Ventral tegmental area dopaminergic involvement in motivation and reward

In order to understand how activity of VTA DA neurons affects the motivation for and anticipation of reward, we will first describe general characteristics of the DA system and the technologies used to determine its function in relation to behaviour.

The mesDA system originates from dopaminergic neurons in the VTA. The VTA itself is a heterogeneous brain area that contains dopaminergic (about 60%), γ-aminobutyric acid (GABA) ergic (about 30%) and glutamatergic (about 10%) neurons⁽ Reference Margolis, Lock and Hjelmstad ^{5 –} Reference Yamaguchi, Wang and Li ^{7 )}. VTA DA neurons typically show three modes of firing, but not all modes are always seen in individual neurons. Neurons can be silent or tonically active, and they can also fire in bursts. When they are tonically active they typically fire action potentials at low frequencies. During both conditions VTA neurons can fire in phasic burst patterns. Typically, bursts consist of an average of 3–4 action potential at a 20–50 Hz frequency⁽ Reference Schultz ^{8 )}.

For its actual function, the VTA relies on release and subsequent receptor binding of DA in target areas, such as the NAc. Compared with other neurotransmitters such as glutamate and GABA, DA levels stay elevated for a longer time, allowing for widespread trans-synaptic effects⁽ Reference Fields, Hjelmstad and Margolis ^{2 )}.

Neurons in DA-projection areas contain different DA receptors that have distinct affinities for DA. Dopamine 1 receptor (D1R) and dopamine 2 receptor (D2R) are the most abundant DA receptors in the striatum and have been best described in the effects of DA signalling. Due to tonic activity of VTA DA neurons, there is a basal level of DA in target areas, which is thought to preferentially activate the high affinity D2R. Burst firing induces a transient increase in DA levels, which also causes the activation of the lower affinity D1R⁽ Reference Wightman and Robinson ^{9 ,} Reference Gonon ^{10 )}.

While in vivo electrophysiological tools allow for direct measurement of VTA DA neuronal activity during behaviour, techniques such as fast scan cyclic voltammetry and microdialysis allow for the measurement of DA release in target areas. Between these latter two it has been suggested that microdialysis is only able to measure changes in tonic DA release, brought about by basal DA firing, while fast scan cyclic voltammetry is able to measure DA transients caused by VTA DA burst firing as well⁽ Reference Wightman and Robinson ^{9 )}.

Functionally, the mesDA system has been implicated in various behaviours, and different modes of VTA DA neuronal activity contribute to different behavioural aspects. While both tonic and phasic activity are important for motivation, tonic activity plays an important role in enabling locomotor activity, while phasic bursting has been linked to cue–reward seeking behaviour⁽ Reference Schultz, Dayan and Montague ^{11 –} Reference Yun, Wakabayashi and Fields ^{13 )}. The mesDA system has also been implicated in coding for salient or aversive events and stimuli, and cognitive functioning⁽ Reference Schultz ⁸ ,Reference Schultz ^{12 ,} Reference Schultz ^{14 ,} Reference Bromberg-Martin, Matsumoto and Hikosaka ^{15 )}. In this review, we only focus on the role of DA in motivation and cue–reward seeking.

The mesDA system is crucial for the development of cue–reward seeking in which neutral cues are associated with rewards they predict. In the context of feeding behaviour, it is important for animals to be able to assess nutritional value of foods when they encounter related cues, such as the smell or sight of particular food. Over time, specific cues illicit specific conditioned anticipatory responses that are coupled to the reward that the cue predicts. DA neurons show burst firing during unexpected primary rewards, such as food, and this burst firing is higher for larger rewards⁽ Reference Tobler, Fiorillo and Schultz ^{16 )}. When a cue is coupled to a reward over multiple sessions, animals learn that the cue predicts the reward and DA burst firing shifts to the cue onset⁽ Reference Schultz, Dayan and Montague ^{11 ,} Reference Waelti, Dickinson and Schultz ^{17 ,} Reference Fiorillo, Tobler and Schultz ^{18 )}. In this context, DA neurons have been shown to encode a reward prediction error that drives learning.

The reward prediction error encodes the difference between a predicted outcome and the actual outcome⁽ Reference Schultz, Dayan and Montague ^{11 )}. These characteristics are also seen in NAc DA release⁽ Reference Day, Roitman and Wightman ^{19 –} Reference Roitman, Stuber and Phillips ^{21 )}, as measured using fast-scan cyclic voltammetry. Importantly, when DA signalling in the NAc is disturbed⁽ Reference Nicola, Taha and Kim ^{22 )} associative learning is impaired. Rather than only encoding the presence of a reward, this signal also correlates with the size of the reward⁽ Reference Tobler, Fiorillo and Schultz ^{16 ,} Reference Roesch, Calu and Schoenbaum ^{23 ,} Reference Cohen, Haesler and Vong ^{24 )}.

