Accumbens dopamine-acetylcholine balance in approach and avoidance

Dopamine

DA in the NAc can increase appetite. Early microdialysis studies found that instrumental feeding responses or appetitive hypothalamic stimulation releases DA in the NAc, and it can stay high for some time afterward [16]. DA release can carry information as to sucrose concentration and thus palatability [17]. Measures of release on a subsecond time scale show that DA is relatively high during food-seeking [18]. DA depletion with 6-hydroxy-dopamine suggests that DA is needed for instrumental work output [19]. DA antagonists briefly increase operant responding for food, which then gradually declines, suggesting a loss of reward and not simply loss of motivation or motor impairment [20,21]. Ahn & Phillips [22] use a random interval schedule of food reinforcement and suggest that reinforcement uncertainty maintains high DA levels in the medial NAc over many days of training. They interpret this microdialysis data as compatible with an incentive motivation view of DA function, as opposed to a reward-value function or response-reinforcement function. The computational approach has lead to a proposal for two ventral striatal systems, one for the primary value of a stimulus (US) via the lateral hypothalamus and another that responds to the learned value (CS) as determined in the amygdala.[23]. Di Chiara [24] recently summarized what DA does and does not do. For example, a novel palatable food stimulus releases DA in the medial NAc shell during the first meal experience, and this effect depends on mu-opioid receptor stimulation in the VTA. This DA response habituates with additional free-feeding exposures, unless the animal is food deprived. As confirmation, we find that ingestion of a sucrose solution releases DA in the medial NAc during the first exposure, and less the second time it is administered. However, under conditions of predictable daily availability, in mildly food deprived, sweet taste releases DA in the NAc, time after time [25].

Acetylcholine

ACh in the NAc has a role in the onset of satiation. As feeding slows down towards the end of a meal, we observe an increase in extracellular ACh in the NAc [26,27]. Its overall satiety effect appears to depend largely on muscarinic receptors. Although ACh can release DA by an action at nicotinic receptors, local injection of the mixed muscarinic agonist arecoline will inhibit feeding, and feeding can be reduced with the relatively specific muscarinic M1 receptor antagonist pirenzepine (P. Rada et al., unpublished data). The mixed, muscarinic agonist oxotremorine fails to inhibit bar pressing for food, but can inhibit self-administration of cocaine [28]

Seemingly contradictory findings by Kelly's group [29] suggest that ACh may have a role in promoting feeding. They find that the muscarinic antagonist scopolamine reduced lever pressing for food. Scopolamine blocks most types of muscarinic receptors, including the M-2 receptor that is both postsynaptic and presynaptic. As a presynaptic autoreceptor on the cholinergic neurons, the M-2 type receptors provide local negative feedback, and blocking them releases ACh [30], which might then suppress eating. Thus it is possible that rats stop eating with scopolamine because of the release of ACh, but more work is needed to understand the mechanisms underlying this issue, as muscarinic receptors of various types have numerous functions within NAc circuitry.

We find that appetite-suppressant drugs release ACh in the NAc, and these therapies may be applicable to drug abuse as well as feeding [31,32]. d-Fenfluramine combined with phentermine, given systemically, increases extracellular ACh in the NAc at a dose that inhibits both eating and cocaine self-administration [31,33,34]. ACh enhancement by acetylcholinesterase inhibitors blocks cue-induced heroin seeking [35] and blocks cocaine and heroin conditioned place preference, which is reversed by selective ablation of ACh neurons [36]. Endogenous appetite suppressants such as hypothalamic DA, serotonin and cholecystokinin (CCK) also act via the NAc by releasing ACh [37].

To further test the ACh-satiety theory, rats were surgically equipped with a gastric fistula so that food could be drained from the stomach. As shown in Figure 1, when they drank 10% sucrose for an hour in the normal fashion (fistula closed), DA was released, followed by ACh. When the sucrose was allowed to drain out of the stomach (fistula open), they drank more sucrose, DA was released as usual, but ACh did not rise until the fistula was closed again [27]. Thus, sham feeding greatly attenuates ACh release. To summarize, we suggest that ACh can act at M1 receptors that oppose dopamine's D2 receptor effects on the indirect GABA output pathway for behavior inhibition, and in this way ACh may contribute to satiation.

