|Home | About | Journals | Submit | Contact Us | Français|
Food intake is regulated by many factors, including sensory information, metabolic hormones, and the state of hunger. In modern humans, the drive to eat has proven to be incompatible with the excess food supply present in industrialized societies. A result of this imbalance is dramatically increased rates of obesity during the last 20 years. The rise in obesity rates poses one of the most significant public health issues facing the United States and yet we do not understand the neural basis of ingestive behavior, and specifically, our motivation to eat. Understanding how the brain controls eating will lay the foundation for systematic dissection, understanding and treatment of obesity and related disorders. The lack of control over food intake bears resemblance to drug addiction, where loss of control over behavior leads to compulsive drug use. Work in laboratory animals has long suggested that there exist common neural substrates underlying both food and drug intake behaviors. Recent studies have demonstrated direct leptin effects on dopamine neuron function and behavior. This provides a new mechanism by which peripheral hormones influence behavior and contributes to a more comprehensive model of neural control over food intake.
Rates of obesity are increasing worldwide (1). The natural drive to eat, combined with a surplus of readily available food, are together partly responsible for this modern epidemic. In order to understand the source of this problem, we need to better define the molecular and neural mechanisms by which the brain regulates food intake, and why it is often difficult to control food ingestion in times of excess. The discovery that peripheral metabolic hormones, normally studied for their activity in the hypothalamus, act directly on dopamine neurons of brain dopamine centers to modulate feeding behavior, has provided new mechanisms for behavioral control. Here, recent findings are reviewed and with an emphasis on the potential implications as well as open questions that remain.
While peripheral signals that communicate metabolic state, such as insulin, had already been isolated and well characterized, the cloning of the mouse obese (ob) mutation initiated a new and productive stage of feeding research. The gene product mutated in ob/ob mice was shown to be a small protein expressed in adipocytes and was later given the name leptin (2). Genetic and molecular studies, along with identification of the leptin receptor (3, 4), confirmed the importance of the hypothalamus in regulating feeding behavior and metabolism, and suggest that it is a primary site of action for circulating factors (5). More recent work has led to a cellular-based hypothesis whereby leptin receptor activity inhibits the orexigenic while exciting the anorexigenic neurons within the arcuate nucleus (6).
While the isolation of critical peripheral metabolic signals has invigorated the field of feeding research, the mechanism by which leptin, or other signals, act on brain circuitry to alter animal behavior remains poorly understood. Current work in the field is focused on identifying which second order feeding centers, such as the paraventricular nucleus (PVN) or lateral hypothalamus (LH), are critical for mediating leptin responses downstream of the arcuate nucleus. However, it is not clear which neural circuits are modified to result in a change in the motivation to seek food or in how that food is experienced. Strikingly, cortico-striatal circuits that are integral for motivated behavior and inhibitory control (7–9) have not been integrated with metabolic hormones that are known to influence food intake. Feeding is a complex behavior that is clearly influenced by many non-homeostatic mechanisms. It seems likely that leptin-initiated signaling circuits eventually converge with circuits that regulate reward and motivation, such as the mesolimbic dopamine system that projects from the VTA to the nucleus accumbens (NAc), as well as the prefrontal cortex.
Administration of leptin has been shown to modulate behaviors that are dependent upon the mesolimbic dopamine circuit. For example, leptin administration alters intracranial self-stimulation (10), suggesting an interaction with dopamine circuits that are thought to underlie the behavior (11). Other data have demonstrated that leptin attenuates the increased propensity to heroin relapse caused by food restriction (12) and can modulate conditioned place preference for sucrose or high fat food (13, 14). These data suggest that leptin can modify reward based behavior and that this may occur via alteration in function of dopamine pathways (15). While this interaction with dopamine pathways could occur downstream from leptin’s effects on the hypothalamus, recent data also suggests a direct action of leptin on dopamine centers of the brain.
