A skeleton in the cupboard in ghrelin research: Where are the skinny dwarfs?

Based on studies delivering ghrelin or ghrelin receptor agonists, we have learned a great deal about the importance of the brain ghrelin signalling system for a wide range of physiological processes that include feeding behaviours, growth hormone secretion and glucose homeostasis. Because these processes can be considered as essential to life, the question arises as to why mouse models of depleted ghrelin signalling are not all skinny dwarfs with a host of behavioural and metabolic problems. Here, we provide a systematic detailed review of the phenotype of mice with deficient ghrelin signalling to help better understand the relevance and importance of the brain ghrelin signalling system, with a particular emphasis on those questions that remain unanswered.


| INTRODUC TI ON
One of those lingering questions in ghrelin research concerns the apparent absence of an obvious phenotype consistent with deficient ghrelin signalling, namely the skinny dwarf, which is both horizontally and vertically challenged and likely has a host of metabolic and behavioural impairments (Figure 1). Subsequent to the discovery of ghrelin in 1999, 1 its receptor in 1996 2 and the first synthetic ligands in the early 1980s [3][4][5] (and please take note of the somewhat unusual reverse timeline here), we have learned a very great deal about the important and diverse roles of the ghrelin signalling system in normal physiology, with its effects on food intake, growth hormone (GH) secretion and glucose metabolism being the most intensely studied. 6 To our knowledge, there are no examples in nature of a spontaneous mutation in the ghrelin system that causes a deficiency phenotype embodying the full breadth of the action of ghrelin. Moreover, the phenotype of genetically modified mice with deficient ghrelin signalling is often rather subtle (Table 1). However, this does not mean that ghrelin is unimportant. Simply, the physiological processes regulated by ghrelin are essential to life and are therefore not dependent solely on ghrelin for their function. It seems that no other known hormone simply steps up, fulfilling the role of ghrelin in its absence, emphasising its unique importance. Note, however, that the ghrelin receptor (the growth hormone secretagogue receptor, GHSR) is constitutively active 7 and may be regulated in ways independent of ghrelin, which include heterodimerisation with other receptors. 8 Here, we explore the apparent mismatch between the predicted skinny dwarf phenotype and what is actually observed in models of deficient ghrelin signalling. The review is inspired by a somewhat serendipitous finding on our part: homozygous Ghsr-IRES-Cre mice have a skinny dwarf phenotype that appears to reflect the absence of a functional GHSR. 9