Overall, VTA DA neuronal firing encodes a reward prediction error that is important for cue–reward associative learning. This role of DA neurons is important in feeding behaviour. In the following section, we will discuss how the mesDA system has been implicated in feeding behaviour.

Dopamine and feeding behaviour

MesDA signalling is thought to be essential for the motivation to earn food rewards⁽ Reference Salamone, Wisniecki and Carlson ^{25 –} Reference Nowend, Arizzi and Carlson ^{28 )}, but not for the actual consumption of these rewards. The consumption of food is mediated by dopaminergic projections from substantia nigra neurons to the dorsolateral striatum⁽ Reference Szczypka, Kwok and Brot ^{29 –} Reference Salamone, Mahan and Rogers ^{32 )}. Furthermore, the anticipation of food rewards is thought to be mediated by a distributed neural circuit that besides the mesDA system includes hypothalamic nuclei⁽ Reference Acosta-Galvan, Yi and van der Vliet ^{33 –} Reference King and Hentges ^{35 )}. In this review, we focus only on the mesDA system.

In general terms, the mesDA system has been implicated in the rewarding effects of natural rewards such as food or sex, and drug rewards, as these have been shown to increase VTA DA firing activity and release of DA in target areas⁽ Reference Schultz, Dayan and Montague ^{11 ,} Reference Schultz ^{36 –} Reference Berridge and Robinson ^{40 )}. Moreover, VTA DA burst firing has been shown to be rewarding, as animals actively self-administer these bursts⁽ Reference Witten, Steinberg and Lee ^{41 )} and direct activation of VTA DA neurons increases operant responses for food rewards⁽ Reference Adamantidis, Tsai and Boutrel ^{42 )}. In addition, induced burst firing of VTA DA neurons elicits conditioned place preference⁽ Reference Tsai, Zhang and Adamantidis ^{43 )}.

In the case of food rewards, it has been shown that food-restricted animals have lower basal DA levels in the NAc⁽ Reference Pothos, Creese and Hoebel ^{44 )}, although these changes are not always observed⁽ Reference Verhagen, Luijendijk and Korte-Bouws ^{45 –} Reference Cadoni, Solinas and Valentini ^{47 )}. However, most studies show an increased DA release when animals are given rewards such as food⁽ Reference Pothos, Creese and Hoebel ^{44 ,} Reference Verhagen, Luijendijk and Korte-Bouws ^{45 ,} Reference Hernandez and Hoebel ^{48 )}. Importantly, these animals also show increased motivation for food during food restriction⁽ Reference Wise ^{37 ,} Reference Carr ^{49 )}. Food-restricted animals exert more effort to obtain food with increasing energetic value, and also respond more in an operant task when they are able to gain more food⁽ Reference Wise ^{37 ,} Reference Fulton ^{50 )}. This increase in motivation can be modulated by changes in DA signalling. When synaptic DA levels are increased through local infusion of amphetamine into the NAc⁽ Reference Sulzer, Chen and Lau ^{51 )}, performance on a progressive ratio (PR) task for a sucrose pellet is increased⁽ Reference Zhang, Balmadrid and Kelley ^{52 )}. Similarly, when VTA DA neuronal firing is increased through optogenetic techniques, sucrose consumption is increased⁽ Reference Domingos, Vaynshteyn and Voss ^{53 )}.

By blocking DA signalling through systemic injection of a D1R or D2R antagonist, food-restricted rats show a decrease in performance on a PR for food⁽ Reference Aberman, Ward and Salamone ^{54 )}. Furthermore, there is a similar decrease in performance when the DA neurotoxin 6-hydroxy is introduced into the NAc. This neurotoxin destroys DA neurons, decreasing NAc DA signalling⁽ Reference Aberman, Ward and Salamone ^{54 )}. This indicates that the increased motivation for food in these food-restricted animals is mediated through mesDA signalling.

Overall these studies indicate that DA signalling is important for feeding behaviour. In food-restricted animals, there is a higher increase in DA signalling when animals are given food or food-related cues are presented. These animals also show more motivation for obtaining food, and this motivation might be mediated by dopaminergic signalling. However, it remains to be determined whether the changes in DA signalling seen in food-restricted animals are due to direct changes in VTA DA neuronal firing and to what extent this is mediated by modulation in neurotransmitter signalling in target areas such as the NAc.

