Hypothalamic neuropeptides and neurocircuitries in Prader Willi syndrome

Abstract Prader‐Willi Syndrome (PWS) is a rare and incurable congenital neurodevelopmental disorder, resulting from the absence of expression of a group of genes on the paternally acquired chromosome 15q11‐q13. Phenotypical characteristics of PWS include infantile hypotonia, short stature, incomplete pubertal development, hyperphagia and morbid obesity. Hypothalamic dysfunction in controlling body weight and food intake is a hallmark of PWS. Neuroimaging studies have demonstrated that PWS subjects have abnormal neurocircuitry engaged in the hedonic and physiological control of feeding behavior. This is translated into diminished production of hypothalamic effector peptides which are responsible for the coordination of energy homeostasis and satiety. So far, studies with animal models for PWS and with human post‐mortem hypothalamic specimens demonstrated changes particularly in the infundibular and the paraventricular nuclei of the hypothalamus, both in orexigenic and anorexigenic neural populations. Moreover, many PWS patients have a severe endocrine dysfunction, e.g. central hypogonadism and/or growth hormone deficiency, which may contribute to the development of increased fat mass, especially if left untreated. Additionally, the role of non‐neuronal cells, such as astrocytes and microglia in the hypothalamic dysregulation in PWS is yet to be determined. Notably, microglial activation is persistently present in non‐genetic obesity. To what extent microglia, and other glial cells, are affected in PWS is poorly understood. The elucidation of the hypothalamic dysfunction in PWS could prove to be a key feature of rational therapeutic management in this syndrome. This review aims to examine the evidence for hypothalamic dysfunction, both at the neuropeptidergic and circuitry levels, and its correlation with the pathophysiology of PWS.

individuals have a certain degree of mental retardation. [1][2][3] In addition, PWS patients suffer from multiple endocrine abnormalities.
PWS infants present with hypotonia which, among others, causes an impaired feeding behaviour due to poor suck and swallowing. By contrast, children and adults are characterised by gross hyperphagia and poor satiety, which lead to a severe obese phenotype. 1,2 The estimated prevalence of PWS is 1/10 000-1/30 000 cases worldwide.
Well-defined, widely accepted diagnostic criteria are available and can strengthen the suspicion as early as in the foetal period; however, genetic testing remains the pillar of PWS diagnosis currently. 2 Clinical studies have provided substantial evidence for hypothalamic dysfunction in PWS like uncontrollable hunger 4,5 and impaired growth 6,7 and sexual development. 8 However, the current shred of evidence cannot pinpoint if the disorder is primarily due to hypothalamic defects or the disruptive hypothalamic function is a consequence of imbalance elsewhere. Further, the molecular mechanisms behind hypothalamic dysfunction in PWS are yet to be determined.
In this review, we aim to discuss the evidence for hypothalamic dysfunction and the correlation with the clinical and pathophysiological aspects of PWS.

| THE G ENE TI C S OF PWS
The PWS phenotype results from the loss of paternally expressed components on chromosome 15q11-q13. The same genes and noncoding RNAs derived from the mother are inactivated by imprinting; and thus, not expressed under normal conditions. So far, it is impossible to attribute the phenotypic traits to a single gene; but rather, the symptomology of PWS is a consequence of the entire deletion.
In this section, we will briefly discuss our current understanding of the role of each gene and its connections with the disruption of hypothalamic function. In addition, a schematic representation of the expression map of chromosome 15 can be found in Figure 1.
Importantly, PWS-associated loss of expression can be extended to a non-imprinted region, resulting in a more severe phenotype. 9 Conventionally, the extension of the deletion led to subdivision of the genotypes. Namely, PWS T1 genotype refers to those that lack expression of both, the critical and non-imprinted region; whereas PWS T2 is associated with deletion exclusively of the critical region. 9

| NIPA1
NIPA1 encodes a transmembrane protein recruited upon fluctuations in intracellular magnesium concentrations. 10,11 There is no evidence for the participation of this gene in energy homeostasis in the hypothalamus or periphery. Its localization with endosomes suggests a role in the secretion pathways. 10 A recent report has demonstrated that neurons from PWS patients have decreased secretory granules and neuropeptides production. However, the authors attribute this defect to the MAGEL2 gene. 12 Since PWS T1 have more severe phenotypic traits it is possible that the additional loss of NIPA1 hampers the neuroendocrine maturation and secretion. 12