| PRED IC TED PHENOT YPE: DWARF
To set the stage, linking the ghrelin signalling system to GH and potentially also growth, it is relevant to mention that the ghrelin field dates back to the early 1980s, a time when pioneers such as Frank Momany and Cy Bowers were searching for the elusive endogenous GH-releasing hormone (GHRH), and had identified a group of short peptide GH secretagogues (GHS) derived from met-enkephalin as potential candidates. 3,4 When GHRH was finally identified, 10,11 it turned out to be entirely different in structure to these GHS, exerting its GH-releasing effects via a completely different receptor signalling system. Therapeutic interest in GHS grew increasingly, however, with the realisation that they greatly amplify the pulsatile pattern of GH release, 12,13 acting by a direct pituitary 4 and hypothalamic action. 14,15 Pulsatility in the GH secretory profile is of paramount importance for stimulating growth, both in childhood and during puberty [16][17][18] and GH has a dose-dependent effect on growth. 19 Admittedly, it remains somewhat of an enigma why nature has invested so much effort into ensuring GH is released in a pulsatile manner to optimise growth, involving a dual neuroendocrine regulatory network provided by GHRH and inhibitory somatostatin, as well as a complex feedback regulation by GH and its effector, insulin-like growth factor 1, that has the capacity to be amplified when the ghrelin signalling system is activated. Nonetheless, GHS (including orally bioavailable non-peptide compounds) offered new possibilities not only to promote growth, [20][21][22] but also to rejuvenate the GH-axis in the elderly, with benefits on lean and bone mass. 5 Although GHS never made it to the clinic, they have provided much insight regarding the physiological roles of the ghrelin signalling system, namely with reference to the hypothalamic-GH-growth axis, even before ghrelin was discovered.
Ghrelin release is episodic, although it is not at all obvious that it operates as an amplifier of the pulsatile pattern of GH secretion.
In humans, ghrelin is released from the empty stomach before meals and in association with hunger and, curiously, also reaches quite high levels in the middle of the night. 23,24 Likewise, in laboratory rodents, ghrelin release is greatest in situations of energy deficit such as fasting. 25 In line with this, there are data indicating that an important physiological role of ghrelin is to help prevent hypoglycaemia, with effects exerted, at least in part, by enhancing GH release during fasting or periods of food restriction. 26 Thus, the pattern of ghrelin secretion links better to the diabetogenic effects of GH (when food is scarce) rather than for its growth-promoting effects (which are suppressed in times of famine).
However, it remains unclear whether the ghrelin signalling system has a role in linear growth. There have been three genetic studies in humans in which GHSR mutations have been associated with short stature, [27][28][29] with five mutations linked to loss of constitutive activity of GHSR. 27,29 Another population-based study (in 3 UK cohorts) found no association between common variations in GHSR and body height in adults or children. 30 In animal models of deficient ghrelin signalling (and there are a very great number to choose from) ( Table 1), extracting growth data is somewhat of a challenge because (i) body length is not usually measured directly, with body weight as a surrogate (and note that rodents continue to grow throughout their life); (ii) the studies are more focused on potential anti-obesity effects rather than body growth; and (iii) any growth-linked differences are often lacking/subtle and/or sexually dimorphic. Some loss-of-function models did not detect a relevant phenotype, 31,32 whereas others found a modest effect (Table 1), for which it is unclear if they are lean or growth retarded, or some combination of the two (ie, skinny dwarfs). Mice deficient in preproghrelin (ghrl −/− ), although having a reduced GH pulse amplitude, have only a mild suppression of growth. 33 In another study, female (but not male) GHSR-null (ghsr −/− ) mice on a chow diet started to differ in body weight at 12 weeks of age, weighing 11.7% less (and with 35.6% reduced body fat) than wild-types by week 19.
However, any effect on body length in these mice was barely detectable. 34 Another study reported that ghsr −/− mice (GHSR-KO here) have a lower body weight from 4 months of age. 35  release pattern phenotype has been characterised in both males and females. 36 It was found that there was a reduction in linear growth in ghsr −/− mice correlated with a reduced pituitary GH content. Curiously, pulsatile GH secretion was decreased in adult but not adolescent ghsr −/− female mice. In male ghsr −/− mice, pulsatile GH release was only diminished during adolescence and did not correlate with linear growth. 36 In a previous study, 9 we noted that homozygous Ghsr-IRES-Cre male mice (that lack GHSR expression and are unresponsive to ghrelin) are lighter (by approximately 10%) in body weight than heterozygous or wild-type mice already at postnatal day (PND) 12, persisting into adulthood. Not only were the homozygotes shorter in length (determined at PND65), dualenergy X-ray absorptiometry scans revealed that they had a decreased bone area and a lower bone mineral content. Although we did not have the possibility to follow their GH secretory pattern, GH levels during fasting were reduced in the homozygous (and also heterozygous) Ghsr-IRES-Cre mice, despite having higher circulating ghrelin levels. 9