Mesolimbic dopamine in obesity

Animals which are kept on a diet that contains a high-fat high-sucrose (HFHS) content typically gain body weight, display increased motivation to obtain sucrose⁽ Reference Figlewicz, Bennett and Naleid ^{55 ,} Reference la Fleur, Vanderschuren and Luijendijk ^{56 )} and show changes in mesDA functioning. It is thought that some of these changes may be attributable to specific diets. Rats on a HFHS diet have decreased NAc D1R expression compared with controls⁽ Reference Alsio, Olszewski and Norback ^{57 )} and when exposed to both regular chow and an energy-dense high-fat diet that resembles cafeteria diets, their striatal D2R levels are decreased⁽ Reference Johnson and Kenny ^{58 )}, similar to human obese patients⁽ Reference Wang, Volkow and Logan ^{59 )}. Furthermore VTA DA transporter levels are increased, while tyrosine hydroxylase levels are decreased in high-fat fed mice⁽ Reference Alsio, Olszewski and Norback ^{57 ,} Reference Li, South and Han ^{60 )}. In these animals, DA turnover in the NAc is decreased⁽ Reference Davis, Tracy and Schurdak ^{61 )}. Further supporting this, animals which are fed a high-fat diet are unable to illicit conditioned place preference for amphetamine⁽ Reference Davis, Tracy and Schurdak ^{61 )}, supporting a possible decrease in NAc DA levels. These molecular changes indicate that DA production and signalling might be decreased.

It has been hypothesised that initial excessive feeding triggers increased DA release, and as a compensatory mechanism DA receptors are then desensitised, resulting in a tolerance to the rewarding effects of excessive feeding. This then causes further excessive feeding in an attempt to increase DA signalling⁽ Reference Wang, Volkow and Fowler ^{62 )}. In accordance with this, rats that are chronically exposed to a HFHS diet are less susceptible to the rewarding effects of lateral hypothalamus (LH) self-stimulation, and this decrease is potentiated when the D2R is knocked down in the striatum⁽ Reference Johnson and Kenny ^{58 )}.

In summary, the mesDA system is important for feeding behaviour, and modulated during changes in energy balance. In the case of food restriction, mesDA signalling increases above free fed animals when subjected to food or tasks that are awarded by food. Opposite to this, obese animals seem to have lower DA-mediated signalling which animals try to compensate for by increasing their behaviour for obtaining food.

It should be taken into account that NAc DA levels do not necessarily reflect VTA neuronal activity, as they are also dependent on reuptake and clearance of synaptic DA. In addition, dopaminergic neurotransmission depends on the number and type of DA receptors. When animals are kept on a high-fat or HFHS diet, NAc DA-mediated signalling is decreased, indicated by lower D1R and D2R levels, lower DA turnover in the NAc and higher DA transporter expression in VTA neurons⁽ Reference Alsio, Olszewski and Norback ^{57 ,} Reference Johnson and Kenny ^{58 ,} Reference Davis, Tracy and Schurdak ^{61 )}. It remains to be investigated whether basal tonic and burst firing is changed in VTA DA neurons in obese animals. To fully understand the role of VTA DA neurons in this process, it is important to be able to record their activity directly in vivo.

Although our current understanding of the changes induced by increased or decreased food availability on VTA DA neurons is incomplete, it is apparent that this system changes when energy balance is disturbed. As leptin and ghrelin are strongly regulated during negative and positive energy balance, these peptides likely modulate the mesDA system, and might do so at the level of the VTA.

Inputs to the ventral tegmental area that are involved in feeding and energy balance

To understand how the mesDA system plays a role in energy balance, it is important to understand how different inputs important for feeding regulate this circuit. In this review, we only focus on the VTA, as the origin of the dopaminergic projections that make the mesDA system. The VTA receives inputs from peptide hormones, and several brain areas that play a role in the regulation of feeding behaviour and energy balance. Among these brain areas there are distinct hypothalamic nuclei that are important for feeding. This review is limited to the effects of the hormones leptin and ghrelin on the VTA, and on hypothalamic areas that are known to project to the VTA⁽ Reference Fields, Hjelmstad and Margolis ^{2 ,} Reference King and Hentges ^{35 ,} Reference Philpot, Dallvechia-Adams and Smith ^{63 ,} Reference Greenwell, Zangen and Martin-Schild ^{64 )} (see Fig. 1).

Fig. 1.

