Hypothalamic control of energy metabolism via the autonomic nervous system

Transcription

1 Ann. N.Y. Acad. Sci. ISSN ANNALS OF THE NEW YORK ACADEMY OF SCIENCES Issue: The Year in Diabetes and Obesity Hypothalamic control of energy metabolism via the autonomic nervous system A. Kalsbeek, 1,2 E. Bruinstroop, 1 C.X. Yi, 1,2 L.P. Klieverik, 1 S.E. La Fleur, 1 and E. Fliers 1 1 Department of Endocrinology and Metabolism, Academic Medical Center (AMC), University of Amsterdam, the Netherlands. 2 Hypothalamic Integration Mechanisms, Netherlands Institute for Neuroscience, Amsterdam, the Netherlands Address for correspondence: Andries Kalsbeek, Department of Endocrinology and Metabolism, G2-136, Academic Medical Center (AMC), University of Amsterdam, Meibergdreef 9, 1105 AZ Amsterdam, the Netherlands. a.kalsbeek@amc.uva.nl The hypothalamic control of hepatic glucose production is an evident aspect of energy homeostasis. In addition to the control of glucose metabolism by the circadian timing system, the hypothalamus also serves as a key relay center for (humoral) feedback information from the periphery, with the important role for hypothalamic leptin receptors as a striking example. The hypothalamic biological clock uses its projections to the preautonomic hypothalamic neurons to control the daily rhythms in plasma glucose concentration, glucose uptake, and insulin sensitivity. Euglycemic, hyperinsulinemic clamp experiments combined with either sympathetic-, parasympathetic-, or sham-denervations of the autonomic input to the liver have further delineated the hypothalamic pathways that mediate the control of the circadian timing system over glucose metabolism. In addition, these experiments clearly showed both that next to the biological clock peripheral hormones may use the preautonomic neurons in the hypothalamus to affect hepatic glucose metabolism, and that similar pathways may be involved in the control of lipid metabolism in liver and white adipose tissue. Keywords: circadian; orexin; PACAP; neuropeptide Y; thyroid hormone; insulin Introduction The hypothalamus has long been appreciated to be fundamental in the control and coordination of peripheral homeostatic activity. Historically, this has been viewed in terms of the extensive neuroendocrine control system resulting from processing of hypothalamic signals relayed to the pituitary. Through these actions, endocrine signals are integrated throughout the body, modulating a vast array of physiological processes. It goes without doubt that our understanding of the responses to endocrine signals is crucial for the diagnosis and management of many pathological conditions. More recently, the control over the autonomic nervous system emanating from the hypothalamus has been increasingly recognized as a powerful additional modulator of peripheral tissues. Moreover, neuroendocrine and neuronal pathways by the hypothalamus are not separate processes. They appear to act as a single integrated regulatory system, far more subtle and complex than when each is viewed in isolation. Consequently, hypothalamic regulation should be viewed as a summation of both neuroendocrine and neural influences. As a result, our endocrine-based understanding of diseases such as diabetes and obesity could be improved by integration of neural inputs to the pathophysiological process. The main focus of our research in the past decade has been the control of the biological clock over daily rhythms in hormone release and glucose homeostasis. The mammalian biological clock resides in the suprachiasmatic nuclei (SCN) located in the anterior hypothalamus. Our studies on the neuronal mechanism used by the biological clock yielded a nice example of the neuroendocrine and neural integration mentioned above, i.e., we were able to show that the SCN uses its projections to both the neuroendocrine and preautonomic hypothalamic neurons for generating the daily rhythm in plasma corticosterone concentrations. 1,2 On the one hand, doi: /j x 114 Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences.

2 Kalsbeek et al. Hypothalamic control of endogenous glucose production the SCN uses its vasopressin containing projections to the subparaventricular nucleus (subpvn) and the dorsomedial hypothalamus (DMH) to modulate the activity of the corticotropin-releasing hormone (CRH) neurons in the PVN and the subsequent release of adrenocorticotropic hormone (ACTH) from the pituitary, 3 5 whereas on the other hand it uses its GABAergic and glutamatergic projections to the preautonomic neurons in the PVN to control the sensitivity of the adrenal cortex to the incoming ACTH signal. 6 Another example was recently provided by the work on the circadian regulation of the osmoregulatory circuit. 7,8 In addition, our investigations on the circadian control of glucose homeostasis yielded a nice example of this two-edged sword used by the SCN. On the one hand, the SCN uses its projections to sympathetic preautonomic neurons in the PVN to control hepatic glucose production, 9 12 whereas on the other hand it uses its projections to parasympathetic preautonomic neurons in the PVN to control the meal-induced insulin responses from the endocrine pancreas. 13 In addition, these experiments provided evidence for an important role of the hypothalamic orexin, pituitary adenylate cyclase activating peptide (PACAP), and vasoactive intestinal polypeptide (VIP) systems in the (circadian) control of hepatic glucose production. 14,15 Next to the intrinsic hypothalamic control of glucose metabolism by the biological clock, it has become clear that the hypothalamus also serves as an important relay center for (humoral) feedback information from the periphery, with the important role for hypothalamic leptin 16 and, more recently, insulin receptors 17,18 as striking examples. In a recent series of euglycemic, hyperinsulinemic clamp experiments combined with the intracerebroventricular (ICV) administration of neuropeptide Y (NPY) and either sympathetic-, parasympathetic-, or sham-denervations of the autonomic input to the liver, we studied the involvement of hypothalamic pathways in the inhibitory effect of hyperinsulinemia on hepatic glucose production. The results of these experiments clearly indicated that circulating insulin can modulate hepatic glucose production through the NPYcontaining projections of arcuate nucleus neurons to the preautonomic hypothalamic neurons controlling hepatic glucose production. 