| NIPA2
NIPA2 shares structural and functional similarities with NIPA1. It is also a transmembrane protein sensitive to magnesium fluctuations. 13 Likewise, there are no in-depth studies that pinpoints a role for this protein in hypothalamic control of metabolism. However, F I G U R E 1 Schematic expression map of the PWS genomic region. PWS is caused by loss of expression of paternally inherited genes located in chrmosome 15. The extension of the deletion is critical for the severity of the phenotype, and patients that lack expression of genes in the non-imprinted region are reported to present more serious symptons. In addition, the contribuition of the PWS-causative genes to the phenotypic traits of the disease is given some reports suggest that NIPA2 has a role in neuronal homeostasis, and thus its deletion might impact optimal neuronal functioning.
Maternal protein restriction leads to increased expression of NIPA2 in the foetal hypothalamus in rats, 14 but the implications of these findings are yet to be clarified. The authors also report an enrichment of mitochondrial metabolism proteins, which reinforces the hypothesis of the participation of NIPA2 in mitophagy. 15

| TUBGCP5
This protein encoding gene has a notorious role in centrosome formation, 22 and therefore impacts the normal process of cellular division. Thus far, it has not been particularly implicated in hypothalamic circuits. Remarkably, disruption of this protein is implicated in microencephaly 22 ; indicating an important role in brain development.
Similar to CYFIP1, dysfunction or absence of TUBGCP5 is associated with the promotion of compulsive behaviours in neuropsychiatric disorders, such as the autism spectrum disorder. 23

| MKRN3
The Disruption of these patterns are tightly associated with obesity and its comorbidities, such as cardiovascular diseases 53 and diabetes. 54

| SNUFF-SNRPN
Little is known about the contribution of this protein to the pathophysiology of PWS. This is perhaps the least explored protein encoding gene in the PWS critical genomic region. However, a report from Cao and colleagues demonstrates that a deletion that causes loss of function of SNUFF-SNRPN is sufficient to promote PWS-like symptoms. 55 The authors speculate this is because SNUFF-SNRPN is responsible for the expression of key small nucleoar RNAs (sno RNAs -discussed next), that are central factors in the PWS symptomatology traits. 55

| sno RNAs
In addition to the protein-coding alleles, non-coding RNAs are found in the critical genomic region of PWS, especially snoRNAs. The exact biology of these RNAs is yet to be defined, but it has been proposed that they are involved in the modification of other RNAs. 56 One special cluster of these biomolecules -the Snord116-is highly implicated in PWS and has gained crescent attention. Snord116del mice are of special interest in PWS research, because those mice in which the snoRNA116 deletion is of paternal heritage display hyperphagia. 57 However, the Snor116del mice do not show increased body weight, even when fed an obesogenic diet. In fact, these mice present characteristically reduced body weight, delayed sexual maturation and high rates of mortality prior to weaning. The hypothalamic dysfunction associated with Snord116del can be explained by endocrine imbalance and sensitivity to adipostatic signals. These will be discussed into more detail in the sections below.

| HYP OTHAL AMIC NEURO CIRCU ITS IN PWS
The insatiable hunger experienced by PWS subjects is strongly indicating a malfunction of the hypothalamic control of feeding be- with stronger function in the pre-meal state. 58 Consistently, upon glucose consumption PWS subjects have delayed satiety-associated neural circuit activation in the hypothalamus and extra hypothalamic areas. 59 These findings suggest that perception of nutritional status is delayed in PWS patients, likely leading to a defective hypothalamic response to nutrients. Interestingly, it has been shown that the hyperfunction of those neuronal networks is particularly associated with high caloric foods rather than low caloric stimuli. 61  tissue and the lack of representative animal models that fully recapitulates its phenotype functional interpretation is a challenge. Of notice, hypothalamic post-mortem material has been extensively used in other fields of research and has proved to be reliable, such as in non-genetic obesity, 112 diabetes, 84 and mood disorders. 113 However, the analysis of neuropeptides' immunoreactivity without other parameters or support of the literature requires caution. This is especially due to alterations in the balance of synthesis, maturation, and secretion of these peptides. Increased levels of a neuropeptide can be explained as increased production or a defective or decreased transport and release. In the same way, reduced immunoreactivity can also indicate rapid turnover of the protein and not necessarily decreased production. This problem can be illustrated by the discrepancy between an increased vasopressin content in the SCN together with an diminished vasopressin production in female depressed patients. 114 Therefore, human post-mortem material is a reliable and stable source of study, especially in rare pathophysiology such as PWS; however the interpretation of data requires a global overview of the studied systems.

