The Ventral Tegmental Area Is Required for the Behavioral and Nucleus Accumbens Neuronal Firing Responses to Incentive Cues

Introduction

Elucidating the actions of dopamine on nucleus accumbens (NAc) neurons is essential for understanding motivated behaviors. Reward-predictive cues trigger reward seeking, activate dopamine neurons (Ljungberg et al., 1992; Schultz et al., 1993), and cause NAc dopamine release (McCullough and Salamone, 1992; Bassareo and Di Chiara, 1999; Weiss et al., 2000; Robinson et al., 2001, 2002; Roitman et al., 2004). These findings suggest that a dopamine-induced change in NAc neuronal firing signals that a stimulus is motivationally salient and increases the probability of a behavioral response (Berridge and Robinson, 1998). In support of this idea, reducing NAc dopamine function inhibits reward seeking in response to reward-associated cues (Blackburn et al., 1992; Di Ciano et al., 2001; Wakabayashi et al., 2004), whereas augmenting NAc dopamine transmission increases operant responding elicited or maintained by reward-associated cues (Taylor and Robbins, 1986; Wolterink et al., 1993; Wyvell and Berridge, 2000). Although NAc dopamine appears critical for goal-directed behavioral responses to predictive cues, it is not yet known which NAc neurons are modulated by dopamine during reward seeking and how this modulation facilitates goal-directed behavior.

NAc neurons in behaving animals exhibit phasic excitations and inhibitions time locked to many task events, such as operant behavior and reward consumption (Apicella et al., 1991; Schultz et al., 1992; Carelli and Deadwyler, 1994; Chang et al., 1994; Peoples and West, 1996; Nicola and Deadwyler, 2000; Nicola et al., 2004a,b). Additional subpopulations are excited or inhibited by reward-predictive cues, suggesting that cue-evoked firing changes in NAc neurons contribute to cue-elicited behavioral responses. For instance, in primate ventral striatum, cue-evoked excitations depend on the magnitude and type of reward predicted by the cue and the temporal proximity of reward (Bowman et al., 1996; Hollerman et al., 1998; Shidara et al., 1998; Hassani et al., 2001; Cromwell and Schultz, 2003). In rat NAc neurons, cue-evoked excitations and inhibitions (incentive cue excitations and inhibitions) are larger for reward-predictive than for nonpredictive cues (Ghitza et al., 2003; Nicola et al., 2004a) and larger when the animal makes a behavioral response to the cue (Nicola et al., 2004a). The possibility that cue-evoked firing of NAc neurons depends on dopamine suggests a specific means of testing the hypothesis that these firing patterns are required for the behavioral response. NAc incentive cue excitations and inhibitions begin at approximately the time that midbrain dopamine neurons are excited by such cues. Furthermore, in well trained animals performing cue–response tasks, the dopamine neuron response is not correlated with either reward acquisition or operant behavior (Ljungberg et al., 1992; Schultz et al., 1993), nor is it evoked by cues that do not predict reward (Mirenowicz and Schultz, 1994). Therefore, it is likely that dopamine selectively modulates responses to reward-predictive cues. If this is the case, then removing the dopamine input to the NAc should reduce the cue-evoked excitation and/or inhibition of NAc neurons. Furthermore, if these excitations and inhibitions are required for reward seeking in response to predictive cues, then interruption of the dopaminergic input to the NAc should also reduce the behavioral response to such cues. In fact, we find that both the cue-evoked firing of NAc neurons and the behavioral response to cues were concomitantly abolished by localized injection of baclofen into the ventral tegmental area (VTA). In addition, we present evidence to localize the effect of reduced dopamine to the NAc by showing that blockade of dopamine receptors in the NAc reduces cue-elicited behavior.

Materials and Methods

Two experiments were performed, involving two separate groups of rats. First, we conducted a behavioral study in which we trained animals on a discriminative stimulus (DS) task, implanted microinjection cannulas in the NAc, and determined how dopamine antagonists injected in the NAc affected responding to predictive cues during the DS task. Next, we used a subset of the rats from our previous electrophysiological studies (Nicola et al., 2004a,b) to determine whether VTA baclofen injection concomitantly affected behavior on the DS task and the firing of NAc neurons. These animals received, in the same surgery, both recording electrodes in the NAc and microinjection cannulas in the VTA.