19 Except for a classic glucoregulatory hormone such as insulin, other classic hormones such as estrogen and thyroid hormone are principal regulators of the glucose and lipid metabolism. It is widely assumed that metabolic changes observed during, for instance, thyrotoxicosis or menopause are mediated via direct actions of thyroid hormone and estrogen, respectively, on nuclear hormone receptors in peripheral organs such as the liver and adipose tissue. However, we and others have shown that the metabolic changes induced by thyroid hormone and estrogen are also mediated via the hypothalamus and the autonomic nervous system, providing clear evidence that a variety of classic hormones may use the preautonomic neurons in the hypothalamus to affect energy metabolism. Hypothalamic neuropeptides and the control of energy homeostasis Nowadays it is evident that the hypothalamic control of hepatic glucose production is an important aspect of energy homeostasis. A host of studies in the last decade has evidenced the involvement of several hypothalamic neuropeptides in the control of hepatic insulin sensitivity and glucose production. Each of these peptides has its distinct physiological role in the control of food intake and energy homeostasis, while in addition affecting common signaling pathways that are involved in the control of energy homeostasis. Therefore, in the remainder of this chapter we will present an overview of the current knowledge on the hypothalamic control of glucose and lipid metabolism, with a special focus on neuropeptidergic pathways involved in the regulation of hepatic glucose production. Neuropeptide Y One of the most familiar hypothalamic neuropeptidergic networks involved in the control of energy metabolism are the NPY-containing neurons in the ARC with their projections to several hypothalamic brain areas, including the PVN. The first report on the glucoregulatory effects of the hypothalamic NPY system appeared in the mid-1990s when it was shown that ICV administration of NPY increases endogenous glucose production in rats, probably by decreasing hepatic insulin sensitivity. 24,25 Later on these results were confirmed in mice. 26 In view of the inhibitory effects of hypothalamic insulin receptors on hepatic glucose production, 17,18 the abundant expression of insulin receptors in the ARC, 27 Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences. 115

3 Hypothalamic control of endogenous glucose production Kalsbeek et al. the inhibitory effect of insulin on NPY neuronal activity, 28,29 and the effects of ICV NPY on sympathetic activity, we decided to test whether NPY could be the hypothalamic intermediate between the insulin receptors in the ARC and the preautonomic neurons in the PVN. Hereto we combined the euglycemic hyperinsulinemic clamp technique with the ICV administration of NPY, and performed these experiments in hepatic sympathetic-, hepatic parasympathetic-, and hepatic sham-denervated rats. Our results confirmed that ICV NPY is able to (partially) block the inhibitory effects of peripheral hyperinsulinemia on hepatic glucose production, but they also showed that selective denervation of hepatic sympathetic nerves blocks the effect of NPY on hepatic insulin sensitivity. 19 Therefore, the brain-mediated inhibitory effect of insulin on hepatic glucose production is probably effectuated via an inhibition of NPY neuronal activity in the ARC. Subsequently, the resulting diminished release of NPY will decrease the stimulatory input to the sympathetic preautonomic neurons in the PVN and thus reduce the sympathetic stimulation of hepatic glucose production. The results of Pocai and colleagues, 33 however, also show that the parasympathetic innervation of the liver is involved in the inhibitory effect of insulin on hepatic glucose production. This means that in addition to the effect of NPY on the sympathetic preautonomic neurons, there is probably another neurotransmitter that is responsible for the transmission of insulin s effects in the ARC to the parasympathetic preautonomic neurons in the PVN. Moreover, the effects of NPY also seem to be specific for glucose production as in none of the above-mentioned studies there was a significant effect on whole body glucose disposal. Several studies have shown that chronic infusion of NPY, mimicking high levels of hypothalamic NPY in animal models of obesity, causes hypertriglyceridemia and lipogenesis in liver and white adipose tissue, even when animals are pair-fed Interestingly, the effect on plasma triglycerides was abolished in adrenalectomized rats, but reappeared in adrenalectomized rats coinfused centrally with NPY and glucocorticoids. 37,38 Van den Hoek and colleagues 26 were the first to describe an acute effect of NPY on liver triglyceride secretion, in mice. In this study hyperinsulinemia decreased the secretion of triglycerides in VLDL particles by the liver. This effect was abolished when the hyperinsulemic clamp was combined with the ICV administration of NPY, suggesting that in the fed state the inhibitory effect of insulin on the activity of NPY neurons is responsible for the decrease in triglyceride secretion. No effects on triglyceride secretion were found in the basal state. More recently, however, Stafford and colleagues 39 showed that ICV administration of NPY in rats increased triglyceride secretion in the post-absorbative state. The changes in hepatic mrna expression suggest that the increased release of hypothalamic NPY causes a mobilization of stored triglycerides in the liver, whereas de novo fatty acid synthesis is inhibited. Moreover, after a longer fast, ICV administration of an Y1 antagonist decreased triglyceride secretion, suggesting a physiological role for the hypothalamic NPY neurons in the ARC in the control of hepatic triglyceride secretion during the transition from feeding to fasting. 39 Pro-opiomelanocortin (POMC) Next to the orexigenic NPY/AGRP neurons, the ARC also contains a population of anorexigenic POMC/CART-containing neurons. The most important POMC-derived peptide with respect to feeding and metabolism is -MSH. The reciprocally antagonistic function of the NPY/AGRP and POMC/CART cell populations is most clearly illustrated by the fact that AGRP acts as an endogenous antagonist of the melanocortin receptors 3 and 4 (MC3R, MC4R), 40 for which -MSH is the main endogenous agonist. The POMC neurons seem to be good candidates for the missing parasympathetic link in the previous paragraph, i.e., the neurons in the arcuate nucleus that are responsible for transmission of insulin s effects to the parasympathetic preautonomic neurons in the PVN, but the experimental evidence is ambiguous. In MC4R knockout mice, plasma insulin levels are increased 41 and central administration of the -MSH agonist melanotan II (MTII) dose dependently inhibits basal insulin release, 42 but ICV administration of -MSH or MTII enhances the action of insulin on both glucose uptake and production 43,44 and transgenic overexpression of -MSH leads to improved glucose metabolism. 45,46 Moreover, the phenotype of the MC4R-deficient mice and humans, i.e., reduced heart rate and diastolic blood pressure in the face of severe obesity, is explained by a decreased sympathetic/parasympathic balance. 47 Surprisingly, 116 Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences.

