Glucosensing is a property of specialized neurons in the brain that regulate their membrane potential and firing rate as a function of ambient glucose levels. These neurons have several similarities to β- and α-cells in the pancreas, which are also responsive to ambient glucose levels. Many use glucokinase as a rate-limiting step in the production of ATP and its effects on membrane potential and ion channel function to sense glucose. Glucosensing neurons are organized in an interconnected distributed network throughout the brain that also receives afferent neural input from glucosensors in the liver, carotid body, and small intestines. In addition to glucose, glucosensing neurons can use other metabolic substrates, hormones, and peptides to regulate their firing rate. Consequently, the output of these “metabolic sensing” neurons represents their integrated response to all of these simultaneous inputs. The efferents of these neurons regulate feeding, neuroendocrine and autonomic function, and thereby energy expenditure and storage. Thus, glucosensing neurons play a critical role in the regulation of energy homeostasis. Defects in the ability to sense glucose and regulatory hormones like leptin and insulin may underlie the predisposition of some individuals to develop diet-induced obesity.
The term “energy homeostasis” is a modification of the second law of thermodynamics whereby the amount of energy taken in as food equals the amount expended as heat (thermogenesis). When intake exceeds expenditure, the excess is stored primarily as glycogen and fat, and these stores are used to supply fuel when food is in short supply. This process is regulated over different time frames and by a variety of physiological and metabolic systems; dysregulation of either intake or expenditure can lead to obesity. Early studies of damage to the hypothalamus pointed to the brain as the primary regulator of energy homeostasis. Lesions of the ventromedial hypothalamus (VMH) produce increased food intake (hyperphagia), obesity (1), and defective autonomic function in organs involved in the regulation of energy expenditure (2,3). On the other hand, electrical stimulation of the VMH leads to generalized sympathoadrenal activation (4) with increased activity in thermogenic tissues (5). Lesions of the lateral hypothalamic area (LHA) reduce food intake and increase sympathetic activity and eventually establish a new lower defended body weight (3,5,6). Whereas such early studies pointed to the hypothalamus as the central controller of energy homeostasis, later studies suggested that energy homeostasis is controlled by a distributed network of specialized neurons that use glucose, as well as a variety of metabolic substrates and hormones, to regulate their membrane potential and firing rate (7–14) (Fig. 1).
These “glucosensing” neurons are localized in a variety of brain sites that are involved in the regulation of energy homeostasis. These central neurons are part of a larger network of glucosensors that are located in peripheral organs. Such peripheral glucosensors are located in the hepatic portal vein (15), carotid body (16), and the gut (17). Their vagal and sympathetic neural afferents terminate predominantly in the nucleus tractus solitarius (NTS) in the medulla (Fig. 1). The neurons in the NTS represent a critical nodal point where hardwired inputs from metabolic, hormone, and peptide signals from the periphery converge and are integrated. Because many NTS neurons are also glucosensing neurons, this allows them to summate the direct effects of glucose, other metabolic substrates, and hormones such as leptin and insulin at the level of their membrane potential with those arriving via neural afferents from peripheral glucosensors (18). NTS neurons project widely to other brainstem and forebrain nuclei such as the rostral and caudal ventrolateral medulla and raphe pallidus and obscurus (RPa/Ob); the hypothalamic paraventricular nucleus (PVN), arcuate nucleus (ARC), ventromedial nucleus (VMN), and dorsomedial nucleus; and LHA, the substantia nigra, and ventral tegmental and central nucleus of the amygdala, most of which contain glucosensing neurons and are also involved in autonomic function and energy homeostasis (Figs. 1 and 2) (19–22).
Because of historical precedents, a majority of early studies focused on hypothalamic neurons as regulators of energy homeostasis. This has led to the realization that manipulations of the VMH most often affected function in both the VMN and ARC (Fig. 2). In fact, it is the ARC that may be the more important of these two nuclei, since it contains two sets of neurons whose primary function appears to be the regulation of energy homeostasis. Medial ARC neurons that express neuropeptide Y (NPY) are classified as anabolic because release of this peptide onto target neurons in the PVN and LHA potently stimulates food intake and inhibits energy expenditure by decreasing sympathetic activity in thermogenic organs. Laterally placed ARC neurons produce proopiomelanocortin (POMC), which is a pro-hormone for α-melanocyte–stimulating hormone. This catabolic peptide interacts with melanocortin 3 and 4 receptors (MC3/4-R) to inhibit intake and stimulate sympathetic activity and thermogenesis by acting on some of the same PVN and LHA neurons that are NPY targets. Some ARC POMC neurons also project to the intermedio-lateral cell column of the spinal cord, which positions them to modulate the activity of sympathetic preganglionic neurons (23). ARC NPY neurons also produce and co-release agouti-related peptide, a unique peptide that acts as a functional antagonist of MC3/4-R. Thus, activation of ARC NPY neurons releases a strong anabolic peptide and an equally potent inhibitor of catabolic pathways. Importantly, ARC NPY and POMC neurons are prototypic metabolic sensing neurons; NPY neurons are inhibited and POMC neurons are excited by glucose, while both leptin and insulin inhibit NPY and stimulate POMC gene transcription. Insulin and leptin also have acute effects on NPY and POMC neuronal activity (19).
