Dopamine and Its Role On Appetitive Learning
Date: Thursday, March 15th, 2001
Title: Dopamine and Appetitive Learning
Speaker: Dr. Jon C. Horvitz, Ph.D., Columbia University, New York, N.Y.
According to traditional views, nigrostriatal dopamine (DA) neurons, originating in the substantia nigra (SN) and projecting to the dorsal striatum, play a role in the expression of motor acts. Conversely, mesolimbic DA neurons, originating in the ventral tegmental area (VTA) and projecting largely to the nucleus accumbens and other ventral striatal regions, play a role in reinforcement or incentive motivational processes. From these views, it might have been expected that nigrostriatal DA neurons would fire in relation to specific motor acts, and that mesolimbic DA neurons would respond specifically to the presentation of reward stimuli. However, single-unit studies show that DA neurons within the SN do not respond to phasic bodily movements, nor do DA neurons within the VTA respond exclusively to reward stimuli. Midbrain DA neurons appear to respond to salient and arousing environmental stimuli.
Dr. Horvitz summarized several experiments that helped determine the respective roles of the DA receptor families, D1 and D2, in appetitive learning. These studies are described below:
Execution of head-entry response to previously-learned food cue:
Rats’ head entries into a food compartment were measured in the presence and absence of a food cue. D1 (0.08 and 0.16 mg/kg SCH23390) and D2 receptor blockade (0.2 and 0.4 mg/kg raclopride) suppressed the emission of spontaneous head entries (i.e., those emitted in the absence of the food cue). Yet, the same behavior, when emitted in response to a well-learned food cue, was unaffected- by either D1 or D2 receptor blockade. In contrast D1 and D2 receptor blockade, at the same doses did disrupt responses to the food cue, prior to over-training. Therefore, the invulnerability of the learned behavior to DA receptor blockade is related to the acquired associate strength of the response-eliciting stimulus.
Acquisition of a head-entry response to a new food cue:
Animals were trained to acquire a head-entry response to a new food cue (a tone that occurred 3 sec prior to food delivery). 0.16 mg/kg of D1 antagonist SCH23390 disrupted the acquisition of the tone/food association, while D2 receptor blockade (0.2 and 0.4 mg/kg of the antagonist raclopride) produced a dose-dependent enhancement of the acquisition of this association. On the day after the tone/food training, animals were presented with the tone alone. The animals that had received the tone/food pairing under the influence of the D1 antagonist were less likely to enter the food compartment during the tone than those that had received the pairing under the influence of the D2 antagonist. These results are consistent with recent evidence that D1 and D2 receptor binding produce opposite effects on a number of cellular events within striatal target neurons.
Nevertheless, DA plays a critical role in the execution of goal-directed behavior, and in the acquisition of new responses to reward-predicting stimuli. The present studies show that while activity at both types of DA receptor families (D1 and D2) is critical for the expression of goal-directed behavior, D1 and D2 receptor activity produces opposing effects on associative learning.
Q. The axon subtypes and loci within the noradrenergic system have classic pre- and post-synaptic connections; is the DA system any different? What about the fusiform connections with DA?
A. DA axons make pre- and post-synaptic connections with target neurons in the striatum and accumbens. DA axons also terminate synaptically and extra-synaptically. The functional significance of these different types of terminal connections from a behavioral standpoint remains to be understood.
Q. Is it clear when DA neurons fire in response to eating? Do the neurons fire before feeding begins?
A. Schultz has shown that if the time of food delivery is predictable on the basis of a prior cue, the midbrain DA neurons will fire to the food cue, and not to the food itself. If the time of food delivery is not predictable, DA neurons will fire in response to the food itself. Whether the stimulus is liquid food squirted into the mouth, or a tone signaling food delivery, or another salient environmental event, such as a light flash, the DA neuron will fire approximately 60 ms after the onset of the event.
Q. Do neurons in the VTA respond similarly to those in the SN?
A. Single-unit data show that VTA and SN DA neurons fire under similar conditions (to unexpected rewards, novel events, intense sensory events) and with similar response latency and duration. On the other hand, microdialysis (with a 10 min sampling period) is more likely to find elevations in DA concentration at VTA terminal sites (nucleus accumbens/ventral striatum and prefrontal cortex) than at SN terminal sites (dorsal striatum). It is possible that highly efficient DA reuptake and feeback regulation mechanisms within the dorsal striatum quickly clear DA from the extracellular space. Therefore, VTA and SN DA neurons may both respond similarly to salient events, but DA within the striatum is not permitted to accumulate over the 10 min microdialysis sampling period.
Q. How similar is this response to that observed in the locus coeruleus, in terms of responding to salient auditory and visual stimuli? And if the responses differ, why do you believe the effect differs from what’s observed in the locus coeruleus?
