
==== Front
Mol Metab
Mol Metab
Molecular Metabolism
2212-8778
Elsevier

S2212-8778(24)00152-2
10.1016/j.molmet.2024.102021
102021
Original Article
Dietary fat content and absorption shape standard diet devaluation through hunger circuits
Sutton Hickey Ames K. 11
Becker Jordan 231
Karolczak Eva O. 21
Lutas Andrew 2
Krashes Michael J. michael.krashes@nih.gov
2⁎
1 Department of Psychology and Neuroscience, Temple University, Philadelphia, PA, USA
2 Diabetes, Endocrinology, and Obesity Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD, USA
3 NIH-Brown University Graduate Program in Neuroscience, Bethesda, MD, USA
⁎ Corresponding author. michael.krashes@nih.gov
1 Ames K. Sutton Hickey,Jordan Becker,Eva O. Karolczak contributed equally to this work.

30 8 2024
11 2024
30 8 2024
89 10202126 7 2024
27 8 2024
27 8 2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Objective

Exposure to 60% high fat diet (HFD) leads to a robust consummatory preference over well-balanced chow standard diet (SD) when mice are presented with a choice. This passive HFD-induced SD devaluation following HFD challenge and withdrawal is highlighted by the significant reduction in SD food intake even in states of caloric deprivation. The elements of HFD that lead to this SD depreciation remains unclear. Possibly important factors include the amount and type of fat contained in a diet as well as past eating experiences dependent on sensory properties including taste and post ingestive feedback. We aimed to explore the role of these components to HFD-induced SD devaluation.

Methods

Wildtype mice were longitudinally presented discrete HFDs in conjunction with SD and feeding and metabolic parameters were analyzed. A separate cohort of animals were assessed for acute HFD preference in 3 conditions: 1) ad libitum fed (sated), 2) overnight fasted (physiologically hungry), and 3) ad libitum fed (artificially hungry), elicited through chemogenetic Agouti-related peptide (AgRP) neuron activation. Population dynamics of AgRP neurons were recorded to distinct inaccessible and accessible diets both before and after consummatory experience. Transient receptor potential channel type M5 (TRPM5) knockout mice were used to investigate the role of fat taste perception and preference to HFD-induced SD devaluation. The clinically approved lipase inhibitor orlistat was used to test the contribution of fat absorption to HFD-induced SD devaluation.

Results

HFD-induced SD devaluation is dependent on fat content, composition, and preference. This effect scaled both in strength and latency with higher percentages of animal fat. 60% HFD was preferred and almost exclusively consumed in preference to other diets across hours and days, but this was not as evident upon initial introduction over seconds and minutes, suggesting ingestive experience is critical. Optical fiber photometry recordings of AgRP activity supported this notion as neuronal suppression by the different diets was contingent on prior intake. While taste transduced via TRPM5 influenced HFD-evoked weight gain, it failed to impact either HFD preference or HFD-induced SD devaluation. Perturbation of post ingestive feedback through orlistat-mediated diminishment of fat absorption prevented HFD-evoked weight gain and abolished HFD-induced SD devaluation.

Conclusions

Post ingestive feedback via fat digestion is vital for expression of HFD-induced SD devaluation.

Highlights

• Hungry mice rapidly transition to exclusively consuming 60% HFD when given a choice.

• AgRP neuronal inhibition is dependent on prior exposure to dietary fat.

• Taste receptor function is dispensable for SD devaluation after HFD exposure.

• HFD-induced SD devaluation requires proper fat absorption.

• AgRP neural readout of HFD-induced SD devaluation necessitates appropriate fatdigestion.

Keywords

Behavioral neuroscience
Feeding
Diet
AgRP neurons
==== Body
pmc1 Introduction

Evolutionary forces have wired our brains to prefer and consume energy-dense foods to aid in our survival. While effective during periods of limited access, the ubiquitous nature of high-fat food sources in society leads to obesity and numerous related health complications. Exacerbating this drive to consume more energy-dense, palatable foods is a devaluation of less appetitive, nutritionally-balanced foods [1,2]. While this preference for calorically-rich foods is well known, significant gaps exist in our understanding of how this develops and leads to devaluation.

Laboratory mice are typically provided with ad libitum access to a well-balanced standard chow diet (SD) in which the macronutrient composition has been formulated for optimal growth. Introduction to ad libitum high fat diet (HFD), but not a high-sucrose diet, leads to rapid weight gain, at least in part due to excessive caloric intake [[3], [4], [5]]. Interestingly, when mice are given a choice between ad libitum access to both SD and HFD, they strongly prefer consumption of the latter at the expense of the former [1,2]. While this predilection for HFD over SD during prolonged exposure is well described [6], how rapidly this transition occurs under physiological or artificial hunger is less known. Removal of HFD from mice given the choice between HFD and SD, akin to a strict human diet, results in rapid weight loss due to the self-restricted consumption of SD [1,2,7,8]. Additionally, mice fed a HFD will forgo SD consumption even in states of physiological or artificially-induced caloric deprivation [1,2]. While this SD devaluation is robustly conserved between sex and subject and independent of fat mass accrual [1,2], the causative nature of this phenomenon is not well understood.

A number of factors may contribute to HFD-induced SD devaluation including the percentage and source of fat (animal versus plant), the food's sensory properties (smell, taste, and texture), and its nutritional value determined as a consequence of ingestion. Whether these elements work collectively or individually to drive HFD-induced devaluation has not been fully explored. Interestingly, mice deemed anosmic following surgical removal of the olfactory bulb remain sensitive to diet-induced obesity [9], while mice with transient receptor potential channel type M5 (TRPM5) gene ablation (which disrupts fat taste perception and preference) [[10], [11], [12], [13]] demonstrate partial protection against diet-induced obesity [14]. However, it is unknown whether this attenuated weight gain in these knockouts affects SD devaluation. Critically, mice exposed to only the smell of HFD (in lieu of ingestion) fail to display HFD-induced SD devaluation, while anosmic mice permitted to consume HFD exhibit SD devaluation, suggesting that the role of olfactory information is minimal to this effect [9]. Further insight into this phenomenon is paramount given that a primary cause of the negative health consequences associated with the obesity epidemic is the robust preference for foods enriched in fat over healthier options [15,16]. Here, we aimed to deconstruct the source of HFD-induced SD devaluation through a minimalistic approach, spotlighting the importance of learning through post-ingestive feedback.

Hypothalamic Agouti-related peptide (AgRP)-expressing cells are sensitive to hunger state and encode the seeking phase of food consumption [[17], [18], [19], [20], [21]]. These neurons exhibit high levels of activity under conditions of caloric deprivation, which is considered relevant to promoting food intake. Furthermore, AgRP cells promptly respond to food presentation, sensory cues of food anticipation, and direct macronutrient infusion into the stomach or intestine, tracking the quantitative but not qualitative nature of calories [[17], [18], [19], [20], [21], [22], [23], [24]]. Notably, this rapid and durable suppression of AgRP neurons is exacerbated in hungry animals when basal activity is elevated, and the response properties are stronger toward palatable, energy-dense food substrates [20]. AgRP neurons display a learned component, as activity changes can be conditioned to visual, gustatory, or auditory cues predicting food availability [18,21,24,25]. However, it is unclear if this enhanced inhibition of AgRP cells in response to HFD precedes post-ingestive experience or is a consequence of familiarity.

Infusion of fat directly into the duodenum both inhibits food intake and AgRP population activity, an effect dependent on fat digestion and intact vagal signaling [26]. Similarly, the gut-secreted satiation signal cholecystokinin (CCK) reduces food intake and suppresses AgRP activity [23,24]. Treatment of mice with a selective CCK-A-receptor antagonist attenuates the lipid-mediated inhibition of AgRP neurons, suggesting this anorectic hormone is required for the drop in AgRP activity by fat, which is consistent with prior observations that fat is the most potent stimulus for CCK secretion in vivo [23,27,28]. Importantly, prolonged exposure to HFD drastically attenuates the AgRP neuron response to SD, independent of weight change [1,2]. This blunting of AgRP sensitivity to nutritionally-balanced SD after HFD experience provides a potential mechanism for the hardships of dieting; namely, the negative valence starvation signal encoded by AgRP neurons [21] cannot be fully silenced by SD. In fact, AgRP activity can only be further inhibited by HFD presentation [1], offering an explanation for why dieting is highly prone to relapse.

Given that AgRP neurons strongly contribute to HFD-induced SD devaluation, vagal afferent signaling is required for fat preference and nutrient-induced suppression of AgRP activity, and blocking fat digestion disrupts AgRP activity to intragastric fat infusions, we hypothesized that perturbing fat absorption in the gut could potentially prevent HFD-induced SD devaluation. To this end, we found that the percentage of animal fat content in HFD scales with the strength and speed of SD devaluation in calorically-restricted mice. Furthermore, we show that 60% HFD, which led to the most robust and rapid SD devaluation, was preferred to all other diets tested over the course of days, weeks, and months but not within the first hour of exposure. This could be partially explained by our finding that presentation of inaccessible novel HFD, which can be visually and olfactorily detected but not ingested, did not inhibit AgRP neurons, suggesting consummatory experience (ie. taste and/or post ingestive feedback) is required for this effect. After ruling out gustatory processing using taste-blind mice, we determined HFD-induced SD devaluation was dependent on fat absorption, highlighting the significance of post-ingestive signaling in the emergence of maladaptive behaviors following HFD exposure.

2 Materials and methods

2.1 Experimental animals

All animal protocols and procedures were approved by the US National Institute of Environmental Health Sciences Animal Care and Use Committee or the National Institutes of Health Animal Care and Use Committee. Mice were housed with a 12-h light/dark cycle and provided ad libitum access to food (standard chow diet (SD), Envigo 7017 NIH-31) and water unless otherwise noted. All experiments were carried out in adult (>8 weeks) mice that were group-housed until experiments began. Some measurements were carried out in the same mouse across conditions (see individual methods sections for further details).