4 Kalsbeek et al. Hypothalamic control of endogenous glucose production however, -MSH does not seem to be involved in the inhibitory effect of hypothalamic insulin on HGP, as coadministration of a melanocortin antagonist failed to block the decrease in HGP induced by hyperinsulinemia. 17,18 Blocking -MSH signaling via ICV infusion of the melanocortin 3/4 receptor (MC3R/MC4R) antagonist SHU9119 has no effects on glucose metabolism, but ICV infusion of -MSH itself has a clear stimulatory effect on EGP via gluconeogenesis (GNG), which can be antagonized by SHU In the liver, the stimulation of GNG was confirmed by the increased expression of G6Pase and PEPCK. It has been proposed that the hypothalamic MC3R/MC4R signaling pathway mediates the effect of systemic leptin on EGP. 49 Central administration of leptin has been proven to be involved in the autoregulation of hepatic glucose output, i.e., an increase in GNG with a concomitant decrease in glycogenolysis without changing total glucose production. 50,51 Recently, it was nicely shown that the adenoviral-induced expression of leptin receptors in the ARC of leptin receptor knock-out animals improves glucose tolerance via enhanced suppression of EGP. 52 The ARCinduced expression of the leptin receptor was associated with a reduced hepatic expression of G6Pase and PEPCK, but again, no significant changes in the insulin-stimulated whole body glucose utilization were apparent. Moreover, the effects of hypothalamic leptin signaling on hepatic insulin sensitivity could be blocked by a selective hepatic vagotomy, providing further support for the idea that ARC projections to preautonomic neurons (in the PVN) are important for the transmission of the effect of leptin on EGP. Further supporting the involvement of POMC neurons in these effects are the observations that re-expression of leptin receptors specifically in POMC neurons improves glucose metabolism, 53 whereas removal of both leptin and insulin receptors specifically from POMC neurons causes severe insulin resistance and increases HGP. 54 Evidently more experiments are needed to unravel the precise hypothalamic pathways. Contrary to the effect of chronic ICV administration of NPY, chronic blockade of the central melanocortin system does not change plasma levels of triglycerides and free fatty acids (FFA) although it does induce an obese phenotype. 35,55,56 However, both chronic blockade of the melanocortin system in the Agouti yellow mice and infusion of a melanocortin antagonist promote lipogenesis and lipid accumulation in liver with a possible role for SREB1c and PPAR. 35,57 58 MC4R knockout mice display increased levels of the lipogenic gene, fatty acid synthase (FAS), as well as hepatic steatosis. 59 In contrast, activation of the melanocortin system with leptin, MTII or NDP- MSH reduces the expression of lipogenic genes in the liver. 60,61 Thus, although the melanocortin system seems to be involved in the control of hepatic lipid metabolism, its interaction with dietary factors and hormones, such as insulin is still under debate. For instance, the significance of insulin s stimulating effect on the expression of lipogenic genes, such as SREBP1c and possibly PPAR is unclear at present, as the increased FAS expression and hepatic steatosis in MC4R knockout mice is abolished in preobese mice not displaying hyperinsulinemia. 59 Likewise, the interpretation of the observations in these mice that increased hepatic lipogenesis and fat content were largely prevented by pair-feeding, 55,56 and that an acute blockade or activation of the central melanocortin system did not significantly alter hepatic triglyceride secretion remains enigmatic. 39 Orexin The neuropeptides orexin-a and orexin-b (also known as hypocretin-1 and hypocretin-2) were initially identified as the endogenous ligands for orphan receptors involved in the pathogenesis of narcolepsy. 62,63 They were subsequently recognized as regulators of feeding behavior and energy metabolism because of the exclusive localization of their cell bodies in the lateral hypothalamus (LH), the induction of feeding upon their ICV administration, their responsiveness to peripheral metabolic cues such as leptin and glucose, and the metabolic phenotype of orexin knock-out models. More recent studies suggested a primary role for the orexin system in the maintenance of wakefulness. 64 However, our data showing that an increased availability of orexin in the central nervous system, either by ICV infusion or by local activation via removal of GABA inhibition, increases plasma glucose concentrations through an increase in hepatic glucose production, have clearly revitalized the concept of regulation of metabolism by the orexin system(fig. 1). Similar to NPY, the stimulatory effect of orexin on EGP could be blocked by a hepatic sympathetic but not Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences. 117

5 Hypothalamic control of endogenous glucose production Kalsbeek et al. Figure 1. Mid-sagital view of the rat brain with a schematic representation of the hypothalamic connections involved in the autonomic control of the daily rhythm in hepatic glucose production. The orexin-containing neurons in the perifornical area are innervated by both glutamatergic and GABAergic projections from the SCN. During the main part of the light period, activation of the orexin neurons by the excitatory glutamatergic inputs is prevented by release of the inhibitory neurotransmitter GABA. The circadian withdrawal of the GABAergic input allows the orexin neurons to become active at the onset of darkness. Subsequently, the excitatory effect of orexin on the preganglionic neurons in the spinal cord will activate the sympathetic input to the liver and result in an increased hepatic glucose production. parasympathetic denervation. 14 Moreover, Shiuchi and colleagues 65 demonstrated that via its action in the VMH orexin also stimulates glucose uptake in skeletal muscle. The effect of orexin on glucose uptake was also mediated via the sympathetic nervous system. These effects of central orexin on hepatic glucose metabolism are nicely in line with the increased hypothalamic orexin expression observed in mice homozygous for the tubby mutation. 66 Mice carrying a tubby mutation develop retinal and cochlear degeneration as well as late-onset obesity and disturbed carbohydrate metabolism. The tubby phenotype of sensory loss coupled with obesity and insulin resistance is similar to that found in two human syndromes, Alstrom and Bardet-Biedl. The ICV infusion experiments and the presence of a pronounced orexin-containing fiber network in the PVN suggest that the main action is at the level of the sympathetic preautonomic neurons in the PVN, but in view of the electrophysiological data of Van Den Top and colleagues, 67 a direct effect of orexin at the level of the sympathetic preganglionic neurons in the intermediolateral column of the spinal cord cannot be excluded at this stage. Unfortunately, the selective liver denervations do not allow for a distinction between these 2 options. In addition, it is not clear yet what are the endogenous triggers for the stimulatory effect of orexin on EGP, but we have proposed 2 possible pathways. First, the orexin neurons could be an alternative target for the output from the ARC, i.e., in addition to the ARC projection to the preautonomic neurons in the PVN. Second, the orexin neurons may integrate circadian information from the SCN. Ours as well as other studies clearly showed that the activity of the orexin neurons is under tight control of a GABAergic input that is probably derived from the circadian system. 13,68 These data indicate that the circadian rhythm in orexin release 69 might be implicated in the genesis of the circadian rhythm in plasma glucose concentrations. In order to test this hypothesis, we administered the orexin antagonist SB , either ICV or i.v., during the final 8 h of the light period and simultaneously 118 Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences.

