Imbalanced Innervation of the Hypothalamic Paraventricular Nucleus by the AgRP and α-Melanocyte-Stimulating Hormone Axonal Projections in Germ-Free Mice

in this issue of the journalare published the first group of papers submitted to the Special Call for Papers on the subject “Peptides that Regulate Food Intake.” These nine papers ([7a][1], [32a][2], [37a][3], [37b][4], [41a][5], [42a][6], [43a][7], [46a][8],[49a][9]) illustrate both the breadth

Loss of GABAergic Signaling by AgRP Neurons to the Parabrachial Nucleus Leads to Starvation

Arcuate NPY Controls Sympathetic Output and BAT Function via a Relay of Tyrosine Hydroxylase Neurons in the PVN

We recently demonstrated that three arcuate nucleus-derived peptides, neuropeptide Y (NPY), agouti-related protein (AGRP), and alphaMSH, are contained in axon terminals that heavily innervate hypophysiotropic TRH neurons in the rat brain and may contribute to the altered set-point of the hypothalamo-pituitary-thyroid axis during fasting. To determine whether a similar regulatory system exists in human brain, we performed a series of immunohistochemical studies using antisera against NPY, AGRP, alphaMSH, and TRH in adult hypothalami obtained within 15 h of death. Numerous small to medium-sized, fusiform and multipolar NPY-, AGRP-, and alphaMSH-immunoreactive (-IR) cells were widely distributed throughout the rostro-caudal extent of the infundibular (arcuate) nucleus. A similar distribution pattern was found for NPY- and AGRP-IR neurons in the arcuate nucleus, whereas alphaMSH-IR cells appeared to form a separate cell population. By double labeling fluorescent immunohistochemistry, 82% of NPY neurons cocontained AGRP, and 87% of AGRP neurons coexpressed NPY. No colocalization was found between alphaMSH- and AGRP-IR neurons. NPY-, AGRP-, and alphaMSH-containing axons densely innervated the hypothalamic paraventricular nucleus and were found in close juxtaposition to TRH-synthesizing cell bodies and dendrites. These studies demonstrate that in man, the NPY-, AGRP-, and alphaMSH-IR neuronal systems in the infundibular and paraventricular nuclei are highly reminiscent of that observed in the rat and may similarly be involved in regulating the hypothalamo-pituitary-thyroid axis in the human brain.

Gene expression for agouti-related protein (AGRP), an endogenous antagonist of melanocortin receptors, has been localized to the hypothalamic arcuate nucleus, where it colocalizes with neuropeptide Y (NPY). Having reported that the NPY innervation of hypophysiotropic TRH neurons in the hypothalamic paraventricular nucleus (PVN) originates primarily from NPY-producing neurons in the arcuate nucleus, here we examined the possibility that TRH neurons in the PVN are similarly innervated by AGRP nerve terminals. Using immunohistochemistry, AGRP-containing cell bodies were found almost exclusively in the arcuate nucleus, but their projections were distributed widely in the hypothalamus, most conspicuously in the paraventricular (PVN), arcuate and dorsomedial nuclei, and the posterior hypothalamic area. Ablation of the arcuate nucleus by the neonatal administration of monosodium glutamate obliterated nearly all AGRP-immunoreactivity in the hypothalamus. In the PVN, double-labeling light and electron microscopic immunohistochemistry revealed that TRH neurons receive dense innervation by AGRP nerve terminals, with the frequent occurrence of axosomatic and axodendritic synapses (mainly of the symmetrical type). These findings provide morphological basis to hypothesize a role for AGRP in the arcuato-paraventricular pathway, in the down-regulation of the hypothalamic-pituitary-thyroid axis, which occurs as an adaptive response to starvation.

Agouti-related peptide (AgRP) neurons of the hypothalamus play a role in hunger-triggered food intake, stability of body weight, and long-term energy balance. A recent study showed that activation of the Gs-linked G protein-coupled receptors (GCPR) expressed by hypothalamic AgRP neurons promotes a sustained increase in food intake. Enhanced AgRP release has been the postulated underlying mechanism. Here, we confirmed that activation of Gs-coupled receptors expressed by AgRP neurons in the arcuate nucleus (ARC) of the hypothalamus, which is the primary brain region for the synthesis and release of AgRP, leads to increased release of AgRP in the paraventricular nucleus of the hypothalamus (PVN). We were unable to confirm changes in AgRP expression or intracellular content using traditional histological techniques. Thus, we developed an assay to measure AgRP in the extracellular fluid in the brain using large molecular weight cut-off microdialysis probes. Our technique enables assessment of brain AgRP pharmacokinetics under physiological conditions and in response to specific pharmacological interventions designed to modulate AgRP signaling.