C57BL/6 J, Agrp-IRES-Cre (“Agrp-iCre”, Jackson Laboratories, JAX stock no. 012899), B6N;129-Tg(CAG-CHRM3∗,-mCitrine)1Ute/J (“LSL-hM3Dq”, JAX stock no. 026220), and B6; 129-Trpm5tm1Csz/J (“TRPM5 KO”, Jackson Laboratories, stock no. 013068) mice were used. Agrp-iCre mice have an internal ribosome entry site (IRES)-Cre inserted after the stop codon of the Agrp gene on chromosome 8 [29].

2.2 Longitudinal high-fat diet exposure

Mice were weighed, single housed, and provided access to pre-weighed SD. Baseline body weight and food intake were measured for two weeks (weeks −2 and −1), afterwhich mice were body weight matched and assigned to separate cohorts, in which SD was provided alongside one of the following diets (weeks 0–8; experimental: 1) 60% lard-based HFD (60% HFD, Research Diets D12492), 2) 45% lard-based HFD (45% HFD, Research Diets D12451), 3) 30% lard-based HFD (30% HFD, Research Diets D20072301), or 4) 50% vegetable-based HFD (50% veg, Bio-Serv F07687 custom made into 5 g pellets) (Table 1; Supp. Fig. 1). A control cohort was only given access to SD throughout the experiment. Body weights and weekly food intake were measured for 8 weeks post-HFD exposure. A fast refeed with SD only was performed at week −1 (baseline), and weeks 1, 4, and 8 post-HFD exposure. After the SD refeed, the same two diets (or just SD for the SD cohort) were pre-weighed and returned to the mouse's cage and weekly body weight and food intake measurements were resumed. The percent change in SD devaluation across the experiment was calculated based on an individual's baseline SD intake during the fast refeed paradigm during week −1. After 8 experimental weeks of the paradigm, mice were maintained on the same diets and metabolic profiling (glucose and insulin tolerance, circulating hormone measurements, CLAMS) was performed in a cohort of the animals. At the end of the experiments, mice were rapidly euthanized and dissected to measure lean and fat pad weights.Table 1 Ingredient breakdown of mouse diets used in these studies. Animals were administered diets manufactured by Envigo (SD; NIH-31), Research Diets (30–60% HFD; D20072301, D12451, D12492, respectively), and Bio-Serv (50% veg; F07687 made into 5 g pellets). While animal-based HFD was generated using lard, the vegetable based diet leverages hydrogenated cottonseed oil as a fat source, while also including a high sucrose content akin to the 45% animal HFD.

Table 1	SD	60% animal HFD	45% animal HFD	30% animal HFD	50% vegetable HFD	
Manufacturer	Envigo	Research Diets	Research Diets	Research Diets	Bio-Serv	
Catalog #	NIH-31	D12492	D12451	D20072301	F07687	
Fat % kcal	14	60	45	30	48.9	
Carb %kcal	62	20	35	50	34.6	
Protein % kcal	24	20	20	20	16.5	
kcal/gm	3.0	5.2	4.7	4.3	4.97	
	
Ingredients (grams per kg)	
Sucrose	0	94.08	206.06	72.93	175	
Hydrogenated Cottonseed Oil	0	0	0	0	250	
Lard	0	316.60	206.88	116.81	0	

2.3 Diet preference score calculations

In animals given access to two diets (SD + one of the HFDs), diet preference scores were determined using the following equation: diet preference score = (HFD intake - SD intake)/(HFD intake + SD intake). In animals given access to more than two diets, diet preference scores were determined using the following equation: ((this diet intake (e.g. SD for SD) - intake of all other diets (e.g. 60% + 45% + 30% for SD))/total intake of all diets.

2.4 SD devaluation using a fast refeed paradigm

Mice were fasted (all food removed from the cage) in the middle of the light cycle. Following ∼18 h of fasting, pre-weighed SD was provided at the beginning of the light cycle when food intake is typically low. SD intake was measured following 1 h of access.

2.5 Glucose and insulin tolerance

Glucose homeostasis was measured using fed blood glucose measurements, as well as glucose and insulin tolerance tests. All blood glucose measurements were taken using tail vein samples with a glucometer. For fed blood glucose measurements and insulin tolerance tests, mice were tested in the randomly fed state. For the glucose tolerance tests, mice were fasted overnight, and then injected with glucose (1 mg/g, i. p.) and blood glucose monitored at 0, 15, 30, 60, and 120 min post-injection. For the insulin tolerance tests, mice were injected with insulin (Humulin R, 0.75 mU/g, i. p.), and blood glucose monitored at 0, 15, 30, and 60 min post-injection.

2.6 Circulating hormone measurements

Circulating hormone levels were measured via collection of blood samples via tail bleed in the randomly fed state. ELISAs were then performed on serum to measure leptin (R&D systems, MOB00) and insulin (Crystal Chem, 90,010) across mice.

2.7 Metabolic profiling using CLAMS

To measure energy expenditure, mice were individually placed in Oxymax/CLAMS (Columbus Instruments) units with continued access to their experimental diets on the cage floor of the units. Mice were acclimated to the chambers for 2 days before the start of the experiment, and experiments were performed at 22 C. Samples were collected every 13 min for 3 days, measured from 12 chambers. Twenty-four hour average parameters were analyzed from data collected during the experimental days.

2.8 Diet preference during homecage presentation of all diets

Mice were single housed one week prior to experiments, after which daily baseline homecage SD food intake was recorded for two days (days −2 and −1). Beginning on day 0, all mice were exposed to all of the diets (SD, 30% HFD, 45% HFD, 60% HFD) simultaneously, which were mixed up in the same wire cage top that they had been used to consuming SD food from during baseline measurements. Following one week of experimental diet exposure, mice were returned to a diet of SD only for one day (day 8).

2.9 Drugs

Clozapine N-oxide (CNO, Tocris, 4936) was dissolved in saline at stock concentrations and stored at −20 °C until use. For in vivo experiments, on the day of testing, the CNO stock solution was thawed, diluted with saline, and delivered intraperitoneally at a volume of 10 ml/kg at a 1.0 mg/kg dose. For in vivo experiments involving orlistat (Cayman Chemical Company, 10005426), on the day of testing, orlistat was mixed in the HFD at a 50.0 mg/kg dose.

2.10 Diet preference during chemogenetic AgRP stimulation

Agrp-iCre control mice or Agrp-iCre, LSL-hM3Dq (AgRP::3Dq) mice were single housed at least one week prior to experiments and assigned to one of three experimental groups: 1) control sated, 2) control fasted, 3) AgRP::3Dq sated. To measure how AgRP stimulation or physiological fasting impacts food preference, mice in the fasted group were fasted overnight (21 h), whereas sated animals in the control or AgRP::3Dq groups continued to have ad libitum access to SD prior to the experiment. Mice were then injected with CNO and placed back in their homecage without access to any food for 30 min to allow appropriate CNO-mediated 3Dq activation in the AgRP::3Dq group prior to novel diet exposure. All diets (SD, 30% HFD, 45% HFD, 60% HFD) were then placed in the wire cage top and 1 h food intake measured for each of the diets across groups.

2.11 Viral vectors and stereotaxic injections

AAV1-hSyn-Flex-GCaMP6s (Addgene, 100,845) was used for all photometry experiments in Agrp-iCre mice. During surgeries, mice were anesthetized with isoflurane, placed in a stereotaxic frame (Stoelting's Just for Mouse), and provided with analgesia (meloxicam, 0.5 mg/kg). Following a small incision on top of the skull and skull leveling, a small hole was drilled for injection. A pulled-glass pipette (20–40 mm tip diameter) was inserted into the brain, and 250–400 nL of virus was injected into the ARC (AP: 1.5, ML: + − 0.23, DV: 5.7) using a micromanipulator (Grass Technologies, Model S48 Stimulator, 25 nL/min).

2.12 Fiber implantation

For in vivo fiber photometry experiments targeting AgRP neurons, an optic-fiber cannula (core = 400 um; 0.48 NA; M3 thread 148 titanium receptacle; Doric Lenses) was placed directly over AgRP neuron somas (AP: 1.5, ML: + − 0.25, DV: 5.6). Fibers were fixed to the skull using C&B-Metabond Quick 150 Adhesive Cement and dental acrylic. After recovery, mice were single-housed and allowed to recover for >2 weeks before further experimentation.

2.13 In vivo fiber photometry set-up

To excite GCaMP6s, <20 uW blue LED light at 470 nm was driven by a multichannel hub (Thorlabs), modulated at 211 Hz, and subsequently delivered to a dichroic mini cube (FMC5, Doric Lenses) that was connected with optic fibers and a rotary joint (FRJ 1x1, Doric Lenses) to the optic cranial implant of the mouse. GCaMP6s calcium GFP signals were collected through the same fibers and dichroic minicube into a Femtowatt Silicon Photoreceiver (2151, Newport). Digital signals were subsequently demodulated, amplified, and collected through a lock-in amplifier (RZ5P, Tucker–Davis Technologies (TDT). LED modulation and data collection were performed using Synapse (TDT), exported via OpenBrowser (TDT), and analyzed in RStudio or GraphPad Prism.