Dopamine

The decision to eat, or not to eat, is clearly based on prior experience with tastes that signal nutrition or poison. Foods for which animals have a conditioned taste preference release DA. Rats were trained to drink a novel, mildly bitter-tasting solution that automatically triggered the injection of nutritious polycose into the stomach. Compared with a control flavor paired with just water, the animals soon develop a preference for the novel flavor in a two-bottle test [38]. When that flavor was squirted into the mouth, it resulted in the release of DA in the NAc. DA was presumably involved in eliciting the approach behavior seen in the preference test [39].

Taste aversion has the opposite effect of taste preference. When the taste of saccharin is made aversive by paring it with lithium chloride-induced nausea, animals avoid drinking saccharin, the taste signal in the NTS changes to resemble that of quinine [40], and microdialysis reveals a shift from saccharin-induced DA release to a decrease below baseline [41]. This conditioned decrease in DA release could be a major factor in conditioned taste aversion.

Acetylcholine

As seen in Figure 2, when rats have a conditioned aversion to saccharin, because of pairing the taste with nausea, the taste significantly increases ACh release in the NAc [42]. Thus a sweet taste can release DA or ACh depending on the learned physiological consequences. Elevated ACh is, by itself, sufficient to cause a taste aversion. This was shown by micro-injecting neostigmine into the NAc to block degradation of ACh and thereby elevate extracellular levels in association with the taste of saccharin. A conditioned taste aversion to saccharin is generated by the local rise in ACh [43]. In summary, there is a decrease in extracellular DA and increased release of ACh in response to a taste that had been associated with nausea. From an evolutionary standpoint, this is sufficient to produce survival-necessitated avoidance behavior such that the animal stops, and then goes to find something else to eat.

Dopamine

Intracranial self-stimulation (ICSS) with lateral hypothalamic electrodes releases DA in the NAc. The hypothalamus contains motivational systems for several behaviors, including feeding [44] and mating [45]. Lateral hypothalamic outputs such as orexin cell groups may be coded for feeding, with leptin receptors [46] on some orexin cells and others for mating with inputs that are sensitive to estrogen or androgen [45]. This suggests the possibility of behaviorally-coded outputs. An electrode site that elicits feeding can, depending on stimuli available in the environment, be made to elicit other behaviors, or it may even elicit oral signs of aversion [47]. Indeed, some electrodes support both ICSS and stimulation-escape, with the animal's choice of two levers shifting from approach to escape as a function of increased satiety, increasing body weight, insulin-induced satiety, or appetite-suppressant drugs [48,49]. Similarly, ICSS at a mating site looses it effectiveness after castration, and ICSS is restored by androgen-replacement therapy [50]. After a male ejaculates, the preponderance of behavior shifts from ICSS to stimulation-escape [49]. The next question is what happens in the NAc when stimulation becomes aversive.

Acetylcholine

Aversive hypothalamic stimulation increases the level of extracellular ACh in the NAc. Figure 3 shows ACh levels in the NAc in rats pretested using one lever and for ICSS and another for responding to escape from recurrent stimulation for 5 s. In the animals with electrode sites where stimulation released ACh, ICSS was relatively slow. If the recurrent stimulation was then turned on automatically, causing a relatively high extracellular level, this increased the animal's lever pressing to escape the ACh-mediated state [51]. As a result, escape behavior effectively lowered ACh (Figure 3). This implies that high ACh is a negative reinforcer (Table 1).

Dopamine

Given the role of DA in motivation, one might expect a depressed animal to display decreased DA. In the forced-swim test, a rat is tested for escape efforts in a water tank two days in a row. They readily give up trying to escape on the second day, as shown by less active swimming behavior and more passive floating. Extracellular DA in the NAc decreases during the swim test when animals become behaviorally depressed [52].

Acetylcholine

We were surprised to find that ACh does not change during passive floating. Instead, it decreases briefly after the animal is removed from the water the first day and then increases, with the interesting discovery that basal ACh is still high before going back in the water the second day when behavior is depressed [53]. Thus, ACh levels reflect a lasting state correlated with behavioral depression.

