Hypothalamic opioid–Melanocortin appetitive balance and addictive craving

Summary

Whilst the parallels between drug and food craving are receiving increasing attention, the recently elucidated complex physiology of the hypothalamic appetite regulatory centres has been largely overlooked in the efforts to understand drug craving which is one of the most refractory and problematic aspects of drug and behavioural addictions. Important conceptual gains could be made by researchers from both appetite and addiction neuroscience if they were to have an improved understanding of each others’ disciplines. It is well known in addiction medicine that the use of many substances is elevated in opiate dependency. There is voluminous evidence of very high rates of drug use in opiate agonist maintained patients, and the real possibility exists that opiate agonist therapy therefore increases drug craving. Conversely, opiate antagonist therapy with naloxone or naltrexone has been shown to reduce most chemical and behavioural addictions, and naltrexone is now being developed together with bupropion as the anti-obesity drug “Contrave”. Hypothalamic melanocortins, particularly α-MSH, are known to constitute the main brake to consumptive behaviour of food. There is a well described antagonism between melanocortins and opioids at many loci including the hypothalamus. Administration of exogenous opiates is known to both suppress α-MSH and to stimulate hedonic food consumption. Opiate maintenance programs are associated with weight gain. As monoamines, opioids and cannabinoids are known to be involved in appetite regulation, and as endorphin opioids are known to be perturbed in other addictions, further exploration of the hypothalamic appetite regulatory centre would appear to be an obvious, albeit presently largely overlooked, locus in which to study drug and other craving mechanisms.

Introduction

Central to the DSM-IV definition of drug addiction is the concept of persistent use of addictive chemicals despite obvious negative consequences or sanctions [1]. It is this preservative and enduring aspect of addiction which is frequently its most troublesome aspect to the individuals, families and communities who are plagued by it and to their treating clinicians. Similar features apply to other addictions centred around behaviours such as eating, gambling or sexual appetites. Similarities such as this suggest that there are a number of similarities between chemical and behavioural addictions, so that the actual subject of addiction can be conceptualized as an addictive focus rather than a completely unique addiction per se [2], [3], [4]. And of course it is well known that over a lengthy career of addiction, patients commonly change focus from one substance or behaviour to another. Others not infrequently pursue multiple addictive foci concurrently. This stubborn and entrenched behaviour apparently regardless of consequence is often spoken of as drug (or other) “craving”.

Similarities between drug and eating addictions have been noted in recent literature [5], [6], [7], [8], [9]. The physiological regulation of appetite is now known to be complexly controlled by multiple factors at both the peripheral and central level. Some of the many orexigenic neurotransmitters are neuropeptide Y (NPY), agouti-related protein (AgRP), melanin concentrating hormone (MCH), orexins A and B, endorphins, galanin, cortisol, prolactin, endocannabinoids, testosterone and ghrelin, as well as the amino acids glutamate and γ-butyric acid (GABA). Anorexigenic hormones include α-, β- and γ-melanocyte-stimulating hormone (α-MSH, β-MSH and γ-MSH), adrenocorticotrophic hormone (ACTH, together referred to as “melanocortins”), leptin, serotonin, estrogen, norepinephrine, adrenaline, dopamine, adiponectin, corticotrophin releasing hormone (CRH), insulin, cholecystokinin, glucagon-like peptide (GLP), cocaine amphetamine related transcript (CART), pro-inflammatory cytokines (particularly IL-1β, tumour necrosis factor-α, and IL-6 and including IκBα, cyclooxygenase-2, nuclear factor-κ beta, toll-like receptor (TLR)-4 -CD14) and peptide YY (PYY), together with arginine, glucose, and lipid breakdown products and fatty acids and distension of the stomach and duodenum [9], [10], [11]. From the arcuate nucleus of the hypothalamus efferent signals are relayed to many brain nuclei and into the remainder of the body by diffuse pathways including sympathoadrenal outflow, vagal efferents, and the hypothalamic–pituitary–adrenal (HPA) axis. Oddly, however it would seem that the complex physiology which is being described to account for consumptive behaviour in relation to food has not been carefully investigated in relation to other agents.

