to bidirectionally regulate hedonic feeding. This potential relationship warrants investigation in future studies and to determine the mechanisms behind differences in diet effects.

Our findings capture the acute contribution of Glp1r^CeA neurons towards Ex-4 induced hypophagia; however, the chronic physiological role of these neurons in feeding and energy homeostasis has yet to be determined. Across other CNS systems, chronic silencing or ablation of GLP-1R-expressing neurons reveals homeostatic contributions that acute manipulation alone does not fully capture. For instance, in the arcuate nucleus (ARC), genetic ablation of GLP-1R expressing neurons fails to attenuate the hypophagic effect of systemically administered semaglutide; however, chronic silencing of thyrotropin-releasing hormone-expressing neurons that co-express GLP-1Rs within the ARC does partially attenuate the hypophagic and weight loss effects of acutely and chronically systemic administration of liraglutide [21,90]. Within the dorsomedial hypothalamus, viral knockdown of GLP-1R expressing neurons fails to produce hyperphagia but decreases energy expenditure and attenuates Ex-4 induced hypophagia [36]. Of the limbic sites, the lateral septum has abundant GLP-1R expression, where knockdown of GLP-1R-expressing neurons attenuates the hypophagic and weight loss effects of liraglutide and chronically silencing these neurons attenuates liraglutide's hypophagic effects. These data suggest that limbic GLP-1R populations may have modulatory roles more clearly revealed under chronic paradigms [15]. Given that clinical usage of GLP-1R agonists are administered chronically; whether chronic silencing or ablation of Glp1r^CeA neurons can produce similar outcomes warrants further investigation. Further studies that incorporate targeted silencing or ablation of Glp1r^CeA neurons will be important to elucidate their role in energy balance and contributions to chronic GLP-1R agonism.

Although it is clear that 1) the majority of peripherally administered GLP-1R agonist will accumulate in circumventricular organs of the brain and 2) above we discuss the evidence that suggests a pathway for indirect activation of the CeA by systemic GLP-1R agonists, our data with Glp1r^CeA neurons suggest that it is possible that Ex-4 or other peripherally-administered GLP-1R agonists may directly activate Glp1r^CeA neurons. The posterior CeA has close proximity to the lateral ventricles and previous whole-brain imaging approaches using fluorescently-conjugated liraglutide and semaglutide may not resolve low-level binding or have differences in brain transit. Consistent with this idea, Godschall et al. have demonstrated that peripheral administration of the small molecule GLP-1R agonists, danuglipron and orforglipron, can activate the CeA expressing humanized GLP-1Rs [80]. Whether GLP-1R expression levels in the CeA are sufficient for direct engagement by GLP-1R agonists such as Ex-4 and various other exogenous agonists has yet to be fully characterized. Collectively, we speculate that CeA activation by peripheral GLP-1R agonists is likely multifaceted, potentially incorporating both indirect and direct signaling. Future studies utilizing site specific methods and rigorous treatment regimens with doses of multiple GLP-1R compounds will be critical to help elucidate this model.

