Localization of brain reinforcement mechanisms: intracranial self-administration and intracranial place-conditioning studies

Transcription

1 Behavioural Brain Research 101 (1999) Review article Localization of brain reinforcement mechanisms: intracranial self-administration and intracranial place-conditioning studies William J. McBride a, *, James M. Murphy a,b, Satoshi Ikemoto a,1 a Department of Psychiatry, Institute of Psychiatric Research, 791 Union Dr., Indiana Uni ersity School of Medicine, Indianapolis, IN , USA b Department of Psychology, Purdue School of Science, Indianapolis, IN , USA Received 27 January 1999; accepted 27 January 1999 Abstract Intracranial self-administration (ICSA) and intracranial place conditioning (ICPC) methodologies have been mainly used to study drug reward mechanisms, but they have also been applied toward examining brain reward mechanisms. ICSA studies in rodents have established that the ventral tegmental area (VTA) is a site supporting morphine and ethanol reinforcement. ICPC studies confirmed that injection of morphine into the VTA produces conditioned place preference (CPP). Further confirmation that activation of opioid receptors within the VTA is reinforcing comes from the findings that the endogenous opioid peptide met-enkephalin injected into the VTA produces CPP, and that the mu- and delta-opioid agonists, DAMGO and DPDPE, are self-infused into the VTA. Activation of the VTA dopamine (DA) system may produce reinforcing effects in general because (a) neurotensin is self-administered into the VTA, and injection of neurotensin into the VTA produces CPP and enhances DA release in the nucleus accumbens (NAC), and (b) GABA A antagonists are self-administered into the anterior VTA and injections of GABA A antagonists into the anterior VTA enhance DA release in the NAC. The NAC also appears to have a major role in brain reward mechanisms, whereas most data from ICSA and ICPC studies do not support an involvement of the caudate-putamen in reinforcement processes. Rodents will self-infuse a variety of drugs of abuse (e.g. amphetamine, morphine, phencyclidine and cocaine) into the NAC, and this occurs primarily in the shell region. ICPC studies also indicate that injection of amphetamine into the shell portion of the NAC produces CPP. Activation of the DA system within the shell subregion of the NAC appears to play a key role in brain reward mechanisms. Rats will ICSA the DA uptake blocker, nomifensine, into the NAC shell; co-infusion with ad 2 antagonist can block this behavior. In addition, rats will self-administer a mixture of a D 1 plus a D 2 agonist into the shell, but not the core, region of the NAC. The ICSA of this mixture can be blocked with the co-infusion of either a D 1 orad 2 antagonist. However, the interactions of other transmitter systems within the NAC may also play key roles because NMDA antagonists and the muscarinic agonist carbachol are self-infused into the NAC. The medial prefrontal (MPF) cortex supports the ICSA of cocaine and phencyclidine. The DA system also seems to play a role in this behavior since cocaine self-infusion into the MPF cortex can be blocked by co-infusing a D 2 antagonist, or with 6-OHDA lesions of the MPF cortex. Limited studies have been conducted on other CNS regions to elucidate their role in brain and drug reward mechanisms using ICSA or ICPC procedures. Among these regions, ICPC findings suggest that cocaine and amphetamine are rewarding in the rostral ventral pallidum (VP); ICSA and ICPC studies indicate that morphine is rewarding in the dorsal hippocampus, central gray and lateral hypothalamus. Finally, substance P mediated systems within the caudal VP (nucleus basalis magnocellularis) and serotonin systems of the dorsal and median raphe nuclei may also be important anatomical components involved in brain reward mechanisms. Overall, the ICSA and ICPC studies indicate that there are a number of receptors, neuronal pathways, and discrete CNS sites involved in brain reward mechanisms Elsevier Science B.V. All rights reserved. * Corresponding author. Tel.: ; fax: Present address: Behavioral Neuroscience Branch, Intramural Research Program, NIDA, 5500 Nathan Shock Dr., Baltimore, MD 21224, USA /99/$ - see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S (99)

2 130 W.J. McBride et al. / Beha ioural Brain Research 101 (1999) Keywords: Intracranial self-administration; Intracranial place conditioning; Place conditioning; Brain reward; Reinforcement; Opioids; Psychostimulants; Morphine; Amphetamine; Cocaine; Alcohol 1. Introduction Intracranial microinjection procedures offer the possibility of localizing brain systems mediating drug reward, as well as localizing and elucidating brain reward mechanisms in general. The idea that there could be distinct pathways mediating reinforcement developed from findings that rats would maintain lever-pressing for the delivery of an electrical current to specific CNS regions [104]. Unfortunately, the intracranial electrical self-stimulation technique does not allow for the direct analysis of neurotransmitter systems and receptors involved in the behavior. The intravenous self-administration technique permits an evaluation of the relative reinforcing effects of a wide variety of compounds. However, compounds self-administered systemically can have multiple actions and act at several CNS sites, not all of which are involved in its rewarding effects. There have been two procedures developed which allow the localization of drug reward mechanisms (see [11,53,112,126,139] for review). One procedure is the drug self-administration paradigm which requires the animal to earn the drug by attaining some predetermined response requirement to receive the drug injection. The characteristic of this paradigm is that the reward is given in a response-contingent manner. In most cases, the animal is required to lever-press to obtain an intracranial injection, whereas, in other cases, the animal may need to nose-poke or enter one of two arms in a Y maze to obtain the injection. In the second procedure, the intracranial administration of the drug is independent of the behavior of the animal and is given in association with a specific environmental stimulus. The learning in this paradigm can be considered to be stimulus learning rather than response learning. The animal learns about the relation between the drug stimulus and environmental stimuli. In this case, the experimenter administers the drug, whereas in the first case, the animal controls the drug administration. Thus, the intracranial selfadministration (ICSA) paradigm is confirmatory to the operant paradigm [124], whereas the intracranial place conditioning (ICPC) paradigm does not quite fit into either the operant [124] or the Pavlovian [108] paradigm. Although the place conditioning paradigm is claimed to assess the rewarding capacity of stimuli by reflecting approach responses and maintenance of contact [26], the validity of the place conditioning paradigm remains ambiguous (for advantages and disadvantages of the place conditioning paradigm see [67,120]). Therefore, an obvious issue is whether or not rewarding effects assessed by the two paradigms are reflected by the same brain mechanisms. This review will focus on the ICSA studies which have been conducted in rodents and attempt to integrate these findings with each other and with results from ICPC