Over the past several decades, multiple studies have established a role for central 5-HT in reducing food intake and promoting satiety. For example, inhibition of central 5-HT synthesis with the TPH inhibitor, para-chlorophenylalanine (pCPA) or chemical lesion of 5-HT neurons with 5,7-dihydroxytryptamine (5,7-DHT) is orexigenic (Breisch et al., 1976; Saller and Stricker, 1976). Conversely, increasing synaptic 5-HT bioavailability, either by facilitating vesicular release with fenfluramine or by inhibiting 5-HT reuptake with selective serotonin reuptake inhibitors (SSRIs), produces an anorexigenic effect (Simansky and Vaidya, 1990; Heisler et al., 1997; Heal et al., 1998; Silverstein-Metzler et al., 2016). In addition to reducing food intake, increased central 5-HT signaling also improves glucose homeostasis, as treatment with fenfluramine or meta-chlorophenylpiperazine (mCPP, a 5-HT_1B/2C receptor agonist) improves glucose tolerance and insulin sensitivity (Storlien et al., 1989; Zhou et al., 2007).

A growing body of literature has explored the central 5-HT receptors that mediate 5-HT’s effect on food intake and glucose homeostasis. Of these receptors, the 5-HT_2C (G_q/11 coupled) receptor has been the most studied. Htr2c^−/− mice develop late onset obesity with hyperphagia, and demonstrate a blunted anorectic response to fenfluramine and mCPP (Tecott et al., 1995; Nonogaki et al., 1998; Vickers et al., 1999; Xu et al., 2008). Moreover, these mice manifest hepatic insulin resistance, independent of body weight gain (Xu et al., 2008). Furthermore, loss of the 5-HT_2C receptor synergistically impairs glucose homeostasis in the diabetic ob/ob mouse model without exacerbating obesity (Wade et al., 2008). Of note, the excessive weight gain and increased risk of type 2 diabetes associated with atypical antipsychotic drugs (AATPs) may be due to antagonism of 5-HT_2C (Godlewska et al., 2009; Laika et al., 2010). Finally, the hyperphagia observed in Prader-Willi syndrome may be partially due to alterations in Htr2c mRNA splicing and editing that reduce the sensitivity of the 5-HT_2C receptor (Kishore and Stamm, 2006; Kawahara et al., 2008; Morabito et al., 2010; Garfield et al., 2016; Zhang et al., 2016).

Several recent studies have also shed light on the neural circuits that mediate central 5-HT’s effects on energy balance, most notably the central melanocortin system, which includes two reciprocal populations of melanocortin neurons within the arcuate nucleus of the hypothalamus (ARC), anorexigenic neurons expressing proopiomelanocortin (POMC) and orexigenic neurons expressing neuropeptide Y/Agouti related peptide (NPY/AgRP; Figure 1; Sohn et al., 2013). Melanocortin receptors (MC₃R and MC₄R) in downstream neurons such as the paraventricular nucleus (PVN) are activated by alpha-melanocyte stimulating hormone (α-MSH), a proteolytic product of POMC, and inhibited by AgRP to reciprocally regulate food intake and glucose homeostasis (Berglund et al., 2014; Garfield et al., 2015; Krashes et al., 2016). Approximately 25% of POMC neurons in the adult mouse brain functionally express 5-HT_2C receptors (Xu et al., 2010b; Sohn et al., 2011). 5-HT_2C receptor activation in POMC neurons both induces Pomc mRNA expression and increases POMC neuronal activity through activation of TRPC5 cation channels (Zhou et al., 2007; Lam et al., 2008; Xu et al., 2010b; Gao et al., 2017). Remarkably, re-expression of 5-HT_2C only in POMC neurons in an otherwise Htr2c^−/− mouse is sufficient to reverse the hyperphagia and liver insulin resistance characteristic of Htr2c deficiency (Xu et al., 2010a,b). Conversely, mice with a POMC neuron-specific deletion of Htr2c are hyperphagic, show a blunted anorectic response to fenfluramine or mCPP, and have impaired glucose homeostasis (Berglund et al., 2013). Together, these studies underscore a critical role for 5-HT_2C in POMC neurons to regulate food intake and hepatic glucose metabolism.

