As reported previously⁶, 2C null mice do not express 5-HT_2CRs. In 2C/POMC mice, 5-HT_2CRs are re-expressed only in brain sites where POMC is expressed (the hypothalamus and brainstem). Further, using electrophysiological recordings, we found that mCPP depolarized 26% POMC neurons in WT mice (Fig. 1). This mCPP-induced depolarization was abolished in POMC neurons from 2C null mice, but restored in 2C/POMC mice (Fig. 1). Therefore, 5-HT_2CRs are selectively re-expressed in POMC neurons in 2C/POMC mice. It should be noted that POMC is also expressed by neurons in the nucleus of solitary tract (NTS)⁷. Thus we cannot rule out that the small number of POMC neurons in the NTS may contribute to the responses outlined below.

Since 2C null mice display late-onset obesity and hyperadiposity⁶, phenotypes that may interfere with glucose homeostasis, all studies were performed in young mice with matched body weight and/or body adiposity (Supplemental Table 1). We first assessed the insulin sensitivity using insulin tolerance tests (ITTs). While intraperitoneal injections of insulin induced robust decreases in blood glucose in chow-fed wildtype (WT) mice, 2C null littermates were significantly resistant to insulin (Fig. 2a). Notably, insulin sensitivity in 2C/POMC mice was rescued to WT level (Fig. 2a), indicating that 5-HT_2CRs expressed by POMC neurons were sufficient to normalize insulin resistance resulting from global 5-HT_2CR deficiency. Despite insulin resistance, 2C null mice showed normal glucose clearance rate in glucose tolerance tests (GTTs) (Fig. 2b), and normal glucose levels at fed and fasted conditions (Fig. 2c). The basal insulin levels at fed condition were slightly but not significantly higher in 2C null mice than in WT and 2C/POMC littermates (Figure 2d). Fasted insulin levels were comparable in these mice (Figure 2d). Similar phenotypes were observed in mice fed with high fat diet (HFD) (Supplemental Fig. 1a–d). Since 2C null mice had normal glucose clearance rate while being insulin resistant, islets in these mice may have undergone compensatory adaptations which allowed elevated insulin secretion in response to hyperglycemia. Supporting this hypothesis, we found that a glucose load (0.75g/kg, i.p.) induced a significantly elevated serum insulin level in 2C null mice than in WT mice, and this hyperinsulinemia was normalized in 2C/POMC mice (Fig. 2e). Consistently, we found that in vitro glucose-stimulated insulin secretion was significantly potentiated in 2C null islets when compared to islets isolated from WT and 2C/POMC mice (Supplemental Figure 1e). We cannot rule out the possibility that the increased islet sensitivity to glucose challenge may be the direct outcome of 5-HT_2CR deficiency. However, it is more likely that this islet phenotype in 2C null mice is a secondary adaptation which occurs to compensate for insulin resistance.

We next performed the hyperinsulinemic-euglycemic clamp studies to assess mechanisms underlying the insulin resistance found in ITT studies. We found the glucose infusion rate (GIR) needed to maintain eugylcemia was reduced in 2C null mice compared to WT littermates (Figure 2f). GIR in 2C/POMC mice was fully restored to WT levels (Fig. 2f). In addition, the basal hepatic glucose production (HGP) and glucose disposal were higher in 2C null versus WT mice (Fig. 2g–h). The expected insulin-mediated suppression of HGP and increase in glucose disposal were evident in WT mice (Fig. 2g–h). A comparable rise in insulin in 2C null mice during the clamp failed to fully suppress HGP whereas glucose disposal rates were comparable (Fig. 2g–h). Notably, insulin-mediated suppression of HGP was fully restored in 2C/POMC mice (Fig. 2g). Collectively, these results demonstrated that insulin resistance caused by global 5-HT_2CR deficiency was due to impaired hepatic insulin action and re-expression of 5-HT_2CRs only in POMC neurons was sufficient to rescue this impairment. These findings support the model that 5-HT_2CR-melanocortin circuits regulate the liver to control peripheral insulin sensitivity and glucose balance.

To further investigate whether 5-HT_2CRs expressed by POMC neurons are sufficient to mediate effects of 5-HT drugs on glucose homeostasis, we tested the effects of mCPP on GTTs and ITTs in DIO mice. Pre-treatments with mCPP (1.5 mg/kg, i.p.) in WT DIO mice significantly improved glucose tolerance and insulin sensitivity (Fig. 2i–j). In contrast, mCPP failed to produce these anti-diabetic effects in 2C null mice (Fig. 2i–j). Remarkably, mCPP-induced improvement in the GTTs and ITTs were restored in 2C/POMC mice (Fig. 2i–j). Therefore, our results support the model that 5-HT_2CRs expressed by POMC neurons are sufficient to mediate the anti-diabetic effects of 5-HT_2CR agonists.

Activation of 5-HT_2CRs in POMC neurons has been shown to stimulate firing activity of these neurons (see ^{reference 4} and Fig. 1). This presumably leads to increased release of neurotransmitters (e.g. α-MSH, glutamate, etc.), which may underlie the mechanisms by which 5-HT_2CR regulates glucose homeostasis. In addition, 5-HT_2CR agonists may also regulate production of neurotransmitters in POMC neurons. Supporting this notion, 5-HT_2CR agonists stimulate POMC expression². Consistent with this observation, we found that 2C null mice had significantly lower POMC expression in the ARC compared to their WT littermates, a phenotype that was restored in 2C/POMC mice (Supplemental Figure 1f).

It is important to note that our results do not exclude the possibility that redundant 5-HT_2CR populations may also mediate similar effects. For example, 5-HT_2CRs are found in other CNS regions implicated in the regulation of glucose homeostasis, including the paraventricular nucleus of the hypothalamus, the ventromedial hypothalamic nucleus, and other sites⁸. These regions all receive serotoninergic projections⁹. The physiological relevance of 5-HT_2CRs in these sites is yet to be characterized. However, our unique mouse model will allow us to directly address the importance of 5-HT_2CRs in any site in which specific expression of Cre-recombinase can be directed. In conclusion, we have provided evidence that 5-HT_2CRs expressed by POMC neurons are physiologically important in regulating hepatic glucose production and insulin sensitivity. Moreover, this 5-HT_2CR-melanocortin circuit is sufficient to mediate the anti-diabetic effects of 5-HT_2CR agonists.

1