Two other groups of metabolic sensing neurons in the LHA produce the anabolic peptides orexin (hypocretin) and melanocyte concentrating hormones. As opposed to ARC NPY and POMC neurons, these LHA neurons are involved in a much wider spectrum of metabolic, physiological, and behavioral processes (24,25) (Fig. 2). Orexin neurons are inhibited (glucose-inhibited [GI] neurons) and melanocyte concentrating hormone neurons are excited (glucose-excited [GE] neurons) by glucose (26), and both receive inputs from ARC NPY and POMC neurons (27). They also receive inputs from thalamic areas involved in nociception, limbic, and striatal areas involved in motivation and reward and hypothalamic and brainstem networks that coordinate foraging, gustatory, feeding, and defensive behaviors. In turn, melanocyte concentrating hormones and orexin LHA neurons project to brain areas involved in nonhomeostatic reward systems such as the nucleus accumbens, ventral tegmental area, prefrontal cortex, amygdale, and bed nucleus of the stria terminalis (which collectively are commonly referred to as the extended amygdala); neocortical and hippocampal areas involved in cognition and memory; striatal areas involved in motor activity; hindbrain areas involved in consciousness and arousal (locus coeruleus, dorsal raphe, central gray); and thalamic areas involved in stress responsivity (ventro-posterior thalamus). They also project to autonomic efferent areas in the central nucleus of the amygdala, PVN, NTS, and dorsal vagal complex (28) (Fig. 2). In addition to neurons in the ARC and LHA, other glucosensing neurons include VMN γ-aminobutyric acid neurons, substantia nigra and ventral tegmental area dopamine neurons, and others in the NTS, rostral and caudal ventrolateral medulla, and dorsal vagal complex cholinergic neurons (19,29–32). There are likely to be other collections of such neurons scattered in other brain sites that remain to be discovered.
HOW DO NEURONS SENSE GLUCOSE?
Whereas most neurons require glucose to fuel their metabolic activity, glucosensing neurons also use glucose in a concentration-dependent manner as a signaling molecule to regulate their membrane potential and neural activity (8,14,33,34). There are two broad categories of glucosensing neurons: GE neurons increase, while GI neurons decrease, their activity as ambient glucose levels rise. They reverse this pattern as glucose levels fall. GE neurons function similarly in many ways to pancreatic β-cells, whereas GI neurons have some similarities to α-cells (14,19,35–39). As with pancreatic β- and α-cells, we know a great deal about the way in which GE neurons sense glucose, but much less about the way in which GI neurons do this. As with most neurons, the majority of GE and GI neurons use the high-capacity high-affinity glucose transporter 3 to import glucose (37). While hexokinase I accounts for the majority of glucose phosphorylation in neurons, many glucosensing neurons also express the pancreatic form glucokinase (GK) (hexokinase IV), the rate-limiting step in β-cell glucosensing (36,38,40–42). The Km for GK activity in the brain, as in the pancreas, is ∼9–11 mmol/l (40,43). Because brain glucose levels are only ∼20% of those seen in blood under most conditions, this means that GK functions at the lower end of its response curve in glucosensing neurons. Also, because hexokinase I is the primary means of phosphorylating glucose in glucosensing neurons, we have postulated that GK may be compartmentalized beneath the plasma membrane together with mitochondria, close to the ion channels that would be affected by the ATP formed by GK-regulated oxidation of glucose (19). Approximately 65% of GE and 45% of GI neurons express GK mRNA (37). In those neurons, pharmacological inhibition of GK activity decreases activity in GE and increases activity in GI neurons at glucose concentrations comparable to those seen in the brain (2.5 mmol/l) under euglycemic conditions in the periphery (36,37). On the other hand, pharmacological activation of GK inhibits activity of GI and enhances activity of GE neurons held at glucose concentrations comparable to those seen in the brain (0.5 mmol/l) under hypoglycemic conditions (42). These data strongly suggest that GK is a critical regulator of neuronal activity in many GE and GI neurons.
