Time-limited modulation of appetitive Pavlovian memory by D1 and NMDA receptors in the nucleus accumbens

Results

Cannula Placements. The location of the tips of injection cannulas within the NAc is shown in Fig. 2. The tips were located in the vicinity of the ventral border of anterior commissure between 0.7 and 1.7 mm forward of bregma. A small number of injector tips were located in the ventrolateral shell region of the NAc.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Diagrammatic sections modified from the atlas of Paxinos and Watson (23) showing injector tip placements in the vicinity of the NAc. The numbers to the right of each section are anterior coordinates forward of bregma (in mm).

Autoshaping.Fig. 3 shows that all subjects receiving either immediate or delayed intra-NAc vehicle infusions acquired discriminated approach behavior to the appetitive CS. Thus, ANOVA revealed significant main effects of CS (F_1,11 = 199.66; P < 0.01) and Block (F_1,14 = 10.31; P < 0.01) and a significant CS × Block interaction (F_14,154 = 27.98; P < 0.01), but there was no distinction between the timing of the infusions and learning (Timing, F_1,11 = 1.95; P = 0.19; CS × Block × Timing, F_14,154 = 1.05; P = 0.411). A later retention test confirmed that both the immediate and delayed control groups acquired this Pavlovian conditioned behavior (see Table 1).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Impaired Pavlovian learning as a consequence of immediate postsession intra-NAc infusions of the D1 receptor antagonist SCH 23390 (0.3 μg per side) and the NMDA receptor antagonist AP-5 (1 μg per side). Delayed infusions of the same compounds produced no significant effect. Discriminated approach to the visual stimulus predictive (CS+) and nonpredictive (CS-) of food reward was evaluated over 15 blocks, each consisting of 10 discrete trials. The maximum number of CS+ or CS- approach responses for each block was 10. Error bars represent ±1 SEM.

Immediate postsession intra-NAc infusions of the D1 receptor antagonist SCH 23390 produced a severe impairment in autoshaping that was not observed in subjects receiving 24-h delayed infusions of this compound. Thus, analysis of the difference scores between CS+ and CS- responses over Block (a measure of discriminative learning) confirmed that subjects in the delayed treatment group acquired autoshaping (vehicle vs. SCH 23390: Group, F_1,13 = 0.01; P = 0.99; Block, F_14,126 = 18.6; P < 0.01; Block × Group, F_14,126 = 1.09; P = 0.37), whereas those in the immediate infusion group did not (vehicle vs. SCH 23390, Group, F_1,13 = 16.9; P < 0.01; Block, F_14,182 = 5.52; P < 0.01; Block × Group, F_14,182 = 5.24; P < 0.01). These differential effects were confirmed by a main effect of CS for the delayed group (F_1,4 = 15.9; P = 0.016) but not the immediate group (F_1,7 = 0.52; P = 0.49) and by a probe test showing impaired and unimpaired discrimination in the immediate (P = 0.716) and delayed (P = 0.004) groups, respectively (see Table 1). By contrast, postsession infusions of the D2 receptor antagonist sulpiride and d-amphetamine were without significant effect on Pavlovian learning (see Fig. 5, which is published as supporting information on the PNAS web site).

Fig. 3 shows that subjects receiving immediate postsession infusions of the NMDA receptor antagonist AP-5 failed to discriminate significantly between the CS+ and CS- during acquisition (CS, F_1,7 = 3.92; P = 0.088; CS × Block, F_14,98 = 1.25; P = 0.251). By contrast, subjects receiving delayed infusions of AP-5 showed significant CS-/CS+ discrimination (CS, F_1,8 = 39.62; P < 0.01; CS × Block, F_14,112 = 3.09; P < 0.01). Although subjects receiving immediate AP-5 infusions were significantly disrupted in acquiring the task (vehicle vs. AP-5, difference score; immediate condition, F_1,13 = 8.56; P = 0.012; delayed condition, F_1,13 = 0.02; P = 0.91), they nevertheless demonstrated significant learning during the subsequent probe retention test, similar to the delayed AP-5 group. Therefore, although immediate postsession intra-NAc infusions of AP-5 significantly retarded appetitive approach learning, they did not abolish it.

