Ghrelin

The gastrointestinal peptide hormone ghrelin was discovered in 1999 as the endogenous ligand of the growth hormone secretagogue receptor. Increasing evidence supports more complicated and nuanced roles for the hormone, which go beyond the regulation of systemic energy metabolism.

In recent years, ghrelin has been found to have a plethora of central and peripheral actions in distinct areas including learning and memory, gut motility and gastric acid secretion, sleep/wake rhythm, reward seeking behavior, taste sensation and glucose metabolism.

In the late 1970s, the work of Cyril Bowers and Frank Momany led to the generation of a group of synthetic opioid peptide derivatives that promoted the release of GH from the anterior pituitary [43,44] . The molecules, which Bowers and Momany referred to as GH releasing peptides (GHRPs), were generated by the chemical modification of met-enkephalin and included growth hormone releasing peptide (GHRP)-6, GHRP-2, and hexarelin [45] . Initially, it was thought that these GHRPs acted only on the pituitary, but soon it became clear that that they also acted on the hypothalamic arcuate nucleus (ARC) [46] , specifically on GH-releasing hormone (GHRH) neurons [41] . The mechanism by which these molecules promoted the release of GH was unknown, but it was distinct from that of the GHRH/somatostatin pathway [46–49] . In 1996, the GHS clinical candidate MK0677 was employed by Roy Smith and Lex van der Ploeg to clone the GH secretagogue receptor (GHSR1a) [50] , at which GHSs and GHRPs were shown to be agonists. In humans, the GHSR1 gene codes for the full-length G-protein coupled seven transmembrane protein GHSR1a, but a truncated isoform (GHSR1b), which has a wide tissue distribution, is also transcribed [51] . GHSR1a has been shown to homodimerize, but the possibility has been raised that GHSR1a and GHSR1b also heterodimerize [52,53] and that the heterodimer inhibits the activation of GHSR1a [53] . GHSR1a is expressed predominantly in the anterior pituitary gland, pancreatic islets, adrenal gland, thyroid, myocardium, ARC, hippocampus, the substantia nigra pars compacta (SNpc), ventral tegmental area (VTA), and raphe nuclei [40,51] . In the ARC, in addition to being expressed in GHRH neurons, GHSR1a is colocalized in neurons that express neuropeptide Y ( Npy ) and Agouti related peptide ( Agrp ), which regulate food intake and satiety [42] . Along with the observation that GHRP-6 induced activation of GHSR1 and increased c-Fos expression in NPY neurons [41] , these data suggested the presence of an unknown but endogenous ligand for GHSR1, one that might regulate systemic metabolism. In the years that followed, extensive research efforts were aimed at identifying the endogenous ligand for GHSR1. The ligand remained elusive until 1999 when Kojima and colleagues identified the cognate agonist for GHSR1. Purified from rat stomach extracts, the 28 amino acid peptide was named ‘ghrelin’, a name originating from ‘ghre’, the Proto-Indo-European root of the word 'grow' [1] .

Intriguingly, the lipids used for ghrelin activation are, at least in part, directly recruited from the pool of ingested dietary lipids [61,63] in a process that may take advantage of the fact that ghrelin-producing X/A-like cells are located within gastric oxyntic glands. A significant number of these cells are apposed to the stomach lumen, allowing for direct access to a supply of dietary lipids [64] . Furthermore, the preferred fatty acid substrates for GOAT are derived from medium-chain-triglycerides, which can be directly absorbed into the circulation without being broken down by lipases and bile acids [65] . Despite this evidence, the relative contribution of de novo synthesized fatty acids in comparison to those directly derived from the diet as substrate of GOAT for ghrelin acylation remains unknown. Mutation studies in the region of the acylated serine 3 have revealed that glycine 1, serine 3, and phenylalanine 4 are critical components of the recognition sequence for GOAT, whereas serine 2, leucine 5, serine 6 and proline 7 seem to be less important [62] . Biochemically, GOAT appears to have two critical substrates, des-acyl ghrelin and short-to mid-chain fatty acids thioesterified with Coenzyme A. Cells expressing both ghrelin and GOAT synthesize serine 3 acyl-ghrelin, with the acyl moiety precursors derived from fatty acids ranging from acetate (C2) to tetradecanoic acid (C14) [56] . The length of the fatty acid used for ghrelin acylation seems to be of importance for ghrelin's metabolic effects, as alterations in the fatty acid length result in differential activation of GHSR1a in vitro and alter ghrelin's effect on food intake and adiposity in vivo [66] . Thus, modulation of the acyl side-chain may also represent an interesting therapeutic control point for future interventions. Octanoyl- and decanoyl-modified ghrelin forms are the optimal ligands for activation of the GHSR1a [57,62] . In vitro studies recreating the acyl-modification of ghrelin with des-acyl ghrelin peptides, fatty acid CoA esters, and GOAT containing microsomes define the substrate specificity for GOAT. These studies support the idea that GOAT requires fatty acid substrates as high energy fatty acid CoA thioesters and that the amino acid sequence GXSFX, where G, X, S, and F correspond to unblocked amino terminal glycine (G), any amino acid (X), serine (S) and phenylalanine (F), respectively, is sufficient as a substrate for GOAT acylation [67] . The structural constraints defined by this amino acid motif appear specific only for ghrelin and suggest that ghrelin may be the principal peptide substrate for GOAT. Most recent studies comparing the in vitro selectivity of hexanoyl- and octanoyl-CoA substrates suggest that GOAT may actually prefer hexanoyl CoA over octanoyl CoA substrates, highlighting the importance of the specific fatty acid metabolism in acyl ghrelin producing cells, responsible for producing circulating levels of octanoyl and decanoyl ghrelin [67] .