We aimed to rigorously perform and analyze the data present in this study; however, we acknowledge several limitations. First, performing fiber photometry recordings in vGat-Cre mice captures a broad range of GABAergic populations in the CeA, of which Glp1r-expressing neurons are a genetically distinct and restrictive subpopulation enriched within the medial CeA [54]. Bulk photometry across this broader population, therefore, likely dilutes responses concentrated within Glp1r+ and other Ex-4 responsive subpopulations. Additionally, variability in fiber placement across the CeA, where some could be located in central, medial, or lateral CeA, could also play a contributing factor in obtained signals. While we observed that Ex-4 resulted in sustained CeA neuronal activation up to a 30 min window post-injection, we are uncertain of the full extent this response is sustained for. Future studies implementing cell-type specific photometry recordings would be beneficial to resolve responsive CeA subpopulations and implement longer recordings determining the duration of Ex-4's effects. Additionally, while fiber optic implantation disrupts the blood–brain barrier at the CeA, we cannot entirely exclude the possibility that local tissue disruption may allow peripheral Ex-4 access to CeA neurons during photometry recordings. Importantly, this does not affect our chemogenetic findings. Second, while DCZ is generally regarded pharmacologically inert in animals lacking transduction of DREADDs in targets of interest, it is not without off-target effects [83]. We selected DCZ over clozapine-N-oxide (CNO) due to its reported minimal off-target effects. This treatment had no effect on any of the mCherry mice used for FR1 refeed and intermittent HFD access except for female vGat^CeA mice during assessment of active pokes, where we report a significant decrease. The basis for this finding is unclear but complicates the interpretation of the female vGat^CeA data specifically. Given that DCZ is relatively recent compared to CNO, future chemogenetic studies should consider and control for potential off-target effects of DCZ. Third, while viral transduction was predominantly contained within the CeA, occasional spread into adjacent regions including the BLA was observed. Cre-dependent expression nonetheless restricts viral spread to genetically defined neurons, even where anatomical containment was imperfect, preserving cell-type specificity. Finally, the present study did not directly assess the role of GLP-1R protein signaling. Future studies will investigate this directly. While our findings demonstrate that select populations are necessary for the full hypophagic response of peripherally administered Ex-4, whether GLP-1R is directly engaged in these neurons or whether recruitment occurs indirectly in upstream circuitry remains to be determined.

For the first time, we show that chemogenetic inhibition of CeA neurons attenuates the complete hypophagic effects of Ex-4 on food intake and food-seeking. Additionally, in a similar manner, population-specific inhibition revealed varying contributions to Ex-4 mediated hypophagia. Our data demonstrates that inhibition of all CeA neurons, Prkcd^CeA neurons, or Glp1r^CeA neurons produced significant attenuation of Ex-4 mediated hypophagia. Inhibition of Sst^CeA neurons, on the other hand, had no effect on Ex-4 mediated hypophagia. These data demonstrate multiple, non-overlapping neuron populations in the CeA participate in gating the actions of exogenous GLP-1R agonists.

