II. Mechanisms of direct detection of nutrients and hormones by the brain

1. Glucose sensing

The brain plays a pivotal role in controlling metabolic homeostasis, including the control of blood glucose. Brain glucose sensing is a key determinant of the counterregulatory responses to hypoglycemia. Both the hypothalamus and the brain stem, areas where the blood-brain barrier is limited, play a crucial role in the response to hypoglycemia. Two populations of glucose-sensitive neurons, either activated (GE) or inhibited (GI) by glucose, have been identified in an increasing number of brain areas. The PVN and ARC nuclei of the hypothalamus, like the NTS and area postrema of the brain stem, possess both GE and GI neurons. GI neurons are particularly present in the LH nucleus and GE in the VMN. The close metabolic coupling between neurons and surrounding glial cells might also be a mechanism involved in glucose sensing. In that case, neurons would be responding to lactate produced by the catabolism of glucose in glial cells rather than to glucose per se.
Glucose sensing depends on mechanisms involving either glucose detection or glucose metabolism. Glucose detection by taste receptors or the sodium-coupled glucose co-transporter SGLT3 triggers electrogenic signals as in intestinal cells. Binding of glucose to SGLT3 promotes Na+ entry and subsequent membrane depolarization (Figure 3A).10 Taste receptor activation leads to the release of Ca2+ from the endoplasmic reticulum where the Ca2+ activates the cation channel TRPM5 (transient receptor potential member 5) leading to membrane depolarization (Figure 3B).11 (Mark Evans, Bernard Thorens, Lectures).
Glucose-sensing mechanisms based on cell metabolism are close to those used by pancreatic β-cells. In β-cells, the low-affinity transporter Glut2 and glucokinase (GK) play a key role in glucose sensing. In the canonical model, glucose catabolism induces the closure of ATP-gated potassium (KATP) channels, allowing membrane depolarization and entry of Ca2+, which triggers insulin release (Figure 3C).12
In the brain, Glut2 has a low level of expression but is dispersed in many regions with sometimes long projections to neighboring structures. In the hypothalamus, Glut2 is necessary for the regulation of the melanocortin pathway by glucose, but Glut2 protein is present in neither POMC nor NPY neurons of the ARC nucleus. However, these neurons are in close contact with numerous Glut2-positive nerve terminals, whose cell bodies have not yet been identified (Bernard Thorens, Lecture). GK also has a low level of expression in the brain, but is expressed in 20% to 30% of POMCGE, NPY-GI, and orexin-GI neurons. Ex-vivo or in-vivo inhibition of GK in the VMN, either with inhibitors or shRNA, largely abolishes GE and GI glucose-sensing activities, whereas GK activation potentiates both GE and GI responses. Outside the hypothalamus, GK-dependent glucose sensing was also identified in the medial amygdala nucleus connected to basomedial hypothalamus. These neurons seem to exert a control loop to increase or decrease counter-regulatory responses to hypoglycemia.
One-third of GK-expressing cells are glial cells, demonstrating the involvement of direct and indirect sensing mechanisms (Mark Evans, Lecture).
Metabolism-dependent glucose sensing also involves AMPK, which is widely expressed in the GE and GI neurons of the brain, and KATP channels, more articularly in the VMN. Glucose sensing in orexin GI neurons of the LHA involves K+ leak channels, but independently of cell metabolism. The response of GI neurons of the VMN depends on metabolism via GK and AMPK, but activates the closure of chloride channels, possibly the CFTR. These examples illustrate the diversity and the complexity of glucose-sensing systems in the brain (Figure 4) (Bernard Thorens, Lecture).
In the NTS, Glut2 is expressed in a small population of GI neurons. In these cells, glucose sensing involves metabolic processes dependent on GK, AMPK, and K+ leak channels.
Glut2 neurons of the NTS are GABAergic and some project to the DMN of the vagus. Specific activation of these neurons induces an increase in vagal nerve firing followed by an increase in glucagon secretion (Bernard Thorens, Lecture). There is no compelling evidence for an effect of brain GK on appetite and energy balance. However, inhibition of GK in the third ventricle stimulates feeding responses to glucoprivation via the activation of NPY and orexigenic cells. As mentioned above, brain Glut2 is involved in the stimulation of food intake induced by glucose via an indirect regulation of the melanocortin pathway. Interestingly, glucose sensing in the hypothalamus controls the early phase of insulin release by a mechanism depending on GK (Mark Evans, Bernard Thorens, Lectures).