4.1 Clinical pharmacology studies on ghrelin's effect on energy metabolism Numerous human studies have evaluated the effect of ghrelin and its analogs on GH secretion [69–72] , food intake [11,34,73–79] , body weight [34,78,80] energy expenditure [81–83] , glucose homeostasis [84–89] , and gastrointestinal motility [90–93] . Peripheral ghrelin or GHRP-2 administration reliably induces the sensation of hunger and increases food intake in lean, obese, healthy and malnourished individuals [94] . Interestingly, iv administration of ghrelin in healthy volunteers increases neural activity in specific brain regions in response to pictures of food. Endogenous fasting ghrelin is positively related to hunger-modulated activity in the hypothalamus, amygdala, and prefrontal cortex in response to palatable food stimuli [95,96] . Activation of these reward centers by ghrelin suggests enhancement of food consumption is a more complex mechanism than the physical sensations of hunger or satiety [97,98] . Interestingly, these ghrelin-brain activity relationships are absent in women with anorexia nervosa, suggesting the possibility of CNS-regulated ghrelin resistance in these individuals [95] .

Both ghrelin and its receptor are widely expressed in multiple regions of the brain [3,108,109] and in peripheral tissues, such as the intestine [110] , pituitary [111,112] , kidney [108,113] , lung [108,114] , heart [110,115,116] , ovaries [108] , and pancreatic islets [17,40] . Expression in pancreatic islets is consistent with a series of human studies showing increased plasma levels of glucose and decreased plasma levels of insulin following ghrelin administration [84–86,88,89] . GHSRs are also expressed by α-cells of the pancreatic islet and likely contribute to the ability of ghrelin to directly stimulate glucagon secretion [117] . Ghrelin inhibits insulin secretion in most animal studies [118–121] , and blockade of pancreatic-derived ghrelin enhances insulin secretion and ameliorates the development of diet-induced glucose intolerance [122] . Supporting these data is the finding that plasma ghrelin and insulin levels seem to be negatively correlated, as the two hormones exhibit reciprocal changes during the day and during a hyperinsulinemic, euglycemic clamp [37,123] . Furthermore, continuous ghrelin infusion for 65 min suppresses glucose-stimulated insulin secretion and impairs glucose tolerance in healthy individuals [124] . In line with this, pharmacological inhibition of GOAT improves glucose disposal by stimulating the release of insulin [60] . The reciprocal relationship of ghrelin and insulin is supported by epidemiologic studies showing an inverse relationship between circulating ghrelin levels and indexes of insulin resistance [39] . A single iv dose of ghrelin significantly increases plasma glucose levels followed by a reduction in fasting insulin levels in lean [84] and obese subjects with or without polycystic ovarian syndrome [87] , suggesting inhibition of insulin secretion. Intriguingly, ghrelin's suppression of insulin secretion in pancreatic β-cells is mediated by a non-canonical GHSR1a signaling pathway in which Gα i rather than Gα q is coupled to the receptor [125] . This modified signaling is dependent upon agonist mediated molecular interactions between GHSR1a and somatostatin receptor subtype-5 (SST5) and the formation of GHSR1a:SST5 heterodimers [126] . Conversely, some studies suggest that ghrelin has positive trophic activity, protecting from β-cell damage in experimental models of type 1 diabetes [127,128] . However, the onset of type 1 diabetes is associated with decreased circulating ghrelin levels [129,130] .

