Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Cocaine-dependent individuals tend to relapse into cocaine use even after long periods of abstinence. It is important to find factors that contribute to such a relapse. Animal studies have shown that drug-seeking behavior can be elicited not only by a low dose of the original self-administered drug (Gerber and Stretch, 1975; de Wit and Stewart, 1981; Stewart and Wise, 1992; Self et al., 1996), but also by drugs from pharmacological groups other than that of the original self-administered drug (Slikker et al., 1984).

Previously, it was shown that caffeine reinstated responding during an extinction phase in rats trained to self-administer cocaine (Worley et al., 1994; Schenk et al., 1996) or potentiated cocaine effects on the reinstatement of self-administration behavior (Self et al., 1996). The nature of this cue is not understood presently. Caffeine does not generalize well to cocaine in a drug-discrimination paradigm, and it is poorly self-administered (Griffiths and Woodson, 1988; Gauvin et al., 1990). However, cocaine substituted for the caffeine-discriminative stimulus in rats trained to discriminate caffeine from saline (Holtzman, 1986). Thus, how caffeine and cocaine can substitute for each other, and to what extent, remains to be determined.

Of all of the known primary biochemical mechanisms of action of caffeine, only antagonism of adenosine effects on adenosine A₁ and A_2A receptors is known to be important at concentrations of caffeine that produce behavioral stimulation (see Fredholm, 1995). Adenosine A₁ receptors are widely distributed in the brain (Johansson et al., 1993) and regulate the release of several neurotransmitters (Fredholm and Dunwiddie, 1988). In contrast, adenosine A_2A receptors are largely restricted to neurons that release -aminobutyric acid that also express dopamine D₂ receptors (Fink et al., 1992; Svenningsson et al., 1997a), and there is a functionally important negative interaction between adenosine A_2A and dopamine D₂ receptors (see Ferré et al. 1997). Perhaps secondarily to its primary effects, caffeine also influences and interacts with a host of transmitter systems in the brain (see Daly, 1993).

Recently, it was shown that an agonist at dopamine D₂ receptors, but not one at D₁ receptors, could reinstate cocaine-seeking behavior (Self et al., 1996). Given that D₂ and A_2A receptors are colocalized, that they are mutually antagonistic (Ongini and Fredholm, 1996; Ferré et al., 1997), and that an A_2A-receptor antagonist can mimic the stimulation of motor behavior produced by a D₂-receptor agonist (Svenningsson et al., 1995), it seemed possible that the ability of caffeine to reinstate the extinguished cocaine-seeking behavior is mainly associated with its ability to block A_2A receptors.

To test this possibility, we used a technique of the extinction of cocaine-associated behavior based on a model of one-session initiation of i.v. cocaine self-administration in naive mice (Kuzmin et al., 1997). With this model, we tested the ability of cocaine and caffeine to prevent the extinction of cocaine reward-associated responding in mice withdrawn from contingent reinforcing infusions of cocaine. Also, we examined the ability of selective adenosine A₁-receptor antagonists, 1,3-dipropyl-8-cyclopentyl xanthine (DPCPX) and 8-cyclopentyl theophylline (8-CPT), a selective A_2A-receptor antagonist, 5-amino-7-(2-phenylethyl)-2-(2-furyl)pyrazolo-[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine (SCH 58261; see Ongini and Fredholm, 1996), and a nonselective, nonxanthine, adenosine-receptor antagonist, 9-chloro-2(2-furyl)[1,2,4]triazolo[1,5-c]quinazolin-5-amine (CGS 15943) to mimic the effect of caffeine.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Animals. All experiments were carried out on male DBA/2 mice (18-22 g). Animals were obtained from the breeding farms in Rappolovo, Russia and Bomholtgård, Denmark and kept under standard laboratory conditions with unlimited access to food and water. They were housed 12 mice per cage in a light-controlled room (12 h light/dark cycle; light on at 10:00 AM) at 21°C and 60% humidity. The experiments were approved by the respective regional animal ethics boards.