6 Kalsbeek et al. Hypothalamic control of endogenous glucose production Figure 2. ICV, but not IV (not shown), administration of the orexin antagonist SB during the latter part of the light period (from ZT6 to ZT12) prevents the endogenous rise of Ra before the onset of darkness (ZT11 ZT12), but not the feedinginduced increase in Ra after the onset of darkness (ZT12 ZT15). P < 0.05 versus the same time period in the vehicle controls; P < 0.05 versus ZT4 6; ΔP < 0.05 versus ZT6 ZT11. measured glucose appearance (Ra) from ZT3-ZT15 with the isotope dilution technique. The ICV but not IV, administration of the orexin-antagonist completely blocked the endogenous increase in Ra until the start of the dark period. Once the animals start eating, in the dark period, Ra also increases in the ICV orexin-antagonist treated animals. This ICV administration of the orexin-antagonist did not inhibit food intake. Together these data strongly suggest that the perifornical orexin neurons are an important link in the circadian control of the daily peripheral glucose rhythm (Fig. 2). Hereby the orexin system provides a nice example of hypothalamic integration, as the increased activity of the orexin system at the end of the light period not only initiates the wake state but at the same time ensures a sufficient supply of energy. This concept may be highly relevant for the recently discovered correlation between sleep duration and type 2 diabetes We hypothesize that short sleep or sleep deprivation may cause an over activation of the orexin system, and thereby a disproportionate increase in EGP. 76 Poli and colleagues 77 compared energy metabolism between patients displaying narcolepsy with cataplexy and patients with idiopathic hypersomnia. Lumbar punctures showed that orexin-a levels in CSF were low in narcoleptic patients. In addition, narcoleptic patients displayed higher waist circumference and higher plasma levels of triglycerides, despite eating less than hypersomnic patients. Conversely, animal experiments have shown that sleep deprivation induces higher prepro-orexin gene expression, hyperphagia, slight weight loss and lower levels of triglycerides. 73 The apparent contradiction between the hyperglycemic effect of ICV orexin described above and the increased weight circumference in narcoleptic patients might be explained by the antiobesity effect of prolonged stimulation of the orexin receptor-2. Transgenic overexpression of orexin in mice made them resistant to diet-induced obesity, 78 probably mediated by an increased metabolic rate and increased leptin sensitivity. Chronic administration of orexin-a in rats, however, did not lead to changes in body mass, plasma triglcyerides, FFA or cholesterol. 79 More recently, Shen and colleagues 80 have shown in an acute experiment that low-dose ICV administration of orexin-a decreased the activity of autonomic nerves innervating white adipose tissue and lipolysis as measured by plasma FFA. This effect was blocked by systemic pretreatment with a muscarinic receptor blocker. By contrast, a high dose of orexin-a resulted in increased autonomic activity and lipolysis, an effect that was blocked by pretreatment with a beta adrenergic receptor blocker. These data further support a role for central orexin pathways in the metabolic control of adipose tissue, possibly via the autonomic nervous system. Melanin-concentrating hormone Melanin-concentrating hormone (MCH) is a cyclic 19-amino-acid polypeptide with an expression pattern that is limited to the LH, zona incerta, and perifornical area, i.e., very similar to orexin. However, despite the almost complete overlap in their distribution, the two peptides do not colocalize. The MCH neurons have been implicated as an important regulator of food intake, because the central administration of MCH promotes feeding, MCH mrna levels rise as a result of starvation and leptin deficiency, knock-out animals are hypophagic and lean, and MCH neurons are essential for the leptin-deficient phenotype. 84 In addition, overexpression of MCH results in hyperglycemia. 85 Despite these earlier observations, we found no effect on glucose metabolism of either ICV administered MCHinwildtyperats 14 or of the MCH knock-out in the MCH knock-out rat. 86 In fact, the reduced metabolic rate we found in the MCH knock-out Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences. 119

7 Hypothalamic control of endogenous glucose production Kalsbeek et al. rats was perfectly adapted to the leaner body composition of these animals. Evidently, these data do not exclude a role for MCH in glucose metabolism as we may have missed the right stimulus to reveal its function. Chronic ICV infusion of MCH in pair-fed mice results in increased plasma triglycerides, with unchanged liver triglycerides. Ex vivo measurements in the MCH treated animals showed an increased lipogenic activity, as measured by the incorporation of 14C-acetate, in both liver and WAT. But as the expression of SREB1c and FAS was not changed, the increased lipogenic activity could not be explained via this pathway. 87 Acute or chronic ICV infusion of a MCH antagonist did not change plasma triglyceride or FFA levels. 88 In both an obese and lean steatosis model, a MCH antagonist reduced hepatic TG contents without affecting the expression levels of lipogenic genes, but the MCH antagonist clearly suppressed gene expression of CYP4A10 and CYP4A14, i.e., two enzymes that are believed to have a key role in the pathophysiology of steatohepatitis. 89 Additionally, MCH1r knockout mice display lower levels of triglcyerides in plasma and liver compared to wild type with lower levels of SREB1c. The MCH1r knockout mice and mice infused with a MCH antagonist, also seem to be protected against hepatic steatosis induced by ovariectomy, indicating a role for MCH in the effect of oestrogen on metabolism. 90 The role of insulin and possible other factors in the changes caused by MCH are yet to be investigated before a direct neural connection can be determined. Pituitary adenylate cyclase activating peptide PACAP is a 38-amino acid, C-terminally amidated neuropeptide, that was originally isolated from the ovine hypothalamus on the basis of its ability to stimulate adenylate cyclase activity in rat anterior pituitary cells. 91 Studies conducted in rodents have shown that PACAP exerts a wide array of biological activities both in the CNS and in peripheral organs. The results from knock-out studies clearly indicated the involvement of PACAP in glucose metabolism However, these studies did not reveal which part of the metabolic phenotype could be attributed to central signaling pathways of PACAP, although some evidence for central effects on energy metabolism was available. Amongst others it has been shown that in the brain PACAP decreases food intake 95,96 and increases plasma glucose. 