Omission Training.Fig. 4 shows the effects of postsession intra-NAc infusions of vehicle, AP-5, and SCH 23390 on performance of an omission schedule where reward was made contingent on the nonoccurrence of a CS+ approach response. Vehicle-treated rats continued to discriminate between the CS+ and CS- during this procedure (CS, F_1,11 = 10.64; P = 0.01), and this did not alter as a function of training (CS × Block, F_4,44 = 0.27; P = 0.895; Block, F_4,44 = 0.744; P = 0.569) or timing of the infusions (Timing, F_1,11 = 1.23; P = 0.296; CS × Timing, F_1,11 = 0.027; P = 0.872). Rats receiving immediate postsession intra-NAc infusions of SCH 23390 were significantly impaired in discriminating between the CS+ and CS- during omission training (CS, F_1,7 = 0.009; P = 0.93), unlike subjects that received delayed intra-NAc infusions of SCH 23390 (CS, F_1,4 = 7.03; P < 0.05) that did not diminish over the course of the omission session (CS × Block, F_4,16 = 0.27; P = 0.89). Despite exhibiting a significant preference for the CS+ during probe retention testing, rats receiving immediate AP-5 infusions failed to discriminate between the CS+ and CS- during omission training (CS, F_1,7 = 0.48; P = 0.512; CS × Block, F_4,28 = 0.24; P = 0.913). By contrast, rats in the delayed AP-5 group continued to approach the CS+ during this procedure (CS, F_1,8 = 18.75; P < 0.01; CS × Block, F_4,32 = 1.52; P = 0.22). Subjects receiving postsession infusions of sulpiride or amphetamine during acquisition were unimpaired during the omission contingency test (see Fig. 6, which is published as supporting information on the PNAS web site).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Inflexible CS+ approach during omission training, where reinforcement was made contingent on the nonoccurrence of a CS+ directed response. A failure to extinguish responding to the CS+ during this procedure indicates that appetitive learning was supported mainly by Pavlovian associative mechanisms. Error bars represent ±1 SEM.

Discussion

The results reported here demonstrate that D1 receptors and NMDA receptors in the NAc contribute to the early consolidation of appetitive Pavlovian conditioning. Specifically, we have shown that learning associated with the normal emergence of a discriminated approach response to a CS predictive of food reward is abolished by the postsession intra-NAc administration of the D1 receptor antagonist SCH 23390. The inability of delayed 24-h infusions of SCH 23390 to interfere significantly with this form of appetitive Pavlovian learning argues strongly that D1 receptors in the NAc play a time-limited role in consolidating appetitive Pavlovian memories. These data thus support the predicted involvement of midbrain DA neurons in appetitive learning based on electrophysiological studies in monkeys (25) and, additionally, demonstrate a causal involvement of D1 receptors in this form of learning. Our findings also implicate NAc NMDA receptors in appetitive Pavlovian learning and therefore are consistent with the widely held concept that D1 and NMDA receptors within distributed corticostriatal networks are important targets underpinning long-term potentiation and other forms of neuronal plasticity hypothesized to underlie associative processes in learning (5, 26-29).

Previous work has established a critical and interactive role of D1 and NMDA receptors in the NAc in appetitive instrumental learning based on lever-press responding for food reward (4). The present results extend these findings in two important ways. First, we used postsession intra-NAc infusions of SCH 23390 and AP-5 to distinguish effects on learning from effects on performance (12). Second, we used what is a principally a Pavlovian task that is minimally sensitive to changes in instrumental contingencies (10, 11, 30). This was demonstrated most convincingly by the failure of control rats to extinguish autoshaped conditioned approach when reinforcement was made contingent on the nonoccurrence of a CS+ approach. In contrast, appetitive instrumental learning generally depends on interacting Pavlovian and instrumental processes that preclude simple inferences to be drawn on the nature of any learning deficit. For example, reinforcing a lever-press response in the presence of a discriminative stimulus such as the illumination of a light stimulus above the lever may, in some circumstances, be represented by a Pavlovian association between a light CS and food (21).