While many studies have investigated leptin signaling in the hypothalamus, evidence existed to suggest that leptin signals directly to many other brain regions. There is extensive extrahypothalamic Lepr expression in brain regions including the hippocampus, cortex and the midbrain (16). Grill and colleagues have shown that leptin targets in the brainstem are important for controlling food intake (17), indicating that leptin’s effects could be mediated via direct signaling to regions of the brain beyond the hypothalamus. Following a report that Lepr is expressed in most TH-positive neurons of the VTA (18), we recently demonstrated a role for leptin in regulation of VTA neuronal function and feeding behavior (19). Interestingly, a companion paper by Fulton et al. showed that leptin administration can modify responses to psychostimulants (20). The work also demonstrated that the VTA neurons are responsive to leptin and that ob/ob mice (missing leptin) have attenuated amphetamine sensitization, which is reversed with leptin administration (20). While both papers establish effects of leptin in the VTA, the Fulton et al study demonstrates that genetic loss of leptin results in attenuated mesoaccumbens function, while Hommel et al. suggest that leptin directly attenuates activity of dopamine neurons.
A possible explanation for the contrasting data is that acute and chronic leptin may have distinct effects on VTA neurons. The ob/ob mutant mice could have altered dopamine neuron synaptic structure or function that was effectively reversed, or rescued, by the long-term (10 day) leptin treatment used in the paper (20). Effects were also seen in wildtype mice, suggesting that chronic leptin exposure in normal animals also increases response to drugs of abuse (20). Leptin has been shown to be critical for normal development of hypothalamic neuronal projections (21), and leptin administration can result in rearrangement of synapses (22). Whereas the acute suppression of dopamine neuron function seen in Hommel et al. is consistent with previous data demonstrating reduced dopamine release (23) and attenuated dopamine-dependent behaviors (10, 12–14) following leptin administration. A recent study also shows that leptin deficient people demonstrate hyperactivated ventral striatal regions in response to food images. Leptin administration suppressed this hyperactivity while also leading to reduced preference ratings of the images (24). Finally, the hormone ghrelin activates VTA dopamine neurons (25), suggesting opposite regulation by orexgenic factors.
However, it should also be emphasized that there likely exists a number of potential mechanisms by which leptin modulates dopamine neurons leading to changes in animal behavior. In fact, earlier studies demonstrated that leptin can have opposing effects on reinforcement depending upon the brain circuits being stimulated (10). More generally, dopamine modulation of the midbrain may influence a number of brain regions, with potentially contrasting effects on behavior. While VTA dopamine neurons have a prominent direct projection to the NAc, the VTA also projects to the prefrontal cortex, that in turn influences drug addiction and related behavior via action on the NAc (26). Other projections, such as those to the amygdala, could also influence reward related behavior. Leptin receptor is also expressed in the substantia nigra (16, 18, 19), which would likely influence the dorsal striatum. Finally, a recent study suggests that the dopamine neurons with unconventional firing properties may influence many of these regions and deserve further study (27).
While the complete behavioral consequences of this new site of metabolic sensing still remain to be elucidated, it is notable that dopamine influences food intake in a number of ways. While genetic disruption of dopamine production results in mice that are aphagic (28), it is difficult to discern the role of dopamine in general motivation and movement versus its specific role in feeding. In fact, dopamine deficient mice drink normal amounts of sucrose, and pharmacological blockade of dopamine in the NAc does not reduce the amount of food consumed in a free-feeding paradigm (29, 30). Previous data with genetic models suggested a predominant role for striatal dopamine in food intake and NAc in locomotor function (31). A more recent report, however, uses an elegant retrograde virus strategy to specifically deliver the rescuing transgene to the midbrain neurons that project to the dorsal striatum (32). The mice have normalized locomotor behavior and only partially normalized food intake, consistent with a role for both dorsal and ventral striatum (NAc) in controlling components of food intake. However, as noted above, pharmacological studies in the NAc have shown little effect of D1 receptor antagonism on ad libitum food intake (30). In contrast, D1 receptor antagonists result in attenuation of operant performance under high work loads (ie., many lever presses required to obtain food) (33–35). These data are consistent with the notion that dopamine in the NAc is necessary for seeking and exerting effort to obtain food, but may not be required for intake under ad libitum, or low effort, conditions. Other studies suggest additional roles for NAc dopamine, some of which could be highly relevant to excessive food intake. For example, intra-NAc amphetamine increases the seeking of sucrose in response to a conditioned cue (36), and dopamine in the NAc plays a role in driving behavioral responses to cues that predict reward (37). These and other data have provoked discussion beyond the scope of this review about what dopamine does and how important it is for food reward (38–40).