A. I think that norepinephrine (NE) cells in the LC and DA cells in the VTA and SN all respond under conditions of salient environmental change. If so, then the different behavioral functions of NE and DA are due, not to the conditions that elicit their release, but to the terminal sites that receive the transmitters. NE neurons of the LC send axons throughout the brain and densely innervate the cerebral cortex. Midbrain DA neurons largely innervate brain regions related to motoric output (e.g., the dorsal and ventral striatum, and frontal cortex). NE is in a position to alert perceptual areas of the brain that may mediate vigilance, while DA alerts brain regions that facilitate motor responses to incoming events (as well as working memory functions within the prefrontal cortex). Both NE and DA appear to facilitate processing of concurrent input signals to target regions by enhancing signal-to-noise ratios. DA suppresses the impact of weak glutamate input signals and boosts the efficacy of strong glutamate inputs to the striatum.
Q. Were animals implanted with electrodes chronically or acutely?
A. Animals were chronically implanted with microwires to the VTA that could be moved dorsal-ventral. Over the course of several months, they were screened for DA unit activity. When an isolated DA unit was identified behavioral tests were carried out.
Q. Can you monitor the same cell more than once?
A. Yes, sometimes we observe actions of a single cell over days.
Q. Your results showing that DA antagonists disrupt the weakly-elicited behavior parallel anectodal evidence that Parkinson’s patients behave normally in the face of strong eliciting stimuli.
A. Yes, the Parkinson’s patients show this phenomenon seen in the rat under D1 and D2 receptor blockade.
Comment: From another perspective the DA antagonist might be said to improve the animal’s discriminative responding by reducing ‘generic noise’; so despite the deficits created with DA antagonists, there are also improvements in some areas of performance.
Q. In your experiments in which the animals looks in the box for the food pellet, can you comment on why the animals doesn’t simply keep their head in the box at all times?
A. I think that there are postural factors that make it difficult for the animal to keep its head in the food compartment. Interestingly, we have tried to train animals to maintain this posture for long periods of time, and found that they had trouble maintaining this fixed posture, particularly under conditions of DA receptor blockade.
Q. In your experiments, where do the rats actually eat the pellets?
A. On the basis of video records, they appear to eat in the chamber, but we haven’t looked at this rigorously.
Q. In your studies, was the interval normally distributed?
A. It was an exponential distribution.
Q. Previous data, along these lines, suggested that neuroleptic-treated animals showed deficits in an initial operant component to a task, but in a latter ‘choice’ component of the response, they were relatively unimpaired.
A. Yes. In that study, as here, DA receptor blockade disrupted responses that were distal to the goal, but proximal responses remain intact.
Q. The D2 antagonist increased the animal’s acquisition of a new S-R association, while the D1 antagonist disrupted the association. But don’t D2 antagonists attenuate reward?
A. There is reason to believe that they may. This study did not directly test that hypothesis. One could argue, here, that animal exposed to 16 days of CS-US (Click-food) training had acquired strong reward-associations to the initial CS (the click). On day 17, when the animal was asked to associate a new Tone with the food, it was paired with the previous CS (click) which had already acquired strong incentive value. These conditioned reward properties may have been intact during the new CS acquisition.
Q. Do VTA DA neurons have an autoreceptor?
A. VTA DA neurons of the mesolimbic system do, but those that project to the frontal cortex, the mesocortical DA neurons, do not.
Q. What results would you predict if you paired new tones with food?
A. Both acquisition and expression would be modulated by the DA antagonist.
Comment: Anthony Azzara recently demonstrated that D1 antagonism blocks acquisition, but not expression, of flavor preferences, while D2 antagonism cannot block either one. Recently this group also found that D1 antagonism blocks new, weak preferences (i.e. acquisition), but has no effect on over-trained preferences (expression).
Comment: The present findings are in agreement with those findings.
Q. How do you account for DA antagonist effects on the maintenance of behavior?
A. DA antagonists produce strong disruption on response maintenance. We find similar deficits in the maintenance of a feeding response and in the maintenance of exploratory locomotor behavior in neuroleptic-treated rats.
Q. Why is the behavior relatively normal at the onset of the session in animals treated with DA antagonists?
A. There are two possibilities that seem likely to me: First, early in the session the incoming stimuli are novel and are “above-threshold” to elicit motor responses even with the raised sensory threshold produced by DA antagonist challenge. Or, you may get a burst in DA release very early in the feeding session due to the novel, salient stimuli present to the animal. This elevated DA concentration in the synapse competes with the DA antagonist for postsynaptic receptor sites, so that normal behavior is seen early in the session.. Actually, you could also get a combination of the above two scenarios.
Comment [Sclafani]: We do see declines in the onset of licking when we block DA action.
Q. If you give a range of tones of decreasing intensities, can you alter the threshold for the learned association, or get a sense of the precise intensity needed to form the association?
A. This has not yet been tested.
Q. Have you compared different pharmacological interactions, for example, to determine how opiates might alter DA-mediated acquisition learning?
A. No. That would be interesting to look at.
Q. What about the effect that neuroleptics have on weight gain? Do you think the effects you’ve investigated could be related?
A. I’m not sure whether or not those effects are related to these findings. I’m not clear on the mechanism by which neuroleptic-treated patients gain weight.