Schematic representation of ventral tegmental area (VTA) inputs from the hypothalamus which are modulated by leptin and ghrelin. VTA dopaminergic neuronal activity is decreased by leptin⁽ Reference Hommel, Trinko and Sears ^{65 )} and increased by ghrelin⁽ Reference Abizaid, Liu and Andrews ^{66 )}. Orexin-producing neurons from the lateral hypothalamus (LH) are inhibited by leptin⁽ Reference Yamanaka, Beuckmann and Willie ^{67 )} and excited by ghrelin⁽ Reference Yamanaka, Beuckmann and Willie ^{67 ,} Reference Lawrence, Snape and Baudoin ^{68 )}. Orexin A increases excitatory projections onto VTA dopamine (DA) neurons⁽ Reference Borgland, Chang and Bowers ^{69 )}. LH neurotensin (Nts)/γ-aminobutyric acid (GABA)-producing neurons are excited by leptin, and project onto LH orexin neurons, as well as VTA DA neurons⁽ Reference Leinninger, Opland and Jo ^{70 )}. VTA GABA neurons directly inhibit VTA DA neurons, and are also modulated by hypothalamic inputs. Endomorphin-producing neurons from the ventromedial hypothalamus (VMH) project to the VTA⁽ Reference Thompson, Canteras and Swanson ^{71 )}, and opioid release in the VTA inhibits VTA GABA neurons⁽ Reference Johnson and North ^{72 )}. It remains to be investigated whether these hypothalamic neurons are also leptin or ghrelin responsive. Pro-opiomelanocortin (POMC) neurons from the arcuate nucleus project to the VTA⁽ Reference King and Hentges ^{35 )}. They produce α-melanocyte stimulating hormone that excites VTA GABA neurons⁽ Reference Korotkova, Sergeeva and Eriksson ^{73 )}, and are themselves excited by leptin⁽ Reference Cheung, Clifton and Steiner ^{74 ,} Reference Schwartz, Woods and Porte ^{75 )}. These POMC neurons are inhibited by arcuate nucleus agouti-related protein (AgRP) neurons that in turn are inhibited by leptin and excited by ghrelin⁽ Reference Schwartz, Seeley and Campfield ^{76 –} Reference Nakazato, Murakami and Date ^{78 )}. Note that this schematic only includes hypothalamic inputs onto the VTA that have been described in the literature, but other hypothalamic inputs may exist. Also, projections shown here do not represent all the known projections of the different neuronal subtypes.

Effects of leptin on the ventral tegmental area and mesolimbic dopamine system

Leptin is an adipose-derived hormone that signals information on adiposity to the brain. It increases energy expenditure, and decreases food intake, resulting in a decrease in body mass⁽ Reference Campfield, Smith and Guisez ^{79 –} Reference Pelleymounter, Cullen and Baker ^{81 )}. Dysfunctional leptin signalling has been implicated in the development of obesity in the spontaneous obese ob/ob mice⁽ Reference Zhang, Proenca and Maffei ^{82 ,} Reference Speakman, Hambly and Mitchell ^{83 )}. Leptin binds to the leptin receptor (LepR), a single pass membrane receptor, leading to dimerisation, the activation of Janus kinase and phosphorylated signal transducer, and activator of transcription 3. Injection of leptin directly into the VTA increases phosphorylated signal transducer and activator of transcription 3 levels and activates Jak-2 signalling pathways⁽ Reference Morton, Blevins and Kim ^{84 ,} Reference Fulton, Pissios and Manchon ^{85 )}. Among its different sites of effects, it also acts on neurons in the arcuate nucleus of the hypothalamus and mediates signalling for a decrease in food intake. It does so by acting on neuropeptide Y (NPY)/Agouti-related protein (AgRP) neurons⁽ Reference Schwartz, Seeley and Campfield ^{76 ,} Reference van den Top, Lee and Whyment ^{77 )} and pro-opiomelanocortin (POMC) neurons⁽ Reference Cheung, Clifton and Steiner ^{74 ,} Reference Schwartz, Woods and Porte ^{75 )}.

The LepR is also expressed in the VTA, where 75–90% of the LepR positive neurons are dopaminergic⁽ Reference Hommel, Trinko and Sears ^{65 ,} Reference Figlewicz, Evans and Murphy ^{86 ,} Reference Leshan, Opland and Louis ^{87 )}, although it was reported that this represents only a fraction of the total amount of VTA DA neurons⁽ Reference Leshan, Opland and Louis ^{87 )}. In VTA acute brain slices, leptin hyperpolarises putative DA neurons and decreases their neuronal firing frequency⁽ Reference Hommel, Trinko and Sears ^{65 )}. Furthermore leptin injections decrease concentrations of extracellular DA levels in the NAc⁽ Reference Krugel, Schraft and Kittner ^{88 )}. Some of the leptin responsive VTA DA neurons project to the NAc⁽ Reference Fulton, Pissios and Manchon ^{85 )}, while it has been argued that the majority of LepR-containing VTA neurons project to central amygdala neurons containing cocaine and amphetamine responsive transcript⁽ Reference Leshan, Opland and Louis ^{87 )}. Supporting a possible implication of amygdala-mediated signalling, mice show an anxiogenic effect when LepR is knocked down in VTA DA neurons specifically, and this phenotype can be saved by local microinjection of a D1R antagonist into the central amygdala⁽ Reference Liu, Perez and Zhang ^{89 )}. Overall LepR is expressed in VTA DA neurons that project to the central amygdala, and to a lesser extent, in neurons that project to the NAc. Activation of this receptor by leptin decreases DA neuronal firing and subsequent NAc DA release.