In GE neurons, glucose metabolism increases the ratio of ATP to ADP. This causes ATP to bind to the ATP-sensitive K+ (KATP) channel composed of a Kir6.2 pore-forming unit for potassium and a sulfonylurea receptor (44). Binding of either ATP or sulfonylureas inactivates (closes) the channel and depolarizes the cell membrane. Depolarization is followed by influx of extracellular calcium through a voltage-dependent calcium channel (36,37) and is often associated with increased action potential frequency. Glucose-induced closure of the KATP channel at nerve terminals on some GE neurons can also release neurotransmitters independently of action potentials propagated from the cell body (45). Whereas the KATP channel appears to be the final common pathway involved in GE glucosensing, much less is known about GI neuronal glucosensing. A Cl− channel (35), the Na+-K+ ATP pump (46), and an ATP-responsive K+ channel (47) have all been proposed as possible final common pathways, but the actual effector of GI glucosensing is unknown. In addition to this uncertainty, it is also unclear how the GE and GI neurons that do not express GK regulate their ability to use glucose as a signaling molecule.
Glucosensing neurons do not function in isolation. They are surrounded by and provided with metabolic support by glia. Astrocytes readily take up and store transported glucose as glycogen, which is hydrolyzed to release lactate into the extracellular space. Extracellular lactate is taken up by neurons and converted to pyruvate, which is then oxidized in mitochondria to provide ATP (37,48). By this mechanism, lactate can reverse the inhibition of GE activity, which occurs at low ambient glucose levels (49). Glucosensing neurons are also responsive to fatty acids (9,50), ketone bodies (51), leptin (12,19), and insulin (13,14). As with lactate, the effects of these other metabolites and hormones on glucosensing neurons depend on ambient glucose levels. For example, both leptin and insulin inhibit neuronal activity in VMH GE neurons held at very high (10 mmol/l) glucose levels (12,13), but they either stimulate or have no effect on activity in these neurons held at glucose levels seen in the brain during euglycemic or hypoglycemic conditions (14,19). In addition, glucosensing neurons are, as expected, subject to modulatory transmitter and peptide inputs from surrounding neurons (14,52). Thus, the response of glucosensing neurons to changes in glucose levels highly depends on the neural, metabolic, and hormonal milieu in which they find themselves. Because of these multiple influences on their activity and because they are critical to the control of energy homeostasis, the term “metabolic sensing” is probably more apt than “glucosensing” to describe their function (19).
GLUCOSENSING NEURONS AND THE REGULATION OF ENERGY HOMEOSTASIS
Glucosensing and the control of feeding.
Mayer (53) proposed that “… the passage of potassium ions into glucoreceptor cells along with the glucose phosphate represents the point at which effective glucose levels are translated into an electric or neural mechanism… ” for the control of feeding. He suggested that these cells were located in the hypothalamus even though it would be another 11 years before such glucosensing neurons were first identified (7,8) and 35 years before the KATP channel would be recognized as the mechanisms by which GE neurons sense glucose (34). Despite Mayer’s “glucostatic hypothesis” for the control of food intake, we still do not know whether normal excursions of blood glucose that occur during the diurnal cycle play an important role in the regulation of normal feeding behavior. Unquestionably, severe glucoprivation can stimulate feeding in rodents (54) and feelings of hunger in humans (55), and high concentrations of glucose placed into the brain can terminate or reduce feeding (56–59). There is also no question that glucosensing neurons can respond to very small changes in ambient glucose concentrations (0.1–0.3 mmol/l) that are comparable to the very small decrements in blood glucose levels (0.5–2 mmol/l) that precede some meals in both humans and rodents (14,35,42,60–63). However, when blood glucose levels are lowered in a stepwise fashion, humans report hunger only at levels slightly higher than those associated with impaired cognitive function (64,65) (Fig. 3). In fact, no one has ever demonstrated that meal initiation or termination can be manipulated by altering brain glucose levels within the limits found during normal ingestive cycles.
Nevertheless, there does appear to be an intrinsic system within the brain (and hepatic portal system [66]) that responds to severe cellular glucopenia by stimulating feeding. Generalized inhibition of brain glucose metabolism by injection of 2-deoxyglucose into the lateral cerebral ventricle stimulates feeding (67). Although Mayer postulated that glucosensing neurons in the hypothalamus were important for feeding, direct unilateral injection of the potent 2-deoxyglucose analog, 5-thioglucose (5TG), into the VMH does not stimulate feeding (30). On the other hand, avid feeding results from small localized injections of 5TG into areas of the caudal ventrolateral medulla and dorsomedial medulla. The former site corresponds to the A1/C1 norepinephrine/epinephrine neurons (30), and destruction of the rostrally projecting catecholamine neurons from this site inhibits feeding in response to systemic injections of 2-deoxyglucose (68). The latter glucosensing site overlaps with serotonin neurons in the raphe pallidus and obscurus that also express GK (Fig. 4A) (30,69). Inhibition of GK activity in neurons in that area with focal injection of pharmacological doses of alloxan stimulates feeding comparably to injecting 5TG (Fig. 4B). This effect is completely inhibited by pretreatment with toxic doses of alloxan, which should selectively destroy GK-expressing neurons. However, such putative destruction of GK neurons does not block 5TG-induced food intake, suggesting that different sets of neurons mediate these feeding effects. While alloxan stimulates feeding when injected into the caudal dorsomedial medulla, similar bilateral injections of alloxan into the VMH have no effect on feeding (Fig. 4D). Despite the fact that neither unilateral 5TG nor alloxan injections into the VMH stimulate feeding, feeding provoked by 5TG injections into the hindbrain serotonin neurons is completely blocked when VMH GK mRNA is upregulated after third ventricular injections of toxic doses of alloxan (70). Such data demonstrate a complex interplay between hindbrain and hypothalamic sites involved in the regulation of glucoprivic feeding. Whereas “emergency” feeding can be stimulated by producing focal glucoprivation in select hindbrain glucosensing neurons, their projections to rostral hypothalamic sites are clearly required for full expression of this response.