Drugs. Caffeine (as free base; Sigma Chemical Co., St. Louis, MO) and cocaine (as hydrochloride; Sigma Chemical Co.) were dissolved in warm saline and injected i.p. just before the experiment. SCH 58261 (a gift from Dr. Ennio Ongini, Schering-Plough, Milan, Italy), CGS 15943 (Research Biochemicals International, Natick, MA) and DPCPX (Research Biochemicals International) were dissolved in dimethyl sulfoxide (DMSO), and stock solutions were diluted in saline so that the final concentration of DMSO was 20% (v/v). 8-CPT (Research Biochemicals International) was dissolved in 0.01 N NaOH and the pH of the solution was adjusted to 7.2 with 0.01 N HCl. Cocaine HCl for i.v. self-administration was dissolved in saline and the pH of the solution was adjusted to 7.2 with 0.01 N NaOH. Doses of cocaine refer to the salt.

Intravenous Self-Administration Method. The model of one-session initiation of i.v. self-administration of cocaine was used (40 µg/kg/infusion; 1.6 µl per infusion; fixed ratio = 1; naive mice; Kuzmin et al., 1997). This unit dose was chosen on the basis of preliminary testing in the range from 10 to 80 µg/kg/infusion. Mice were tested in pairs in cages (8 × 8 × 8 cm) that had a frontal hole (1.5 cm) for nose poking fitted with infrared sensors (3 mm into the hole from the inner surface of the cage) interfaced to an operating computer that controlled the automatic syringe pump. Mice were partially immobilized by fixing their tails with adhesive tape to the horizontal surface. The tails protruded through a vertical slot in the back wall of the box.

Extinction of Cocaine Reward-Associated Responding. On day 2, extinction sessions were carried out. Just before placement into boxes, both active and yoked control mice were injected i.p. (1 ml/kg) with saline or one of the compounds investigated: cocaine HCl (2.5-20 mg/kg); caffeine (3-30 mg/kg); CGS 15943 (1.25-5 mg/kg); SCH 58261 (0.312-2.5 mg/kg); DPCPX (1.25-5.0 mg/kg); or 8-CPT (5-20 mg/kg). There were six pairs in the treatment groups except in the groups given cocaine and caffeine treatment (n = 8). After the injection, pairs of mice were placed into exactly the same experimental boxes and partially immobilized as in the initiation experiments (day 1), but the needles were not introduced and cocaine infusions were not activated. As in the initiation session, the number of nose pokes of each mouse in the pair was counted and analyzed.

Data Analysis and Statistics. The number of nose pokes for active and yoked control mice in each treatment group was analyzed with two-way ANOVA; mode (active mice versus yoked control mice) and dose (saline/vehicle and drug treatments) were the independent factors. Next, paired comparison (two-tailed Student's t test) was performed to compare active and yoked control activity within each group of drug-injected animals and nose-poke activity in active and yoked control animals between the drug-treated group and a saline/vehicle-treated group. The whole study was designed as a between-subjects (independent groups) experiment (i.e., each treatment we describe was done on a single set of animals).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Initiation of Cocaine/Saline Self-Administration. The initial nose-poke activities (in a 10-min trial without infusions) did not differ significantly between the different groups of animals. The mean number of nose-poke responses in the various groups ranged from 23.88 ± 14.13 to 31 ± 12.87 during 10 min of testing (lowest and highest levels for different groups; mean ± S.E.M.). However, it is important to note that the difference in nose-poke activity between individual pairs was significant (from 16 to 88 in 10 min; lowest and highest activity in pairs). Two-way ANOVA showed no difference between treatment groups (p = .49) or between active and yoked control mouse subgroups (p = .86), and there was no interaction between group and subgroup factors. From this, it can be concluded that the matching between and within experimental groups was successful. Thus, any reinforcing action of the contingent infusions of the drug solution was expected to result in a larger number of nose pokes by the active mouse than by the yoked control.