For both experiments, the subjects were male Long–Evans rats [Harlan Sprague Dawley (Indianapolis, IN) or Charles River Laboratories (Wilmington, MA)], which weighed ∼350 gm on arrival. They were individually housed on a 12 hr light/dark cycle; experiments were conducted during the light phase. After receipt, rats were allowed at least 1 week of ad libitum food and water, followed by 1 week of restricted food and water before training. Throughout all experiments (both behavioral and electrophysiological), restriction was accomplished by allowing the animals 1 hr of ad libitum food and water per day, at the end of experimental manipulations. Animal handling and experiments conformed to National Institutes of Health and Ernest Gallo Clinic and Research Center animal care and use policies.

Behavioral study (NAc microinjection). Rats (n = 12) were trained on the DS task exactly as described by Nicola et al. (2004a). The task is diagrammed in Figure 1. Standard operant chambers (23.5 × 30.5 cm) contained two nose pokes, one on each side of a reward receptacle, two houselights, a white-noise speaker, and a tone speaker (Med Associates, St. Albans, VT). Liquid 10% sucrose reward was delivered into a well in the reward receptacle by a syringe pump. In the DS task, the DS was presented on a variable interval schedule with an average interval of 2 min. The DS consisted of an intermittent tone (200 msec tone-on intervals, 550 msec tone-off intervals) coupled with dimmed houselights (by turning off one of the two). Each DS presentation lasted up to 20 sec. If the animal responded to the DS by making a nose poke in the active nose-poke hole, the DS was terminated, 50 μl of 10% sucrose was delivered into the reward receptacle, and a 20 sec conditioned stimulus (CS) was presented. The CS consisted of an 8 kHz intermittent tone (200 msec on, 550 msec off) with continued dimmed houselights. In addition to the DS, a non-rewarded stimulus (NS) was presented on an independent variable interval 2 min schedule. The NS also consisted of an intermittent tone (200 msec on, 550 msec off) and dimmed houselights. For six rats, the DS tone was 6 kHz and the NS tone was 4 kHz, and, for the remaining six rats, the DS was 4 kHz and the NS was 6 kHz. Responding during the NS, the CS, or in the absence of the DS was never rewarded. Animals were given one 2 hr session per day.

When animals were trained to criterion performance (>90% DS response ratio, defined as the proportion of all DSs in the session to which the animal responded), they were stereotaxically implanted with microinjection cannulas in the NAc. Anesthesia was induced with ketamine and xylazine (intraperitoneally) and maintained with isoflurane. Target coordinates of the injection cannulas were as follows (in mm relative to bregma): anteroposterior (AP), 1.6; mediolateral (ML) ±1.1; dorsoventral (DV), –7.5. Guide cannulas were 26 ga double cannulas separated by 2.2 mm (Plastics One, Roanoke, VA). Obturators flush with the bottom of the guide cannulas were inserted and remained in place at all times except during drug injection. The guide cannulas were secured to the skull with dental acrylic and bone screws. Animals were allowed 1 week to recover before retraining to criterion.

After retraining, animals received one of seven injections before behavioral testing sessions. Each injection session was followed and preceded by a recovery session in which no drugs were injected. Three different doses of the D₁ antagonist SCH23390 and the D₂ antagonist raclopride (Sigma, St. Louis, MO) were injected. For SCH23390, these were 1, 2, and 4 μg/side. For raclopride, these were 2, 4, and 6 μg/side. Drugs were dissolved daily in saline and injected bilaterally in a volume of 500 nl/side. A vehicle control injection was also given. The order of the seven injections was randomized and different for each rat. In the injection procedure, the animal was gently restrained while the obturators were removed and the injection cannulas were inserted into the guides. After a 1 min wait, the total volume was injected over 2 min by a programmable syringe pump. After an additional 1 min wait, the injection cannulas were removed, the obturators were replaced, the animal was immediately placed in the behavioral chamber, and the session was started.

