Cocaine Self-Administration Experience Induces Pathological Phasic Accumbens Dopamine Signals and Abnormal Incentive Behaviors in Drug-Abstinent Rats

Abstract

Chronic exposure to drugs of abuse is linked to long-lasting alterations in the function of limbic system structures, including the nucleus accumbens (NAc). Although cocaine acts via dopaminergic mechanisms within the NAc, less is known about whether phasic dopamine (DA) signaling in the NAc is altered in animals with cocaine self-administration experience or if these animals learn and interact normally with stimuli in their environment. Here, separate groups of rats self-administered either intravenous cocaine or water to a receptacle (controls), followed by 30 d of enforced abstinence. Next, all rats learned an appetitive Pavlovian discrimination and voltammetric recordings of real-time DA release were taken in either the NAc core or shell of cocaine and control subjects. Cocaine experience differentially impaired DA signaling in the core and shell relative to controls. Although phasic DA signals in the shell were essentially abolished for all stimuli, in the core, DA did not distinguish between cues and was abnormally biased toward reward delivery. Further, cocaine rats were unable to learn higher-order associations and even altered simple conditioned approach behaviors, displaying enhanced preoccupation with cue-associated stimuli (sign-tracking; ST) but diminished time at the food cup awaiting reward delivery (goal-tracking). Critically, whereas control DA signaling correlated with ST behaviors, cocaine experience abolished this relationship. These findings show that cocaine has persistent, differential, and pathological effects on both DA signaling and DA-dependent behaviors and suggest that psychostimulant experience may remodel the very circuits that bias organisms toward repeated relapse.

Significance statement: Relapsing to drug abuse despite periods of abstinence and sincere attempts to quit is one of the most pernicious facets of addiction. Unfortunately, little is known about how the dopamine (DA) system functions after periods of drug abstinence, particularly its role in behavior in nondrug situations. Here, rats learned about food-paired stimuli after prolonged abstinence from cocaine self-administration. Using voltammetry, we found that real-time DA signals in cocaine-experienced rats were strikingly altered relative to controls. Further, cocaine-experienced animals found reward-predictive stimuli abnormally salient and spent more time interacting with cues. Therefore, cocaine induces neuroplastic changes in the DA system that biases animals toward salient stimuli (including reward-associated cues), putting addicts at increasing risk to relapse as addiction increases in severity.

Keywords: Incentive salience; prediction error; sign tracking; ventral striatum; voltammetry.

Figure 1.

A, Schematic of task phases. Self-administration sessions were 2 h per day for 14 d. Final extinction session was also 2 h. B, Rewarded self-administration presses for controls (open squares; water delivered to receptacle, yoked intravenous saline infusion; n = 20) and cocaine-administering subjects (black circles; 0.33 mg/inf i.v. cocaine; n = 15). C, Presses on the self-administration lever under extinction conditions after extended abstinence (control: n = 13; cocaine, n = 7). *p < 0.05, last 5 d self-administration versus extinction; †p < 0.05, control vs cocaine.

Figure 2.

Pavlovian conditioning after 1 month of abstinence from self-administration. First-order sessions were each 83 min, whereas second-order sessions were 3 h each. A, Number of food cup entries during 10 s cue presentations for the CS+ and CS− stimuli (minus the rate during the immediately preceding 10 s baseline) over 10 d of FOC in controls (n = 20) and cocaine (n = 10) rats. B, Number of food cup entries during the 10 s SOC cues during SOC sessions. *p < 0.05, CS+ vs CS−, SOC+ vs SOC−.

Figure 3.

Histological electrode placement for valid FSCV recordings. Black circles indicate core placements; gray circles indicate shell placements.

Figure 4.

Development of differential PCA behaviors across FOC sessions in controls (n = 13) and cocaine (n = 9) subjects. The percentage of time the rats spent either in ST (A) or GT (B) during the cue presentations varied by drug experience. C, Total amount of conditioned approach behavior (GT plus ST) during cues was the same between groups. D, The first contact index indicates the proportion of times the rats made their first response to either the cue light (+1) or the food cup (−1) during cue presentations. E, PCA index (the averaged indices of percentage of time, number of entries, probability of entry, and first contact for the cue and goal) during cue presentations. F, Percentage time in the goal cup in the 5 s immediately after reward delivery did not differ between groups. *p < 0.05, **p < 0.01, control vs cocaine; †p < 0.05, ††p < 0.01, cocaine CS+ vs cocaine CS−; @p < 0.05, @@p < 0.01, control CS+ vs control CS−.

Figure 5.