2.14 AgRP fiber photometry experiments with high-fat diets

All mice were habituated to handling and patch cord hook up for at least 3 days before experimentation. Habituation was also performed to the tea ball used for inaccessible diet exposures by placing an empty tea ball in the homecage overnight prior to the day of experimentation across all mice. Experiments (Figure 3) were performed in female mice only. Following an 18 h overnight fast, mice were hooked up to the fiber patch cord during the early to middle stages of the light cycle and allowed to habituate to the set-up in their homecage for 3–5 min. For inaccessible food photometry assays, following 3 min of baseline recordings, a diet (SD, 45% HFD, or 60% HFD) was placed in a tea ball and placed in a corner of the homecage, and recordings continued for the following 3 min. Mice were then unhooked from the patch cord, the tea ball removed from the cage and food removed, and the tea ball cleaned with water and dried. Mice were then given a break of at least 15 min before a separate inaccessible diet was tested in the same manner to allow AgRP population activity to return to baseline levels. In separate experiments with the same mice, photometry experiments were repeated in response to the presentation of one of the diets for the first time in a corner of their homecage and the subsequent consumption of each individual diet (detected by the auditory sound of food consumption and visual appearance of intake), separated by at least 1 week in between diet exposures to allow AgRP baseline activity to return to normal. At least 1 week following the completion of these experiments, pre-exposure to one of the diets (counterbalanced across each diet) was then achieved by placing it in the mouse's homecage overnight (alongside their normal SD) for ∼16 h. High fat diet was then removed and the animals remained on ad libitum SD exposure for three days, followed by an overnight fast refeed prior to photometry experiments the following day (as before). AgRP population activity was then measured using the same paradigms as before, with the same diet that was in the homecage 3 days prior being placed in a tea ball to measure AgRP responses to inaccessible food after pre-exposure, followed by a 15-minute break in recording, and then recordings during accessible and consumption of the same diet. Experiments were then repeated with a separate HFD following a 1 week break. Exposure to individual diets in all of these experiments were counterbalanced across mice. Synchronized high-definition videos were recorded for time-locked data analysis in Synapse, which was downsampled to 8 Hz.Figure 1 HFD-induced weight gain and SD devaluation are dependent on dietary fat content.

(a) Following a baseline period of SD access in their homecage (Base), separate cohorts of mice were given access to a choice of SD and HFD (HFD groups only), and weekly body weight and food intake was measured and compared across cohorts. Weekly body weight (b), change in body weight (c), (calculated based on the average body weight during baseline weeks 0–2), (d) standard diet (SD) food intake, (e) high fat diet (HFD) food intake, and (f) total food intake (calculated by adding SD + HFD intakes) are altered by HFD exposure in a stepwise fashion (repeated measures two-way ANOVA followed by Bonferroni multiple comparisons; b,c: F (30,570) = 9.4, d: F (27,360) = 35.5, e: F (14,231) = 3.32, f: F (27,360) = 8.16). Food intake of SD and HFD across the experimental days was averaged and plotted for each animal as a stacked circular bar graph (g), with SD in darker bars, stacked on top with HFD in more opaque colors (HFD groups only). Preference for HFD over SD (indicated by a diet preference score>0) is stepwise increased dependent on HFD percentage both across the paradigm (h; repeated measures two-way ANOVA followed by Bonferroni's multiple comparisons; F (14,231) = 2.42) and when averaged across experimental weeks (i), with exposure to 60% HFD eliciting the highest preference, followed by 45% HFD, and then 30% HFD (i; one-way ANOVA followed by Bonferroni's multiple comparisons (F (2,91) = 29.26). While (j) SD devaluation (measured by SD intake after a fast refeed) is not observed in SD cohorts, exposure to 30% HFD in the homecage elicits SD devaluation by week 8 (RM two-way ANOVA followed by Bonferroni's post-hoc; F (3,30) = 5.02). In contrast, 45% and 60% HFD groups show SD devaluation beginning on week 1 of HFD exposure and continuing throughout week 4 and week 8 (RM two-way ANOVA followed by Bonferroni's multiple comparisons; 45% HFD: F (3,48) = 27.18, 60% HFD: F (3,18) = 24.43). Similar to homecage diet preference, SD devaluation is stepwise increased based on the content of homecage HFD exposure (k), with a more negative percent change of SD intake (calculated based on SD intake on week −1 before HFD exposure) in 30% HFD groups than SD, followed by 45% HFD, with the 60% HFD group demonstrating the most robust SD devaluation (one-way ANOVA followed by Bonferroni multiple comparisons; week 1: F (3,57) = 18.93, week 4: F (3,57) = 28.53, week 8: F (3,40) = 6.11). (l) The change in body weight observed by week 1 of diet exposure is significantly correlated to the SD devaluation during a fast refeed in the 45% HFD cohort (Pearson correlation coefficient), but not in the 30% HFD or SD cohorts. Significance is determined by p < 0.05, and denoted by: ∗(SD vs. 60%/45%/30%HFD), #(60%/45% HFD vs. 30% HFD), + (45% HFD vs. 60% HFD) (b-f, h), or ∗(groups indicated by line on graph) (i–m). Data represented as mean ± SEM (b-h, j), a box and whiskers plot with lines at the median, the first and third quartiles and whiskers at the highest and smallest values of the interquartile range (IQR) ∗1.5 (i, k), or as a correlation plot with lines indicating a linear model lowess smooth line with 95% confidence intervals (l). Sample sizes (a–i): SD: n = 16, 30% HFD: n = 11–12, 45% HFD: n = 18, 60% HFD: n = 16–17; (j–k): SD: n = 8–16, 30% HFD: n = 12, 45% HFD: n = 18, 60% HFD: n = 7–15, (l): SD: n = 16, 30% HFD: n = 11, 45% HFD: n = 18, 60% HFD n = 15.

Figure 1

Figure 2 Preference for dietary fat differs across access and physiologic versus artificial AgRP-mediated hunger states.

(a) Following a baseline period of access to SD only, animals were given access to four diet types (SD, 60% HFD, 45% HFD, 30% HFD) simultaneously in their homecage followed by a post-HFD withdrawal period. All animals opted for the 60% HFD diet (b), demonstrated across all days of HFD access (linear mixed model followed by repeated measures Bonferroni post-hoc; ∗ = 60% HFD vs. all other diets; # = SD vs. 30% and 45%). A heat map of food intake (c) across the experiment illustrates that all mice preferentially chose to eat 60% HFD, with minimal intake of other diets. (d) Diet preference scores indicate that preference for 60% HFD was robust across all days of multiple HFD homecage access (linear mixed model followed by repeated measures Bonferroni's multiple comparisons; ∗ for 60% HFD vs. each other diet type, # for SD vs. 30% HFD, $ for 45% HFD vs. 30% HFD). (e) Following 1 week of multiple HFD exposure, homecage SD intake was decreased (e; paired t-test: t (15) = 8.91, P < 0.0001). To test how physiologic versus AgRP-mediated hunger coordinates diet preference, groups of mice were distributed among three cohorts (f): (1) Agrp-iCre (−/−) mice were sated or (2) fasted, or (3) Agrp-iCre, LSL-hM3Dq mice were sated (AgRP::3Dq). 30 min following CNO injection (to activate AgRP neurons in the AgRP::3Dq group), diet preference assays were performed in which animals were given simultaneous access to SD, 30% HFD, 45% HFD, and 60% HFD in their homecage. One hour food intake measurements (g) demonstrate that both fasting and AgRP neuron activation promotes overall food intake (one-way ANOVA followed by Bonferroni's multiple comparisons; F (2,37) = 42.85). While both sated (h, left) and fasted (h, middle) mice demonstrate increased intake of 60% and 45% HFD, AgRP::3Dq (h, right) mice increase intake of SD, 45% HFD and 60% HFD similarly over 30% HFD (two-way RM ANOVA followed by Bonferonni's multiple comparisons; F (2.26, 88.19) = 20.27). (j) Diet preference scores further demonstrate that AgRP stimulation in AgRP::3Dq mice is distinct from physiologic hunger, with higher preference for SD over HFD in AgRP::3Dq mice than sated or fasted groups (i; one-way ANOVA) (one-way ANOVA followed by Bonferroni multiple comparisons; SD: F (2,37) = 9.63, 60% HFD: F (2,37) = 1.44, 45% HFD: F (2,37) = 0.60, 30% HFD: F (2,37) = 0.72), and no difference in diet preference scores across all of the diets (j, right), in contrast to a higher preference score for 45% HFD and 60% HFD in both sated (j, left) and fasted (j, middle) conditions (one-way RM ANOVA followed by Bonferroni's multiple comparisons; sated: F (1.2,13.19) = 24.57, fasted: F (2.07,31.07) = 17.81, AgRP::3Dq: F (1.7,18.7) = 3.39). Data represented as mean ± SEM (b, d, g, h), or a box and whiskers plot with lines at the median, the first and third quartiles and whiskers at the highest and smallest values of the interquartile range (IQR)∗1.5 (e, i, j) and significance denoted by × p < 0.05. Sample sizes (a–e): n = 16, (f–j): Sated (control): n = 12, Fasted (control): n = 16, AgRP::3Dq: n = 12).

Figure 2

Figure 3 Feeding-induced AgRP neuronal inhibition is dependent on prior exposure to dietary fat.

To measure AgRP neuronal population activity, photometry recordings were performed in mice upon interaction with inaccessible food (b; administered via a tea ball in the homecage), which reduces AgRP neuronal activity, but only to standard diet (SD) and not high fat diets (HFDs) that have not yet been tasted before (c; one-way ANOVA followed by Bonferroni's multiple comparisons; F (2,8) = 15.18; SD n = 4, 45% HFD n = 4, 60% HFD n = 3). In contrast, the first ever consumption of both 45% or 60% HFD decreases AgRP population activity to similar levels as SD consumption (e-f; one-way ANOVA: F (1.89, 5.68) = 1.71; SD n = 6, 45% HFD n = 4, 60% HFD n = 5). Measurement of AgRP population activity indicates that AgRP neurons decrease their activity in similar levels across 45% and 60% HFD in response to inaccessible (h; pairwise t-test; 45% HFD n = 6, 60% HFD n = 6) or consumption (i; pairwise t-test; 45% HFD n = 6, 60% HFD n = 6) of the diets after an overnight binge access period to the diets three days prior to recordings (g). When compared across experiments, AgRP neuronal activity is unchanged in response to inaccessible or consumption of SD (j-k, left; pairwise t-test). Prior exposure to 45% HFD in the homecage (j, middle) decreases AgRP neuronal activity in response to approach to the food in an inaccessible situation or during consumption (k, middle; two-way ANOVA followed by Bonferroni's multiple comparisons; F (1,16) = 6.17), and access to 60% HFD drives decreases in AgRP population activity across accessibility (l, right; one-way ANOVA followed by Bonferroni's multiple comparisons; F (1,16) = 7.94). Significance was determined by × p < 0.05. Data are represented as a mean z-score (calculated based on minutes −3 to −1 of recordings) ± SEM (b, e, h, i, j), or the average z-score calculated from minutes 2–3 (c, f, h-i (insets), or minutes 0–3 (k) post interaction/consumption.

