ENDOCRINE LINKS BETWEEN FOOD REWARD AND CALORIC HOMEOSTASIS

Abstract

Both intrinsic and extrinsic (endocrine) inputs to the CNS modulate motivation for feeding. Endocrine inputs such as insulin and leptin can have very rapid effects, but also the potential for chronic actions to decrease rewarding attributes of food. Future studies should elucidate the neural and cellular mechanisms which underlie these endocrine actions in the CNS.

To consider the concept of caloric homeostasis, one must consider the organism as a member of a species. Evolutionary perspective tells us that the “job” of a species is to successfully replicate itself and to that end, an animal must survive long enough to achieve reproductive competence and, at a minimum, mate successfully. In many animal species, survival of a parent to support and nourish a young animal to a state of adequately functional independence is an additional requirement. Presumably with that goal, the homeostasis of food intake, as well as fluid balance, evolved in a physically challenging environment where adequate supply of food (or fluid) could never be assured. Hence, we live as animals with brains constantly monitoring for food, and the avoidance of potential starvation. Highly sensitive and effective neural systems have developed that allow us to find the activity of eating reinforcing and motivating, and to learn and remember very rapidly about stimuli and environments which signal the likely availability of more food. Recent research suggests that the reward value of food is not only assessed by intrinsic central nervous system (CNS) circuitry but can be modified by hormones that signal acute and chronic caloric status to the CNS (Figlewicz, 2003).

It is clear that nutritional status has a dramatic effect upon the rewarding aspect(s) of many stimuli, including food. Specifically, food restriction or food deprivation increase the rewarding attributes of stimuli assessed in several types of behavioral paradigms. This has been documented repeatedly in the drug abuse literature (Carroll & Meisch, 1984). It has been demonstrated, for example, that both drug self-administration (motivated work for a dose of drug) and the propensity to relapse to drug-taking are enhanced in animals that are acutely food restricted or more chronically food deprived (Comer et al., 1995;Shaham et al., 2000). The paradigm of lateral hypothalamic (LH) self-stimulation has been interpreted as a behavioral activity which accesses ‘reward circuitry’ and brain self-stimulation is significantly increased in association with food deprivation (Margules & Olds, 1962; Carr, 2007). Food as a stimulus is, likewise, more rewarding in food-deprived animals, when evaluated in paradigms that assess reward. The physiological advantage of this is obvious.

From the psychological perspective, one might invoke increased drive; enhanced salience of a food stimulus, or of cues that signal the availability of food; and decision-making that involves increased effort. Numerous behavioral approaches have been used to assess ‘food reward’. Thus, self-administration of food (pressing a lever to obtain a small amount of food) (Figlewicz et al., 2006); moving down a runway or maze for food (Salamone et al., 2007); the conditioning of a place preference with a drug (Bell et al., 1997) or with food (Figlewicz et al., 2001; Lepore et al., 1995); the hedonic valuation of food (Levine, 2006); and cue-conditioned overfeeding (Petrovich & Gallagher, 2003) elicit stronger performances when an animal is food-deprived. Performance in these tasks may represent enhanced reinforcing and motivating properties of food, and learned association of being in a location in which food was available. That the tasks evaluate different aspects of reward and motivation is reflected in findings that the CNS circuitry engaged by the different tasks is overlapping but not identical (Ikemoto & Wise, 2004; McBride et al., 1999).