To analyze the behavioral data, the measured parameters (DS response ratio, NS response ratio, and latency to respond to the DS) were compared across drug doses using one-way within-subjects ANOVA. We also examined the time course of the drug effects over the behavioral session by using two-way within-subjects ANOVA, with one factor being drug dose and the other the time since the start of the session (broken into 20 min bins) (see Fig. 2 J, K).

Electrophysiological study (NAc recording and VTA baclofen injection). A subset of 14 of the animals described in our previous studies (Nicola et al., 2004a,b) was used for this experiment. Animals were trained exactly as described in the previous papers and were run on a DS task that was identical to the one described above for the behavioral study, with two exceptions. First, for the electrophysiology–VTA baclofen injection study, the operant chambers were larger (40.6 × 40.6 cm), and second, for most animals, the sucrose reward was delivered with a dipper instead of a syringe pump.

Fully trained animals received both microwire recording electrodes (described in detail by Nicola and Deadwyler, 2000; Nicola et al., 2004a) implanted in the NAc and guide cannulas for microinjection in the VTA during the same stereotaxic surgery. Arrays (NB Labs, Denison, TX) consisted of two parallel rows of four electrodes (eight electrodes total per array; electrode diameter, 50 μm) and were ∼0.7 mm long and 0.3 mm wide. They were implanted bilaterally with the long dimension in the saggital plane. Target coordinates of the medioposterior electrode were as follows (in mm relative to bregma): AP, 1.0; ML, ±0.5; DV, –6.5 to –8. Target coordinates of the VTA injection cannulas were as follows (in mm): AP, –4.5 to –5.0; ML, ±0.25 to ±0.5; DV, –7.8 to –9.5. Cannulas were placed in the rostral VTA because, compared with caudal VTA, the activity of rostral VTA neurons appears to contribute preferentially to reinforcement (Ikemoto et al., 1998; Carlezon et al., 2000). Guide cannulas were 27 ga stainless steel tubes. They were implanted at an angle of ∼10° from vertical in the coronal plane, to allow the experimenters access to both the electrode connectors and the cannulas. Obturators (0.38 mm stainless steel wire) were inserted and remained in place at all times except during injection. The electrodes and cannulas were secured to the skull with dental acrylic and bone screws. Animals were allowed to recover from surgery for 1 week before commencement of experiments.

Electrophysiological recording was conducted as described previously (Nicola et al., 2004a). Animals were connected to the recording apparatus [consisting of a head stage with field effect transistors, cable, commutator to allow the animal free movement within the chamber, and Plexon Inc. (Dallas, TX) spike sorting hardware and software], and units were isolated as described previously. A 60–90 min baseline recording session, during which animals were allowed to perform the DS task, was run every day for each rat (5 d/week). At the end of the baseline session, animals were removed from the behavior–recording chamber and injected with 100 nl of saline or (+)baclofen into the VTA. Injections were made over 1 min, with a 1 min postinjection wait before removal of the injector. After the injection, the animals were immediately reconnected to the recording apparatus and placed in the chamber for the postinjection session. This lasted at least 2 hr or until behavioral recovery. Recovery was defined as regular behavioral responding to the DS. Fresh baclofen stock solution was dissolved in saline weekly or more frequently.