Figure 3

2.15 Chow devaluation experiments with TRPM5 KOs

TRPM5 KOs and their littermate controls were single-housed in standard housing conditions with ad libitum access to standard diet (SD) and water. Baseline body weight and SD measurements were taken at weeks −2 and −1, then a baseline SD fast-refeed was conducted (precluding any food intake measurements for Week 0). For each SD fast-refeed, mice were fasted for ∼18 h overnight, and SD intake was measured for 1 h near the beginning of the light cycle. Subsequently (starting week 0), all mice were given ab libitum access to both 60% HFD and SD, and food intake and body weight were measured across 6 weeks. Notably, SD fast-refeeds were performed after 1 and 5 weeks of exposure to HFD, precluding food intake measurements from occurring during the weeks labeled 2 and 6. Finally, a single withdrawal week (where mice were taken off their experimental diets) preceded a final fast-refeed.

2.16 Orlistat dosing

On average, mice eat approximately 2.4 g of HFD daily [1,30]. To control the dosage, orlistat (50 mg/kg) was administered by mixing the drug with 2.4 g of HFD per mouse and pressing it into pellets using a silicone baking mat (Webake, AKLHLKK) on the day of dosing. Control mice in the HFD-only groups were given HFD pellets prepared in the same way without orlistat. Each pellet weighed approximately 0.6 g, and the mice ate all the pellets daily.

2.17 Chow devaluation experiments with orlistat

Mice were single-housed and separated into two groups: C57BL/6 J mice given HFD and C57BL/6 J mice given HFD with orlistat. Both groups were housed in standard housing conditions and had ad libitum access to standard diet and water in addition to their experimental diets. Baseline body weight and SD intake was measured for 3 days before the start of the protocol. After this baseline period, mice were age and weight-matched, and a fast-refeed was done to measure initial chow intake following an overnight fast. Briefly, mice were fasted for ∼18 h overnight, and chow intake was measured for 1 h near the beginning of the light cycle. After this, SD intake and body weight were measured for 1 week, during which mice were also given access to HFD or HFD/orlistat. Following this experimental phase, another fast refeed was administered. Finally, a withdrawal week (where mice were taken off their experimental diets) separated a final fast-refeed.

2.18 AgRP fiber photometry experiments with orlistat

Agrp-iCre mice expressing GCaMP6s in ArcAgRP neurons were handled by the experimenter at least 3 times before the start of recordings. Photometry recordings were conducted during the light cycle, and mice were transferred to a new home case the night before testing. Mice were fasted for ∼18 h overnight before all recordings. In all conditions, mice were recorded for 5 min before adding food to the cage. During recordings, following the 5-minute baseline period, a chow pellet was added to the corner of the cage. 5 min later, either a second chow pellet (Standard Diet group for BSLN, EXP, and WTH sessions, HFD and HFD + orlistat for BSLN session) or HFD pellet (HFD and HFD + orlistat group for EXP and WTH sessions) was added to the cage.

2.19 Data analysis for fiber photometry

A 2 (Figure 3; minutes −3 → −1) or 5 min (Figure 5; minutes 0 → 5) baseline period was defined prior to the introduction of any experimental manipulations or stimuli. This period was used to calculate the baseline fluorescence (Fbaseline) and the standard deviation (σbaseline) of the signal.Figure 4 Taste receptor function is dispensable for SD devaluation after HFD exposure.

An experimental cohort containing homozygous TRPM5 knockouts (“taste-blind” mice) and littermate controls (n = 13 each, 8 males/5 females) was established. (a) After a baseline period of SD access only, animals were given ab libitum access to both 60% HFD and SD in their homecage (experimental period), followed by a post-HFD withdrawal period. (b) Weekly body weight measurements indicate that TRPM5 KOs are more susceptible to HFD-induced obesity (n = 13 for TRPM5, n = 13 for control, repeated-measures two-way ANOVA with Šídák multiple comparisons; F (9,216) = 5.30). Weekly measurements of (c) standard diet (SD) intake reveal a similar level of decreased SD intake following HFD availability for both groups (repeated-measures mixed effects analysis; x2 = 21.48 (1)), with (d) this depreciation remaining during the withdrawal (W/D) period (n = 13 for TRPM5, n = 13 for control, two-way RM ANOVA (F (1,24) = 0.30), main effect of time (p < 0.0001; F (1,24) = 52.84). Measurements of (e) 60% high fat diet (HFD) intake and (f) total food intake reveal that TRPM5 KOs consume fewer kcals during the experimental period than controls (n = 13 for TRPM5, n = 13 for control, repeated-measures mixed-effects analysis with Šídák multiple comparisons; F (5,118) = 4.36), driven by relatively less consumption of HFD (n = 13 for TRPM5, n = 13 for control, repeated-measures mixed-effects analysis with Šídák multiple comparisons; F (5,118) = 6.53). Nonetheless, TRPM5 KOs and controls (g) exhibit a comparably strong preference for HFD over SD across the experimental period (as indicated by a diet preference score>0; n = 13 for TRPM5, n = 13 for control, unpaired t-test; t (24) = 0.32). To determine if TRPM5 KOs and controls display similar devaluation of SD following HFD exposure, (h) at select time points throughout the paradigm (before HFD exposure (Base), after 1 week on HFD (1 wk), after 5 weeks on HFD (5 wks), and after the withdrawal period (W/D)), animals were fasted for ∼18 h overnight and SD intake was measured at the beginning of the light cycle. One hour food intake measurements indicate that (i) TRPM5 KOs and controls both display rapid SD devaluation, consuming less SD at both experimental timepoints compared to baseline (n = 13 for TRPM5, n = 13 for control, repeated-measures two-way ANOVA with Šídák multiple comparisons; F (3,48) = 3.57). Both groups also display significant revaluation of SD following the withdrawal period, consuming more SD compared to both experimental timepoints. Further, (j) TRPM5 KOs and controls display equivalent extents of SD devaluation, consuming comparable amounts of SD at each timepoint (n = 13 for TRPM5, n = 13 for control, repeated-measures two-way ANOVA with Šídák multiple comparisons; F (3,48) = 3.57). Data represented as mean ± SEM (b-f, i, j), or a box and whiskers plot with lines at the median, the first and third quartiles and whiskers at the highest and smallest values of the interquartile range (IQR)∗1.5 (g) and significance denoted by × p < 0.05.

Figure 4

Figure 5 The behavioral and neural readout of SD devaluation requires fat absorption.

(a) Following a baseline period of access to SD only, animals were given daily 2.4 g 60% HFD (HFD group) or 2.4 g 60% HFD mixed with the lipase inhibitor orlistat (HFD + orlistat group) in their home cage (alongside continued ab libitum access to SD) for 1 week (experimental period), followed by a post-HFD withdrawal (W/D) period. At three timepoints (baseline, experimental, and withdrawal), fast-refeeds were conducted in which 1-hour SD intake was measured following an overnight fast. Average body weights (b) across the paradigm reveal significant weight gain in the HFD group compared to the HFD + orlistat group during the experimental week (n = 14 for HFD + orlistat, n = 14 for HFD, repeated-measures two-way ANOVA with Bonferroni's multiple comparisons; F (3.85,100.21) = 12.31). Longitudinal SD fast-refeeds (c) reveal significantly less cumulative SD intake for the HFD group compared to the HFD + orlistat group during the experimental (Exp) fast-refeed (n = 20 for HFD + orlistat, n = 20 for HFD, repeated-measures two-way ANOVA with Bonferroni's post-hoc; F (2,36) = 7.95). (d) To assess whether this phenomenon could be captured at the level of AgRP activity, Agrp-iCre mice expressing GCaMP6s in AgRP neurons in the ARC were split into three groups: 1) SD group receiving SD throughout the paradigm, 2) HFD group given 2.4 g 60% HFD during the experimental period, and 3) HFD + orlistat group given 2.4 g 60% HFD mixed with orlistat during the experimental period. After a baseline period of access to SD only, animals were dosed according to their group (SD, HFD, or HFD + orlistat) for one week, followed by a post-HFD withdrawal period. At three timepoints (base, exp, and with), fast-refeeds were conducted while recording AgRP activity; at baseline, all mice received 2 consecutive SD pellets, whereas during the experimental and withdrawal timepoints the HFD and HFD + orlistat groups received HFD as their second pellet. For the SD group, (e) averaged individual traces of AgRP neuron z-score responses to food presentation reveal full inhibition of AgRP activity after SD presentation across all timepoints. For the HFD group (f), averaged individual traces of AgRP neuron z-score responses to food presentation reveal diminished inhibition of AgRP activity after SD presentation at the experimental timepoint, with full inhibition only achieved through subsequent presentation of HFD; notably, this effect persists after the withdrawal period. In contrast, for the HFD + orlistat group (g) averaged individual traces of AgRP neuron z-score responses to food presentation resemble the SD group, displaying complete inhibition of AgRP activity at all timepoints. (h) Average AgRP activity during Pellet 1 presentation across groups (SD, HFD, HFD + orlistat) and time (base, exp, with) confirms diminished AgRP inhibition seen in the HFD group during HFD exposure (exp) (n = 7 for SD, n = 6 for HFD, n = 7 for HFD + orlistat, repeated-measures one-way ANOVA with Tukey's multiple comparisons (HFD and SD groups only); SD: F (1.51, 9.03) = 0.57, HFD: F (1.33, 6.65) = 10.69, HFD + Orlistat: F (1.13, 6.78) = 0.64). (i) Change in AgRP activity from Pellet 1 to Pellet 2 presentation across groups (SD, HFD, HFD + orlistat) and time (base, exp, with) underscores diminished inhibition seen in HFD group (purple, middle), unlike SD/HFD + orlistat groups (n = 7 for SD, n = 6 for HFD, n = 6 for HFD + orlistat, repeated-measures one-way ANOVA with Tukey's multiple comparisons (HFD group only); SD: F (1.83, 10.99) = 0.19, HFD: F (1.51, 7.55) = 14.73, HFD + Orlistat: F (1.19, 5.96) = 3.15). Data represented as mean ± SEM and significance determined at p < 0.05.