Extinction. On day 2, mice were given i.p. injections of different drugs or saline/vehicle and placed into exactly the same experimental boxes as in the initiation experiment. In the groups injected with saline/vehicle, the number of nose pokes was not significantly different between active and yoked control mice (see Figs. 1-3). The ratio between the response rate in the two paired mice was close to 1 and, hence, the r criterion did not differ from 0. This was interpreted as the extinction of the behavior acquired on day 1 in the absence of response-contingent cocaine infusions. Injection of the vehicle for adenosine antagonists (20% DMSO) produced a parallel decrease in nose pokes in both active and yoked control mice as compared with saline-treated animals. However, the r criterion remained on the same level after vehicle treatment as after saline treatment.

View larger version (26K): [in this window] [in a new window]   Fig. 1.   Influence of cocaine and caffeine on the extinction of cocaine-seeking behavior in mice. Top, nose-poke responses (mean ± S.E.M.; n = 8). Dark bar and filled circle, active mice. Light bar and open circles, yoked control mice. Doses of cocaine (left) and caffeine (right) are given in mg/kg. *p < .05; Student's t test, comparison between active and yoked control mice within group of animals injected with drug during day 2 of the experiment. Bottom, r criteria in animals injected with cocaine or caffeine on day 2; mean ± S.E.M.; n = 8. The r criterion is defined as the logarithm (Lg) of the ratio in nose-poke activity of active (A) animals over yoked control (YC) animals. The mean ± S.E.M. of the r criterion in the acquisition experiment (day 1) is presented as a crossed field. p < .05; Student's t test, comparison between the ratio on day 2 in the drug- versus saline-treated groups of pairs of mice. #p < .05; Student's t test, comparison between the r criterion obtained in the extinction experiment (day 2) and the r criterion obtained in the acquisition experiment (day 1).

Effects of Cocaine in the Extinction Phase. Injection of cocaine had a dose-dependent influence on the operant pattern in mice trained to self-administer cocaine (Fig. 1). Administration of cocaine at the dose of 5.0 mg/kg increased specifically the nose-poking response of the active animals that had learned to self-administer cocaine on day 1 (p < .05 as compared with the saline-treated group). ANOVA revealed a significant effect of cocaine on the number of nose pokes in the active mice (F_(4,29) = 4.31; p = .007; n = 8 animals per group). However, this effect of cocaine (5.0 mg/kg) was not seen in the active animals trained for saline self-administration on day 1 (data not shown) or in yoked control animals that were trained with either cocaine or saline infusion on day 1. With the higher doses of cocaine (10 and 20 mg/kg), there were no significant differences in nose-poking activity between the animals in the pair.

Effects of Caffeine in the Extinction Phase. ANOVA failed to reveal a significant effect of caffeine treatment on the number of nose pokes in groups. However, there was a significant influence of caffeine dose on the level of the r criterion in the extinction experiment (F_(3,26) = 7.04; p = .001; n = 8 animals per group) with a significant increase of the r criterion (p < .01, compared with the saline-injected group) in the group injected with caffeine (3 mg/kg). Paired comparison revealed a significant difference between r criteria in the initiation versus the extinction experiment in the groups injected with saline (p = .002) or with caffeine at the doses of 10 mg/kg (p = .0002) and 30 mg/kg (p = .002; Fig. 1, bottom). There was no difference between the r criteria in initiation and extinction sessions in the group treated with caffeine at the dose of 3 mg/kg. Thus, it was concluded that i.p. administration of caffeine at a dose of 3 mg/kg maintained the cocaine-oriented pattern of behavior in mice in the absence of contingent, reinforcing cocaine infusions, and prevented the extinction of cocaine-oriented behavior, whereas higher doses of caffeine lacked this effect. To find the reason for these effects, paired comparison of the nose pokes in particular groups was performed. Among the active mice trained to self-administer cocaine, those injected with caffeine (3 mg/kg) exhibited more nose-poke responses than did those injected with saline (p = .05; Fig. 1). No such difference was seen in active mice trained for saline self-administration or in either of the yoked control groups. The highest dose of caffeine (30 mg/kg) increased responding in both active mice and yoked controls, but the effect did not reach statistical significance (p = .08). Thus, caffeine at the low dose (3 mg/kg) tended to increase specifically nose-poke responses of the active animals that had learned to self-administer cocaine on day 1.