Orexins/hypocretins are newly discovered neuropeptides synthesized by neurones located mainly in the lateral hypothalamus. They were originally believed to have an important role in the regulation of appetite. However, nerve fibres from orexin/hypocretin neurones are widely distributed in the brain suggesting that orexins/hypocretins have multiple functions. In particular, these peptides have now been implicated as important regulators of sleep, arousal and locomotor activity. Animal studies suggest these peptides are also involved in the regulation of pituitary hormone secretion and autonomic nervous system activity. The aim of this review is to describe how these fascinating peptides were discovered, what their structures are, the available evidence on their putative actions and their possible relevance to clinical medicine. Orexins (A and B), or hypocretins (1 and 2), have recently been discovered as a result of two independent modern approaches to the identification of neuropeptides. The hypocretins were discovered, using directional tag PCR subtraction, in an effort to determine abundant messenger RNAs that are selectively expressed in the rat hypothalamus (de Lecea et al., 1998). In this study, one clone (clone 35) was found to be particularly abundant in the dorsolateral hypothalamus and resulted in the identification of the prepro-hypocretin and its possible peptide products (hypocretin 1 and 2). These peptides were named 'hypocretins' (hypothalamic incretins) because of some amino acid sequence homology to the peptide hormone secretin. The orexins were discovered through the reverse pharmacology approach in search of endogenous ligands for 'orphan' G protein coupled receptors (GPCRs; Sakurai et al., 1998). Transfectant cell lines, expressing the orphan GCPR HFGAN72 (now orexin-A/hypocretin-1 receptor, OX1R/Hcrt1R) cDNA, were challenged with chromatographic fractions derived from rat brain extracts (Fig. 1). The fractions that elicited an increase in cytoplasmic Ca2+ levels in cells were further purified and analysed to identify and isolate the peptide orexin A. A smaller activity peak was identified as orexin B and led to a search for a different orphan receptor now known as the orexin-2/hypocretin-2 receptor (OX2R/Hcrt2R). The orexins ('stimulators of appetite') were named because orexin neuronal cell bodies were found to be largely confined to the lateral hypothalamus, an area classically described as the 'feeding' centre, and based on observations that when administered into the cerebral ventricles (intracerebroventricular; ICV), orexins potently increased food intake in rats (Sakurai et al., 1998). The 'reverse' pharmacology approach to identification of ligands for orphan G protein-coupled receptors. In the discovery of orexins, transfectant cells (HEK-293) expressing the orphan receptor (HFGAN72) were challenged with chromatographic fractions derived from rat brain extracts. The observed response was a rise in cytoplasmic calcium. Hypocretin-1 and -2 are essentially the same peptides as orexin A and B, respectively (Fig. 2) and the prepro-orexin and prepro-hypocretin genes are the same. Much confusion has occurred because the structure of biologically active hypocretin-1 was not fully determined when the hypocretins were first described. Thus, hypocretin-1, as originally reported (Fig. 2), is longer than orexin A, lacks the intrachain disulphide bonds and carboxy-terminal amidation, and is inactive at the orexin/hypocretin receptors (Smart et al. 2000). It is not surprising that in the original description (de Lecea et al., 1998) only hypocretin-2 was reported to have neuroexcitatory activity since its structure (only different from orexin B by one amino acid), unlike hypocretin-1, was the biologically active form. Whether we should use the name 'hypocretins' or 'orexins' for these neuropeptides is a matter of debate. These peptides have only weak homology to secretin and it is unlikely that they belong to the secretin family of peptides; therefore the name 'hypocretins' may be inappropriate. On the other hand, these peptides are now known to have multiple functions beside the regulation of appetite (Fig. 3) thus making the name 'orexins' confusing. In this review, we will refer to the biologically active peptide products of prepro-orexin/hypocretin as orexin A/Hcrt-1 and Orexin B/Hcrt-2. Schematic representation of the structure of the prepro-orexin/hypocretin gene, mRNA and its prepro-peptide product, and the amino acid structures of orexins (A and B) and hypocretins (1 and 2; as originally described). The amino acids underlined are shared between orexin A and B (and between hypocretin 1 and 2). The main reported actions of orexins/hypocretins. The human prepro-orexin/hypocretin gene is located on chromosome 17q21 (Sakurai et al., 1998). The gene, spanning 1432 bp, consists of 2 exons and 1 intron (Fig. 2; Sakurai et al., 1999). Orexins/hypocretins are derived from a 131 amino acid human (130 amino acids in the rat) precursor prepro-orexin/hypocretin. The first 33 amino acids of prepro-orexin/hypocretin have the characteristics of a signal sequence. Orexin A/Hcrt-1 is a 33 amino acid carboxy-amidated peptide of 3562 Da with an N-terminal pyroglutamyl residue and two intrachain disulphide bonds (Fig. 2). The human orexin A/Hcrt-1 sequence is identical to the mouse, rat, bovine and porcine orexin A/Hcrt-1. Orexin B/Hcrt-2 is also C-terminally amidated, but is a linear peptide of 28 amino acids with a molecular weight of 2937 Da. Human orexin B/Hcrt-2 has two amino acid substitutions compared with rodent orexin B/Hcrt-2 and one substitution compared to porcine orexin B/Hcrt-2. There is 46% amino acid sequence identity between orexin B/Hcrt-2 and orexin A/Hcrt-1 (Fig. 2). Orexin A/Hcrt-1, but not orexin B/Hcrt-2, is lipophilic and crosses the blood–brain barrier (Kastin & Akerstrom, 1999). The OX1R/Hcrt1R receptor has structural homology to several neuropeptide receptors including the NPY Y2 receptor, TRH receptor, CCK-A receptor and NK2 neurokinin receptor. Orexin A/Hcrt-1 has two to three times greater affinity than orexin B/Hcrt-2 for the human OX1R/Hcrt1R receptor. Both orexin A/Hcrt-1 and orexin B/Hcrt-2 have similar affinities for the human OX2R/Hcrt2R receptor. Using oligonucleotide probes, OX1R/Hcrt1R receptor mRNA has been reported to be most abundant in the ventromedial hypothalamus, but has also been detected in the tenia tecta, hippocampus, dorsal raphe and locus coeruleus (Trivedi et al., 1998). OX2R/Hcrt2R receptor mRNA is expressed in the paraventricular hypothalamic nucleus, the subthalamic and thalamic nuclei, the septum, the cerebral cortex, nucleus accumbens, anterior pretectal nucleus and several regions in the medulla oblongata. The subcellular location of these orexin/hypocretin receptors is interesting. Immunohistochemical studies have shown that the location of OX1R/Hcrt1R is mainly cytosolic while that of OX2R/Hcrt2R is mainly nuclear (G. Bewick, unpublished data). OX1R/Hcrt1R and OX2R/Hcrt2R have been detected in the pituitary gland (Date et al. 2000) while OX2R/Hcrt2R mRNA has also been detected in the adrenal medulla (Lopez et al., 1999), where it may alter catecholamine secretion, but the physiological relevance of these observations is at present unclear. Orexins