Research from physiological psychologists and behavioral neuroscientists has been elucidating the neural substrates mediating or modulating the enhanced drive and effort for food, in circumstances of food deprivation. Berridge and colleagues (Berridge, 2004) have presented the differential constructs of ‘wanting’ food (motivation) and ‘liking’ food (hedonic valuation) and provided substantial evidence that these may be mediated in the central nervous system (CNS) by somewhat unique circuitries. Nutritional status has been shown to modulate both motivation for food as well as, in certain paradigms, the efficacy of signals relating to food palatability (Barbano & Cador, 2005). Critical for the mediation of motivation for food are the mesolimbic dopamine (DA) neurons, with cell bodies in the ventral tegmental area (VTA) projecting to the prefrontal cortex and to the striatum, including the nucleus accumbens (NAc) (Smith, 1995; Smith, 2004; Wise, 2005). Critical for mediation of palatability and palatability-based food choices are populations of endogenous opioidergic neurons and their receptors (Bodnar, 2004; Glass et al., 1999; Levine, 2006; Olszewski & Levine, 2007; Zhang & Kelley, 2000). Exogenous administration of mu, delta, and kappa opioids has been shown to increase feeding; mu opioids have been studied most extensively, increasing food intake when administered into a number of CNS sites. Synaptic connections between the VTA, NAc, and medial hypothalamus provide the anatomic basis for direct ‘crosstalk’ between CNS circuitry which assesses energy balance and caloric homeostasis, and the lateral hypothalamus may be a key intermediary site, with reciprocal synaptic connections to all of these regions (Berthoud, 2004; Berthoud, 2007). Little is currently known about intrinsic CNS mechanisms that may mediate the enhanced rewarding value of food in association with caloric deprivation, although both dopaminergic and opioidergic pathways have been implicated (Carr, 2007).

Our focus has been upon the contribution of extrinsic signals to the CNS. Food deprivation and caloric restriction invoke a pattern of well-characterized endocrine changes, which is modified by the duration and severity of calorie deprivation. Key endocrines include the adrenal glucocorticoids and ghrelin, secreted in increased amounts in association with calorie deprivation; and insulin and leptin, which are secreted in association with food intake and increased adipose mass, and conversely, are decreased in the circulation in association with calorie deprivation. The glucocorticoids have been implicated in increased drug taking (Piazza & LeMoal, 1997) but have not been extensively studied in formal paradigms of food reward. Their involvement in stress-associated intake of ‘comfort foods’ has been recently described (LaFleur, 2006). Ghrelin, an orexigenic hormone (Cummings, 2006), has been shown to increase feeding when administered directly into the VTA (Naleid et al., 2005), and can increase sucrose self-administration when given into the third cerebroventricle (Overduin, Cummings, and Lattemann, unpublished). The actions of insulin and leptin have been much better characterized and they are implicated in the change of reward behavior in general, and food reward specifically, both in the scenario of the fed vs. calorie-deprived states, and when administered exogenously during the fed state. Both insulin and leptin have been identified as ‘candidate adiposity signals’ which inform the CNS regarding the energy balance of the organism (Baskin et al., 1999; Niswender et al., 2004).

In 1994, we reported that intraventricular (IVT) administration of insulin in the rat increased expression of an important regulatory protein for dopamine (DA) activity, the re-uptake transporter (DAT) in the VTA and substantia nigra compacta (SNc) (Figlewicz et al., 1994). The DATs are expressed on the synaptic membrane of DA neurons, and act to clear DA from the synapse, thus terminating signaling. A great deal has been learned in the intervening fifteen years about the specific neuronal expression, and regulation of both synthesis and activity (Blakely et al., 2005) of the DAT. Our initial observation intrigued us, and we made the subsequent observations that DAT expression was decreased in association with acute food restriction, and further, that activity of the DAT measured by in vitro voltammetric uptake of DA was decreased in striatal slices (containing DA nerve terminals) taken from food restricted animals (Patterson et al., 1998; Zhen et al., 2006). However, if insulin was added to the in vitro bath, DA uptake by the ‘food restricted’ striatal slices was normalized. Thus, it appears that insulin regulates both synthesis and functional activity of the DAT. The further implication of this is that insulin increases clearance of, and decreases synaptic concentrations of DA, hence DA activity. This line of research has now been extended by Galli and colleagues. Using cultured cells transfected with DATs and expressing insulin receptors, they have shown that the insulin effect is due to enhanced plasma membrane recruitment of DATs, and is dependent upon activation of a PI3 kinase signaling pathway (Garcia et al., 2005). PI3 kinase signaling has been identified as critical for the energy-regulatory actions of insulin at the hypothalamus (Niswender et al., 2003).