4. Materials and methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-mCherry	Invitrogen, Thermo Fisher Scientific	Cat# PA5-34974; RRID:AB_2552323
Cy™5 AffiniPure® Donkey Anti-Rabbit IgG (H + L)	Jackson ImmunoResearch Labs	Cat# 711-175-152; RRID:AB_2340607
Chicken anti-GFP	Aves Labs	Cat# GFP-1010; RRID:AB_2307313
Alexa Fluor® 488 AffiniPure® Donkey Anti-Chicken IgY (IgG) (H + L)	Jackson ImmunoResearch Labs	Cat# PA5-34974; RRID:AB_2552323
Viral strains
pAAV-hSyn-DIO-mCherry (6.25∗1012 GC/ml)	Addgene	Cat# 50459-AAV9; RRID:Addgene_50459
pAAV-hSyn-DIO-hM4D(Gi)-mCherry (5.75∗1012 GC/ml)	Addgene	Cat# 44362-AAV9; RRID:Addgene_44362
pGP-AAV-syn-FLEX-jGCaMP7f-WPRE (2.8∗1013 GC/ml)	Addgene	Cat# 104492-AAV9; RRID:Addgene_104492
Chemicals, peptides, and recombinant proteins
Deschloroclozapine dihydrochloride	HelloBio	Cat# HB9126
Exendin-4	MedChemExpress	Cat# HY-13443
Exendin(9–39) amide	MedChemExpress	Cat# HY-P0264
Normal Donkey Serum	Jackson ImmunoResearch Labs	Cat# 017-000-121; RRID: AB_2337258
Bovine Serum Albumin	Sigma-Aldrich	Cat# A2153
VECTASHIELD® HardSet™ Antifade Mounting Medium with DAPI	Vector Laboratories	Cat# H-1500-10
Triton X-100	Fisher Scientific	Cat# BP151-500
Kynurenic acid	MedChemExpress	Cat# HY-100806
Experimental models: Organisms/strains
Mouse: B6J.129S6(FVB)-Slc32a1tm2(cre)Lowl/MwarJ	The Jackson Laboratory	RRID:IMSR_JAX:028862
Mouse: STOCK Tg(Prkcd-glc-1/CFP,-cre)EH124Gsat/Mmucd	MMRRC	RRID:MMRRC_011559-UCD
Mouse: STOCK Ssttm2.1(cre)Zjh/J	The Jackson Laboratory	RRID:IMSR_JAX:013044
Mouse: STOCK Glp1rtm1.1(cre)Lbrl/RcngJ	The Jackson Laboratory	RRID:IMSR_JAX:029283
Software and algorithms
Graphpad Prism v10	GraphPad	https://www.graphpad.com; RRID:SCR_002798
ImageJ/FIJI	NIH	https://fiji.sc; RRID:SCR_002285
Adobe Illustrator	Adobe	https://www.adobe.com/products/illustrator.html; RRID:SCR_010279
pCLAMP v10	Molecular Devices	http://www.moleculardevices.com/products/software/pclamp.html; RRID:SCR_011323
Keyence BZ-X810 Fluorescent Microscope	Keyence	https://www.keyence.com/ss/products/microscope/bz-x800_research/; RRID:SCR_023617
Microsoft Excel	Microsoft	www.microsoft.com/en-gb/; RRID:SCR_016137
FED3_Viz	Lex Kravitz Lab	https://github.com/earnestt1234/FED3_Viz
Bonsai	Bonsai Foundation CIC	https://bonsai-rx.org/; RRID:SCR_017218
Other
MLABRODENT TABLET 20 mg, Grain-Based Rodent Tablet	TestDiet, Lab Supply Orders	CAT# 1811143 5TUM
Fiber Optic Cannula	Newdoon	Cat# FOC-C-B-200-1.25-0.37-10
Rodent Diet With 60 kcal% Fat, D12492	Research Diets Inc	Cat# D12492
C&B Metabond® Quick Adhesive Cement System	Parkell	Cat# S380
Dental Cement	Stoelting Co.
Nanoject III	Drummond Scientific Company	Cat# 3-000-207
Nanoject Glass Capillaries	Drummond Scientific Company	Cat# 3-000-203-G/X
Screws	McMaster-Carr	Cat# CS-335
Fiber optic cannulae holder	This paper	https://github.com/OpenBehavior/Modular-Stereotaxic-Holders
Feeding Experimental Device 3	Handmade or OpenEphys	https://open-ephys.org/
Fiber-optic Patch Cords	Doric Lenses	Cat# BBP(2)_200/220/900/900-0.37_3 m_FCM-2xMF1.25(F)
FP3002	Neurophotometrics, MBF Bioscience	https://neurophotometrics.com/fp3002

4.1. Animals

We used 8–16 week old C57BL6/J, Slc32a1^tm2(cre)Lowl (vGat; MGI: 5751790), Tg(Prkcd-glc-1/CFP,-cre)EH124Gsat (PC; MGI: 3844446), Ssttm2.1-Cre (SC; MGI: 4838416) and Glp1r^{tm1.1(cre)Lbrl} (GL; MGI: 5776617) male and female mice throughout this study (Jackson Labs, Bar Harbor, ME) [[91], [92], [93], [94]]. Unless otherwise specified, mice were group housed with 3–5 mice per cage on standard bedding and provided with food and water ad libitum. A 12:12 light cycle was maintained with lights on at 7AM and off at 7PM. All procedures were performed with approval from the University of Alabama at Birmingham Institutional Care and Use Committee (IACUC).