| PRED IC TED PHENOT YPE: S K INNY
Growth hormone is lipolytic. In the late 1990s, our laboratory obtained funding for a project based on the idea that GHS treatment, by engaging the GH axis, should reduce body fat and hence body weight. Sometimes it is good to be wrong (or perhaps naïve). We found that chronically daily GHS-injected mice increased their body weight and body fat content, clearly via a GH-independent mechanism. 37 Although it may be beneficial to mobilise stored fat upon fasting or when food is unavailable, it is also important to replenish these stores. Indeed, one of the first things we learn about ghrelin is that it stimulates food intake, 38,39 operating as an orexigenic hunger hormone. 23 There are thousands of articles backing these claims, although it is important to point out that the actual role for this hunger hormone may be to initiate meals or to organise food intake into meals, rather than to cause over-eating per se.
There is no doubt whatsoever that acute peripheral ghrelin injection is orexigenic, as first described by Wren and colleagues in 2000. 39 In this particular study, the orexigenic effect of ghrelin was only detected at 1 hour after peripheral injection but not at later time points, raising the question of whether its orexigenic effects are rather short-lived. This is unlikely, however, because later studies reported that cumulative food intake was elevated after a single i.p. ghrelin injection at the 2-, 3-and 4-hour time points (at a dose of 1 or 10 nmol) 40 and at various time points between 30 min and 5 hours (at a dose of approximately 1 nmol). 41 We have also observed a heightened feeding response at > 2 hours after a single i.p. ghrelin (approximately 100 μg) injection to rats returned to their home cage after a behavioural task during which access to their regular chow was denied. 42 Collectively, these data suggest that, in situations where food is available, ghrelin administered by the i.p. route rapidly increases food intake to a level at which satiation can be reached but that the overall time window for the orexigenic action of ghrelin could be much longer if food is scarce or not immediately available. This profile of action, taken together with data showing that the circulating levels of ghrelin are highest preprandially in humans, 24 strongly supports a physiological role for ghrelin in hunger and meal initiation.
Perhaps, because ghrelin is orexigenic, it is assumed to be "obesity-promoting". The evidence suggests, however, that obese individuals are not hyperghrelinemic, 43 with the notable exception being Prader-Willi patients, for whom elevated ghrelin may contribute to their obesity. 44 Paradoxically, dietary obesity appears to be associated with impaired ghrelin signalling. 45 Daily peripheral injections of ghrelin or GHS to mice can cause an increase in body fat and weight gain, 25,37 although the mechanism may relate more to a decrease in fat utilisation rather than an increased in food intake. 25 Indeed, the actual amount of food eaten by rats at 1 hour after a single ghrelin injection may appear to be a large increase (ie 3.5-4.5 fold) but this is actually a small increase in the amount of food eaten (from 0.35 g to around 1.2-1.6 g) and insufficient to impact on 24-hour food intake (which is around 18-20 g). 39 Ghrelin injection studies such as these are normally performed during the light phase when rodents are inactive and have a low level of spontaneous eating and when endogenous circulating ghrelin levels are assumed to be lowest. 46,47 One reason why the feeding response to peripheral ghrelin is minimal during the light phase could be that this corresponds to a time when the action of ghrelin is opposed by the high circulating levels of liverenriched antimicrobial peptide 2 (LEAP2), a recently discovered endogenous GHSR antagonist, 48 for which circulating levels are determined by metabolic status. 49 The central ghrelin signalling system is powerfully orexigenic.
Ghrelin is able to drive a feeding response when delivered to most brain areas where GHSR is located (for localisation of GHSR in rats and mice, see Zigman et al 50 ). This includes not only areas essential for energy homeostasis in the hypothalamus and brainstem, but also areas involved in cognition, memory and emotional reactivity (hippocampus, amygdala), as well as areas involved in reward (nucleus accumbens and tegmental areas). 51 Indeed, by engaging these pathways, the effects of ghrelin on feeding extend beyond food intake to include food choice, 52 binge-like eating, 53 food anticipation, 54 food reward, 55 food motivation 42,51,[56][57][58] and food intake during stress exposure, 59,60 and may even convey the negative valence signal of hunger. 61 With all of these reported effects of ghrelin, it can be asked why mice with deficient ghrelin signalling are not skinny and anorectic.
Even if some studies in models of deficient GHSR signalling have found body weight to be unaffected, 35,56,62-65 many others have reported the opposite. 34,36,[66][67][68][69][70][71] This skinny phenotype was especially noticed at older ages [71][72][73] or when challenged with a high-fat diet (HFD) early in life 34,66,74 or an activity-based anorexia (ABA) protocol. 54 However, daily food intake in these models was generally unaffected 9,35,36,62,69,70,72 with the exception of HFD and ABA studies 34,54,74 and a GHSR-mutant model that lacks GHSR constitutive activity, as well as a ghrelin-induced feeding response and GH release. 73 TA B L E 1 Studies using rodent models with genetic alterations in the ghrelin signalling system, grouped by model (from large scale to specific alterations) and in chronological order         our study, 9 whereas life-threatening hypoglycaemia develops with more chronic caloric restriction. 26,70,73,80,81 This finding is consistent across models and hence points towards a critical role of the ghrelin system in the prevention of life-threatening hypoglycaemia.

| CON CLUDING COMMENTS
Skinny dwarfs, the "ghremlins" in ghrelin research, may exist but have remained somewhat hidden from view because diverse models of deficient ghrelin signalling did not cause a consistent and/or dramatic phenotype in line with that predicted from ghrelin injection studies. Homozygous Ghsr-IRES-Cre mice emerge, however, as a novel model for exploring deficient ghrelin signalling, in which GH secretion, body growth and glucose homeostasis are impaired. 9 omLäkarutbildning och Forskning (ALFGBG-723681). We thank Professor Jeffrey M. Zigman for reading and providing valuable comments on the first version of the review.

CO N FLI C T O F I NTE R E S T S
The authors declare that they have no conflicts of interest.