Ghrelin is the only circulating hormone that, upon systemic and central administration, potently increases adiposity and food intake [2] . Similar to other GH secretagogues [136] the effect of ghrelin on adiposity is GH-independent and involves neural circuits that control food intake, energy expenditure, nutrient partitioning, and reward [3] . In the ARC, a key hypothalamic center regulating food intake and satiety [42] , ghrelin increases the activity of Npy and Agrp expressing neurons while inhibiting the activity of proopiomelanocortin ( Pomc ) neurons [3] . NPY and AgRP are crucial for ghrelin's effect on feeding behavior as ghrelin fails to increase food intake in mice lacking both Npy and Agrp [137] . In line with this notion, AgRP neuron-selective GHSR re-expression in otherwise GHSR deficient mice partially restores the orexigenic response to administered ghrelin and fully restored the lowered blood glucose levels observed upon caloric restriction [138] . Of appreciable note, hyperphagia observed under pathophysiological conditions, such as streptozotocin-induced diabetes, is mediated by increased ghrelin release, which targets the ghrelin receptor on NPY and AgRP neurons [139] . A crucial regulator of Npy and Agrp expression is the hypothalamic homeobox domain transcription factor Bsx [140] , which is also essential for ghrelin's ability to stimulate food intake [140] .

4.7.1 Studies in mice with adult-onset ablation of ghrelin-producing cells The group of Joseph Goldstein and Michael Brown recently generated transgenic mice in which the diphtheria toxin receptor is expressed in ghrelin-secreting cells. Adult-onset ablation of ghrelin-producing cells in these mice, following administration of diphtheria toxin, had no effect on food intake and body weight, indicating that while ghrelin has potent pharmacological effects on food intake, energy metabolism, and body weight, it is not an essential endogenous regulator of those endpoints [144] . Despite this finding, diet-induced obesity renders NPY/AgRP neurons unresponsive to the stimulatory actions of ghrelin on food intake [143,145] , an effect that is reversed with diet-restricted weight loss [146] . However, mice continue to gain adiposity on a high-fat diet (HFD) [143] ; thus, reinstatement of ghrelin sensitivity with diet-induced weight loss may provide a physiological means to protect a higher body weight set point established after prolonged HFD exposure. Moreover, selective reduction of the expression of GHSR1a in the paraventricular nucleus of the hypothalamus (PVH) reduces body weight without affecting food intake [147] , which supports the idea of two parallel ghrelin-responsive hypothalamic circuits that regulate food intake and adiposity independently. As yet, the differences in these neural circuits remain unknown.

The orexigenic effect of ghrelin is specifically modulated through GHSR1a, as exogenous ghrelin fails to promote food intake in mice lacking this receptor [158] and in rats treated centrally with GHSR1a antagonists [147] . Despite the well-described actions of ghrelin on NPY and AgRP neurons, very little is known about the function of other hypothalamic neuronal populations expressing the GHSR. These include the ventromedial nucleus of the hypothalamus (VMH), dorsomedial nucleus of the hypothalamus and medial preoptic area [159,160] . Elucidating the function of these populations will highlight the role of ghrelin as being more than simply a “hunger hormone.” As in mice lacking ghrelin, mice deficient for GHSR are protected from diet-induced obesity (DIO) when fed a HFD. This might be explained, in part, by a mild hypophagia and preferential utilization of fat as an energy substrate in these mice [151,152] . Expression of GHSR antisense RNA under the TH promoter in the ARC of rats results in hypophagia and decreased body weight and body fat [161] . Compared to wt littermate controls, GHSR deficient mice also show improved glucose disposal and insulin sensitivity upon HFD exposure. However, body weight and fat mass are not affected when male GHSR deficient mice are maintained on a standard chow diet [151,158] . Consistently, ablation of ghrelin receptor reduces adiposity and improves insulin sensitivity during aging by regulating fat metabolism in white and brown adipose tissues [25] . The effects of ghrelin on glucose metabolism during aging might be associated to GH levels, as it is known that circulating GH levels, which cause insulin resistance [162] , are decreased in later stages. Interestingly, simultaneous deletion of both ghrelin and GHSR results in lower body weight and fat mass even when the double mutant mice are fed a standard chow diet [153] . One possible explanation for this finding is the potential existence of additional ligands for the GHSR and/or additional receptors for ghrelin, which may exacerbate the metabolic phenotype of double mutant mice where ghrelin and GHSR are inactivated [153] . Another possibility is that GHSR may affect food intake independently of ghrelin signaling, e.g. by heterodimerization with other receptors such as the dopamine receptor [163] or GPR83 [164] . Of note, the impact of GHSR signaling on food intake, body weight, or energy and glucose homeostasis might be influenced by the receptor's intrinsic constitutive activity [165] , thus complicating the direct comparison of metabolic phenotypes of ghrelin and GHSR deficient mouse models. However, given the level of GHSR1a expression found in native tissues, it is doubtful that basal activity is a contributing factor. Furthermore, this interpretation might be confounded by the observation that GHSR1a, but not ghrelin, is essential for appetite regulation by dopamine receptor subtype-2 [166] .