Drug craving is classically studied in characteristic paradigms in preclinical laboratories such as place preference, which typically rates the preference of a rodent for a drug exposed box over a non-drug associated space. Craving can also be studied clinically by PET and fMRI scanning [12]. Using such paradigms careful pharmacological or genetic studies are often performed to examine the manner in which such manipulations alter measurable drug preference. Positive associations have been shown for drug preference with many alterations of both neuroscience [12], [13] and neuroimmune [14], [15], [16], [17] parameters. The neurobiological substrate of craving is usually conceptualized as being mediated by mesocorticolimbic structures particularly those of the ventral tegmental area, the amygdala, the nucleus accumbens, the cingulate gyrus and the medial prefrontal cortex [12], [13].

The lateral hypothalamic nucleus stimulates eating whilst the ventromedial nucleus signals satiety and halts eating behaviours. Crosstalk between these two major centres generally determines food intake behaviour. The paraventricular hypothalamic nucleus also signals satiety and the dorsomedial hypothalamic nucleus stimulates eating. Gut hormones and peripheral inputs converge on the arcuate nucleus. In the arcuate nucleus a mutually antagonistic balance is maintained between orexigenic AGRP/NPY expressing and anorexigenic proopiomelanocortin (POMC)/CART expressing neurons with both sets of neurons suppressing each other’s activity. Appetitive behaviour is frequently characterized as the output of this balanced system. The major neurotransmitter output of the POMC neurons is α-MSH which signals via MCR3 and MCR4. Defects in this signalling system are associated with obesity [9], [18]. POMC neurons are exquisitely sensitive to high plasma glucose levels, and increase their level of firing in response [9].

Interestingly major centres known to be involved in addiction have also been identified to play critical roles in eating. The amygdala has close ties to the olfactory signalling system and can powerfully impact appetitive drives [10]. Midbrain dopaminergic neurons also express leptin and ghrelin receptors, and leptin administration increases brain self-stimulation, and satiety – reward behaviour is associated with a dopamine spike in the nucleus accumbens [9].

The overall schema of the physiology of the hypothalamic–pituitary–adrenal axis has been understood for several decades. Hypothalamic corticotrophin releasing hormone (CRH) is released onto corticotrophs of the anterior pituitary which induces the production and release of the peptide preproopiomelanocortin. In its turn, POMC is cleaved by a series of proteolytic steps into a variety of smaller peptides including adrenocorticotrophic hormone (ACTH), the μ-agonist β-endorphin, α-, β- and γ-melanocyte-stimulating hormone (MSH) and α-, β- and γ-lipotropin (LTN) [9]. ACTH and the MSH’s are also known as melanocortins. The major proteases responsible for these post-translational cleavages are the prohormone convertases 1 (or 3) and 2 (PC1/PC3 and PC2). By regulating the expression of these enzymes, tissue specificity can be achieved in terms of the products released. Hence ACTH (from PC1/3) is the principal product of the anterior lobe pituitary corticotrophs and α-MSH is the major product of the pituitary neurointermediate lobe melanotrophs and produced by the activity of PC2 [9], [19]. The anterior lobe is the major source of endorphins (which include met-enkephalin) which arise from PC1/3 activity. As CRH stands at the apex of the stress response pathways, it is important to appreciate that β-endorphin is released as part of the stress response pathway. This system is controlled by negative feedback loops at multiple levels. CRH haplotype has been shown to be related to alcohol consumption in rhesus macaques [20].

The POMC gene promoter occurs in three distinct leader exons prior to the transcription start site. POMC transcription is stimulated by CRH, cytokines (especially leukaemia inhibitory factor (LIF), interleukin (IL) -6, IL-1β), vasopressin, leptin, estrogen, catecholamines, vasoactive intestinal peptide, high levels of glucose (by a free radical dependent mechanism), signal transducer and activator of transcription-3 (STAT3), Specificity protein 1 (Sp1), forkhead box O1 (FoxO1), starvation, and neurotrophic factors (brain derived neurotrophic factor (BDNF) and ciliary derived neurotrophic factor (CDNF)) [21], [22], [23], [24], [25], [26], [27], [28]. Many of these stimuli synergize together. Importantly components of the innate immune system such as the toll-like receptor (TLR)-4 and the nucleotide oligomerization domain (NOD)’s are expressed in normal anterior pituitary cells [29] and lipopolysaccharide has been shown to directly stimulate POMC expression [27], [30]. POMC transcription is inhibited by cortisol. LIF is a cytokine produced in the pituitary and hypothalamus and signals via the JAK–STAT pathway and acts synergistically with CRH and induces direct STAT3 binding to the POMC promoter [9], [31], [32].