In addition to 5-HT_2C, the 5-HT_1B receptor has been implicated in body weight regulation. Agonists for 5-HT_1B suppress food intake (Lee and Simansky, 1997), whereas genetic ablation of Htr1b leads to increased food intake and body weight (Halford and Blundell, 1996; Lee et al., 1998; Lucas et al., 1998). Moreover, the anorectic effect of fenfluramine is blunted in Htr1b null mice (Lucas et al., 1998). Collectively, these findings suggest that Htr1b contributes to 5-HT’s effect on food intake and body weight. Activation of 5-HT_1B seemingly inhibits NPY/AgRP neurons to promote satiety, as administration of 5-HT or a 5-HT_1B agonist hyperpolarizes NPY/AgRP neurons and subsequently reduces inhibitory postsynaptic currents (IPSCs) on POMC neurons (Figure 1; Heisler et al., 2006). Thus, 5-HT appears to promote satiety by simultaneously activating anorexigenic POMC neurons and inhibiting orexigenic AgRP/NPY neurons, due to the expression of 5-HT_2C and 5-HT_1B, respectively (Heisler et al., 2006). As a result, the reciprocal functions of these two receptors synergize to promote the activation of downstream melanocortin receptor-expressing neurons. Indeed, it has been shown that intact MC₄R signaling is required for the anorectic effects of 5-HT compounds (Heisler et al., 2002, 2003, 2006; Lam et al., 2008; Xu et al., 2010b).

Less is known about the role of 5-HT₆ (G_s coupled) receptors in food intake and satiety. Most of the current evidence comes from pharmacological studies using 5-HT₆ receptor agonists and antagonists, as well as Htr6 siRNA experiments (Woolley et al., 2001; Dudek et al., 2015; Higgs et al., 2016). Unlike 5-HT_2C and 5-HT_1B receptors, where activation leads to hypophagia, antagonism of 5-HT₆ reduces food intake and promotes satiety (Heal et al., 2008; Dudek et al., 2015; Higgs et al., 2016; Oh et al., 2016). Consistent with this, Htr6^−/− mice are partially protected from diet-induced obesity due to reduced food intake (Frassetto et al., 2008). Although the metabolically relevant neuronal circuits modulated by 5-HT₆ receptors remain unclear, a recent study mapping 5-HT₆ antagonist-induced c-Fos activity suggests that the PVN is an important site of action (Garfield et al., 2014). A potential role for 5-HT₆ in energy expenditure and glucose homeostasis warrants further study.

Thermogenesis through mitochondrial uncoupling occurs primarily in brown adipose tissue (BAT) and in beige adipocytes, a type of thermogenic adipocyte that appears in subcutaneous white adipose tissue (WAT) during prolonged cold exposure (Cohen and Spiegelman, 2015). Adaptive thermogenesis has recently gained widespread attention as a potential therapy to combat obesity by increasing energy expenditure. Central 5-HT appears to promote thermogenesis, since pharmacological or genetic depletion of central 5-HT has been found to impair thermogenic adaptation to cold (Fuller et al., 1987; Alenina et al., 2009; Hodges et al., 2011; Cerpa et al., 2014; McGlashon et al., 2015). For instance, mice with central depletion of 5-HT have reduced adaptation to cold exposure, diminished thermogenic function of BAT, and decreased recruitment of beige adipocytes (Alenina et al., 2009; Hodges et al., 2011; Cerpa et al., 2014; McGlashon et al., 2015). Central 5-HT appears to increase BAT and beige adipocyte thermogenic function by modulating sympathetic outflow to these tissues. Transynaptic retrograde tracing from BAT synaptic terminals show that glutamatergic and 5-HT neurons of the rostral raphe pallidus synapse onto sympathetic fibers in the intermediolateral nucleus (IML) of the spinal cord (Figure 2; Bowker et al., 1981; Bamshad et al., 1999). Systemic or IML injections of 5-HT or fenfluramine increase sympathetic firing of these fibers, while IML injections of 5-HT₇ receptor antagonists decrease sympathetic tone (Arase et al., 1988; Madden and Morrison, 2006, 2010; Morrison, 2016). Collectively, these data suggest that inputs from central 5-HT pathways play a significant role in the adaptation to cold exposure through the sympathetic activation of thermogenic adipose tissue.