It is important to point out that glucose does not necessarily act alone on central glucosensing neurons to alter feeding. First, infusions of glucose into the portal vein can reduce food intake (71). Second, the complex interaction between central glucosensing neurons and other metabolic substrates and hormones is illustrated by studies demonstrating the role of insulin and the VMH KATP channel activity in feeding. First, chronic (2-day) carotid infusions of glucose alone have little effect on feeding, independent of the caloric load infused, whereas addition of insulin to the infusate decreases intake out of proportion to the infused calories (72). Additionally, central infusions of oleic acid reduce food intake, and this effect is antagonized by pharmacologic inhibition of KATP channel activity in the VMH (73).
Glucosensing and the control of energy expenditure and glucose homeostasis.
It has been known for several years that infusions of glucose can both increase general sympathetic activity (as evidenced by increased plasma norepinephrine levels) (74,75) and produce an increase in thermogenesis, which is partly due to such sympathetic activation (76). While hyperinsulinemia has been evoked as a stimulant of sympathetic activity, it is clear that glucose can evoke this activity when given alone (in insulin-deficient animals) (75) and directly into the forebrain via the carotid artery without altering plasma insulin levels (77,78). Such forebrain infusions activate neurons in several hypothalamic areas known to contain glucosensing neurons such as those in the PVN that project directly to autonomic outflow areas of the medulla and spinal cord (79,80). Intracarotid glucose infusions also increase efferent vagal nerve activity in the pancreas (81). Although such activation should increase insulin secretion, absence of change in plasma insulin levels after such infusions is probably because of the concomitant sympathetic activation that inhibits insulin secretion (78,82). The thermogenic effects of glucose are also probably mediated by hypothalamic glucosensing neurons, since intracarotid and direct injections of glucose into the VMH and PVN produce increased activity in the sympathetic efferents to brown adipose tissue in the rat (83,84).
Again, the interaction of hormones with glucosensing neurons in the control of energy homeostasis is suggested by studies showing that third cerebral ventricular insulin infusions reduce hepatic glucose production, and this effect is inhibited by central administration of KATP channel inactivators (85) or mimicked by third ventricular infusion of a KATP channel activator (86). However, these results alone do not definitively implicate glucosensing neurons, since pharmacological manipulation of the KATP channel can alter neuronal activity even in GI and nonglucosensing neurons, neither of which normally use the KATP channel to sense glucose (36,37).
As with food intake, cellular glucopenia can stimulate a counterregulatory response associated with increased plasma norepinephrine, epinephrine, glucagon, and corticosteroid levels that helps mobilize blood glucose levels by promoting hepatic glycogenolysis and fatty acids by promoting lipolysis (65,87). The brain appears to be an important mediator of this effect, since clamping the brain at euglycemic levels greatly attenuates the counterregulatory response to systemic hypoglycemia (88). Unlike glucoprivic feeding, the VMH appears to play a major role in the counterregulatory response to glucoprivation. Bilateral injections of 2-deoxyglucose into the VMH mimic the effects of systemic hypoglycemia (89), whereas VMH infusions of glucose greatly attenuate the counterregulatory response to systemic hypoglycemia (90). Also, inhibition of KATP channel activity in the hypothalamus attenuates (91), whereas enhancing channel activity selectively in the VMH increases, the counterregulatory response to hypoglycemia (92). Even a single bout of hypoglycemia leads to blunting of subsequent counterregulatory response to hypoglycemia (93). GK may be an important modulator of this reduced counterregulatory response, since its expression in the VMH is elevated commensurate with this blunting (36). Similarly, the counterregulatory response to systemic glucoprivation is greatly attenuated 4 days after third ventricular administration of toxic doses of alloxan, which increase VMH GK mRNA expression (70). Finally, VMH GK expression is elevated in obesity-prone rats, which also demonstrate a reduced adrenomedullary response to hypoglycemia (36,94).