97 We showed that ICV administered PACAP causes a strong increase of EGP. 15 Additional tracing and denervation experiments provided strong evidence that the effects of PACAP are mediated through the preautonomic neurons in the hypothalamus. Moreover, ICV administration of PACAP causes augmented sympathetic nerve activity, whereas parasympathetic nerve activity is decreased. 98 Contrary to the neuropeptidergic systems discussed above, PACAP-producing neurons do not show a restricted localization, but are widespread throughout the CNS. Prominent populations of PACAP neurons can be found in the ARC and the VMH, but part of the PACAP innervation in the PVN is also derived from other sources such as the brainstem and the bed nucleus of the stria terminalis (BNST). 99 Since at present only little is known about the stimuli that modulate PACAP release, it is not clear what the physiological function of PACAP could be. We speculated that it could be involved in the counterregulatory response to hypoglycemia, as PACAP knockout animals have a defective counterregulatory response. 92 Recovery from hypoglycemia is believed to be mainly processed by the VMH and the sympatho-adrenal pathway, 100 and the VMH contains a prominent population of PACAP containing neurons 99,101 (Fig. 3). In line, PACAP is involved in the chronic and acute stress response via the PVN CRH neurons, the HPA-axis and the sympathetic nervous system Thus, the PACAP system may be an important gateway to control hepatic glucose production during stressful conditions including hypoglycemia. PACAP / mice show no expression of PACAP in the brain and die after several days, possibly due to cardiovascular distress. They display high levels of FFA, triglycerides and cholesterol compared to wildtype animals, with high amounts of microvascular fat in liver and heart cells. Subcutaneous white adipose tissue deposits were totally depleted at time of death, possibly indicating a function of PACAP in the mobilization of fatty acids. The site of action of PACAP in these metabolic changes has yet to be determined. 92 To our knowledge, the only study to have addressed the possible central role of PACAP in lipid metabolism was performed in chickens. It showed that ICV injections of either PACAP or 120 Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences.

8 Kalsbeek et al. Hypothalamic control of endogenous glucose production Figure 3. Mid-sagital view of the rat brain with a schematic representation of proposed VIP- and PACAP-containing hypothalamic connections involved in the control of hepatic glucose production. The VIP-containing input to the PVN is almost exclusively derived from the SCN. Therefore we suggest that the observed effects of ICV administered VIP on hepatic glucose production reflect an involvement of the SCN in the daily control of hepatic glucose production. The PACAP-containing input to the PVN on the other hand can be derived from different sources. We propose that the effects of ICV administered PACAP on hepatic glucose production reflect an activation of PACAP-containing neurons in the VMH that are activated by for instance a hypoglycemic event. VIP increased NEFA, with a tendency to decrease triglycerides, indicating increased lipolysis. 105 Mice lacking the Adycap1 / gene, which encodes PACAP, showed a markedly reduced white adipose tissue mass, 106 raising the possibility that PACAP signaling pathways favor energy storage. Vasoactive intestinal polypeptide VIP is a 28 amino acid peptide expressed at multiple sites throughout the body. It was discovered as a potent muscle relaxant with vasodilatory activity and as a stimulator of secretory activity in the gut. Mice defective in VIP signaling show dysglycemia and overtly altered daily rhythms of metabolism and feeding behavior. 107,108 With VPAC2 being one of the main receptors for VIP, unsurprisingly, similar finding were obtained in mice defective in VPAC2 signaling. However, a close comparison of these two knockout genotypes revealed that whereas daily food intake and metabolic rate were significantly reduced in the VPAC2 receptor knockout mice, no such reduction was seen in the VIP knockout mice. These results are completely in line with the fact that virtually all of the VIP projections in the hypothalamus are derived from the circadian oscillator in the SCN, but that the VPAC2 receptors (in the PVN) bind PACAP with equal affinity as VIP. The VPAC2 receptor may thus contribute to the regulation of feeding and metabolism independently from its role in the circadian clock. Therefore, our results with ICV VIP on glucose metabolism most likely reflect the influence of the clock, 15 whereas the effects of PACAP probably are a reflection of the activity of PACAP-containing neurons in the VMH as discussed above (Fig. 3). The proposed stimulatory role of a VIP-containing projection from the SCN on plasma glucose concentrations is fully supported by an elegant series of experiments by the Nagai group. 109 Preliminary evidence from the same group indicates that also this SCN-VIP effect may be mediated via the sympathetic innervations of the liver. 110 Remarkably the hereditary blind microphtalmic rat (with a greatly reduced VIP content in the SCN) shows a greatly increased fat deposition, without significant differences in body weight or naso-anal length. 111 Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences. 121

9 Hypothalamic control of endogenous glucose production Kalsbeek et al. Thyrotropin-releasing hormone The tripeptideamide pyroglu-his-pro-nh2, was originally isolated as the first hypothalamic hormone and named thyrotropin-releasing hormone (TRH) based on its capacity to stimulate the release of thyroid-stimulating hormone (TSH) from the anterior pituitary. Later studies showed that TRH is also a potent prolactin releasing hormone. TRH positive neurons and fibers are widely distributed throughout the brain. Although the hypothalamus contains the highest concentration, actually over 70% of total brain TRH is found in extra hypothalamic areas, such as the olfactory system, cortex, thalamus, amygdala, brainstem, spinal cord, and pineal gland. The expression of the TRH receptor type 1 (TRH-R1) in the brain is very restricted and mainly confined to the hypothalamus, brainstem and spinal cord. By contrast, the type 2 receptor (TRH-R2) is widely distributed throughout many brain areas, including cerebellar and cerebral cortex and the reticular information, which may explain central effects of TRH on cognitive functions and arousal. 