Deletion of the Opioid Receptor Gene Impairs Place Conditioning But Preserves Morphine Reinforcement

Background: Converging experimental data indicate that opioid receptors contribute to mediate drug reinforcement processes. Whether their contribution reflects a role in the modulation of drug reward or an implication in conditioned learning, however, has not been explored. In the present study, we investigated the impact of receptor gene knockout on reinforced conditioned learning under several experimental paradigms.

W ithin the endogenous opioid system, opioid receptors are necessary for most drugs of abuse to exert their rewarding properties (1). Pharmacological evidence suggests that ␦ opioid receptors play a similar role. Delta agonists elicit conditioned place preference (CPP) (2)(3)(4)(5), whereas antagonists attenuate CPP to cocaine, methamphetamine, or morphine (6 -8) and alter cocaine self-administration (9). In these studies, however, ␦ compounds might produce their effects via partial activation of receptors (10,11). Gene knockout therefore represents a unique approach to tackle the role of ␦ receptors in vivo. At present however, behavioral characterization of ␦ receptor knockout mice (Oprd1 Ϫ/Ϫ ) failed to clarify the implication of ␦ receptors in drug reinforcement. Although morphine-induced CPP was decreased in Oprd1 Ϫ/Ϫ mice (6), cannabinoid-induced CPP (12) and morphine self-administration into the ventral tegmental area (13) were preserved.
A major difficulty when assessing drug reinforcement in animal models lies in the tight intertwining of reward and learning processes. Most animal models used to measure rewarding properties of drugs also assess conditioned learning (14), and pharmacological data point toward a role of ␦ receptors in learning processes (15,16). Delta receptor blockade or genetic inactivation might cause deleterious effects on learning rather than reward, which might explain the previously observed deficits in drug reinforcement. To challenge this hypothesis, we examined acquisition of either appetitive (morphine-induced) or aversive (lithium-induced) place conditioning in Oprd1 Ϫ/Ϫ mice. We assumed that altered opioid reward in mutant mice would specifically result in decreased morphine-induced CPP, whereas a deficit in learning would impact on aversive as well as appetitive conditioning. To further assess ␦ receptor contribution to conditioning processes, we tested Oprd1 Ϫ/Ϫ mice for operant morphine self-administration. We evaluated morphine efficiency as a positive reinforcer with a fixed ratio 1 (FR1) schedule of reinforcement and motivation to earn morphine injections under a progressive ratio (PR) schedule.

Methods and Materials
Independent groups of Oprd1 Ϫ/Ϫ mice (17) and their wild-type (WT) control subjects were trained for morphine-induced (5, 10, 20 mg/kg, 6 days) or lithium-induced (3 mEq/kg, 3 days) place conditioning and tested either drug-free or under drug effects (statedependency). Other groups of mice were trained to self-administer morphine (.25 or .5 mg/kg/infusion) under an FR1 schedule of reinforcement for 10 days. Animals that completed self-administration criteria were moved to a PR schedule of reinforcement on Day 11. The PR session was performed only once and allowed the determination of a breaking-point for each animal (see Supplement 1 for complete protocols and surgical methods).

Intact Spatial Novelty Discrimination
We first performed a novelty discrimination test to examine whether mutant mice discriminate spatial cues under the same conditions as for place conditioning. During choice session, WT and

Decreased CPP to Both Appetitive and Aversive Stimuli
We submitted Oprd1 Ϫ/Ϫ mice to several place conditioning paradigms. First, distinct groups of animals were tested in a drug-free state after conditioning to saline or morphine (5-20 mg/kg). The WT but not mutant mice acquired morphine CPP [gender: F (1,48) ϭ .14,  Figure 1E). Altogether the data indicate that both morphine CPP and lithium CPA are impaired in Oprd1 Ϫ/Ϫ mice.