are not found in plasma. However, it is possible that they may be present in nerve endings, in sufficiently low density to be undetectable by radioimmunoassay, in peripheral tissues. Alternatively, peripheral orexin receptors may function as receptors for another ligand. Curiously, orexin/hypocretin neurones appear to be labelled by an antiserum raised against ovine prolactin (Risold et al., 1999). A small proportion of orexin/hypocretin neurones are immunopositive for galanin (Hakansson et al., 1999). Orexin A/Hcrt-1 and Orexin B/Hcrt-2 are found in secretory vesicles at neuronal synapses and are neuroexcitatory (de Lecea et al., 1998; van Den Pol et al., 1998). These peptides cause a phospholipase C-mediated release of calcium from intracellular stores, with subsequent calcium influx in vitro (Smart et al., 1999; Lund et al. 2000). Orexin/hypocretin immunoreactivity and immunoreactive fibres are widely distributed throughout the central nervous system (Peyron et al., 1998; Mondal et al., 1999; Nambu et al., 1999; Taheri et al., 1999), but have only occasionally been reported in peripheral tissues (Kirchgessner & Liu, 1999; Date et al. 2000). Orexin/hypocretin immunoreactive fibres are densely distributed in the hypothalamus, septum, thalamus, brainstem and spinal cord (Peyron et al., 1998; Cutler et al., 1999; Date et al., 1999; Nambu et al., 1999). Orexin fibres innervating neurones in the locus coeruleus appear to represent the greatest peptidergic input into this pontine region (Horvath et al., 1999a). The locus coeruleus is a major monoaminergic site within the central nervous system involved in the regulation of arousal. Elevated c-fos (a measure of neuronal activation) immunoreactivity in response to orexins (Date et al., 1999; Edwards et al., 1999; Mullett et al. 2000) has been detected in hypothalamic areas involved in food intake, neuroendocrine regulation and sleep and circadian rhythms (the lateral hypothalamus, the posterior and dorsomedial hypothalamus, anterior hypothalamus, the perifornical, arcuate and paraventricular nuclei), in the lateral septal area, the central nucleus of the amygdala, the shell of the nucleus accumbens, the bed nucleus of the stria terminalis, and the nucleus of the solitary tract (involved in autonomic and visceral regulation). Orexin neurones in the lateral hypothalamus have reciprocal connections with neuropeptide Y (NPY) and agouti-related peptide (AgRP) neurones in the arcuate nucleus (Broberger et al., 1998; Horvath et al., 1999b). The lateral hypothalamus has classically been considered to be the 'feeding' centre with lesions of this area being associated with a marked decrease in food intake. The ventromedial hypothalamus is known as the 'satiety' centre since lesions of this area result in increased food intake and obesity. These early experiments have been clarified by the discovery of neuropeptide mediators of food intake (Table 1; Kalra et al., 1999). Several neuropeptides synthesized in the arcuate nucleus of the mediobasal hypothalamus are mediators of food intake including NPY, the pro-opiomelanocortin derivative α-melanocyte-stimulating hormone (α-MSH), AgRP and the peptide product of cocaine and amphetamine-regulated transcript (CART). When administered ICV to rats, NPY and AgRP stimulate food intake, while α-MSH and CART reduce food intake. These neuropeptides are regulated by leptin, a cytokine that signals the extent of fat stores to the central nervous system (Meister 2000). The lateral hypothalamus contains neurones, distinct from orexin/hypocretin neurones, expressing melanin-concentrating hormone (MCH), a central stimulator of food intake (Elias et al., 1998). The anatomical location of neurones expressing orexins/hypocretins, led to the hypothesis that they are important regulators of food intake. When originally administered into the lateral ventricle of rats, both orexin A/Hcrt-1 and orexin B/Hcrt-2 dose-dependently and potently stimulated food intake in rats and prepro-orexin/hypocretin mRNA was shown to be upregulated with fasting (Sakurai et al., 1998). The stimulation of food intake by orexins/hypocretins has been difficult to consistently repeat in several laboratories, particularly for orexin B/Hcrt-2 (Edwards et al., 1999). While orexin A/Hcrt-1 may increase food intake in the first 4 hours after ICV administration, it actually decreases food intake in the subsequent 20 hours (Taheri et al. 2000a). Furthermore, chronic administration of orexin A/Hcrt-1 does not result in obesity (Haynes et al., 1999; Yamanaka et al., 1999). It may be that orexins/hypocretins are not as important in food intake as originally believed, or that they may be important in food intake only in particular circumstances (e.g. in response to hypoglycaemia and/or in the regulation of circadian food intake). Orexin/hypocretin neurones have leptin receptors and are immunoreactive for STAT3, a transcription factor activated by leptin (Hakansson et al., 1999). However, the interaction of leptin with orexins/hypocretins is unlike the interaction observed with the potently orexigenic peptide NPY. NPY mRNA is upregulated in leptin-deficient ob/ob and leptin receptor deficient db/db mice, while prepro-orexin/hypocretin mRNA is downregulated, but increases with starvation in these animals (Yamamoto et al., 1999; Yamamoto et al. 2000). Although one group has recently reported moderately lower prepro-orexin/hypocretin mRNA in obese Zucker rats (Cai et al. 2000), which have defective leptin receptor action compared to their lean controls, we have noted no differences in prepro-orexin/hypocretin mRNA or orexin A/Hcrt-1 immunoreactivity in these strains (Taheri et al., in press). It is therefore unlikely that orexins/hypocretins are directly involved or are altered in the obesity of Zucker rats. These findings do not mean that orexins/hypocretins are not regulated by leptin since the effects of the other neuroendocrine abnormalities present in these animals may mask the effects of leptin deficiency on orexin/hypocretin neurones. Orexins/hypocretins appear to have an important role in the regulation of metabolic rate. Using indirect calorimetry, orexin A/Hcrt-1, but not orexin B/Hcrt-2, increased metabolic rate (Lubkin & Stricker-Krongrad, 1998). Mice overexpressing prepro-orexin/hypocretin have reduced body weight despite increased food intake since they also have a concomitant increase in metabolic rate (Inui 2000). Mice with targeted disruption of the prepro-orexin/hypocretin are hypophagic with reduced metabolic rate, but are prone to diet induced obesity, presumably due to their lower metabolic rate (Willie et al., in press). Orexin/hypocretin fibres innervate hypothalamic regions involved in the regulation of pituitary hormone release. Orexin A/Hcrt-1 releases several neuropeptides and releasing factors from mediobasal hypothalamic explants in vitro (Russell et al. 2000). Centrally administered orexin A/Hcrt-1 alters plasma prolactin, growth hormone and corticosterone. In the studies that originally reported these endocrine responses, orexin A/Hcrt-1 was shown to increase arousal and to activate neurones in the locus coeruleus (Hagan et al., 1999). The profound inhibition of plasma prolactin by ICV orexin A/Hcrt-1 is partially attenuated by the administration