studies into a reasonably cohesive picture of brain reward mechanisms Methodological considerations Intracranial injection procedures are highly problematic and careful attention must be given to overcome the many obstacles. For one, the injected solutions may cause local non-specific damage which could alter brain function by actions unrelated to the normal pharmacological action of the drug. Consequently, careful pharmacological characterization needs to be carried out. In addition, the number of sessions that can be conducted before site viability is lost is limited and needs to be determined for each agent. Secondly, intracranially injected drugs can diffuse from the injection site and act at more distal locations. Therefore, small injection volumes and careful neuroanatomical control experiments need to be conducted to minimize diffusion and establish the site of action of the injected drug. Third, for the ICSA procedure, the agent must be accurately delivered to discrete sites with minimal delay to initiate and maintain self-administration behavior. In addition, controls should be built into the experimental designs to evaluate non-specific motor effects from goal-directed motivated responding. Typical pharmacological controls would be obtaining dose-response effects, antagonizing the action of the drug (with co-infusion of an antagonist), mimicking the action of the drug with another agent known to have similar properties, testing an inactive isomer of the drug, and examining the effects of vehicle alone. Ideal anatomical controls would be to do injections of the agent at sites dorsal, ventral, anterior and posterior to the target region. In addition, injections of the radioactive form of the drug, if available, should be carried out to measure the spread of radioactivity from the injection site. To reduce the spread of agents small injection volumes, e.g. 100 nl or less, could be used. Accurate delivery of the agent is critical for ICSA experiments. Self-administration may be erratic or not

3 W.J. McBride et al. / Beha ioural Brain Research 101 (1999) occur at all if changes in drug flow occur during an ICSA session. An ideal system would be one that rapidly delivers a reproducible small volume, with moderate pressure to reduce pressure-induced tissue damage. In ICSA experiments, injection volumes of 100 nl or less are typically employed. The electrolytic microinfusion transducer (EMIT) system [13,32] circumvents many of the problems inherent in small volume drug delivery systems. The EMIT system is a self-contained unit that includes a gas-tight drug reservoir, which mounts directly onto a guide cannula aimed at the region of interest. Microinfusions are produced by passing a direct current between the anode and cathode in the drug reservoir. The current flow results in the evolution of hydrogen gas which increases the pressure within the gas-tight reservoir, thereby forcing the drug solution out through the injection cannula. The volume of solution infused is directly proportional to the amount of hydrogen gas produced, which is related to the current intensity and duration [32]. A flexible spring-covered lead connects the reservoir to a counter-balanced electrical commutator permitting relatively unrestrained movement of the animal during testing, while maintaining electrical contact with the constant current generator (EMIT system). In addition to the above pharmacological and neuroanatomical control experiments, behavioral control data need to be generated which clearly establish that the animal is responding for the reinforcing value of the agent and that responding is not a result of general motor activation. Ideally, such experiments would have to establish that the animal (a) can discriminate the active infusion lever (or arm or nose-poke hole) from the inactive lever; (b) demonstrates extinction of responding on the active lever when vehicle is substituted for the agent; (c) will maintain reinforcements with increased response requirements; and (d) demonstrates bar reversal when the roles of the two levers are switched. The inherent technical difficulties of the ICSA technique itself using the EMIT system (i.e. chronic stereotaxic cannula implant, inserting injector and reservoir into the guide without damaging it, finding conditions for stabilizing currents which differs for each agent, ensuring a gas tight fit of the electrodes onto the reservoir, ensuring proper electrical contact between the electrodes and the constant current generator, careful attention to preventing obstruction of the injector during insertion into the tissue, establishing appropriate operant conditions, use of expensive custom made constant current generators, etc.) coupled with the large number of pharmacological, anatomical and behavioral controls has prevented the wide-scale use of this technique. However, the few laboratories that have been able to successfully apply the ICSA technique have produced critical data toward understanding drug reward mechanisms in particular and brain reward mechanisms in general. The ICPC technique suffers from many of the same neuroanatomical and pharmacological requirements as the ICSA procedure. Moreover, because injection volumes greater than 100 nl are typically employed with the ICPC technique, careful attention must be especially given to the neuroanatomical control experiments. With the ICPC method, the experimenter administers the agent, which eliminates the need for conducting the behavioral control experiments associated with operant responding, and circumvents the technical problems inherent with the use of the EMIT system. In addition, the ICPC procedure can be used to demonstrate the rewarding as well as the aversive properties of injected compounds. However, major weaknesses of the ICPC procedure are that the animal has no control over the administration of the agent, and the learning principles that explain place conditioning are not well understood [67]. 2. Ventral tegmental area as a major site for morphine and alcohol reinforcement, and for brain reward in general The dopamine system arising from the ventral tegmental area (VTA) has received considerable attention as a major neurobiological substrate involved in mediating the reinforcing actions of drugs of abuse and in the brain reward system. This region has been the most extensively studied using ICSA and ICPC techniques to understand neural mechanisms involved in initiating and maintaining morphine reinforcement. Thus far, the VTA is the only region that has been shown to support direct alcohol self-administration. In addition, data will be reviewed suggesting that the VTA is not a functionally homogeneous anatomical unit Morphine and opioid receptors Reward relevant CNS injection sites for the opiates and opioid peptides have been the most extensively explored. The studies of Broekkamp et al. [17] gave one of the first ideas where opiates might be having their rewarding actions. These investigators [17] reported that VTA morphine injections at low doses were effective at facilitating electrical self-stimulation of the medial forebrain bundle and that this effect occurred with a short latency. Subsequently, a number of ICSA studies supported these initial findings and observed that rodents will self-administer morphine or mu-agonists directly into the VTA. In general, ICPC results are in accord with the ICSA findings.