We next explored the possible in vivo or behavioral relevance of our observations with the DAT. Work from Shaham and colleagues (2000) called attention to the well-documented observation, discussed above, that food restriction or deprivation enhanced the addictive and relapse potencies of a number of types of abused drugs (Shalev et al., 2001), and that one key anatomic substrate is the midbrain DA neurons. We speculated that insulin might be one specific factor contributing to this phenomenon. We thus begin to test the hypothesis that insulin would decrease ‘reward’ behavior for food, evaluated in several behavioral paradigms. Coinciding with this, studies of the role of leptin in blunting reward were begun in other labs. Independently, Carr and colleagues (2000), and Fulton and colleagues (2000) reported that IVT insulin and leptin could reverse the enhanced responding for brain electrical stimulation that occurs with food restriction. These studies, along with our work demonstrating insulin and leptin effects on food-conditioned place preference (Figlewicz et al., 2004) and sucrose self-administration (Figlewicz et al., 2006), have provided substantive support for our hypothesis and additionally implicated leptin in this modulatory action.

The observation of decreased food reward with IVT insulin or leptin administration leads to the question of target CNS sites of action. In 2003 we reported the expression of both insulin receptors and leptin receptor immunoreactivity on TH-positive (presumptively, DA) neurons in the VTA and substantia nigra compacta (Figlewicz et al., 2003). Leptin receptor mRNA on TH neurons has been reported by Hommel and colleagues (2006). We have also observed immunoreactivity for both insulin and leptin receptors on TH-negative neurons (presumptively, GABAergic), and we have obtained preliminary evidence for insulin receptor immunoreactivity on some identified GABA neurons in the VTA. In particular we find co-expression with TH (hence, within DA neurons) at a high level in the anterior VTA. The behavioral relevance of this receptor expression has been demonstrated by a few labs, including our own: insulin and leptin administered directly into the VTA decrease both baseline, and mu opioid-stimulated, feeding (Figlewicz et al., 2007; Fulton et al., 2006; Hommel et al., 2006; Morton et al., 2007).

Several groups have begun to elucidate cell signaling pathways activated by insulin and leptin within the VTA. We have observed increased PIP3 immunoreactivity in response to direct injection of insulin or leptin into the VTA (Figlewicz et al., 2007). DiLeone and colleagues (Hommel et al., 2006) and Fulton and colleagues (2006) have identified activation of the JAK-Stat3 pathway in response to peripheral or local (intra-VTA) leptin administration, and leptin administration and activation of the JAK-Stat pathway is also critical for the decrease in food intake when leptin is injected into the VTA (Morton et al., 2008). Since both leptin and insulin can activate multiple signaling pathways including PI3 kinase (Carvalheira et al., 2005; Mirshamsi et al., 2004; Niswender et al., 2004), the connection between cell signaling pathways and behavioral effects will require further exploration. From a broader perspective it is clear that the VTA can serve as a direct anatomical substrate for either leptin or insulin action. Further, although the feeding paradigms used to obtain the results described above do not allow us to state definitively that insulin or leptin are decreasing the rewarding attributes of food, they imply a physiological role for this major component of CNS reward circuitry in feeding.

A second CNS site which has already been recognized as an important anatomical substrate for the energy regulatory actions of insulin and leptin is the arcuate nucleus (ARC). We have followed our initial observation that intraventricular insulin (and leptin) decreases sucrose self-administration, with a study to determine the anatomical locus(loci) mediating this effect. Insulin was administered directly into the NAc, paraventricular nucleus of the hypothalamus (PVN), VTA, and ARC. We observed an anatomically-specific effect of insulin to decrease sucrose self-administration given into the ARC, which quantitatively accounts for the effect of insulin given IVT (Figlewicz et al., 2008). This result suggests that endocrine input at the medial hypothalamus must then be relayed to the brain reward circuitry to modulate sucrose self-adminstration. It should be emphasized that this effect occurs at a dose of insulin which does not have chronic effects to decrease 24-hr food intake or body weight. Thus, the effect of insulin on motivation or effort for food may be a contributing mechanism to, and not a consequence or correlate of, the well-defined effect of insulin to regulate body weight and adiposity.