Although most VTA injections were unilateral, the VTA straddles the midline (cannulas were placed accordingly), and the injected drug was likely to affect both hemispheres. Bilateral injections were made in some cases but were not included in the dose–response analysis of behavior shown in Figure 2. They were, however, included in the analysis of the effects of VTA baclofen injection on the firing of NAc neurons. Unilateral injections alternated daily between the implanted cannulas, and the dose was adjusted each day to determine a dose that resulted in transient but pronounced behavioral effects that recovered within the postinjection session. The result was that many different doses were administered, which allowed us to construct the dose–response curves shown in Figure 2. However, not enough neurons of each class were recorded in each dose range for an accurate dose–response analysis of the effects of VTA baclofen on NAc neuronal firing. Nevertheless, the question we wanted to investigate was whether there was a significant effect on firing that paralleled a reduction in behavioral responding to the DS. Therefore, each postinjection session was determined to be either the result of a “behaviorally effective” injection or a “behaviorally ineffective” injection. This determination was based on whether the animal's DS response ratio (the proportion of DSs presented to which the animal made a nose-poke response) was reduced by VTA injection to <50% (effective) or remained at >50% (ineffective) after the injection. Although higher doses tended to be effective, effective injections also often resulted from the lower baclofen doses, and ineffective injections sometimes resulted from higher doses. Therefore, dividing the sessions into those with behaviorally effective and ineffective injections provided us with an independent criterion for choosing sessions to analyze the effects of VTA baclofen injection on NAc neuronal firing. Analysis of NAc firing after behaviorally effective injections regardless of the dose of baclofen minimized the variability introduced by differences in injection site, possible clogging of the cannula, and backflow of injectate along the cannula track, all of which could potentially reduce the effectiveness of the injections.

With the microwire technique, electrodes are fixed at the time of surgery and cannot be moved subsequently by the experimenter. Microwire recordings are usually quite stable, and thus the same neuron can often be recorded for several days on the same wire. In cases in which the waveform of a neuron was similar on consecutive days, the spike firing data from the neuron was averaged across days provided that the neuron maintained its phasic response in the preinjection baseline period. When a neuron was lost to the noise, neurons recorded subsequently from the electrode that displayed the same response type were not used for analysis. This eliminates the possibility that a single neuron could have contributed n > 1 to any given analysis.

The neurons used for each analysis were those that exhibited the response type of interest in the preinjection condition. The magnitude of event-related excitation of each neuron during the preinjection period was computed by taking the average firing rate in the window of interest (for a list of analysis windows for each response type, see Table 1) and subtracting the average firing rate during the 10 sec pre-DS baseline. (The magnitude of event-related inhibition was computed by subtracting the firing rate in the window of interest from the pre-DS baseline.) This was repeated for each condition (preinjection, postinjection, and recovery). The excitation or inhibition of each neuron in each of these conditions was used to compute the median excitation or inhibition across neurons in each condition. The median excitations or inhibitions in each condition were then compared with nonparametric within-subjects ANOVA or paired Wilcoxon signed rank tests. A separate ANOVA was used to compare baseline firing rates in the 10 sec before each DS presentation across conditions. Nonparametric tests were used because the firing rates of NAc neurons were not normally distributed.

Histology. Animals were deeply anesthetized with Euthasol and then perfused first with saline and 10% Formalin. Electrode placements were determined by passing a current through each electrode before perfusion to deposit iron at the tip and completing the perfusion with potassium ferrocyanide solution to develop the Prussian blue deposit (Nicola et al., 2004b). Cannula placements were verified by injection of methylene blue or by deposition of iron through an electrode cut to the same length as the injection cannulas and developing the Prussian blue deposit. Brain sections (40 μm) were cut on a microtome. NAc and VTA sections were stained with neutral red. In some cases, alternate VTA sections were immunostained with antityrosine hydroxylase antibody to visualize dopamine-containing neurons.

Results

Discussion

Microinjection of D₁ or D₂ receptor antagonists into the NAc substantially reduced operant responding to reward-predictive cues. Therefore, NAc dopamine must affect the firing of at least one subpopulation of NAc neurons, and this effect on firing must contribute to the behavioral response to cues. VTA baclofen injection reduces NAc dopamine release (Tanner, 1979; Westerink et al., 1996) and should therefore produce effects similar to those of NAc dopamine antagonism. In fact, the behavioral effects of VTA baclofen injection and NAc dopamine receptor antagonism were strikingly similar. The most dramatic effects of VTA baclofen on NAc neuronal firing patterns were a reduction in incentive cue excitation and inhibition (and receptacle exit excitations). Phasic excitations and inhibitions associated with operant and consummatory behaviors were not affected by VTA baclofen injection. Thus, the only phasic firing patterns that both preceded the goal-directed behavior and were reduced by VTA baclofen were incentive cue excitations and inhibitions. These results strongly suggest that NAc cue-evoked excitations and inhibitions depend on dopamine and are required for the animal's goal-directed behavioral response to cues.