Figure 5

Calculation of Z-Scores:z=F−Fbaselineσbaseline

Where:

F is the fluorescence signal at each time point.

Fbaseline is the mean fluorescence signal during the baseline period.

σbaseline is the standard deviation of the baseline fluorescence signal.

2.20 Perfusions and histology

After completing behavioral experiments, mice with viral injections and/or optical implants were terminally anesthetized using chloral hydrate (Sigma–Aldrich, 301-17-0) and transcardially perfused first with phosphate-buffered saline (PBS) followed by 10% neutral buffered formalin (Fisher Scientific, SF100). Brains were removed, post-fixed, and dehydrated in 30% sucrose before sectioning into 30–50 um slices using a freezing sliding microtome (Leica Biosystems). Coronal sections were collected and stored at 4 °C. Slices were mounted with a mounting medium containing DAPI (Vectashield, H-1200-10), and images were captured using a 10× objective on an Olympus VS200×Scanscope and 20× objective on a Zeiss Observer Z1 confocal microscope.

2.21 Statistical analysis

GraphPad Prism 10 and RStudio were used for statistical analysis, and GraphPad Prism10, RStudio, and Adobe Illustrator 2023 were used to generate graphs. Two-tailed t-tests were used for discrete comparisons between two groups. For comparisons across groups or between groups over time, repeated measures one-way or two-way ANOVAs were used, respectively, with corresponding post hoc tests adjusted for multiple comparisons. Normality and equal variances were tested and appropriate tests were applied if assumptions were not met (Friedman test, Greenhouse-Geisser correction). Mice were body weight matched across groups before experiments began. Experimenters were not blinded to conditions during testing and analysis. Power analyses were not used to determine sample sizes, however, group sizes were chosen to match previous similar studies.

3 Results

3.1 HFD-induced SD devaluation is dependent on fat content, composition, and preference

Recent work has demonstrated a strong depreciation of SD after both acute and chronic exposure to a 60% HFD, even in states of physiological or artificial hunger [1,2]. This 60% HFD is the most used food source to stimulate diet-induced obesity (DIO) in mice, as its composition elicits maximal overconsumption [31]. To ascertain whether SD devaluation still occurs in response to diets with varying percentages of fat derived from lard, we ran a longitudinal study in which C57BL/6 J male and female mice were exposed to either a 60%, 45%, or 30% animal fat diet in conjunction with SD (Figure 1A; Table 1). Importantly, all food was provided ad libitum. Introduction of either the 60% or 45% HFD led to significant weight gain due to increased fat but not lean mass (Supplemental Figs. 1a–b) compared to animals remaining on SD, with significance occurring after 1 and 2 weeks of HFD exposure, respectively (Figure 1B–C). However, the 30% HFD failed to promote obesity as animals exhibited body weights comparable to SD controls (Figure 1B). Shortly after HFD presentation, mice consumed very few kilocalories (kcal) of SD (Figure 1D), instead opting to overeat the HFD (Figure 1E), which resulted in a higher weekly amount of ingested kcals in the 60% and 45% HFD cohorts (Figure 1F). This robust preference for HFD could be seen across animals and time (Figure 1G–H) and scaled with the percentage of lard (Figure 1I). The higher preference for the 60% and 45% HFD, resulting in enhanced caloric intake (Figure 1D), glucose levels (Supplemental Figs. 1c–d), and leptin levels (Supplemental Fig. 1e), were at least in part responsible for the increased weight gain in these groups. Indirect calorimetry was used to explore the putative contribution of energy expenditure (Supplemental Fig. 1f) and locomotor activity (Supplemental Fig. 1g) in these cohorts, but no significant differences were found. Respiratory exchange ratio shifted from primarily carbohydrate to fat intake in accord with higher percentages of HFD (Supplemental Fig. 1h). Furthermore, when normalizing energy expenditure to lean mass [32], no significant difference was observed in total energy expenditure across diet exposures (Supplemental Fig. 1i).

Next, to determine if this home cage preference for the different HFDs could result in HFD-induced SD devaluation during periods of energy deficiency, we longitudinally measured SD intake during a 1 h refeeding session following an overnight fast across time. Both the 45% and 60% HFD caused a strong depreciation of SD at all time points assessed, whereas the 30% HFD only reached significance at 8 weeks of diet exposure (Figure 1J–K). This finding was independent of sex (Supplemental Figs. 2a–b). Notably, the change in body weight observed by week 4 of diet exposure was significantly correlated to the fast-refeed SD devaluation in the 45% and 60% HFD cohorts, but not in the 30% HFD or SD cohorts (Figure 1L). Interestingly, the highest preferred HFDs resulted in the strongest SD devaluation (Supplemental Fig. 2c). This notion is further reinforced by the lack of HFD-induced SD devaluation observed in mice with prolonged exposure to a 50% plant-based HFD (Supplemental Fig. 2d).

3.2 Hungry mice rapidly transition to exclusively consuming 60% HFD when given a choice

Mice demonstrated a clear preference for the 60% and 45% HFDs (Figure 1H–I), consuming little to no SD as early as 1 week after the HFD was introduced to the home cage (Figure 1D). To evaluate both the speed and identity of the most preferred diet, mice were simultaneously presented with a buffet of HFDs in conjunction with SD, and total intake of each diet was recorded daily (Figure 2A). We found that every male and female animal tested formed a robust preference for the 60% HFD after just 1 day of exposure, which persisted across the entire week (Figure 2B–D). As expected, these mice exhibited a strong SD devaluation when the HFDs were removed from the home cage (Figure 2B–D), as SD consumption significantly decreased ∼35% in the withdrawal (post-HFD) versus baseline (pre-HFD) phase (Figure 2E).

To further capture the temporal kinetics of this preference, we measured food intake after the first hour of the HFD buffet and SD exposure in three groups of mice: 1) a cohort that had ad libitum access to SD right before the assay, 2) an overnight fasted cohort, and 3) a cohort that had ad libitum access to SD right before the assay with chemogenetic activation of AgRP neurons (Figure 2F). Notably, all mice received the same dose of the otherwise pharmacologically inert ligand clozapine-n-oxide (CNO). As previously demonstrated, AgRP chemogenetic stimulation led to voracious total food intake, comparable to fasted mice (Figure 2G). Both the sated and fasted cohorts displayed a preference for the 45% and 60% HFDs, whereas the AgRP activation mice continued to eat comparable amounts of SD in addition to the 45% and 60% HFDs (Figure 2H–J), likely due to the inability of AgRP neurons to be silenced during consumption in these animals. This delay in the permanent switching to consumption of the 60% HFD potentially suggests that mice need to consume the different diets to make informed decisions on future intake.

3.3 AgRP neuronal inhibition is dependent on prior exposure to dietary fat

To test this notion at the level of AgRP cells, population dynamics were assessed using optical fiber photometry. All Agrp-iCre mice received both a unilateral injection of Cre-dependent GCaMP6s and a fiber implant targeting the ARC. Mice that had only been exposed to SD were presented with inaccessible SD, 45%, or 60% HFD (Figure 3A). While inaccessible SD resulted in a rapid suppression of AgRP activity [20], neither HFD had any effect (Figure 3B–C). However, AgRP neurons were inhibited to a similar degree as SD during the first exposure to each HFD when it was accessible and mice were permitted to consume it (Figure 3E–F). After an initial overnight consummatory exposure, AgRP activity was suppressed the very next time these animals were presented with either inaccessible or accessible HFDs (Figure 3G–I). This implies that AgRP inhibition to the sensory detection and consumption of discrete food substrates is contingent on prior ingestive experience, which could stem from taste and/or post-ingestive information (Figure 3J–K).

3.4 Taste receptor function is dispensable for SD devaluation after HFD exposure

If prior HFD consumption is required for AgRP neuron inhibition, then HFD-induced SD devaluation may necessitate taste. To assess the relevance of gustatory perception for HFD preference and SD depreciation, “taste-blind” TRPM5 knockouts (KO) and weight-, sex-, and age-matched controls were supplemented with ad libitum access to 60% HFD in addition to SD (Figure 4A). Both groups gained weight across the experimental period with TRPM5 KOs displaying resistance to HFD-induced obesity compared to controls (Figure 4B), which was mostly driven by changes observed in males (Supplemental Fig. 3a), as previously described [14]. Notably, both cohorts almost completely ceased SD intake when HFD became available (Figure 4C; Sup Figure 3b), and this SD depreciation remained during the withdrawal period (Figure 4D; Sup Fig 3c). The decreased weight gain observed in the TRPM5 KOs could be explained by the reduction in HFD consumption (Figure 4E; Sup Figure 3d) and total caloric consumption (Figure 4F; Sup Figure 3e) compared to controls. In summary, TRPM5 KOs exhibited similar preference for HFD over SD as control animals (Figure 4G; Sup Figure 3f) supporting the relevance of taste-independent motivations to consume HFD.

Next, we assessed whether these “taste-blind” mice displayed HFD-induced SD devaluation under physiological hunger. To this end, TRPM5 KOs and their control counterparts were fasted overnight and then presented with SD for 1 h the next morning across the experimental timeline (Figure 4H). Both groups of mice exhibited a robust HFD-induced SD devaluation 1 week and 5 weeks after home cage HFD exposure, significantly reducing their 1 h fast-refeed SD consumption compared to baseline (pre-HFD exposure) (Figure 4I; Supp Fig. 3g). This HFD-induced SD devaluation was rescued after a 1 week withdrawal phase in both groups (Figure 4I; Supp Fig. 3g). No differences in the fast-refeed SD consumption were observed between groups across the longitudinal measurements (Figure 4J; Supp Fig. 3h), suggesting taste encoded via TRPM5 is not a factor in meeting caloric demand.