4 132 W.J. McBride et al. / Beha ioural Brain Research 101 (1999) Intracranial self-administration Bozarth and Wise [14] were the first to demonstrate that rats would self-administer morphine directly into the VTA; in a preliminary study, van Ree and de Wied [128] reported that rats will self-infuse fentanyl into the VTA. Bozarth and Wise [14] designed their study to determine if experimentally naive (no prior operant training) rats would learn to press a lever for the direct infusion of morphine into the VTA. For these studies, a single lever was employed and light cues were used. The electrolytic microinjection transducer (EMIT) system was used to inject 100 ng (150 pmol) morphine in a volume of 100 nl into the VTA. This study demonstrated that male, Long Evans rats readily learned to respond for the self-infusion of 150 pmol/100 nl morphine into the VTA on a continuous reinforcement schedule. The highest response rates occurred during the 1st hour of the 4-h session. Animals had much lower response rates for vehicle and there was no difference between the response rates of the yoked control (for general motor activation) and the vehicle control groups. Moreover, an i.p. injection of naloxone, 1 h into the 4-h session, reduced responding for morphine to vehicle control levels. Overall, these results suggest that morphine is reinforcing in the VTA of Long Evans rats. However, although this landmark study provided valuable information that rats readily learned to self-infuse morphine into the VTA, it also had a number of weaknesses, i.e. no dose-response experiments were done, there were insufficient anatomical controls, and there was no second lever to properly control for a general increase in locomotor activity. Welzl et al. [132] designed a micropipette system consisting of a glass micropipette (with a tip diameter of 5 m) connected by flexible tubing and a swivel to a pressure source, which had the potential of delivering injection volumes of nl. Testing was conducted with stereotaxic placements generally in the anterior and middle (along the anterior-posterior axis) VTA. Nose-poking through a hole in one wall of the chamber was the operant task. A light was presented for each nose-poke. Rats readily learned to nose-poke for 75 pmol/5 nl morphine into the VTA on both a continuous reinforcement (CRF) and fixed-ratio 2 (FR2) schedule of reinforcement. Response rates for the morphine groups were significantly higher than for the saline control group or the yoked control groups. The strengths of this study were better anatomical controls, smaller injection volumes to reduce diffusion and the use of an FR2 response requirement. In addition, these investigators demonstrated extinction of responding on the second session after substituting vehicle for morphine in the two morphine groups. Data that the VTA was a site supporting the reward-relevant actions of morphine was provided by two additional self-administration studies. Self and Stein [122] used an experimental chamber containing two nose-poke holes, with the active hole indicated by a white cue light above it. A nose-poke response at the active hole resulted in the infusion of 300 pmol morphine in 100 nl over 5 s; a tone was sounded concurrent with the infusion. The infusion period was followed by a 30-s time-out during which the cue light was extinguished and responses at the active hole had no programmed consequences. Most of the injection sites were in the middle to anterior portions of the VTA. With this procedure, rats reliably responded more for morphine than saline, and responding decreased when the saline solution was substituted for morphine. Although this study supported the hypothesis that morphine is rewarding in the VTA, it suffered from several weaknesses, including the fact that no dose-response experiments were conducted, no information was given on discrimination of the active from the inactive hole, and no antagonist effects were reported. The study of Devine and Wise [38] examined the ICSA of morphine, the mu-opioid agonist DAMGO, and the delta-opioid agonist DPDPE into the VTA of Long Evans rats, using the EMIT system. Rats were tested in 4-h sessions on each of 8 consecutive days in single-lever operant chambers. A white cue light was situated above the lever. Rats were naive at the start of testing. The first five sessions constituted the acquisition phase, the sixth session was the extinction session (vehicle substituted for drug solution), and the next two sessions were the reinstatement phase (drug solution replaces the vehicle). Rats were assigned to groups which received either vehicle or one of the opioids (300 pmol/120 nl morphine; 3 pmol/120 nl DAMGO; or 300 pmol/120 nl DPDPE) over a 5-s period. Placements appeared to be mainly in the anterior and middle portions of the VTA. Using a continuous reinforcement schedule, rats responded significantly more for the morphine solution than for vehicle during acquisition sessions 2 5. Responding was greater in the first 2 h than in the last 2 h of the acquisition sessions. Substituting vehicle for morphine in session 6 reduced responding; higher lever presses were reinstated when morphine was reintroduced. Similar results were obtained for the single doses of DAMGO and DPDPE. This study demonstrated that rats would lever press more for morphine, DAMGO or DPDPE than for vehicle, extinguish responding when vehicle was substituted for the opioid agents, and reinstate responding with the return of the opioid agents, suggesting that morphine acting at mu- and/or delta-opioid receptors is reinforcing in the VTA. Although the data with the two selective agonists suggest that the effects of morphine may be mediated