Several SNPs within the first intron of the fat mass and obesity-associated gene ( FTO) are robustly associated with increased BMI and adiposity across different ages and populations [171–176] . Subjects homozygous for the obesity-risk (A) allele of SNP rs9939609 have a 1.7-fold increased risk for obesity and exhibit overall increased ad libitum food-intake [177–179] , particularly fat consumption [177,179–181] , and impaired satiety [182,183] compared to subjects homozygous for the low-risk (T) allele. Recently, a series of studies implicated ghrelin in mediating this altered feeding behavior. In two independent cohorts of normal-weight, adiposity-matched individuals with either FTO rs9939609 TT or the obesity risk AA genotype [184] , AA subjects exhibited attenuated post-meal suppression of both hunger and circulating acyl-ghrelin levels. Using fMRI, these studies demonstrated that FTO rs9939609 genotype modulated the neural responses to food images in homeostatic and reward brain regions. Furthermore, AA and TT subjects exhibited divergent neural responsiveness to circulating acyl-ghrelin within brain regions that regulate appetite, reward-processing and incentive motivation. At the molecular level, FTO directly demethylates N 6 -methyladenosine (m 6 A), a naturally occurring adenosine modification in RNA and ghrelin mRNA has been identified as an FTO target [184,185] . FTO over-expression in MGN3-1 cells, a validated ghrelin cell line, reduced ghrelin mRNA m 6 A methylation, increased ghrelin mRNA abundance and the synthesis and secretion of acyl-ghrelin. Furthermore, subjects with the A allele of rs9939609 exhibit increased FTO expression [184,186] and decreased ghrelin m 6 A methylation coupled with increased ghrelin expression [184] . Interestingly, FTO also regulates the m 6 A methylation and expression of key molecular components of the mid-brain dopaminergic system, which is known to play a key role in mediating the rewarding effects of ghrelin [187] . Altered dopaminergic signaling may account for the altered neural ghrelin sensitivity reported in rs9939609 FTO AA subjects [184] . This suggests that the known actions of acyl-ghrelin, increased food intake, increased adiposity, preference for high-fat food, enhanced operant responding for food rewards, induced conditioned place preference for food rewards and a role in cue-potentiated feeding are strikingly similar to the feeding phenotype of rs9939609 AA subjects. However, while these findings of altered ghrelin function in FTO rs9939609 AA subjects provide a parsimonious explanation for the obesity risk phenotype seen in these subjects, given the pleiotropic effects of FTO a number of other mechanisms could also be implicated.

5.1 Hypothalamic effects of ghrelin on energy metabolism Although an important site of action of ghrelin on the control of food intake is the ARC, ghrelin administration into other hypothalamic sites, including the PVH [201,202] and the lateral hypothalamus [203] also promote a positive energy balance. In the hypothalamus, ghrelin triggers endocannabinoid release [204] , leading to activation of the calcium/calmodulin-dependent protein kinase 2 (CaMKK2) and increased phosphorylation of the energy sensor AMP-activated protein kinase (AMPK) [205–207] . Ghrelin mediated activation of GHSR1a also triggers hypothalamic sirtuin 1 (Sirt1) [208,209] , which deacetylates p53, leading to increased phosphorylated levels of AMP-activated protein kinase (AMPK) [206] and to the inactivation of enzymatic steps of de novo fatty acid biosynthetic pathway in the VMH [207] . These molecular events induce changes in uncoupling protein 2 (UCP2) [210] and the upregulation of the transcription factors Bsx [140] , forkhead box O1 (FoxO1), and cAMP response-element binding protein (pCREB) [211] followed by subsequent activation of downstream signaling pathways. Within the hypothalamus, ghrelin increases expression of the prolyl carboxypeptidase (PRCP) a negative regulator of the melanocortin 4 receptor agonist α-melanocyte stimulating hormone (α-MSH) [212] and the mechanistic target of rapamycin (mTOR) in the ARC. In fact, central inhibition of mTOR signaling with rapamycin decreases ghrelin's orexigenic action [213] .