While there are many ways to evaluate the importance of a molecular pathway in biology, one approach is to look for regulation under different physiological states. For example, voltammetry experiments have revealed increased dopamine release before and during lever pressing for food with a return to baseline after food ingestion (41), consistent with a potential functional role at different points of the behavior. Likewise, it will be critical to determine how dopamine neuron leptin signaling pathways are regulated and how this interfaces with other neuronal pathways that are known to be altered by metabolic state.
The VTA dopamine neurons show prominent projections to NAc, which has been well studied for its role in drug addiction (42). Because of this, it is often assumed that this pathway will influence food intake via effects on hedonic perception, or response to palatable food. However, animal model data suggests that the mesolimbic reward centers may influence both hedonic and non-hedonic intake. A set of studies in the NAc has also shown a role for glutamate, GABA, and opioids in regulation of food intake. While μ-opioid receptor agonists lead to a selective increase in intake of highly palatable food (43), glutamate antagonists and GABA agonists lead to a general increase in food intake (44, 45). As with GABA and glutamate, it does not appear that NAc dopamine, or dopamine in general, influences hedonic responses or “liking” (36, 46). In fact, this is a core component of the incentive-sensitization theory of addiction, whereby liking and wanting are dissociable with dopamine playing a role in the latter (47). It is notable that infusion of μ-opioid agonists into the NAc shell dramatically increases the effort, assessed by lever pressing, that animals will make to obtain food (48), suggesting that opioids in the NAc can modulate multiple components of behavior relevant to food intake. Microinfusion studies with opioids have suggested that a small region of the NAc shell mediates this enhancement of hedonic “liking”, whereas a larger area of the shell mediates feeding responses (49).
In all, these data suggest that a main role of NAc shell may be as a behavioral switch whereby silencing of the neurons would allow feeding to occur without clear effects on hedonic components (8). While the studies have relied on relatively blunt pharmacological approaches, recent in vivo recordings support this model since most NAc neurons are inhibited before food intake (50, 51). Interestingly, the lateral hypothalamic peptide MCH also promotes general food intake (52), and appears to inhibit neurons of the NAc (Sears et al., under review).
Direct action of leptin on dopamine circuits has profound implications on feeding and addiction research, as it provides a molecular and neuronal mechanism for how the state of the body may modulate the drive to eat. Moreover, this connection to peripheral signals places the mesolimbic system within a larger physiological framework of eating regulation. As noted above, other results have shown that additional metabolic factors, such as insulin (18, 54) and ghrelin (25, 55), appear to act on the VTA and modulate neuron function or food intake and reward. Thus, as with the hypothalamus and brain stem, we now consider the midbrain region as a potential site for integration of peripheral metabolic signals.
The full behavioral consequences of metabolic hormones acting on dopamine neurons will require much more work to appreciate. Recent data from our lab have shown leptin-mediated suppression of drug seeking during withdrawal from psychostimulants. These and other observations have provoked renewed attention toward defining potential common mechanisms between food intake and drug addiction. Likewise, imaging data suggesting that obese patients resemble drug addicts (53) has also inspired discussion and set up a potential convergence between animal models and clinical conditions. At this point, it seems even more critical that we fully describe the molecular and neural players so that we can interpret and extend the findings to humans.