But the VMH is not the only site from which the counterregulatory response can be stimulated. Injection of 5TG into the caudal dorsomedial medulla evokes a brisk hyperglycemic response, as do injections into the nearby caudal ventrolateral medulla (30,70) (Fig. 4C). The hyperglycemic response to systemic 2-deoxyglucose can be blocked by lesions of the rostral C1 epinephrine–containing neurons that project to the sympathetic preganglionic neurons in the spinal cord (68). However, unlike the stimulatory effect that alloxan injections into the caudal dorsomedial medulla have on feeding, such injections do not produce hyperglycemia (Fig. 4C).
As with feeding, peripheral glucosensing sites are also important contributors to the effect of glucose on energy homeostasis and the counterregulatory response to hypoglycemia. Raising portal vein glucose levels leads to a decrease in vagal afferent discharges impinging upon NTS neurons, which are themselves inhibited by direct application of glucose. Sympathetic efferents to the adrenal, liver, splanchnic bed, and pancreas are activated, whereas pancreatic vagal afferents are inhibited after such infusions. Because all of these reflex efferent outputs are blocked by hepatic vagotomy, it appears that signals from high levels of portal glucose are transmitted to the brainstem through hepatic vagal afferents (18,95). On the other hand, counterregulatory response to moderate systemic hypoglycemia is attenuated (but not completely blocked) by clamping the liver at euglycemic levels, and this effect is disrupted by interruption of sympathetic (but not vagal) afferents from the hepatic portal circulation (96,97). Such data suggest that the hepatic portal vein, like the brain, may have different glucose sensors that respond to either high or low glucose levels. Other glucosensors are also present in carotid body glomus cells that are activated by low glucose levels and contribute to the release of glucagon from the pancreas in response to hypoglycemia (98,99). These signals reach the brainstem via glossopharyngeal afferents whose cell bodies reside in the inferior petrosal ganglion (Fig. 1). Finally, the small intestinal myenteric plexes contain glucosensing neurons that express the KATP channel, which is the same channel used by pancreatic β-cells and GE neurons in the brain (17,33).
Glucosensing and the predisposition to obesity.
In humans, functional magnetic resonance imaging demonstrates inhibition of hypothalamic indexes of neuronal activation after ingestion of an oral glucose load. Interestingly, this response is both delayed and attenuated in obese individuals (100). Unfortunately, human studies involving brain function are seriously hampered by the low resolution and relative nonspecificity of most imaging techniques. For this reason, rodent models of obesity are useful surrogates for the studies of human obesity, particularly those involving the brain. We have used a polygenic model of rodent obesity that has many similarities to human obesity and has the advantage of being able to study such animals before the onset of obesity to identify potential predisposing factors. One of the most striking features of this model is the fact that obesity-prone rats have a number of defects in their ability to sense and respond to glucose. These include defective glucose-induced activation of the sympathetic nervous system to both peripheral and central glucose infusions (74,101), reduced glucose-induced activation of hypothalamic neurons (similar to that seen in obese humans) (80), reduced sensitivity of the KATP channel (102,103) and the number of VMN glucosensing neurons (35), attenuated adrenomedullary response to hypoglycemia (94), and increased hypothalamic GK expression (36). Despite this large number of defects in central glucosensing, it is unclear how these contribute to the predisposition of these rats to develop diet-induced obesity when the caloric density and fat content of their diets are increased. However, these glucosensing defects appear to be part of an overall reduction in the ability of these animals to respond to the negative feedback signals from the periphery, since they also are resistant to the anorectic effects of leptin and centrally administered insulin (104–106). Such raised thresholds might contribute to their inability to downregulate their caloric intake when diet caloric density is increased despite rising leptin and insulin levels (107). As with all other aspects of glucosensing, it is likely that the interaction between the ability of neurons to sense glucose and participate in the regulation of energy homeostasis is a function of their ability to integrate the responses to glucose and the simultaneous input of a multitude of other incoming signals.
SUMMARY AND CONCLUSIONS
Glucosensing neurons are specialized neurons that reside in a distributed network throughout the brain in sites critical to the regulation of energy homeostasis. They use glucose, other metabolic substrates, hormones, and neural signals from glucosensors in the periphery to control their activity. The output of these neurons is involved in all aspects of energy homeostasis—intake, expenditure, and storage. Reduced sensitivity of these neurons to a variety of substrates and hormones may be a predisposing factor to the development of obesity in some individuals.