3.5 HFD-induced SD devaluation requires proper fat absorption

Given our findings that 1) HFD preference requires post-ingestive feedback, 2) HFD-mediated silencing of AgRP neurons requires consummatory experience, and 3) taste does not play an obligatory role in HFD-induced SD devaluation, we next considered the contribution of dietary fat absorption on these processes. Accordingly, we separated mice into two groups: one receiving HFD alone and one receiving HFD mixed with orlistat. This lipase inhibitor prevents the breakdown of triglycerides in the intestine, resulting in reduced fat absorption [33]. After a baseline period of ab libitum access to SD only, both groups were dosed daily across 1 week with pre-measured amounts of HFD or HFD/orlistat while maintaining ab libitum access to SD (experimental period), followed by HFD removal for 1 week (withdrawal period) (Figure 5A). Importantly, all animals from both groups consumed the same daily HFD presented to them (2.4 g). Body weight measurements across the experimental period revealed significant weight gain in the HFD group compared to mice receiving HFD/orlistat (Figure 5B). After removal of HFD, body weights for both groups rapidly converged, with mice previously on HFD losing weight and mice previously on HFD/orlistat gaining weight (Figure 5B).

To determine the relevance of appropriate fat digestion on HFD-induced SD devaluation, we longitudinally tested fast-refeed SD consumption in both groups before HFD access (baseline), at the conclusion of the one-week HFD exposure (experimental period), and 1 week after HFD removal (withdrawal period). While both groups displayed similar baseline levels of SD intake, the HFD group consumed significantly less SD during the experimental period compared to the HFD/orlistat group (Fig. 5C), demonstrating the importance of appropriate fat absorption in the development and expression of HFD-induced SD devaluation. Notably, the two groups exhibited comparable fast-refeed SD intake at the end of the withdrawal phase (Figure 5C).

3.6 AgRP neural readout of HFD-induced SD devaluation necessitates appropriate fat digestion

Prior research demonstrated that HFD-induced SD devaluation could be detected at the level of AgRP neurons, whereby the suppression of AgRP population activity in response to SD was dampened in hungry mice previously exposed to HFD, independent of body weight differences [1,2]. To evaluate the relevance of effective fat absorption to this alteration in AgRP neural dynamics, we performed longitudinal AgRP fiber photometry GCaMP recordings. Animals were distributed into three groups: 1) those that remained on SD throughout the length of the experiment, and those exposed to measured quantities of 2) HFD or 3) HFD/orlistat in conjunction with ad libitum SD across the 7-day experimental period before a 7-day withdrawal period. AgRP activity was recorded in response to two separate food pellet presentations separated by 5 min. At baseline, all mice were exposed to two consecutive SD pellets. During both the experimental and withdrawal phases, HFD naïve animals continued to receive 2 SD pellet drops, while mice with prior exposure to HFD received an SD pellet followed by a HFD pellet (Figure 5D). As expected, mice in the SD group displayed rapid AgRP suppression during SD presentation longitudinally across time (Figure 5E,H), with no further inhibition upon the introduction of a second SD pellet (Figure 5E,I). In contrast, while mice in the HFD group showed a robust silencing of AgRP activity during SD presentation at baseline, this inhibition by SD was significantly diminished at the end of the experimental period (Figure 5F,H). This inability of SD to fully quell AgRP activity likely contributes to the struggles of dieting, as nutritionally balanced food substrates are no longer capable of switching off the negative consequences associated with hunger. Vitally, introduction of a subsequent HFD pellet in these animals enabled the complete inhibitory response of AgRP neurons (Figure 5F,I). The drive to seek out and consume palatable foods to completely switch off these starvation sensitive cells serves to elucidate why dieting is susceptible to regression. Remarkably, mice in the HFD/orlistat group, which consumed identical quantities of homecage HFD as the HFD group, revealed complete suppression of AgRP activity upon presentation of SD alone across all time points (Figure 5G,H), suggesting that intact fat digestion is required for HFD-induced SD devaluation, at least in part through AgRP signaling. Moreover, AgRP neurons in the HFD/orlistat group were non-responsive to a subsequent HFD pellet (Figure 5G,I), implicating that maximal inhibition had already been met. In conclusion, we found that HFD preference and subsequent SD devaluation relied on a number of factors including the amount, source, length of exposure, and post-ingestive feedback of fat.

4 Discussion

The obesity epidemic presents a complex and multifaceted challenge, affecting millions globally. Its roots lie in a combination of factors including poor dietary habits, sedentary lifestyles, genetic predispositions, and socioeconomic influences. The rise of highly processed, calorie-dense foods has contributed significantly to increasing obesity rates, leading to chronic conditions like diabetes, heart disease, and certain cancers, while placing a tremendous burden on healthcare systems. Restricting individual caloric intake causes weight loss, but continued, persistent dieting is difficult due to a combination of physiological, psychological, and environmental factors. Physiologically, the body's mechanisms to maintain weight (such as hormone regulation and metabolic adaptation) can make sustained weight loss challenging. When calorie intake is reduced, the body often responds by slowing metabolism and increasing hunger signals, which spurs overeating. We and others have previously shown that exposure to 60% HFD in mice promotes a depreciation of energy sources once deemed a valuable source of calories [1,2,9]. Understanding the causes of this HFD-induced SD devaluation could lead to more successful dieting strategies that reduce the frustration and discouragement associated with adhering to a diet.

We assessed the capacity of distinct HFDs in promoting SD devaluation and found both the source (lard vs vegetable) and amount of fat in the diet are key factors. Higher percentages of animal fat content led to higher intake and preference over SD when both were provided ad libitum in the home cage; this predilection for HFD occurred rapidly but required ingestive experience. Importantly, mice exposed to lard-based HFDs demonstrated a robust SD devaluation even in times of caloric need, opting to consume less SD than they previously did at baseline (when they were naive to HFD). We found that the strength and speed of this SD devaluation scaled with the proportion of animal fat in the diet, suggesting fattier food substrates are more efficient in driving this phenomenon. This HFD-induced SD devaluation observed under physiological hunger was directly correlated with the specific HFD preference over SD in the home cage. Interestingly, prolonged exposure to a plant-based HFD failed to elicit SD devaluation in our fast-refeed assay, potentially due to the lack of preference for this diet over SD.

Our observation that preference for HFD occurred over the course of hours to days (in contrast to seconds and minutes) suggests the importance of familiarity and experience with different food substrates. While we don't know precisely at which time point this preference fully develops, we found every animal tested, regardless of sex, exhibited a robust 60% HFD preference after 24 h that was not as obvious in the first hour of exposure. It is likely that over the course of a day, the mice were able to make associations based on post-ingestive feedback/taste/texture/smell that resulted in this preference. When presented with novel HFDs, physiologically hungry mice still consume moderate levels of SD, presumably due to the knowledge that it is a safe and ample source of energy. Notably, artificial hunger evoked through chemogenetic AgRP activation resulted in no definitive HFD preference during initial presentation, a key distinction to fasted mice. This may be due to the non-physiological nature of this manipulation in which the inability of appropriate AgRP activity suppression during food consumption locks in feeding behavior toward the closest/current supply of calories. This notion that prior exposure is required to develop a selective HFD preference was further reinforced by our AgRP population activity recordings. Despite the vigorous preference for HFD over SD, we found that AgRP dynamics were unaltered in mice naive to HFD upon the presentation of inaccessible HFD. However, AgRP neurons are rapidly suppressed during the first exposure to accessible HFD and inaccessible HFD thereafter, highlighting the significance of learned experience acquired from eating in hunger circuits. These findings build on seminal studies supporting a learned contingency between sensory cues and the nutritive value of food whereby both a transient, sensory-mediated anticipatory inhibition and a sustained, calorie-mediated inhibition was observed when these properties were isolated [21,24].

The requirement of prior HFD consumption to effectively silence AgRP activity implies orosensory information (i.e. taste, texture, and/or post ingestive feedback) is vital for this preference and subsequent HFD-induced SD devaluation. Both elements likely contribute to AgRP inhibition, as population dynamics are promptly diminished during the initial bout of HFD intake and remain durably suppressed during feeding. Gustatory information may mediate this quick drop in AgRP activity, and post-ingestive signaling throughout the meal ensures inhibition remains constant. Reinforcing this idea, a recent study showed that AgRP neurons are rapidly inhibited during each bout of ingestion by a signal linked to the taste of food [34]. Moreover, intragastric infusion of nutrients has been shown to stably suppress AgRP activity with a delayed response [[22], [23], [24]]. Therefore, we sought to isolate and evaluate the contributions of taste or post-ingestive information in establishing HFD-induced SD devaluation.

We determined that “taste-blind” TRPM5 KOs still exhibited a biased HFD preference, consuming little to no SD when given a choice in the home cage and displaying HFD-induced SD devaluation under a caloric deficit (similar to controls). These animals have abolished sweet, amino acid, and bitter taste reception [10], in addition to perturbed fat taste perception and preference [12,13]. However, other studies have suggested these null mice still respond to sweet, umami, and bitter compounds [11], implicating taste may be abrogated but not entirely nullified during our assay. This potential partial blockade of HFD taste may have contributed to HFD-induced SD devaluation. One way in which the sense of taste regulates feeding is by supplying the positive feedback that propels a meal forward, often leading to overeating. Interestingly, we found that TRPM5 KOs consumed less 60% HFD than controls, which granted them partial resistance to diet-induced obesity. This suggests these animals may be eating smaller or less frequent meals due to the inability to process gustatory information. Unfortunately, we were unable to rule out the role of HFD texture in our study; it is possible that TRPM 5 KOs continue to show HFD preference and SD devaluation due to the softer consistency of lard in the HFDs.