Neurons with incentive cue excitations and inhibitions exhibit critical properties that are consistent with the interpretation that their firing promotes a specific behavioral response to cues. Incentive cue excitations and inhibitions are triggered at short latency by predictive cues and are often sustained throughout the behavior required to obtain the reward predicted by the cue. These changes in firing are larger in response to reward-predictive cues than non-predictive cues (Ghitza et al., 2003; Nicola et al., 2004a) and larger when the animal makes a behavioral response than when no response is made or when the behavioral response occurs at very long latency (Nicola et al., 2004a). After VTA baclofen injection, these cue-elicited excitations and inhibitions were dramatically reduced at the same time that behavioral responding was severely impaired. The most parsimonious interpretation is that the direct VTA to NAc dopamine projection is required for these firing patterns and that the firing of these NAc neurons is required for the animal to perform the initial phase(s) of the cue response (orienting, approach, and subsequent nose poke).

One alternative interpretation is that a projection from the VTA to another nucleus affects the probability and latency of the behavioral response, and thus the VTA affects cue-elicited firing patterns in the NAc indirectly. For example, VTA dopamine neurons project to the prefrontal cortex and basolateral amygdala, both of which innervate the NAc. Indeed, the basolateral amygdala is required to respond to DSs under some conditions (Kantak et al., 2002; Yun and Fields, 2003). VTA baclofen injection undoubtedly reduced dopamine release in these nuclei as well as the NAc, and therefore it is conceivable that the reduction in incentive cue excitation and inhibition was attributable, in part, to a change in firing in afferent structures other than the VTA. However, the fact that injection of dopamine receptor antagonists into the NAc profoundly reduced cue responding means that the firing of at least one subpopulation of NAc neurons is both dopamine dependent and necessary for the performance of the task. Because VTA baclofen injection reduces NAc dopamine (Tanner, 1979; Westerink et al., 1996), it should also affect those firing patterns that are dopamine dependent. Even if the effects on firing caused by VTA baclofen injection were partly attributable to the altered firing of neurons in other nuclei that project to the NAc, it is unlikely that reduction of dopamine release in the NAc (a major target of VTA dopamine neurons) contributed nothing to the observed effects on firing. Similarly, another alternative interpretation, that inhibition of the GABAergic projection from the VTA to the NAc (Van Bockstaele and Pickel, 1995) is responsible for the observed effects of VTA baclofen injection on NAc neuronal firing, is not supported by the similarity of the behavioral effects of VTA baclofen injection and NAc dopamine receptor antagonism. Thus, although it remains a formal possibility that dopamine receptor antagonism in the NAc had its behavioral effects by changing some aspect of NAc neuronal firing not measured by our electrodes after VTA baclofen injection, the profound reduction of incentive cue excitations and inhibitions by VTA baclofen injection, together with the fact that they are tightly time locked to the cue onset and both encode the predictive value of the cue and the probability of a behavioral response, suggest that these firing patterns are dopamine dependent and promote the behavioral response to predictive cues.

VTA baclofen injection reduced the baseline firing rate of neurons with operant excitation and inhibition. These effects could have resulted in reduced DS responding if the tonic firing of NAc neurons is permissive for the behavior. However, this possibility is not supported by previous studies of the effects of dopamine iontophoresis onto NAc and striatal neurons in awake animals, which showed that dopamine application reduced (Rolls et al., 1984; Kiyatkin and Rebec, 1996) or did not consistently change (Inase et al., 1997) the baseline firing rate, in contrast to our observations that inactivating dopamine neurons reduced baseline firing of some neurons. The iontophoretically applied dopamine should mimic tonically released dopamine because, in these studies, the dopamine was applied for prolonged periods of time. Therefore, although a permissive role for tonic NAc neuronal firing cannot be definitively ruled out, the available evidence is most consistent with the idea that incentive cue excitations and/or inhibitions are dopamine dependent and required for the behavioral response to predictive cues.