Mice that lack leptin (ob/ob mice) are obese, are hyperphagic and have decreased energy expenditure. They have reduced DA levels in the NAc and decreased tyrosine hydroxylase levels in the VTA⁽ Reference Fulton, Pissios and Manchon ^{85 )}. Moreover, similar to obese patients⁽ Reference Wang, Volkow and Logan ^{59 )}, they also have reduced D2R in the VTA⁽ Reference Roseberry, Painter and Mark ^{90 )}, and this can be restored by leptin treatment⁽ Reference Pfaffly, Michaelides and Wang ^{91 )}. Interestingly, rats that are unable to produce the LepR, termed fa/fa rats or Zucker rats also become obese, and, similar to ob/ob mice, their VTA D2R levels are reduced⁽ Reference Thanos, Michaelides and Piyis ^{92 )}. Overall, these studies indicate that DA signalling originating in the VTA is reduced in leptin-deficient animals. Curiously, intra-VTA administration of leptin also reduces DA release in the NAc⁽ Reference Krugel, Schraft and Kittner ^{88 )}, an effect that may be caused by a direct inhibitory effect of leptin on VTA DA neurons⁽ Reference Hommel, Trinko and Sears ^{65 )}. Apart from possible developmental changes that occur in ob/ob mice, the reduced DA release in these mice may be due to a lack of leptin signalling in neurons projecting to the VTA, such as those from the LH (see section ‘Indirect effects of peptide hormones from hypothalamic inputs’). These studies, using leptin-deficient mice clearly show that leptin signalling is important for mesDA signalling.

This requirement of leptin for mesDA signalling is supported by studies in which leptin was administered (rather than removed) and food-related behaviours were investigated. Leptin injections in the ventricle and in the VTA decrease food consumption, while overall locomotor activity is not impaired⁽ Reference Morton, Blevins and Kim ^{84 ,} Reference Leshan, Opland and Louis ^{87 ,} Reference Krugel, Schraft and Kittner ^{88 )}. Ventricular leptin injections decrease performance on a PR task for food⁽ Reference Figlewicz, Bennett and Naleid ^{55 )}, indicating that the motivation for a food reward is decreased. Taken together these data suggest an involvement of the mesDA system in leptin's effect on reduced food intake and food reward. In animals on a HFHS diet, ventricular injections have no effect on food consumption⁽ Reference Figlewicz, Bennett and Naleid ^{55 )}, suggesting a leptin resistance. Mice with leptin over-expression in the VTA gained less body weight over time when compared with controls, and interestingly, they preferred a high-fat diet more than controls. Phosphorylated signal transducer and activator of transcription 3 activation was lower in the VTA of these leptin over-expressing animals, while also being lower than chow controls in HFHS-fed animals⁽ Reference Matheny, Shapiro and Tumer ^{93 )}. This indicates that chronically increased VTA leptin levels induced by HFHS diets or by leptin overexpression in the VTA blunt VTA LepR-mediated signalling.

Viral knock down of the LepR in the VTA using RNA interference increases food intake under both balanced and high-fat diets⁽ Reference Hommel, Trinko and Sears ^{65 )}. These animals also show increased locomotor activity and preference for sucrose and high-fat food. However, in contrast to this, transgenic animals that have the LepR knocked out specifically in VTA DA neurons show no change in body weight or food intake on a chow or high-fat diet⁽ Reference Liu, Perez and Zhang ^{89 )}. It is possible that these discrepancies arise from the specificity of the knockdown. While Hommel and co-workers most likely knocked down the LepR in all VTA neurons, including GABA and glutamatergic neurons⁽ Reference Hommel, Trinko and Sears ^{65 ,} Reference Figlewicz, Evans and Murphy ^{86 ,} Reference Leshan, Opland and Louis ^{87 ,} Reference Liu, Perez and Zhang ^{89 )}, Liu et al.⁽ Reference Liu, Perez and Zhang ^{89 )} knocked out the LepR specifically in DA neurons. Furthermore, the use of a transgenic knock out might facilitate compensatory changes. Further investigation into the role of leptin in the VTA remains necessary to understand its role on VTA-mediated feeding behaviour.

Overall, based on electrophysiological evidence in vitro and microdialysis studies in vivo leptin decreases the activity of VTA neurons, and subsequent DA release for instance in the NAc. By doing so it decreases motivation for food resulting in lower food consumption. However, when leptin levels are constantly elevated by diet or over-expression, leptin-mediated signalling is decreased and the effect of additional leptin is blunted (leptin resistance). Therefore it can be concluded that the effects of leptin-signalling are likely dependent on metabolic state and diet.