4.2. Histology and fluorescence immunohistochemistry

A lethal dose of 2,2,2-tribromoethanol (250 mg/kg, Sigma-Aldrich) was administered via intraperitoneal injection prior to sacrifice. After an appropriate anesthetic depth was confirmed by loss of the toe pinch reflex, animals were transcardially perfused with 50 ml of 0.01 M PBS, then 50 ml of 4% paraformaldehyde (Electron Microscopy Sciences, Cat No. 19210) in 0.01 M PBS. Brains were extracted, post-fixed in 4% paraformaldehyde overnight then transferred to 30% sucrose (Thermo Fischer Scientific, Cat No. 036508. A1) for 36–48 h. Once sufficiently cryoprotected, brains were sectioned at 40 μm on a Leica CM1850 cryostat.

Free-floating coronal tissue sections were washed three times in 0.01 M PBS for 5 min each, then permeabilized for 30 min in 0.5% Triton X-100/PBS (Fisher Scientific, Cat No. BP151-500), followed by another three washes. Sections were then incubated in a blocking solution comprised of 0.1% Triton X-100/PBS, 1% Bovine Serum Albumin (Sigma, Cat No. A2153), and 10% Normal Donkey serum (Jackson ImmunoResearch, Cat No. 017-000-121) for 1 h. Following blocking, sections were then transferred into a separate blocking solution containing rabbit anti-mCherry polyclonal antibody (1:500; Invitrogen, Thermo Fisher Scientific Cat No. PA5-34974) and incubated overnight at 4 °C. The following day, sections were then washed three times in 0.01 M PBS for 5 min each. Sections were then incubated for 2 h in PBS containing donkey anti-rabbit conjugated with Cy™3 AffiniPure® (1:200; Jackson ImmunoResearch, Cat No. 711-165-152), then washed three times in 0.01 M PBS for 5 min each again. Once completed, sections were then mounted on SuperFrost slides (Fisher Scientific, Cat No. 12-550-123), allowed to dry, then coverslipped with Vectashield Hard-set mounting medium with DAPI (Vector Labs, Cat No. H-1500-10).

4.3. Stereotaxic surgery

Surgical procedures were performed on 6–8-week-old male and female mice. Animals received Buprenorphine SR (1.0 mg/kg, s. c.) and Meloxicam SR (5 mg/kg, s. c.) prior to surgery and again on post-op day 2. Mice were anesthetized with isoflurane inhalation (0.5−5%) and secured in a stereotaxic frame. Eye lubricant was applied immediately and as needed throughout surgery. The respiratory rate was visually monitored throughout surgery with a target of 55–60 breaths/minute. Internal temperature was maintained at 35 °C using a homoeothermic biodynamic feedback system (Harvard Apparatus; Holliston, MA). Anesthetic depth was confirmed by a loss of toe pinch reflex. The scalp was depilated and sterilized with rotating application of 70% ethanol and betadine. Triple antibiotic ointment and 4% topical lidocaine ointment was also applied. A midline incision was made and the scalp parted to expose the skull surface. After verifying appropriate orientation of the head, 1 mm burr holes were drilled for injection/fiber placement and screw stabilization.

For viral injections, we used a pulled glass capillary secured in Nanoject III (Drummond Scientific; Broomall, PA). Viruses for injection were either AAV-hSyn-DIO-mCherry (Addgene, Cat No. 50459-AAV9), AAV-hSyn-DIO-hM4d(Gi)-mCherry (Addgene, Cat No. 44362-AAV9), pGP-AAV-syn-FLEX-jGCaMP7f-WPRE (Addgene, Cat No. 104492-AAV9) and were tittered from stock concentration into sterile PBS at the following concentrations respectively, 6.25∗10¹² GC/ml, 5.75∗10¹² GC/ml, and 2.8∗10¹³ GC/ml. For CeA, 100–300 nl of each virus was injected at +/− 3.0 mm lateral, −1.3 mm posterior, −4.6 mm ventral of bregma.