of the dopamine receptor antagonist domperidone suggesting that while in the presence of domperidone centrally administered TRH, neurotensin and vasopressin lose their prolactin lowering effect, the effect of orexin A/Hcrt-1 is partly independent of dopamine (Russell et al. 2000). Orexin A/Hcrt-1 increases plasma ACTH and corticosterone and there is increased c-fos mRNA in the paraventricular nucleus (PVN) of the hypothalamus where corticotrophin-releasing factor (CRF) neurones are located (Hagan et al., 1999; Ida et al. 2000; Kuru et al. 2000). It is therefore likely that orexin A/Hcrt-1 is an important regulator of the hypothalamo-pituitary-adrenal axis, which is consistent with its effect on increasing arousal. The mechanism through which orexin A/Hcrt-1 inhibits GH secretion remains to be determined, but may involve the release of somatostatin. Orexins/hypocretins may regulate LH secretion via GnRH neurones in the hypothalamus. In one study, the effects of orexin A/Hcrt-1 and orexin B/Hcrt-2 on LH secretion were evaluated in ovariectomized (ovx) and ovarian steroid-treated ovx rats (Pu et al., 1998). Central injection of orexin A/Hcrt-1 or orexin B/Hcrt-2 rapidly stimulated LH secretion in ovx animals pretreated with oestradiol and progesterone. However, these peptides inhibited LH release in unprimed ovx rats. Interestingly, we have observed changes in orexin A/Hcrt-1 immunoreactivity in the rat CNS with the oestrous cycle (Taheri & Russell, unpublished data). The available evidence for the importance of orexins/hypocretins in the hypothalamic regulation of pituitary hormone release is derived from animal experiments in which these peptides have been administered into the cerebral ventricles. This experimental approach, however, will result in the activation of several circuits in which orexins/hypocretins are involved and any hormonal changes in the plasma will represent the net effect of the circuits activated. It is therefore not surprising that intracerebroventricular administration of orexins/hypocretins stimulates the HPA axis, while paradoxically inhibiting prolactin release. The neuronal circuitry through which orexins/hypocretins may regulate pituitary hormone release remain to be defined. Further, the physiological circumstances in which orexins/hypocretins regulate the neuroendocrine system and the existence of any negative or positive feedback loops remain to be determined. The distribution of orexin/hypocretin immunoreactive fibres within the CNS raised the possibility that they may be involved in the regulation of sleep, arousal and activity. Targeted disruption of the prepro-orexin gene in mice results in a phenotype with features similar to the human narcolepsy syndrome (Chemelli et al., 1999). In the same report, modafinil, a drug used in the treatment of narcolepsy through an unknown mechanism, was shown to activate orexin/hypocretin neurones. Narcolepsy is a debilitating sleep disorder characterized by excessive daytime sleepiness, cataplexy (loss of muscle tone in response to emotional stimuli) and disturbance of rapid eye movements (REM) sleep. Canine models of narcolepsy, which inherit narcolepsy in an autosomal recessive fashion with full penetrance through the canarc-1 gene, have been shown to have mutations in the OX2/Hcrt-2 receptor gene (Lin et al., 1999). In rats, ICV administration of orexin A/Hcrt-1 increases arousal while reducing REM sleep and prolonging the latency to the first occurrence of REM sleep (Piper et al. 2000). These studies suggested that orexins have a role in the regulation of arousal and the sleep-wake cycle. Indeed, orexin A/Hcrt-1 immunoreactivity shows diurnal variation in areas of the brain involved in the regulation of sleep, arousal and circadian hormone release (Taheri et al. 2000a). When orexin A/Hcrt-1 immunoreactivity was measured in the cerebrospinal fluid of patients with narcolepsy, it was shown that the majority of the patients examined had undetectable orexin A/Hcrt-1 compared to controls (Nishino et al. 2000; Taheri et al. 2000b). This is now further supported by undetectable prepro-orexin mRNA in autopsy specimens from patients with narcolepsy (Peyron et al. 2000). It is likely that in human narcolepsy, unlike canine narcolepsy, there is an abnormality in orexin production or prepro-orexin processing rather than mutations in the OX2/Hcrt-2 receptor gene. Since orexin A/Hcrt-1 abnormalities have not been detected in all patients diagnosed with narcolepsy suggests that there may be different subsets of patients with narcolepsy. Interestingly, little is known about endocrine abnormalities in both canine and human narcolepsy except a greater propensity towards type 2 diabetes mellitus in human narcolepsy (Honda et al., 1986). It is not surprising that orexins/hypocretins are involved in the regulation of hormones such as prolactin, GH and corticosterone, which are intimately associated with sleep and arousal. Orexins have been implicated in the CNS regulation of gastric acid secretion (vagally mediated), sympathetic activation, cardiovascular function and drinking behaviour. Both orexin A/Hcrt-1 and orexin B/Hcrt-2, when injected ICV increase blood pressure and heart rate in rats (Samson et al., 1999; Shirasaka et al., 1999). ICV orexin A/Hcrt-1 increases water intake, while prepro-orexin/hypocretin mRNA increases with water deprivation (Kunii et al., 1999). In the spinal cord, orexins/hypocretins may influence the sensory and autonomic nervous systems (van Den Pol, 1999). There are several reports on the effect of orexins/hypocretins on peripheral tissues such as the gastrointestinal tract, the pancreas and adrenal gland (Kirchgessner & Liu, 1999; Malendowicz et al., 1999; Nowak et al. 2000). Orexins/hypocretins are good examples of how modern molecular biology techniques have contributed to the discovery and greater understanding of the physiological actions of neuropeptides. It is remarkable how quickly these peptides have made the transition from basic scientific research to clinical medicine. Since the secretion of most pituitary hormones is intimately linked to sleep and arousal, it is likely that under basal conditions, orexins/hypocretins link sleep and arousal to pituitary hormone release. It now remains to study the orexin/hypocretin neuronal circuitry in conditions that disturb this basal state, for example, ageing, sleep loss, night or shift work, jet lag, affective disorders and endocrine diseases (e.g. Cushing's syndrome). Manipulation of the orexin/hypocretin ligand-receptor system may prove therapeutically useful not only in the treatment of sleep disorders such as narcolepsy and insomnia, but also in the treatment of several medical and psychiatric disorders associated with sleep disturbance. However, a greater understanding of the role of the individual peptides and receptors is required combined with an effort to tease out the neuronal circuits involved. Further study of these fascinating peptides may also clarify how hormone release is linked to sleep and arousal. Shahrad Taheri is funded by the Wellcome Trust. We would like to thank members of our laboratory who have contributed to the study of orexins/hypocretins and Dr Mignot, Dr Yanagisawa and Dr Chemelli for providing us with information regarding their work.