Ghrelin engages reward neurocircuits that are activated by drugs of abuse [15,226–232] . In particular, the central ghrelin signaling system seems to be important for the rewarding properties of alcohol [228] , nicotine [233,234] and cocaine [235] . One of the neurocircuits involved in these effects is the mesolimbic, dopaminergic pathway that projects from the VTA to the nucleus accumbens (NAc) [229,236] , a pathway with a key role in reward-seeking behavior. Acting on this pathway, ghrelin affects the motivation and drive to eat. Ghrelin administration to the VTA and the NAc increases both food intake [232,237] and extracellular dopamine [13,238] . Underscoring the importance of dopamine for ghrelin's orexigenic effects is the finding that intra-VTA delivery of a ghrelin antagonist blocks the ability of parenteral ghrelin to increase feeding [239] . Dopamine modulates the incentive salience of food [240] and the animal's willingness to work for food [241] . In short, it increases feeding by increasing the drive, arousal, foraging, and motor hyperactivity that occur during food anticipation. For example, ghrelin ko mice do not show the normal anticipatory locomotion to scheduled meals [242,243] . Ghrelin also has direct effects on two other regions implicated in the control of feeding: the hippocampus and amygdala, where it facilitates learning and memory [188] and emotional arousal [244] and cue-potentiated feeding [189] . GHSR1a and dopamine receptor-2 (D2R) are present as GHSR1a:D2R heterodimers in native hypothalamic neurons and the inhibitory effects of D2R signaling on food intake is dependent on the presence of GHSR1a. GHSR ko mice and wt mice treated with a selective GHSR1a antagonist are resistant to the anorexigenic effects of a DRD2 agonist. Remarkably, ghrelin ko mice are fully sensitive to DRD2 agonist suppression of food intake, demonstrating a dependence on GHSR1a but not on ghrelin [163] . At the level of the VTA, but not the NAc, ghrelin increases motivation for food, reflected by an increased lever pressing for sucrose pellets in a progressive ratio task [231] . Interestingly, the VTA-driven effects of ghrelin on food motivation involve different neurocircuits than those involved in food intake. NAc delivery of dopamine receptor (D1R and D2R) antagonists blocked the effects of intra-VTA infused ghrelin on food motivation/reward behavior but not food intake, suggesting that the VTA-NAc dopamine reward pathway is important for food motivation but not food intake [245] . Given that central blockade or stimulation of the dopamine receptors 1, 2, and 3 suppress the effects of icv delivered ghrelin on food intake [246] , it can be inferred that dopamine has a role outside of this classic reward pathway to regulate ghrelin's orexigenic effects. Consistent with this, GHSR1a and D2R have been shown to interact within hypothalamic neurons blunting the anorexigenic actions of D2R agonism [163] . Divergence in the mesolimbic circuitry mediating ghrelin's orexigenic versus and food reward effects also occur at the VTA level and can be parsed using opioid and NPY Y1 receptor antagonists [232,236] . Interactions between ghrelin and the opioid system occur not only in the mesolimbic dopamine system but also in the hypothalamus. More precisely, GHSR1a and kappa opioid receptor colocalize in hypothalamic areas and the blockade of the kappa opioid receptor in the ARC is sufficient to blunt ghrelin-induced food intake [247] . Collectively, studies linking ghrelin to the mesolimbic reward circuitry suggest that ghrelin's role in hunger and meal initiation may extend to reward-driven behaviors, including food motivation.

Despite a growing body of literature characterizing ghrelin action and the distribution of ghrelin cells, relatively little is known about the exact molecular pathways responsible for the biosynthesis and release of ghrelin. Instead, most of what is known regarding the control of circulating ghrelin is on a broader, systemic level. 6.1 Ghrelin secretion in response to fasting and feeding It has been known for several years that ghrelin levels rise pre-prandially and decrease to baseline levels within the first hour after a meal [37] , a pattern that can be entrained by artificial meal schedules [251] . The magnitude of ghrelin reduction is proportional to the caloric load and macronutrient content, and ingested lipids are the least effective suppressor of plasma ghrelin [252] . Also, it is well established that plasma levels of both acyl and des-acyl ghrelin rise with prolonged food deprivation, increases that can be blocked by reserpine, which depletes adrenergic neurotransmitters from sympathetic neurons [253] . Sham feeding also suppresses ghrelin levels [254] . Furthermore, the recovery of ghrelin levels does not seem to be an important determinant of intermeal intervals [255] , and mice that lack ghrelin have normal meal intervals.

Recent data demonstrate variable effects of various bariatric surgery procedures (i.e. Roux-en-Y gastric bypass, vertical sleeve gastrectomy, laparascopic adjustable banding) on ghrelin levels (generally demonstrating decreases post-surgery) [273–279] , shedding light on how ghrelin exerts its mechanistic effects in the gastrointestinal tract (reviewed in [280] ). Low ghrelin levels have been reported for individuals after weight loss induced by Roux-en-Y gastric bypass, initially believed to play a key role in the decreased appetite observed after this surgery [257] . Subsequent data, however, show that ghrelin levels rise within the first year after this surgery in humans [281] and within 6 weeks after surgery in mice [282] . Additionally, compared to wt control mice, in ghrelin ko mice, vertical sleeve gastrectomy is equally efficient in lowering body weight [283] , indicating a ghrelin independent effect in this type of bariatric surgery.