We tested the cholinergic receptor types, nicotinic and muscarinic, that might be responsible for this depression-related effect. Nicotinic receptors are unlikely to be involved, because we have shown that local nicotine releases DA [54]. Turning to the muscarinic receptors, local bilateral injection of a muscarinic agonist arecoline depresses swimming behavior, and the relatively specific M-1 antagonist pirenzepine increases escape efforts (Figure 4) [30]. This is not a locomotor effect because there was no hyperactivity in an open field. Thus the M-1 type receptor is a candidate for accumbens-induced depression.

Antidepressants are well known to restore escape swimming [55]. We find that fluoxetine infused directly into the NAc by reverse dialysis causes an immediate increase in escape efforts [56]. To achieve the same effect with systemic administration requires two weeks of treatment. If the accumbens ACh-depression hypothesis is correct, then an antidepressant drug should lower ACh. It does. When fluoxetine is infused by reverse-microdialysis into the NAc, it lowers extracellular ACh [56]. A serotonin-1A receptor agonist 8-0H-DPAT, given locally has the same effect [57]; thus identifying the type and location of one of the relevant receptors for antidepressant therapy.

In summary, we suggest that behavioral depression, as modeled by the forced swim test, is associated with low DA and a protracted rise in ACh. The rapid onset of immobility on the second day of the test is mediated in part by ACh acting at postsynaptic M1 receptors. Local fluoxetine raises extracellular serotonin, which then acts at serotonin-1A receptors to lower ACh, thereby reducing M1 stimulation and alleviating behavioral depression.

Dopamine

Morphine, phencyclidine, amphetamine and nicotine are examples of drugs that can increase extracellular DA, even when infused locally in the NAc [58–61]. Numerous explanations for addiction to opioid and DA receptor stimulation have been offered. Examples include intracellular changes involving gene expression in the NAc [62], habit learning [63], loss of cortical executive control [64], abstinence-induced incubation of the reinstatement of responding for drug- related cues [65], and a cycle of unpleasant withdrawal effects, including conditioned withdrawal, that leads to self-medication with the drug in an effort to regain normalcy [66]. These processes are not mutually exclusive, and several may involve DA in one site or another.

Acetylcholine

Withdrawal signs can be elicited in rats by administering the appropriate receptor antagonist after prolonged drug use. For example, rats received injections of morphine daily for a week, causing DA release each time [60,61]. Then they received naloxone to block opioid receptors, which caused DA to decrease and ACh to be released instead. Koob and colleagues [66,67] have argued that withdrawal is an aversive state. They show that the threshold for brain self-stimulation reward rises during withdrawal from addictive drugs. If this state is due in part to accumbens ACh, then naloxone given systemically, or methyl-naloxonium infused locally in the accumbens, should cause the release of ACh in morphine-dependent rats. This is in fact the case. An opioid antagonist injected into the NAc causes a conditioned place aversion [68] and lowers extracellular DA while releasing ACh [60,69]. It is unexpected that rats will self-inject the cholinergic agonist carbachol into the NAc [70]. However, we find that carbachol decreases ACh release by stimulating M2-muscarinic autoreceptors, and it also releases DA, which may help explain self-injection [71].

In summary, most addictive drugs activate accumbens dopaminergic functions that foster approach. In cases of pharmacologically induced withdrawal, DA decreases and ACh is released. It is as if ACh were disinhibited, revealing its function as an opponent process. Based on the results cited above with conditioned taste aversion, behavioral depression and brain stimulation-escape behavior, we conclude that high ACh during drug withdrawal, especially when DA is low, generates an aversive state. This DA/ACh imbalance may be an important aspect of drug addiction. Theoretically, any type of behavior could be addictive if it repeatedly releases surges of DA and leads to withdrawal-induced loss of dopaminergic tone and increased extracellular ACh.