To explore the effect of refeeding on recovery of TRH gene expression in the hypothalamic paraventricular nucleus (PVN) and its correlation with the feeding-related neuropeptides in the arcuate nucleus (ARC), c-fos immunoreactivity (IR) in the PVN and ARC 2 h after refeeding and hypothalamic TRH, neuropeptide Y (NPY) and agouti-related protein (AGRP) mRNA levels 4, 12, and 24 h after refeeding were studied in Sprague-Dawley rats subjected to prolonged fasting. Despite rapid reactivation of proopiomelanocortin neurons by refeeding as demonstrated by c-fos IR in ARC alpha-MSH-IR neurons and ventral parvocellular subdivision PVN neurons, c-fos IR was present in only 9.7 +/- 1.1% hypophysiotropic TRH neurons. Serum TSH levels remained suppressed 4 and 12 h after the start of refeeding, returning to fed levels after 24 h. Fasting reduced TRH mRNA compared with fed animals, and similar to TSH, remained suppressed at 4 and 12 h after refeeding, returning toward normal at 24 h. AGRP and NPY gene expression in the ARC were markedly elevated in fasting rats, AGRP mRNA returning to baseline levels 12 h after refeeding and NPY mRNA remaining persistently elevated even at 24 h. These data raise the possibility that refeeding-induced activation of melanocortin signaling exerts differential actions on its target neurons in the PVN, an early action directed at neurons that may be involved in satiety, and a later action on hypophysiotropic TRH neurons involved in energy expenditure, potentially mediated by sustained elevations in AGRP and NPY. This response may be an important homeostatic mechanism to allow replenishment of depleted energy stores associated with fasting.