![FIG. 1. Location and afferent inputs to central glucosensing neurons. Vagal, glossopharyngeal, and sympathetic afferents from metabolic sensors (denoted by four-pointed stars) in the portal vein, carotid body, stomach, and small intestines converge on NTS metabolic sensing neurons (denoted by five-pointed stars) in the medulla and are integrated along with metabolic and hormonal signals from the periphery, which are transported across the blood-brain barrier or cross through the fenestrated barrier in the area postrema adjacent to the NTS. In addition to the NTS, noradrenergic, adrenergic, neuropeptide Y (NPY), glucagon-like peptide 1 (GLP-1), and α-melanocyte–stimulating hormone (α-MSH) metabolic sensing neurons in the caudal ventrolateral medulla (CVLM) and rostral ventrolateral medulla (RVLM) also integrate these incoming signals and project to various hypothalamic nuclei (arcuate [ARC], ventromedial [VMN], dorsomedial [DMN], paraventricular [PVN]) and the LHA. Metabolic sensing serotonin neurons in the raphe pallidus and obscurus (RPa/Ob) project to sympathetic preganglionic neurons in the intermediolateral cell column of the spinal cord. Metabolites and hormones also interact with metabolic sensing neurons in the hypothalamus. Neurons in the ARC project to additional metabolic sensing neurons in the PVN and LHA, and the LHA also projects to the PVN, which, along with the LHA, gives rise to descending outputs to autonomic outs in the medulla and spinal cord. Adrenal medullary cells also are responsive to low glucose levels as one possible mechanism for their release of epinephrine to mobilize hepatic glucose stores during hypoglycemia (L.K., B.E.L., unpublished data). Dotted lines denote polysynaptic pathways.](/view-large/figure/2329408/zdb1120646260001.jpeg)
Location and afferent inputs to central glucosensing neurons. Vagal, glossopharyngeal, and sympathetic afferents from metabolic sensors (denoted by four-pointed stars) in the portal vein, carotid body, stomach, and small intestines converge on NTS metabolic sensing neurons (denoted by five-pointed stars) in the medulla and are integrated along with metabolic and hormonal signals from the periphery, which are transported across the blood-brain barrier or cross through the fenestrated barrier in the area postrema adjacent to the NTS. In addition to the NTS, noradrenergic, adrenergic, neuropeptide Y (NPY), glucagon-like peptide 1 (GLP-1), and α-melanocyte–stimulating hormone (α-MSH) metabolic sensing neurons in the caudal ventrolateral medulla (CVLM) and rostral ventrolateral medulla (RVLM) also integrate these incoming signals and project to various hypothalamic nuclei (arcuate [ARC], ventromedial [VMN], dorsomedial [DMN], paraventricular [PVN]) and the LHA. Metabolic sensing serotonin neurons in the raphe pallidus and obscurus (RPa/Ob) project to sympathetic preganglionic neurons in the intermediolateral cell column of the spinal cord. Metabolites and hormones also interact with metabolic sensing neurons in the hypothalamus. Neurons in the ARC project to additional metabolic sensing neurons in the PVN and LHA, and the LHA also projects to the PVN, which, along with the LHA, gives rise to descending outputs to autonomic outs in the medulla and spinal cord. Adrenal medullary cells also are responsive to low glucose levels as one possible mechanism for their release of epinephrine to mobilize hepatic glucose stores during hypoglycemia (L.K., B.E.L., unpublished data). Dotted lines denote polysynaptic pathways.
Location and afferent inputs to central glucosensing neurons. Vagal, glossopharyngeal, and sympathetic afferents from metabolic sensors (denoted by four-pointed stars) in the portal vein, carotid body, stomach, and small intestines converge on NTS metabolic sensing neurons (denoted by five-pointed stars) in the medulla and are integrated along with metabolic and hormonal signals from the periphery, which are transported across the blood-brain barrier or cross through the fenestrated barrier in the area postrema adjacent to the NTS. In addition to the NTS, noradrenergic, adrenergic, neuropeptide Y (NPY), glucagon-like peptide 1 (GLP-1), and α-melanocyte–stimulating hormone (α-MSH) metabolic sensing neurons in the caudal ventrolateral medulla (CVLM) and rostral ventrolateral medulla (RVLM) also integrate these incoming signals and project to various hypothalamic nuclei (arcuate [ARC], ventromedial [VMN], dorsomedial [DMN], paraventricular [PVN]) and the LHA. Metabolic sensing serotonin neurons in the raphe pallidus and obscurus (RPa/Ob) project to sympathetic preganglionic neurons in the intermediolateral cell column of the spinal cord. Metabolites and hormones also interact with metabolic sensing neurons in the hypothalamus. Neurons in the ARC project to additional metabolic sensing neurons in the PVN and LHA, and the LHA also projects to the PVN, which, along with the LHA, gives rise to descending outputs to autonomic outs in the medulla and spinal cord. Adrenal medullary cells also are responsive to low glucose levels as one possible mechanism for their release of epinephrine to mobilize hepatic glucose stores during hypoglycemia (L.K., B.E.L., unpublished data). Dotted lines denote polysynaptic pathways.