Pre- and neo-natal overfeeding is known to induce a metabolic syndrome including both insulin- and leptin-resistant phenotype later in development in both childhood and adulthood. One fascinating study in this area looked at the epigenetic effects of such overfeeding on the methylation state of the POMC promoter. CpG dinucleotides within two Sp1 promoter regions were found to be very significantly hypermethylated, and these changes in the promoter were found to be directly related to insulin- and leptin- hypo-responsiveness and thereby the other functional changes which underlie the diabetic and cardiovascular complications of this syndrome. Remarkably blood glucose levels correlated closely with the level of POMC promoter hypermethylation [28].

α-MSH has many actions [33], [34], [35], but two prominent ones include a potent anorectic stimulus which acts to suppress appetite in the hypothalamic appetite centre [9], [36], [37], [38], [39], and significant immunosuppression [14], [33], [35], [40], [41], [42], [43], [44]. α-MSH also inhibits morphine analgesia, and its antagonists reduce morphine tolerance [14], [45], [46], [47], [48]. Indeed chronic stress has been shown to overload pituitary melanotrophs and result in the demise of these cells [49]. The immunosuppressive effects of α-MSH have been noted to oppose those of brain pro-inflammatory cytokines such as IL-1 [14].

Disruption of the melanocortin signalling system either at the ligand or receptor level has been defined as an important cause of obesity [9], [50] and has also been linked with chronic stress states [51], such as are common amongst methadone and opiate dependent clients [52], [53], [54]. This relative melanocortin deficiency explains the described stimulation to appetite which occurs with opiate administration [38], [55], [56], [57] doubtless exacerbated by a preference for high energy and sweet foods [56], [58], [59], [60], [61].

Mutually antagonistic interactions have been described in several loci including the spinal dorsal horn, amygdala [34], [35], [38], [46], [48], [59], [62], [63], [64], [65], [66], [67], [68], [69] and hypothalamus [33], [37], [39], [70], [71]. Opiate withdrawal has been precipitated by melanocortin agonists and opiate tolerance has been shown to be antagonised by melanocortin antagonists [34], [48], [63], [64]. Inhibition of melanocortin signalling alleviates chronic pain [14], [45], [46], [47], [64], [72]. And alterations in the α-MSH content of various hypothalamic nuclei have been noted in chronic ethanol exposure (see following discussion) [73].

Opiates have been noted to powerfully stimulate the immune system both by TLR4 [74], [75] and via chemokines [76], [77], [78]. Cannabinoids also stimulate chemokines release [76]. Activation of the innate immune system by these routes TLR4 activation results in pro-inflammatory cytokine release which, as noted above, is likely to increase POMC transcription and post-translational processing. The strongly anti-inflammatory actions of α-MSH directly oppose such activities [33], [34], [35]. Similarly, systemic opiate administration is also associated with indices of elevated immune activity in the general circulation including CRP, γ-globulins, lymphocytes, monocytes, IL-1β, Il-6 and TNF-α [78], [79], [80], [81], [82]. These cytokines gain access to the CNS through a variety of mechanisms including directly through fenestrated capillaries in the hypothalamus, via the CSF in the circumventricular organs lining the third ventricle including the hypothalamic nuclei, by active transport across or synthesis within the endothelium particularly when these cells are inflamed or activated, and by other routes [9]. Again the immunosuppressive actions of the melanocortins will tend to suppress the pro-inflammatory actions of opiates whether such stimuli arise extrinsic or intrinsic to the neuraxis.