5 W.J. McBride et al. / Beha ioural Brain Research 101 (1999) through mu-opioid receptors, a more thorough dose-response effect coupled with the use of selective antagonists for the mu- and delta-opioid receptors would be required to resolve this issue. There have also been two studies with mice using a self-administration procedure, the results of which support the hypothesis that morphine is reinforcing in the VTA. In the first study [33], BALB/c mice were unilaterally implanted with guide cannulas aimed at the VTA. On each experimental day, an injection cannula is inserted into the guide and is connected by flexible polyethylene tubing to the micro-injection system. By interrupting one photocell beam in a Y -maze, mice could receive an injection of morphine sulphate (either 6.5 or 65 pmol in 50 nl over 4 s). Entrance into the other arm had no programmed consequences. Each session consisted of ten trials, and mice were confined to the chosen arm for 10 s. Trial intervals were 1 min in duration. A separate group of mice received vehicle alone. For both the 6.5- and 65-pmol doses, mice readily learned to discriminate the active from the inactive arm, whereas with vehicle alone there was no discrimination. Systemic administration of naloxone, 10 min before the sessions, eventually reduced morphine reinforcements by the fifth treatment session. This experimental approach used a combination of conditioned place preference with self-administration behavior, i.e. the mouse enters the compartment associated with the infusion of morphine. However, the mouse does not have control over the number and pattern of reinforcements. In addition, it is difficult with this technique to obtain a dose-response effect, determine response rates, increase the response requirements, evaluate extinction behavior, or examine patterns of drug self-administration behavior. In a second study, David and Cazala [35] implanted BALB/c mice with two guide cannulas, aimed at either the amygdala and VTA in one group, or the amygdala and 2.3 mm above the VTA in a second group. Following four training sessions, mice were tested in subsequent sessions in which entrance in one arm resulted in 6.5 or 65 pmol morphine in 50 nl being injected into the VTA (or dorsal to the VTA) whereas entrance into the other arm produced an injection into the amygdala. At the lowest morphine dose, the mice readily distinguished between the two injection sites, and showed a marked preference for the VTA. Moreover, when the infusion sites associated with arm entries were reversed, the mice rapidly differentiated the arm associated with the infusion of 6.5 pmol morphine into the VTA. At the higher dose of morphine, the mice did not discriminate between the amygdala and the VTA. On the other hand, mice did show preference for the low dose morphine infusions into the amygdala compared to sites dorsal to the VTA. Overall, the results suggest that the VTA may be more sensitive than the amygdala to the reinforcing effects of morphine Intracranial place conditioning It has been demonstrated in four separate laboratories that injection of morphine into the VTA produces conditioned place preference. Phillips and LePiane [113] demonstrated that infusion of 0.2 g (260 pmol) or 1.0 g (1300 pmol) of morphine in 500 nl produced conditioned place preference in male Wistar rats. Neither injection of saline alone into the VTA nor of morphine into sites 2.5 mm dorsal to the VTA produced place preference. Pretreatment with systemic administration of naloxone antagonized the morphine induced place preference. Bozarth [12] in an elegant study used the place conditioning paradigm to map the neuroanatomical boundaries of the VTA mediating morphine reinforcement. For this study, morphine (750 pmol/500 nl) was infused unilaterally into different sites along the anterior-posterior axis of the VTA of male, Long Evans rats. Injection sites along the entire anterior-posterior axis of the VTA produced conditioned place preference, whereas sites approximately 0.6 mm beyond the VTA in either the rostral or caudal direction did not produce conditioned place preference. These results suggest that activating opioid receptors throughout the VTA along its anterior-posterior axis is rewarding. Jaeger and van der Kooy [76] established that there are separate neural substrates mediating the motivating and discriminative stimulus actions of morphine. Infusion of morphine into the parabrachial nucleus, but not into the VTA, served as a stimulus for the acquisition of discrimination learning, whereas infusion into the VTA, but not into the parabrachial nucleus, produced conditioned place preference. Unfortunately, in these studies a relatively high dose of morphine (approximately 3 nmol) was used to produce the conditioned place preference. Control experiments conducted with the inactive isomer of morphine did not, however, produce place preference. In a fourth study, Olmstead and Franklin [106] reported that unilateral injection of 1.5 nmol morphine into the VTA produced conditioned place preference, whereas injections 1 mm dorsal to the VTA did not. Overall, the results from the conditioned place preference experiments are consistent with the idea that activating opioid receptors within the VTA is rewarding. In addition to morphine, Phillips and LePiane [114] demonstrated that bilateral injections of 100 or 250 ng (approximately pmol) (D-ala 2 )-met 5 -enkephalinamide into the VTA of male Wistar rats produced conditioned place preference, whereas injections dorsal to the VTA did not. Systemic injection of the opioid antagonist naloxone blocked the local effects of the agonist in the VTA. Bals-Kubik et al. [4] showed that unilateral infusion of the mu-opioid agonist DAMGO at doses of 0.05 or 0.1 g/1.0 l (approximately pmol) into the VTA produced conditioned place prefer-