Interactions of identified first-order ST and GT phenotype with self-administration behavior and higher-order conditioning. Cocaine rats that showed greater ST behaviors during FOC were found to have self-administered cocaine at a higher rate (A) and to have emitted more extinction responses (B). C, Individual differences in ST or GT behaviors during FOC had no effect on the ability to perform the SOC discriminations; both GT and ST controls were able to discriminate the stimuli, whereas ST and GT cocaine subjects were both impaired at the discrimination. §p < 0.05, GT vs ST; *p < 0.05, SOC+ vs SOC−; †p < 0.05, control vs cocaine.

Figure 6.

Phasic DA recordings from the NAc core during a well learned first-order Pavlovian discrimination in controls (n = 8) and cocaine-experienced subjects (n = 8). Representative color plots of single CS+ trials from a control (A) and a cocaine-experienced (B) subject. Inset in top left corner shows the cyclic voltammogram (current vs voltage; CV) at the oxidation potential for DA from a representative transient release event on the color plot. Subject averages for controls (C) and cocaine (D) rats in the task showing real-time concentrations of DA aligned to the CS+ and CS− trials. E, Peak DA release immediately after cue onset shows decreased and nonspecific DA release in cocaine rats (red) that was reliably decreased relative to controls (white) during the CS+, but not CS−. F, Peak DA release during the reward period was selectively enhanced for the reward receipt in cocaine subjects relative to controls. Inset, Average difference between CS+ and reward DA was significantly above zero in controls and below zero in cocaine subjects. G, Total volume of DA release (summed concentration of DA during the 10 s cue period plus 2 s reward period) was greater for CS+ than CS− trials, but did not vary between cocaine and controls. *p < 0.05, CS+ vs CS−; †p < 0.05, control vs cocaine, §p < 0.05 vs baseline.

Figure 7.

Phasic DA recordings from the NAc shell during a well learned first-order Pavlovian discrimination in controls (n = 7) and cocaine-experienced subjects (n = 9). Representative color plots of single CS+ trials from a control (A) and a cocaine-experienced (B) subject. Inset in top left corner shows the CV from a representative transient release event on the color plot. Subject averages for controls (C) and cocaine (D) rats in the task show real-time concentrations of DA aligned to the CS+ and CS− trials. E, Peak DA release immediately after cue onset show decreased and nonspecific DA release in cocaine rats (blue) that was reliably decreased relative to controls (white) during the CS+, but not CS−. F, Peak DA release during the reward period was selectively enhanced for the reward receipt in controls relative to cocaine subjects. Inset, Average difference between CS+ and reward DA was similar to zero in both drug groups. G, Total volume of DA release was greater for CS+ than CS− trials only in controls. *p < 0.05, CS+ vs CS−; †p < 0.05, control vs cocaine.

Figure 8.

NAc core population DA concentrations (i.e., average of trials) for subjects classified as either ST (controls: n = 56; cocaine: n = 84) or GT (controls: n = 56; cocaine: n = 28). In controls, the CS+ cue generally elicited more DA release than the CS− in both GT (A) and ST (B) subjects. C, Peak DA for controls showed greater DA at the CS+ in ST than GT, but similar DA at the reward. Cocaine-experienced animals showed less discrimination between cues in both the GT (D) and ST (E) groups. F, Peak DA in cocaine subjects did not differ at the cue, but did differ at the reward between ST and GT. §p < 0.05, cue vs reward; *p < 0.05, ST vs GT; †p < 0.05, control vs cocaine.

Figure 9.

NAc shell population DA concentrations (i.e., average of trials) for ST (controls: n = 28; cocaine: n = 70) and GT subjects (controls: n = 70; cocaine: n = 56). In controls, the CS+ cue generally elicited more DA release than the CS− in both GT (A) and ST (B) subjects. C, Peak DA for controls showed greater DA at the CS+ in ST than GT, but similar DA at the reward. Cocaine-experienced animals showed no discrimination between cues in both the GT (D) and ST (E) groups. F, Peak DA in cocaine subjects did not differ at the cue or reward between ST and GT. #p < 0.08, cue vs reward; *p < 0.05, ST vs GT; †p < 0.05, control vs cocaine.

Figure 10.

Correlations between peak DA release in both core and shell and approach behavior in ST (A, B) and GT (C, D) for the CS+ cue and the reward delivery. ST controls showed a significant correlation between cue-evoked DA and the rate of ST behavior, whereas ST cocaine subjects did not (A), but DA did not correlate with reward delivery in either group (B). The rate of GT behavior did not correlate with DA release to either the CS+ (C) or reward (D) regardless of drug treatment.

Figure 11.

DA signals of prediction-error-type signal (i.e., DA at reward minus DA at cue) from both the core and shell were significantly correlated with the total amount of cocaine ingested during self-administration (A) and the average number of drug loading presses (average number of presses made during the first 5 min of a session on the last 4 self-administration days) during stable self-administration (B).