Dopamine

Our data show, in essence, that bingeing on sugar can be, in many ways, like taking a low dose of an addictive drug. Feeding behavior and cocaine both increase extracellular DA in the NAc [72]. The effects of cocaine and heroin on extracellular DA are more potent than those of food. Cocaine acts everywhere there are DA reuptake sites and does so every time the drug is administered. Heroin acts in many sites, from the brain to the gut. Psychostimulants and opiates cause local homeostatic compensations in terms of receptor trafficking and synthesis of neurotransmitters and receptors. Food would not be expected to produce addictive effects of the magnitude caused by drugs. Perhaps this is one reason the DA system that projects to the NAc evolved to stop firing when the food taste is no longer novel or salient. Recordings from DA cells in the VTA of monkeys, suggests that even the conditioned response to DA wanes with repetition [73,74]. Moreover eating has postingestional satiation effects that gradually raise ACh levels to counteract DA [26,27]. Thus under most mealtime circumstances we are well protected from developing an “addictive” state with regard to food. However, alternating feasting with famine is another story.

A surge of DA can be released in the NAc every time the animal binges if trained with alternate food deprivation and access to chow and sugar [25]. A control group with ad libitum access to 10% sucrose solution does not show the effect, nor does a group that has sugar to drink only twice, on Day 1 and again on Day 21. However daily 12-h food deprivation followed by excessive sucrose intake causes a rise in DA, even on Day 21. This effect in the NAc is analogous to receiving a low dose of an addictive drug every day for 21 days. Signs of dependency also appear. In brief, they fall into three behavioral categories, First “excessive intake”, which refers to an unusually large meal in the form of a “binge, ” that is larger than an animal would take when feeding ad libitum [25], second, “withdrawal, ” which occurs in response to fasting, or when naloxone is administered, [75], and third, “craving, ” which, for lack of a better word, is defined as long-lasting signs of enhanced motivation seen during abstinence. These protracted effects seen after more than a week of sugar abstinence include: first, “cross-sensitization, ” in which the animals are hyperactive in their exploration of an open-field when given a low dose of amphetamine or cocaine [75,77], second, the “deprivation effect, ” in which the animals engage in more lever pressing for sugar than ever before [78], third the “gateway effect”, shown as an increased intake of alcohol compared with control rats [79], and (d) the “incubation effect, ” which refers to the growing motivation to lever press for sugar- associated cues during abstinence [80].

There are concomitant neurochemical changes in sugar-bingeing rats. An enkephalin system in the NAc shows decreases in enkephalin mRNA [81] as Kelley's group observed using intermittent access to a palatable sweet-fat drink (Ensure) [82]. With sugar bingeing, we find that there is also an increase in mu-opioid receptor binding [83]. The DA system adapts with increased D1 and decreased D2 receptor binding in the striatum relative to chow-fed controls [82]. Drug-dependent rats show many of these same effects, and in human drug addicts and people who are obese, brain imaging reveals a decrease in D2 receptor binding, which is interpreted as dopaminergic involvement in compulsive drug or food intake [84].

Acetylcholine

Naloxone or 36-h food deprivation can cause opiate-like withdrawal signs in rats trained to binge on sugar. These signs includes first, teeth-chattering and anxiety as measured by avoidance of the open arms of an elevated plus-maze [75], second, behavioral depression in the swim test (MN Avena et al., data unpublished), and in terms of neurochemistry, (c) the predicted decrease in DA and increase in extracellular ACh in the NAc [75]. This is much like naloxone-precipitated withdrawal from morphine.

In summary, repeated daily bingeing on a sugar solution can cause “surges” of DA in the NAc that causes adaptations in both DA and opioid systems.[85] These changes contribute to opiate-like withdrawal signs, which include behavioral manifestations of increased ACh release combined with relatively low extracellular DA. As explained above, we assume that this causes an aversive state involving anxiety and a tendency to behavioral depression. Thus, sugar, when it is taken in a feast- and-famine fashion, may promote a feedback cycle similar to an addiction, with repetitions of high extracellular DA followed by low DA and relatively high ACh. This opioid-related high DA, followed by a deprivation-induced, cholinergic-mediated avoidance state, may be advantageous when food is scarce. It can be troublesome in times of plenty.

It is an interesting question whether social and scientific norms will be promulgated such that the term “addiction” is applied to natural behaviors. The popular press already has many books and articles about “natural addiction” to a variety of excessive behaviors. The difference between strong motivation and well-defined addiction has yet to be determined. The evidence outlined above suggests that, under certain circumstances, bingeing on a sugar solution can produce effects that are at least analogous, perhaps homologous, to the changes caused by addictive drugs.