6 134 W.J. McBride et al. / Beha ioural Brain Research 101 (1999) ence, whereas infusion of the kappa-opioid agonist U50,488H (at doses of 0.3 and 1.0 g, but not 3.3 g) produced conditioned place aversion. These results suggest that activating mu-opioid receptors within the VTA is reinforcing but that activating kappa-opioid receptors may affect different neuronal ensembles [110] to produce place aversion Alcohol Only limited studies have been conducted to examine neuroanatomical substrates mediating alcohol reinforcement [43,119]. The rationale for initially examining the ICSA of alcohol into the VTA was based upon the in vivo [46] and in vitro [15] electrophysiological findings that ethanol could increase the firing rates of VTA dopamine neurons in the rat, and microdialysis studies indicating that alcohol could stimulate the release of dopamine in the nucleus accumbens (NAC) [39]. Because increased VTA dopamine neuronal activity is associated with reinforcement processes [ ], and alcohol can activate this system, it was reasoned that the VTA may be a prime candidate as a neuroanatomical substrate mediating the reinforcing effects of ethanol. Gatto et al. [43], using unilateral placements in the middle-posterior regions of the VTA, observed that rats of the selectively bred alcohol-preferring line (P rats) self-administered mg% ethanol directly into the VTA. These P rats readily discriminated the active from the inactive lever, showed extinction of responding when vehicle was substituted for ethanol, and reinstated responding on the active lever when ethanol was restored. These data suggested that ethanol is rewarding in the VTA of P rats. Non-specific membrane effects are not likely producing this behavior because dose-related self-administration was obtained at concentrations that are pharmacologically relevant and within the range of blood and brain alcohol concentrations achieved by P rats, orally or intragastrically, self-administering alcohol [100,130]. Sites dorsal to the VTA did not support ICSA by the P rat. In contrast, the selectively bred alcohol non-preferring NP line of rats did not self-infuse any of the ethanol concentrations. The results of this study not only indicated that the VTA may be one site initiating and maintaining ethanol reinforcement but it also indicated that genetic factors could influence this response. However, the preliminary study of Rodd et al. [119] indicated that stock Wistar rats would also self-administer mg% ethanol directly into the VTA. Ethanol was self-administered in the posterior VTA (sites posterior to 5.2 mm bregma, according to [109]), and subjects showed lever discrimination, demonstrated extinction when vehicle was substituted for ethanol, and reinstated responding on the active lever when alcohol was restored. These results with the stock Wistar rats add support to the hypothesis that the VTA is a site mediating ethanol reinforcement, and further suggest that genetic factors associated with selective breeding for low alcohol consumption influence the self-administration of ethanol into the VTA. There is insufficient data to determine whether there are differences in the reinforcing properties of ethanol in the VTA of the alcohol-preferring P line compared to the stock Wistar rats. The genetic factors which determine whether or not ethanol is reinforcing within the VTA are unknown. Moreover, the mechanisms underlying ethanol self-infusion into the VTA are also unknown. One possible mechanism underlying the ICSA of ethanol into the VTA may involve the serotonin-type 3 (5-HT 3 ) receptor. There is evidence that ethanol can potentiate the depolarizing effects of 5-HT at the 5-HT 3 receptor [90]. Campbell et al. [20] reported that local perfusion of a 5-HT 3 agonist through a microdialysis probe increased the somatodendritic release of dopamine in the VTA, suggesting an enhancement of dopamine neuronal firing rate following activation of local 5-HT 3 receptors. Furthermore, local perfusion of a 5-HT 3 antagonist through the microdialysis probe prevented the ethanolstimulated somatodendritic release of DA, suggesting that the effects of ethanol on DA neuronal activity are being mediated in part by 5-HT 3 receptors within the VTA. However, the involvement of GABA A [57] and NMDA [131] receptors in mediating the actions of alcohol within the VTA should also be considered GABA A and excitatory amino acid receptors The activity of the VTA DA neurons is governed by a balance between inhibitory and excitatory inputs, with a significant portion of these inputs being mediated through GABA A and excitatory amino acid (EAA) receptors (reviewed in [78]). A decrease in inhibition and/or an increase in excitation would enhance the activity of VTA DA neurons, which, in turn, may produce reinforcing effects. Therefore, if the reinforcing action of intra-vta morphine is to inhibit GABA interneurons, thereby disinhibiting DA neurons, then local administration of GABA receptor antagonists should block tonic GABA-mediated inhibition and enhance the activity of VTA DA neurons. Furthermore, agents known to increase directly the activity of VTA DA neurons should also produce reinforcing effects when administered into the VTA. The ICSA of GABA A antagonists was studied in rats [72] and mice [36], whereas only the GABA A agonist has been evaluated in rats [73]. Using unilateral infusions into the VTA, Ikemoto et al. [72] demonstrated that Wistar rats readily self-administered the GABA A antagonist picrotoxin into the anterior VTA, discriminated the infusion lever from the inactive lever, extin-

7 W.J. McBride et al. / Beha ioural Brain Research 101 (1999) guished responding when vehicle was substituted, and resumed responding upon reinstatement of the picrotoxin. In addition, increasing the response requirements from FR1 to FR4 enhanced responding on the infusion lever but not the inactive lever; increasing the response requirements did not alter the number of reinforcements given. Furthermore, the results of this study demonstrated that picrotoxin was self-infused into the anterior VTA (anterior to 5.2 mm bregma, according to [109]) but not into the posterior VTA (at and posterior to 5.6 mm bregma), the substantia nigra, or sites dorsal to the VTA. Co-infusion of an equimolar concentration of muscimol, a GABA A agonist, significantly reduced the ICSA of picrotoxin into the anterior VTA; replacing picrotoxin with another GABA A antagonist, bicucculline methiodide, also supported ICSA behavior in the anterior VTA [72]. These results suggest that blocking tonic GABA A mediated inhibition within the anterior VTA is reinforcing, and that inhibiting GABA A receptors results in an enhancement of VTA DA neuronal activity. To assess this possibility, Ikemoto et al. [71], using in vivo microdialysis, examined the effects of microinjecting GABA A antagonists into the anterior VTA on DA release in the nucleus accumbens (NAC). The results of this study established that microinjection of either picrotoxin or bicuculline into the anterior VTA increased the release of DA in the