For virus-only injected animals, we closed the scalp using veterinary adhesive (3M, Cat No. 1469SB). For mice used for photometry, optical fibers (Newdoon, Cat No. FOC-C-B-200-1.25-0.37-10) were secured in a custom-made ferrule holder (https://github.com/OpenBehavior/Modular-Stereotaxic-Holders) and implanted at a 5° angle +/− 3.05 mm lateral, - 1.3 mm posterior, −4.45 mm ventral of bregma. Fibers were secured with setscrews and Metabond (Parkell, Cat No. S380) applied over the skull surface, covering the base of the fibers and the screws. When the Metabond was dry, dental cement (Stoelting Co., Cat No. 10-000-786) was applied and allowed to harden before rehousing the animal. Virally injected animals were re-housed with their littermates and fiber implanted animals were singly housed. Post-operative care included daily weight monitoring for 5 days or until their weight returned to within 5% of their initial preoperative weights. All mice recovered for 3 weeks under normal husbandry conditions to allow for viral expression and recovery prior to any experiment.

4.4. Subject inclusion criteria

Validation of cells transduced by Cre-dependent viruses in the CeA and the termination sites of fiber optic implants above the CeA were visually confirmed first under an epifluorescent microscope and then a Keyence BZ-X800 Fluorescent Microscope and then noted within written records. For photometry experiments, CeA hemispheres that either did not have transduced cells in the CeA or did not have their fiber optic located near the CeA were excluded from statical analyses. For feeding-related experiments, animals that did not have expression of transduced cell present within the CeA and in both hemispheres were excluded from our statical analyses. For animals feeding from FED3s, they needed to sufficiently learn to feed from their FED3 under an FR1 paradigm; meaning that following a fasted condition and prior to any treatment regimes, they needed to successfully poke the active port, retrieve a pellet, and consume at least 0.2 g (10 precision pellets) within the 1st hour to ensure they are trained for refeeding experiments. Mice that failed to learn to feed from their FED3 or displayed pellet-hoarding behaviors anytime throughout FR1 feeding assays were excluded from our studies. Pellet hoarding is defined as pellets being successfully being retrieved from the FED3 but were dropped and not consumed during experimental treatments. For intermittent HFD access feeding studies, mice also needed to consume atleast 0.2 g of the HFD pellet within the 1st hour following saline injections prior to any treatment regimes. Mice that failed to consistently consume this amount were excluded from statical analyses.

4.5. Electrophysiology

4.5.1. Slice preparation

Electrophysiological experiments were performed as previously described [54]. Briefly, animals were removed from their cage and brought to the lab for brain slice preparation. The animal rested in a quiet chamber for 30 min prior to slice preparation to dissipate stress associated with animal transport from the vivarium. Mice were treated with a lethal dose of 2,2,2-tribromoethanol (250 mg/kg, i. p.), and, after a deep plane of anesthesia was reached, animals were transcardially perfused with cold, sodium free N-methyl-d-glucamine (NMDG) artificial cerebrospinal fluid (aCSF)[(in mM) 93 N-methyl-d-glucamine, 2.5 KCl, 1.2 NaH₂PO₄, 30 NaHCO₃, 20 HEPES, 25 Glucose, 5 l-ascorbic acid, 2 Thiourea, 3 sodium pyruvate, 10 MgSO₄ X 7H₂O, 0.5 CaCl₂ X 2H₂O]. All solutions were saturated with 95% O₂ and 5% CO₂. The brain was rapidly dissected and coronal 300 μM sections prepared in ice cold, oxygenated NMDG aCSF using a Leica VT1200S at 0.07 mm/s. Slices were immediately transferred to 36 °C NMDG aCSF for 10 min, and then normal 36 °C aCSF [(in mM): 124 NaCl, 4.4 KCl, 2 CaCl₂, 1.2 MgSO₄, 1 NaH₂PO₄, 10.0 glucose, and 26.0 NaHCO₃]. Slices rested in normal aCSF for at least 30 min prior to recordings.