![FIG. 2. Efferents of glucosensing neurons involved in energy homeostasis. Metabolic sensing neurons (stars) are localized throughout the hypothalamus and brainstem as well as other areas, including the substantia nigra and ventral tegmental area (VTA). ARC NPY and POMC (α-melanocyte–stimulating hormone [α-MSH]) neurons are prototypic metabolic sensing neurons that project to the PVN and to orexin and melanin concentrating (MCH) metabolic sensing neurons in the LHA (heavy dotted lines). ARC POMC neurons, which also express cocaine- and amphetamine-related transcript, also project to the intermediolateral (IML) preganglionic sympathetic neurons. PVN efferents project to the median eminence and pituitary to modulate neuroendocrine outputs and to sympathetic and parasympathetic output (dorsal vagal complex [DVC]) neurons in the medulla for control of visceral functions involved in energy homeostasis. Unlike the relatively restricted output of ARC NPY and POMC neurons, those of the orexin and MCH LHA neurons project widely to brain areas involved in a large variety of behavioral and physiological functions, some of which are involved in the regulation of various aspects of energy homeostasis. The dotted box around the VMN and ARC denotes the area generally referred to as the VMH (ventromedial hypothalamus) in many stimulation and lesion studies of energy homeostasis.](/view-large/figure/2329409/zdb1120646260002.jpeg)
Efferents of glucosensing neurons involved in energy homeostasis. Metabolic sensing neurons (stars) are localized throughout the hypothalamus and brainstem as well as other areas, including the substantia nigra and ventral tegmental area (VTA). ARC NPY and POMC (α-melanocyte–stimulating hormone [α-MSH]) neurons are prototypic metabolic sensing neurons that project to the PVN and to orexin and melanin concentrating (MCH) metabolic sensing neurons in the LHA (heavy dotted lines). ARC POMC neurons, which also express cocaine- and amphetamine-related transcript, also project to the intermediolateral (IML) preganglionic sympathetic neurons. PVN efferents project to the median eminence and pituitary to modulate neuroendocrine outputs and to sympathetic and parasympathetic output (dorsal vagal complex [DVC]) neurons in the medulla for control of visceral functions involved in energy homeostasis. Unlike the relatively restricted output of ARC NPY and POMC neurons, those of the orexin and MCH LHA neurons project widely to brain areas involved in a large variety of behavioral and physiological functions, some of which are involved in the regulation of various aspects of energy homeostasis. The dotted box around the VMN and ARC denotes the area generally referred to as the VMH (ventromedial hypothalamus) in many stimulation and lesion studies of energy homeostasis.
Efferents of glucosensing neurons involved in energy homeostasis. Metabolic sensing neurons (stars) are localized throughout the hypothalamus and brainstem as well as other areas, including the substantia nigra and ventral tegmental area (VTA). ARC NPY and POMC (α-melanocyte–stimulating hormone [α-MSH]) neurons are prototypic metabolic sensing neurons that project to the PVN and to orexin and melanin concentrating (MCH) metabolic sensing neurons in the LHA (heavy dotted lines). ARC POMC neurons, which also express cocaine- and amphetamine-related transcript, also project to the intermediolateral (IML) preganglionic sympathetic neurons. PVN efferents project to the median eminence and pituitary to modulate neuroendocrine outputs and to sympathetic and parasympathetic output (dorsal vagal complex [DVC]) neurons in the medulla for control of visceral functions involved in energy homeostasis. Unlike the relatively restricted output of ARC NPY and POMC neurons, those of the orexin and MCH LHA neurons project widely to brain areas involved in a large variety of behavioral and physiological functions, some of which are involved in the regulation of various aspects of energy homeostasis. The dotted box around the VMN and ARC denotes the area generally referred to as the VMH (ventromedial hypothalamus) in many stimulation and lesion studies of energy homeostasis.
Levels of blood and brain extracellular (EC) glucose at which various components of the counterregulatory response, self-reported hunger, and defects in cognitive function develop during a hyperinsulinemic-hypoglycemic clamp in humans. Data are extrapolated from studies in which simultaneous blood and brain EC glucose levels were measured in rats (62,63) and from studies in humans (64,65).
Levels of blood and brain extracellular (EC) glucose at which various components of the counterregulatory response, self-reported hunger, and defects in cognitive function develop during a hyperinsulinemic-hypoglycemic clamp in humans. Data are extrapolated from studies in which simultaneous blood and brain EC glucose levels were measured in rats (62,63) and from studies in humans (64,65).