NAC, and thus supported the hypothesis that disinhibiting anterior VTA DA neurons could be reinforcing. In another ICSA study, Ikemoto et al. [73] examined whether there were regional differences within the rat VTA for muscimol self-infusions. In this study, muscimol was self-infused into the posterior VTA ( 6.3 to 6.8 mm bregma, according to [109]), but not into sites anterior to 5.3 mm bregma. Co-infusion of picrotoxin with muscimol into the posterior VTA reduced the number of infusions; in addition, increasing the response requirements from FR1 to FR3 enhanced responding and maintained infusion levels. Although the data suggest that activating GABA A receptors in the posterior VTA may be reinforcing, there was no evidence for lever discrimination nor were extinction experiments conducted. However, the findings that picrotoxin enhanced responding in the anterior VTA, but not in the posterior VTA, whereas muscimol enhanced responding in the posterior VTA, but not in the anterior VTA, suggest that different GABA A mediated circuits may be operating between the anterior and posterior VTA. In the anterior VTA, DA neurons may be under tonic GABA A mediated inhibition, whereas this may not be the case in the posterior VTA. Instead, in the posterior VTA, activating GABA A receptors may increase the activity of DA neurons by inhibiting GABA interneurons [78,79,84]. Micro-dialysis studies, in fact, support the idea that activating GABA A receptors within the VTA can increase somatodendritic [84] and terminal [79] DA release. One possible explanation of how a GABA A agonist and antagonist can apparently produce the same effects on terminal DA release may depend upon the microinjection site within the VTA. The ICSA studies [72,73] and another behavioral study [2], in which locomotor activity was measured, support the concept that different GABA A mediated neuronal mechanisms may be operating in the anterior versus the posterior VTA. David et al. [36], using unilateral injections into the VTA and a spatial discrimination task in a Y maze, reported that BALB/c mice readily discriminated between the arm enabling the microinjection of the GABA A antagonist bicuculline (3 pmol/50 nl) and the neutral arm. Systemic injection of the D 2 antagonist sulpiride, 30 min before the test, prevented the acquisition of bicuculline self-administration into the VTA; when given after acquisition of self-administration behavior had been acquired, sulpiride reduced self-administration and eventually extinguished this behavior. The size of the VTA in the mouse precludes any possibility of delineating anterior from posterior effects. However, the results with the mouse confirm that blocking GABA A receptors within the VTA can be reinforcing and that these affects may be mediated through activation of postsynaptic D 2 receptors. Activation of NMDA and non-nmda receptors within the VTA increases firing rates of DA and non- DA neurons, increases somatodendritic and terminal field DA release, and enhances locomotor activity (reviewed in [78]). Therefore, it would be expected that NMDA and/or non-nmda receptor agonists would be self-administered into the VTA and/or microinfusion of these agonists would produce conditioned place preference. Thus far, no studies have been reported examining the effects of NMDA and non-nmda receptor agonists in ICSA or ICPC studies. However, one study examined the ICSA of AMPA and NMDA receptor antagonists into the VTA of mice. Contrary to the expected results, David et al. [37] reported that 12 pmol/50 nl of AP-7, an NMDA receptor antagonist, or DNQX, an AMPA receptor antagonist, were unilaterally self-administered into the VTA of BALB/c mice. Using a Y maze task, mice readily discriminated between the arm producing microinjection of the antagonists and the neutral arm. Systemic injection of the DA D 2 antagonist sulpiride produced extinction of the correct arm response; discrimination was restored in subsequent sessions when injections of saline were given in place of the sulpiride. These results suggest that, in these mice, blocking NMDA and AMPA receptors enhances VTA DA neuronal activity and produces reinforcing effects. In the rat, infusion of NMDA or AMPA into the VTA enhanced DA release in the NAC, whereas co-infusion of the antagonists with the

8 136 W.J. McBride et al. / Beha ioural Brain Research 101 (1999) agonists blocked the stimulated release [81,133]. In addition, infusions of 0.3 and 1.0 nmol of the NMDA receptor antagonist AP-5 into the VTA of rats was without intrinsic effect on place conditioning; these doses, however, did block the conditioned place preference produced by infusions of sulpiride into the perifornical region of the lateral hypothalamus [98]. Therefore, it is difficult to reconcile the differences between the findings with the mouse and the results with the rat. It is possible that there are differences between the mouse and the rat in mechanisms regulating the activity of VTA DA neurons. It is also possible that the low doses of antagonists used in the mouse study may be preferentially blocking excitatory amino acid inputs on inhibitory interneurons. A third possibility is that, because of the small size of the mouse VTA, there may be diffusion of the antagonists to adjacent regions. Comparable ICSA studies need to be conducted in rats, in which a wide dose-response range is examined, to help resolve this issue Neurotensin and cholinergic receptors Neurotensin-immunoreactive fibers and receptors are found in high density throughout the VTA; 6-OHDA lesioning studies suggest that the majority of neurotensin receptors are located on dopamine neurons (reviewed in [78]). Several electrophysiological studies indicate that neurotensin increases the firing frequencies of DA neurons in vivo and in vitro (reviewed in [78]). Therefore, on this basis, neurotensin should be rewarding in the VTA. This possibility was examined in conditioned place preference and self-injection experiments [47,48]. Bilateral microinjection of neurotensin (5 g/0.5 l) into the VTA of rats produced conditioned place preference [47]. In a subsequent study, Glimcher et al. [48] reported that Sprague Dawley rats would self-infuse 2.5 g/50 