Discussion
In our study, Oprd1 Ϫ/Ϫ mice failed to express morphine CPP over a 5-20-mg/kg dose range. Mutant animals as WT mice, however, were able to discriminate novel from familiar compartment in the place conditioning apparatus. This suggests that decreased morphine CPP in mutant mice does not result from altered recognition of spatial contexts, at least in the CPP apparatus. Furthermore, when tested under the effects of morphine, Oprd1 Ϫ/Ϫ mice displayed preference for the morphine-paired context, similar to WT animals. Morphine-induced CPP was thus state-dependent in Oprd1 Ϫ/Ϫ mice. Otherwise, state-dependent morphine-induced CPP was not observed in mice lacking opioid receptors, consistent with the essential role of receptors as the primary target for morphine in vivo. State-dependency qualifies a behavioral response that can only be retrieved when the animal experiences the same (drug) state as during the acquisition of this response (18). Interoceptive drug cues can then function as conditioned stimuli and contribute to contextual information together with external cues. In the present study, Oprd1 Ϫ/Ϫ mice and not their WT counterparts needed both internal and external cues to express a morphine-induced CPP. Together, these data suggest that the ability to form drug-context associations and/or retrieve such associations was blunted in Oprd1 Ϫ/Ϫ mice.
To challenge this hypothesis, we tested Oprd1 Ϫ/Ϫ mice in a place aversion-conditioning paradigm with lithium chloride (3 mEq/kg) as a potent nonopioid aversive stimulus. Knockout mice failed to display lithium-induced CPA when tested drug-free but showed strong aversion for the lithium-paired compartment when tested under the effects of lithium. Thus, as for morphine-induced CPP, lithium-induced CPA was state-dependent in Oprd1 Ϫ/Ϫ mice, a finding that strengthens the hypothesis of deficient drug-context associations. Altogether, impaired appetitive and aversive place conditioning demonstrate that reinforced learning is altered in Oprd1 Ϫ/Ϫ mice, regardless of the affective value of the stimuli. This is consistent with previously reported decreased morphine CPP in Oprd1 Ϫ/Ϫ mice (6) as well as in WT animals injected with ␦ antagonists before conditioning sessions (6,8). Notably, such effects were not detected in a cannabinoid CPP paradigm (12). In the latter study, increased number of drug-pairings (five sessions vs. three) could have facilitated drug-context associations.
Our place conditioning data provide evidence for impaired drug-context association but do not exclude that drug reward is also reduced in ␦ receptor knockout mice. Hence we further addressed consequences of the gene knockout on opioid conditioned reinforcement with operant morphine self-administration. The WT and Oprd1 Ϫ/Ϫ mice similarly acquired morphine self-administration and reached the acquisition criteria (FR1). This result is consistent with previous studies showing preserved operant responding for food (19) or intra-ventral tegmental area morphine self-administration (13) in Oprd1 Ϫ/Ϫ mice. Hence, morphine served as an efficient positive reinforcer in both WT and knockout mice. Moreover, mutant mice achieved a similar or even higher breaking-point than WT control subjects under a PR schedule, strongly suggesting that motivation for morphine reinforcement is preserved in the absence of ␦ receptors. This result contrasts with the prevailing view of comparable roles of and ␦ opioid receptors in mediating drug reinforcement (see introductory text) and further documents the notion of divergent ␦ and receptor activities previously established for emotional responses (17) and motor impulsivity (19).
In conclusion, the present study demonstrates that ␦ receptor gene knockout alters place conditioning, whereas morphine reinforcement is maintained. The place conditioning phenotype sug- Mice from WT and Oprd1 Ϫ/Ϫ lines discriminate between the active and the inactive holes and acquire self-administration for intravenous morphine at (A) .25 or (B) .5 mg/kg/infusion. The data represent the number of nose-pokes in each hole/1-hour daily session. Data are expressed as mean Ϯ SEM (.25 mg/kg/infusion: n ϭ 3-7/gender and genotype; .5 mg/ kg/infusion: n ϭ 6 -10/gender and genotype). 1 solid star/*p Ͻ .05; 2 solid stars/**p Ͻ .01, comparison between holes (one-way analysis of variance). (C) The Oprd1 Ϫ/Ϫ mice achieved a similar breaking-point as WT mice under a progressive ratio schedule of reinforcement. Data analysis at each dose separately shows increased breaking-point at the high dose only, 1 open star p Ͻ .05, comparison between genotypes (one-way analysis of variance).
gests that ␦ receptor activation might facilitate drug-context associations, an effect that might influence contextual aspects of addiction-related behaviors. Neural mechanisms through which these receptors influence conditioning, however, remain to be elucidated. Remarkably, these receptors are expressed in all key brain regions controlling conditioned learning processes (1,20), where they can fine-tune reinforced memory processes.