2. Fatty acid sensing

Fatty acid−sensitive neurons are present in the brain, especially in the hypothalamus, and participate in the control of energy homeostasis. Infusion of fatty acids in the hypothalamus decreases food intake, sympathetic activity, and hepatic glucose production and is associated with exaggerated glucose-stimulated insulin secretion. The brain can detect free fatty acids crossing the blood-brain barrier, but also free fatty acids delivered locally from the hydrolysis of triglycerides. Inactivation of lipoprotein lipase in the hypothalamus induces an increase in body weight through a ceramidedependent pathway.
As with glucose sensing, fatty acid sensing involves pathways dependent on and independent of cellular metabolism. Neurons of the VMN express fatty acid transporters and the main enzymes involved in fatty acid metabolism (such as long-chain acyl-CoA synthase-ACS, carnitine palmitoyltransferase-CPT1, and uncoupling protein 2-UCP2). Fatty acid catabolism is associated with an increase in the acyl-CoA intracellular pool, which is considered as the “final” satiety signal rather than fatty acids themselves. Fatty acids are indeed detected via the production of ATP through β-oxidation.
The central inhibition of CPT-1 (rate-limiting step in mitochondrial β-oxidation) mimics the effect of fatty acids on energy homeostasis.
Malonyl-CoA is a potent inhibitor of CPT-1 and is produced by fatty acids via β-oxidation or by glucose via glycolysis. An increase in malonyl-CoA and acyl-CoA levels is associated with reduced food intake, suggesting that malonyl-CoA may be the metabolic sensor of energy levels in the hypothalamus. Since a selected population of fatty acid−sensitive neurons also sense glucose, the responses to fatty acids is dependent upon the metabolic state of the animal. In addition, since so much fatty acid oxidation takes place in the astrocytes, the latter probably contribute to brain fatty acid sensing. Astrocytes might have their primary effects by the production of ketone bodies, which are further utilized by neurons to alter their fatty acid and glucose sensing (Figure 5).
The hypothalamus contains two populations of fatty acid−sensitive neurons: one excited by oleic acid and another inhibited by oleic acid. Neurons of the ARC nucleus detect fatty acids mainly by mechanisms dependent on cell metabolism. The excitatory effect of oleic acid is due to the closure of chloride channels leading to membrane depolarization and increased action potential frequency. The inhibitory effect of oleic acid involves KATP channels. The β-oxidation of oleic acid induces an increase in POMC and a decrease in NPY expression in the ARC nucleus. In VMN neurons, cell metabolism accounts for a relatively small percentage of fatty acid sensing. Fatty acid sensing is almost completely abolished by inhibition of the fatty acid transporter CD36. In most fatty acid−sensitive neurons, CD36 may act primarily as a receptor for long-chain fatty acid−activating store-operated calcium channels to alter membrane potential (Figure 5). Reduction of CD36 in VMN neurons decreases expression of AgRP and POMC. Finally, some fatty acids, such as palmitic acid, alter cell signaling molecules by altering their function (covalent attachment) or their location in the membrane.

3. Brain glucose sensing and glucose effectiveness

The regulation of hepatic glucose uptake by insulin through an indirect mechanism involving the brain is still a source of controversy. However, several findings point to a role of the brain in the decrease of hepatic glucose uptake by insulin. For example, the inactivation of the insulin receptor in the hypothalamus by antisense oligonucleotides impairs the suppression of HGP by insulin. The mechanism involves KATP channels, which are targeted by the IRS-PI3K pathway in some neurons and which induce the activation of efferent vagal fibers. In the DVC, some neurons also respond to insulin, but the mechanism involves the ERK pathway. Moreover, in streptozotocin-induced diabetic mice, the capacity of insulin to decrease blood glucose levels is attenuated by a central PI3K inhibitor. The collective data suggest that both the direct and the indirect pathways are sufficient to control HGP and that when one does not function properly, the other can compensate. Insulin might regulate HGP indirectly via a FoxO1-independent mechanism, which is blocked by excessive FoxO1 signaling (Michael Schwartz, Lecture).
Secretion of FGF15/19 by enterocytes is induced by bile acids via FXR signaling. FGF19 improves glucose tolerance in diet-induced obesity mice, while inactivation of FGF19 impairs glucose tolerance. FGF19 signaling might depend on FGFR1, which is expressed in the brown and white adipose tissue and in the mediobasal hypothalamic areas. Infusion of FGF19 in the brain of diet-induced obese mice or ob/ ob mice improves glucose tolerance. The systemic effect of FGF19 is blunted by 50% with the infusion of an FGFR antagonist in the brain, demonstrating the key role of the central nervous system. The beneficial effect of FGF19 depends on an increase in glucose effectiveness, or the ability of glucose to stimulate its own disposal independently of insulin. Glucose effectiveness contributes at least as much as insulin to normal glucose tolerance and is altered in type 2 diabetes and in ob/ob mice.

“Neural Orchestration of Metabolism and Islet function”
I. General points on the central control of energy balance and food intake
II. Mechanisms of direct detection of nutrients and hormones by the brain
III. Gastrointestinal and vagal detection of nutrients
IV. Control of β-cell function by the brain
V. Conclusion
References
Lectures during IGIS meeting