nl neurotensin directly into the VTA, show lever discrimination, extinguish responding when vehicle was substituted for neurotensin, and reinstate responding on the active lever when neurotensin was reintroduced. These studies suggest that the activating effects of neurotensin on VTA DA neurons is reinforcing. The subregional effects of neurotensin within the VTA were not evaluated; however, most of the placements appear to be in the middle and posterior portions of the VTA [47]. Cholinergic innervation of the VTA is thought to arise mainly from the pedunculopontine tegmental and laterodorsal tegmental nuclei (reviewed in [78]). Both muscarinic and nicotinic receptors are found in the VTA and substantia nigra, with nicotinic receptors being most densely localized within the VTA and substantia nigra par compacta (reviewed in [78]). Systemic or local infusion of nicotine stimulates the firing rate of DA neurons in the VTA or substantia nigra (reviewed in [78]). In addition, blocking muscarinic receptors within the VTA reduced the excitation of DA neurons produced by cholinergic projections [89]. Therefore, cholinergic agonists might be reinforcing in the VTA. Bilateral microinjections of the nicotinic agonist cytisine (10 nmol/500 nl per side) into the anterior-middle portions of the VTA of Long Evans rats produced conditioned place preference [101]. Neuroanatomical controls indicated that cytisine injections dorsal to the VTA did not produce conditioned place preference. In addition, microinjection of the muscarinic agonist carbachol into the VTA also caused conditioned place preference [141]. These results suggest that activating cholinergic receptors in the VTA may be reinforcing Summary Overall, the studies support the hypothesis that activating VTA DA neurons is reinforcing. Thus far, the results from ICSA and ICPC studies appear to be in general agreement. The results with ethanol self-infusions and the self-administration of GABA A agents suggest that there may be different mechanisms mediating reinforcement in the anterior portion of the VTA versus the posterior portion of the VTA. However, this does not seem to be the case for morphine because evidence indicates that both the anterior and posterior VTA support morphine self-administration. Insufficient data are available to evaluate whether there are subregional differences within the VTA in mechanisms mediating the reinforcement of the other agents tested. Fig. 1 shows an overall hypothetical simplified diagram of mechanisms mediating reinforcement within the VTA. In the anterior VTA, activating neurotensin, nicotinic or muscarinic receptors on DA neurons increases their firing rates and promotes reinforcement. By the same token, activating mu-opioid receptors, or blocking tonic excitation at NMDA or AMPA receptors on GABA interneurons causes disinhibition of VTA DA neurons and promotes reinforcement. In addition, antagonizing tonic GABA A mediated inhibition of DA neurons will also result in their activation and the promotion of reinforcement. In the posterior VTA (Fig. 1), the same processes may function for neurotensin, nicotinic, muscarinic, mu-opioid and NMDA receptors, but not for GABA A receptors. In the posterior VTA, DA neurons may not be under tonic GABA A mediated inhibition. Instead, the DA neurons in the posterior VTA may be under tonic inhibition mediated by GABA B receptors, whereas the activity of GABA interneurons may be mediated by GABA A receptors. In this case, activation of GABA A receptors would inhibit the firing of the GABA interneurons, resulting in disinhibition of the DA neurons in the posterior VTA. The reinforcing effects of ethanol in the posterior VTA could be medi-

9 W.J. McBride et al. / Beha ioural Brain Research 101 (1999) ated through activation of a subclass of GABA A receptors [57] on GABA interneurons. In addition, the reinforcing effects of ethanol in the posterior VTA could be mediated through inhibition of a subclass of NMDA receptors [131] on GABA interneurons, or via activation of 5-HT 3 receptors [90] present on DA neurons in this subregion of the VTA. In conclusion, agents which disinhibit VTA DA neurons, block inhibitory receptors on VTA DA neurons, or activate excitatory receptors on VTA DA neurons should all produce reinforcing effects. Conversely, agents which do not activate VTA DA neurons or result in their inhibition should not be reinforcing. Additional studies will need to be carried out to provide data to support these hypotheses. 3. Nucleus accumbens as a major site for psychostimulant reinforcement and for dopamine receptors in reward mechanisms The nucleus accumbens (NAC) has also been suggested to play a key role in brain reward processes [40,140]. There is considerable evidence in the literature Fig. 1. Hypothetical simplified circuits within the VTA mediating reinforcement. In the anterior portion of the VTA, activating mu-opioid receptors on GABA interneurons, or activating neurotensin (NT), nicotinic (NIC) or muscarinic (M) receptors on DA neurons results in increased firing rates of VTA DA neurons and initiates reinforcement processes. Blocking excitatory amino acid (EAA) receptors on GABA interneurons, or blocking GABA A receptors on DA neurons reduces tonic inhibition of the DA neurons; this disinhibition results in increased firing rates of VTA DA neurons and initiates reinforcement processes. In the posterior VTA, a similar circuit is present with the exceptions that (a) tonic inhibition of DA neurons is mediated by GABA B receptors, (b) GABA A receptors mediate the activity of GABA interneurons, (c) 5-HT 3 receptors are present on DA neurons, and (d) NMDA receptors sensitive to ethanol are present on GABA interneurons. In the posterior VTA, activating GABA A or mu-opioid receptors on GABA interneurons results in disinhibition of DA neurons and produces reinforcement. Ethanol (ETOH) is reinforcing in the posterior VTA through three possible mechanisms: (a) enhancing the excitatory effects of 5-HT at 5-HT 3 receptors located on DA neurons, (b) potentiating the inhibition of GABA at GABA A receptors on GABA interneurons, and/or (c) inhibiting the EAA effects at NMDA receptors located on GABA interneurons.