To characterize the impact of gut microbiota on host metabolism, we investigated the multicompartmental metabolic profiles of a conventional mouse strain (C3H/HeJ) (n=5) and its germ-free (GF) equivalent (n=5). We confirm that the microbiome strongly impacts on the metabolism of bile acids through the enterohepatic cycle and gut metabolism (higher levels of phosphocholine and glycine in GF liver and marked higher levels of bile acids in three gut compartments). Furthermore we demonstrate that (1) well-defined metabolic differences exist in all examined compartments between the metabotypes of GF and conventional mice: bacterial co-metabolic products such as hippurate (urine) and 5-aminovalerate (colon epithelium) were found at reduced concentrations, whereas raffinose was only detected in GF colonic profiles. (2) The microbiome also influences kidney homeostasis with elevated levels of key cell volume regulators (betaine, choline, myo-inositol and so on) observed in GF kidneys. (3) Gut microbiota modulate metabotype expression at both local (gut) and global (biofluids, kidney, liver) system levels and hence influence the responses to a variety of dietary modulation and drug exposures relevant to personalized health-care investigations.

The melanocortin system coordinates the maintenance of energy balance via the regulation of both food intake and energy expenditure. Leptin, a key adipogenic hormone involved in the regulation of energy balance is thought to act by stimulating production, in the hypothalamic arcuate nucleus, of alpha-melanocyte stimulating hormone (alphaMSH), a potent agonist of MC3/4 melanocortin receptors located in the paraventricular nucleus of the hypothalamus. Additionally leptin inhibits release of agouti-related protein (AgRP), an MC4R antagonist. During periods of caloric restriction, weight loss is not sustained because compensatory mechanisms, such as reduced resting metabolic rate (RMR) are brought into play. Understanding how these compensatory systems operate may provide valuable targets for pharmaceutical therapies to support traditional dieting approaches. As circulating leptin is reduced during caloric restriction, it may mediate some of the observed compensatory responses. In addition to decreases in circulating leptin levels, circulating AgRP is increased during fasting in rodents while alphaMSH is decreased. As central administration of AgRP depresses metabolism, we hypothesised that the peripheral rise in AgRP might be involved in signalling the depression of RMR during food restriction. We hypothesised that changes in plasma AgRP and alphaMSH may coordinate the regulation of changes in energy expenditure acting through central MC4 melanocortin receptors via the sympathetic nervous system.We show here that acute peripherally administered AgRP at supra-physiological concentrations in both lean (C57BL/6) and obese leptin-deficient (ob/ob) mice does not depress RMR, possibly because it crosses the blood-brain barrier very slowly compared with other metabolites. However, in vitro AgRP can decrease leptin secretion, by approximately 40%, from adipocytes into culture medium and may via this axis have an effect on energy metabolism during prolonged caloric restriction. In contrast, peripheral [Nle4,D-Phe7]-alpha MSH produced a large and sustained increase in resting energy expenditure (0.15 ml O2/min; P < 0.05) with a similar response in leptin-deficient ob/ob mice (0.27 ml O2/min) indicating that this effect is independent of the status of leptin production in the periphery. In both cases respiratory exchange ratio and the levels of energy expended on spontaneous physical activity were unaffected by the administration of peripheral [Nle4,D-Phe7]-alpha MSH. In conclusion, alphaMSH analogues that cross the blood-brain barrier may significantly augment dietary restriction strategies by sustaining elevated RMR.