There has been significant interest in unveiling the mechanisms involved in postprandial suppression of circulating ghrelin levels. The placement of a pyloric cuff in rats to block normal flow of gastric contents into the duodenum prevents drops in circulating ghrelin usually observed following intragastric infusion of glucose [293] . Furthermore, stomach distention by infusion of water into animals whose gastric outflow was occluded at the level of the pylorus also was ineffective in changing ghrelin levels [294] . Thus, it appears that neither nutrient detection by the stomach nor gastric distention is sufficient for eliciting the usual postprandial fall in circulating ghrelin. On the other hand, both intraduodenal and intrajejunal administration of nutrients via intestinal cannulas lower circulating ghrelin levels. Within the stomach, considered the predominant source of circulating ghrelin, ghrelin cells tend to cluster towards the base of the gastric mucosal glands and are of the round, closed-type variety that do not have direct contact with gastric luminal contents (reviewed in [295] ). There is also credible evidence that ghrelin cells exist, although in fewer numbers, throughout the entirety of the gastrointestinal tract, including the duodenum, where more elongated, opened-type ghrelin cells, which have direct contact with the intestinal lumen and may be regulated differently than their gastric counterparts.

Recent studies have demonstrated that increases in ghrelin levels also occur upon acute or chronic stress that is not necessarily related to negative energy balance [reviewed in [26,28,297,298] ]. Indeed, sympathetic activation increases ghrelin secretion [192] . Both in vitro and in vivo studies have demonstrated release of ghrelin in response to sympathetic stimulation mediated viaβ1-adrenergic receptors present on the ghrelin cell [253,284,287,288,299] . To more directly study the determinants of ghrelin secretion, a recent study performed local infusion of candidate compounds into the gastric submucosa followed by measurement of ghrelin mobilization via implanted microdialysis probes [300] . Using this method, epinephrine, norepinephrine, endothelin, and secretin were found to stimulate ghrelin release. In contrast to the stimulation of ghrelin release by activation of the sympathetic nervous system (SNS), the inhibition of ghrelin release seems primarily mediated by gastrointestinal hormones released during nutrient digestion, such as somatostatin and gastrin releasing peptide/bombesin. Numerous other hormones as well as many neurotransmitters, neuropeptides, glucose, and amino acids had no effect. Although this microdialysis technique and others, such as an ex vivo stomach explant culture system [294,301] , help focus on locally acting compounds that influence ghrelin release, the techniques do not discriminate between compounds that act directly on ghrelin cells versus those that act indirectly via effects on neighboring cells. In summary, the regulation of ghrelin release is a complex process that is tightly controlled by both, the SNS and the gastrointestinal tract and which involves hormonal stimuli not necessarily involved in energy balance regulation. The observation that also factors not involved in systems metabolism regulate ghrelin secretion speaks for a broader physiological role of ghrelin and is in line with a multitude of ghrelin effects beyond the regulation of hunger and satiety. A potential beneficial effect of this complexity is the possibility to target the ghrelin system for the treatment of pathological conditions not necessarily related to a negative energy balance, such as e.g. gastroparesis. Recent studies showed that that the ghrelin cell is chemosensory and contains taste receptors similar to those located in the tongue. Indeed, the ghrelin cell is co-localized with the gustatory G-proteins, α-gustducin and α-transducin. Studies in α-gustducin ko mice show that α-gustducin partially mediates the effect of bitter tastants on ghrelin release [302] . Similarly, the taste 1 receptor subtype, TAS1R3, involved in sensing both sugars and amino acids, is co-localized with ghrelin cells in the antrum [303] . The closed-type ghrelin cells in the stomach may receive chemosensory input from the bloodstream while the opened-type cells in the duodenum may respond to luminal stimuli. The long-chain fatty acid sensing receptor GPR120 is co-localized with ghrelin containing cells in the duodenum but not in the stomach and has been shown to play a role in the lipid-sensing cascade of the ghrelin cell [68] . Free fatty acid receptor 1 (FFAR1), involved in sensing of long/medium chain fatty acids, is expressed only in the des-acyl (non-active)-containing ghrelin cell population in the stomach, and its function is unclear [68] . More studies are warranted to elucidate the role of taste receptors in the effects of nutrients on ghrelin secretion. It is likely that recent findings and new tools will provide greater insight into the regulation of ghrelin secretion (as reviewed in [304] ). For instance, the development of novel, high-throughput sandwich enzyme linked immunosorbent assays or radioimmunoassays for the specific and sensitive detection of acyl-ghrelin as well as new mass spectrometry methods will permit accurate means to detect the different forms of circulating ghrelin and determine how various manipulations influence the levels of these different forms [305] . Another key development is the identification of GOAT, the enzyme responsible for catalyzing ghrelin's unique post-translational modification [56,57] . Recent work involving genetic manipulations of GOAT expression and the aforementioned mass spectrometry methods are challenging some of the accepted dogma about ghrelin secretion and regulation and suggest that ghrelin acylation and the secretion of acylated ghrelin represent two independent processes [61] .