Effects of the Nonselective Adenosine-Receptor Antagonist CGS 15943 in the Extinction Phase. First, we examined whether the effect of caffeine could be reproduced by CGS 15943, a structurally unrelated antagonist, at adenosine A₁ and A_2A receptors. CGS 15943 is approximately 1000-fold more potent than caffeine at both adenosine A₁ and A_2A receptors (see Ongini and Fredholm, 1996). However, it is much less water-soluble and, hence, its apparent volume of distribution is much higher. For this reason, it must be given in vivo in doses close to those of caffeine. As seen in Fig. 2, in none of the CGS 15943-treated groups (6 animals per group) was the r criterion different from that obtained in the initiation experiments, whereas such a difference was observed in the vehicle-treated group. Although ANOVA failed to show a significant influence of the drug on the nose-poke activity in the extinction experiment, we tested whether the increase in the r criterion with the highest dose of CGS 15943 was because of an increase in nose-poke responding in active mice or because of a decrease in activity in yoked control mice. The former was the case (p < .05; active and yoked control groups were compared with Student's t test). Thus, we concluded that CGS 15943 tends to prevent the extinction of the cocaine-oriented behavior.

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Influence of CGS 15943 and SCH 58261 on the extinction of cocaine-seeking behavior in mice. Top, nose-poke responses (mean ± S.E.M.; n = 6). Dark bar and filled circles, active mice. Light bar and open circles, yoked control mice. Doses of CGS 15943 (left) and SCH 58261 (right) are given in mg/kg. *p < .05; Student's t test, comparison between active and yoked control mice within the group of animals injected with a drug during day 2 of the experiment. +p < .05; Student's t test, comparison between the drug-treated group and the vehicle-treated group. Bottom, r criteria in animals injected with CGS 15943 (CGS) or SCH 58261 (SCH) on day 2 (mean ± S.E.M.; n = 6). The mean ± S.E.M. of the r criterion in the acquisition experiment is presented as a crossed field. #p < .05; Student's t test, comparison between the r criterion obtained in the extinction experiment (day 2) and the r criterion obtained in the acquisition experiment (day 1).

Effects of the Adenosine A_2A-Receptor Selective Antagonist SCH 58261 in the Extinction Phase. Next, we examined the effect of a highly selective adenosine A_2A-receptor antagonist, SCH 58261, which is structurally related to CGS 15943 (see Ongini and Fredholm, 1996). This compound is at least as potent as CGS 15943 as an antagonist at A_2A receptors, but at least 100 times less potent at A₁ receptors. Therefore, it was given in doses similar to those of CGS 15943. As seen in Fig. 2, lower doses of this compound had little effect on nose-poke responding, whereas higher doses produced a marked decrease. ANOVA revealed a significant effect of SCH 58261 injection on nose-poke responses in both active and yoked control mice (F_(4,25) = 22.5; p < .0005 and F_(2,25) = 20.5; p < .0005, respectively; 6 animals per group). There was a significant reduction (compared with vehicle-injected group) of the operant activity in the groups injected with SCH 58261 at doses of 1.25 or 2.5 mg/kg (p < .01 for both doses). Treatment with 0.312 and 0.625 mg/kg had no effect on operant responses in either active or yoked control mice (comparison with vehicle-treated group). ANOVA also failed to reveal any significant effect of SCH 58261 on the r criteria in the extinction experiment. Paired comparison of the levels of r criteria in the initiation versus the extinction experiments revealed that in all of the groups injected with SCH 58261, the r criteria in the extinction experiments differed significantly from those in the initiation experiments. Thus, at none of the doses used did the selective adenosine A_2A-receptor antagonist SCH 58261 prevent the extinction of cocaine-associated behavior.