112 TRH may play a dual role in energy homeostasis, first by controlling the secretion of thyroid hormones that are a key determinant of metabolism, and secondly through central mechanisms that are independent of its endocrine actions. Already in the 1980s, it was found that the hypothalamic or ICV administration of TRH induces hyperglycemia through pathways involving the adrenal gland, the pancreas and the liver Remarkably, TRH / mice also show a marked hyperglycemia, and this has been attributed to impaired pancreatic insulin secretion. 116 In the past two decades the physiological role of TRH in the autonomic regulation of visceral functions has been further established and seems to involve both sympathetic and parasympathetic effects The major site of action for these visceral TRH effects seems to be the brainstem and spinal cord, and may predominantly involve the TRH-R1. TRH containing projections from the raphe nuclei innervate the dorsal vagal complex, the nucleus of the solitary tract, the ventrolateral medulla and the intermediolateral column of the spinal cord 121,122 and the same areas abundantly express the TRH-receptor Thus in spite of the pronounced hypothalamic effects of TRH on glucose metabolism and the considerable number of centrally projecting TRH neurons in the PVN, the contribution of TRH-containing projections from the PVN to these brainstem and spinal cord areas is very limited, 121, although probably not completely absent Despite the ubiquitous distribution of TRH within the central nervous system, the exact mechanism via which TRH regulates visceral and endocrine functions remains to be fully elucidated, probably related to the absence of reliable selective TRH agonists and antagonists. In the thyrotoxic state, major changes in lipid metabolism of liver and WAT occur to supply fuel for increased energy demands, with a major role for fatty acids supplied by de novo lipogenesis and lipolysis. 130,131 Riedel and colleagues 132 showed that TRH injected into the cisterna magna in rabbits increased plasma FFA levels, indicating lipolysis. This increase of FFA was higher after transsection of the spinal cord at level C6 7, contrary to the increase of plasma concentrations of glucose and insulin, which was significantly lower in the spinal cord transected group. After thyroidal denervation the TRHinduced rise in FT3 and FT4 levels was not present, but the increase in FFA was even higher. Abdominal vagotomy did not affect this increase in FFA levels. Therefore, despite this pronounced effect of central TRH on lipolyis its neural mechanism remains unclear, as it seems independent of thyroid hormone, insulin and autonomic nervous input. Arginine-vasopressin Vasopressin is a neuropeptide hormone that is involved in diverse functions, including the regulation of osmotic homeostasis, coagulation, vasomotor tone, ACTH release and circadian rhythms. Several reports indicated the involvement of arginine-vasopressin (AVP) in plasma glucose homeostasis. It has been known for a long time that circulating AVP affects glucose metabolism by promoting gluconeogenesis and glycogenolysis in the liver, 133,134 and by modulating insulin and glucagon release secretion from the endocrine pancreas. 135 Indeed, the vasopressin-deficient Brattleboro rat has difficulty maintaining euglycemia during restricted feeding. 136 Moreover, both vasopressin V1a and V1b receptor knockout animals show an altered glucose homeostasis. 137,138 As till now, however, no clear evidence has been presented for a modulatory role of the hypothalamic vasopressin systems on glucose metabolism, despite the direct indications from an excitatory effect of vasopressin on glucose-responsive VMH neurons Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences.

10 Kalsbeek et al. Hypothalamic control of endogenous glucose production In line, the moderate hyperglycemia we observed previously upon administration of the vasopressin V1a antagonist in the PVN, but not the DMH, may involve increased release of circulating vasopressin. 9 The only direct evidence so far for a role of central AVP in glucose homeostasis is the hyperglycemia induced by local administration of AVP in the nucleus of the solitary tract, 140 which may point at a role for the descending AVP-containing projection from the preautonomic neurons in the PVN. In addition to glucose metabolism, AVP also exerts effects on lipid metabolism. Vasopressin V1a receptor knockout mice exhibit a phenotype with hypermetabolism of fat, while V1b receptor knockout mice show suppressed lipolysis and enhanced lipogenesis, resulting in increased fat weight. 141,142 However, as for glucose metabolism, no clear evidence exists for a modulatory role of hypothalamic vasopressin neurons in lipid metabolism. The effects of AVP on lipid metabolism may be attributed solely to its peripheral effects. 142 Oxytocin Oxytocin is a nonapeptide hormone synthesized in neurons of the PVN and supraoptic nucleus (SON). Thebulkofthepeptideistransportedfromthemagnocellular oxytocin neurons of the PVN and SON via the internal zone of the median eminence to the posterior pituitary where it is secreted into the periphery.thetypicalactionsofperipheraloxytocin are stimulation of uteral smooth muscle contraction during labor and milk secretion during lactation. The central oxytocin-containing projections are mainly derived from the parvocellular neurons in the PVN. The parvocellular oxytocin neurons project to various hypothalamic, limbic, and brainstem regions, as well as to the spinal cord. Oxytocin receptors are abundantly expressed in the spinal cord where oxytocin excites sympathetic preganglionic neurons. 143,144 Early studies showed an inhibitory effect of centrally administered oxytocin on food intake. 145,146 It has been proposed that the preautonomic oxytocin neurons in the PVN are a component of the leptin-sensitive signaling circuit between the hypothalamus and the brainstem, 147 with some oxytocin neurons projecting polysynaptically to brown and white adiposetissue,aswellasliverorpancreas The strong decrease in the number of oxytocin expressing neurons in the PVN of Prader-Willi patients, 152 and the decreased oxytocin expression in the obese Sim1 mice 153 are in total agreement with its proposed role as satiety neurons. More recently it has been shown that both the oxytocin- and the oxytocin-receptor deficient mice show an obese phenotype. 