Dynamic processing of hunger and thirst by common mesolimbic neural ensembles

Abstract

The nucleus accumbens (NAc) is a canonical reward center that regulates feeding and drinking but it is not known whether these behaviors are mediated by same or different neurons. We employed two-photon calcium imaging in awake, behaving mice and found that during the appetitive phase, both hunger and thirst are sensed by a nearly identical population of individual D1 and D2 neurons in the NAc that respond monophasically to food cues in fasted animals and water cues in dehydrated animals. During the consummatory phase, we identified three distinct neuronal clusters that are temporally correlated with action initiation, consumption, and cessation shared by feeding and drinking. These dynamic clusters also show a nearly complete overlap of individual D1 neurons and extensive overlap among D2 neurons. Modulating D1 and D2 neural activities revealed analogous effects on feeding versus drinking behaviors. In aggregate, these data show that a highly overlapping set of D1 and D2 neurons in NAc detect food and water reward and elicit concordant responses to hunger and thirst. These studies establish a general role of this mesolimbic pathway in mediating instinctive behaviors by controlling motivation-associated variables rather than conferring behavioral specificity.

Keywords: feeding behavior; motivation; need states; reward; two-photon calcium imaging.

Fig. 1.

c-Fos labeling identifies NAc neurons are activated by both food and water in a state-dependent manner. (A) Quantification of c-Fos protein expression in fed, fasted, and refed animals (above images, n = 6 mice per group for refeeding assay, one-way ANOVA, with Tukey’s multiple comparisons). (B) Quantification of c-Fos expression in hydrated, dehydrated and rehydrated animals (bottom images, n = 5 mice per group for rehydration assay, one-way ANOVA, with Tukey’s multiple comparisons). (C–F) Coexpression of D1 (Drd1) and D2 (Drd2) receptor mRNA with c-Fos mRNA in NAc after refeeding and rehydration assay, caged food, and caged water gel assay (Left). Representative images (Right) showing colocalization of neurons expressing c-Fos (green), Drd1 (red) and Drd2 (white); DAPI, blue. (Scale bar, right images: 100 μm.) (C and D) Quantification of percentages of colocalization between c-Fos, Drd1, and Drd2 mRNA-expressing neurons (n = 5 sections from three mice for both refeeding and rehydration assays, one-way ANOVA, with Tukey’s multiple comparisons). (E and F) Quantification of percentages of colocalization between c-Fos-, Drd1-, and Drd2-expressing neurons (n = 5 sections from three mice for caged food assay, n = 3 sections from three mice for caged water gel assay, one-way ANOVA, with Tukey’s multiple comparisons). All error bars represent mean ± SEM. NS, not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 2.

Sensory cues activate highly overlapping sets of NAcc^D1 and NAcc^D2 neurons. (A and B) Percentages of NAcc^D1 and NAcc^D2 neurons responding to the presence of food or water in the hungry or thirsty state (n = 5 D1:GCaMP6s and n = 5 D2:GCaMP6s mice). (C) Overlap percentages of neurons activated by the presence of food compared to neurons activated by the presence of the water gel. (D) Average NAcc^D1 and NAcc^D2 neural responses by sensory stimuli from food in hungry mice and from water gel in thirsty mice. (E) Average PCC of NAcc^D1 and NAcc^D2 neural responses to the presence of food and water (n = 5 D1:GCaMP6s and n = 5 D2:GCaMP6s mice, PCCs were averaged across three trials per condition per mouse, one-way ANOVA, with Tukey’s multiple comparisons). (F) Population activity vectors in principal component space for D1-Cre (n = 693 neurons from five mice averaged across three trials) and D2-Cre (n = 900 neurons from five mice averaged across three trials) mice. AU, arbitrary unit. All error bars represent mean ± SEM. NS, not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 3.

Highly overlapping NAcc^D1 and NAcc^D2 neurons respond similarly to food vs. water during consummatory phase. (A) Heatmap of all D1 neuronal responses to food presentation, n = 693 neurons. (B) Heatmap of D1 neuronal responses to food consumption from one example mouse and averaged neural traces from k-means clustering, n = 120 neurons. (C) t-SNE map of D1 neuronal states labeled by clustering from the example mouse. (D) Percentage of D1 neurons activated by both vs. by food or water, n = 5 mice. (E) Heatmap of all D2 neuronal responses to water gel presentation, n = 900 neurons. (F) Heatmap of D2 neuronal responses to water consumption from one example mouse and averaged neural traces from k-means clustering, n = 109 neurons. (G) t-SNE map of D2 neuronal states labeled by clustering for visualization. (H) Percentage of D2 neurons activated by both vs. by food or water, n = 5 mice.

Fig. 4.