| Hypothalamic orexigenic neuropeptides in PWS
NPY neuron numbers are consistently downregulated in obese and PWS-obese subjects. 103 At first glance, reduction in NPY cell counts seems counterintuitive since an insatiable hunger is a major feature of PWS. 4 The molecular mechanisms behind this reduction are yet to The authors found a tendency for reduction of AgRP immunoreactivity in the INF of PWS and non-genetic obese individuals, which was lost when the values were adjusted according to premorbid illness duration. This finding is consistent with reports of murine models of obesity, in which the AgRP expression is unaltered. 120 In contrast, a recent study examined the transcriptomic signature of PWS patients and found a 3-fold upregulation in the AgRP transcript. 121 Additionally, genes that are overrepresented in the PWS hypothalami overlap with the murine AgRP identity. 121 The discrepancy between both studies might be explained by a defective post translational processing of AgRP in PWS. The enzyme prohormone convertase PC1 is responsible for the posttranslational cleavage of the AgRP transcript. 122 Furthermore, PWS patients have reduced levels of prohormone convertase PC2, 123,124 largely implicated in the posttranslational processing of hypothalamic neuroendocrine mediators. 125 Interestingly, Burnett et al 126  gain. Interestingly, hypocretin has a role in sleep regulation, 135,136 which is also markedly disturbed in PWS. 137 Therefore, the involvement of the hypocretin system in PWS is beyond the regulation of energy metabolism and may play a more complex role in PWS pathophysiology.
An overview of the main findings regarding the neuropeptides explored so far is summarized in Table 1.

| Hypothalamic anorexigenic neuropeptides in PWS
Surprisingly, although impaired satiety is a hallmark in PWS, 5 The authors propose that the element of neurodegeneration in the pathogenesis of PWS might be associated with this phenomenon.
BDNF and its receptor TrKB (encoded by ntrk2) expression were found to be decreased in the ventromedial nucleus (VMH) of the hypothalamus. 121 Furthermore, plasma levels of BDNF in fasting conditions are decreased in PWS compared to healthy controls. 146 Notably, beyond its trophic role, BDNF and its receptor have been implicated in suppression of feeding behavior. 147 Diminished levels of this peptide seem to be consistent with the phenotype observed in PWS, reported to have an unhealthy microenvironment for neuronal populations and satiety deficiencies. 148,149 An overview of the alterations in neuropeptides can be found in Table 1.

| HYP OTHAL AMUS -REG UL ATORY ME TABOLI C HORMONE S IN PWS
As discussed so far, the hypothalamic neuronal populations engaged with energy homeostasis are severely affected in PWS.
Hypothalamic neuronal malfunction might be directly induced by the genetic defects of PWS but can also be indirectly caused by the neuroendocrine dysregulation which occurs in conjunction with the primary PWS phenotype. Circulating metabolic hormones (i.e. ghre-

lin, leptin, insulin and adiponectin) inform the hypothalamic neuro-
circuits about the nutritional status of the organism. 150 Here we will discuss the current understanding of the role of endocrine factors in the aetiology of PWS.