Alternatively, it may indeed be post-ingestive feedback that is integral for this effect, which was strengthened by our observation that hindering fat absorption eliminates HFD-induced SD devaluation at both the behavioral and AgRP activity level. Notably, several studies have reported that ingested nutrients acutely trigger a transient surge of dopamine to the striatum, and recently it was shown that intragastric delivery of nutrients led to an increase of dopamine release in the nucleus accumbens (NAc) [[35], [36], [37], [38], [39], [40]] and the basolateral amygdala (BLA) [35]. Furthermore, discrete subpopulations of ventral tegmental are (VTA)-DA neurons track the internal, post-ingestive consequences of eating and help determine which foods will keep an animal satiated [35].

Orlistat is a clinically-approved medication used to aid weight loss and prevent weight regain by inhibiting the action of lipase, an enzyme in the digestive system responsible for breaking down fats [33]. Inhibiting lipase activity results in ∼30% of the fat consumed in a diet to pass directly through the body without nutritive absorption [41]. Despite the incomplete prevention of fat absorption, we found that mice administered orlistat with HFD failed to exhibit HFD-induced SD devaluation, implying that even a partial disruption of fat digestion can alter future feeding episodes. Reinforcing this notion, the blunted AgRP suppression to SD observed in HFD-exposed mice was absent in orlistat-treated animals. It should be noted that orlistat has gastrointestinal side effects after consuming high-fat meals. Therefore, it is possible that these animals fail to gain weight or submit to SD devaluation while on HFD due to visceral malaise and subsequent avoidance of the HFD. However, we think this limitation is unlikely for several reasons: 1) every test subject in our study consumed all the measured HFD presented to them daily, 2) each cage had solid, blue-dyed fecal deposits associated with consumption of 60% HFD, and 3) we never observed behaviors linked with malaise such as hunching or decreased movement. Nevertheless, the GI issues coupled to this drug are an important caveat to highlight in our experiments. Relatedly, it should be noted that HFD-orlistat-treated mice showed pronounced weight gain following the cessation of treatment, likely due to overcompensating for lost calories. While we didn't record basal levels of AgRP dynamics during this withdrawal period, we speculate that activity is higher at this time and the hyperphagia displayed may be blunted with artificial inhibition of AgRP neurons.

Collectively, we concluded that HFD preference and SD devaluation, as determined by food intake and AgRP activity changes, scales with the degree of animal fat content, necessitates learned information following consumption, is independent of gustatory signaling via TRPM5, and requires post ingestive feedback via proper fat absorption. A better understanding of dieting challenges may help in devising strategies to facilitate weight loss and adherence to a healthier lifestyle.

CRediT authorship contribution statement

Ames K. Sutton Hickey: Writing – review & editing, Formal analysis, Data curation, Conceptualization. Jordan Becker: Writing – review & editing, Formal analysis, Data curation, Conceptualization. Eva O. Karolczak: Writing – review & editing, Formal analysis, Data curation, Conceptualization. Andrew Lutas: Writing – review & editing, Conceptualization. Michael J. Krashes: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A Supplementary data

The following is the Supplementary data to this article:Multimedia component. 1

Supp Fig. 1: Metabolic parameters following exposure to various HFDs in the homecage.Following 8–12 weeks of diet exposures, metabolic analyses demonstrate that homecage access to 45% or 60% HFD increases fat mass (a) (Welch one-way ANOVA followed by Tukey's multiple comparisons, F(3,27) = 21.4) without changing lean mass (b; Kruskal–Wallis test, x2(3) = 2.84), whereas this is not observed in the 30% HFD cohort. Glucose homeostasis, measured by fed blood glucose (c; Kruskal–Wallis test, x2(3) = 14.3, followed by Dunn's multiple comparisons using Bonferroni adjustments), and a glucose tolerance test (d; RM two-way ANOVA followed by Bonferroni's multiple comparisons; F(12,228) = 5.50) is decreased primarily in the 60% HFD cohort. Longitudinal access to any HFD increases circulating leptin (e; Kruskall–Wallis test, x2(3) = 24.1, followed by Dunn's multiple comparisons using Bonferroni adjustments). Energy expenditure measurements such as oxygen consumption (VO2) (f) and physical activity (g) are unaffected across cohorts (f: Kruskal–Wallis test, x2(3) = 4.38, g: one-way ANOVA, F(3,55) = 3.47). Energy utilization, demonstrated by respiratory exchange ratio (RER) (h), moves towards the breakdown of fat (indicated by a lower RER) in 45% and 60% HFD cohorts (one-way ANOVA followed by Tukey's multiple comparisons, F(3,56) = 2.82). (i) 24-hour normalized total energy expenditure (TEE) was similar across groups (calculated using an ANCOVA with each animal's lean mass as the covariate; F(3,55) = 0.87). Significance determined by p < 0.05 and indicated by × 60% vs SD, # 60% vs. 45%, % 60% vs. 30%, and ˆ 30% vs. SD. Data represented as a box and whiskers plot with lines at the median, the first and third quartiles and whiskers at the highest and smallest values of the interquartile range (IQR) ∗1.5 (a-c), or the 5th and 95th percentiles (f-h) or mean ± SEM (d). Sample sizes are SD: n = 16–24, 30% HFD: n = 11, 45% HFD: n = 18, 60% HFD: n = 15–24.

Supp Fig. 2: SD devaluation after exposure to various HFDs occurs across biological sex.SD consumption following a fast-refeed experiment across the experimental timepoints indicates that mice with access to SD (orange) or 30% HFD (blue) do not devalue SD in females (a; RM two-way ANOVA; F(3,15) = 0.161) or males (b; two-way ANOVA; F(3,19) = 1.301). In contrast, exposure to both 45% HFD (green) and 60% HFD (purple) in the homecage decreases SD intake in both females (a; RM two-way ANOVA followed by Bonferroni's multiple comparisons; 45% HFD: F(3,27) = 11.26, 60% HFD: F(3,12) = 14.01) and males (b; two-way ANOVA followed by Bonferroni's multiple comparisons; 45% HFD: F(3,18) = 22.51, 60% HFD: F(3,19) = 67.54), beginning 1 week after exposure and continuing until 8 weeks post-HFD exposure. (c) In response to exposure to a vegetable-based HFD (veg) alongside SD in their homecage, both females and males show less preference for the veg-HFD than those that are exposed to lard-based HFDs (30%–60% anim HFD), indicated by a lower diet preference score (one-way ANOVA followed by Bonferroni's multiple comparisons; females: F(3,70) = 17.17, males: F(3,48) = 12.86). (d) SD devaluation, measured by SD intake following a fast-refeed, is not altered by homecage veg-HFD exposure (RM two-way ANOVA, F(3,18) = 2.59). Significance determined by p < 0.05 and indicated by ∗. Data represented as mean ± SEM (a-b, d), or as a box and whiskers plot with lines at the median, the first and third quartiles and whiskers at the highest and smallest values of the interquartile range (IQR) ∗1.5 (c). Sample sizes are females: SD: n = 6–9, 30% HFD: n = 8, 45% HFD: n = 10, 60% HFD: n = 5–8, 50% veg: 5–8, males: SD: n = 2–7, 30% HFD: n = 4, 45% HFD: n = 9, 60% HFD: n = 2–7, 50% veg: 2–15.

Supp Fig. 3: Taste receptor function is dispensable for SD devaluation after HFD exposure (data segregated by sex).(a, top) Weekly body weight measurements indicate that male TRPM5 KOs are less susceptible to HFD-induced obesity than controls (n = 8 per group, repeated-measures two-way ANOVA with Šídák multiple comparisons; F(9,126) = 5.97) whereas (a, bottom) female TRPM5 KOs and controls are comparably susceptible to HFD-induced obesity (n = 5 per group, repeated-measures two-way ANOVA with Šídák multiple comparisons; F(7,56) = 1.37). Weekly measurements of (b) standard diet (SD) intake reveal a comparable cessation of SD intake during HFD availability for males and females across both groups (males: n = 8 per group; repeated-measures mixed effects analysis; x2 = 17.55(1); females: n = 5 per group, repeated measures two-way ANOVA; F(5,40) = 1.10), with (c) this depreciation remaining during the withdrawal (W/D) period across genotypes for males (top, n = 8 per group, repeated-measures two-way ANOVA; (F(1,14) = 0.01), main effect of time (p < 0.0001; F(1,14) = 89.42) and females (bottom, n = 5 per group, repeated-measures two-way ANOVA; F(1,8) = 1.40, main effect of time (p < 0.05, F(1,8) = 8.32). Measurements of (d) 60% high fat diet (HFD) intake and (e) total food intake reveal that both male and female TRPM5 KOs consume fewer kcals during the experimental period than controls (e, top, n = 8 per group, males, repeated-measures mixed effects analysis with Šídák multiple comparisons; F(5,118) = 4.36) (e, bottom, n = 5 per group, females, repeated measures mixed effects analysis with Šídák multiple comparisons; F(5,47) = 3.45), which is driven by relatively less HFD consumption for both sexes (d, top, n = 8 per group, males, repeated-measures mixed-effects analysis with Šídák multiple comparisons; F(5,69) = 4.19) (d, bottom, n = 5 per group, females, repeated-measures mixed-effects analysis with Šídák multiple comparisons; F(5,39) = 6.77). Nonetheless, both male and female TRPM5 KOs and controls (f) exhibit a comparably strong preference for HFD over SD across the experimental period (as indicated by a diet preference score>0, calculated using the following equation: (HFD intake - SD intake)/(HFD intake + SD intake)) (top, n = 8 per group, males, unpaired t-test, t(14) = .32) (bottom, n = 5 per group, females, unpaired t-test, t(8) = 0.06). Longitudinal fast-refeed measurements indicate that (g, top) male TRPM5 KOs and controls both display rapid SD devaluation, consuming less SD at both experimental timepoints compared to baseline (n = 8 per group, males, repeated-measures two-way ANOVA with Šídák multiple comparisons; F(3,42) = 0.34). Both male groups also display revaluation of SD following the withdrawal period, consuming more SD than during the final experimental timepoint. Meanwhile, (g, bottom) females across groups show devaluation by time, without respect to genotype (n = 5 per group, females, repeated-measures two-way ANOVA; F(3,24) = 1.80). Critically, (h) both male and female TRPM5 KOs display comparable extents of SD devaluation to controls (top, n = 8 per group, males, repeated-measures two-way ANOVA with Šídák multiple comparisons; F(3,42) = 0.34) (bottom, n = 5 per group, females, repeated-measures two-way ANOVA; F(3,24) = 1.80). Data represented as mean ± SEM (b-f, i, j), or a box and whiskers plot with lines at the median, the first and third quartiles and whiskers at the highest and smallest values of the interquartile range (IQR)∗1.5 (g) and significance denoted by × p < 0.05.