10 138 W.J. McBride et al. / Beha ioural Brain Research 101 (1999) to support such a role in drug reinforcement mechanisms [87,136,138]. Intravenous self-administration studies utilizing 6-hydroxydopamine lesioning and microinjection techniques suggest that the NAC is a site initiating the rewarding effects of the psychostimulants [18,91,111,117,118], and that activation of dopamine D 1 and D 2 receptors are involved in this process [19,94,116]. The findings obtained with ICSA and ICPC studies are consistent with these results. Moreover, the ICSA and ICPC studies provided additional information for the involvement of opioid, NMDA, muscarinic and neuropeptide Y receptors within the NAC in reinforcement Psychostimulants: amphetamine and cocaine Several ICSA and ICPC studies are consistent with the NAC as a site mediating the reinforcing effects of amphetamine [16,27,140]. On the basis of the results with amphetamine, one would have predicted a similar action for cocaine. However, the findings with cocaine have been somewhat controversial, with early results indicating that cocaine was not reinforcing in the NAC [51], and more recent findings supporting the hypothesis that cocaine is reinforcing within the NAC [23,95,96]. The discrepancy between the ICSA results obtained in the early studies and those more recently published, may be a result of unit doses employed and location of injection sites within the NAC Intracranial self-administration Hoebel et al. [66] reported that rats would self-administer 3 nmol/65 nl d-amphetamine directly into the NAC, and discriminate the active from the inactive lever. Rats maintained amphetamine responding when the roles of the inactive and active levers were switched, and reduced the number of self-infusions when amphetamine was experimenter-administered. Although only one dose was evaluated and no antagonist experiments were undertaken, the results of this study suggested that the NAC supported amphetamine reinforcement. Furthermore, injection sites, mostly in the medial shell portion of the NAC, supported amphetamine self-infusions, whereas placements in the lateral ventricle or caudate-putamen did not [66]. In general agreement with these results, Phillips et al. [115,116] reported that rats would self-administer 1 10 nmol/100 nl d-amphetamine bilaterally into the medial core of the NAC near the shell-core boundary. Rats readily discriminated the active from the inactive lever, extinguished responding when vehicle was substituted for amphetamine, and resumed responding on the active lever when amphetamine was reinstated. In addition, co-infusion with D 1 or D 2 antagonists, either individually or together, partially antagonized d-amphetamine self-infusions [115,116]. These results support the idea that activation of both D 1 and D 2 receptors is required for the self-administration of amphetamine into the NAC, and that the medial core region near the core-shell interface may also support amphetamine reinforcement. In contrast to the ICSA studies with amphetamine, Goeders and Smith [51] found that microinjections of cocaine at doses as high as 5 nmol into the NAC or VTA did not maintain responding on the active lever, whereas these investigators were able to demonstrate in the same study that 100 pmol cocaine maintained responding in the medial prefrontal cortex. In agreement with these results, Carlezon et al. [23] reported that, using a single-lever operant paradigm, doses of 90 or 180 pmol did not maintain responding for cocaine in the shell portion of the NAC during the first sessions. However, doubling the dose in later sessions did produce reliable responding for the intra-nac infusions of cocaine in the four animals tested. The reason why 12 or more sessions were required to obtain cocaine self-administration behavior is difficult to explain. One possibility is that sensitization needed to develop before animals responded for the low doses of cocaine. Unfortunately, the small number of animals tested and the number of sessions required to produce ICSA behavior, although encouraging, provide weak support for cocaine being reinforcing in the NAC. However, two other studies support the idea that cocaine may be reinforcing in the NAC shell [95,96]. In one preliminary study, using a two-lever operant paradigm, McKinzie et al. [95] reported that the selectively bred alcohol-preferring P line of rats would selfinfuse 800 pmol/100 nl of cocaine into the shell region of the NAC. Rats receiving cocaine infusions responded significantly more than the control group receiving only vehicle, responded significantly more on the active than inactive lever for cocaine, showed extinction when vehicle was substituted for cocaine, resumed responding on the active lever when cocaine was reinstated, and successfully demonstrated lever reversal when the roles of the active and inactive levers were reversed. These data support the idea that doses of cocaine, comparable to the doses of d-amphetamine found to be rewarding [66], are reinforcing in the shell portion of the NAC. However, because these experiments were undertaken with selectively-bred lines of rats, the difference in the results may be due to selective breeding factors. Similar studies conducted with Wistar rats indicated that by the fourth acquisition session, rats receiving 800 and 1200 pmol/ 100 nl of cocaine responded four- to fivefold higher for cocaine than the control group given only vehicle [96]. Moreover, these rats readily discriminated the active from the inactive lever when cocaine was given, decreased responding when vehicle was substituted for cocaine, and resumed responding on the active lever when cocaine was reinstated. On the other hand, when