VTA baclofen injection caused a trend toward a reduction of the baseline firing rate of neurons with incentive cue excitations and inhibitions. This raises the question of whether the reduction in baseline rate prevented further excitation or inhibition by ceiling or floor effects. However, the reduction in baseline firing rate was clearly smaller (as a proportion of the preinjection baseline rate) than the reduction in cue-evoked excitation and inhibition (Table 1). Furthermore, neurons with operant excitations and inhibitions, whose baseline firing was reduced by VTA baclofen injection, exhibited normal event-related phasic excitations and inhibitions, and receptacle exit excitations were reduced without a corresponding reduction in baseline firing. A more likely hypothesis is that the phasic firing patterns that were dependent on the VTA (incentive cue excitation and inhibition, and receptacle exit excitation) were dependent on the phasic release of dopamine. This possibility is consistent with the properties of midbrain dopamine neurons, which fire short bursts in response to predictive cues (Ljungberg et al., 1992; Schultz et al., 1993). Recent measurements of phasic dopamine release in awake animals indicate that predictive cues do in fact cause the rapid release of dopamine within the NAc (Robinson et al., 2001, 2002; Roitman et al., 2004), suggesting that DS presentation increased NAc dopamine release rapidly and transiently.

That the cue-evoked firing responses of NAc neurons depend on dopamine is supported by previous findings that the phasic responses to cues of primate striatal tonically active neurons (TANs) are abolished by depletion of dopamine from the striatum (Aosaki et al., 1994). Although our neurons are not likely to be TANs (Nicola et al., 2004a), dopamine could exert similar response-enhancing effects on NAc projection neurons. This idea is further supported by findings of dopamine-mediated enhancement of evoked versus baseline firing rate in striatal (Rolls et al., 1984; Kiyatkin and Rebec, 1996) and prefrontal cortical (Sawaguchi et al., 1986, 1988, 1990; Sawaguchi, 1987) neurons recorded in behaving animals. A number of cellular mechanisms could underlie such enhancement of incentive cue excitation (Gonon and Sundstrom, 1996; Gonon, 1997; Hernandez-Lopez et al., 1997; Nicola and Malenka, 1997; Nicola et al., 2000; West and Grace, 2002; Hopf et al., 2003; O'Donnell, 2003). Dopamine-mediated enhancement of incentive cue inhibition could result from enhancement of incentive cue excitation in neurons that exert collateral inhibition onto incentive cue-inhibited neurons or from direct inhibitory effects of dopamine on excitatory synaptic transmission (Harvey and Lacey, 1996; Nicola et al., 1996; Nicola and Malenka, 1998).

Although VTA baclofen injection reduced receptacle exit excitations, this effect is not likely to explain the reduced cue responding, because these excitations did not precede the operant response. Because a subset of these excitations last for tens of seconds after the exit, and this prolonged excitation is correlated with decreased uncued operant responding (Nicola et al., 2004b), these excitations may actively inhibit the operant response in the absence of the DS. Therefore, the effect of dopamine released at receptacle exit may be to facilitate the switch between consummatory and other behaviors, consistent with proposals for a role of dopamine in the basal ganglia in switching from ongoing behavior to behaviors that are more beneficial given the changing environmental conditions (Oades, 1985; Redgrave et al., 1999). Thus, the VTA dependence of incentive cue excitation and inhibition and receptacle exit excitation suggests that these firing patterns are the critical dopamine-dependent elements within the NAc that are necessary to promote the behavioral response to predictive environmental information.

In summary, both VTA baclofen injection and NAc dopamine receptor antagonism reduced operant responding to a predictive cue. The most prominent effects on NAc neuronal firing caused by VTA baclofen injection were a reduction in the excitation and inhibition evoked by the cue. These results strongly suggest that incentive cue excitations and inhibitions are dopamine dependent and facilitate the goal-directed behavioral response to cues.