![FIG. 4. Sites of hindbrain glucokinase (GK) expression and responses to 5TG- vs. alloxan-induced feeding and hyperglycemia. A: In situ hybridization autoradiograms demonstrating GK mRNA expression in the nucleus tractus solitarius (NTS), raphe obscurus (ROb), and raphe pallidus (RPa) and in the caudal medulla and the site at which alloxan and 5TG were injected (vertical dotted line). Top autoradiogram is rostral and bottom is caudal. B: Food intake over 3 h after 200-nl injections of 24 μg 5TG, pharmacological dose of alloxan (1 μg in saline, pH 3.0), and saline (pH 3.0) into the caudal dorsomedial medulla in satiated rats (n = 7 per group) before and 7 days after injection of a toxic dose of alloxan (40 μg) into the same site. Each injection was separated by at least 4 days. Coordinates were based on those of Ritter et al. (30). Both alloxan and 5TG stimulated food intake comparably, but only the response to alloxan was prevented by pretreatment with a toxic dose of alloxan to destroy GK-expressing neurons. *P < 0.05 when saline group was compared with the others. C: Blood glucose levels over 120 min after the caudal dorsomedial medulla injection of the same amount of 5TG (pre- and post-toxic alloxan dose [40 μg]), alloxan, or saline injected into a separate group of seven rats. The 5TG but not alloxan injections provoked hyperinsulinemia both before and after putative destruction of GK-expressing neurons. D: Bilateral 1-μl injections of 4 μg alloxan into the ventromedial hypothalamus (VMH) had no effect on 3-h food intake (compared with saline; n = 13 per group).](/view-large/figure/2329411/zdb1120646260004.jpeg)
Sites of hindbrain glucokinase (GK) expression and responses to 5TG- vs. alloxan-induced feeding and hyperglycemia. A: In situ hybridization autoradiograms demonstrating GK mRNA expression in the nucleus tractus solitarius (NTS), raphe obscurus (ROb), and raphe pallidus (RPa) and in the caudal medulla and the site at which alloxan and 5TG were injected (vertical dotted line). Top autoradiogram is rostral and bottom is caudal. B: Food intake over 3 h after 200-nl injections of 24 μg 5TG, pharmacological dose of alloxan (1 μg in saline, pH 3.0), and saline (pH 3.0) into the caudal dorsomedial medulla in satiated rats (n = 7 per group) before and 7 days after injection of a toxic dose of alloxan (40 μg) into the same site. Each injection was separated by at least 4 days. Coordinates were based on those of Ritter et al. (30). Both alloxan and 5TG stimulated food intake comparably, but only the response to alloxan was prevented by pretreatment with a toxic dose of alloxan to destroy GK-expressing neurons. *P < 0.05 when saline group was compared with the others. C: Blood glucose levels over 120 min after the caudal dorsomedial medulla injection of the same amount of 5TG (pre- and post-toxic alloxan dose [40 μg]), alloxan, or saline injected into a separate group of seven rats. The 5TG but not alloxan injections provoked hyperinsulinemia both before and after putative destruction of GK-expressing neurons. D: Bilateral 1-μl injections of 4 μg alloxan into the ventromedial hypothalamus (VMH) had no effect on 3-h food intake (compared with saline; n = 13 per group).
Sites of hindbrain glucokinase (GK) expression and responses to 5TG- vs. alloxan-induced feeding and hyperglycemia. A: In situ hybridization autoradiograms demonstrating GK mRNA expression in the nucleus tractus solitarius (NTS), raphe obscurus (ROb), and raphe pallidus (RPa) and in the caudal medulla and the site at which alloxan and 5TG were injected (vertical dotted line). Top autoradiogram is rostral and bottom is caudal. B: Food intake over 3 h after 200-nl injections of 24 μg 5TG, pharmacological dose of alloxan (1 μg in saline, pH 3.0), and saline (pH 3.0) into the caudal dorsomedial medulla in satiated rats (n = 7 per group) before and 7 days after injection of a toxic dose of alloxan (40 μg) into the same site. Each injection was separated by at least 4 days. Coordinates were based on those of Ritter et al. (30). Both alloxan and 5TG stimulated food intake comparably, but only the response to alloxan was prevented by pretreatment with a toxic dose of alloxan to destroy GK-expressing neurons. *P < 0.05 when saline group was compared with the others. C: Blood glucose levels over 120 min after the caudal dorsomedial medulla injection of the same amount of 5TG (pre- and post-toxic alloxan dose [40 μg]), alloxan, or saline injected into a separate group of seven rats. The 5TG but not alloxan injections provoked hyperinsulinemia both before and after putative destruction of GK-expressing neurons. D: Bilateral 1-μl injections of 4 μg alloxan into the ventromedial hypothalamus (VMH) had no effect on 3-h food intake (compared with saline; n = 13 per group).
This article is based on a presentation at a symposium. The symposium and the publication of this article were made possible by an unrestricted educational grant from Servier.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