154,155 Part of this phenotype might be the result of a low sympathetic tone due to the absence of oxytocinergic signaling in the hypothalamic projections to the spinal cord. However, the possibility that lack of oxytocin signaling in peripheral organs, such as adipose tissue and pancreas, is contributing to the phenotype can certainly not be excluded Oxytocin deficient mice display mild normophagic obesity, increased WAT weight and elevated fasting plasma triglycerides, 155 suggesting a role for oxytocin in lipid metabolism. But again, at present no evidence is available to directly link hypothalamic oxytocin to these effects. Conclusion Most of the neuropeptides discussed in this review in relation to their role in the control of energy metabolism share a common hypothalamic effector area, the PVN. However, it has not been proven definitely that they all regulate energy metabolism specifically via the PVN. But in our studies and studies of others, central administration of orexin-a, PACAP-38, NPY, 14,15,159 and synthetic MC3R and MC4R agonist 160 are all associated with Fos immunoreactivity in this nucleus. Moreover, the PACAP-38 induced Fos-ir neurons in the PVN project to the sympathetic preganglionic neurons in the spinal cord. Since some of these peptides are orexigenic (orexin, NPY, and MCH) while others (POMC and PACAP) are anorexigenic, we assume that the mechanism of feeding regulation is separated from that of energy metabolism. Sympathetic and parasympathetic preautonomic neurons in the PVN are separated. 161 Limited circumstantial evidence for the involvement of peptidergic transmitters contained in these preautonomic neurons, such as AVP, oxytocin and TRH, in the control of glucose and lipid metabolism is available, but clearly further experiments are needed to pinpoint their exact role. This also brings about the question whether the neuropeptidergic effects described above can be categorized into two groups, depending on their Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences. 123

11 Hypothalamic control of endogenous glucose production Kalsbeek et al. specific effects on either sympathetic or parasympathetic output pathways. In our studies, the hyperglycemia induced by orexin-a and PACAP- 38 can only be blocked by selective hepatic sympathetic, but not parasympathetic denervation. In the case of NPY, the suppressive effect on hepatic insulin sensitivity is only blocked by hepatic sympathetic denervation. On the other hand, it has been shown that insulin and leptin signaling in ARC influence hepatic insulin sensitivity via the vagal nerves. 17,18,52 Clearly, further studies combining neuroanatomical and physiological approaches are necessary to further unveal this parasympathetic pathway. Finally, the NPY containing projections from the arcuate nucleus have been clearly implicated in the central effects of insulin, whereas the POMC neurons have been proposed as an important mediator for the central effects of leptin. The antagonizing mechanism of these neuropeptides is extremely important to adapt the hypothalamic pituitary thyroid axis to the prevailing food/energy status, i.e., the fasting-induced suppression of TRH mrna in the PVN follows from the reduction in hypothalamic -MSH and the increase in NPY/AGRP. 162 As it is clear by now that many hormones exert part of their effects via a central route, a major question is which hypothalamic neuropeptides are involved in mediating the diverse effects of these hormones. Only few data are available for a limited set of hormones such as estrogen, resistin and corticosterone. Singhal and colleagues 163 nicely showed that the adipocyte hormone resistin needs an intact NPY system in order to exert its inhibitory effect on hepatic glucose production. Also for the glucocorticoids there are strong indications that the hypothalamic NPY system is an important mediator for the central effects of glucocorticoids on energy metabolism. 164,165 PACAP expression in the VMH has been reported to be under the influence of estrogen, 166 and by inference the above described effectsofpacaponglycogenolysismightbepartofa brain circuit that connects the reproductive system with energy metabolism. Although a large part of this evidence is still circumstantial, it clearly shows that there is no such thing as a 1 hormone 1 neuropeptide relation, and that the NPY and POMC neurons (at least at the system level) integrate the information from a whole spectrum of humoral factors. Acknowledgments The authors thank Dr. Mariette T. Ackermans at the Academic Medical Center in Amsterdam for her help with the stable isotope measurements, Henk Stoffels for preparation of the images and Wilma Verweij for correction of the manuscript. Special thanks are dedicated to Ewout Foppen for his superb technical assistance in most of the work just described and to Jilles Timmer for animal husbandry. Parts of the work presented were financially supported by the Dutch Diabetes Research Foundation and by Servier. Conflicts of interest The authors declare no conflicts of interest. References 1. Buis, R.M. & A. Kalsbeek The day-night balance of everyday life: central and peripheral clocks interacting in the hypothalamus. Nat. Neurosci. Rev. 2: Kalsbeek, A. & R.M. Buis Output pathways of the mammalian suprachiasmatic nucleus: coding circadian time by transmitter selection and specific targeting. Cell Tissue Res. 309: Kalsbeek, A., R.M. Buis, J.J. Van Heerikhuize, et al Vasopressin containing neurons of the suprachiasmatic nuclei inhibit corticosterone release. Brain Res. 580: Kalsbeek, A., J.J. Van Heerikhuize, T.P. Van Der Woude, et al. 1996a. A diurnal rhythm of stimulatory input to the hypothalamo-pituitary-adrenal system as revealed by timed intrahypothalamic administration of the vasopressin V 1 -antagonist. J. Neurosci. 16: Kalsbeek, A., J. Van Der Vliet & R.M. Buis. 1996b. Decrease of endogenous vasopressin release necessary for expression of the circadian rise in plasma corticosterone: a reverse microdialysis study. J. Neuroendocrinol. 8: Buis, R.M., J. Wortel, J.J. Van Heerikhuize, et al Anatomical and functional demonstration of a multisynaptic suprachiasmatic nucleus adrenal (cortex) pathway. Eur. J. Neurosci. 11: Trudel, E. & C.W. Bourque Central clock excites vasopressin neurons by waking osmosensory afferents during late sleep. Nat. Neurosci. 13: Colwell, C.S Preventing dehydration during sleep. Nat. Neurosci. 13: Kalsbeek, A., S.E. La Fleur, C. Van Heijningen & R.M. Buijs Suprachiasmatic GABAergic inputs to the paraventricular nucleus control plasma glucose concentrations in the rat via sympathetic innervation of the liver. J. Neurosci. 24: Kalsbeek, A., M. Ruiter, S.E. La Fleur, et al The hypothalamic clock and its control of glucose homeostasis. Prog. Brain Res. 153: Cailotto, C., S.E. La Fleur, C. Van Heijningen, et al The suprachiasmatic nucleus controls the daily variation of 124 Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences.