Effects of ghrelin on the ventral tegmental area and mesolimbic dopamine system

Ghrelin is a twenty-eight-amino acid peptide hormone that is primarily produced by the stomach. Its plasma levels rise during fasting and decreases after a meal, and it promotes feeding⁽ Reference Nakazato, Murakami and Date ^{78 )}. It acts on the growth hormone secretagogue receptor (GHSR), a G protein-coupled receptor that couples to Gα_q signalling⁽ Reference Wettschureck, Moers and Wallenwein ^{94 )}. It is expressed in different brain areas closely involved in feeding behaviour, including hypothalamic areas (e.g. arcuate nucleus, dorsomedial hypothalamus and ventromedial hypothalamus and paraventricular nucleus), but, importantly, also in the VTA⁽ Reference Zigman, Jones and Lee ^{95 ,} Reference Mondal, Date and Yamaguchi ^{96 )}. Within the arcuate nucleus, the GHSR is expressed on NPY and AgRP neurons, through which ghrelin stimulates feeding behaviour by increasing meal frequency, but not meal size⁽ Reference Nakazato, Murakami and Date ^{78 ,} Reference Tschop, Smiley and Heiman ^{97 ,} Reference Horvath, Diano and Schop ^{98 )}. In addition, ghrelin is involved in the anticipation of meals. While its plasma levels are correlated with food anticipatory activity (FAA)⁽ Reference Verhagen, Egecioglu and Luijendijk ^{99 )}, FAA is decreased in animals in which ghrelin signalling is blocked, either in GHSR knock out mice or by injection of a GHSR antagonist⁽ Reference Verhagen, Egecioglu and Luijendijk ^{99 –} Reference Blum, Patterson and Khazall ^{101 )}. Chronic ghrelin injections furthermore increase c-Fos expression in different hypothalamic areas, in a pattern that is similar to that induced by food restriction⁽ Reference Blum, Patterson and Khazall ^{101 )}.

In the VTA, the GHSR is co-localised with tyrosine hydroxylase in 35–60% of the GHSR positive neurons⁽ Reference Abizaid, Liu and Andrews ^{66 ,} Reference Zigman, Jones and Lee ^{95 ,} Reference Guan, Yu and Palyha ^{102 )}. Opposite to leptin-induced inhibition of putative VTA DA neurons, ghrelin increases the firing rate of DA neurons in slices. Interestingly, this effect is abolished by blockade of glutamate, but not GABAergic neurotransmission⁽ Reference Abizaid, Liu and Andrews ^{66 )}, indicating that these effects are mediated by presynaptic glutamatergic inputs, and not by direct effects of GHSR on DA neurons. Furthermore, 90 min after intraperitoneal injection of ghrelin, electron microscopy shows that VTA DA neurons have increased incoming glutamatergic synapses and decreased GABAergic synapses. These results indicate that the effects of ghrelin might mediate an increased excitation and decreased inhibition. This is supported by the fact that ghrelin increases the frequency of miniature excitatory postsynaptic currents, but decreases miniature inhibitory postsynaptic currents onto VTA DA neurons⁽ Reference Abizaid, Liu and Andrews ^{66 )}.

Systemic ghrelin administration increases DA levels in the NAc shell⁽ Reference Quarta, Di Francesco and Melotto ^{103 )}, and DA turnover, an effect that is absent in GHSR knock out animals⁽ Reference Abizaid, Liu and Andrews ^{66 )}. More specifically, local infusion of ghrelin into the VTA increases the number of lever presses for a sucrose reward on a PR task and fixed ratio tasks, and also increases food intake in free-fed rats⁽ Reference Skibicka, Hansson and Alvarez-Crespo ^{104 –} Reference Naleid, Grace and Cummings ^{106 )}. Furthermore, intra-VTA ghrelin injections increase locomotor activity and DA release in the NAc⁽ Reference Jerlhag, Egecioglu and Dickson ^{107 )}. Chronic ghrelin infusion in the VTA also dose-dependently increases food intake⁽ Reference King, Isaacs and O'Farrell ^{105 )}. Importantly, increases in food intake by systemic injection of ghrelin are blocked by local ghrelin antagonist injection into the VTA and chronic blockade of VTA GHSR abolishes food intake increase on a high-fat diet⁽ Reference Abizaid, Liu and Andrews ^{66 ,} Reference King, Isaacs and O'Farrell ^{105 )}. These results indicate that the VTA at least partly mediates the increases in motivation and consumption of food by ghrelin.

In food-restricted animals, ghrelin receptor activation contributes to FAA⁽ Reference Verhagen, Egecioglu and Luijendijk ^{99 )} and increases PR performance for food⁽ Reference Skibicka, Hansson and Alvarez-Crespo ^{104 )}. This increase in performance can be blocked by VTA injection of a GHSR antagonist⁽ Reference Skibicka, Hansson and Alvarez-Crespo ^{104 )}. Finally, ghrelin deficient and GHSR knock out animals do not increase their food intake over days when they are food restricted overnight⁽ Reference Abizaid, Liu and Andrews ^{66 )}.

Ghrelin signalling in the VTA also appears to play a role in food choice behaviour. When given the choice between standard chow and more palatable food, both transgenic GHSR knock out mice, as well as rats systemically treated with a ghrelin antagonist, show suppressed intake of palatable food, but not of chow intake⁽ Reference Egecioglu, Jerlhag and Salome ^{108 )}. Furthermore, ghrelin administration in the VTA increased intake of palatable food, but not chow, during an hour exposure to both types of food⁽ Reference Egecioglu, Jerlhag and Salome ^{108 )}.