| Ghrelin
Ghrelin is an orexigenic hormone produced by enteroendocrine cells. 151,152 Central or peripheral administration of ghrelin induces eating and promotes adiposity. 153,154 In fasted conditions, ghrelin levels are elevated and by contrast, re-feeding or oral glucose administration reduces the total plasma concentrations of this hor- are reduced in human non-genetic obesity. 158 Whether humans also have hypothalamic resistance to ghrelin is yet to be determined.
Differently from non-genetic obesity, ghrelin is found to be upregulated in PWS, and has been implicated as an underlying cause of hyperphagia in PWS. [159][160][161][162] DelParigi et al 160 reported a 2.5-fold increase in PWS plasma levels of ghrelin compared to lean controls; and a 4.5-fold increase compared to obese subjects. The difference remained significant even after adjustment for percent of body fat. 160 Ghrelin levels remained elevated in PWS patients in comparison to matched controls even after the consumption of satiating dose liquid meals. 160 This finding matches with the persistent urge to eat found in PWS. Moreover, plasma ghrelin levels and subjective rating of hunger have a positive correlation in PWS. 160 It would be interesting to study the ghrelin levels in PWS infants before the onset of hyperphagia. However, no differences were found between non-obese PWS infants (under the age of 5) and matched controls regarding their ghrelin levels. 163 In a different study, PWS obese children (average age 9.5 years) had elevated plasma ghrelin. 164 It is still possible that hyperghrelinemia occurs prior to the onset of obesity. 163  leanness), this model seems to be particularly interesting to comprehend, at least partially, the hyperphagic state of the syndrome. 168 It is interesting to note that Snordel116 mice have reduced PC1 expression. 126 This protein acts in ghrelin posttranslational maturation, 127 as in POMC. 169 Although ghrelin levels are elevated in Snord116del mice, there is an increased ratio between the pro-ghrelin and the mature hormone, due to diminished expression of PC1. 126

| Leptin
Leptin is an adipokine that regulates energy homeostasis by signalling hypothalamic centers. 115 Leptin is a potent anorexigenic hormone, and it is well recognized for activation of POMC-expression neurons while suppressing the NPY and AgRP neuronal activity. 85 Leptin also exerts its functions through glial cells. 89, 174 Recently the importance of leptin receptors in both astrocytes 174 and microglia 89 has been demonstrated. It is well-known that lack of leptin leads to a severe obese phenotype and deletion of its receptor also leads to obesity and diabetic traits. 175 179 Of note, leptin is a major regulator of reproduction. 180 Goldstone et al 181 also did not find differences in plasma leptin between lean and PWS adult women. These differences on plasma levels of leptin cannot be explained by defective secretion or defects in functionality of the receptor. 181 Snord116del mice have unaltered leptin levels. 168 Hypothalamic genes associated with leptin signalling genes also remain unchanged in these mutants compared to their wild type littermates. 168 Interestingly, adenovirus-mediated deletion of Snord116 in the hypothalamus leads to increased expression of the suppressor of cytokine signalling 3 (SOCS3) in obese animals. 168 This gene is responsible for the suppression of leptin signaling. 182 However, the cited work cannot dissociate the contribution of the obese phenotype versus the deletion itself in the elevated SOCS3 expression. 168 Another animal model that is often employed in PWS research is the MAGEL2 null mouse. 183 This knockout is insensitive to the anorexigenic effects of leptin, 184