The ability of energy regulatory signals to restrain body weight and adiposity can be modified under a number of circumstances. Similarly, the ability of energy regulatory signals to restrain food reward can also be modified. We have observed that feeding rats a high fat baseline diet (chow with 30% or more fat) results in both an increase in baseline responding for sucrose, and prevention of the ability of IVT insulin and leptin to decrease responding (Figlewicz et al., 2006). A parallel impairment of the energy-regulatory action of insulin and leptin at the medial hypothalamus had been made by several laboratories (Clegg et al., 2005; Munzberg & Myers, 2005). Recent studies suggest that impaired intracellular signaling may underlie or contribute to this impairment. Since we have determined that the effect of insulin to decrease sucrose self-administration is likewise mediated through the medial hypothalamic ARC, it is possible that similar changes of cell signaling modify insulin effects on both food reward and caloric homeostasis. Alternatively, downstream or independent changes may occur in brain reward circuitry as a result of different baseline diet composition or its perceived hedonic value. Given the propensity of Westernized diets to be high in fat (e.g., >30% of daily caloric intake), further mechanistic insight into the interaction between high(er) fat diets and motivation for food treats is warranted.

Persuasive as the collective evidence with exogenous insulin and leptin is, before we conclude more definitively that very acute effects of these adiposity signals to decrease food reward are physiological, and are a component of their more chronic effect to decrease food intake and body weight, the parameters of their efficacy need to be addressed and defined more clearly. This has already been done in the work of Fulton (2000) and Carr (2000) and their respective colleagues, in which food restriction-induced increases of lateral hypothalamic self-stimulation are reversed to baseline levels by intraventricular leptin or insulin. Comparable studies specifically testing food reward paradigms have not been carried out. Note that the effects of insulin or leptin given IVT or into specific brain regions usually occur within 10–30 minutes.

The goal of our studies has been to simulate a ‘dessert’ or ‘between meal snack’ neuroendocrine milieu. Data on transport kinetics have suggested that elevations of plasma insulin or leptin are associated with increased CNS levels in a 20–30 minute timeframe (Banks & Kastin, 1998; Figlewicz, 2003). Theoretically, a rise of insulin at the onset of a meal would result in elevated brain levels by ‘dessert time’ in a reasonably paced meal, that could provide a brake on reward value of dessert. The onset of leptin action is less easy to fit into a behavioral framework as leptin levels show a circadian pattern associated with the taking of meals (Kalsbeek et al., 2003) but leptin secretion is not tied to the ingestive bout in the way that insulin is. Intriguingly Morton and colleagues (2007) have found that intra-VTA leptin enhances the ability of the meal-regulatory peptide cholecystokinin to decrease size of acute feeding bouts, and thus in the context of normal circadian behaviors this may be a truly physiological role for leptin. Another possibility to be explored is that leptin in the CNS sets a ‘tone’ and the shorter-timeframe, meal-related, excursions of insulin can amplify that tone or perhaps require it, for insulin efficacy. This potential synergy remains to be explored.

Granted that these issues are addressed, and we arrive at the conclusion that endocrine signals, such as insulin and leptin, can serve to restrain any of the aspects of food reward (motivation, effort-based decision making, hedonic valuation, reinforcement), we still are left with the question of why it is advantageous for a homeostatic system to evolve such mechanisms. Perhaps the answer lies in the need of an organism to engage in other behaviors, in order to survive and reproduce, and hence once an animal has eaten enough to generate signals of plenty (insulin being relevant for shorter-term repletion and leptin being relevant for longer-term calorie storage), food becomes a less relevant stimulus, and the animal turns to other critical behaviors: drinking, for fluid homeostasis; social or affiliative behaviors as species-appropriate; mating behaviors as appropriate; or even seeking a safe and quiet place to rest to allow the body an opportunity to digest and assimilate calories. The ‘setpoint’ of the CNS, as currently evolved, to strongly bias humans to eat whenever food is available pushes modern man to a new and somewhat maladaptive homeostasis from the perspective of being longer-lived. Understanding the brain reward system and its modulation may lead us to design behavioral strategies (the diversion of other activities that are as engaging and rewarding as eating) in addition to the currently limited medical approaches to curb eating. Respect for ‘the wisdom of the body’, rather than denial, should guide our future basic and translational research on food reward contributions to caloric homeostasis.