Effects of the Adenosine A₁ Receptor Selective Antagonists DPCPX and 8-CPT in the Extinction Phase. DPCPX is slightly (about 3-fold) more potent than CGS 15943 on adenosine A₁ receptors, but almost 1000-fold less potent than CGS 15943 at A_2A receptors (see Ongini and Fredholm, 1996). Because the physicochemical properties of the two compounds are similar, the doses used were also similar. The results of experiments with DPCPX are presented in Fig. 3. ANOVA revealed significant differences between nose-poke responses in active and yoked control animals after treatment with DPCPX (F_(1,40) = 5.68; p = .02; 6 animals per group), although the effect was not dependent on the dose of DPCPX administered. Subsequent paired comparison showed that at 5.0 mg/kg DPCPX, there was a significant increase in the active group of mice (p = .04, comparison with the vehicle-treated group; p = .005, comparison with the corresponding yoked control group).

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Influence of DPCPX and 8-CPT on the extinction of cocaine-seeking behavior in mice. Top, nose-poke responses (mean ± S.E.M.; n = 6). Dark bar and filled circles, active mice. Light bar and open circles, yoked control mice. Doses of DPCPX (left) and 8-CPT (right) are given in mg/kg. *p < .05; Student's t test, comparison between active and yoked control mice within the group of animals injected with drug during day 2 of the experiment. Bottom, r criteria in animals injected with DPCPX or 8-CPT on day 2 (mean ± S.E.M.; n = 6). The mean ± S.E.M. of the r criterion in the acquisition experiment is presented as a crossed field. p < .05; Student's t test, comparison between the ratio on day 2 in the drug- versus saline-treated groups of pairs of mice. #p < .05; Student's t test, comparison between the r criterion obtained in the extinction experiment (day 2) and r criterion obtained in the acquisition experiment (day 1).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

This study shows that the initiation and the extinction of cocaine-related behavior can be studied not only in rats, but also in mice. A single, 30-min session in mice established self-administration of cocaine by means of directed nose pokes. This behavior was rapidly extinguished in mice that received only saline/vehicle injections during testing on a subsequent day. However, an i.p. injection of cocaine at 5.0 mg/kg, a dose known to exert a reinforcing effect in a conditioned place-preference paradigm in mice (Kuzmin et al., 1997), was able to stabilize the higher level of nose-poke activity in active animals compared with the yoked control animals even when withdrawn from the contingent cocaine infusions. The ability of cocaine to effectively maintain cocaine-oriented behavior in the absence of response-contingent injections of cocaine was not surprising because others have shown that noncontingent administration of the self-administered drug produces potent reinstatement of extinguished drug-taking behavior (Gerber and Stretch, 1975; de Wit and Stewart, 1981; Slikker et al., 1984; Self et al., 1996). Conversely, the highest dose of cocaine used in the present study (20 mg/kg), a dose that does not induce conditioned place preference in mice (Kuzmin et al., 1997), lacked a priming effect during the extinction session. Thus, it can be speculated that only reinforcing doses of the drugs are able to prevent an extinction of drug-related behavior.

In our experiments, caffeine at the dose of 3 mg/kg i.p. also effectively prevented the extinction of a cocaine-induced operant pattern. This finding confirms and extends previous data showing the ability of caffeine to reinstate cocaine-seeking behavior in rats (Worley et al., 1994; Schenk et al., 1996). However, Self and coworkers (Self et al., 1996) found that caffeine at the dose we used for this study (3 mg/kg) lacked appreciable priming ability, whereas caffeine at 10 mg/kg tended to enhance cocaine priming. A probable explanation is that in our mouse strain, the caffeine dose-response curve is slightly shifted to the left compared with the rat strain used by Self and coworkers (Self et al., 1996), but there also are differences in the experimental procedure. Whereas in our study there was a 24-h delay after the cocaine self-administration session, in the cited study, the extinction and the effects of priming caffeine injections were studied 2 to 4 h after the cocaine self-administration session.