Decision letter: Identification of a GABAergic neural circuit governing leptin signaling deficiency-induced obesity

Editor's evaluation: Identification of a GABAergic neural circuit governing leptin signaling deficiency-induced obesity

Agouti-related protein (AgRP) is coexpressed with neuropeptide Y (NPY) in a population of neurons in the arcuate nucleus (ARC) of the hypothalamus and stimulates food intake for up to 7 days if injected intracerebroventricularly. The prolonged food intake stimulation does not seem to depend on continued competition at the melanocortin-4 receptor (MC4R), because the relatively specific MC4R agonist MTII regains its ability to suppress food intake 24 h after AgRP injection. Intracerebroventricular AgRP also stimulates c-Fos expression 24 h after injection in several brain areas, so the neurons exhibiting delayed Fos expression might be particularly important in feeding behavior. Thus we aimed to identify the neurochemical phenotype of some of these neurons in select hypothalamic areas, using double-label immunohistochemistry. AgRP-injected rats ingested significantly more chow (10.2 +/- 0.6 g) vs. saline controls (3.4 +/- 0.7 g) in the first 9 h (light phase) after injection. In the lateral hypothalamus (particularly the perifornical area) 23 h after injection, AgRP induced significantly more Fos vs. saline in orexin-A (OXA) neurons (25.6 +/- 4.9 vs. 4.8 +/- 3.1%), but not in melanin-concentrating hormone (MCH) or cocaine- and amphetamine-regulated transcript (CART) neurons. In the ARC, AgRP induced significantly more Fos in CART (40.6 +/- 5.9 vs. 13.4 +/- 1.8%) but not NPY neurons. In the paraventricular nucleus, there was no significant difference in Fos expression induced by AgRP vs. saline in oxytocin and CART neurons. We conclude that the long-lasting hyperphagia induced by AgRP is correlated with and possibly partially mediated by hyperactive OXA neurons in the lateral hypothalamus and CART neurons in the ARC, but not by NPY and MCH neurons. The substantial increase in light-phase food intake by AgRP supports a role for the arousing effects of OXA. Activation of CART neurons in the ARC (which likely coexpress proopiomelanocortin) could indicate attempts to activate counterregulatory decreases in food intake.