4.5.2. Recordings

Whole cell or cell-attached patch clamp recordings were performed in the CeA guided by DIC microscopy and mCherry fluorescence. Slices were then transferred to a recording chamber (Warner Instruments), submerged in normal, oxygenated aCSF and maintained at 32 °C with a flow rate of 2 ml/min. For determining hM4d expression, we patched mCherry-expressing cells in the CeA. For hM4d recordings we used a potassium gluconate internal solution [(in mM): 135 C₆H₁₁KO₇, 5 NaCl, 2 MgCl₂, 20 HEPES, 0.6 EGTA, 4 Na₂ATP, 0.4 Na₂GTP at a final osmolarity of 290 mOsm at a pH of 7.3]. For voltage clamp recordings, neurons were voltage clamped at −70 mV using a Multiclamp 700 B and currents were digitized with an Axon 1550 B digitizer (Molecular Devices, Fremont, CA). For hM4d experiments, we included tetrodotoxin (500 nM), KA, picrotoxin (25 μM), and kynurenic acid (3 mM) in the bath to isolate the cell autonomous effects of the hM4d.

4.5.3. Data analysis

Data were analyzed in Clampfit 11.1 (Molecular Devices, San Jose, CA). Membrane capacitance and resistance were determined online using a −10mV square pulse after the cell stabilized (∼1 min). We did not correct for liquid junction potential.

4.6. Fiber photometry recordings

4.6.1. Freely behaving recordings

In preparation for fiber photometry recordings in combination with drug treatments, mice were habituated to the behavior room, handling, exposure to i. p. injections, and patch cable attachment to their implanted fiber optics. In addition, mice individually took turns acclimating to the behavior chamber where photometry recordings take place. On experimental days, mice were brought into the behavior room and left undisturbed for 1 h prior to recordings. A patch cord (200 μm diameter, 0.37 NA; Doric; Quebec, Canada) was then attached to the photometry system (FP3002, Neurophotometrics (MBF Biosciences); Williston, VT) and to mice's fiber implants using a ceramic mating sleeve before placing them into the behavior chamber. A 415-nm LED (50 μW) was used to obtain isosbestic signal and 470-nm (50 μW) used to excite GCaMP7f where each wavelengths values were transcribed by Bonsai (Bonsai Foundation CIC, https://bonsai-rx.org). Alternating 415 and 470 nM LED stimulation was administered through the patch cable at 40 Hz. Each trial was 2.5 h long where the 1st treatment was administered, i. p., at hour 1, the 2nd treatment administered at hour 2, followed by no more intervention and the recordings ending at hour 2.5. Treatment regimens were planned under a Latin-square design in order to prevent drug order effects. Each trial was separated at least 2 days to allow for drug wash-out.

4.6.2. Fiber photometry data analysis

Fiber photometry data were analyzed using custom python notebooks and functions. Briefly, raw photometry data were deinterleaved to isolate 415 nm and 470 nm signals. Raw data were then visually inspected to identify major aberrations or signal disruptions due to patch cable hairpins or loss of signal due to optic fiber cannulae separation. Raw 415 nm and 470 nm signals were then independently fit using a biexponential function to correct for photobleaching of signal intensity. Fitted 415 nm signals were then regressed from the fitted 470 nm signals using Huber regression to remove motion-related isosbestic signals from the Ca-dependent signal. Processed 470 nm signals were then used for downstream alignment and normalization. We aligned user-generated timestamps for each injection to the processed 470 nm signal using a nearest-neighbor search function. Aligned timestamps are depicted as time point 0. We then selected 30-minute ranges prior to and after time point 0 and calculated raw and Z-Scored ΔF/F. We examined the change of Z-Scored fluorescent values, looking at 30 min pre-/post-injection of the 2nd treatment administrations to examine what effect pretreatment of the 1st treatment had on these values. We additionally examined 30 min pre-/post-injection of the 1st treatment administrations to determine any effects to CeA neuronal activity prior to 2nd treatment. For Net AUC analysis, we examined the change of Z-Scored fluorescent values 15–30 min post 1st- and -2nd treatment administration for each individual animal, where the animal's response to injection has subsided, for each respective trials and performed mixed-effects model with Tukey's multiple comparisons with the mean for each treatment compared to each other. For Z-Scored peak response, we found the max Z-Scored value 0–30 min post 1st- and 2nd-treatment injection for each individual animal for each respective trials and performed mixed-effects model with Tukey's multiple comparisons with the mean for each treatment compared to each other. Net AUC was used to capture sustained, integrated change in neuronal activity in response to treatments' effect on tonic activity. In GraphPad Prism, the Net AUC incorporates peaks that deviate below baseline (y = 0) and the difference is computed by subtracting the area of the peaks below the baseline from the area of the peaks above baseline. All data was analyzed in GraphPad Prism. Python notebooks are available at https://github.com/Hardaway-Lab/FED3-Photometry_Workflows.