Other models developed to assess ghrelin secretion include genetically-engineered mouse models in which green fluorescence protein reports on the location of ghrelin-expressing cells. This method enables direct visualization of ghrelin cells, and fluorescence activated cell sorting-mediated isolation of ghrelin cells for expression analyses and cell culture [295,307,308] . Primary cell cultures of dispersed gastric mucosal cells from adult mice and 8-day-old rat pups also have been developed to investigate ghrelin secretion [288,307,309] . Ghrelin-secreting immortalized cell lines developed from ghrelinomas in the stomachs (SG-1, MGN3-1) and pancreatic islets (PG-1) of transgenic mice expressing SV40 large T-antigen under the control of preproghrelin promoter are now available [253,310] . These ghrelinoma cell lines retain many of the key, phenotypic features of ghrelin cells and respond to many of the same regulators of ghrelin secretion that have been described in vivo and in primary culture systems [253,288,307,310] . Using these models, the modulation of ghrelin release by peptide hormones, monoaminergic neurotransmitters, glucose, fatty acids, second messengers, potential downstream effector enzymes, and channels has now been investigated. Insulin, glucagon, oxytocin, somatostatin, dopamine, glucose, and long-chain fatty acids all have been shown to regulate ghrelin secretion through their direct interaction with ghrelin cells [253,288,307–310] . In addition, all of these models, as well as related in vivo studies, have been used to confirm that the catecholamines norepinephrine and epinephrine act as direct ghrelin secretagogues [253,287,288,299,307] . These data are supported by high levels of β1-adrenergic receptor expression in ghrelin cells enriched from the stomach of ghrelin-green fluorescent protein reporter mice as well as in the SG-1 and PG-1 ghrelin cell lines [253] . Forskolin, a potent activator of adenlyl cyclase, mimics the effect of norepinephrine [253] , suggesting that activation of adenylyl cyclase and an ensuing elevation of cAMP occurs following engagement of β1-adrenergic receptors, as has been shown in other cell systems [311,312] . Interestingly, neuronal and endocrine signals have stimulatory effects on ghrelin secretion whereas paracrine signals and macronutrient metabolites such as fatty acids inhibit ghrelin release [306] . Altogether, these findings, along with the aforementioned microdialysis experiments, link fasting-induced stimulation of the sympathetic nervous system and ensuing release of norepinephrine locally in the stomach wall to the release of ghrelin [168,299] . New transgenic and cell culture models should allow for many more discoveries in the regulation of ghrelin secretion and other aspects of ghrelin cell physiology. In summary, recent studies reveal a comprehensive picture of the receptor repertoire expressed on the ghrelin cells, which allows for deeper analyses of the physiological properties and pharmacological potential of the ghrelin cell.

Nevertheless, several studies suggest that des-acyl ghrelin promotes differentiation and fusion in C2C12 skeletal muscle cells [29] , prevents muscle atrophy [30] , elicits GHSR1 independent effects on energy and glucose metabolism [316–318] , and exerts a cardioprotective effect on endothelial cells and cardiomyocytes [319,320] . Central infusion of des-acyl ghrelin into the third ventricle of rats increases short-term food intake, whereas peripheral administration of des-acyl ghrelin seems to have no direct effect on food intake [317] . The effect of des-acyl ghrelin on food intake is independent of GHSR1a and Npy signaling and might be orexin mediated [317] . Other studies, however, suggest that des-acyl ghrelin decreases food intake in rats and disrupts stomach motor activity under conditions of fasting [321] . Transgenic mice overexpressing des-acyl ghrelin have reduced body fat mass when fed a regular chow diet and are protected from diet-induced obesity when challenged with a HFD [321] . Although there are speculations about a potential receptor for des-acyl ghrelin located in the cardiovascular system [320] , to date no such receptor has been identified. Growing evidence points to a GHSR1a-independent role of des-acyl ghrelin in glucose metabolism, possibly antagonizing the effect of acyl-ghrelin. The often reported increase of plasma glucose levels and decrease of plasma insulin levels upon ghrelin administration [84–86,88,89] seem to be antagonized by co-administration of des-acyl ghrelin [322] . Several human studies report a positive relationship between des-acyl ghrelin and insulin sensitivity [323,324] , although other studies do not support this finding [325] . The effect of des-acyl ghrelin on glucose metabolism might be triggered indirectly via modulation of lipid metabolism, as transgenic mice overexpressing des-acyl ghrelin have lower body fat mass, lower body weight gain, and improved insulin sensitivity compared to wt controls [318,326] . Furthermore, administration of des-acyl ghrelin decreases activation of gene programs regulating lipogenesis [327] . A more recent study in mice shows that chronic, subcutaneous administration of des-acyl ghrelin prevents the typical metabolic alterations caused by chronic HFD exposure, such as increased expression of pro-inflammatory cytokines and the development of HFD-induced glucose intolerance and insulin resistance [328] . Conversely, mice overexpressing ghrelin driven by the neuron-specific enolase (NSE) promoter develop age-related glucose intolerance despite having lower body weight [18] . Recent observations suggest a direct role on glucose metabolism based on observations indicating that des-acyl ghrelin promotes survival of pancreatic β-cells and as des-acyl ghrelin prevents the diabetogenic effect of streptozotocin [329–332] . Other data suggest that des-acyl ghrelin administered centrally to mice at high pharmacological doses acts to increase adiposity and glucose-stimulated plasma insulin through a GHSR-dependent mechanism [333] .