Modulation of NAcc^D1 and NAcc^D2 neurons has complementary effects on goal-seeking behaviors, irrespective of the specific need states. (A) Schematic of caged food assay with optogenetic apparatus. (B) Activating NAcc^D1 neurons significantly increases approaches to caged food chow (n = 9 mice per group, P = 0.0058, two-tailed Mann–Whitney U test). Silencing NAcc^D1 neurons in hungry mice decreases caged food visits and duration (n = 9 mice per group, two-tailed Mann–Whitney U test). Neither manipulation affects total locomotor activity. (C) Activating NAcc^D2 neurons significantly down-regulates caged-food seeking behaviors in ad libitum mice (n = 5 mice per group, two-tailed Mann–Whitney U test). Silencing NAcc^D2 neurons decreases time spent proximity to caged food in hungry mice (n = 7 mice per group, two-tailed Mann–Whitney U test). Activating D2 neurons also decreases total locomotor activity (D) Schematic of caged water gel assay with optogenetic apparatus. (E) Activating NAcc^D1 neurons significantly increases approaches to the caged water gel (n = 7 mice per group, two-tailed Mann–Whitney U test). Activating D2 neurons also decreased locomotor activity. Silencing NAcc^D1 neurons in thirsty mice decreases caged water gel visits and duration spent on caged water gel (n = 5, 8 for control, D1:Arch3.0 group; two-tailed Mann–Whitney U test). (F) Activating NAcc^D2 neurons down-regulates approach frequency to a caged water gel (n = 7 mice per group, two-tailed Mann–Whitney U test). Silencing NAcc^D2 neurons decreases time spent in proximity to a caged water gel in thirsty mice (n = 5 mice per group, two-tailed Mann–Whitney U test). Box plots show mean (+), median, quartiles (boxes), and range (whiskers). NS, not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 5.

Modulation of NAcc^D1 and NAcc^D2 neurons reciprocally regulates consumption, irrespective of the specific need states. (A) Schematic of food consumption assay with optogenetic apparatus. (B) Activating NAcc^D1 neurons in ad libitum mice does not affect food intake. However, there was a significant increase of food but not water consumption 20-min post-activation (n = 7 mice per group; blue box, laser on; two-way ANOVA, with Sidak’s multiple comparisons). Silencing NAcc^D1 neurons in hungry mice decreases food consumption (n = 7 mice per group; green box, laser on; two-tailed Student’s t tests). (C) Activating NAcc^D2 neurons decreases food intake in ad libitum mice (n = 5, 7 for control and D2:ChR2 group; blue box, laser on; two-way ANOVA, with Sidak’s multiple comparisons). Silencing NAcc^D2 neurons does not affect food intake in hungry mice (n = 8 mice per group; green box, laser on; two-tailed Student’s t tests). (D) Schematic of water consumption assay with optogenetic apparatus. (E) Activating NAcc^D1 neurons does not affect water consumption, during or after stimulation (n = 5 mice per group; blue box, laser on; two-way ANOVA, with Sidak’s multiple comparisons). Silencing NAcc^D1 neurons decreases water consumption in thirsty mice (n = 9 mice per group; green box, laser on; two-tailed Student’s t tests). (F) Activating NAcc^D2 neurons decreases water consumption (n = 7 mice per group; blue box, laser on; two-way ANOVA, with Sidak’s multiple comparisons), whereas silencing them doesn’t affect water consumption (n = 8 mice per group; green box, laser on; two-tailed Student’s t tests). All error bars represent mean ± SEM. NS, not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 6.

Chemogenetic modulation of NAcc^D1 and NAcc^D2 neurons coregulates food and water consumption. (A) Chemogenetic activation of D1 neurons does not regulate food or water consumption in a physiological ad libitum state. Both groups of animals receive CNO injections. (n = 7, five mice per group; two-way ANOVA, with Sidak’s multiple comparisons). (B) Chemogenetic activation of D2 neurons reduces food and water consumption in a physiological ad libitum state (n = 10 mice per group; two-way ANOVA, with Sidak’s multiple comparisons). Both groups of animals receive CNO injections. (C) Chemogenetic silencing of D1 neurons reduces food and water consumption in fasted animals (n = 7,10 mice per group; two-way ANOVA, with Sidak’s multiple comparisons). Both groups of animals receive CNO injections. (D) Chemogenetic silencing of D2 neurons does not affect food and water consumption in fasted animals (n = 5 mice per group; two-way ANOVA, with Sidak’s multiple comparisons). Both groups of animals receive CNO injections. (E) Chemogenetic silencing of D1 neurons reduces food and water consumption in water-deprived animals (n = 8, five mice per group; two-way ANOVA, with Sidak’s multiple comparisons). Both groups of animals receive CNO injections. (F) Chemogenetic silencing of D2 neurons does not affect food and water consumption in water-deprived animals (n = 5 mice per group; two-way ANOVA, with Sidak’s multiple comparisons). Both groups of animals receive CNO injections.