| Insulin
Insulin is a hormone secreted by β-cells of the pancreatic islets.
Molecular resistance to insulin or plasma insufficiency leads to the development of type 2 diabetes mellitus (T2DM). 185 Hypothalamic insulin action takes place in synergism with leptin signalling, and therefore has also a potent anorexigenic effect. 186 The morbid Moreover, PWS patients were reported to have hyperinsulinemia response to an oral glucose tolerance test (OGTT) when compared to normal weight controls, 193 while obese non-PWS controls displayed comparable OGTT readout to PWS individuals in this study. 193 In another study, an intravenous glucose tolerance test (IGVTT) showed comparable glucose assimilation coefficient between PWS and nonsyndromic obese population. 194 These data on glucose assimilation and insulin concentration upon oral glucose challenge indicate that the insulin sensitivity among non-syndromic obese and PWS should also be comparable. Interestingly, after a protein meal ingestion, PWS patients showed a similar insulin peak to healthy weighted controls, whereas obese non-PWS controls had a clear insulin peak upon the meal consumption. 195 Schuster et al 196 demonstrated that PWS infants have lower insulin response to oral glucose test compared to BMI and age matched individuals. In the same study, no differences were found in glucose or insulin levels on PWS adults when compared to lean or obese controls. The authors discuss that the PWS limited group size might influence the interpretation of this data. Interestingly, on OGTT, PWS presented delayed peak of glucose and insulin when compared to obese controls. It was proposed that this difference might be explained by reduced pancreatic β-cell responsiveness to glucose fluctuations in PWS population. Consistently with previous literature, the authors found no differences in glucose assimilation assessed by IVGTT. 194 However, insulin and peptide C plasma levels were reduced in PWS. 196 This led to reduced insulin to glucose ratios during IGVTT in PWS compared to obese control group. Furthermore, the authors were the first to demonstrate increased hepatic insulin extraction and insulin clearance in PWS compared to obese controls.
A recent study described a heightened fasting insulin levels and sensitivity in PWS infants compared to BMI and age-matched controls. 197 However, this study lacks the comparison between PWS and lean subjects. Therefore, it is impossible to determine if in this cohort, whether PWS are normoinsulinemic during fasting. In concordance with the insulin sensitivity data, a more recent report described that PWS individuals have lower fasting insulin levels when compared to obese controls. Not only insulin concentrations were lessened, but also PWS individuals were also proven to have greater insulin sensitivity, which were compared to lean controls. 191 Thus, although the prevalence of T2DM is greater in PWS, an unexpected increase in insulin sensitivity is found within this population as well. The link between insulin resistance and T2DM is long recognized. 198 In addition, molecular resistance to insulin signalling is one of the most accurate predictors of development of T2DM and the main therapeutic target for this disorder. 199,200 There are potential mechanisms that are used to explain this paradoxical clinical feature of PWS. Firstly, although PWS patients display abnormal fat deposition, that occurs preferentially in subcutaneous depots rather in visceral ones. 191,[201][202][203] The visceral fat deposition, which is observed in obese controls, has a positive correlation with lowered sensitivity to insulin signalling. 204 Moreover, it is known that PWS patients are deficient in growth hormone. 6,205 This hormone has glucoregulatory roles and, it is particularly interesting to pinpoint that there is a transient induction of insulin resistance in puberal development due to elevated secretion of growth hormone. 206 Lastly, increased insulin sensitivity in PWS population can eventually be explained by increased levels of adiponectin. 202 This is an adipose tissue-derived hormone and targets insulin-producing cells in the pancreas. 207 Adiponectin is associated with increased insulin sensitivity and has anti-inflammatory properties. 208 Compared to non-syndromic obese population, PWS individuals have higher plasma concentrations of this hormone. This is further correlated with insulin sensitivity in a PWS cohort. 202 It is proposed that increased overall fatty acid oxidation promoted by adiponectin signalling in skeletal muscle leads to increased insulin sensitivity, as previously stated. 209 However, whether this is true for the pathophysiology of PWS remains unknown.
Our current understanding of the endocrine imbalance on the hypothalamic malfunction is still extremely poor. One of the biggest gaps to be filled in that sense is the "chicken-or-egg question" between the endocrine imbalance and the defective neuronal functioning. Is disrupted hypothalamic function the primordial cause of the obese phenotype, which is accompanied by endocrine alterations? Or is the defective endocrine production and/or secretion the main cause of the hypothalamic function?

| HYP OTHAL AMIC G LIAL CELL S AND INFL AMMATORY PATHWAYS IN PWS
The consumption of an obesogenic diet leads to structural and functional damage of the neuronal populations engaged with energy homeostasis regulation. 82,87,210 The malfunction of those neurons is closely related to the activation of microglia cells. 83 rather than in the hypothalamus. 228 In brief, NPY is capable to inhibit LPS-induced proinflammatory mediators (such as nitric oxide production and cytokines/interleukins) and has a suppressor effect on phagocytic capacity of microglia. 229

ACK N OWLED G EM ENT
The authors would like to thank Irina Milanova for careful revision of the manuscript.