4.7. Feeding behavioral assays

4.7.1. FED3 habituation and fixed-ratio 1 training

Prior to feeding assays involving Feeding Experimentation Devices (FED3), each mouse was singly housed in OneCages (Plexx; Netherlands) with their own FED3, where standard chow pellets were removed and instead, they fed on 20 mg precision grain pellets (Lab Diet; Cat no. 1811143 5TUM) [82]. In order to receive pellets, mice learned to operate the FED3 under a Fixed-Ratio 1 (FR1) paradigm, where they must nose-poke an active port on the FED3 to dispense an individual pellet. We selected an FR1 paradigm over free feeding because FR1 provides discrete, time-stamped nose poke and pellet retrieval events that enable precise quantification of both food-seeking (active pokes) and consumption (pellet retrieval) within defined event windows, while closely approximating free feeding given its minimal operant demand [82]. This allows the precise quantification of active pokes and pellets retrieved while detecting treatment-induced changes during FR1 feeding assays. FED3s emit an auditory tone and deliver an LED light stimulus whenever the active port is selected to reinforce training. Mice were housed with their own FED3 and trained at least 1-week prior to any feeding assay. Mice were then tested following a fast where they needed to consume at least 0.2 g (10 precision pellets) within the 1st hour to ensure they are trained for refeeding experiments. Mice that failed to learn to feed from their FED3 or displayed food-hoarding behaviors were excluded from our studies. Unless otherwise specified, FED3s remained with the mice and fed ad libitum under FR1 paradigm for the remainder of our studies.

4.7.2. FR1 refeed

For food-deprived refeeding combined with drug treatments, mice were food-deprived for 24 h prior to experimental days. On experimental days, 1 hour before the 24-hour fasted time point, mice were weighed and assigned their randomized treatment regimen. The first injection was given 40 min prior to FED3 access, followed by the second injection 20 min later. Finally, at the 24-hour fasted time point, access to their FED was allowed where pellet retrieval and active port selection was recorded by their FED3. After the 1st hour, pellets retrieved and consumed were visually confirmed by the experimenter. Treatment regimens were planned under a Latin-square design in order to prevent drug order effects, with the exception of saline/saline administration which was performed first and served as a baseline for each mouse before drug exposure. Injections were given 40 min before food onset to allow for DCZ onset prior to Ex4 administration (−20 min). We selected −20 min administration of Ex-4 to allow for drug onset prior to food access. Each session was separated at least 2 days to allow for drug wash-out.

4.7.3. FED3 data analysis

Data obtained from FR1 refeed studies were pulled from SD cards from each animal's assigned FED3. The csv files obtained were then processed through FEDViz to bin their active pokes and pellets retrieved as a 1-hour timepoint [82]. These values were then recorded into spreadsheets for each treatment regime they received. Normalized values for active pokes and pellets retrieved were calculated individually for each animal by subtracting the values recorded at the 1-hour timepoint for each treatment regime it received, respectively, from the value recorded at their Veh/Veh treatment and then dividing it by their Veh/Veh value. These data were then converted into percentage and recorded into a spreadsheet, along with their raw values. The FR1 refeed assays are a repeated measures design and baseline food intake varies between individual mice due to differences in body weight, sex, and session-to-session treatments. Within-subject normalization to each animal's Veh/Veh sessions controls for inter-animal variance and isolates the relative magnitude of treatment-induced changes (expressed as percent change from baseline). This improves the ability to detect consistent within-animal treatment effects that could be obscured in raw data due to between-animal variability. Raw values were left as is. Both normalized and raw values were then inputted into GraphPad Prism and analyzed using two-way repeated measures ANOVA with Tukey's multiple comparisons with the mean for each treatment compared to each other within their respective group for both pellets retrieved and active port pokes. When performing this analysis for normalized data sets, Veh/Veh values were set as zero and not included in the graphical plot. We present normalized data in our main figure to demonstrate the degree of treatment-induced changes relative to baseline (Veh/Veh). Raw values are presented in the supplemental figures and demonstrate absolute intake.

