Inhibitory Interplay between Orexin Neurons and Eating

Abstract

In humans and rodents, loss of brain orexin/hypocretin (OH) neurons causes pathological sleepiness [1-4], whereas OH hyperactivity is associated with stress and anxiety [5-10]. OH cell control is thus of considerable interest. OH cells are activated by fasting [11, 12] and proposed to stimulate eating [13]. However, OH cells are also activated by diverse feeding-unrelated stressors [14-17] and stimulate locomotion and "fight-or-flight" responses [18-20]. Such OH-mediated behaviors presumably preclude concurrent eating, and loss of OH cells produces obesity, suggesting that OH cells facilitate net energy expenditure rather than energy intake [2, 21-23]. The relationship between OH cells and eating, therefore, remains unclear. Here we investigated this issue at the level of natural physiological activity of OH cells. First, we monitored eating-associated dynamics of OH cells using fiber photometry in free-feeding mice. OH cell activity decreased within milliseconds after eating onset, and remained in a down state during eating. This OH inactivation occurred with foods of diverse tastes and textures, as well as with calorie-free "food," in both fed and fasted mice, suggesting that it is driven by the act of eating itself. Second, we probed the implications of natural OH cell signals for eating and weight in a new conditional OH cell-knockout model. Complete OH cell inactivation in adult brain induced a hitherto unrecognized overeating phenotype and caused overweight that was preventable by mild dieting. These results support an inhibitory interplay between OH signals and eating, and demonstrate that OH cell activity is rapidly controllable, across nutritional states, by voluntary action.

Figure 1

Impact of Eating on Natural OH Cell Dynamics In Vivo (A) A hypothesis for temporal modulation of OH cells during eating. (B) Left: targeting scheme of GCaMP6s to OH cells for obtaining the data shown in this figure (data using alternative targeting of OH cells are shown in Figures S2C–S2F). Right: localization of injection site and path of the optical fiber. 3V, third ventricle; L, D, M, V, lateral, dorsal, medial, ventral; VMH, ventromedial hypothalamus; Arc, arcuate nucleus. Representative image of n = 5 brains. (C) Left: recording scheme. Right: fluorescence trace during cage exploration for mice expressing GCaMP6s or eGFP in OH neurons. Typical examples of n = 5 and n = 3 mice, respectively. (D) Fluorescence trace during introduction of food into the cage and its subsequent consumption (orange-shaded area). Food was a drop of strawberry milkshake. Typical example of n = 5 mice. (E) Left: fluorescence trace during repeated bouts of food contact (orange-shaded areas; food is strawberry milkshake). Typical example of n = 5 mice. Right: quantification of fluorescence change during the first 2 s of consecutive food-contact bouts (means ± SEM, n = 3 mice). (F) Fluorescence change during food licking detected with a touch sensor (food is strawberry milkshake). Typical example of n = 5 mice across eight foods shown in (H), right. (G) Top: probability density of OH cell activity. Bottom: distribution of the bootstrap differences of the same data. Typical example of n = 3 mice. (H) Left: peri-event plots aligned to the onset of licking bouts (dashed line). The heatmap shows individual bouts (two per mouse), and the trace below the heatmap shows the mean of trial averages from each mouse (red line; gray lines represent SEM); n = 5 mice. Right: quantification of the experiment shown on the left, for different foods. Each column shows fluorescence change during the first 4 s of a licking bout (mean signals during 3.5 to 4 s minus signal during −0.5 to 0 s, times relative to the first lick). Data are means ± SEM of n = 4 mice in each group. Left column is control (OH-eGFP mice); other columns are OH-GCaMP6s mice; for food abbreviations, see the Supplemental Experimental Procedures; fast, overnight fasted before the experiment; fed, ad libitum feeding before the experiment. All changes in OH-GCaMP6s mice were significant (p < 0.05 in one-sample t tests of response to each food, DF = 3, t > 3.4). See also Figures S1–S3 and Movie S1.

Figure 2

Impact of Natural OH Cell Activity on Spontaneous Feeding Rhythm (A) Strategy for destroying OH neurons in adult mice. (B) Immunostaining for OH (green) and melanin-concentrating hormone-containing (MCH) (red) neurons in DTR⁻ (top) and DTR⁺ (bottom) littermates 10 days after DT injection. The DT injection led to the loss of OH neurons in all brains tested (n = 8). Scale bars represent 500 μm (left) and 100 μm (right). Dashed boxes in the left-hand panels indicate the areas shown in corresponding right-hand panels. (C) Time course of OH cell loss after DT injection in DTR⁺ and their DTR⁻ littermates (n = 5 mice in each group). (D) Quantification of OH cell number >21 days after DT injection in DTR⁻ and DTR⁺ littermates. Unpaired t test, t(13.3) = 16.41, p = 3.2e-10, n = 8 mice in each group. (E) Body weight time series of DTR⁻ and DTR⁺ littermates after DT injection. ANCOVA, F(1, 12) = 12.07, p = 0.005, n = 7 mice in each group. (F) Daily rhythm of eating in DTR⁻ and DTR⁺ mice after DT injection, across days. n = 7 mice in each group. (G) Mean daily rhythm of eating (average of 14 days; gray box is lights off) in DT-injected DTR⁻ and DTR⁺ mice. Repeated-measures ANOVA, interaction: F(7, 84) = 2.38, p = 0.029. Significant differences were found only at the time of day indicated (^∗∗∗p < 0.001, Holm correction for multiple comparisons). n = 7 mice in each group. (H) Total food consumed after DT injection, relative to the time of day. Repeated-measures ANOVA, interaction: F(7, 84) = 3.07, p = 0.006. Pairwise comparisons revealed statistical differences at the time of day indicated (^∗∗∗p < 0.001, Holm correction for multiple comparisons). n = 7 animals in each group. See also Figure S4.

Figure 3

Impact of Natural OH Cell Activity on Rebound Eating after Fasting (A) Eating responses to a 1-day fast in control mice (OH⁺, DTR⁻ mice injected with DT) and their OH cell-deficient littermates (OH⁻, DTR⁺ mice injected with DT). n = 6 mice in each group. (B) Daily eating rhythms before and after a 1-day fast in OH⁺ and OH⁻ littermates. n = 6 mice in each group. (C) Daily eating rhythms 3 days before and after food restriction. Repeated-measures ANOVA, interaction: F(7, 35) = 3.36, p = 0.008 (left) and F(7, 35) = 3.53, p = 0.006 (right). Follow-up tests showed significant differences only at times marked with asterisks (^∗∗p < 0.01, ^∗∗∗p < 0.001, Holm correction for multiple comparisons). n = 6 mice in each group. (D) Total food consumed 3 days before (fed) and 3 days after (fast) food restriction in OH⁺ and OH⁻ mice (n = 6 in each group) during early night (ZT12–14) compared to that consumed during late night (ZT18–20). Paired t tests.

Figure 4

Prevention of Weight Gain Caused by OH Cell Loss by Dieting (A) Strategy for pair-feeding experiment (OH⁺, DTR⁻ mice injected with DT; OH⁻, their DTR⁺ littermates injected with DT). (B) Weight gain of OH⁻ and OH⁺ littermates during weeks 2 and 3 after DT injection, and during free feeding (unpaired t test, t(11.93) = −4.327, p = 0.0009, n = 7 mice in each group) and pair feeding (unpaired t test, t(5.41) = −0.598, p = 0.574, n = 4 animals in each group).