Indirect effects of peptide hormones from hypothalamic inputs

Thus far we have discussed how the mesDA system is involved in motivated behaviour, feeding, and how VTA DA neurons might be affected directly through the actions of leptin and ghrelin. However, it is likely that these neurons are also affected by inputs from leptin- and ghrelin-responsive neurons in other brain areas. Among these brain areas, we focus on the effects on different hypothalamic nuclei, because of their importance in homoeostatic control of feeding behaviour.

Lateral hypothalamic control of the ventral tegmental area

The LH consists of multiple nuclei that receive inputs not only from the arcuate nucleus POMC/cocaine and amphetamine responsive transcript and NPY/AgRP neurons but also from NAc shell, cortical areas, amygdala, hippocampus and multiple brain stem nuclei. It projects to various brain areas, including the VTA, paraventricular and arcuate nuclei, amygdala, hippocampus and different cortical areas⁽ Reference Berthoud and Munzberg ^{109 )}. As such it is ideally situated to serve as a centre that integrates feeding-related information. LH neurons express different neuropeptides that are involved in feeding behaviour, including melanocortin-concentrating hormone, orexin/hypocretin, galanin, neurotensin (Nts), and cocaine and amphetamine responsive transcript⁽ Reference Berthoud and Munzberg ^{109 )}. Among these LH neurons projecting to the VTA, about 20% contain orexin⁽ Reference Fadel and Deutch ^{110 )}, a peptide stimulating food intake⁽ Reference Cason, Smith and Tahsili-Fahadan ^{111 )}. Some VTA DA neurons increase their firing rate to Orexin A and Orexin B, while others are unresponsive⁽ Reference Korotkova, Sergeeva and Eriksson ^{73 )}. However, it has also been reported that presumed VTA GABAergic neurons are similarly excited by orexins⁽ Reference Korotkova, Sergeeva and Eriksson ^{73 )}. Interestingly, LH orexin neurons play a role in the rewarding properties of food, as food-induced place preference increases c-Fos activation in LH orexin neurons and pharmacological blockade of the orexin receptor decreases PR performance for high-fat foods⁽ Reference Borgland, Chang and Bowers ^{69 )}. Their effects on the mesDA system can be modulated by changes in diet, as high-fat food self-administration increases orexins excitability of DA neurons⁽ Reference Borgland, Chang and Bowers ^{69 )}. Thus, LH orexin neurons that project to the VTA might contribute to dopaminergic responses for food rewards.

Indirect leptin effects on the ventral tegmental area via the lateral hypothalamus

Within the LH, the LepR is expressed on GABAergic neurons that project to the VTA and do not express orexin or melanocortin-concentrating hormone. In vitro, leptin induces a depolarisation in one-third of these neurons, while in vivo leptin increases expression of c-Fos in these neurons⁽ Reference Leinninger, Jo and Leshan ^{112 )}. With respect to feeding, leptin injections into the LH decrease body weight and food intake in rats and leptin-deficient ob/ob mice. In these mice, LH leptin injections also increase VTA tyrosine hydroxylase expression and NAc DA levels⁽ Reference Leinninger, Jo and Leshan ^{112 )}. In a follow-up study, it was shown that some of these LepR-positive LH neurons also express the neuropeptide Nts. These LepR- and Nts-expressing LH neurons are depolarised by leptin and they project both to orexin neurons in the LH, as well as to the VTA⁽ Reference Leinninger, Opland and Jo ^{70 )}. Direct application of Nts into the VTA dose-dependently decreases performance on an operant responding for food, mimicking the effects of systemic leptin⁽ Reference Kelley, Cador and Stinus ^{113 )}. Possibly, these LepR and Nts expressing GABA neurons inhibit the mesDA system by direct actions on the VTA, and by inhibiting LH orexin neurons. Supporting this, in slices a population of orexin-positive neurons in the LH are inhibited by leptin application⁽ Reference Yamanaka, Beuckmann and Willie ^{67 )}. A specific knock-out mouse line in which the LepR is deleted from Nts positive neurons in the LH shows lower NAc DA levels, and less amphetamine-induced locomotor activation, a process that is thought to be dependent on striatal DA levels⁽ Reference Leinninger, Opland and Jo ^{70 )}. Taken together these results suggest that leptin can reduce VTA DA activity, and subsequent DA release, through stimulation of inhibitory projections originating from the LH.

Indirect ghrelin effects on the ventral tegmental area via the lateral hypothalamus

Ghrelin might also have indirect effects on DA neurons through the LH, although this has not been studied extensively. Ghrelin activation increases food intake and activates orexin, but not melanocortin-concentrating hormone-positive neurons in the LH⁽ Reference Lawrence, Snape and Baudoin ^{68 )}. In acute brain slices, ghrelin also activates LH orexin neurons⁽ Reference Yamanaka, Beuckmann and Willie ^{67 )}. Ghrelin is found in terminals forming synapses with orexin-positive LH neurons. Ghrelin-stimulated food intake is reduced when orexin signalling is diminished and blocking NPY signalling completely abolishes ghrelin-mediated increase in food intake⁽ Reference Toshinai, Date and Murakami ^{114 )}. Overall, this indicates that interplay between direct ghrelin effects on LH orexin neurons and arcuate nucleus originating NPY and ghrelin might contribute to ghrelin effects on food intake.