Ghrelin and its agonists appeal to those who desire to exploit the potent anabolic biology of the hormone. Such attention is directed at cachexic and frail states. Ghrelin has a very short half-life but the peptide can be engineered for a more sustained delivery and better pharmacokinetic properties. Treatment with ghrelin by infusion may be indicated in very acute circumstances when a short-term anabolic state is desired, such as prior to an elective surgery. The non-peptide GHSR1a agonists developed prior to the discovery of ghrelin are orally bioavailable; most importantly, they produce significant exposure levels for up to 24 h. The extended half-life of these compounds is of metabolic and functional importance, because, in contrast to ghrelin, chronic administration of non-peptide agonists results in sustained but modest increases in GH and IGF-1 superimposed on endogenous GH/IGF-1 without an increase in cortisol [338] . Interestingly, in obese subjects the anabolic effects of MK0677 produce an increase in the ratio of lean/fat mass [339] . Chronic therapy with orally active stable GHSR1a agonists may have utility in frail, elderly subjects because MK0677 rejuvenates the GH/IGF1 axis by enhancing pulse amplitude of episodic GH release to match the physiological profile of young adults [338] . This effect is consistent with rescue of the epigenetic-mediated age-dependent decline in GHSR expression that reduces ghrelin sensitivity [340] . Indeed, encouraging results were observed in elderly patients recovering from hip fracture [341] , with beneficial effects on skeletal muscle and bone density [80] . Other positive effects of ghrelin are reported in patients with cachexia, sarcopenia (muscle wasting due to aging [342] ), myopenia (muscle wasting due to chronic illness [343] ), and frailty states [343,344] . Among the first applications of ghrelin in human chronic illness were studies in CHF [34] and COPD [78,345] . A pilot study in 10 CHF patients reported decreased levels of norepinephrine, improved cardiac function and exercise capacity [34] . In animal models using a placebo control the cardiac effects could not be confirmed [346] , but these models validated weight gain and the skeletal muscle anabolic effects of ghrelin and ghrelin analogs [110,347,348] . Of interest, in CHF, ghrelin secretion is modulated by application of brain natriuretic peptide, which is produced in the heart and generally increased in heart failure [349] . In advanced CHF, ghrelin resistance has been observed [350] . The first studies of ghrelin in COPD, focused on cachexia [78] . Preliminary studies suggested that ghrelin increases skeletal muscle mass and improves exercise capacity. In a recent double-blind controlled trial, ghrelin improved symptom scores and increased respiratory muscle strength [345] . An additional indication for ghrelin treatment may be replacement therapy to patients who have undergone total gastrectomy due to gastric cancer. Theoretically, this is based on the fact that most ghrelin is produced in the stomach and that total gastrectomy results in loss of appetite, body fat and also lean body mass. Indeed, proof of concept studies in mice and humans have yielded positive results [224,225] . Inhibition of GOAT-mediated ghrelin acylation is considered an interesting opportunity to tackle obesity. Several GOAT inhibitors have been developed [60,62] , and their intraperitoneal administration promotes weight loss while improving glucose tolerance in wildtype (wt) mice but not in mice deficient for ghrelin [60] . Ghrelin receptor agonists such as BIM-28131 (a.k.a. RM-131) have sustained effects in increasing body weight at long-term [351] . Chronic therapy with ghrelin agonists, however, is associated with weight gain, fat attrition, and insulin resistance. Such observations have led to drug discovery efforts designed to block ghrelin action, inducing a negative energy balance with a goal of treating obesity and insulin resistance. Although ghrelin levels are lower in obesity, that circulating ghrelin levels increase during a negative energy state may suggest that a method inhibiting ghrelin activity may be useful for preventing weight regain after diet and exercise (or another weight loss treatment) rather than as a weight loss therapeutic. The greatest utility of des-acyl ghrelin, in fact, appears to be for the treatment of insulin resistance, but only when injected in combination with ghrelin. Several synthetic ghrelin mimetics are being pursued in clinical trials for diverse indications ( Table 1 ). Three compounds are currently in development. Macimorelin is in clinical trials for the diagnosis of GH deficiency, relying on the stimulation of the hypothalamic pituitary axis, as described earlier in this review [352] . A second compound, Anamorelin, is in clinical trials for the treatment of cancer cachexia in the treatment of non-small cell lung cancer. Anamorelin mechanistically relies on the anabolic

The authors wish to declare that Dr. L Van der Ploeg is an employee of Rhythm Pharmaceuticals, a privately held Biotechnology company, which is developing RM-131 for the treatment of diverse functional gastrointestinal disorders.