Much previous evidence has implicated the NAc in appetitive Pavlovian and instrumental learning (1, 2, 31) in ways that are readily distinguishable from its well known involvement in motivation, reinforcement, and motor activity (32). In particular, the NAc has been implicated in the acquisition and/or consolidation of spatial information but not in its retrieval (1, 33). In intact rats, place learning in the aversively motivated Morris water maze is impaired by pretraining intra-NAc infusions of haloperidol (34) and by posttraining intra-NAc infusions of sulpiride (3) and AP-5 (7), but not by posttraining intra-NAc infusions of the AMPA receptor antagonist DNQX (7). Consistent with these data, posttraining intra-NAc infusions of AP-5 in the present study retarded learning of a spatially conditioned appetitive approach response. In our hands, however, posttraining intra-NAc infusions of sulpiride neither impaired nor facilitated autoshaping, despite previous evidence that spatial escape learning is impaired by a comparable posttraining intra-NAc dose of sulpiride (3). The reasons for this are unclear but may be due to differences in underlying learning processes engaged by the two tasks, for example appetitive vs. aversive, in addition to apparent differences in the timing, frequency, and site of the intra-NAc sulpiride infusions. For example, Setlow and McGaugh (3) infused sulpiride into more rostral and dorsal sites of the NAc core compared with the present study. Moreover, posttraining infusions of d-amphetamine (10 μg) into the dorsomedial shell of the NAc but not the core, as here, have recently been shown to facilitate appetitive Pavlovian conditioning (6). Therefore, variations in the precise effects of DA receptor agents on appetitive learning and memory may be explained by the existence of functionally distinct neuroanatomical loci within the NAc, possibly consistent with the heterogeneous distribution of inputs to the NAc from the prefrontal cortex, hippocampus, and amygdala (35).

Learning-related plasticity has often been attributed to D1 and NMDA receptor interactions within broadly distributed but interconnected brain regions, including the NAc, basolateral amygdala, and prefrontal cortex (2, 29, 36). Recently, distinct and interactive effects of NMDA and D1 receptor antagonists on appetitive instrumental learning have been reported with impaired learning after pretraining intra-NAc infusions of AP-5 and SCH 23390 (4). In addition, pretraining systemic and intra-NAc NMDA and D1 receptor antagonists have been reported to impair Pavlovian learning in the context of the conditioned appetitive approach (9, 24). Because intra-NAc AP-5 infusions appear to have no significant effect on the expression of the appetitive Pavlovian approach (9), it is likely that NMDA receptors in the NAc contribute to the acquisition and/or consolidation of appetitive Pavlovian learning. The present data provide evidence that NMDA receptors in the NAc play a significant role in the early consolidation of discriminated approach learning. Intriguingly, however, a summary of recent unpublished findings suggests that posttrial infusions of either AP-5 or SCH 23390 do not interfere with appetitive instrumental learning (2). The basis for these discrepant results is unclear, but it is assumed that differences in the associative processes and/or distributed brain networks that support learning in these tasks may be contributory factors.

Although AP-5 retarded significantly the acquisition of autoshaping after its immediate postsession intra-NAc administration, rats nevertheless displayed significant evidence of learning during a subsequent retention test involving simultaneous presentations of the CS+ and CS-. These findings suggest that the probe test is probably a less-sensitive method for detecting impairments in discriminative approach learning; indeed, simultaneous procedures have previously been shown to be more sensitive than successive procedures in revealing discrimination learning (22). AP5-treated rats, however, failed to maintain discriminative performance when an instrumental contingency was introduced that prevented food delivery after a CS+ approach. The most parsimonious explanation for this finding was that AP-5 infusions blocked the predominantly Pavlovian elements of this task to the point that they were no longer sufficient to maintain discriminated approach behavior. In view of the fact that many learning situations likely engage competing Pavlovian and instrumental influences (21, 22), it is also possible that a small residual instrumental component supported appetitive approach learning in AP-5 rats that was subsequently disrupted by the omission contingency. Due to the severe learning deficits induced by SCH 23390, no inferences can be drawn on the relative involvement of D1 receptors in this process. However, pretraining depletion of NAc DA produces effects on omission performance that are broadly opposite to the effects of AP-5. Specifically, DA-lesioned rats exhibit an inflexible CS+ approach rather than the normal fluctuations observed in sham rats that may indicate transient effects of instrumental contingencies (10).

Conclusion

It is hypothesized that converging afferent inputs to the NAc from glutamatergic neurons arising from limbic-cortical structures, as well as dopaminergic neurons from the midbrain (35), contribute to, and are necessary for, distinct aspects of memory consolidation relating to Pavlovian conditioned approach learning. Additional studies are required to understand how multiple interacting but dissociable neural systems involving the anterior cingulate cortex (18, 30, 37), orbitofrontal cortex (38), and amygdala (39) contribute to this learning process.