Relative melanocortin deficiency is an important cause of hypothalamic inflammation. Melanocortins improve attention, concentration and memory, and effect oposite to that of opiates [34]. The sedation and somnolence accompanying opiate treatment is well known [79], [83]. This implies that opiate associated weight gain will occur preferentially into the abdominal compartment. Adipose deposits in the visceral distribution have been shown in many studies to be strongly pro-inflammatory, and to release many pro-inflammatory cytokines into the systemic circulation including TNF-α, IL-1β, IL-6, and insulin resistance mediating adipokines including leptin and resistin, and suppressed levels of anti-inflammatory adiponectin [84]. Elevated levels of IL-1β has been shown in opiate dependent patients [82], and it has been noted that IL-1β haplotype is related to the incidence of alcohol and heroin dependence in clinical populations [85].

Administration of xenobiotic opiates to humans has been demonstrated to suppress POMC and its derived opiates [39], [48], [86], in a similar fashion to exogenous glucocorticoids. Blockade or reversal of the normal nycthemeral rhythm by opiates has been demonstrated in sensitive studies [87]. Contrariwise, conditions of adrenal insufficiency such as Addison’s Disease are marked by hyperpigmentation relating to increased levels of α-MSH and related melanocortins (β-MSH, γ-MSH and ACTH) [9]. It therefore follows that α-MSH levels in opiate dependency would be expected to be reduced, and this has been confirmed by several workers [37], [38], [39], [46], [62], [63] including after chronic administration [45], [47], [48], [64].

In a paradigm of acute stimulation by opiates rats experience stimulation of the HPA axis whereas in humans the effect on the HPA axis is suppression. In both species tolerance to chronic adminstration is quickly induced [88], [89], [90] but particularly in man this is only partial [87], [91], [92].

The role of opiates to stimulate the consumption of calorie rich sugar- and lipid- laden foods is well defined [55], [56], [58], [60], [61], [65], [93], [94], [95]. This action has been associated with hedonic and pleasure driven eating which can override metabolic and homeostatic signals of satiety [38], [58]. Intraaccumbens injection of the pure μ-opiate agonist D-Ala2-N-Me-Phe4-gly5-ol-enkephalin (DAMGO) is a well characterized model stimulating rats to consume a high fat diet. Opiate stimulation of orexin signalling in the ventral tegmental area is key to driving this behaviour. Furthermore nucleus accumbens DAMGO stimulates expression of neuronal activation markers in the perifornical hypothalamus where efferent neurons make anatomically intimate contact with orexin neuronal bodies [65]. There is evidence of significant weight gain on long term opiate maintenance programs [83], [96], [97], [98]. Similarly there is evidence of high rates of consumption of heroin, cocaine, benzodiazepines, cocaine, cannabis, methadone and amphetamines along with other high risk behaviours such as unprotected sexual contacts in opiate maintained patients [99], [100], [101], [102], [103], [104], [105], [106], [107], [108], [109]. Anecdotal clinical evidence suggests that some forms of body disfigurement including cutting the forearms, tattooing and deliberate burns from cigarettes and tobacco lighters are also common in this group. This contrasts with antagonist treated patients where naltrexone has been used to treat self-mutilation (see below).

Contrariwise classical opiate receptor antagonists, notably naltrexone, have been associated with reduction of food intake in many models [110], [111], [112], [113], [114]. The combination of bupropion and naltrexone has been shown to be particularly efficacious [115] and will now be marketed as the combination anti-obesity drug “Contrave” [116]. Naltrexone is well known to block the effects of exogenous administered opiates and this is predictable from its effects as an antagonist of classical opiate receptors. It is also highly effective in the management of chronic alcoholism especially where its administration is supervised by a significant other [117], [118], [119], [120], [121]. It has also been described as having useful activity in a variety of other addictions including cannabis, amphetamine, cocaine, self-mutilation, overeating-obesity and gambling [2], [111], [116], [122], [123], [124], [125], [126], [127], [128], [129], [130], [131], [132]. Naltrexone has been described as mildly stimulating the HPA axis in humans [133], an effect opposite to that described for opiate agonists (see above). This wide range of activity of naltrexone against diverse addictive foci raises the question as to its real site of action, particularly in the light of exactly reciprocal and equally diverse effects of full opiate agonists.