Eating for hunger or pleasure: a Serotonin Model

Obesity, resulting from an imbalance between energy intake and expenditure, represents a major health crisis to our society, due to its alarmingly high prevalence and comorbidities, including diabetes, cardiovascular diseases, cancer, and COVID-19. Better understanding the neurobiological mechanisms for feeding behavior is essential for developing rational strategies to combat obesity and related comorbidities.

Agouti-related peptide (AgRP) neurons located in the arcuate nucleus of the hypothalamus (ARH) have received perhaps the most attention as a master regulator of feeding behavior. It is well known that AgRP neurons are regulated by multiple hormones that reflect the body’s energy storage or nutritional state, e.g., leptin, insulin, ghrelin, and asprosin (Yang and Xu, 2020). AgRP neurons are activated in a calorie-deficient state (Takahashi and Cone, 2005; Yang et al., 2011; Liu et al., 2012), and activations of AgRP neurons can drive eating (Aponte et al., 2011; Krashes et al., 2011). Together, these findings support a physiological feedback pathway that regulates feeding: a calorie-deficient state (e.g. hunger) activates AgRP neurons, which in turn drive eating. However, this ‘AgRP model’ faces a challenge, as recent in vivo recordings revealed that AgRP neurons decrease their activities dramatically within a few seconds after feeding starts, or even without the food actually being consumed (Betley et al., 2015; Chen et al., 2015; Mandelblat-Cerf et al., 2015). This rapid diminishment of AgRP neuron activity (Figure 1A) raises a question regarding how feeding behavior, which usually lasts for minutes to hours, is sustained. Based on our observations reported in a recent Molecular Psychiatry article (He et al., 2021), we propose an alternative ‘Serotonin Model’, which provides physiological feedback signals for feeding control. The brain serotonin, a neurotransmitter also known as 5-hydroxytryptamine (5-HT), is mainly synthesized by neurons in the midbrain dorsal Raphe nucleus (DRN). We demonstrated that the activation of these 5-HT^DRN neurons can inhibit eating (He et al., 2021). Importantly, using the in vivo recordings, we found that 5-HT^DRN neurons gradually increase their activities during the 2-h feeding period (Figure 1A). In sharp contrast to the rapid and sustained inhibition of AgRP neurons, the activation of 5-HT^DRN neurons occurs in a gradual and slow fashion (He et al., 2021). Importantly, the level of 5-HT^DRN neuron activity is correlated to the quantity of food intake (He et al., 2021). Thus, we suggest that 5-HT^DRN neurons function as a key component of a negative feedback loop. Low 5-HT^DRN neuron activity permits animals to eat; as animals continue eating, 5-HT^DRN neurons slowly elevate their activities to eventually terminate the meal.

Figure 1.

The ‘Serotonin Model’ illustrates physiological feedback signals to regulate both hunger-driven feeding and non-hunger-driven feeding. (A) A schematic illustration of changes in activities of AgRP neurons or 5-HT^DRN neurons during feeding. (B) A subgroup of 5-HT^DRN neurons project to the ARH, inhibiting AgRP neurons via the 5-HT_1BR and activating POMC neurons via the 5-HT_2CR, to suppress hunger-driven feeding; another subgroup of 5-HT^DRN neurons project to and inhibit DA^VTA neurons via the 5-HT_2CR to suppress non-hunger-driven feeding. The GABA_A receptor and the SK3 potassium channel mediate changes in activities of the ARH-projecting and VTA-projecting 5-HT^DRN neurons, respectively, during feeding. 5-HT_1BR, 5-HT 1B receptor; 5-HT_2CR, 5-HT 2C receptor; DA, dopamine; DRN, dorsal Raphe nucleus; GABA, γ-aminobutyric acid; POMC, proopiomelanocortin; SK3, small conductance calcium-activated potassium channel 3.

Feeding can be driven by hunger (a state of nutritional deficit) to ensure survival. Feeding can also be triggered by the hedonic properties of certain foods in the absence of nutritional deficit. Dysregulations of hunger-driven feeding and hedonic feeding both contribute to the development of obesity (Kenny, 2011; Alonso-Alonso et al., 2015). It has been suggested that neurocircuits controlling these two types of feeding behaviors are not completely dissociable (Rossi and Stuber, 2018). Consistent with this notion, we found that 5-HT^DRN neurons can regulate a hunger-driven feeding and a non-hunger-driven feeding in animals (He et al., 2021). However, our study further illustrated two largely segregated subgroups of 5-HT^DRN neurons: one subgroup send projections to the ARH and specifically inhibit feeding behavior driven by hunger, while the other subgroup of 5-HT^DRN neurons project to the ventral tegmental area and reduce the intake of a high palatable diet in the non-hungry state (Figure 1B). Interestingly, these two subgroups of 5-HT^DRN neurons both display slow activation during the course of hunger-driven feeding and non-hunger-driven feeding, respectively; however, they use distinct ion channels to achieve these changes (Figure 1B).

In summary, our findings support a ‘Serotonin Model’ that provides physiological feedback signals to regulate both hunger-driven feeding and non-hunger-driven feeding. We further identified distinct 5-HT^DRN-originated neurocircuits and disparate ion channels that can regulate these two types of feeding behaviors. These results provide a necessary framework for the development of a precision medication approach to treat obesity resulting from overeating for hunger or for pleasure.

[We acknowledge Dr Wei Wang (Zhongkai University of Agriculture and Engineering) for the illustration in Figure 1B. The investigators were supported by grants from the National Institutes of Health (NIH; P01DK113954, R01DK115761, R01DK117281, R01DK125480, and R01DK120858 to Y.X.; P20GM135002 to Y.H.), USDA/CRIS (51000-064-01S to Y.X.), and the American Diabetes Association (1-17-PDF-138 to Y.H.).]