The hypothalamus controls food intake [1]. It regulates food intake by integrating information on nutritional status from periphery through nutrients, hormones, and autonomic nervous system. Previous studies revealed that numerous neuropeptides are expressed in the hypothalamus and contribute to the neural networks that regulate food intake. Among them, the melanocortin system—which consists of proopiomelanocortin (POMC), agouti-related peptides (AgRP), and melanocortin receptors—is the most fundamental system regulating food intake. POMC is a precursor for alpha-melanocyte-stimulating hormone (a-MSH), which is an agonist for melanocortin type 4 receptor (MC4R), and AgRP is an inverse agonist for MC4R and counteracts the function of a-MSH. In fact, POMC-null mice [2] and MC4R-null mice [3] are obese, and cases of highly obese humans with MC4R gene mutations have been reported [4–6]. Within the hypothalamus, the arcuate nucleus (ARC), paraventricular nucleus (PVN), and lateral hypothalamic (LH) area are centers for controlling food intake. ARC is the ‘‘first-order center’’ for regulating food intake. It is located at the mediobasal hypothalamus, where the blood– brain barrier is quite permissive, making ARC the place for sensing nutrients and hormone levels. ARC contains two types of neurons: anorexigenic POMC and orexigenic AgRP. These neurons project axons to the ‘‘second-order centers,’’ which are located in PVN and LH, and competitively regulate the activity of these nuclei [7, 8]. The most-studied feeding-related hormone is leptin. In the hypothalamic neurons, leptin activates the Janus kinase 2–signal transducer and activator of transcription 3 (JAK2STAT3) pathway, leading to nuclear translocation of phosphorylated STAT3. STAT3 suppresses food intake by transactivating the anorexigenic Pomc gene and transrepressing the orexigenic Agrp gene. Insulin is also known as a central regulator for food intake. Neuron-specific insulin receptor knockout mice exhibit increased food intake and obesity [9]. Insulin signaling is transmitted from the insulin receptor to phosphoinositide 3-kinase (PI3K), which subsequently activates a serine/threonine kinase protein kinase B (Akt). FoxO1 is a transcription factor and one of the substrates for Akt. Phosphorylation of FoxO1 by Akt results in cytoplasmic shuttling from the nucleus, thereby inactivating FoxO1 as a transcription factor [10]. FoxO1 is expressed in ARC AgRP and POMC neurons, and FoxO1 in these neurons are located in the nucleus under fasted condition but is shuttled to cytoplasm by feeding [11, 12]. Overexpression of constitutively active FoxO1 in the mediobasal hypothalamus of rats by adenoviral microinjection leads to loss of feeding inhibitory effect of leptin and results in body weight gain [11]. Hypothalamus-specific constitutively active FoxO1 knockin mice also have increased food intake and decreased energy expenditure, and consequently these mice develop obesity [13]. Silent mating type information regulation 2 homolog (sirtuin 1; Sirt1) is a nicotinamide adenine dinucleotide (NAD)-dependent deacetylase and serves as an energy sensor [14]. Sirt1 is the mammalian ortholog of Sir2, which is crucial for caloric-restriction-induced longevity [15–17]. Sirt1 is expressed in POMC and AgRP neurons in ARC and T. Kitamura (&) T. Sasaki Metabolic Signal Research Center, Institute for Molecular and Cellular Regulation, Gunma University, 3-39-15 Showa-machi, Maebashi, Gunma 371-8512, Japan e-mail: kitamura@gunma-u.ac.jp

Growth Hormone Receptor Deletion Reduces the Density of Axonal Projections from Hypothalamic Arcuate Nucleus Neurons