Multimedia component. 1

Data availability

Data will be made available on request.

Acknowledgements

We thank the mouse metabolism core at NIDDK, including director Dr. Oksana Gavrilova, for the assistance in measuring metabolic parameters in mice including indirect calorimetry, glucose homeostasis, and circulating hormone levels. This research was supported by the Intramural Research Program of the National Institutes of Health, the National Institutes of Diabetes and Digestive and Kidney Diseases (DK075088 to M.J.K. and DK075087-06 to M.J.K.) and the Nancy Nossal Fellowship (NIH-NIDDK; to A.K.SH.).

Appendix A Supplementary data to this article can be found online at https://doi.org/10.1016/j.molmet.2024.102021.
==== Refs
References

1 Mazzone C.M. Liang-Guallpa J. Li C. Wolcott N.S. Boone M.H. Southern M. High-fat food biases hypothalamic and mesolimbic expression of consummatory drives Nat Neurosci 23 2020 1253 1266 32747789
2 Beutler L.R. Corpuz T.V. Ahn J.S. Kosar S. Song W. Chen Y. Obesity causes selective and long-lasting desensitization of AgRP neurons to dietary fat Elife 9 2020
3 Black B.L. Croom J. Eisen E.J. Petro A.E. Edwards C.L. Surwit R.S. Differential effects of fat and sucrose on body composition in A/J and C57BL/6 mice Metabolism 47 1998 1354 1359 9826212
4 Hu S. Wang L. Yang D. Li L. Togo J. Wu Y. Dietary fat, but not protein or carbohydrate, regulates energy intake and causes adiposity in mice Cell Metabol 28 2018 415 431.e4
5 Sumiyoshi M. Sakanaka M. Kimura Y. Chronic intake of high-fat and high-sucrose diets differentially affects glucose intolerance in mice J Nutr 136 2006 582 587 16484528
6 Drewnowski A. Greenwood M.R. Cream and sugar: human preferences for high-fat foods Physiol Behav 30 1983 629 633 6878464
7 Briggs D.I. Enriori P.J. Lemus M.B. Cowley M.A. Andrews Z.B. Diet-induced obesity causes ghrelin resistance in arcuate NPY/AgRP neurons Endocrinology 151 2010 4745 4755 20826561
8 Briggs D.I. Lockie S.H. Wu Q. Lemus M.B. Stark R. Andrews Z.B. Calorie-restricted weight loss reverses high-fat diet-induced ghrelin resistance, which contributes to rebound weight gain in a ghrelin-dependent manner Endocrinology 154 2013 709 717 23307790
9 Boone M.H. Liang-Guallpa J. Krashes M.J. Examining the role of olfaction in dietary choice Cell Rep 34 2021 108755 33596417
10 Zhang Y. Hoon M.A. Chandrashekar J. Mueller K.L. Cook B. Wu D. Coding of sweet, bitter, and umami tastes: different receptor cells sharing similar signaling pathways Cell 112 2003 293 301 12581520
11 Damak S. Rong M. Yasumatsu K. Kokrashvili Z. Pérez C.A. Shigemura N. Trpm5 null mice respond to bitter, sweet, and umami compounds Chem Senses 31 2006 253 264 16436689
12 Liu P. Shah B.P. Croasdell S. Gilbertson T.A. Transient receptor potential channel type M5 is essential for fat taste J Neurosci 31 2011 8634 8642 21653867
13 Sclafani A. Ackroff K. Fat preference deficits and experience-induced recovery in global taste-deficient Trpm5 and Calhm1 knockout mice Physiol Behav 246 2022 113695 34998826
14 Larsson M.H. Håkansson P. Jansen F.P. Magnell K. Brodin P. Ablation of TRPM5 in mice results in reduced body weight gain and improved glucose tolerance and protects from excessive consumption of sweet palatable food when fed high caloric diets PLoS One 10 2015 e0138373
15 Hill J.O. Peters J.C. Environmental contributions to the obesity epidemic Science 280 1998 1371 1374 9603719
16 Swinburn B.A. Sacks G. Hall K.D. McPherson K. Finegood D.T. Moodie M.L. The global obesity pandemic: shaped by global drivers and local environments Lancet 378 2011 804 814 21872749
17 Takahashi K.A. Cone R.D. Fasting induces a large, leptin-dependent increase in the intrinsic action potential frequency of orexigenic arcuate nucleus neuropeptide Y/Agouti-related protein neurons Endocrinology 146 2005 1043 1047 15591135
18 Mandelblat-Cerf Y. Ramesh R.N. Burgess C.R. Patella P. Yang Z. Lowell B.B. Arcuate hypothalamic AgRP and putative POMC neurons show opposite changes in spiking across multiple timescales Elife 4 2015
19 Krashes M.J. Koda S. Ye C. Rogan S.C. Adams A.C. Cusher D.S. Rapid, reversible activation of AgRP neurons drives feeding behavior in mice J Clin Invest 121 2011 1424 1428 21364278
20 Chen Y. Lin Y.-C. Kuo T.-W. Knight Z.A. Sensory detection of food rapidly modulates arcuate feeding circuits Cell 160 2015 829 841 25703096
21 Betley J.N. Xu S. Cao Z.F.H. Gong R. Magnus C.J. Yu Y. Neurons for hunger and thirst transmit a negative-valence teaching signal Nature 521 2015 180 185 25915020
22 Bai L. Mesgarzadeh S. Ramesh K.S. Huey E.L. Liu Y. Gray L.A. Genetic identification of vagal sensory neurons that control feeding Cell 179 2019 1129 1143.e23 31730854
23 Beutler L.R. Chen Y. Ahn J.S. Lin Y.C. Essner R.A. Knight Z.A. Dynamics of gut-brain communication underlying hunger Neuron 96 2017 461 475.e5 29024666
24 Su Z. Alhadeff A.L. Betley J. N. Nutritive Post-ingestive signals are the primary regulators of AgRP neuron activity Cell Rep 21 2017 2724 2736 29212021
25 Reed F. Reichenbach A. Dempsey H. Clarke R.E. Mequinion M. Stark R. Acute inhibition of hunger-sensing AgRP neurons promotes context-specific learning in mice Mol Metabol 77 2023 101803
26 Goldstein N. McKnight A.D. Carty J.R.E. Arnold M. Betley J.N. Alhadeff A.L. Hypothalamic detection of macronutrients via multiple gut-brain pathways Cell Metabol 33 2021 676 687.e5
27 Berthoud H.-R. The vagus nerve, food intake and obesity Regul Pept 149 2008 15 25 18482776
28 Tolhurst G. Reimann F. Gribble F.M. Intestinal sensing of nutrients Handb Exp Pharmacol 2012 309 335 22249821
29 Tong Q. Ye C.-P. Jones J.E. Elmquist J.K. Lowell B.B. Synaptic release of GABA by AgRP neurons is required for normal regulation of energy balance Nat Neurosci 11 2008 998 1000 19160495
30 Wang X. Li H. Chronic high-fat diet induces overeating and impairs synaptic transmission in feeding-related brain regions Front Mol Neurosci 15 2022 1019446 36226318
31 Licholai J.A. Nguyen K.P. Fobbs W.C. Schuster C.J. Ali M.A. Kravitz A.V. Why do mice overeat high-fat diets? How high-fat diet alters the regulation of daily caloric intake in mice Obesity 26 2018 1026 1033 29707908
32 Speakman J.R. Measuring energy metabolism in the mouse - theoretical, practical, and analytical considerations Front Physiol 4 2013 34 23504620
33 Bray G.A. Purnell J.Q. Feingold K.R. Anawalt B. Blackman M.R. Boyce A. An historical review of steps and missteps in the discovery of anti-obesity drugs Feingold K.R. Endotext 2022 MDText.com, Inc. South Dartmouth (MA)
34 Aitken T.J. Liu Z. Ly T. Shehata S. Sivakumar N. La Santa Medina N. Negative feedback control of hunger circuits by the taste of food bioRxiv 2023 10.1101/2023.11.30.569492
35 Grove J.C.R. Gray L.A. La Santa Medina N. Sivakumar N. Ahn J.S. Corpuz T.V. Dopamine subsystems that track internal states Nature 2022 10.1038/s41586-022-04954-0
36 Alhadeff A.L. Goldstein N. Park O. Klima M.L. Vargas A. Betley J.N. Natural and drug rewards engage distinct pathways that converge on coordinated hypothalamic and reward circuits Neuron 103 2019 891 908.e6 31277924
37 Zhang L. Han W. Lin C. Li F. de Araujo I.E. Sugar metabolism regulates flavor preferences and portal glucose sensing. Front. Integr Neurosciences 12 2018 57
38 Han W. Tellez L.A. Perkins M.H. Perez I.O. Qu T. Ferreira J. A neural circuit for gut-induced reward Cell 175 2018 665 678.e23 30245012
39 Hsu T.M. McCutcheon J.E. Roitman M.F. Parallels and overlap: the integration of homeostatic signals by mesolimbic dopamine neurons Front Psychiatr 9 2018 410
40 de Araujo I.E. Oliveira-Maia A.J. Sotnikova T.D. Gainetdinov R.R. Caron M.G. Nicolelis M.A. Food reward in the absence of taste receptor signaling Neuron 57 2008 930 941 18367093
41 Drent M.L. van der Veen E.A. First clinical studies with orlistat: a short review Obes Res 3 Suppl 4 1995 623S 625S 8697067