Pancreatic islets, or islets of Langerhans, contain hormone-secreting endocrine cells. The two major endocrine cells in pancreatic islets are the beta cells, which secrete insulin, and the alpha cells, which secrete glucagon. In the fed state, insulin inhibits endogenous glucose production and stimulates glucose uptake and conversion into glycogen and lipid. Conversely, glucagon acts to increase blood glucose levels during fasting by stimulating glycogenolysis and gluconeogenesis. Insulin also acts on the alpha cells to inhibit glucagon secretion (Cooperberg and Cryer, 2010).

5-HT is synthesized in pancreatic islets and co-secreted along with insulin, potentially acting as a local autocrine/paracrine signal (Ekholm et al., 1971; Lundquist et al., 1971; Gylfe, 1978; Richmond et al., 1996; Ohta et al., 2011; Almaca et al., 2016; Bennet et al., 2016). 5-HT appears to be required for normal insulin secretion, since the loss of 5-HT synthesis within beta cells impairs insulin secretion (Paulmann et al., 2009; Kim et al., 2015). Consistent with this, posttranslational modification with 5-HT (known as serotonylation) activates the small GTPases Rab3a and Rab27a in the insulin secretion pathway, leading to increased insulin exocytosis (Paulmann et al., 2009). In addition, activation of 5-HT_2B receptor in isolated pancreatic islets has been found to augment glucose-stimulated insulin secretion (Bennet et al., 2016). Notably, deletion of Tph1 in intestinal enterochromaffin cells in mice does not alter insulin secretion, which further supports the notion that locally synthesized 5-HT in the pancreas serves as an autocrine signal to support normal insulin secretion (Sumara et al., 2012). In addition to regulating insulin secretion, there is evidence that 5-HT also regulates glucagon secretion (Marco et al., 1977; Adeghate et al., 1999). For example, human islet beta cells reportedly release physiological levels of 5-HT in response to glucose (Almaca et al., 2016). The release of 5-HT significantly decreased glucagon secretion, and this effect was mediated by the 5-HT_1F (G_i/o coupled) receptor on alpha cells. Interestingly, a 5-HT_1F receptor agonist, LY344864, was able to inhibit glucagon release from human islets. Systemic administration of this drug reduced plasma glucagon levels in hypoglycemic mice, suggesting that this pathway is active physiologically.

Pancreatic 5-HT may also play an important role under conditions of metabolic stress. For example, it has been suggested that islet 5-HT functions during pregnancy to increase beta cell mass and glucose-stimulated insulin secretion (Kim et al., 2010; Ohara-Imaizumi et al., 2013). Very early in pregnancy, the islet cell expression of Tph1 and Tph2 increases, resulting in a 400-fold increase in 5-HT levels (Kim et al., 2010). This 5-HT may augment glucose-stimulated insulin secretion in beta cells through activation of the ligand-gated cation channel 5-HT₃ receptor, increasing beta cell depolarization in response to glucose. Moreover, 5-HT signaling through 5-HT_2B (G_q coupled) has been implicated in pregnancy-induced maternal beta cell proliferation, which is later reversed by upregulation of the 5-HT_1D receptor (G_i/o coupled) in the perinatal period (Kim et al., 2010). In addition to its adaptive role in pregnancy, pancreatic 5-HT also appears to influence the metabolic adaptation to conditions of high fat diet-induced obesity and insulin resistance. In mice fed a high fat diet, the beta cell-specific loss of either Tph1 or the Htr3a receptor subunit results in impaired glucose tolerance (Kim et al., 2015). In addition, there is a positive correlation between body mass index (BMI) and the number of 5-HT immunoreactive cells in the human pancreas, further supporting an role for pancreatic 5-HT signaling during metabolic stress (Almaca et al., 2016).