Indirect effects via the ventromedial and dorsomedial hypothalamus

In food-restricted animals, cues that are associated with food can induce increased locomotor activity. Both the ventromedial hypothalamus and dorsomedial hypothalamus have been implicated in this FAA⁽ Reference Verhagen, Luijendijk and de Groot ^{115 )}, which is said to reflect both hunger and the motivation to eat⁽ Reference Merkestein, Brans and Luijendijk ^{116 )}. Interestingly, both the ventromedial hypothalamus and dorsomedial hypothalamus are sensitive to leptin and ghrelin, and have been implicated in the expression of FAA. Whereas decreased leptin levels increases FAA⁽ Reference Ribeiro, Ceccarini and Dupre ^{117 –} Reference Persons, Stephan and Bays ^{119 )}, reduced ghrelin signalling attenuates FAA⁽ Reference Verhagen, Egecioglu and Luijendijk ^{99 –} Reference Blum, Patterson and Khazall ¹⁰¹ Reference Verhagen, Egecioglu and Luijendijk ^{99 –} Reference Blum, Patterson and Khazall ^{101 ,} Reference Davis, Choi and Clegg ^{120 )}. Neurons in and near the dorsomedial hypothalamus and ventromedial hypothalamus project to the VTA⁽ Reference Greenwell, Zangen and Martin-Schild ^{64 )}, see also⁽ Reference Thompson, Canteras and Swanson ^{71 )}, and could, in principle, modulate VTA activity during FAA. These neurons projecting to the VTA produce the endogenous opioids endomorphin-1 and -2. These opioids act on μ opioid receptors that are expressed on VTA GABA neurons⁽ Reference Garzon and Pickel ^{121 )}, and upon activation inhibit VTA GABA neurons, thus causing a disinhibition of VTA DA neurons⁽ Reference Johnson and North ^{72 )}. Whether these neurons express receptors for leptin or ghrelin remains to be determined.

Indirect effects of arcuate nucleus inputs on the ventral tegmental area

The arcuate nucleus has been extensively implicated in the regulation of energy balance. Two prominent neuronal subtypes are NPY/AgRP-producing neurons and POMC-producing neurons. Activation of the POMC-positive neurons induces decreases in food intake. They primarily release α-melanocyte stimulating hormone and β-endorphin, although they can also release other neurotransmitters. A subpopulation of the arcuate nucleus POMC neurons projects to the VTA⁽ Reference King and Hentges ^{35 )}. In slice recordings of VTA neurons, α-melanocyte stimulating hormone was not seen to have an effect on DA neurons, but excited a subpopulation of VTA GABAergic neurons⁽ Reference Korotkova, Brown and Sergeeva ^{122 )}. Supporting this, it has also been found that local injection of α-melanocyte stimulating hormone into the VTA increases NAc DA levels, an effect that was found to be melanocortin receptor 4 receptor dependent⁽ Reference Lindblom, Opmane and Mutulis ^{123 )}. Leptin is known to activate POMC and doing so could alter VTA DA activity as well. Although ghrelin does not affect POMC neurons directly it can inhibit these neurons through inhibitory projections from AgRP cells onto POMC neurons⁽ Reference Nakazato, Murakami and Date ^{78 )}.

Concluding remarks

Among the circuits involved in feeding behaviour, the mesDA system is important for the motivational drive for food rewards and this circuit is affected by disturbances of energy balance. VTA DA neurons encode a reward prediction error that guides associative learning processes, but it remains unknown what the direct in vivo effects are of a disturbed energy balance on activity of VTA DA neurons. Ghrelin has been shown to directly activate VTA DA neurons and inhibition of ghrelin signalling in the VTA reduces motivation and intake of palatable food. However it remains unclear to what extent VTA DA neurons are indirectly affected by distal projections of ghrelin-responsive neurons onto these DA neurons. Potentially VTA GABAergic neurons or hypothalamic neurons from the arcuate nucleus and LH play an important role in mediating the effects of ghrelin on the mesDA system. Although there are LepR expressed on VTA neurons, it remains unclear to what extent these receptors contribute to the effects of leptin on feeding behaviour. Possibly leptin affects VTA DA neuronal activity predominantly via LepR on Nts positive neurons in the LH. In addition, leptin may affect VTA DA activity via LepR expressed on POMC and NPY/AgRP neurons projecting from the arcuate to the VTA. Further insights into the effects of a disturbance on energy balance, and the effects of leptin and ghrelin on the mesDA system will require in vivo recordings from VTA DA neurons to understand the role of this system in feeding behaviour.