Whereas the lower dose (3 mg/kg) of caffeine had a clear-cut priming effect, a high dose (30 mg/kg) of caffeine induced an increase in nose poking in both active and yoked control mice, but this activation was not selective and lacks priming effect, according to the r criteria. Because caffeine has a biphasic reinforcing/aversive activity depending on the dose used (Brockwell et al., 1991), the dose of 30 mg/kg in mice might have some aversive properties. In fact, our preliminary experiments with the aid of the conditioned place-preference technique also showed that whereas caffeine at the dose of 3 mg/kg produced conditioned place-preference in mice, the dose of 30 mg/kg produced a strong place-aversion reaction (A.K., unpublished observation).

The ability of a low dose of caffeine to prevent the extinction of a cocaine-related behavior cannot be explained simply by its psychomotor stimulant properties (Schenk et al., 1989) because an increased responding was observed only in active animals and not in the yoked control mice. Thus, the effect of caffeine is related to an association of the operant responses with cocaine infusions on day 1 of the experiment. Moreover, it has been shown previously that the motor-activating effect of caffeine is not correlated with its ability to reinstate cocaine-associated responding (Schenk et al., 1989), and that with repeated administration, the effectiveness of a dose of caffeine in increasing motor activity remains unchanged, whereas its ability to reinstate cocaine-associated responding decreases (Schenk et al., 1996).

It is important to note that the nose-poke activity does not mirror the general motor activity in mice. In fact, administration of high doses of cocaine (Kuzmin et al., 1997) that produce definite stimulatory effect usually resulted in a decrease in the number of nose pokes in the technique used. This gives a unique opportunity to study selectively the reinforcing effects of drugs because the increase in nose-poke activity was found only in mice that actively self-administer cocaine and not in animals from the yoked control group.

It is possible that some component of the effects of caffeine can modify cocaine-sensitive systems, possibly by the activation of the central dopaminergic systems (Ferré et al., 1991; Garrett and Holtzman, 1994). As mentioned, it is generally believed that a major primary neurochemical effect of caffeine is the blockade of adenosine receptors (Daly, 1993; Fredholm, 1995), and this effect has been demonstrated as critical for the acute stimulant effects of caffeine (Kaplan et al., 1989; Holtzman, 1991). It also has been shown that activation of adenosine receptors can antagonize the effects of dopamine at both D₁ and D₂ receptors (see Ferré et al., 1997). There is a high density of A_2A receptors in the accumbens nucleus (Parkinson and Fredholm, 1990; Johansson et al., 1997), where they can modulate striatal dopaminergic neurotransmission. Furthermore, there are antagonistic adenosine A_2A-dopamine D₂ receptor interactions: activation of adenosine A_2A receptors decreases both the affinity of dopamine D₂ receptors and the signal transduction from dopamine D₂ receptors to the guanine nucleotide-binding protein (see Ferré et al., 1997). The finding that a dopamine D₂ agonist, but not a dopamine D₁ agonist, reinstated cocaine-seeking behavior in rats (Self et al., 1996) raised the possibility that A_2A-receptor antagonism in this terminal region of the mesolimbic dopamine system underlies the reinstatement effects produced by caffeine. Therefore, it was surprising that in our study, the selective adenosine A_2A antagonist could not exhibit the same behavioral effect as a low dose of caffeine in the extinction experiment despite the fact that this dose of the antagonist can mimic the stimulatory effect of caffeine on motor behavior in rats (Bertolli et al., 1996; Pinna et al., 1996). In fact, the A_2A-receptor antagonist caused a dose-dependent decrease in nose-poke responses in both active and yoked control mice and actually decreased the difference between them. In contrast, the drugs that inhibit adenosine A₁ receptors (i.e., DPCPX, CGS 15943, and, to a lesser extent, certain doses of 8-CPT) did prevent, or tended to prevent, the extinction of cocaine-induced behavior and maintained cocaine-associated responses in active mice on a higher level than the responses of the yoked control mice.