In the fed state, the liver converts excess glucose into glycogen and activates de novo lipid synthesis. Conversely, in the fasted state, the liver liberates glycogen stores, produces glucose through de novo synthesis (gluconeogenesis), and generates ketone bodies from fatty acids. Studies examining the effect of 5-HT on hepatic gluconeogenesis, glycogen storage, glucose uptake, and glycolysis have produced conflicting results, likely due to differences in routes of 5-HT administration and the discrepancy between in vivo and in vitro models (Zabala et al., 1992; Moore et al., 2004a,b; An et al., 2009; Watanabe et al., 2010; Tudhope et al., 2012; El-Merahbi et al., 2015). However, Sumara et al. recently used genetic models to elegantly show that circulating 5-HT synthesized in intestinal enterochromaffin cells can signal through the 5-HT_2B receptor in hepatocytes to stimulate liver gluconeogenesis and inhibit glucose uptake by the liver during the fasted state (Sumara et al., 2012). In addition, 5-HT has been shown to regulate hepatic bile acid turnover and lipid metabolism (Watanabe et al., 2010). 5-HT treatment in mice accelerated the turnover of bile acids (excretion by the gallbladder followed by reabsorption by the intestine) and increased the concentration of circulating bile acids. 5-HT treatment also decreased liver triglyceride levels and increased liver cholesterol levels. However, it remains unclear whether this reflects a physiological role of 5-HT in hepatic lipid metabolism. In addition, systemic deficiency or chemical inhibition of TPH1 in high fat diet-fed mice reduced hepatic steatosis, secondary to decreased body weight and increased adipose thermogenesis (Crane et al., 2015). Thus, a direct link between physiological levels of 5-HT and hepatic lipid metabolism is not as clear as the role of 5-HT in hepatic glucose metabolism. Nevertheless, these studies overall suggest that gut-derived 5-HT plays an important role in the liver’s control of metabolic homeostasis.

Adipose tissue is a complex organ with multiple depots. WAT stores excess energy as triglycerides and releases non-esterified fatty acids (NEFA) and glycerol through lipolysis during fasting. WAT also acts as a major endocrine organ by secreting key hormones, including leptin and adiponectin, which regulate systemic metabolic homeostasis. There is evidence that both gut-derived circulating 5-HT and adipocyte-derived 5-HT play important roles in adipose tissue function. It has been known for decades that 5-HT administration increases circulating NEFA and glycerol levels (Carlson et al., 1967; Sumara et al., 2012). Fasting increases circulating 5-HT levels, and intestine-specific deletion of Tph1 in mice blunts fasting-induced plasma glycerol and NEFA levels (Sumara et al., 2012). 5-HT has also been implicated in adipose tissue lipogenesis. Recently, the recruitment of beige adipocytes in subcutaneous WAT, also known as browning of WAT, has generated widespread interest as a potential target for treating obesity (Cohen and Spiegelman, 2015). Interestingly, pharmacological inhibition of 5-HT synthesis protects mice from high fat diet-induced obesity through decreased adipose tissue lipogenesis, increased browning in subcutaneous WAT, and increased BAT thermogenesis (Crane et al., 2015; Oh et al., 2015). Of note, it appears that central 5-HT increases adipose tissue thermogenesis, whereas peripheral 5-HT inhibits it, highlighting the distinct roles of central and peripheral 5-HT. Interestingly, 5-HT appears to be locally synthesized as an autocrine factor in adipocytes, since adipocyte-specific deletion of Tph1 results in a similar phenotype as the systemic loss of Tph1 (Oh et al., 2015). Collectively, this evidence suggests that peripheral 5-HT stimulates adipose tissue lipolysis during fasting, promotes lipogenesis in adipose tissue in response to high fat diet, and inhibits adaptive thermogenesis. Specific 5-HT receptors have been implicated in each of these functions. 5-HT modulation of lipolysis appears to be mediated by the 5-HT_2B receptor, since adipose deletion of Htr2b results in blunted fasting-induced lipolysis and nearly complete ablation of serotonin-induced lipolysis (Sumara et al., 2012). Treatment with a 5-HT_2A receptor antagonist blocks lipid accumulation in 3T3-L1 adipocytes (Oh et al., 2015). Finally, Htr3a knockout mice exhibit increased thermogenesis and reduced weight gain (Oh et al., 2015).

It is interesting to note the reported major source of 5-HT in each of the metabolic tissues discussed here: islet-derived 5-HT in the pancreas, gut-derived 5-HT in the liver, and a combination of gut-derived 5-HT and adipocyte-derived 5-HT in adipose tissue. These differences, combined with the large number of 5-HT receptors, highlight the complexity of the peripheral serotonin system and the need for continued work to better understand the role of serotonin in metabolism.