4.7.4. Intermittent HFD assay

To assess the consumption of palatable, calorie-dense food in the presence of neutral food, we generated a separate cohort of mice. Mice were singly housed 3 weeks after surgeries and kept on standard chow diet, ad libitum, throughout the study. Prior to the week of assays, crumbs of 60 kcal% fat rodent diet (HFD; Research Diets, Cat No. D12492) were sprinkled into their cages once every 4 days to prevent neophagia on test days. On experimental days, mice were weighed and randomly assigned their treatment regimen. Similar to the injection timeline and rationale for FR1 refeed assays, mice were then injected, i. p., with their randomly assigned treatments before pre-weighed HFD and standard chow pellets were given 40 min later. We measured HFD and standard chow pellets weights an hour after HFD presentation and manually recorded the values for each mouse. Each session was separated at least 3 days to allow for drug wash-out and to avoid entrainment of HFD consumption.

4.7.5. Intermittent HFD data analysis

Data of HFD and standard chow consumed for each individual mouse during intermittent HFD assays were recorded into spreadsheets to calculate the amount gram amount consumed, their normalized values, the percentage of standard chow consumed, and total caloric intake. Normalized values for HFD consumption were calculated individually for each animal by subtracting the values recorded at the 1-hour timepoint for each treatment regime it received, respectively, from the value recorded at their Veh/Veh treatment and then dividing it by their Veh/Veh value. Similarly, percentage of standard chow consumed was calculated by dividing the gram amount of standard chow consumed by the same gram amount of standard chow consumed plus the gram amount of HFD consumed for mouse and respective session. These data were recorded into a spreadsheet along with their raw values. Total caloric intake (kCal) was calculated for each animal in each treatment session as the sum of HFD and chow consumption, in grams, by their respective caloric density derived from manufacturer specification: 5.24 kcal/g for HFD (Research Diets, Cat No. D12492) and 3.0 kcal/g for standard chow (NIH-31 open formula). Raw values were left as is. Values were then inputted into GraphPad Prism and analyzed using two-way repeated measures ANOVA with Tukey's multiple comparisons with the mean for each treatment compared to each other within their respective group as the multiple comparisons. When performing this analysis for normalized data sets, Veh/Veh values were set as zero and not included in the graphical plots.

4.8. Imaging

Imaging was performed on a Keyence BZ-X810 under 20X magnification. For tissue sections obtained from C57BL6/J mice, tiled z-stacks of the CeA and surrounding region were captured using optical sectioning. For tissue sections obtained from any Cre mice, tiled z-stack images of the CeA and surrounding regions were captured using widefield fluorescence imaging. Raw images were then stitched and a maximum intensity projection made in Keyence Analyzer. Stitched raw images were obtained for individual channels including DAPI and 550 nm. Final images were assembled in Adobe Illustrator 25.2.3.

CRediT authorship contribution statement

Miguel Duran: Writing – review & editing, Writing – original draft, Validation, Investigation. Ningxiang Zeng: Investigation. Elam J. Cutts: Investigation. Anusha Polamarasetty: Investigation. Melissa Rodriguez: Investigation. Kirk K. Habegger: Supervision, Conceptualization. J. Andrew Hardaway: Writing – review & editing, Writing – original draft, Visualization, Project administration, Methodology, Conceptualization.

Disclosure statement

The authors have nothing to disclose.

Declaration of competing interest

The authors declare no competing interests.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Data will be made available on request.