A critically important question is whether the present results are confounded by behavioral activation unrelated to cocaine seeking. It can be argued that a nonspecific motor stimulation might enhance any drug-related activity. This can only be partially controlled for by yoked controls for the initiation of such behaviors, and it would be particularly important if drug-induced activity persists at the time of the experiment. Unpublished experiments with the present method show that drug-related behavior is lost 10 to 20 min after the beginning of the extinction session and that the level of activity remaining is not different from that observed during a second or third extinction session (48 or 72 h after initiation; A.K., E.E.Z., unpublished data). Thus, the present method is, in this respect, not significantly different from other methods used to study reinstatement. Of course, it also could be argued that a motor stimulant would be able to induce an activity that the animal has learned to perform previously, even if that activity is not overt under basal conditions. This is true, and we have not been able to control for this. However, the same problem occurs in every method for the study of reinstatement. There is, finally, one important finding from a previous study from our laboratory that provides strong evidence against nonspecific motor stimulation being the explanation for the major finding in our study. Svenningsson and coworkers (Svenningson et al., 1997b) showed that DPCPX, the adenosine A₁ antagonist that enhanced cocaine-seeking behavior in the present study, is not a motor stimulant, but actually tended to reduce locomotor activity. Conversely, SCH 58261, which did not enhance cocaine-seeking behavior, acted as a motor stimulant in that study. For all of these reasons, we feel that nonspecific motor activation cannot explain the effects of A₁- and A_2A-receptor antagonists on cocaine-associated responding.

Adenosine A₁ receptors have a widespread distribution in the brain of most species (Fastbom et al., 1987) and are present in the nucleus accumbens, where they selectively counteract responses to dopamine D₁-receptor agonists (Ferré et al., 1996). In addition, adenosine A₁ receptors are located on the nerve terminals of several types of neurons and inhibit the release of many neurotransmitters (Fredholm and Dunwiddie, 1988). Thus, an adenosine A₁ antagonist might increase the release of endogenous dopamine (Stoner et al., 1988), which is known to activate dopamine receptors. Furthermore, adenosine A₁ antagonists enhance signaling via such receptors (see Ferré et al., 1996). Such activation of dopaminergic transmission is known to be part of the response to cocaine (Self and Nestler, 1995). However, recent findings have shown that reinstatement of cocaine-associated behavior by cocaine priming does not correlate with the dopaminergic neurotransmission in the accumbens nucleus (Neisewander et al., 1996). Furthermore, as noted above, a D₁ agonist did not reinstate cocaine-seeking behavior (Self et al., 1996), although D₁ agonists were shown to maintain self-administration behavior in rats (Weed et al., 1997). Thus, the schemes for explaining the actions of caffeine as well as selective adenosine-receptor antagonists in terms of interactions with known mechanisms involving dopamine receptors, schemes that work very well in explaining the actions of these compounds on motor behavior (see Ferré et al., 1997), break down when trying to explain the effects of these drugs in the extinction model of cocaine-seeking behavior. Consequently, the mechanism of the maintenance of drug-seeking behavior also must incorporate other neurotransmitter systems and probably other neuronal circuits (Koob and Le Moal, 1997). It also might be speculated that the receptors involved in reinforcement (D₁ for cocaine and, possibly, A_2A for caffeine) are different from the receptors involved in the processes of extinction and reinstatement (D₂ for cocaine and A₁ for caffeine).

In summary, the present results show that a mouse model can be successfully used to study the mechanisms of the extinction of drug-seeking behavior during withdrawal from the response-contingent reinforcement. Given the increasing availability of genetically modified mouse strains, this type of experimental model might prove valuable in future studies of the genetic mechanisms involved in establishment, maintenance, extinction, and reinstatement of drug-seeking behavior. With this model, we demonstrated a clear-cut biphasic effect on the extinction of cocaine-seeking behavior when using both the self-administered drug (cocaine) and a drug from a completely different drug class (caffeine). Finally, low, behaviorally stimulant doses of caffeine can prevent the extinction of cocaine-seeking behavior, possibly via the blockade of A₁ adenosine receptors.