12 Kalsbeek et al. Hypothalamic control of endogenous glucose production plasma glucose via the autonomic output to the liver: are the clock genes involved? Eur. J. Neurosci. 22: Cailotto, C., C. van Heijningen, J. Van Der Vliet, et al Daily rhythms in metabolic liver enzymes and plasma glucose require a balance in the autonomic output to the liver. Endocrinology 149: Kalsbeek, A., E. Foppen, I. Schalij, et al Circadian control of the daily plasma glucose rhythm: an interplay of GABA and glutamate. PLoS One 3: e Yi, C.X., M.J. Serlie, M.T. Ackermans, et al A major role for perifornical orexin neurons in the control of glucose metabolism in rats.diabetes. 58: Yi, C.-X., N. Sun, M.T. Ackermans, et al Pituitary adenylate cyclase-activating polypeptide stimulates glucose production through glycogenolysis and via the hepatic sympathetic innervation in rats. Diabetes. 59: Campfield, L.A., F.J. Smith, Y. Guisez, et al Recombinant mouse OB protein: evidence for a peripheral signal linking adiposity and central neural networks. Science 269: Obici, S., Z. Feng, G. Karkanias, D.G. Baskin & L. Rossetti. 2002a. Decreasing hypothalamic insulin receptors causes hyperphagia and insulin resistance in rats. Nat. Neurosci. 5: Obici, S., B.B. Zhang, G. Karkanias & L. Rossetti. 2002b. Hypothalamic insulin signaling is required for inhibition of glucose production. Nat. Med. 8: Van Den Hoek, A.M., C. Van Heijningen, J.P. Schroder-Van Der Elst, etal Intracerebroventricular administration of NPY hampers insulin action to suppress endogenous glucose production via the sympathetic nervous system. Diabetes 57: Klieverik, L.P., S.F. Janssen, A. van Riel, et al Thyroid hormone modulates glucose production via a sympathetic pathway from the hypothalamic paraventricular nucleus to the liver. Proc. Natl. Acad. Sci. USA 106: Fliers, E., L.P. Klieverik & A. Kalsbeek Novel neural pathways for metabolic effects of thyroid hormone. Tr. Endocrinol. Metab. 21: Musatov, S., W. Chen, D.W. Pfaff, C.V. Mobbs, et al Silencing of estrogen receptor alpha in the ventromedial nucleus of hypothalamus leads to metabolic syndrome. Proc. Natl. Acad. Sci. USA 104: Clegg, D.J., L.M. Brown, S.C. Woods & S.C. Benoit Gonadal hormones determine sensitivity to central leptin and insulin. Diabetes 55: Marks, J.L. & K. Waite Intracerebroventricular neuropeptide Y acutely influences glucose metabolism and insulin sensitivity in the rat. J. Neuroendocrinol. 9: Marks, J.L. & K. Waite Some acute effects of intracerebroventricular neuropeptide Y on insulin secretion and glucose metabolism in the rat. J. Neuroendocrinol. 8: Van Den Hoek, A.M., P.J. Voshol, B.N. Karnekamp, et al Intracerebroventricular neuropeptide Y infusion precludes inhibition of glucose and VLDL production by insulin. Diabetes 53: Unger, J., T.H. McNeill, R.T. Moxley III, et al Distribution of insulin receptor-like immunoreactivity in the rat forebrain. Neuroscience 31: Schwartz, M.W., J.L. Marks, A.J. Sipols, et al Central insulin administration reduces neuropeptide Y mrna expression in the arcuate nucleus of food-deprived lean (Fa/Fa) but not obese (fa/fa) Zucker rats. Endocrinology 128: Van Den Top, M. & D. Spanswick Integration of metabolic stimuli in the hypothalamic arcuate nucleus. Prog. Brain Res. 153: Hu, Y. & J.C. Dunbar Intracerebroventricular administration of NPY increases sympathetic tone selectively in vascular beds. Brain Res. Bull. 44: Van Dijk, G., A.E. Bottone, J.H. Strubbe & A.B. Steffens Hormonal and metabolic effects of paraventricular hypothalamic administration of neuropeptide Y during rest and feeding. Brain Res. 660: Egawa, M., H. Yoshimatsu & G.A. Bray Neuropeptide Y suppresses sympathetic activity to interscapular brown adipose tissue in rats. Am.J.Physiol.260: R328 R Pocai,A.,T.K.Lam,R.Gutierrez-Juarez,et al Hypothalamic K(ATP) channels control hepatic glucose production. Nature 434: Zarjevski, N., I. Cusin, R. Vettor, et al Chronic intracerebroventricular neuropeptide-y administration to normal rats mimics hormonal and metabolic changes of obesity. Endocrinology 133: Baran, K., E. Preston, D. Wilks, et al Chronic central melanocortin-4 receptor antagonism and central neuropeptide-y infusion in rats produce increased adiposity by divergent pathways. Diabetes 51: Mashiko, S., A. Ishihara, H. Iwaasa, et al Characterization of neuropeptide Y (NPY) Y5 receptor-mediated obesity in mice: chronic intracerebroventricular infusion of D-Trp(34)NPY. Endocrinology 144: Sainsbury, A., I. Cusin, F. Rohner-Jeanrenaud & B. Jeanrenaud Adrenalectomy prevents the obesity syndrome produced by chronic central neuropeptide Y infusion in normal rats. Diabetes. 46: Zakrzewska, K.E., A. Sainsbury, I. Cusin, et al Selective dependence of intracerebroventricular neuropeptide Y-elicited effects on central glucocorticoids. Endocrinology 140: Stafford, J.M., F. Yu, R. Printz, et al Central nervous system neuropeptide Y signaling modulates VLDL triglyceride secretion. Diabetes 57: Ollmann, M.M., B.D. Wilson, Y.K. Yang, et al Antagonism of central melanocortin receptors in vitro and in vivo by agouti-related protein. Science 278: Fan, W., D.M. Dinulescu, A.A. Butler, et al The central melanocortin system can directly regulate serum insulin levels. Endocrinology 141: Banno, R., H. Arima, M. Hayashi, et al Central administration of melanocortin agonist increased insulin sensitivity in diet-induced obese rats. FEBS Lett. 581: Obici, S., Z. Feng, J. Tan,et al Central melanocortin receptors regulate insulin action. J. Clin. Invest. 108: Ann. N.Y. Acad. Sci (2010) c 2010 New York Academy of Sciences. 125